Wan Lei · Anthony C.K. Soong Liu Jianghua · Wu Yong · Brian Classon Weimin Xiao · David Mazzarese Zhao Yang · Tony Saboorian 5G System Design An End to End Perspective Second Edition

Wireless Networks

Series Editor Xuemin Sherman Shen, University of Waterloo, Waterloo, ON, Canada The purpose of Springer's Wireless Networks book series is to establish the state of the art and set the course for future research and development in wireless communication networks. The scope of this series includes not only all aspects of wireless networks (including cellular networks, WiFi, sensor networks, and vehicular networks), but related areas such as cloud computing and big data. The series serves as a central source of references for wireless networks research and development. It aims to publish thorough and cohesive overviews on specifc topics in wireless networks, as well as works that are larger in scope than survey articles and that contain more detailed background information. The series also provides coverage of advanced and timely topics worthy of monographs, contributed volumes, textbooks and handbooks.

** Indexing: Wireless Networks is indexed in EBSCO databases and DPLB ** More information about this series at http://www.springer.com/series/14180 Wan Lei - Anthony C. K. Soong Liu Jianghua - Wu Yong - Brian Classon Weimin Xiao - David Mazzarese Zhao Yang - Tony Saboorian

5G System Design

An End to End Perspective Second Edition

ISSN 2366-1186 ISSN 2366-1445 (electronic) Wireless Networks ISBN 978-3-030-73702-3 ISBN 978-3-030-73703-0 (eBook) https://doi.org/10.1007/978-3-030-73703-0 © Springer Nature Switzerland AG 2020, 2021 This work is subject to copyright. All rights are reserved by the Publisher, whether the whole or part of the material is concerned, specifcally the rights of translation, reprinting, reuse of illustrations, recitation, broadcasting, reproduction on microflms or in any other physical way, and transmission or information storage and retrieval, electronic adaptation, computer software, or by similar or dissimilar methodology now known or hereafter developed. The use of general descriptive names, registered names, trademarks, service marks, etc. in this publication does not imply, even in the absence of a specifc statement, that such names are exempt from the relevant protective laws and regulations and therefore free for general use. The publisher, the authors, and the editors are safe to assume that the advice and information in this book are believed to be true and accurate at the date of publication. Neither the publisher nor the authors or the editors give a warranty, expressed or implied, with respect to the material contained herein or for any errors or omissions that may have been made. The publisher remains neutral with regard to jurisdictional claims in published maps and institutional affliations. This Springer imprint is published by the registered company Springer Nature Switzerland AG The registered company address is: Gewerbestrasse 11, 6330 Cham, Switzerland

To all the global health-care and other front-line workers who worked so tirelessly to take care of us during the Covid-19 pandemic.

Preface To The First Edition

It is hard to overstate the impact that HSPA and LTE technology have had on our global society. Mobile subscriptions for these technologies are now counted in the billions, touching lives and changing business everywhere on the planet. How did this come to pass? The dominance of these 3rd Generation Partnership (3GPP) technologies was not known a priori when their studies were frst approved. There are many factors to explain the success, but surely these factors include the quality of the standards and technology as well as economies of scale from global deployment. For 5G, the ffth generation of telecommunication systems, the situation is different in that it is expected, even before the frst equipment delivery, that 3GPP technology will dominate future global deployments. The vision for 5G also goes beyond traditional mobile broadband services. In 2015, ITU-R (International Telecommunication Union-Radio communications sector) established the 5G requirements for IMT-2020 (International Mobile Telecommunication system2020), targeting diverse requirements from three key usage scenarios: enhanced mobile broadband (eMBB), massive machine-type communication (mMTC), and ultrareliable and low-latency communication (URLLC).

Mobile broadband services like web browsing, social apps with text messaging, fle sharing, music downloading, and video streaming are already very popular and supported by 4G communication systems. In the 5G era, these and other applications such as ultrahigh-defnition (UHD) video, 3D video, and augmented reality (AR) and virtual reality (VR) will be better served with data rates up to hundreds of megabits per second or even gigabits per second. In addition, the demand of uplink high data rate service is also emerging, for example with HD video sharing. These service requirements, together with the anywhere anytime experience requirements with high user density and user mobility, defne new limits for the eMBB scenario.

In the future, any object that can beneft from being connected will be connected, either partially or dominantly, through wireless technologies. This trend poses a huge demand for connecting objects/machines/things in a wide range of applications. Driverless cars, enhanced mobile cloud services, real-time traffc control optimization, emergency and disaster response, smart grid, e-health, and industrial communications, to name just a few, are all expected to be enabled or improved by wireless connectivity. By a closer observation of these applications, there are two major characteristics of these services: one is the number of desired connections; the other is the requested reliability within a given latency budget. These two signifcant characteristics drive the defnitions of the mMTC and URLLC scenarios.

Accordingly, ITU-R defnes the IMT-2020 requirements for all the above potential use cases, including eight key capabilities: 20 Gbps peak rate, 100 Mbps user perceived data rate at cell edge, 3 times spectrum effciency of IMT-Advanced, mobility up to 500 km/h, low latency with less than 1 ms air interface round-trip time (RTT), connectivity density of 10 M connections per square kilometer, 100 times energy effciency of IMT-Advanced, and traffc density with 10 Mbps per square meter. It is anticipated that the area traffc capacity is predicted to be increased by at least 100 times in 5G, with available bandwidth increased by 10 times. Altogether, the 5G system can provide a basic information infrastructure to be used by both people and machines for all of these applications, similar to the transportation system and electric power system infrastructure that we use now.

Each region in the world is planning their 5G spectrum. In Europe, multiple countries have allocated 5G spectrum mainly in C-band and released 5G operational licenses among mobile network operators. The UK has already auctioned 190 MHz sub-6 GHz spectrum for 5G deployment, with another 340 MHz low-frequency spectrum auction ongoing. In Asia, China has newly allocated 390 MHz bandwidth in 2.6 GHz, 3.5 GHz, and 4.9 GHz frequency bands among the three major operators for 5G deployment by 2020; Japan has allocated totally 2.2 GHz 5G spectrum including 600 MHz in C-band and 1.6 GHz in 28 GHz mm-wave spectrum among four mobile carriers, with 5G investment around JPY 1.6 trillion ($14.4 billion) over the next 5 years; South Korea has auctioned 280 MHz bandwidth in 3.5 GHz and 2.4 GHz bandwidth in 28 GHz spectrum for 5G network among the three telco carriers, with SKT having already released the frst 5G commercial services since April 3rd, 2019. In the USA, 5G licenses are permitted in the existing 600 MHz, 2.6 GHz, and mm-wave frequency bands of 28 GHz and 38 GHz, with additional 3 millimeterwave spectrum auctions in 2019 on 28 GHz, 24 GHz, as well as higher mm-wave spectrum at 37 GHz, 39 GHz, and 47 GHz. Europe, Asia, and North America have all announced early 5G network deployment by 2020.

This treatise elaborates on the 5G specifcations of both the 5G new radio (5GNR) and 5G new core (5G-NC) and provides a whole picture on 5G end-to-end system and key features. Additionally, this book provides the side-by-side comparison between 5G-NR and long-term evolution (LTE, also called as 4G) to address the similarities and the differences, which benefts those readers who are familiar with LTE system. 3GPP Release-15, i.e., the frst release of 5G standard, has completed the standardization of both 5G non-stand-alone (NSA) and stand-alone (SA) architecture. 5G deployment will eventually go to SA deployment based on 5G new carrier (NC) with advanced core network features as slicing, MEC, and so on. Some operators, however, due to their business case balance between the signifcant investments and quick deployment, may consider NSA deployment from the beginning, i.e., with a primary connection to LTE and a secondary connection to NR. In addition to the network architecture, NR has built-in provisions in confguration and operation for coexistence with LTE. These include same-band (and even same channel) deployment of NR and LTE in low-band spectrum. An especially important use case is a higher frequency NR TDD deployment that, for coverage reasons, includes a supplemental uplink (SUL) carrier placed in an existing LTE band.

The book is structured into six main chapters. The frst chapter looks at the use cases, requirements, and standardization organization and activities for 5G. These are 5G requirements and not NR requirements, as any technology that meets the requirements may be submitted to the ITU as 5G technology including a set of radio access technologies (RATs) consisting of NR and LTE, with each RAT meeting different aspects of the requirements. A second chapter describes, in detail, the air interface of NR and LTE side by side. The basic aspects of LTE that NR builds upon are frst described, followed by sections on the NR-specifc technologies such as carrier/channel, spectrum/duplexing (including SUL), LTE/NR coexistence, and new physical layer technologies (including waveform, polar/LDPC channel coding, MIMO, and URLLC/mMTC). In all cases, the enhancements made relative to LTE are made apparent. The third chapter contains description of NR procedures (IAM/ beam management/power control/HARQ), protocols (CP/UP/mobility, including grant-free), and RAN architecture. The fourth chapter has a detailed discussion related to end-to-end system architecture, and the 5G Core (5GC), network slicing, service continuity, relation to EPC, network virtualization, and edge computing. The ffth chapter describes the ITU submission and how NR and LTE meet the 5G requirements in signifcant detail, from the rapporteur responsible for leading the preparation and evaluation. Finally, the book concludes with a look at the 5G market and the future.

Beijing, China Wan Lei Plano, TX, USA Anthony C. K. Soong Beijing, China Liu Jianghua Beijing, China Wu Yong Rolling Meadows, IL, USA Brian Classon Rolling Meadows, IL, USA Weimin Xiao Beijing, China David Mazzarese Shanghai, China Zhao Yang Plano, TX, USA Tony Saboorian

Preface To The Second Edition

In the 15 months since the publication of the frst edition and the writing of this treatise, 3GPP has completed Release-16 of 5G. It has enhanced 5G with many features that were requested by the vertical industries. Loosely speaking, 5G Release-15 can be characterized as being optimized for the cellular eMBB service while 5G Release-16 is the beginning of the optimization of 5G for the vertical industries. It mainly focused on the support of the vehicular vertical and industrial Internet of Things. As such, we have signifcantly altered the frst edition to cover the key features standardized in Release-16.

In particular, since the key focus of 5G Release-15 was on eMBB, it made sense to describe the key physical layer of NR in one chapter. That now no longer makes sense, and we have broken the chapter up into fve chapters. The frst two, Chaps. 2 and 3, describe the fundamental aspects of the LTE and NR air interface, respectively. By fundamental, we mainly mean the features that were needed to support eMBB services. The NR discussion in Chap. 3 has been enhanced with new discussions on MIMO enhancements, unlicensed access, positioning, and power savings.

The other three chapters are new to the 2nd edition; Chaps. 7, 8, and 9 discuss the beginning of our journey on 5G support for the verticals (URLLC, V2X, industrial/ manufacturing) after a complete end-to-end discussion of the fundamental aspects of 5G. The chapters are organized by verticals with descriptions of both Release-15 and -16 features that were developed in 3GPP mainly for that vertical contained in one place.

Chapter 4 contains the 5G RAN procedures, protocols, and architecture. It has been updated with the newly standardized beam management techniques and the IAB features. Other aspects standardized in Release-15 and -16 for the vertical industry have now been moved to their respective vertical chapters.

The end-to-end architecture is described in Chap. 5. This chapter has been enhanced with the newly standardized feature that accounts for the fact that there will exist a number of wireless technologies, some non-3GPP, for the user to access the Internet. As such, access traffc steering, switching, and splitting features were standardized in Release-16. These features enable a multi-access PDU connectivity service that allows one PDU session to be established with two simultaneous access connections, where one uses 3GPP access network and the other uses non-3GPP access network. The mechanisms for supporting non-3GPP access technology in 5G via untrusted non-3GPP access networks, trusted non-3GPP access networks, and wireline access networks are given. The chapter also discusses 5GC support for data analytics, cellular IoT, location services, and IMS.

Chapter 6 is a totally new chapter on 5G security. To ensure the entirety of 5G is secure, each architectural deployment model sports its own security architecture, features, and capabilities. The 5G security features, considerations, services, and capabilities are frst described. Security enhancements that mainly targeted at the support of URLLC, CIoT, service-base architecture, network slicing, 5WWC, radio voice call continuity, vertical industries, and IAB are then each described. Not only does 5G need to be secure, but it has to also give assurance to other entity of its security in order to function as the wireless access part of a larger global network. Consequently, the last part of the chapter investigates 5G security assurance.

The 5G features that support URLLC are elucidated in Chap. 7. Physical layer changes to the PDCCH, UCI, and PUSCH that allow for transmissions with short latency and enhance reliability are introduced. Since URLLC services are likely to coexist with other services, mechanisms exist in the standards to allow for an effcient support of services with different requirements in the same band. DL preemption indication and CBG-based retransmissions to facilitate DL inter-UE multiplexing of UEs with different services have been specifed in Release-15 while UL intra-UE prioritization and inter-UE multiplexing were specifed in Release-16. From an end-to-end system perspective, Release-16 added redundant transmission and QoS monitoring for the URLLC service.

Chapter 8 focuses on the features for V2X service. It frst gives a detailed description of the LTE V2X feature and then introduces the NR V2X sidelink feature by comparing and contrasting it to the V2X feature in LTE. The discussion will give an in-depth understanding of the physical structure, synchronization, physical layer and power control procedures, resource allocation and QoS and congestion control mechanism, layer 2/3 enhancements, as well as the architecture of the V2X feature. The operation was designed to work in both FR1 and FR2, but no effort was spent in Release-16 to optimize the FR2 operation. This optimization is widely expected to be the topic of future Releases of 5G.

Chapter 9 focuses on the features for the support of the OT (operational technology) industry. It starts with a discussion on how 5G enables the promise of Industrie 4.0 and the ecosystem that is forming to facilitate 3GPP to support the OT industry. As countries recognize the importance of OT for their economy, spectrum bands are being allocated by the respective regulatory bodies for nonpublic networks to support the smart factories of the future. The bulk of the chapter will detail the enhancement in Release-16 for IIoT including the support of TSN by the 5GC.

Chapter 10 is now the chapter that describes the ITU submission and how NR and LTE meet the 5G requirements in signifcant detail, from the rapporteur responsible for leading the preparation and evaluation. A new section has been added to the chapter on the ITU evaluations since the last edition.

The last chapter was highly reworked. It starts, as in the last edition, with a discussion on the 5G market and argues that vertical industries are critical to the robust growth of the wireless industry. Instead of detailing the 5G trials next, it now contains a detailed discussion on global 5G deployments. It next discusses the trials for different verticals that are ongoing at different parts of the world. The chapter then ends with a look into the future and gives a description of what is anticipated to be in 5G Release-17 that is expected to be fnalized in June of 2022.

Acknowledgement

The authors would like to thank our colleagues from all over the world that participated in the different standardization and industrial fora for 5G development. It is through our collective hard labor in harmonization from which 5G was born. In addition, the efforts made by operators from all the three regions are appreciated very much, by identifying focused use cases, key features, and solutions, as well as prioritizing architecture option, spectrum bands, and terminal confgurations, especially for the early deployment. Without this support, 3GPP can hardly complete the 5G standardization within such a short time. The frst release of 5G standardization was done within 3 years, including study item and work item, which breaks the record in the history of 3GPP. Such an astonishing speed is a testimony to the joint cooperation and collaboration from all the members in the ecosystem, including operators, network vendors, devices, and chipset vendors. All from the industry are expecting 5G, and all have contributed to 5G standardization.

We also thank the editorial and publication staff at Springer Nature for their support of this manuscript; chief among them are Susan Lagerstrom-Fife, our editor for the frst edition, as well as Mary James, our editor, and Shabib Shaikh, our project coordinator for the second edition.

Most importantly, we thank the support of our spouse/signifcant others and family members for putting up with the many days that we were away from home at the various meetings of the standard bodies and industry fora. But for 2020, we thank them for taking care of us while we were at home for almost the entire year with our virtual standards meeting that runs at all hours of the day and night.

xviii

Contents

xxii Contents Contents

xxiii

Contributors

Mazin Al-Shalash Futurewei Technologies, Plano, TX, USA Brian Classon Futurewei Technologies, Rolling Meadows, IL, USA Shao Jiafeng Huawei Technologies, Beijing, China Liu Jianghua Huawei Technologies, Beijing, China John Kaippallimalil Futurewei Technologies, Plano, TX, USA Wan Lei Huawei Technologies, Beijing, China David Mazzarese Huawei Technologies, Beijing, China Tony Saboorian Futurewei Technologies, Plano, TX, USA Philippe Satori Futurewei Technologies, Rolling Meadows, IL, USA Thorsten Schier Huawei Technologies, Lund, Sweden Anthony C. K. Soong Futurewei Technologies, Plano, TX, USA Marcus Wong Futurewei Technologies, Bridgewater, NJ, USA Amanda Xiang Futurewei Technologies, Plano, TX, USA Weimin Xiao Futurewei Technologies, Rolling Meadows, IL, USA Cheng Yan Huawei Technologies, Beijing, China Zhao Yang Huawei Technologies, Shanghai, China Wu Yong Huawei Technologies, Beijing, China

Abbreviations

xxxii

xxxiv

Chapter 1

From 4G To 5G: Use Cases And Requirements

This chapter investigates the motivations and driving forces of 5G development as well as introduces the 5G use cases and technical requirements. 5G is the frst generation that devotes itself to connecting both humans and machines. Accordingly, the service requirements and the technical performance requirements are extended from mobile broadband (MBB) to the new use cases. The diverse requirements pose signifcant challenges to system design.

This chapter also presents how 5G development is made based on industry collaboration, where ITU-R and 3GPP play the central role in this process. It introduces and reviews the ITU-R procedure on IMT-2020 development and 3GPP 5G standardization process. With the guidance of ITU-R and the well-harmonized technical development in 3GPP, 5G technology is well developed; which is one of the major keys for 5G success.

1.1 Introduction

Mobile cellular network has been developing since the 1970s.1 The frst-generation (1G) mobile network was based on frequency division multiple access (FDMA) (Fig. 1.1). It provided analog voice service to mobile users. After approximately 10 years, time division multiple access (TDMA) was developed in the second-generation (2G) network which enabled the digital voice service and low data rate service. In mid-1990 to 2000s, coding division multiple access (CDMA) was employed to develop the third-generation (3G) mobile network. The CDMA access enabled more effcient multiple user access through the specifed bandwidth. By this means, the data rate can reach several kilobits per second to several megabits per second, which enables fast data transmission for multimedia.

By the mid-2000s, ubiquitous data began to transform the meaning of mobile telephony service. The ever-increasing demand on data service stressed the capability of the 3G mobile network. Users are expecting to connect to the network anywhere, on the go or at home. More frequent data transmission happens, and faster data rates are becoming essential. Users start demanding broadband experience from their wireless mobile network service.

In 2005, the industry started the development of the fourth-generation (4G) mobile network that aims to provide ubiquitous mobile broadband (MBB) service. 3GPP developed the Long Term Evolution (LTE) that employs Orthogonal Frequency Division Multiple Access (OFDMA) to offer a good compromise on multiuser data rates and complexity. LTE development received a wide range of industry support. The OFDMA concept is well incorporated with multiple-input multiple-output (MIMO) technology, and the complexity is signifcantly reduced.

International Telecommunication Union (ITU), especially its Radiocommunication Sector (ITU-R), has played an important role in mobile network development since 3G. Due to the great success of 1G and 2G mobile networks, the industry and research interests were increased exponentially for 3G development. Therefore, many stakeholders were involved in 3G development with respect to previousgeneration mobile networks. A global standardization was becoming necessary due to the involvement of a variety of network vendors, user terminal manufacturers, and chipset providers. Global spectrum harmonization also becomes a critical issue for the successful development and deployment of the mobile network. In order to harmonize the spectrum use in different regions for appropriate technologies for 3G mobile network, ITU-R established procedures to address the allocation of spectrum for 3G mobile network, and to identify the appropriate radio interface technology that could be deployed on those spectrums globally. In this context, International Mobile Technology-2000 (IMT-2000) was specifed by the ITU, where a family of technologies were identifed as radio interface technologies for 3G mobile network (IMT-2000 system). Such procedures provide fair opportunity for the proponents that are interested in mobile network development, as well as set the necessary performance requirements to guarantee that candidate technology can effectively meet the requirements. The procedure is further developed and applied to 4G and 5G

development in ITU-R. This has resulted in the IMT family specifcation: IMT-2000 for "3G," IMT-Advanced for "4G," and IMT-2020 for "5G." While ITU-R plays the central role in defning appropriate technology for each generation of mobile network, the technology development is conducted in standard development organizations (SDOs). In 1998, the third-generation partnership project (3GPP) was initiated by the key players of the mobile network development, and is currently supported by seven regional SDOs from Europe, China, Japan, Korea, America, and India. It laid the foundation of global development for mobile technologies, and attracted the participation of a variety of industry and academy players. 3GPP has grown to be the essential standard organization for technology development for mobile networks since 3G.

1.2 Global 5G Development

In July 2012, ITU-R started the development of the vision for IMT for 2020 and beyond, which is later known as "IMT-2020." Following the ITU-R activity, in 2013–2015, several regional promotion groups and research forums were established in China, Europe, Korea, Japan, and America for 5G development. The regional studies provided extensive investigations of 5G use cases and capability requirement, which formed a global foundation for 5G vision developments in ITUR. In 2015, the 5G vision was set up in ITU-R based on the regional convergence.

Along with the gradual maturity of the 5G vision, in late 2014, 5G technology studies received increasing attention from industry and academy: with many new technologies and concepts proposed. In 2015, when ITU-R created the 5G vision, 3GPP as one of the most widely supported global standardization organizations started the technical requirement study and deployment scenario investigation, targeting to fulflling the 5G vision. In 2016, 3GPP initiated the 5G new radio (NR) technology study. Industry members, institutions, and universities are actively engaged in 3GPP study, which forms the solid foundation for 5G radio interface technology that encompasses a wide range of usage scenarios and features.

In December 2017, 3GPP accomplished the frst milestone of 5G specifcation.

An initial characteristic description was submitted to ITU-R in February 2018. The ongoing global 5G development opens the gate to the fully connected world. The global harmonization and coordination are the essential keys throughout the 5G development.

1.2.1 Itu-R Development On 5G/Imt-2020

For different radio services to coexist, it must rely on the allocated spectrum resources to deliver their capability. ITU-R is responsible for coordinating the spectrum allocation and spectrum usage for many different radio services, including satellite, broadcast, scientifc research, and mobile service. The radio spectrum resources are rather limited and, thus, effcient use of spectrum by each of the radio services should be guaranteed. This is also part of the responsibility of the ITU-R. It conducts technology studies to continuously improve the effciency of spectrum usage, and compatibility studies are conducted among the radio services to make sure that different radio services with appropriate spectrum allocation can coexist in an effcient manner. To this end, a range of study groups (SGs) and, under a specifc SG, a number of working parties (WPs) are established in ITU-R. The technical and spectrum experts with rich expertise on related felds are gathered under WPs and SGs for the study of applying appropriate technologies for the specifc radio services, and to investigate the spectrum allocation for specifc radio service purposes.

WP 5D is the expert group under SG 5 for the study of International Mobile Technology (IMT) services in ITU-R. To guarantee that the IMT technologies can utilize the spectrum effciently, and can well coexist with other radio services, the expert group devotes itself to develop a globally implemented radio interface standard that has high spectral effciency and other key capabilities, which are widely supported by a variety of network vendors, user terminal manufacturers, and chipset providers.

The history of the development of IMT-2000 (3G), IMT-Advanced (4G), and IMT-2020 (5G) is depicted in Fig. 1.2 (see also [1]).

As seen from Fig. 1.2, ITU-R usually spent around 10 years for the development of each generation of the mobile network (or IMT network under ITU-R context). From IMT-Advanced, a vision development was always made before the technical development. The vision study usually covers the investigation of high-level requirements, use cases, and key capabilities desired by the next-generation mobile

network. This is becoming an important step to converge the regional desires from different parts of the globe. It is observed that one ITU-R study period (3–4 years) is usually spent for this purpose. After that, another 1–2 (or even more) study periods are used to develop the technical aspects and to conduct the compatibility study if the new IMT spectrum that is requested is already used by other radio services.

The time plan might be depending on the technical complexity of technology to achieve the vision, and the new spectrum it may utilize to reach the demanded capability.

For 5G development, the vision study was from 2012 to 2015. During this time period, regional 5G promotion groups and research forums were established to gather and converge the regional interest of 5G vision, and the regional views are contributed to ITU-R through the representative administration members and sector members.

From 2015, ITU-R started the technical development of 5G which aimed to defne an IMT-2020 (5G) global specifcation by November 2020. The technical development phase usually contains the defnition of minimum technical requirements and evaluation guidelines, as well as an IMT-2020 submission and evaluation procedure. The minimum technical requirements guarantee that the candidate proposal can achieve the vision and can utilize the spectrum (including the potential new IMT spectrum) effectively. The submission and evaluation procedure defnes the criteria of acceptance for IMT-2020 proposal, the procedure of inviting external organizations to submit IMT-2020 proposal and inviting the independent evaluation groups to evaluate the proposal, as well as ITU-R's procedure of approving the proposal. Following the ITU-R procedure, external organizations like 3GPP initiated their technical development of 5G from 2015 which will be reviewed in Sect. 1.2.3. More detailed discussion on the 5G timeline and the submission and evaluation procedure can be found in Sect. 1.4. Concurrently ITU-R also started the compatibility study of potential new spectrum for 5G deployment. Such studies consist of three parts. Part 1 is for the new IMT spectrums that were allocated in world radio communication conference (WRC) 2015. For example, the frequency ranges of 3.3–3.8 GHz and 4.5–4.8 GHz (usually referred to as C-band) were identifed as IMT spectrum globally or regionally. Part 2 is for potential new IMT spectrums in WRC-19. For example, the frequency ranges of 24.25–27.5 GHz (usually referred to as millimeter wave), 37–43.5 GHz, and 66–71 GHz and others were identifed as IMT spectrum globally or regionally. ITU-R continued to study the remaining issues on compatibility study on these bands in WP 5D. Part 3 is for potential new IMT spectrums, such as 6425–7125 MHz, which will be discussed at WRC-23. ITU-R has already begun its work on evaluating new spectrum by requesting technical development on feasible technologies to guarantee the effcient use of these potential new bands, and the compatibility studies to guarantee that they can effciently coexist with other radio services.

The 5G spectrum will be discussed in more detail in Sect. 3.2.1.1.

1.2.2 Regional Development/Promotion On 5G

After ITU-R started 5G vision study in 2012, several regional promotion groups and research forums were founded in China, Europe, Korea, Japan, and America. These regional activities include the study of 5G requirements, use cases, and deployment scenarios, as well as exploring the key technologies and the nature of 5G spectrum.

These activities are briefy discussed in this section.

1.2.2.1 Ngmn

Next-Generation Mobile Networks (NGMN) is an alliance with worldwide leading operators and vendors that aims to expand the communications experience which will bring affordable mobile broadband services to the end user. It has a particular focus on 5G while accelerating the development of LTE-Advanced and its ecosystem.

NGMN published the "5G white paper" [2] in February 2015 which provided a list of requirements from the operator perspective to 5G networks. The requirements indicated a demand for capacity increase and the uniform user experience data rate across urban areas to rural areas. It also indicated that 5G networks should be capable of delivering diverse services including massive Internet of Things like sensor networks, extreme real-time communications like tactile Internet, and ultrareliable communications like e-Health services, to name a few. The services should be delivered in diverse scenarios, including high-speed trains, moving hotspots, aircrafts, etc.. Such diversity fts in ITU's envisioned 5G use cases as described in Sect. 1.3.

1.2.2.2 Imt-2020 (5G) Promotion Group

In February 2013, the IMT-2020 (5G) promotion group was established in China by three ministries: the Ministry of Industry and Information Technology (MIIT), the National Development and Reform Commission, and the Ministry of Science and Technology. It is the major platform to promote the research and development of 5G in China. The promotion group consists of leading Chinese operators, network equipment vendors, research institutions, and universities.

The IMT-2020 (5G) promotion group published its 5G vision white paper [3] in May 2014. The white paper indicated two important 5G usage categories: mobile broadband and Internet of Things. The mobile broadband use cases will continue to be addressed by 5G network that provides 1 Gbps user-experienced data rate in many deployment scenarios. The Internet of Things use cases on the other hand would be the other major driver to 5G network deployment, which requires a very capable 5G network to provide massive connections, very low latency, and high reliability. This forms a frst overview of 5G use cases that were later developed into three 5G usage scenarios identifed by ITU-R.

1.2.2.3 Europe: 5G Ia

The 5G Infrastructure Public Private Partnership (5G PPP) is a joint initiative between the European Commission and European ICT industry for 5G study in Europe. It is a huge collaborative research program with a lifetime of 2014 to 2020 and is addressing all major building blocks of 5G. Then, its name is replaced by 5G Infrastructure Association (5GIA).

Before the frst phase of 5G PPP initiative, an important 5G project was founded in November 2012, called Mobile and wireless communications Enablers for Twentytwenty (2020) Information Society (METIS). In April 2013, METIS published their study on 5G use cases, requirements, and scenarios (see [4]). In this study, METIS mentioned a number of new industrial and machine-type communications for 5G besides mobile broadband applications. In April 2015, METIS summarized these use cases into three categories in [5], in accordance with ITU-R development on 5G use cases. The three use cases are extreme mobile broadband (xMBB), massive machinetype communications (mMTC), and ultrareliable machine-type communications (uMTC), which are converged to ITU-R vision on 5G requirement and use cases as will be discussed in Sect. 1.3.1. The second-phase projects started around June/July 2017; many of these organizations are actively participating in international standardization and the ITU-R process toward IMT-2020.

1.2.2.4 Korea: 5G Forum

The 5G Forum was founded by the Ministry of Science, ICT and Future Planning and mobile industries in Korea in May 2013. The members of 5G Forum consisted of mobile telecommunication operators, manufacturers, and academic professionals. The goal of the 5G Forum is to assist in the development of the standard and contribute to its globalization.

Five core 5G services were foreseen by the 5G forum, including social networking services; mobile 3D imaging; artifcial intelligence; high-speed services; ultraand high-defnition resolution capabilities; and holographic technologies. Such new services will be enabled by the 5G network with powerful capabilities that provide ultrahigh capacity and data rates.

TTA 5G Technology Evaluation Special Project Group (TTA SPG33) is under 5G Special Technical Committee (5G STC) of TTA and responsible for evaluating the proposed candidate IMT-2020 RIT(s)/SRIT(s).

1.2.2.5 Japan: 5Gmf

The Fifth Generation Mobile Communications Promotion Forum (5GMF) was founded in September 2014 in Japan. 5GMF conducts research and development related to 5G including the standardization, coordination with related organizations, and other promotion activities.

5GMF published the white paper "5G Mobile Communications Systems for 2020 and beyond" [6] in July 2016, which highlighted the 5G use cases of high data rate services, self-driving, location-based services, etc. It foresaw that 5G network needs to be extremely fexible to reach the requirements of these diverse requirements.

1.2.2.6 North And South America: 5G Americas

5G Americas is an industry trade organization composed of leading telecommunications service providers and manufacturers. It was continued in 2015 from the previously known entity: 4G Americas. The organization aims to advocate for and foster the advancement and full capabilities of LTE wireless technology and its evolution beyond 5G, throughout the ecosystem's networks, services, applications, and wirelessly connected devices in the Americas. 5G Americas is invested in developing a connected wireless community while leading 5G development for the Americas.

5G Americas published its white paper on 5G Services & Use Cases [7] in November 2017. It provided an insightful report on 5G technology addressing new trends in a broad range of use cases and business models, with technical requirements and mappings to 5G capabilities. This is a continued study on 5G use cases for future uses.

1.2.2.7 Global 5G Event

The 5G regional developments call for a global coordination to form a unifed 5G standard that is applicable worldwide. The regional 5G developers, including IMT-2020 (5G) PG, 5G IA, 5G forum, 5GMF, 5G Americas, and 5G Brazil, have answered this call by setting up a global 5G event to share the views and development status in each region. These events will be instrumental in building global consensus on 5G with the world's 5G promotion organizations. This series of events have been putting its efforts in promoting the usage of 5G for different vertical industries and 5G ecosystems and inviting key industry players, administrations, and regulators to participate in the discussion.

The frst global 5G event was hosted by IMT-2020 PG in Beijing, May 2016, and the event is held twice a year in rotation. The most recent event was held July 2-3, 2020 at the Shanghai International Convention Center, Shanghai China.

1.2.3 Standard Development

Along with ITU-R and regional activities toward 5G development, standardization organizations have also focused their attention on 5G since 2014. The global partnership project, 3GPP, has become the key standard organization of 5G development. It has become a global initiative for mobile cellular standard since the development of the 3G network. The current 4G, LTE mobile broadband standard is one of the most successful mobile standards and it is used worldwide.

3GPP is an organization that unites telecommunications standard development organizations (SDOs) all over the world. These SDOs are known as organization partners (OPs) in 3GPP, and currently there are seven OPs: ARIB and TTC from Japan, ATIS from America, CCSA from China, ETSI from Europe, TSDSI from India, and TTA from Korea. The seven OPs provide the members with stable environment to develop 3GPP technology. The OP members include key industry players, leading operators, vendors, user terminal manufacturers, and chipset developers.

Research institutes with regional impacts, academic organizations, and universities are also included. It covers almost all the key parties as far as a mobile cellular standard is concerned. The members of the OPs are heavily involved in the technical standard development, which ensures that the 3GPP technology addresses the concerns and issues from different parties and different regions. Based on the consensus of the members from different OPs, the technical specifcations defned by 3GPP are transposed by the OPs into their regional specifcations. By this means, global mobile standards are developed. It can be argued that this is the essential key for the success of 3GPP standard development.

LTE is an early example under this consensus-based spirit of development. The global and wide-range participation laid the foundation for the success of LTE development, standardization, and implementation. Due to the great success of LTE, 3GPP has become the essential standard development body for 5G. In late 2014, 3GPP initiated 5G studies and development along with the gradual maturing of the 5G vision.

During late 2015 to early 2017, when 3GPP was in its 14th release time frame (known as Release-14), 5G studies on technical requirements and deployment scenarios were conducted. These studies aimed to achieve the 5G vision as defned by ITU-R in June 2015. The study of a new radio (NR) interface was initiated following the requirement study. The key technical components were identifed for the NR development, which form the basis of the specifcation work in the next release (i.e., Release-15) that lasts from early 2017 to June 2018. The full-capability 3GPP 5G technology, including NR and LTE, was developed in Release-16 during the time frame from early 2018 to September of 2020. In particular, the September 2020 TSG-RAN meeting agreed that a UE or network vendor wishing to support Release-16 must use version 16.2.0 or later of TS 38.331/TS 36.331 [8], because it agreed to some non-backward compatible changes. With this phased approach, 3GPP will bring its 5G solution to ITU-R as IMT-2020 in the year 2020. The ITU approved the frst version of IMT-2020 in February of 2021.

1.3 Use Case Extensions And Requirements

Cellular communication system, from its frst generation, is focused on connecting humans. The 1G and 2G communication system provided ubiquitous voice service between persons, which enables us to talk freely to our friends when we are at home, in offce, or on the move. From 3G to 4G, multimedia and other mobile broadband applications are supported, and we are able to do much more, such as browsing the Web, sharing nice pictures with our friends, chatting on short videos, etc., while on the move. For 5G, however, the study of the use cases reveals some new demands that are beyond the mobile broadband (MBB) which primarily aims to connect human beings and machines. This section provides a review of 5G use cases, requirements, and key capabilities.

1.3.1 5G Usage Cases And Service Requirement

5G communication is envisioned to enable a full connected world for 2020 and beyond. This full connectivity is not only targeted at communication of the people, but also enables the communication of machines and things that can bring added value to improving the operational effciency of the society, and facilitate our everyday life. Under this vision, 5G use cases are extended from mobile broadband (MBB) to Internet of Things (IoT).

1.3.1.1 Extended Usage Scenarios: From Embb To Iot (Mmtc And Urllc)

ITU-R established the 5G vision in 2015 through its Recommendation ITU-R M.2083 [1], indicating that 5G will extend its usage scenario from enhanced mobile broadband (eMBB) to massive machine-type communication (mMTC) and ultrareliable and low-latency communication (URLLC). The mMTC and URLLC services are a subset of Internet of Things (IoT) services. They are considered as the frst step for 5G network stepping into the broad range of IoT services that are characterized by the key service requirements. Consequently, 5G will be the frst-generation that targets to extend wireless connection to other than human-to-human connections.

The extension of 5G from eMBB to mMTC and URLLC comes from the observation and the demand of the user and application trend. On the one hand, high data rate video streams both down to the people (e.g., video streaming downloading) and up to the cloud server (e.g., users share their produced videos to their friends) are expected, and instantaneous and low-latency connectivity becomes very vital to the user experiences for including augmented reality (AR) and virtual reality (VR). The user density with such high demand also increases, especially in urban areas, while in rural and/or high-mobility cases, satisfactory end-user experiences are also desired by the users. Therefore, the challenging per-user high data rate request combined with high user density and user mobility becomes a driven force for 5G development, which requests signifcantly enhanced capability for mobile broadband services.

On the other hand, in the future, any object that can beneft from being connected will be connected, though, partially or dominantly, wireless technologies. This trend poses a huge amount of connectivity demand for connecting objects/machines/ things in a wide range of applications. For example, driverless cars, enhanced mobile cloud services, real-time traffc control optimization, emergency and disaster response, smart grid, e-health, or effcient industrial communications are what are expected to be enabled or improved by wireless technologies/connections to name just a few [1]. By a closer observation of these applications, one can fnd two major characteristics of these services: one is the number of desired connections, and the other is the requested reliability within a given latency budget. These two signifcant characteristics provide the nature of mMTC and URLLC.

Therefore, 5G puts itself on the target to support diverse usage scenarios and applications including eMBB, mMTC, and URLLC. The following is a frst glance on what the three usage scenarios indicate, which comes from [1]: - Enhanced mobile broadband: Mobile broadband addresses the human-centric use cases for access to multimedia content, services, and data. The demand for mobile broadband will continue to increase, leading to enhanced mobile broadband. The enhanced mobile broadband usage scenario will come with new application areas and requirements in addition to existing mobile broadband applications for improved performance and an increasingly seamless user experience. This usage scenario covers a range of cases, including wide area coverage and hotspot, which have different requirements. For the hotspot case, i.e., for an area with high user density, very high traffc capacity is needed, while the requirement for mobility is low and user data rate is higher than that of wide area coverage. For the wide area coverage case, seamless coverage and medium to high mobility are desired, with much improved user data rate compared to existing data rates. However, the data rate requirement may be relaxed compared to hotspot.

• Ultrareliable and low latency communications: This use case has stringent requirements for capabilities such as throughput, latency, and availability. Some examples include wireless control of industrial manufacturing or production processes, remote medical surgery, distribution automation in a smart grid, transportation safety, etc.
• Massive machine-type communications: This use case is characterized by a very large number of connected devices typically transmitting a relatively low volume of non-delay-sensitive data. Devices are required to be of low cost, and should have a very long battery life. Additional use cases are expected to emerge, which are currently not foreseen.

For future IMT, fexibility will be necessary to adapt to new use cases that come with a wide range of requirements.

In the following, we will frst give a survey on the diverse services that are envisioned as 5G use cases, and then investigate the service requirements. Then technical performance requirements are extracted by grouping the similar service requirements to one characterized technical requirement. The importance of the technical requirement will be mapped to different usage scenarios based on the group of services under the specifc usage scenario.

1.3.1.2 Survey Of Diverse Services Across 5G Usage Scenarios And The Diverse Requirements

There are a broad range of emerging services that are expected to appear for the year 2020 and beyond that are under the scope of the 5G study. Generally, these 5G services are categorized into three groups according to the three usage scenarios elucidated in the last section.

Embb Services

Mobile broadband services like web browsing, social apps with text messaging, fle sharing, and music downloading are already very popular and well supported by 4G communication systems. In the future, it is expected that higher data rate services like ultrahigh-defnition (UHD) video, 3D video, and augmented reality and virtual reality will be dominating the human-to-human communication requirements. In addition to the abovementioned downlink high data rate services, the demand of uplink high data rate service is also emerging, e.g., the HD video sharing from the users. These service requirements defne new horizon limits for eMBB development, together with the anywhere anytime experience requirements.

Uhd/3D Video Streaming

The 4 K/8 K UHD video streaming requires up to 300 Mbit/s experienced data rate. From the study of technology enablers of enhanced mobile broadband conducted in 3GPP [9], the requested data rates of 4 K and 8 K UHD video are listed in Table 1.1.

Video Sharing

With the increased popularity of social applications, video sharing from the users to the cloud is becoming popular. The full HD (1080p) video can be assumed, and 4 K/8 K UHD video sharing can be expected in the 2020 and beyond. The requested data rates of 4 K and 8 K UHD video are the same as those listed in Table 1.1. The required data rates of 1080p and 720p video are given in Table 1.2 [10].

Ar/Vr Delivery To The User

Augmented reality (AR) and virtual reality are the applications that bring fresh new experience to the eMBB users. Augmented reality provides the users with interactive experience of a real-world environment with "augmented" views generated by computer graphics. Virtual reality (VR) provides another interactive experience with an immersive environment created by computer graphics which can be fantastical or dramatic compared to physical reality. Both AR and VR require very high data rate and low latency to deliver the computer-generated graphics and multimedia contents to the end users with guaranteed and smooth experience.

The required data rate and round-trip delay of VR application are listed in Table 1.3 (see [11]).

In general, it is observed from the above examples that future eMBB services will require very high data rate compared with the requirement today. In addition, for AR/VR, the latency requirement is becoming more important. Therefore, the high data rate and low latency would become dominant for 5G eMBB services.

Mmtc Services

Massive machine-type communication (mMTC) refers to a group of emerging services that typically use massive number of sensors to report the sensing data to the cloud or a central data center in order to make smart decisions and/or reduce the human workload for collecting these data.

In the study of technology enablers of massive Internet of Things conducted in 3GPP (see [12]), various services are investigated. Here we just mention some of them.

For example, the electric company deploys a large number of smart meters for each apartment within an apartment complex, and the smart meters report electricity usage to the company periodically.

Another example is the number of video recorders that are installed along or at the corner of the streets. The camera records continuous video, and stores the content for some period of time. The device periodically sends a status update to the traffc police indicating how the traffc is moving. When an accident occurs at the intersection, the device begins sending high-quality video to the traffc police of the accident and ensuing traffc congestion.

Farm machinery is increasingly being automated. Farm machinery can report various sensor data, such as soil condition and crop growth, so that the farmer can remotely monitor the farm condition and control machinery.

There are other examples that have very similar paradigm to the abovementioned applications. The common service requirements can be summarized as follows:

• These services usually require large amount of sensors within a specifc area. If we further consider the network needs to provide connection to multiple types of sensor or sensor applications, the amount will soon become incredibly large. In this sense, it is not diffcult to imagine that there might be 1 sensor per square meter for the year 2020 and beyond. For example, if an apartment is of the size of 80 m2, and in one apartment building we have ten foors, then if one apartment has eight sensors (for example, various smart meters for electricity use, water use, indoor air quality monitoring, temperature monitoring, etc.), we will have 80 sensors in the apartment building. It indicates 1 sensor per square meter.
• These services have a range of data rate request; however, the recent applications might be dominant by small data packet sizes. For example, the smart metering reports usually small amount of data.2 The farm machinery also reports small amount of data at a time. In the video recorder application, although sometimes video needs to be transmitted when the accident occurs, the small data is dominant in daily case.
• Device battery life is critical to the economic success of these applications.

Otherwise frequent reinstallation of the sensors will be required. Considering the huge amount of sensor deployment, such reinstallation can be very costly. In [13], it is required that the device battery life should be longer than 10 years.

• Coverage is very important in this use case. The connection service needs to be provided to the sensors deployed in deep coverage environment, e.g., in basement. Therefore, it is important to guarantee that the network can reach such sensors. In summary, the service requirements for mMTC applications are primarily on connection density (the number of devices per unity area) under a specifc quality of service (e.g., the data rate), the battery life, and the coverage capability of the sensors. There might be different specifc values for the above requirements. Nevertheless, high connection density, long battery life, and deep coverage capability are foreseen as the major service requirements for mMTC.

Urllc Services

Ultrareliable and low-latency communication (URLLC) is the group of emerging services that are very sensitive to latency and loss of data packets. In [1], some examples are given, which include wireless control of industrial manufacturing or production processes, remote medical surgery, distribution automation in a smart grid, transportation safety, etc. In the study of technology enablers of massive Internet of Things conducted in 3GPP (see [14]), similar examples are mentioned, with more detailed descriptions on the service applications and envisaged application scenarios.

In [14], the service requirements for this type of applications are summarized. As indicated by its name "URLLC," the primary requirements are low latency and ultrareliability. They need to be achieved at the same time. That is, for a transmitted data packet, high reliability (very low loss of data) should be guaranteed within a given time duration.

1.3.1.3 Supporting Requirements And Operational Requirements To Enable 5G Service Deployment

The service requirement is from the perspective of an individual user or device, and it is independent to the deployment environment. However, from system design perspective, how the network can guarantee, as much as possible, a service that is supported is needed, especially in a cost-effective manner. Therefore, besides the service requirement that is raised from a service perspective, it is also desired to defne the related supporting and operational requirements to guarantee certain level of service support from system design and operating perspective.

Embb

For eMBB services, it is needed to consider to what extent the high data rate can be supported. From a system perspective, there are multiple users in a given environment. If one assumes that data transmission of a specifc service type simultaneously happens on these users, it is straightforward to consider the area traffc volume requirement, as well as to defne a "cell edge" user-experienced data rate requirement such that most of the users can achieve the required data rate for this service.

To this end, area traffc capacity and edge user-experienced data rate requirement are defned.

Edge User-Experienced Data Rate

Edge user (e.g., ffth percentile)-experienced data rate is defned to guarantee that most of the users (e.g., at least 95% of the users) can exceed a specifc user data rate. In NGMN white paper [2], the edge user-experienced data rate requirement is given in Table 1.4. By recalling the different service requirements in eMBB, it is seen that with such requirements, UHD video is expected to be well supported in dense urban environment, and advanced or "ultimate" AR/VR are expected to be well supported in indoor. For crowded environment as well as in rural and high-speed vehicle environment, 1080p and 720p video is expected to be well supported.

It is also noted that mobility is defned to support the specifc user-experienced data rate requirement. This is from the motivation that specifc services need to be well supported also on the go.

Area Traffc Capacity

Area traffc capacity is the amount of the total traffc in the given area. The total amount of traffc is related to the number of simultaneous users with specifc service requirement. It provides the capability requirement on the number of supported users with a guaranteed average user data rate.

For example, for indoor environment, if we assume that 20 users per 1000 m2 are simultaneously transmitting data with service required data rate of 500 Mbps, then the area traffc capacity requirement would be 20 × 500 Mbps/1000 m2 = 10 Mbps/m2.

Another perspective is from the prediction of the amount of increased traffc. It is predicted that for the year of 2020 and beyond, the total traffc amount will be increased by 100 times to 1000 times compared to that in 2010s. In this case, by investigating the traffc amount in the 2010s, and by applying the increasing time prediction, one can derive the area traffc capacity requirement for 5G time frame. Besides the requirements that support successful deployment of eMBB services from system perspective, the operational requirements that aim to improve the resource utilization effciency so as to reduce the cost are also defned.

Spectral Effciency

Unlike wireline communication, radio spectrum is a scarce resource. Consequently, spectral effciency is very critical for eMBB services since the 3G era.

For 5G, it is anticipated that the requested data rate is signifcantly increased. The area traffc capacity is predicted to be increased by at least 100 times. This means in a given area, the total amount of data rate will be increased by at least 100 times. The total amount of data rate is given by averaged spectrum effciency (bps/Hz per base station) multiplying with the available system bandwidth (Hz) and the number of base stations in the area. If we assume that the base station deployment within the given area can be increased by three times, and the available bandwidth can be increased by ten times, then the average spectrum effciency should be increased by at least three times.

Besides the average spectral effciency, there is a strong desire in the industry to improve the edge user spectral effciency. Here we note that it is a desideratum and leave the quantitative requirement to be discussed in Chap. 10.

Energy Effciency

Energy effciency is motivated by recognizing that it is not affordable to provide 100 times network area traffc capacity with 100 times energy consumption increase. Indeed, the network energy consumption increase should be very limited, e.g., see analysis in [15].

Specifcally, it is reported that currently the radio access network infrastructure consume about 0.5% of the global energy consumption [16], which is already a considerable portion. To avoid the 5G network from becoming a major consumer of the global energy production, it is reasonable to constrain the increase of energy consumption of 5G radio access network, such that its energy consumption ratio is below or at least at the same level compared to today. That is to say, the energy consumption increase of radio access needs to follow the pace of global energy consumption increase, or even slower.

Based on the data from [17] reproduced in Fig. 1.3, one can see that between 1990 and 2013, the CAGR of global energy consumption is about 1.91%, which indicates 1.2 times increase in one decade. Thus, one can forecast that in 2020 the global energy consumption would increase no more than 20% compared to 2010. Therefore, the power consumption of 5G network in 2020 should be no more than 1.2 times as compared to 4G network in 2010.

This implies that if the area traffc capacity increases by 100 times, with the assumption of very limited or no energy consumption increase, the network energy effciency measured as bit per Joule (or bit/s per Watt) needs to be improved by approximately 100 times.

Energy effciency is also important for mMTC and URLLC as operating requirements. However, eMBB may come as the frst urgency in terms of this requirement.

Mmtc

For mMTC, area traffc capacity can also be used as one of the requirements for system design. However, if the current focus is on small data packet transmission, the requirement would be very loose. However, for the services with larger devicewise data rate requirement (e.g., video recorder applications), the area traffc capacity would become important in addition to the service requirement.

Urllc Availability

To guarantee the users in different regions within the area achieving the service requirement (low latency and ultrahigh reliability), the availability is defned in [2] as follows: the network is available for the targeted communication in 95% of the locations where the network is deployed and 100% of the time. Within the 95% of the locations the URLLC service requirement is achieved in 1 ms latency. This is a very stringent target for system design for URLLC applications.

General

In addition to the above requirements, coverage capability of a network is the key for network operational cost.

Coverage

Coverage is defned as the maximum geographical range of a radio access point to provide a certain quality of service (QoS). For different QoS level, the coverage of a network will be different. For example, the uplink coverage will be more challenging for a service requirement of data rate of 1 Mbps than that of 100 kbps. Coverage is also highly related to site density. If the coverage for a certain QoS is small, operators need to deploy much dense network to provide the desired QoS. For example, in eMBB, if the coverage for providing UL data rate of 10 Mbps is 50 m, then the inter-site distance (ISD) could be hardly larger than 100 m. In this case, the number of sites required would be huge, leading to unacceptable cost. In this sense, coverage capability is fundamental for eMBB, URLLC, and mMTC usage scenarios.

1.3.2 5G Key Capabilities And Technical Performance Requirements

The 5G services form a solid basis for the defnition of 5G key capability and the related technical requirements. In ITU-R, the key capabilities for 5G are defned together with 5G envisaged usage scenarios. Technical performance requirements are defned based on the 5G vision and key capabilities.

1.3.2.1 Key Capabilities For 5G

The key capabilities for 5G (also known as IMT-2020 in ITU-R context) are shown in Fig. 1.4 (see [1]). Figure 1.4(a) demonstrates the targets of 5G key capabilities as well as the enhancement compared to 4G (also known as IMT-Advanced).

Figure 1.4(b) gives the importance of each key capability in the three usage scenarios. In the following, the key capabilities for each usage scenario are discussed.

Embb

In ITU-R, the user-experienced data rate and area traffc capacity are identifed as

part of the key capabilities most relevant for eMBB. As discussed in previous section, the two capabilities guarantee the system support of a specifc eMBB service to be successfully delivered to most of the users, with a given number of users in the system.

User-Experienced Data Rate

User-experienced data rate is defned as the achievable data rate that is available ubiquitously across the coverage area to a mobile user/device (in Mbps). The term "ubiquitous" is related to the considered target coverage area and is not intended to relate to an entire region or country [1]. It can be seen that ten times user-experienced data rate is expected compared to 4G. It can also be seen that 100 Mbps edge userexperienced data rate is indicated for downlink in dense urban areas (Fig. 1.4).

Although the target capability is a bit lower than the NGMN requirement, it still guarantees that most users experience good UHD video and have partial AR/VR capabilities. Undoubtedly this capability is well beyond the service requirement of 1080p and 720p video.

Area Traffc Capacity

Area traffc capacity is defned [1] as the total traffc throughput served per geographic area (in Mbps/m2).

Area traffc capacity of 10 Mbps/m2 is expected in high user density scenarios, e.g., indoor. This means that if we assume 20 users per 1000 m2 simultaneously transmitting data, the service required data rate of 500 Mbps can be supported.

This capability demonstrates a 100 times improvement compared to 4G. This is the result of ten times user-experienced data rate improvement together with ten times connection density capability.

Mobility

Mobility is an important capability that allows the support of high data rate transmission inside high-speed vehicles.

It is defned as the maximum speed at which a defned QoS and seamless transfer between radio nodes which may belong to different layers and/or radio access technologies (multilayer/−RAT) can be achieved (in km/h) [1].

The capability of up to 500 km/h is to be supported. This mobility class is 1.4 times of 4G (where up to 350 km/h mobility class is defned).

Peak Data Rate

The peak data rate is defned as the maximum achievable data rate under ideal conditions per user/device (in Gbps) [1]. It indicates the maximum capability of the data rate of a device. It is seen that 10–20 Gbps is expected. This means the device under the ideal condition can support AR/VR applications. Besides the above capabilities, the effciency capabilities that are helpful to reduce the operational cost are identifed as key capabilities for 5G. Energy Effciency Network energy effciency is identifed as one of the key capabilities that enable the economics of 5G.

Energy effciency has two aspects [1]: - On the network side, energy effciency refers to the quantity of information bits transmitted to/received from users, per unit of energy consumption of the radio access network (RAN) (in bit/Joule)

• On the device side, energy effciency refers to the quantity of information bits per unit of energy consumption of the communication module (in bit/Joule) For network energy effciency, it is required that the energy consumption should not be greater than the radio access networks deployed today while delivering the enhanced capabilities. Therefore, if area traffc capacity is increased by 100 times, the network energy effciency should be improved by a similar factor.

Spectral Effciency

Spectral effciency includes two aspects. The average spectrum effciency is defned as the average data throughput per unit of spectrum resource and per transmitreceive point (bps/Hz/TRxP). The edge user spectral effciency is defned as the ffth percentile user data throughput per unit of spectrum resource (bps/Hz).

As discussed in Sect. 1.3.1.3, the requirement on average spectral effciency improvement is related to the required amount of increase in area traffc capacity. ITU-R defnes 100 times improvement capability for area traffc capacity. In this case, the total amount of data rate will be increased by at least 100 times. The total amount of data rate is given by averaged spectrum effciency (bps/Hz/TRxP) multiplying with the available system bandwidth (Hz) and the number of transmit-receive point (TRxP) in the area. If we assume that the TRxP deployment within the given area can be increased by three times, and the available bandwidth can be increased by ten times, then the average spectrum effciency should be increased by at least three times.

For edge user spectral effciency, the requirement of improvement is related to the supported edge user-experienced data rate, the number of users under one TRxP, and the available bandwidth. If we assume that the connection density within the area is increased by ten times (see below on "connection density" capability) while the number of TRxPs within the area can be increased by three times, then the number of users in one TRxP will be increased by three times. On the other hand, one may assume that the available bandwidth in one TRxP is increased by ten times.

This means for an individual user, the available bandwidth is increased by 10/3 = 3.3 times. In this case, to consider the support of ten times edge user-experienced data rate (see above on "user-experienced data rate" capability), the edge user spectral effciency should be improved by three times.

Mmtc

For mMTC usage scenario, connection density and energy effciency are identifed as the two most relevant key capabilities. Besides key capabilities, operational lifetime is also identifed as desired capabilities for 5G network for mMTC. These capabilities will now be discussed in ordinem.

Connection Density

Connection density is the total number of connected and/or accessible devices per unit area (per km2) [1]. For mMTC usage scenario, connection density is expected to reach 1,000,000 devices per km2 due to the demand of connecting a vast number of devices in the time frame of 2020 and beyond. This is a ten times improvement compared to 4G (IMT-Advanced).

Network Energy Effciency

Network energy effciency is also identifed as one of the important key capabilities for mMTC. This is because the large coverage provided for mMTC devices should not be at the cost of signifcantly increased energy consumption.

Operational Lifetime

Operational lifetime refers to operation time per stored energy capacity. This is particularly important for machine-type devices requiring a very long battery life (e.g., more than 10 years) whose regular maintenance is diffcult due to physical or economic reasons [1].

Urllc

As discussed in the previous sections, latency, mobility, and reliability are identifed as the two most relevant key capabilities of URLLC.

Latency

In [1], user plane latency is highlighted. It is defned as the contribution by the radio network to the time from when the source sends a packet to when the destination receives it (in ms).

The 1 ms latency is desired. Reliability is further defned in technical performance requirements and will be discussed in the following sections.

Mobility Mobility is relevant to URLLC usage scenario because the transportation safety applications, etc. are usually under high-speed mobility.

Besides the above key capabilities, reliability and resilience are also identifed as desired capabilities for 5G network for URLLC.

Reliability

Reliability relates to the capability to provide a given service with a very high level of availability [1].

Resilience

Resilience is the ability of the network to continue operating correctly during and after a natural or man-made disturbance, such as the loss of mains power [1]. Other Capabilities Other capabilities for 5G are also defned in [1]. Spectrum and Bandwidth Flexibility Spectrum and bandwidth fexibility refer to the fexibility of the system design to handle different scenarios, and in particular to the capability to operate at different frequency ranges, including higher frequencies and wider channel bandwidths than today.

Security And Privacy

Security and privacy refer to several areas such as encryption and integrity protection of user data and signaling, as well as end-user privacy preventing unauthorized user tracking, and protection of network against hacking, fraud, denial of service, man-in-the-middle attacks, etc.

The above capabilities indicate that 5G spectrum and bandwidth fexibility, as well as security and privacy, will be further enhanced.

1.3.2.2 Technical Performance Requirements For 5G

Based on the key capabilities and IMT-2020 vision defned in [1], technical performance requirements are defned in Report ITU-R M.2410 (see [18]).

The technical performance requirements are summarized in Tables 1.5, 1.6, and 1.7. The detailed defnition of the technical performance requirements can be found in [18].

To reach the 5G vision defned by ITU-R, 3GPP further studied the deployment scenarios and the related requirements associated with the three usage scenarios as documented in 3GPP TR 38.913 (see [13]). These requirements are usually higher than ITU's technical performance requirement, showing 3GPP's ambition of providing higher capability than ITU required. A detailed description of the 3GPP requirements is beyond the scope of this book. This interested reader is encouraged to consult [13].

1.3.3 Summary On 5G Requirements

From the analyses in the previous sections, it can be seen that 5G has diverse requirements. For eMBB usage scenario, the edge user-experienced data rate should

be enhanced to deliver high-quality video at anywhere and anytime to the end users,

Table 1.7 mMTC technical performance requirements

and high data rate transmission should also be possible under high mobility class. The area traffc capacity is required to be improved by 100 times or more to enable more users to enjoy high data rate service. Latency is also found to be of increasing importance for eMBB: for example, in AR/VR applications. The above performance requirements need to be provided in an affordable manner, which in turn requires spectral effciency to be improved at least by three times, and energy effciency to be signifcantly enhanced. Note that not all of these requirements are unique to the eMBB service; for example, energy effciency is also desired for network providing mMTC and URLLC services.

For the mMTC usage scenario, connection density of the order of 1,000,000 devices per km2 should be supported to enable mMTC services that are delivered to a huge number of sensors, with variable data rates. Long battery life and deep coverage are desired.

For URLLC usage scenario, low latency together with high reliability is required.

High availability should also be targeted from a long-term perspective which further ensures that most of the locations within the network coverage are able to achieve the URLLC service requirement.

Coverage is a fundamental requirement when providing eMBB, URLLC, and mMTC services, because otherwise the site deployment would become very dense. A balance will need to be struck between density and network cost to make the system design economically feasible.

Other requirements including privacy and security enhancement, which makes 5G a safer network, are also necessary. This is of vital importance especially given that 5G targets to connect everything.

1.4 Standard Organization And 5G Activities

5G development is a vast process that needs wide support from both industries, SDOs, and administrations. As is seen from Sect. 1.2, different regions show their visions for 5G development and applications. Consequently, for effcient 5G development, global efforts are needed to harmonize the regional concepts, from both technology perspective and spectrum perspective, so as to develop a unifed 5G technology that is applicable on 5G potential bands.

To achieve this goal, the following organizations play a vital role in 5G standardization and development. ITU-R, as the leading organization of identifying qualifed 5G technology (known as IMT-2020 in ITU-R context) for effcient spectrum use, develops strict procedures for IMT-2020 submission and evaluation. These procedures are open to all technical standard development organizations, and they guarantee that the proposed technology could fulfll a specifc set of requirements that are suitable for 5G deployment and to reach the vision of the envisaged network system for 2020 and beyond. On the other hand, 3GPP, as a very active technical partnership project, becomes the essential standard development organization for 5G development for the SDOs. This section discusses the ITU procedures for IMT-2020 submission and 3GPP developments toward ITU-R submission.

1.4.1 Itu-R Procedure/Process Of Imt-2020 Submission

ITU-R is the leading organization for IMT-2020 development. ITU-R's role is to identify the IMT-2020 (5G) vision which will serve as the overall target for 5G development for 2020 and beyond, and to invite the technical standard development organizations to submit their candidate technologies that are capable to achieve the IMT-2020 vision. The qualifed technology proposals are allowed to be deployed on IMT bands that are licensed to operators.

ITU-R Working Party 5D (WP 5D), which is the responsible working party in ITU-R for IMT system development, made an overall time plan for the above process in Fig. 1.5.

There are, generally, three stages to develop IMT-2020:

Stage 1: Imt-2020 Vision Development (2012–2015).

From July 2012 to June 2015, ITU-R WP 5D developed the IMT-2020 vision (Recommendation ITU-R M.2083) to defne the framework of IMT-2020. At the same time, the study on technology trends and IMT feasibility for above 6 GHz was conducted in preparation for 5G development.

Stage 2: IMT-2020 technical performance and evaluation criteria development (2015–2017).

After the IMT-2020 vision is developed, WP 5D started the defnition of IMT-2020 requirements in 2016. The requirements include three aspects: technical performance requirements, service requirements, and spectrum requirements. These requirements are used to evaluate whether a candidate proposal is a qualifed technical proposal that has the potential to be included as IMT-2020 standard for global deployment. At the same time, the evaluation criteria and methods are developed by ITU-R to evaluate whether a specifc requirement is achieved by the candidate proposal. Submission templates are also developed such that a unifed submission format can be used among the proponents, which is helpful to ease the work of ITU-R and independent evaluation groups when evaluating the received candidate technologies. In 2017, ITU-R published three reports related to the above aspects: - Report ITU-R M.2411: Requirements, evaluation criteria, and submission templates for the development of IMT-2020, which defned IMT-2020 requirements, evaluation criteria, and submission templates. Technical performance requirements are detailed in Report ITU-R M.2410. And evaluation criteria and method are detailed in Report ITU-R M.2412.

• Report ITU-R M.2410: Minimum requirements related to technical performance for IMT-2020 radio interface(s), which defnes the 13 technical performance requirements for IMT-2020 development.
• Report ITU-R M.2412: Guidelines for evaluation of radio interface technologies for IMT-2020, which defnes the test environment, evaluation method, and detailed evaluation parameters for the technical performance requirement. Evaluation criteria on how to conduct these test environment, method, and parameters, and how to determine that a requirement is fulflled, are defned in suffcient details.
• Besides the above three important reports, Document IMT-2020/02 (Rev. 1) [19] further defnes the criteria for accepting a proposal to enter the IMT-2020 evalu-

ation procedure and the criteria for accepting a proposal to be approved as IMT-2020. It is, currently, required that the proposed radio interface technology should meet the requirement of at least two eMBB test environments and one URLLC or one mMTC test environment when it is considered for acceptance to enter the IMT-2020 evaluation process (see below). Furthermore, it is required that the requirements should be met for all test environments in eMBB, URLLC, and mMTC when it is considered to be approved as IMT-2020. By this means, the developers have the chance to enter IMT-2020 development process at the beginning of IMT-2020 submission phase, and further develop itself to achieve the full vision of IMT-2020 when approved.

Stage 3: Imt-2020 Submission, Evaluation, And Specifcation Development (2016–2020).

In February 2016, the ITU-R issued the Circular Letter that invited the submission of IMT-2020 technologies among ITU members and external organizations. This Letter offcially announces that the evaluation and submission procedure is initialized. This procedure includes several steps that span from 2017 to 2020. As can be roughly seen in Fig. 1.5, there are generally three phases: - "Submission phase" (see "Proposals IMT-2020" in Fig. 1.5): In this phase, the proponents can propose IMT-2020 candidate technologies. Self-evaluation will need to be provided by the proponent to demonstrate that their proposed technology can pass the minimum set of requirements requested by ITU-R.

• "Evaluation phase": In this phase, independent evaluation groups are invited to evaluate the qualifed candidate technologies. Evaluation reports from the evaluation groups will be reviewed by WP 5D to determine which candidates pass the requested minimum set of requirements based on a predefned criteria. These technologies will go into specifcation phase.
• "Specifcation phase": In this phase, the qualifed technologies will be specifed and captured in a recommendation of ITU-R for global IMT-2020 standard. The technologies captured in IMT-2020 standard will be able to be deployed on IMT bands licensed to operators.
• Consensus building is brought across the procedure to achieve a unifed IMT-2020 technology.

The detailed procedure is given in Document IMT-2020/02 (Rev. 2) [19].

Based on the guidance of ITU-R procedure for IMT-2020 submission, the technical standard development organizations can submit their candidate technologies to ITU-R, and ITU-R will invite independent evaluation groups to make an independent evaluation to all the received candidates that declare themselves capable of meeting IMT-2020 requirements. 3GPP as one of the technical standard partnership projects has made such efforts for IMT-2020 submission.

1.4.2 3Gpp Development Toward Itu-R Submission

3GPP is a partnership project formed during the 3G development era that targets to develop a global harmonized technical specifcation for wireless cellular systems. It unites seven telecommunications standard development organizations from different regions worldwide. The seven organizations, also known as "Organizational Partners" (OPs) of 3GPP, as discussed in Sect. 1.2.3. provide their members with a stable environment to produce the Reports and Specifcations that defne 3GPP technologies.

It should be observed that the seven OPs have a good representation and worldwide coverage for the global development on wireless telecommunications. Through the seven OPs, the member companies, universities, and research institutes can join the technical discussions of 3GPP and contribute to the technology development. This is a very essential base that 3GPP can gather a wide range of support across industry, academy, and other stakeholders for wireless system development.

Therefore, 3GPP continues to be the essential technology standard partnership project for 5G.

In 2015, after ITU-R published the IMT-2020 vision (Recommendation ITU-R M.2083), 3GPP started the new radio (NR) study in Release-14 time frame (Fig. 1.6), which targets to fulfll IMT-2020 requirements. This study item (SI) included new numerology and frame structure design, coding scheme development, fexible duplexing study, and multiple access study. At the frst stage, NR study focuses on eMBB and URLLC usage scenarios as defned in IMT-2020 vision. NR mMTC may be developed in a later stage, since the earlystage mMTC applications are expected to be supported by NB-IoT or eMTC developed earlier in LTE.

In 2017, 3GPP started the normative work in Release-15 time frame for NR. The NR work item (WI) was set up to specify the technologies identifed useful for 5G development in Release-14 NR SI. The NR numerology, frame structure, duplex, and multiple access were specifed. NR MIMO was discussed and developed. The specifcation considers the IMT-2020 requirement and the potential spectrum deployment for 5G, including the high data rate and high spectral effciency

requirement, low latency requirement and high reliability requirement, etc., together with wide area coverage. Specifcally, the C-band characteristics played a key role in 5G development. To address the high data rate and low latency target simultaneously with large coverage issue, 3GPP developed several key technical components.

These key components include numerologies and frame structures that are suitable for large bandwidth, massive MIMO technology that is suitable for C-band deployment, and downlink uplink decoupling that is suitable for achieving high data rate (especially for UL) on C-band with existing inter-site distance.

In the Release-16 time frame, 3GPP will continue to enhance its NR capability to address higher spectral effciency for eMBB, and to enhance URLLC in multiuser scenarios. 3GPP also targets to extend NR usage to more vertical scenarios, including vehicular communications.

5G self-evaluation study was set up in 2017 to conduct the self-evaluation of 3GPP technologies developed in Release-15 and Release-16 for IMT-2020 submission. The preliminary self-evaluation was submitted to ITU-R in October 2018, and the fnal evaluation results were submitted in June 2019.

1.4.3 Independent Evaluation Groups To Assist Itu-R Endorse Imt-2020 Specifcation

To assist ITU-R on evaluating whether a candidate technology can fulfll the IMT-2020 requirements, ITU-R invites independent evaluation groups to conduct evaluation on the received candidates and reports to WP 5D on their conclusions. As of February 2020, there are 14 registered independent evaluation groups: - 5G Infrastructure Association - ATIS WTSC IMT-2020 Evaluation Group - ChEG Chinese Evaluation Group. - Canadian Evaluation Group - Wireless World Research Forum - Telecom Centres of Excellence, India. - The Fifth Generation Mobile Communications Promotion Forum, Japan - TTA 5G Technology Evaluation Special Project Group (TTA SPG33). - Trans-Pacifc Evaluation Group - ETSI Evaluation Group - 5G India Forum - Africa Evaluation Group. - Beijing National Research Center for Information Science and Technology (Bnrist EG) (application form—restricted access).

• Chinese Industry and Research Alliance of Telecommunications (CIRAT).

1.4.4 Status In Itu

As of the writing of this treatise, the independent evaluation entities have all completed their evaluations and submitted their results to the ITU. At the November 23rd, 2020, meeting, ITU-R SG 5 approved WP5D's draft recommendation [20], IMT 2020.SPECS which is the frst offcial 5G RAN standard. The IMT 2020.

SPECS was approved by the 193 member states of the ITU in February of 2021. The specifcation, "Detailed specifcations of the radio interfaces of IMT-2020" (ITU-R M.[IMT-2020.SPECS]), is now available on the ITU website [21].

1.5 Summary

In this section, the motivation and driving forces of 5G development are investigated, and the 5G use cases and technical requirements are introduced. Cellular communication system, from its frst generation, is focused on connecting humans. From 5G, however, the use cases are extended beyond MBB for connecting human beings to mMTC and URLLC for connecting different machines. 5G will therefore be the frst generation that devotes itself to connecting both humans and machines.

Accordingly, the service requirements and the technical performance requirements are extended from eMBB to the new use cases (or usage scenarios). For eMBB, it requires signifcant increase of data rate to adapt to the emerging highdefnition video and AR/VR transmission, at the cost of affordable energy consumption and spectrum occupation. This implies that both energy effciency and spectral effciency should be considerably increased. Furthermore, the cell edge data rate should be signifcantly increased to several or tens of megabits per second, with the similar coverage of existing network deployment, which poses a big challenge for eMBB 5G design. For URLLC, very low latency is required and at the same time high reliability should be guaranteed. On the other hand, such capability should be provided under various conditions including mobility. For mMTC, the large connection density is desired to provide massive connection in an effcient manner.

Besides, in both use cases, coverage is needed to enable an affordable network deployment. Finally, the above requirements should be achieved by a unifed air interface framework from a cost effciency perspective. In summary, 5G requirements are diverse, but at the same time the cost of energy, spectrum, and deployment should be restricted to an affordable level. All these issues pose signifcant challenges to system design.

Like previous generations, 5G development is based on industry collaboration, where ITU and 3GPP play the central role in this process. ITU-R defned the 5G vision, technical performance requirements, and evaluation criteria for the candidate 5G technologies, from the perspective of effcient use of spectrum and other resources, e.g., energy, site and deployment, to address the diverse requirements from 5G use cases. ITU-R has a well-established procedure of the IMT-2020 development, which guarantees that only the technologies that pass the abovementioned requirements could be approved as IMT-2020 (the name of 5G in ITU-R), and only such technologies can be allowed to use IMT spectrums. By this means, the technologies that are able to be deployed on IMT spectrum are guaranteed to be of high performance and high effciency when using the scarce spectrum and other resources. On the other hand, 3GPP is the major organization for 5G technical development which unites the telecommunications standard development organizations (SDOs) all over the world. By the support of the seven SDOs (also known as organization partners, OPs): ARIB and TTC from Japan, ATIS from America, CCSA from China, ETSI from Europe, TSDSI from India, and TTA from Korea, 3GPP provides the members with stable environment for 5G technology development. The OP members cover almost all the key parties, which ensure that the 3GPP technology addresses the concerns and issues from different parties and different regions. With the guidance of ITU-R and the well-harmonized technical development in 3GPP, 5G technology and 5G standardization are well developed. This is the key for 5G success.

Now that we have an understanding of the requirements of 5G and the process by which it was developed, the next chapter will begin our study of the enablers of the 5G wireless system. The discussion will begin with technical overview of LTE. This will naturally lead us to understand the 5G air interface by comparing and contrasting with the LTE design.

References

1. ITU-R, "Recommendation ITU-R M.2083," September 2015. [Online]. Available: https:// www.itu.int/rec/R-REC-M.2083.
2. NGMN, "NGMN 5G White Paper," February 2015. [Online]. Available: https://www.ngmn.

org/5g-white-paper/5g-white-paper.html.

3. IMT-2020 (5G) PG, "5G vision and requirement white paper," May 2014. [Online]. Available: http://www.imt-2020.cn/zh/documents/download/1.
4. METIS, "D1.1 Scenarios, requirements and KPIs for 5G mobile and wireless system," 2013 April. [Online].
5. METIS, "D6.6 Final report on the METIS 5G system concept and technology roadmap," 2015 April. [Online].
6. 5GMF, "5G Mobile Communications Systems for 2020 and beyond," 2016 July. [Online].

Available: http://5gmf.jp/en/whitepaper/5gmf-white-paper-1-01/.

7. 5G Americas, "5G services and use cases"," 2017 November. [Online]. Available: http:// www.5gamericas.org/fles/9615/1217/2471/5G_Service_and_Use_Cases__FINAL.pdf.
8. 3GPP TSG-RAN Secretariat, "3GPP RAN RP-202124: Draft report of electronic meeting RAN #89e held 14.09.-18.09.2020," December 2020. [Online]. Available: https://www.3gpp. org/ftp/tsg_ran/TSG_RAN/TSGR_90e/Docs.
9. 3GPP, "3GPP TR22.863: Feasibility Study on New Services and Markets Technology Enablers - Enhanced Mobile Broadband," September 2016. [Online]. Available: https://portal.3gpp.org/ desktopmodules/Specifcations/SpecifcationDetails.aspx?specifcationId=3015.
10. Wikipedia, "Bit Rate," [Online]. Available: https://en.wikipedia.org/wiki/Bit_rate. [Accessed 23 May 2019].
11. Huawei iLab, "Cloud VR Bearer Networks," 2017. [Online]. Available: https://wwwfle.huawei.com/-/media/CORPORATE/PDF/ilab/cloud_vr_oriented_bearer_network_white_ paper_en_v2.pdf?source=corp_comm.
12. 3GPP, "3GPP TR22.861: Feasibility Study on New Services and Markets Technology Enablers for Massive Internet of Things", September 2016. [Online]. Available: https://portal.3gpp.org/ desktopmodules/Specifcations/SpecifcationDetails.aspx?specifcationId=3013.
13. 3GPP, "3GPP TR 38.913: STudy on scenarios and requirements for next generation access technologies v15.0.0," June 2018. [Online]. Available: https://portal.3gpp.org/desktopmodules/Specifcations/SpecifcationDetails.aspx?specifcationId=2996.
14. 3GPP, "3GPP TR22.862: Feasibility Study on New Services and Markets Technology Enablers for Critical Communications," 2018 September. [Online]. Available: https://portal.3gpp.org/ desktopmodules/Specifcations/SpecifcationDetails.aspx?specifcationId=3014.
15. China (Peoples Republic of), "ITU Document 5D/757: Discussion on energy effciency and the requirement for future IMT," September 2014. [Online]. Available: https://www.itu.int/md/ meetingdoc.asp?lang=en&parent=R12-WP5D-C&PageLB=300.
16. The Climate Group, "Smart 2020: Enabling the low carbon economy in the information age,"
17. [Online]. Available: https://www.theclimategroup.org/sites/default/fles/archive/fles/ Smart2020Report.pdf.
18. Enerdata, "Global energy statistical yearbook 2014," 2014. [Online]. Available: http://yearbook.enerdata.net/.
19. ITU-R, "ITU-R Report M.2410: Minimum requirements related to technical performance for IMT-2020 radio interface(s)", November 2017. [Online]. Available: https://www.itu.int/ pub/R-REP-M.2410-2017.
20. Document IMT-2020/02 (Rev. 2), "Submission, evaluation process and consensus building for IMT-2020," July 2019. [Online]. Available: https://www.itu.int/md/R15-IMT.2020-C-0002.
21. ITU, "ITU completes evaluation for global affrmation of IMT-2020 technologies," 26 November 2020. [Online]. Available: https://www.itu.int/en/mediacentre/Pages/pr26-2020evaluation-global-affrmation-imt-2020-5g.aspx. [Accessed 1 December 2020].
22. ITU, "ITU-R Report M.2150: Detailed specifcation of the terrestrial radio interfaces of International Mobile Telecommunications-2020 (IMT-2020)", February 2021. [Online]. Available: https://www.itu.int/rec/R-REC-M.2150/en.

Chapter 2

4G Lte Fundamental Air Interface Design

Before we start our investigation of NR, we will frst give a detailed review of LTE. The focus will be on the basic aspects of LTE that NR builds upon. This will serve as a solid foundation for us to understand the 5G NR air interface. Those readers that are professionals in LTE may choose to skip this chapter and move directly to the NR air interface.

2.1 Lte Air Interface Overview

As the NR air interface borrows some design from LTE, one straightforward way to understand NR for those who are at least passingly familiar with LTE is to look at the differences between LTE and NR. This chapter reviews the signifcant parts of LTE that are critical to the understanding of NR. The interested reader on details of LTE is referred to some literatures such as [1, 2].

LTE standardization started from Release-8 in 3GPP, and then many features were standardized in the following releases from Release-8 to Release-14. The evolution in the standard from Release-15 which parallels NR is still going on to support different new features from industry interests. The designed LTE network can support diverse services including eMBB, Voice, Multimedia Broadcast Multicast Services (MBMS), D2D, IoT, Vehicle, etc. These services can be transmitted in the form of unicast transmission, or single-cell point-to-multipoint transmission (SC-PTM), or MBSFN (MBMS Single-Frequency Network) transmission. The transmission is mainly operated in the licensed spectrum, but eMBB can also be transmitted in the unlicensed spectrum. For conciseness of description, the overview of LTE air interface in the following sections will focus on unicast transmission for both downlink and uplink in the licensed spectrum.

2.2 Lte Frame Structure

The frame structure is dependent on the duplex scheme applied in the spectrum. Two duplex schemes FDD and TDD are supported. FDD is operated in the paired spectrum (see Sect. 3.2.1), where the downlink and uplink transmissions are performed in different carrier frequencies. For FDD, whether downlink and uplink transmission can occur simultaneously depends on UE capability. Half-duplex FDD is for the case when the UE cannot transmit and receive at the same time while there are no such restrictions for full-duplex FDD. TDD is operated in the unpaired spectrum; see Sect. 3.2.1, where the downlink and uplink transmissions occur in different time instances of the same carrier frequency. For TDD, switching time is needed to guarantee the switching from downlink to uplink transmission.

Two different frame structures, denoted as type 1 and type 2, are supported, which are applicable to FDD and TDD operated in licensed spectrum, respectively.

For frame structure type 1, one radio frame consists of ten sub-frames and each sub-frame is 1 ms consisting of two slots. There are ten sub-frames available for downlink transmission and ten sub-frames available for uplink transmission in each 10 ms interval as shown in Fig. 2.1.

For frame structure type 2, the length of one radio frame is 10 ms and each radio frame consists of ten sub-frames with 1 ms each. Each sub-frame consists of two slots of length 0.5 ms. The sub-frames in one radio frame are reserved for downlink transmissions, switching from downlink to uplink transmission and uplink transmissions separately. The sub-frame(s) reserved for switching transmission direction is called special sub-frame. There are three felds in the special sub-frame which are DwPTS, GP, and UpPTS, where DwPTS and UpPTS are used for downlink and uplink transmission separately as shown in Fig. 2.2. GP is the time duration without both downlink and uplink transmission, which is to mainly avoid the overlap of

uplink reception and downlink transmission at the base station side. The length of GP is related to the cell size as determined by the round-trip time of propagation. As there are different deployment scenarios in the practical network, several GP confgurations are supported in [3]. The supported downlink-to-uplink switching periodicities are 5 ms and 10 ms.

The uplink-downlink sub-frame confguration in one radio frame depends on the UL/DL traffc ratio. Normally, the DL/UL traffc is asymmetric and DL has more traffc than UL statistically [4], and therefore more DL sub-frames are needed in one radio frame. There are seven different uplink-downlink sub-frame confgurations defned in [3] to match different DL/UL traffc patterns, which are provided in Table 2.1 [3]. In the existing TDD commercial networks over the world, uplinkdownlink confguration 2 is a typical confguration.

As downlink and uplink transmissions are operated in the same carrier frequency for TDD, serious cross-link interference may occur without network synchronization; for example, uplink transmission suffers from downlink transmission of neighboring cell as illustrated in Fig. 2.3. In other words, the same uplink-downlink confguration needs to be confgured among the different cells of the same operator or between different operators. Due to the requirement of such strict confguration, the uplink-downlink confguration is semi-static and cannot be changed dynamically; otherwise it will cause severe cross-link interference.

2.3 Physical Layer Channels

The properties of the physical layer channels are described in this section starting with the multiple-access schemes.

2.3.1 Multiple-Access Scheme

Orthogonal Frequency-Division Multiplexing (OFDM)-based scheme is the basic transmission scheme for LTE downlink and uplink transmission. The basic principle of OFDM can be found in some textbooks [1, 5, 6]. OFDM is a multi-carrier technology which generates a number of orthogonal subcarriers with small bandwidth. Each subcarrier with small bandwidth faces relatively fat fading in the propagation, which can effectively remove the impact of frequency-selective fading and simplify the equalizer at the receiver especially for wideband transmission. OFDM can be implemented by IFFT which is illustrated in Fig. 2.4, and DFT is used for OFDM reception.

Some of the main aspects that led to the selection of OFDM for the downlink transmission include low cost and complexity implementation for wide (20 MHz) bandwidth, inherent and simple multipath protection through the use of a cyclic prefx, and a naturally scalable implementation where multiple system bandwidths can be supported.

Due to the multi-carrier transmission, the instantaneous OFDM signal has a large variation and results in high peak-to-average power ratio (PAPR). For the LTE uplink, PAPR was a hot discussion, soon to be replaced with the more accurate but not-as-straightforward-to-calculate cubic metric (CM) [7]. The power amplifer is required to have a large linear dynamic range, which will cause high cost and low effciency of power amplifer. Alternatively, the nonlinearity of the device results in

the distortion of the signal with large variation distortion, and power back-off is needed which reduces the cell coverage. Hence OFDM is not a good choice for UE in the uplink transmission. The selection of a single-carrier waveform allowed for a substantially lower CM for the uplink, thus facilitating lower cost lower power consumption UE. The SC-FDMA waveform used is "DFT"-spread OFDM (DFT-SOFDM), using with an extra DFT block making the net signal single carrier, which is illustrated in Fig. 2.5. SC-FDMA with cyclic prefx in the form of DFT-S-OFDM is used to achieve uplink inter-user orthogonality and to enable effcient frequencydomain equalization at the receiver side. This allows for a relatively high degree of commonality with the downlink OFDM scheme and the same parameters, e.g., clock frequency, can be reused [8].

Based on the downlink and uplink transmission scheme described above, UE multiplexing in the downlink and uplink transmission is operated by OFDMA and SC-FDMA as the multiple-access scheme, respectively. As DFT-S-OFDM is based on OFDM, different UEs in the downlink or uplink transmission can be multiplexed over different subcarriers within a certain time interval in terms of several OFDM symbols. An exemplary downlink and uplink UE multiplexing is illustrated in Fig. 2.6.

The downlink supports the transmission to UE in either localized or distributed manner. For localized transmission, a set of resources being consecutive in the frequency domain is allocated to achieve scheduling gain. In this case, it implies that the base station knows the part of the wideband channel that is of good quality for a certain UE, and then schedules the corresponding part for transmitting data to the UE. The distributed transmission is that a set of resources allocated for a certain UE where the resources are distributed in the frequency domain to achieve frequency diversity gain. Normally, due to the high mobility or the worse channel condition, the UE cannot accurately track or timely report the channel state information to the base station. In this case, since scheduling gain cannot be achieved, it is better to schedule for frequency diversity gain. It can be seen that knowledge of the channel information is key to making good scheduling decision and, thus, the UE is required to report the channel state information to the base station.

In the uplink, in order to keep the single carrier property with low CM, the UE multiplexing is different from downlink. All the subcarriers allocated for one UE have to be localized.

2.3.2 System Bandwidth

LTE targets for operating below 6 GHz and most of the existing LTE commercial networks are deployed in sub-3GHz licensed spectrum. As the spectrum has multiple channel bandwidths existing in sub-3 GHz, LTE is designed to support scalable channel bandwidth in order to facilitate the deployment. The supported channel bandwidth is shown in Table 2.2 [9].

If all the subcarriers corresponding to a certain channel bandwidth are utilized, there is the problem of out-of-band emissions because the added window for the OFDM symbol cannot completely remove the side lobes in the frequency domain. The out-of-band emissions will interfere with the adjacent channel. To avoid the serious impact of out-of-band emissions, a number of subcarriers at the edge of channel bandwidth need to be used as guard band without data transmission. In

LTE, at least 10% of channel bandwidth is required for the guard band. The transmission bandwidth is the actual bandwidth available for data transmission, which is smaller than the channel bandwidth; see Fig. 2.7 [9]. The transmission bandwidth is expressed in terms of the number of resource blocks (RB), and the bandwidth of one resource block is 180 KHz. The spectrum utilization is defned as the ratio of transmission bandwidth over channel bandwidth. It can be observed that there is 90% spectrum utilization for LTE except for the 1.4 MHz bandwidth.

Table 2.2 Channel bandwidth and the transmission bandwidth confguration

2.3.3 Numerology

The subcarrier spacing of OFDM in LTE is selected as 15 kHz for both downlink and uplink unicast transmission. It is noted that LTE also supports 7.5 kHz subcarrier spacing for dedicated MBMS network and 3.75 kHz subcarrier spacing for NB-IoT uplink. The selection of subcarrier spacing takes into account the impact of Doppler spread which is related to the candidate carrier frequency of LTE network and the supported velocity of up to 350 km/h [10]. In case of 15 kHz subcarrier spacing, the number of subcarriers for a channel bandwidth can be determined. A cyclic prefx is used to mitigate the impact of delay spread at the cost of overhead. Two cyclic prefx lengths including normal cyclic prefx and extended cyclic prefx are supported. The extended cyclic prefx is applied for the very large cell, e.g., 100 km cell radius and MBSFN transmission by using large number of cells simultaneously. The selection of cyclic prefx is a balance of overhead and mitigating delay spread impact.

In case of normal cyclic prefx, there are 7 OFDM symbols in one slot with 0.5 ms (Fig. 2.8). For the extended cyclic prefx, the number of OFDM symbols per slot is 6. A resource block (RB) is defned as 7 consecutive OFDM symbols with normal cyclic prefx in the time domain and 12 consecutive subcarriers in the frequency domain. A Resource element (RE) is defned as one subcarrier over one OFDM symbol, which is the smallest resource unit. There are 84 (7*12) REs in one resource block in the case of normal cyclic prefx while there are 72 REs in the case of extended cyclic prefx. The selection of RB size is a trade-off between resource allocation size and padding for small packets.

A resource grid consists of a number of NRB,T RBs, where 6 ≤ NRB, T ≤ 110. The functionality defned in the specifcation [3] supports the downlink and uplink transmission from 6 to 110 RBs; however, the actual number of used RBs is restricted by the transmission bandwidth defned in Table 2.2.

Fig. 2.8 Resource grid of normal cyclic prefix

The scheduling unit in the time domain is one sub-frame. A physical resource block (PRB) pair is defned as the two physical resource blocks in one sub-frame having the same frequency position. For the downlink and uplink localized transmission, the resource allocation unit is a physical resource block pair; that is, the same resource blocks in the two slots within one sub-frame are allocated. However, for the downlink distributed transmission or uplink transmission with intra-subframe frequency hopping, the allocated resource blocks for a certain UE in the two slots of one sub-frame are different.

2.3.4 Physical Channel Defnition

A downlink physical channel is defned as a set of resource elements carrying information originating from higher layers [3]. The following downlink physical channels are defned:

• Physical Downlink Shared Channel, PDSCH - Physical Broadcast Channel, PBCH - Physical Multicast Channel, PMCH - Physical Control Format Indicator Channel, PCFICH - Physical Downlink Control Channel, PDCCH - Physical Hybrid ARQ Indicator Channel, PHICH The information from higher layer is carried on the channels defned in higher layer including logical channels and downlink transport channels. The mapping from higher layer channels to physical channels is given in Fig. 2.9 [11].

The following uplink physical channels are defned [3, 11] (Fig. 2.10):

• Physical Uplink Shared Channel, PUSCH

- Physical Uplink Control Channel, PUCCH - Physical Random Access Channel, PRACH

2.4 Reference Signal

Reference signal refers to the so-called pilot signal used for channel estimation by the receiver. Specifcally in LTE, a reference signal is a predefned signal transmitted over a set of predefned resource elements in a resource grid. The concrete purpose of the estimated channel depends on the defnition of the corresponding reference signal. From the receiver perspective, the reference signal is utilized to obtain estimates of the channel of the resource grid conveying it, and therefore the channel of symbols conveyed on the same resource grid can be inferred from the derived channel. Each reference signal is associated with one antenna port.

2.4.1 Downlink Reference Signals

Downlink reference signals are used for downlink channel measurement and/or coherent demodulation of downlink transmission. The following reference signals are defned in downlink [3]:

• Cell-specifc reference signal (CRS) - UE-specifc reference signal (DM-RS) - CSI reference signal (CSI-RS) - Discovery reference signal (DRS) - MBSFN reference signal - Positioning reference signal

2.4.1.1 Cell-Specifc Reference Signal

The CRS is a basic signal defned in Release-8 and is used by UEs accessing LTE network. The demodulation of PBCH, PDCCH/PCFICH/PHICH, and PDSCH in transmission mode (TM) 1–6 [12] uses the channel estimation based on the CRS. When precoding is used for PDSCH transmission in case of multiple CRS, the used precoding matrices for PDSCH are signaled in downlink control information (DCI). The UE utilizes the estimated channel from the CRS and the signaled precoding matrices to generate the equivalent channel for coherent demodulation of PDSCH. The CRS is also used to derive the channel state information (CSI) in the form of CQI/RI/PMI for PDSCH scheduling in TM 1–8. In addition, the CRS can be used for both synchronization tracking after initial synchronization by PSS/SSS and RRM measurement for cell selection.

The CRS is transmitted in all downlink sub-frames for frame structure type 1, and all downlink sub-frames and DwPTS for frame structure type 2. In the normal subframe, the CRS is distributed over the two slots within one sub-frame. Futhermore, the CRS is only transmitted in the frst two OFDM symbols of MBSFN sub-frame and is defned for 15 kHz subcarrier spacing only. The number of CRS in one cell can be one, two, and four, which corresponds to one, two, and four antenna ports. The CRS antenna ports are named as antenna port 0, 1, 2, and 3. The structure of the CRS in one resource block pair in case of normal cyclic prefx is illustrated in Fig. 2.11 [3]. The CRS spans over the whole system bandwidth in its located OFDM symbol. It is noted that the frequency position of different CRS in Fig. 2.11 is relative.

For each CRS port, the distance between two adjacent reference signals in one OFDM symbol is 6 REs, and the starting position of the CRS has 6 frequency shifts.

The frequency shift is cell specifc, which is obtained by performing Ncell ID mod 6. Ncell ID is the cell ID index selected from [0, 503], where there are 504 cell IDs defned. The frequency shift is to enable the adjacent cells to avoid the CRS intercell interference. The CRS for antenna port 0/1 is located at the frst and ffth OFDM

symbols of one slot in a staggered manner; however, the CRS for antenna port 2/3 is only located in the second OFDM symbol of each slot in order to reduce the overhead of reference signal. The CRS overhead in case of two antenna ports is 16/168 = 9.52%, and it would be 19.05% if assuming the same density applied for antenna ports 2 and 3 in case of 4 antenna ports which is too high. The lower density of antenna port 2/3 may degrade the channel estimation performance at very high velocity.

The reference signal of antenna ports 0 and 1 is separated by FDM. The resource elements on antenna port 0 are not used for transmission when these resource elements correspond to the positions of the reference signal of antenna port 1. The power of the resource elements not used for transmission can be transferred to the resource elements of reference signal within the same OFDM symbol, which can improve the channel estimation performance. This is applied for antenna ports 2 and 3 as well.

In addition to using the CRS for CSI measurement and coherent demodulation of physical channels, the CRS is also used for measurement for cell selection/reselection and cell handover. CRS port 0 is used for determining reference signal received power (RSRP), and CRS port 1 can be optionally used depending on whether the UE can reliably detect if CRS port 1 is available. RSRP is defned as the linear average over the power contributions (in [W]) of the resource elements that carry cellspecifc reference signals within the considered measurement frequency bandwidth [13]. The network makes the decision whether to perform cell selection/reselection or cell handover based on the reported RSRP and another parameter, the reference signal received quality (RSRQ) [13].

2.4.1.2 Ue-Specifc Reference Signal

UE-specifc reference signal1 is used for demodulation of PDSCH in TM 7–10 [12].

The UE-specifc reference signal is only present in the scheduled resource blocks of the PDSCH transmission. For the UE-specifc reference signal-based PDSCH transmission, the used precoding matrix for PDSCH is transparent to UE and, hence, the signaling in DCI is not needed. As the UE does not have the information on how the precoding matrix is used over different resource blocks, it cannot perform channel interpolation among different scheduled resource blocks. In order to balance the precoding gain and channel estimation performance loss, PRB bundling is supported in TM9 and it is assumed that the same precoding matrix is applied in precoding resource block groups (PRGs). The PRG size depends on the system bandwidth; see Table 2.3 [12].

UE-specifc reference signal-based PDSCH transmission TM7 was frst introduced in Release-8, which only supports single-layer beamforming. The defned UE-specifc reference signal corresponds to antenna port 5 shown in Fig. 2.12 [12].

Table 2.3 PRG size

Fig. 2.12 Structure of antenna port 5 with normal cyclic prefx

In Release-9, TM8 was introduced to support dual-layer beamforming and two new reference signals were defned corresponding to antennas 7 and 8 in Fig. 2.13. Due to the support of backward compatibility, the introduced UE-specifc reference signal shall avoid the collision with cell-specifc reference signal, PDCCH and PBCH/PSS/SSS. The position is located at the last two OFDM symbols of each slot within one sub-frame. There are 12 REs for each UE-specifc reference signal in one resource block pair. The two UE-specifc reference signals occupy the same resource elements but are separated by code division multiplexing (CDM). The two codes [+1 +1 +1 +1] and [+1 –1 +1 –1] spanning in the time domain are used for antenna ports 7 and 8, respectively. The designed UE-specifc reference signal in TM8 supports the dynamic switching between SU-MIMO and MU-MIMO. The MU-MIMO transmissions are transparent to UE; that is, SU-MIMO operation is always assumed from the UE perspective. There are two candidate predefned sequences for the reference signal in TM8, and the two sequences are quasi-orthogonal. For each of the predefned sequences, the generated UE-specifc reference signals corresponding to antenna ports 7 and 8 are orthogonal to each other; however, the UE-specifc reference signals generated from the two predefned sequences are quasi-orthogonal.

Which sequence is used for the UE-specifc reference signal of a certain UE is indicated by DCI. Hence, it can support up to 4 UEs for MU-MIMO in TM8.

In Release-10, TM9 was introduced to support up to eight-layer transmission based on UE-specifc reference signal and thus up to eight UE-specifc reference signals need to be defned. The UE-specifc reference signals are extended from antenna ports 7 and 8 by extending the number of orthogonal cover code (OCC) from 2 to 4 and the number of resource elements from 12 to 24. The corresponding antenna ports are 7, 8, 9, 10, 11, 12, 13, and 14. For each group of 12 resource elements, there are 4 antenna ports which are separated by 4 codes. The mapping between the antenna port and the OCC code is shown in Table 2.4. The mapping of UE-specifc reference signal for antenna ports 7,8, 9, and 10 in case of normal cyclic prefx is illustrated in Fig. 2.14.

2.4.1.3 Csi Reference Signal

CSI reference signal is used for measuring the channel state information, which was introduced in LTE Release-10 and applicable for both TM9 and TM10. To increase the peak data rate, eight-layer transmission support was a requirement for Release-10 development. Extending the methodology of CRS antenna ports 0/2/4 design for 8 antenna ports is ineffcient and complicated. First, the resource overhead of 8 CRS ports would not be acceptable. As described in Sect. 2.4.1.1, the overhead of 4 CRS antenna ports is already 14.28% and the overhead of 8 antenna ports may be about 30%. Furthermore, in a practical network, the number of UEs operating larger than rank 4 transmission is small. Second, due to the requirement of backward compatibility, the legacy UEs operating in LTE Release-8 are required to work in LTE Release-10 and the legacy CRS antenna ports have to be presented in Release-10 network. It would be complicated for the newly defned 8 antenna ports and the legacy CRS antenna ports to coexist. These considerations necessitated a new methodology for reference signal design that is different from the CRS design. The functionalities of reference signal including channel measurement and demodulation are performed by the CSI reference signal and UE-specifc reference signal separately. The UE-specifc reference signal is described in Sect. 2.4.1.2.

Table 2.4 The mapping between antenna port and OCC code

R7R7 R7R7 R8 R8 R8 R8

Antenna Port 7 Antenna Port 8 Antenna Port 9 Antenna Port 10 Although the CSI reference signal is introduced because of supporting eightlayer transmission, the CSI reference signal is also defned for 1, 2, and 4 antenna ports in Release-10. The CSI reference signal is transmitted on one, two, four, or eight antenna ports, and the corresponding antenna ports are 15, 15/16, 15/16/17/18, and 15/16/17/18/19/20/21/22, respectively. As the CSI reference signal is only for channel measurement, low-density CSI reference signal is desired to reduce the overhead. Optimizing the trade-off between CSI reference signal overhead and channel measurement performance, the transmission of each CSI reference signal is wideband and the average number of RE per port in one resource block pair is 1 in the frequency domain. The periodicity in the time domain is confgurable with values of 5, 10, 20, 40, and 80 sub-frames.

The CSI reference signals in one resource block pair are multiplexed by CDM or hybrid FDM and TDM. Each CSI reference signal spans over two consecutive REs in the time domain, and the two consecutive REs can convey two CSI reference signals multiplexed by CDM. The two OCC codes are [+1 +1] and [+1 –1], respectively. The locations of CSI reference signal should avoid collisions with the CRS, UE-specifc reference signal, positioning reference signal, PDCCH, PBCH, and PSS/SSS. There are 40 REs reserved for CSI reference signal in one resource block pair for both frame structure type 1 and type 2, which correspond to 20, 10, and 5 CSI reference signal confgurations for the number of CSI reference signals being 1 or 2, 4, and 8, respectively. The CSI reference signal confgurations for different number of CSI reference signals applicable for both frame structure type 1 and type 2 in case of normal cyclic prefx are illustrated in Fig. 2.15, where the resource elements marked by the same color correspond to one CSI reference signal confguration and the number represents the index of CSI reference signal confguration; the resource elements marked with grey and "x" represent the positions of CRS or UE-specifc reference signal. In addition, there are additional 12, 6, and 3 CSI reference signal confgurations corresponding to a number of 1 or 2, 4, and 8 CSI reference signals defned for frame structure type 2 only [3].

The number of CSI reference signal and the corresponding CSI reference signal confguration are confgured by higher layer in a UE-specifc manner. The CSI reference signals within a CSI reference signal confguration are multiplexed by CDM and FDM, where the four pairs of antenna ports 15/16, 17/18, 19/20, and 21/22 are multiplexed by FDM and the two antenna ports of each pair are code multiplexed. Figure 2.16 takes CSI reference signal confguration 0 as an example to illustrate the CSI reference signal of different antenna ports.

As the CSI reference signal is a wideband transmission, it may collide with the resource blocks scheduled for Release-8 legacy UEs. In this case, data puncturing is performed for Release-8 legacy UEs. Since the base station knows the CSI reference signal confguration and Release-8 legacy UEs' scheduling, the impact of data puncturing on the legacy UEs can be mitigated by the scheduling and link adaption. For the Release-10 and later release UEs, rate matching is operated around the CSI

reference signal. Due to the UE-specifc CSI reference signal confguration, a UE does not know the confgured CSI reference signal of other UEs and therefore cannot do rate matching around the confgured CSI reference signal. To enable the rate matching, a zero-power CSI reference signal confguration is signaled to a UE and the UE assumes zero-power transmission in these resource elements. The zeropower CSI reference signal confguration reuses the CSI reference signal confguration of four CSI reference signals. The CSI reference signal confgurations for which the UE shall assume zero transmission power in a sub-frame are given by a 16-bit bitmap. For each bit set to one in the 16-bit bitmap, the UE shall assume zero transmission power for the resource elements corresponding to the CSI reference signal confguration.

The number of antenna ports was extended to 12, 16, 20, 24, 28, and 32 in Release-13 and -14 in the context of supporting full-dimension MIMO (FD-MIMO) [14, 15]. The indices for 12, 16, 20, 24, 28, and 32 antenna ports are labeled as p = 15,…, 26, p = 15,…, 30, p = 15,…, 34, p = 15,…, 38, p = 15,…, 44, p = 15,…, 38, and p = 15,…, 46, respectively. In case the number of antenna ports is larger than 8, the corresponding CSI reference signals are generated by aggregating Nres CSI > 1 CSI reference signal confgurations within one sub-frame. The number of antenna ports NCSIres ports per CSI reference signal confguration is 4 or 8. Although 2 antenna ports per CSI reference signal confguration is the most fexible confguration for aggregation, it will result in signifcant number of confgurations to increase signaling overhead and confguration complexity. The supported confgurations of aggregated CSI-RS confgurations are shown in Table 2.5.

The resource elements in one resource block pair corresponding to each aggregated CSI reference signal confguration with 4 or 8 antenna ports are same as shown in Fig. 2.15. In the case of CDM where OCC = 2 is used, each reference signal spans over two resource elements and the two reference signals multiplexed on the two same resource elements are separated by [+1 +1] and [+1 –1], which is shown in Fig. 2.16. When CDM OCC = 4 is applied, each reference signal spans over four resource elements within one resource block pair and is separated by [+1 +1 +1 +1], [+1 –1 +1 –1], [+1 +1 –1 −1], and [+1 –1 −1 +1]. An example of CSI reference signal confguration 0 with 4 and 8 antenna ports in the case of CDM OCC = 4 is illustrated in Figs. 2.17 and 2.18, respectively.

The mapping between the antenna ports of one CSI reference signal confguration and the code of OCC = 4 is provided in Tables 2.6 and 2.7.

Fig. 2.17 CSI reference signal confguration 0 with 4 antenna ports (CDM OCC = 4, normal cyclic prefx) R22 R22 R22 R22 R21 R21 R21 R21 R20 R20 R20 R20 R19 R19 R19 R19 Fig. 2.18 CSI reference signal confguration 0 with 8 antenna ports (CDM OCC = 4, normal cyclic prefx)

Table 2.7 OCC code assignment for one CSI reference signal confguration with 8 antenna ports

When CDM OCC = 8 is applied, each reference signal spans over eight resource elements within one resource block pair and is separated by the code with length equal to 8. It should be noted that CDM OCC = 8 is only applicable for 32 and 24 antenna ports. For the case of CDM OCC = 8 and the number of antenna ports equal to 32, the aggregated 4 CSI reference signal confgurations with 8 antenna ports each are restricted to one of {0, 1, 2, 3}, or {0, 2, 3, 4}, or {1, 2, 3, 4}; for the number of antenna ports equal to 24, the aggregated 3 CSI reference signal confgurations with 8 antenna ports each are restricted to {1, 2, 3}. The aggregated CSI reference signal confgurations can be found in Fig. 2.15. The OCC code assignments for different antenna ports in case of 32 and 24 antenna ports are illustrated in Tables 2.8 and 2.9, respectively.

For the number of antenna ports larger than 8, the total number of antenna ports equals Nres N CSICSIres ports which are labeled as p Nres N CSICSIres ports = … 15, , 16 15 + −1. The number of antenna ports per CSI reference signal confguration is NCSIres ports , and the antenna ports are indexed as p N ′ = … + − CSIres ports 15, , 16 15 1. The relation between p and p′ is [3], if CDM OCC = 2:

$p=\left\{\begin{array}{ccc}p^{\prime}+\dfrac{N_{\mathit{CSI\,res}}^{\mathit{pous}}}{2}\,i&p^{\prime}\in\left\{15,16,\ldots,15+\dfrac{N_{\mathit{CSI\,res}}^{\mathit{pous}}}{2}-1\right\}\\ p^{\prime}+\dfrac{N_{\mathit{CSI\,res}}^{\mathit{pous}}}{2}\left(i+N_{\mathit{res}}^{\mathit{CSI}}-1\right)&p^{\prime}\in\left\{15+\dfrac{N_{\mathit{CSI\,res}}^{\mathit{pous}}}{2},\ldots,15+N_{\mathit{CSI\,res}}^{\mathit{pous}}-1\right\}\\ i\in\left\{0,1,\ldots N_{\mathit{res}}^{\mathit{CSI}}-1\right\}\end{array}\right.$

15 17 19 21 +1 +1 +1 +1 +1 +1 +1 +1

16 18 20 22 +1 −1 +1 −1 +1 −1 +1 −1

23 25 27 29 +1 +1 −1 −1 +1 +1 −1 −1

24 26 28 30 +1 −1 −1 +1 +1 −1 −1 +1 31 33 35 37 +1 +1 +1 +1 −1 −1 −1 −1

32 34 36 38 +1 −1 +1 −1 −1 +1 −1 +1

39 41 43 45 +1 +1 −1 −1 −1 −1 +1 +1 40 42 44 46 +1 −1 −1 +1 −1 +1 +1 −1

else

$p=p^{\prime}+N_{\text{CStress}}^{\text{ports}}i\quad p^{\prime}\in\left\{15,16,...,15+N_{\text{CStress}}^{\text{ports}}-1\right\}$ $i\in\left\{0,1,...N_{\text{res}}^{\text{CSI}}-1\right\}$ end.

For each CSI reference signal confguration using CDM OCC = 2, the set of

54 Teacher's sigma $p'=15,16,\ldots,15$ .

antenna ports p′ = … + − $N_{\text{C}}^{\text{parts}}-1$ and $p^{\prime}=15+\frac{N_{\text{C}}^{\text{parts}}}{2},...,15+N_{\text{C}}^{\text{parts}}$ 15 16 152 , , , 1 and p′ = + … + − NN CSIres ports 152 , ,15 1 represent the antenna ports corresponding to two different polarizations in case of cross-polarization antenna confguration. Hence, the mapping between p and p′ can enable the overall antenna ports:

enlobe the overall antenna PMS. $p=15,16,\ldots,15+\frac{N_{\text{CNT}}^{\text{PWN}}N_{\text{rc}}^{\text{CSI}}}{2}-1$ and $p=15+\frac{N_{\text{CNT}}^{\text{PWN}}N_{\text{rc}}^{\text{CSI}}}{2},\ldots,15+N_{\text{CIS}}^{\text{PWN}}N_{\text{rc}}^{\text{CSI}}-1$ represent two different polarizations.

2.4.1.4 Discovery Signal

Discovery signal was introduced in Release-12 for small cell deployment scenario. In order to mitigate the interference between small cells and save energy, adaptive cell on/off that is adaptive with traffc load was defned [16]. Discovery signal is used to facilitate the fast and effcient discovery of small cells, which can reduce the transition time.

The discovery signal is transmitted in the downlink sub-frame(s) or DwPTS region of special sub-frame. A discovery signal consists of primary synchronization signal (PSS)/secondary synchronization signal (SSS), CRS port 0, and confgurable CSI-RS. The period of a discovery signal occasion for a cell can be confgured. The period can be 1–5 consecutive sub-frames for frame structure type 1 and 2–5 consecutive sub-frames for frame structure type 2. The periodicity of the discovery signal occasion is confgured by higher layer parameter.

During the period of a discovery signal occasion, PSS is in the frst sub-frame of the period for frame structure type 1 or the second sub-frame of the period for frame structure 2, SSS is in the frst sub-frame of the period, CRS port 0 is in all downlink sub-frames and in DwPTS of all special sub-frames in the period, and the CSI-RS is confgured in some sub-frames within the period if there is.

2.4.1.5 Other Downlink Reference Signals

In addition to the reference signals described above, there are some other downlink reference signals in LTE, including MBSFN reference signal which are only applicable for PMCH transmission, positioning reference signal for positioning, demodulation reference signals associated with EPDCCH, etc. Due to space limitations, these reference signals will not be discussed here and the details can be found in [3].

2.4.2 Uplink Reference Signals

Similar to the downlink reference signal, uplink reference signals are used for coherent demodulation and uplink channel measurement. In the downlink, the reference signal can be common for all the UEs in a broadcast manner, e.g., CRS; however, for uplink the reference signals are UE specifc. Two types of uplink reference signals are supported: - Uplink demodulation reference signal (DM-RS) - Sounding reference signal (SRS)

2.4.2.1 Uplink Demodulation Reference Signal

Uplink DM-RS is used for channel estimation by the base station to perform coherent demodulation of PUSCH and PUCCH. There are two types of DM-RS; one type is associated with the PUSCH and the other with the PUCCH transmission. For DM-RS, only single-antenna port is supported for PUSCH and PUCCH transmission in Release-8. In Release-10, it is extended to support 2 and 4 antenna ports for PUSCH transmission and 2 antenna ports for PUCCH transmission. The maximum number of DM-RS ports for PUSCH or PUCCH transmission is confgured by higher layer in a UE-specifc manner. The DM-RS antenna port labeling for PUSCH and PUCCH transmission is summarized in Table 2.10 [3].

DM-RS is described starting from single-antenna port transmission for simplicity and then extended to the case with multiple-antenna port transmission. As it is required to have low cubic metric transmission in uplink, DM-RS and data are time multiplexed in the assigned resource block(s) within one slot to preserve the single carrier property. DM-RS spans the same bandwidth as data in the frequency domain to estimate the channel of data transmission. For PUSCH, DM-RS is transmitted in the fourth OFDM symbol of the scheduled resource blocks within one slot and the remaining 6 OFDM symbols are used for data. The PUCCH transmission is constrained to one resource block. The multiplexing of PUCCH DM-RS and data in the resource block depends on PUCCH format; for example, for PUCCH format 1/1a/1b, the third, fourth, and ffth OFDM symbols in one resource block are used for DM-RS transmission. The multiplexing of DM-RS and data is illustrated in Fig. 2.19.

Different from downlink reference signal, the uplink DM-RS sequence shall have constant amplitude in both frequency and time domain. The constant amplitude of DM-RS sequence in the frequency domain is for obtaining the evenly estimated channel performance over different subcarriers. The constant amplitude in the time domain is to keep low cubic metric. However, due to oversampling in IFFT operation, there would be some amplitude variation in the time domain but still with low cubic metric. In addition, it is also desirable that the DM-RS sequence have zero autocorrelation for optimizing the channel estimation and low cross-correlation property to mitigate inter-cell interference.

Zadoff-Chu (ZC) sequence was, therefore, selected as the uplink DM-RS sequence because of its good auto- and cross-correlational properties [17]. ZC sequence is defned by

$x_{q}\left(m\right)=e^{-j{\frac{\pi q m\left(m+1\right)}{N_{2C}^{E S}}}}\quad0\leq m\leq N_{2C}^{E S}-1$

where NZC RS is the length of ZC sequence and q represents the qth root ZC sequence.

For a certain ZC sequence length NZC RS, the number of available root sequence equals to the number of integers being relative prime to NZC RS. That is to say the number of root sequence if the ZC sequence length is a prime number is maximized. More root sequences for a certain length means that different root sequences can be allocated to more different cells to mitigate inter-cell interference. For this reason, ZC sequence length is selected to be prime number in LTE.

DM-RS sequence spans the same bandwidth as PUSCH. The bandwidth of PUSCH depends on the number of assigned consecutive resource blocks, i.e., Msc = 12 ∗ NRB, where NRB is the number of scheduled resource blocks. The length of DM-RS sequence is equal to Msc. As the ZC sequence length is a prime, a ZC sequence with length close to Msc can be extended or truncated to generate the DM-RS sequence. The method of ZC sequence cyclic extension is adopted because of better cubic metric property [18]. The length NZC RS of the ZC sequence is determined as the largest prime number being smaller than Msc. Then the ZC sequence is extended to generate the DM-RS base sequence equal to Msc by

$1\leq M_{s c}-1$

Ru V N Xq Z N N C N M Rs , Mo Sc ( ) = ( ) D , 0 1 ≤ ≤ −

There are multiple base sequences corresponding to each possible scheduled bandwidth. The base sequences are divided into 30 groups with u ∈ {0, 1, …, 29} denoting the group number. The relationship between the group number u and the qth root sequence can be found in [3]. In each group, it is comprised of the base sequences with different lengths corresponding to different scheduled bandwidth NRB. The number of base sequences for each length in one group depends on the scheduled bandwidth NRB. When 1 ≤ NRB ≤ 2, the number of available base sequences of length Msc = 12 ∗ NRB is less than 30 because the length of ZC root sequence is smaller than 30. Consequently, it is not possible for each group to have one base sequence of such length. Instead, computer search generation method is used to generate 30 groups of base sequences of length Msc = 12 ∗ NRB, 1 ≤ NRB ≤ 2 in the form of ru(n) = ejφ(n)π/4, 0 ≤ n ≤ MSC − 1, where φ(n) is defned in [3]. As PUCCH transmission is constrained within one resource block, DM-RS sequence for PUCCH will use the computer search sequences. The length NZC RS of ZC sequence is larger than 30 when 3 ≤ NRB which allows for the generation of more than 30 base sequences for each length of Msc. In this case, each group contains one base sequence (v = 0) of each length Msc = 12 ∗ NRB, 3 ≤ NRB ≤ 5 and two base sequences (v = 0, 1) of each length Msc = 12 ∗ NRB, 6 ≤ NRB.

The DM-RS is located at the same position among different cells, which may cause serious inter-cell interference if the DM-RS base sequences are not assigned well. The base sequences are grouped to mitigate the inter-cell interference. The base sequences of each length are assigned to the different group, and one group is comprised of the base sequences with different lengths. Each cell is associated with one group. In this case, the DM-RS interference between the cells using different groups can be mitigated because of low cross-correlation property of ZC sequence. As there are only 30 base sequence groups, it can only enable 30 cells to associate with different base sequence groups. In case of 504 cells with different cell IDs, each base sequence group will be reused by 17 cells. Clearly two cells associated with the same base sequence group will result in very serious interference. This situation is solved by cell planning; that is, the adjacent cells are planned to associate different base sequence groups. This method can avoid the adjacent cells using the same base sequence group, but it is infexible and complicated because all the cell IDs need to be updated once a new cell is added in the network. Another method is to introduce base sequence group hopping over different time instances; that is, the base sequence group of one cell varies over different slots to randomize the inter-cell interference.

There are 17 different hopping patterns and 30 different sequence shift patterns.

The base sequence group number u is determined by

U F N F = ( ) Gh ( )S S + S Mod30

The sequence group hopping can be enabled or disabled by the cell-specifc higher layer parameter. The sequence group hopping pattern is generated by a pseudorandom sequence and the pseudorandom sequence generator is initialized with NcellID 30 . There will be 30 cells associated with each sequence group hopping pattern.

Then 30 sequence shift patterns associated with the 30 cells are used to generate different base sequence group numbers. The sequence shift pattern fss defnition differs between PUSCH and PUCCH. The sequence shift patterns for PUSCH and PUCCH are provided by fss = (Ncell ID + Δss) mod 30, where Δss ∈ {0, 1, …, 29} and fss = Ncell ID mod 30, respectively. Here TM10 uplink CoMP is not assumed; otherwise the physical cell ID Ncell ID can be replaced by a higher confgured virtual cell ID. The detail can be found in [3].

In one cell, when the UEs are scheduled in different resource blocks for PUSCH transmission, the corresponding DM-RS base sequences from the same group are transmitted in FDM which are orthogonal to each other. However, when uplink multiuser MIMO or virtual MIMO is applied, multiple users are scheduled on the same resource blocks to transmit simultaneously. For PUCCH, multiple users are multiplexed by CDM in one resource block. In this case, the DM-RS sequences of different UEs need to be orthogonal. Since the DM-RS base sequences have the zero autocorrelation property, the cyclic shift of the base sequence in the time domain can be assigned to different UEs. The length of cyclic shift shall be larger than the delay spread in order to differentiate different UEs. With the cyclic prefx length selected to be larger than the delay spread, the cyclic shift length can be determined to be larger than the cyclic prefx. Based on such considerations, it can support a maximum of 12 cyclic shifts. The cyclic shift in the time domain is equal to the phase shift in the frequency domain. The phase shift of base sequence is given by

$r_{u,v}^{\alpha}\left(n\right)=e^{j\alpha n}r_{u,v}\left(n\right),0\leq n\leq M_{s c}-1$ e of $n_{\mathrm{c}}$ is where $\alpha=\frac{2\pi n_{\alpha}}{12},n_{\alpha}\in\{0,1,...11\}$. The 0 1 11 n n cscs , , , . The value of ncs is signaled by DCI for PUSCH and higher layer confgured for PUCCH. The detail of signaling can be found in [19].

In Release-10, uplink multiple-antenna transmission is introduced to support up to four-layer PUSCH transmission. It is needed to defne up to 4 DM-RS and each DM-RS is associated with one layer. Similar to uplink virtual MIMO, the cyclic shifts of the base sequence can be used for different layers. In addition, orthogonal cover code (OCC) in time domain is also used to generate two-dimension DM-RS. The OCC codes [+1 + 1] and [−1–1] are used for the DM-RS in the two slots of one sub-frame as shown in Fig. 2.20. As there are 12 cyclic shifts and 2 OCC codes in time domain, in principle there are up to 24 DM-RS available. The cyclic shift is sensitive to frequency selectivity and the OCC code is sensitive to mobility. To achieve the better orthogonality of the DM-RS corresponding to different layers, there are eight combinations of cyclic shift and OCC codes are defned for up to four layers [3].

2.4.2.2 Uplink Sounding Reference Signal (Srs)

Uplink sounding reference signal is not associated with PUSCH transmission; that

is, there is no need to transmit SRS and PUSCH or PUCCH together. The SRS is used by base station for uplink channel measurement. The channel measurement is for uplink adaptation, uplink scheduling, power control, etc. In addition, the SRS can also be used for downlink beamforming due to channel reciprocity with TDD system.

For SRS, single-antenna port transmission was supported in Release-8, and then up to 4 SRS antenna port transmissions were introduced in Release-10. The number of SRS antenna ports is confgured by higher layer, which can be confgured as 1, 2, and 4. The SRS antenna port number and the number of confgured SRS antenna ports are given in Table 2.11.

Srs Transmission In Time Domain

The sub-frames used for SRS transmission are confgured by a cell-specifc higher layer parameter, which enables all the UEs in the cell to know the position of the SRS. There are 16 SRS sub-frame confgurations defned for both frame structure type 1 and 2 [3]. An extreme case is that all the uplink sub-frames can be confgured for SRS transmission in the context of high load in the cell. For all the sub-frames other than special sub-frames confgured for SRS, SRS is transmitted in the last symbol of the sub-frame. In the special sub-frames, it supports up to two symbols for SRS transmission before Release-13, and then was extended to support up to 6 symbols for SRS in Release-13 because of the requirement of SRS capacity increase.

For each UE, a subset of sub-frames within the confgured cell-specifc SRS subframes are confgured for its own SRS transmission. The starting sub-frame and periodicity of SRS transmission for each UE are confgured by a UE-specifc parameter. The location of SRS in time domain is illustrated in Fig. 2.21. In the sub-frames confgured for SRS transmission, PUSCH and PUCCH will not be transmitted in the last symbol to avoid the collision.

Srs Transmission In Frequency Domain

There is a cell-specifc SRS bandwidth confgured by a cell-specifc parameter which is common for all the UEs in the cell. There are eight cell-specifc SRS bandwidth confgurations for different system bandwidths. For each UE, there is one UE-specifc SRS bandwidth confguration. The UE-specifc SRS bandwidth is in multiples of four resource blocks. There are two methods to transmit SRS which are

wideband transmission and frequency hopping (Fig. 2.22). The wideband transmission is that the UE transmits the SRS spanning the whole UE-specifc SRS bandwidth in one shot. It can obtain the whole-channel measurement result in one sub-frame. However, the power spectrum density is lower which will degrade the channel estimation performance especially for power-limited UE. Hence, the wideband SRS transmission is useful in the scenarios with good channel condition and the UE power is not limited. In the UE power-limited scenario, SRS frequency hopping can be used; that is, UE transmits SRS in a subset of SRS bandwidth in one shot and spans multiple time instance to obtain the whole-channel measurement result. It can improve the channel estimation performance in each shot, but there is latency to obtain the whole-channel information.

SRS is transmitted in every second subcarrier within the SRS transmission bandwidth in the form of comb transmission. The SRS on the two combs is multiplexed by FDM and the SRS on the same comb is multiplexed by CDM as shown in Fig. 2.23.

If the SRS bandwidth is MSRS resource blocks, then the length of SRS sequence is MSC = MSRS ∗ 12/2. The SRS sequence is defned by

$r_{s R S}^{P}\left(n\right)=e^{j\alpha_{p}n}r_{u,v}\left(n\right),0\leq n\leq M_{s c}-1$

where u is the base sequence group number and v is the sequence number. The determination process of u is same as that with DM-RS, but the difference is that the sequence shift pattern is fss = Ncell ID mod 30. The cyclic shift of SRS is given as

$\alpha_{_{p}}=2\pi\,\frac{n_{_{C S}}^{p}}{8},n_{_{C S}}^{p}\in\left\{0,1,\ldots7\right\}$ $n_{_{C S}}^{p}=\left(n_{_{S S}}^{C S}+\frac{8p}{N_{\mathrm{ap}}}\right)\mathrm{mod}8$ $p\in\left\{0,1,\ldots,N_{\mathrm{ap}}-1\right\}$

where Nap is the number of confgured SRS antenna ports, and nSRS CS ∈{ } 0 1, ,234 , , , , 5 6,7 is the cyclic shift signaled by the higher layer parameter. For the comblike transmission, the signal is a repetition in time domain which reduces the number of available cyclic shifts.

Aperiodic Srs

In addition to periodic SRS transmission, aperiodic SRS transmission was introduced in Release-10. Since uplink MIMO is supported in Release-10, more SRS resources were needed. Given the existing SRS capacity, the periodicity of SRS needs to be increased in order to support multiple antennas. In this case, the channel measurement will not track the channel variation timely due to the longer SRS periodicity especially for high mobility. To obtain accurate channel information, aperiodic SRS is supported. Different from periodic SRS transmission, aperiodic SRS is triggered by uplink grant and not with a higher layer parameter. The confgurations of aperiodic SRS transmission are still higher layer confgured, e.g., the comb, SRS bandwidth, and cyclic shift. Once it is triggered, aperiodic SRS is transmitted in the confgured SRS sub-frame immediately after the PDCCH UL grant. With proper aperiodic trigger, the channel variations can be properly tracked.

2.5 Downlink Transmission

Now that the role of the reference signals is understood, the different physical channels as introduced that are essential to grasp the operation of LTE in Sect. 2.3 can be described in detail. In this section, the transmissions of physical downlink broadcast channel, control channels, and data channel are introduced. There are also some other physical channels including PMCH, EPDCCH, MPDCCH, etc., which will not be discussed here for simplicity.

2.5.1 Pbch

Before the UE sets up the connection with a cell, the UE frst acquires the cell synchronization (including time/frequency/frame synchronization and cell ID) based on Primary Synchronization Signal (PSS) and Secondary Synchronization Signal (SSS). Then the UE needs to determine the system information, such as system bandwidth and number of antenna ports. The system information is divided into the Master Information Block (MIB) and a number of System Information Blocks (SIBs) [11]. The MIB is transmitted on the PBCH and includes a limited number of the most essential and most frequently transmitted parameters that are needed to acquire the other information from the cell.

The information in the MIB includes [11]:

• Downlink bandwidth: indicates the system (for both downlink and uplink) bandwidth in terms of the number of resource blocks from {6, 15, 25, 50, 75, 100}, 3 bits.
• PHICH confguration: indicates the PHICH resource used to derive the resources for control channel and data channel, 3 bits.
• System frame number (SFN): eight most signifcant bits of SFN are included in the MIB, which indicates the number of the starting frame with PBCH transmission.
• Spare bits: 10 spare bits are reserved.

Since the UE does not know the system information before PBCH detection, the PBCH is transmitted in a predefned position so that the UE can detect it after cell synchronization. The PBCH is located at the central 72 subcarriers (i.e., central 6 resource blocks) of the carrier which is same as that of PSS/SSS, and the frst four consecutive OFDM symbols in slot 1 in sub-frame 0 of each radio frame. The transmission time interval (TTI) of the PBCH is 40 ms (4 consecutive radio frames).

Consequently, it is transmitted on each radio frame fulflling nf mod 4 = 0. Recalled from the description of the CRS, some of the OFDM symbols within PBCH are for the transmission of the CRS, and therefore rate matching around the CRS is needed when performing resource mapping for PBCH. However, the number of CRS antenna ports is unknown before PBCH detection. To simplify the UE receiver design, it is assumed that there are always 4 CRS antenna ports for PBCH resource mapping regardless of the actual number of CRS antenna ports. Then the number of available resource elements for PBCH within one TTI is 240*4 = 960. The resource mapping of PBCH is illustrated in Fig. 2.24.

The system designer will need to guarantee that the performance of PBCH transmission is such that all the UEs in the cell can detect the PBCH. As the payload size of PBCH is small with only 40 bits (24 information bits+16 CRC bits), convolutional coding and QPSK modulation are used because of their better performance.

The rate matching is done by using a 1/3 mother coding rate with repetition to obtain a very low coding rate (40/1920 = ~0.02) for the PBCH, which enables the PBCH in each radio frame within one TTI to be decodable. In addition, transmit diversity scheme space-frequency block coding (SFBC) and SFBC+frequency shift transmit diversity (FSTD) are used in the case of 2 and 4 CRS antenna ports, respectively. The information on the number of CRS antenna ports is embedded in the CRC mask of the PBCH; that is, there are three different CRC masks that correspond to 1, 2, and 4 CRS antenna ports [19]. The number of CRS antenna ports is blindly detected by the UE.

2.5.2 Control Channel

The downlink physical control channel and data channel are time multiplexed as shown in Fig. 2.25. Normally, in one sub-frame, the control region is located within the frst n = 1/2/3 OFDM symbols or n = 2/3/4 for bandwidth smaller than 10 RBs.

The maximum number of OFDM symbols in the control region is dependent on sub-frame type [3]. The number of OFDM symbols in the control region is dynamically changing. Time division multiplexing (TDM) of the control region with the data region has the beneft of low latency and power saving for the UE. For example, with TDM, once the UE completes the PDCCH detection, it can immediately start decoding the scheduled data and reduce the waiting time. Furthermore, if the UE detects that there is no PDCCH transmission for it, it can go to micro sleep for power saving. In the control region, three downlink physical control channels PCFICH, PDCCH, and PHICH are transmitted. The function and processing of each downlink physical control channel are described in the following sections.

2.5.2.1 Pcfich

The function of the PCFICH is to transmit an indicator to the UE of the number of OFDM symbols used for the transmission of the PDCCHs in a sub-frame. Moreover, the starting position of PDSCH can be implicitly derived from the PCFICH transmission. Since the number of OFDM symbols used for PDCCH is dynamically Fig. 2.25 Multiplexing of

control region and PDSCH (e.g., 3 OFDM symbols) changing, the PCFICH is always transmitted in the frst symbol of each sub-frame so that the UE knows where to detect it.

To guarantee the reliability of PCFICH detection, three code words with a length of 32 bits are used to represent 1, 2, or 3 (or 2, 3, or 4) OFDM symbols [19]. The 32 bits are modulated to 16 QPSK symbols which are divided into 4 groups. Each group with 4 QPSK symbols is mapped onto one resource element group (REG) which is the resource unit for control channel to resource element mapping [3]. There are 6 consecutive resource elements in one REG in the frst OFDM symbol, where two resource elements are reserved for CRS. The four REGs are almost evenly distributed over the whole bandwidth to achieve frequency diversity shown in Fig. 2.26. In addition, the PCFICH is transmitted on the same set of antenna ports as PBCH. In case of 2 or 4 CRS antenna ports, SFBC or SFBC+FSTD is applied to achieve transmit diversity gain.

As the PCFICH is located as the frst OFDM symbol of each sub-frame for all the cells, it may cause interference between the PCFICH of different adjacent cells. To randomize the interference, cell-specifc scrambling sequence and cell-specifc starting position of the frst REG of PCFICH are used.

2.5.2.2 Pdcch

The PDCCH transmits downlink control information (DCI) including downlink scheduling assignments, uplink scheduling assignments (i.e., UL grant), and power control information to the UE. A number of DCI formats are defned such as format 0/1/1A/1B/1C/1D/2/2A/2B/2C/3/3A/4, which depends on downlink or uplink scheduling assignment, carried control information (e.g., power control, MCC change), transmission scheme, and payload size. For simplicity, the DCI formats in Release-10 are summarized in Table 2.12. It should be noted that some new DCI formats are introduced after Release-10. These new DCI formats and the detail of the information feld in each DCI format can be found in [19].

A PDCCH is transmitted on an aggregation of one or several consecutive control channel elements (CCE), where one CCE corresponds to 9 REGs and there are 4 available REs per REG. The supported CCE aggregation levels are 1, 2, 4, and 8. The number of CCEs assigned for one PDCCH depends on the channel quality and the payload size of PDCCH. The channel quality is based on the reported CQI from the scheduled UE, and the payload size depends on the DCI format.

The number of available CCEs in the system is NCCE = ⌊NREG/9⌋ and they are enumerated from 0 to NCCE − 1, where NREG is the number of REGs not assigned to PCFICH and PHICH; see Sect. 2.5.2.1. When assigning the CCEs for PDCCH, a PDCCH consisting of n consecutive CCEs can only start on a CCE fulflling i mod n = 0, where i is the CCE number. In principle, the number of candidate positions in the system for the PDCCH with aggregation level n would be Ln = ⌊NCCE/n⌋. Consequently, for aggregation level n, a UE has to blindly detect Ln candidate PDCCH positions. To reduce the complexity of PDCCH blind detection, the number of candidate positions for a PDCCH with aggregation level n is restricted to be a small value and the CCEs of all the candidate positions are required to be consecutive. For example, the numbers of candidate positions for aggregation levels 1, 2, 4, and 8 are 6, 6, 2, and 2, respectively. How to place the PDCCHs on the CCEs to minimize the number of PDCCH blind detection is part of the search space design by the system designer [12].

Given that the PDCCHs to be transmitted in a sub-frame are placed on the CCEs as aforementioned, the block of information bits on each PDCCH is multiplexed. If there is no PDCCH on some CCEs, elements are inserted during the multiplexing to enable the PDCCHs to start at the required CCE position. The block of multiplexed bits is scrambled with a cell-specifc scrambling sequence and then modulated to QPSK symbols. The QPSK symbols are grouped into a sequence of symbol quadruplets, where each symbol quadruplet consists of 4 consecutive QPSK symbols. Then the sequence of symbol quadruplets is interleaved by using a subblock interleaver [19]. The block of symbol quadruplet output from the interleaver is cyclically shifted, where the cyclic shift size is determined by the physical cell identity. The interleaving can enable CCEs of one PDCCH to be distributed in the frequency domain to take advantage of frequency diversity gain. The cell-specifc scrambling sequence and cell-specifc shift size are provided to the system designer to randomize the inter-cell interference to the PDCCH.

An illustration of REG is given in Fig. 2.27. The cyclically shifted symbol quadruplets are sequentially mapped onto the REGs not assigned to either the PCFICH or the PHICH. The mapping is time frst and then going to the frequency domain. The size of REG depends on which OFDM symbol is located in the control region and the number of CRS antenna ports. In the frst OFDM symbol, one REG consists of six consecutive REs in one resource block and there are two REGs per resource block, where two REs in each REG are reserved for CRS regardless of one or two CRS antenna ports. In case of four CRS antenna ports, the REG defnition in the second OFDM symbol within control region is same as that of the frst OFDM symbol. With two CRS antenna ports, each REG in the second OFDM symbol consists of four consecutive REs in one resource block and there are three REGs per resource block. For the third OFDM symbol, the REG is defned as that of the second OFDM symbol with two CRS antenna ports.

The same set of antenna ports as the PBCH are used for PDCCH transmission.

In case of two or four CRS antenna ports, SFBC or SFBC+FSTD is applied to achieve transmit diversity gain. The processing of PDCCH is illustrated in Fig. 2.28.

2.5.2.3 Phich

The PHICH carries the hybrid-ARQ (HARQ) ACK/NACK in response to uplink

PUSCH transmission. For proper uplink HARQ system operation, PHICH performance should be maintained with very low error probability to reduce the number

of retransmission and avoid the fail reception of a transport block in MAC layer.2 The target performance is up to the system designer but it is nominally acknowledged that the probability of ACK erroneously detected as NACK is in the order of 10−2 and the error rate of NACK detected as ACK shall be at least lower than 10−3.

In the case of PUSCH transmission with one or two transport blocks, there are the corresponding one or two bits HARQ ACK/NACK. Multiple PHICHs are mapped to the same set of resource elements and code multiplexed by using different orthogonal sequences nPHICH seq , which are known as a PHICH group nPHICH group . A PHICH resource is identifed by the index pair n n PHICH group PHICH seq ( , ).

The number of PHICH groups supported in the system is confgurable and is transmitted to the UE in the MIB of the PBCH. In the PBCH, PHICH confguration includes 1-bit PHICH duration and 2-bit PHICH resource. The PHICH duration indicates that PHICH spans 1 or 3 OFDM symbols in the control region. If the PHICH is only located at the frst OFDM symbol of the control region, the room for power boosting of PHICH would be limited because of the power use to transmit the PCFICH and PDCCH if they also exist in the frst OFDM symbol and therefore the coverage of PCFICH would be restricted. LTE, thus, provides the PHICH resource to indicate the number of PHICH groups expressed as a fraction of the downlink system bandwidth in terms of resource blocks so that the system designer can spread the PHICH over multiple OFDM symbols. For frame structure type 1, the number of PHICH groups is constant in all sub-frames according to the confguration from PBCH. However, for frame structure type 2, the number of PHICH groups may vary between downlink sub-frames because each downlink sub-frame may associate with different number of uplink sub-frames. For example, in uplink-downlink confguration 0, the PHICHs in sub-frame 0 associate with the PUSCHs from two uplink sub-frames [12]; for other cases, the PHICHs in one downlink sub-frame associate with the PUSCHs from one uplink sub-frame.

In a PHICH group, there are up to eight orthogonal sequences in the case of normal cyclic prefx and four orthogonal sequences in the case of extended cyclic prefx. The orthogonal sequences used in one PHICH group are shown in Table 2.13.

For one UE, each of the HARQ ACK-NACK bits transmitted on one PHICH in one sub-frame is encoded with three repetitions resulting in a block of coded bits. The block of coded bits is modulated using BPSK to generate a block of modulation

symbols. There are three modulation symbols corresponding to one HARQ ACK/ NACK bit and each of the modulation symbols is spread with the assigned orthogonal sequence. The spread modulation symbols from different UE are multiplexed and mapped onto REGs as shown in Fig. 2.29. The distance of two adjacent REGs of one PHICH group in the frequency domain is almost 1/3 the system bandwidth to achieve the frequency diversity gain.

In the case of two CRS antenna ports, SFBC is used for PHICH transmission.

However, in case of four CRS antenna ports, SFBC+FSTD cannot be directly used. To keep the orthogonality of PHICH sequences in one REG, the four modulation symbols mapped onto the same REG have to be transmitted on the same antenna port. Hence it only performs SFBC on two antenna ports for the modulation symbols within one REG of PHICH, which causes the power imbalance among different antenna ports. To achieve more transmit diversity gain and enable the power balance over different antenna ports, SFBC operation on (port 0, port 2) or (port 1, port 3) is switched over the frst, second, and third REGs of one PHICH group.

2.5.3 Pdsch

PDSCH carries the data to be transmitted in the downlink with the following transmission schemes [12]:

- Single-Antenna Port Scheme:

• PDSCH transmission is on one antenna port which can be on port 0 (transmission mode (TM) 1), port 5 (TM7), port 7, or port 8 (TM8).
• Transmit diversity scheme:
• It is an open-loop rank 1 transmission scheme based on CRS antenna ports. In case of two and four CRS antenna ports, the transmit diversity schemes are SFBC and SFBC+FSTD, respectively. Transmit diversity scheme is supported in TM 2. It is also the fallback transmission scheme in TM3–10.

• Large-delay CDD scheme:
• It is CRS based open-loop spatial multiplexing with rank = 2/3/4 transmission, which is supported in TM3.
• Closed-loop spatial multiplexing scheme:
• CRS-based closed-loop spatial multiplexing with up to four layers, which is supported in TM4/6.
• Multiuser MIMO scheme:
• Up to two users are multiplexed together for MU-MIMO, which is CRS-based transmit scheme and supported in TM5.
• Dual-layer scheme:
• It was introduced in Release-9 for dual-layer beamforming. PDSCH is performed with two transmission layers on antenna ports 7 and 8 in TM8.
• Up to eight-layer transmission scheme:
• It is introduced in Release-10 and DM-RS-based transmission scheme. PDSCH would be performed with up to 8 transmission layers on antenna ports 7–14. It is supported in TM 9 and 10. The UE is semi-statically confgured via higher layer signaling to receive PDSCH data transmissions signaled via PDCCH according to one of ten transmission modes, denoted as TM 1–10. The general structure of PDSCH processing is shown in Fig. 2.30.

Transmission of up to two transport blocks is supported in one TTI. After the formation of each transport block, CRC and channel coding are performed. At this point, there is a one-to-one mapping between transport block and code word. The number of code word (i.e., transport block) is determined by the value of rank. If the rank in the initial transmission equals to 1, there is only one code word; otherwise two code words are supported. For code word to layer mapping, one code word can be mapped onto more than one layer when rank is larger than 2. The selection of up to two code word transmission is a trade-off of performance and signaling overhead.

If each code word is mapped onto one layer, it can enable UE to perform SIC operation for performance improvement; however, the overhead of required DL/UL

Fig. 2.30 General structure of PDSCH processing

2 4G LTE Fundamental Air Interface Design signaling, e.g., MCS/CQI, would be proportional to the number of code words. The detailed mapping of the code word to layers is rather complicated and is beyond the scope of this treatise. The interested reader should consult the standard [3].

The precoding operation is different among the aforementioned transmission schemes. For the transmission schemes in TM 1–6, the antenna ports after the precoding in Fig. 2.30 refer to CRS antenna port which can be confgured as 1/2/4. In this case, the precoding matrix is predefned or signaled to UE by DCI; that is, UE knows what precoding matrix is used at the eNB side. The precoding matrices are defned in [3]. However, for the transmission schemes in TM 7–10, the antenna ports are DM-RS ports and the number of DM-RS ports is equal to the number of layers. The precoding is transparent to UE and the signaling of the precoding matrices is not needed anymore. The precoding information is embedded in the DM-RS. The precoding matrix used for PDSCH is up to eNB implementation, which gives the freedom for eNB operation. Such operation is friendly to TDD because of utilizing channel reciprocity and without the need of precoding matrix indicator (PMI) feedback. For FDD, precoding matrices are defned only for CSI reporting. The UE performs channel measurement based on CSI-RS and reports the selected precoding matrices to the eNB for reference. Regarding how to use the reported precoding matrices, this is left for eNB implementation.

The resource mapping is the same for all the transmission schemes. The precoded data is mapped onto the resource elements of each antenna port. The resource mapping is performed from frequency frst and then time within the assigned PDSCH resource blocks. The mapping skips the resource elements occupied by different kinds of reference signal, e.g., CRS, DM-RS, and CSI-RS.

2.5.4 Modulation Coding Scheme (Mcs)

Scheduling based on the reported channel state information (CSI), such as rank indication (RI), precoding matrix indicator (PMI), and channel quality indication (CQI), is an important characteristic of LTE system. CQI refects the channel quality and is reported via the modulation scheme and coding rate. A number of 15 CQI values are defned, and the step size in terms of SINR is about 2 dB. When UE reports a CQI index based on the channel measurement, there is the assumption that "a single PDSCH transport block with a combination of modulation scheme and transport block size corresponding to the reported CQI index, and occupying a group of downlink physical resource blocks termed the CQI reference resource, could be received with a transport block error probability not exceeding 0.1" [12]. Then eNB performs PDSCH scheduling based on the reported CSI.

In order for the UE to receive the PDSCH transmission, it needs to determine the modulation order, transport block sizes (TBS), and assigned resource blocks. The combination of modulation order and TBS index is indicated by a 5-bit MCS feld in the DCI. The assigned resource blocks are also signaled in the DCI according to the resource allocation types. The 5-bit MCS table includes 29 combinations of modulation and TBS index and three states reserved for implicit modulation order and TBS signaling for retransmissions. The 29 MCS correspond to 29 spectral effciency with fner granularity compared to the CQI values, where the spectral effciency has an overlap between QPSK and 16QAM and also for 16QAM and 64QAM. In total there are 27 different spectral effciencies because some values of the 5-bit MCS table have the same spectral effciency. Although the spectral effciency of two modulation schemes is same, different modulation order shows better performance in different fading channels. It is up to system designer to select the modulation order given the same spectral effciency based on the fading channels.

There is a basic TBS table defned, the dimension of which is 27 × 110 for the case one transport block not mapping to more than one layer. From the signaled MCS in the DCI, the modulation order and TBS index can be determined. The TBS of PDSCH is determined by looking up the basic TBS table according to the TBS index and the number of assigned resource blocks. If the transport block is mapped to more than one layer, the determined TBS from the basic TBS table is then translated to a fnal TBS corresponding to the number of layers that the transport block is mapped to. The three entries for retransmission in the 5-bit MCS table represent that the TBS is assumed to be as determined from DCI transported in the latest PDCCH for the same transport block.

In principle, the TBS can be directly calculated according to the MCS and the number of available resource elements, which results in an arbitrary TBS. The arbitrary TBS would not match the size of Turbo QPP interleaver (the padding or depadding is needed) and the MAC payload size is normally calculated with bytes [20, 21]. In addition, the TBS may vary during the initial transmission and retransmission because the number of available resource elements could be different, e.g., the control region is changed. Hence, it is desirable that TBS corresponding to different number of resource blocks is constant across different sub-frames, and the TBS should be aligned with QPP sizes to remove the need for padding/depadding.

Calculating the TBS from MCS requires a resource assumption: three OFDM symbols for control region, two CRS antenna ports, no PSS/SSS/PBCH, and normal CP [20]. The generated TBS tables can be found in [12]. It is noted that the actual resource allocations may not be same as the resource assumptions for generating TBS tables, e.g., the number of OFDM symbols in the control region is not 3 and the existence of CSI-RS or DM-RS. The resulted coding rate may not be the exact coding rate of MCS, but is close to the MCS.

2.6 Uplink Transmission 2.6.1 Pucch

Uplink control information (UCI) including scheduling request (SR), HARQ-ACK, and periodic CSI is transmitted on PUCCH when there is no PUSCH transmission in the same sub-frame. In the case of concurrent transmission of PUCCH and PUSCH, UCI is piggybacked on PUSCH if the simultaneous transmission of PUCCH and PUSCH is not confgured. The PUCCH is located at the resource blocks at the edge of the transmission bandwidth within one slot and frequency hopping across the two slots within one sub-frame is performed as shown in Fig. 2.31.

The placement of PUCCH at the edge is to facilitate the continuous resource block allocation for PUSCH and to achieve frequency diversity gain. The resource blocks allocated for PUCCH are defned by the scheduler and are transmitted via higher layer signaling.

Several PUCCH formats are defned according to the conveyed UCI on PUCCH. Some PUCCH formats (e.g., format 3) are extended to accommodate new UCI as the releases evolve. For this treatise, only the original function of such kind of PUCCH format is introduced for simplicity. The extended function can be found in [12]. The conveyed UCI of different PUCCH formats is summarized in Table 2.14.

2.6.1.1 Pucch Formats 1/1A/1B

For PUCCH format 1, scheduling request (SR) is carried by the presence or absence

of transmission of PUCCH from the UE. The periodic resource for SR is UE-specifcally confgured by higher layer signaling. If there is the scheduling request in the SR time instance, UE transmits the bit "1" in the form of BPSK on

PUCCH; otherwise, UE does not transmit anything. PUCCH format 1a/1b carries 1-bit and 2-bit HARQ-ACK, and the corresponding modulation schemes are BPSK and QPSK, respectively.

The modulation symbol is multiplied with a cyclically shifted sequence with length 12, where the sequence is the same as the uplink DM-RS sequence defned for one resource block. The multiplied sequence is block-wise spread with an orthogonal sequence with length 4 or 3 depending on the number of OFDM symbols for PUCCH 1/1a/1b within one slot. Then the two-dimensional spread symbol is mapped onto one resource block. The three OFDM symbols in the middle of the resource block are reserved for PUCCH DM-RS for coherent detection, and the remaining OFDM symbols are for PUCCH data as shown in Fig. 2.32. The PUCCH DM-RS is also a two-dimensional sequence, which is a cyclic shift (CS) of one sequence with length 12 that is block-wise spread with an orthogonal sequence with length 3.

The PUCCH format 1/1a/1b from different UEs is multiplexed in one resource block by using different two-dimensional sequences. The PUCCH resource is indicated by a combination of CS index and orthogonal code covering (OCC) index.

The sequence in the frequency domain is a CS of one sequence. The number of available CS depends on the CS distance with three values, 1, 2, or 3, which is confgured by higher layer and dependent on the delay spread. The confgured CS distance is applied for both PUCCH data and PUCCH DM-RS. The CS varies with the symbol number and slot number in a cell-specifc manner to randomize the intercell interference. Considering the additional 4 OCC in time domain, the maximum number of available two-dimensional sequence for PUCCH data is 48, 24, 16 in principle. The OCC length for PUCCH DM-RS is 3, and therefore the maximum number of available DM-RS is 36, 18, 12 , respectively. Due to the requirement of coherent detection and the restriction of the number of PUCCH DM-RS, only three OCCs are needed for PUCCH data. The sequences of OCC = 4 and 3 are summarized in Table 2.15. For PUCCH data, OCC = 4 is applied for both slots within one sub-frame or OCC = 4 is applied for the frst slot and OCC = 3 is applied for the second slot in case the last OFDM symbol is reserved for SRS.

Fig. 2.32 PUCCH format 1/1a/1b

Table 2.15 Orthogonal sequences with OCC for PUCCH format 1/1a/1b

2.6.1.2 Pucch Format 2/2A/2B

PUCCH format 2 is to convey periodic CSI including RI, PMI, and CQI. The Reed– Muller (RM) code is used to encode the periodic CSI with up to 11 bits and the output of encoding is 20 bits [19]. QPSK modulation is performed to generate ten QPSK symbols. Each modulation symbol is multiplied with a cyclically shifted sequence with length 12 and it is mapped onto one symbol within one resource block. For the case of normal CP, there are two symbols reserved for DM-RS and the remaining fve symbols for PUCCH data in one resource block as shown in Fig. 2.33. In the case of extended CP, there is only one symbol for DM-RS located at the fourth symbol of the resource block.

The total ten QPSK symbols are mapped onto two resource blocks from two different slots within one sub-frame. Similar to PUCCH format 1/1a/1b, the CS of one base sequence is assigned to different UEs for multiplexing and the number of available CS depends on the confgured CS distance. Typically, it can support up to 6 UEs multiplexed together. The resource of PUCCH format 2 is semi-statically confgured via higher layer signaling.

Fig. 2.33 Transmission structure of PUCCH format 2/2a/2b in case of normal CP PUCCH format 2a/2b is used when periodic CSI and ACK/NACK feedback from a certain UE occurs at the same sub-frame, and it is only applicable in the case with normal CP. The ACK/NACK modulation symbol is multiplied by the second DM-RS sequence within one resource block; that is, the ACK/NACK information is conveyed by the relative phase between the two DM-RS symbols. PUCCH format 2a/2b corresponds to 1-bit or 2-bit ACK/NACK, respectively. In the case of extended CP, the periodic CSI and 1- or 2-bit ACK/NACK are concatenated and then encoded by the RM coding.

2.6.1.3 Pucch Format 3

In Release-10, carrier aggregation with up to fve component carriers was supported. As there is an independent HARQ entity per carrier, the number of reported ACK/NACK bits increased. For example, there are up to ten-bit ACK/NACK for FDD and even more for TDD depending on the DL-UL sub-frame confguration. PUCCH 1a/1b/2 is not able to convey so many ACK/NACK bits. PUCCH format 3 was introduced to support up to 11 bits for FDD corresponding to up to 10-bit ACK/ NACK and 1-bit positive/negative SR, and up to 21 bits for TDD corresponding to up to 20-bit ACK/NACK and 1-bit positive/negative SR.

PUCCH format 3 uses RM (32, O) as the channel coding scheme [19], where O is the number of ACK/NACK bits with the maximum value being 11. When the number of ACK/NACK/SR bits is less than or equal to 11 bits, the output of encoding is 32 bits, and then circular repetition is performed to obtain 48 bits corresponding to 24 QPSK symbols. Each modulation symbol is spread in time domain by a sequence with length 5 and mapped onto the 5 REs with the same sub-carrier index in one resource block. The mapping of 24 modulation symbols spans two resource blocks of two slots within one sub-frame. The DM-RS location is same as that of PUCCH format 2/2a/2b. As the maximum size of O is 11, in case the number of ACK/NACK/SR bits is larger than 11 bits, the ACK/NACK/SR bits are evenly divided into two blocks and the size of each block is no more than 11 bits. The same channel coding scheme using RM with (32, O) is separately performed for each block, i.e., dual-RM code [22], and the output is restricted to be 24 bits corresponding to 12 QPSK symbols. To achieve the frequency diversity gain, the modulation symbols from these two blocks are alternately mapped onto the REs in the frequency domain. This enables the total 12 QPSK symbols of each block to be distributed over the two PUCCH resource blocks. The transmission structure of PUCCH format 3 is demonstrated in Fig. 2.34.

2.6.1.4 Pucch Format 4/Format 5

In Release-13, massive carrier aggregation with up to 32 component carriers was

introduced. Normally, it is not possible for one operator to have so many licensed component carriers for aggregation. However, there is larger unlicensed spectrum bandwidth available for use below 6 GHz. Due to the larger number of carriers, the payload size of periodic CSI and/or ACK/NACK is further signifcantly increased, and therefore PUCCH format 4/5 was introduced as the container.

The transmission of the PUCCH format 4 is similar to the PUSCH. Up to eight resource blocks can be confgured for PUCCH format 4, and frequency hopping across two slots of one sub-frame is applied to reap the frequency diversity gain. Only QPSK modulation is supported for the robust transmission of the PUCCH. Similar to the PUSCH, DFT-based transform precoding is applied for the modulation symbols. The DM-RS transmission is also the same as that of PUSCH. Normal PUCCH format and shortened PUCCH format are defned corresponding to whether SRS is present at the last symbol of the second slot. For normal PUCCH format, there are 12 (normal CP) and 10 (extended CP) symbols for the PUCCH data transmission. For the shortened PUCCH format, there are 11 (normal CP) and 9 (extended CP) symbols for PUCCH data. The number of coded bits conveyed by PUCCH format 4 is NRB N PUCCHsymbol 4 4 PUCCH ∗ ∗ 12 ∗2, where NRB PUCCH 4 is the number of confgured resource blocks and Nsymbol PUCCH 4 is the number of symbols. Tail biting convolutional code (TBCC) with CRC is used as the channel code for PUCCH format 4.

PUCCH format 5 transmission is restricted to one resource block and with frequency hopping across the two slots within the one sub-frame. The operation of channel coding, modulation, and DFT-based transform precoding is similar to that of PUCCH format 4. The difference is that there is a block-wise spread operation before DFT-based transform precoding. A block of six modulation symbols is spread using [+1 + 1] or [+1 − 1] to obtain 12 modulation symbols and then the 12 modulation symbols are operated on by DFT-based transform precoding. The number of coded bits conveyed by PUCCH format 5 is 6 2 5 ∗ ∗ Nsymbol PUCCH , where NRB PUCCH5 is the number of symbols depending on normal or shortened PUCCH format. PUCCH format 5 is applicable for the medium payload size. Since two spread sequences are used, it can support up to two UEs multiplexed.

2.6.2 Pusch

The two transmission schemes for the PUSCH are:

- Single-Antenna Port Scheme:

• It is the transmission scheme in uplink TM1 and fallback transmission scheme in TM2.
• Closed-loop spatial multiplexing:
• Up to four-layer transmission is supported and it is the transmission scheme in TM2.

The general structure of PUSCH processing is shown in Fig. 2.35. For the singleantenna port scheme, there is only one transport block. In case of closed-loop spatial multiplexing, similar to the downlink, up to two transport blocks are supported. The number of layers (i.e., rank) and the precoding matrix used for the PUSCH transmission are determined by the eNB and then signaled to UE via the UL grant. The maximum number of layers supported is 4. The code word-to-layer mapping is the same as that of the PDSCH. The cubic metric preserving (CMP) codebook was introduced in order to preserve the single carrier property of PUSCH transmission on each antenna port. There is only one nonzero element in each row of the precoding matrix, which avoids the mix of the two signals on the same antenna port. In addition, antenna selection precoding matrices for 2 and 4 antenna ports are

Fig. 2.35 General structure of PUSCH processing 2 4G LTE Fundamental Air Interface Design supported. It may occur that part of transmit antennas are blocked due to hand grabbing for mobile phone, which causes signifcant loss in the radio signal. In this case, antenna selection is used to save transmission power. The details of CMP codebook for different ranks are beyond the scope of this treatise and can be found in [3].

PUSCH transmission is based on DM-RS ports, and the number of DM-RS ports is equal to the rank signaled in the UL grant. It is different from downlink that the precoding of the PUSCH transmission is not transparent. In uplink, the precoding matrix and the corresponding MCS signaled to the scheduled UE are determined by the eNB. The UE is mandated in the standards and is tested via test specifed by RAN 4 to use the signaled precoding matrix. MU-MIMO is also supported for the PUSCH transmission. That is, multiple UEs transmitting PUSCH data on the same time and frequency resources can be separated by eNB because of its multiple receive antenna capability. Uplink MU-MIMO is also known as virtual MIMO because the multiple transmit antennas are from different UEs. From the UE perspective, it does not know whether there are some other UEs multiplexed with it together. That is, MU-MIMO is transparent to UE. To separate the PUSCH from different UEs, uplink DM-RS is assigned to different UEs by using different cyclic shifts of one DM-RS sequence. As there are eight cyclic shifts available, it can, in principle, support up to eight UEs for MU-MIMO in principle. The assignment of cyclic shifts for different UEs is applicable for the case of aligned resource allocation among the MU-MIMO UEs. In Release-10, OCC over time domain for uplink DM-RS was introduced (see Sect. 2.4.2), which supports two UEs with unaligned resource allocation for MU-MIMO. In a practical system, especially for TDD systems, there are a large number of receive antennas, e.g., 8 or 16, and the uplink resource is limited; consequently there is the desire to support more UEs with unaligned resource allocation for MU-MIMO. In Release-14, uplink DM-RS in the form of comb as SRS is introduced to support up to four UEs with unaligned resource allocation.

2.6.3 Modulation

In order for the eNB to receive the PUSCH transmission, the modulation order and transport block sizes (TBS) need to be determined. The procedure of determining MCS and TBS for PUSCH is same as that for PDSCH. The combination of modulation order and TBS index is also indicated by a 5-bit MCS feld in the uplink grant. The 5-bit MCS table for PUSCH is originated from that of PDSCH with some changes. The change is that for the frst four MCS indices corresponding to 64QAM the modulation scheme is treated as 16QAM. The reason is that many UE categories do not support 64QAM or eNB does not support 64QAM transmission for PUSCH. In this case, the highest MCS corresponding to 16QAM is 2.41 bits/symbol which restricts the peak data rate, and therefore the related changes are done and the MCS can reach 3.01 bits/symbol [23].

2.7 Harq Timing

HARQ uses the stop-and-wait protocol for transmission. After a transport block transmission, the transmitter stops to wait until receiving the confrmation from the receiver. The receiver reports an indicator (i.e., ACK or NACK) to the transmitter based on the transport block detection result. If ACK is received, the transmitter will transmit a new transport block; otherwise, the erroneously detected transport block is retransmitted. Each stop-and-wait protocol transmission forms one HARQ process. The downlink HARQ operation is illustrated in Fig. 2.36.

Multiple parallel HARQ processes are supported to improve the overall system performance. The maximum number of HARQ process depends on the round-trip time (RTT) as shown in Fig. 2.36, where RTT is restricted by the eNB and UE processing time [1]. In LTE, it is assumed that the eNB processing time is 3 ms and UE processing time is 3 − TA ms in case of 1 ms TTI for PDSCH [24]. Based on that assumption, the maximum of HARQ process for FDD DL/UL was set to be 8. The maximum number of HARQ processes for TDD depends on DL/UL confguration and the eNB/UE processing capability [12, 25], which are summarized in Table 2.16.

There are two types of HARQ mechanisms defned as synchronous HARQ and asynchronous HARQ [8]: - Synchronous HARQ: the (re)transmissions for a certain HARQ process are restricted to occur at known time instants.

• Asynchronous HARQ: the (re)transmissions for a certain HARQ process may occur at any time, but the duration between the transmission and retransmission is larger than the RTT. For synchronous HARQ, the time instants for the (re)transmission for a certain HARQ process are predefned, and therefore the HARQ process number can be derived from the time instant directly.3 In this case, the retransmission can reuse the same transmission format as the initial transmission including the resource allocations and MCS without control signaling, which is known as nonadaptive HARQ. This has the beneft of controlling signaling overhead reduction and simple

0 4 7 1 7 4 2 10 2 3 9 6 4 12 3 5 15 2 6 6 1 scheduling. On the contrary, asynchronous HARQ requires the explicit signaling of HARQ process number, but has the advantage of scheduling fexibility for the retransmission(s). This is suitable for downlink transmission because it can fexibly schedule the retransmission to avoid the MBSFN sub-frames or in the sub-frames with paging or system information. In LTE, synchronous HARQ is used for uplink and asynchronous HARQ is used for downlink.

Soft combing schemes for the HARQ retransmission include chase combining and incremental redundancy (IR). Chase combining implies that the retransmission bits are exactly identical to the initially transmitted information bits. For IR, however, it is not necessary that the retransmission bits must be the same as those of the initial transmission. Instead, compared to the coded bits transmitted in the previous transmission (initial or retransmission), there are incremental coded bits transmitted in the next retransmission. In both cases soft combining after the retransmission will result in a lower code rate to increase the detection probability. The incremental coded bits in the retransmission are named as redundancy version (RV) which is

implicitly indicated by the MCS index in the DCI. In LTE, HARQ is based on IR. It is noted that CC is a special case of IR. 4 Regardless of synchronous or asynchronous HARQ, the HARQ timing in LTE is predefined; that is, upon the detection of the transport block at sub-frame n , the corresponding ACK/NACK is transmitted in sub-frame n + T HARQ as illustrated in Fig. 2.36 . For FDD, the HARQ timing THARQ for both DL and UL is 4, which is shown in Fig. 2.37 . For TDD, the HARQ timing depends on the DL/UL configuration and the location of the PDSCH/PUSCH. It can be seen that the HARQ ACK/ NACK corresponding to the PDSCHs transmitted in several downlink sub-frames is transmitted in one uplink sub-frame, but it requires that T HARQ ≥ 4, e.g., the HARQ timing for DL/UL configuration 2 in Fig. 2.37 .

2.8 Carrier Aggregation (Ca) And Band Combinations

For LTE Release-8, the maximum channel bandwidth of a single carrier is 20 MHz and the peak data rate can reach ~300 Mbps assuming 64QAM and four layers [9].

In order to fulfll the requirement for peak data rate of 1 Gbps [26], CA was introduced in Release-10 to support up to fve component carrier aggregation and the maximum channel bandwidth of each component carrier is 20 MHz. In Release-12, CA was enhanced to support the aggregation of FDD and TDD carrier, which is a kind of FDD and TDD convergence because the merits of both FDD and TDD can be jointly utilized. In Release-13, CA was further enhanced to support up to 32 component carrier aggregation, and in principle it can support up to 32 × 20 = 640MHz. However, it is diffcult for one operator to have 640 MHz licensed spectrum in the practical system, and the aggregated component carriers can also be on unlicensed spectrum. Based on whether the aggregated component carriers are within the same band or not, intra-band CA and inter-band CA are defned. The LTE bands and different CA bands are defned in [9]. CA is a very important feature for LTE and there are already 241 commercial networks in the world [27].

As mentioned, an important target of CA was to increase the peak data rate.

However, other important advantages of CA are scheduling gain and load balancing. There are many UE categories defned corresponding to different combinations of the number of component carriers, layers, and modulation order. The UE categories with the typical peak data rates are listed in Table 2.17 [9].

2.9 Initial Access And Mobility Procedures

Before a UE can operate on a LTE carrier, the UE needs to frst perform the initial access procedure to access the network and establish radio resource control (RRC) connection. The initial access procedure includes cell search, random access, and RRC connection establishment.

Cell search is the procedure by which a UE acquires the time (symbol/slot/frame) and frequency synchronization with a cell as well as physical layer cell identity (cell ID). There are 504 physical layer cell IDs supported and many layer 1 transmission parameters such as reference signal sequence generation and scrambling sequence are dependent on the physical layer cell ID. The 504 physical layer cell IDs are grouped into 168 unique physical layer cell ID groups and each group contains three IDs. The primary and secondary synchronization signals (PSS and SSS) are defned to facilitate the cell search. Since there are not any prior knowledge before the UE performs the cell search, the synchronization signals are always transmitted in the center six resource blocks of the system bandwidth, so that the UE always knows where it is. The PSS is transmitted in the last OFDM symbol of slots 0 and 10 for frame structure type 1 and the third OFDM symbol in slots 2 and 12 for frame structure type 2. The SSS is transmitted in the second last OFDM symbol of slots 0 and 10 for frame structure type 1 and the last OFDM symbol in slots 1 and 11 for frame structure type 2.

The PSS transmitted in the two slots of one radio frame is same. This transmission is used to acquire the symbol synchronization and part of the cell ID information. There are three different ZC root sequences with length 63 defned for PSS to represent the three unique IDs within one physical layer cell ID group. The middle element of the sequence is punctured when mapped onto the center six resource blocks due to the existence of direct current (DC) subcarrier. The mapping results in the symmetric property of PSS in time domain which can reduce the computation complexity at the UE [28]. Once the PSS is acquired, UE can detect SSS based on the time relationship between PSS and SSS. As the time relationship between PSS and SSS is different for frame structure types 1 and 2, the detection of SSS can also obtain the information of the frame structure and cyclic prefx. Due to the close distance between PSS and SSS symbol, the PSS may be used for coherent detection of SSS, which is up to UE implementation. The SSS is represented by two short sequences with length 31 and the 168 unique combinations of two short sequences represent the physical layer cell ID groups. For all the combinations of the two short sequences, the index of the frst short sequence is always less than that of the second short sequence, which can reduce the ambiguity of cell group ID detection [29]. The positions of two short sequences of SSS are switched across the two slots of one radio frame, which can be used to acquire the sub-frame synchronization.

After the cell search is done, UE needs to acquire the system broadcast information conveyed on PBCH and SIBs. The PBCH is detected to obtain the system bandwidth, number of CRS antenna ports, SFN, and PHICH duration described in Sect. 2.5.1. The SIBs are mapped to the PDSCH in physical layer, and the corresponding PDCCH with CRC scrambled by the system information radio network temporary identifer (SI-RNTI) is transmitted in the common search space for all the UE's detection. The PDCCH indicates the scheduled resources, MCS, etc. of SIBs. There is a periodicity to the transmission of SIBs with different periodicity. The SIB1 uses a fxed periodicity of 80 ms and there are four repetitions within 80 ms. The frst transmission of SIB1 is scheduled in sub-frame 5 of radio frames for which the SFN mod 8 = 0, and the repetitions are scheduled in sub-frame 5 of all other radio frames for which SFN mod 2 = 0 [11]. The SIB1 contains the information to determine whether a UE is allowed to access this cell and defnes the scheduling of other system information (SI). The SI message is used to convey one or more SIBs other than SIB1. All the SIBs included in one SI message are transmitted with the same periodicity, and one SIB can only be mapped to one SI message. The SIB2 is always mapped to the frst SI confgured in SIB1. The SIB2 contains radio resource confguration information that is common for all UEs, e.g., MBSFN subframe confguration and PRACH confguration information. The detail information of each SIB can be found in [11].

The UE needs to perform random access to achieve uplink synchronization for the PUSCH transmission and establish RRC connection after obtaining the system information related to random access including PRACH confguration, frequency position, root sequences, cyclic shift, set type (restricted or unrestricted), and so on. The transmission of PRACH occupies six resource blocks in a sub-frame or more consecutive sub-frames, which is scalable to different system bandwidths. Contention base random access procedure is used during initial access phase because the UE is in idle state, which has four steps as shown in Fig. 2.38 [30]. Step 1: Random Access Preamble Transmission: The UE frst selects one preamble sequence from the confgured preamble sequence set(s), and transmits it in the confgured frequency resource with the desired power.

Step 2: Random Access Response (RAR): Upon the successfully detected preamble sequences, the eNB will transmit the RAR in downlink in response to these transmitted preamble sequences. The RAR is transmitted in the scheduled PDSCH indicated by a PDCCH marked with random-access RNTI (RA-RNTI). The RAR conveys at least the identity of the detected random-access preamble, timing alignment information, and Random Access Response Grant (RARG) [12] and the assignment of Temporary cell RNTI (Temporary C-RNTI). If there are multiple random-access preambles detected simultaneously, the intended information in RAR can be contained in one PDSCH.

Step 3: PARG-based UL Transmission: After the reception of RAR, the UE would be synchronized in uplink using the signaled timing alignment information. However, the UE does not establish the connection with the cell yet. The PUSCH scheduled by RARG conveys some request information from higher layer, e.g., RRC Connection Request and NAS UE identifer [30]. The PUSCH is scrambled by the Temporary C-RNTI.

Step 4: Contention Resolution on Downlink: It may happen that multiple UEs transmit the same random-access preambles simultaneously and then receive the same RAR, which results in the collision of different UEs using the same Temporary C-RNTI. To solve the collision, the eNB transmits a downlink message in PDSCH scheduled by a PDCCH marked with Fig. 2.38 Contentionbased random access

procedure the Temporary C-RNTI in Step 2. The UEs will check whether the UE identity signaled in the downlink message is the same as the identity conveyed in Step 3. If it is matched, it implies that the random access is successful and the Temporary C-RNTI will be promoted to C-RNTI for the UE; otherwise, the UE needs to restart the random access procedure from Step 1.

In addition to the contention-based random access procedure, there is also noncontention-based random access procedure applicable to only handover and DL data arrival during RRC_CONNECTED requiring random access procedure [30].

There are two states defned for the UE which are RRC_IDLE and RRC_ CONNECTED (Fig. 2.39) [30]. In RRC_IDLE state, there are no RRC context and specifc cell attachment. The UE under this state does not have data to receive/transmit and can be in sleep mode for power saving. Once it is triggered to establish RRC connection, the random access procedure as described above is performed and it establishes the RRC context to move to RRC_CONNECTED state. In case of RRC_CONNECTED state, the normal communication between the UE and eNB can be performed under the RRC context known by both UE and eNB. If there is no data to transfer, the RRC connection can be released to go to RRC_IDLE state.

2.10 Summary

When LTE was initially designed, the considered aspects of the system are signifcantly simpler and more focused comparing to the recent situation. Relatively lowfrequency bands with FDD were the main consideration while TDD was largely an add-on feature by augmenting the FDD frame structure. System bandwidth is relatively narrow even though 20 MHz carrier bandwidth was deemed wide (in terms of implementation complexity and needed data rate) at that point of time. Deployment scenario is for homogeneous network topology of macro base stations. While data rate improvement was a priority for LTE, coverage and robustness of the network were certainly the essential factors. MIMO was introduced in LTE from the beginning as an advanced feature for improving spectrum effciency and peak data rate.

However, the number of antennas at the base station was limited to up to 4 and only SU-MIMO was supported in the frst release. OFDMA for downlink and DFT-SOFDM for uplink were selected for fexible multiplexing of channels and much easier support of MIMO transceiver schemes. As a result, the initial LTE design was that of a simple and rigid framework. In a nutshell, the basic air-interface design of LTE can be described as a single carrier with fxed bandwidth (with a small set of candidate values) of a single numerology and rigid frame structure, with fxed always-on CRS for all transmission schemes, and infexible control region and initial access channels.

Though a very-well-designed system, the simple and infexible framework of LTE soon became problematic after its frst 2–3 releases when new use cases and requirements need to be met. For MBB traffc, much higher system capacity demands pushed for wider system bandwidth (which drove the networks to higher frequency bands where TDD instead of FDD bands are more prominent), dense deployments of macro base station as well as small cells (where overhead and interference of always-on CRS transmissions became a signifcantly limiting factor to performance and compatibility issue to new feature introduction), and introduction of MU-MIMO and massive MIMO (which demands a DM-RS- and CSI-RS-centric design for transmission and CSI feedback). In addition to MBB, support of new applications such as machine-type communication, D2D/V2X, and URLLC in LTE requires substantial effort to work under its frame structure. Features such as carrier

aggregation, FD-MIMO, sidelink/V2X, NB-IoT/eMTC, small cell on/off, and HRLLC were introduced in later releases of LTE to address these issues.

All these are like lessons learned or exploration into new domains which point to elements that are needed when it is time to design a new air interface. Even without changing the fundamental waveform and similarity to basic frame structure, NR adopted a drastically different framework of multiple/scalable numerologies, fexible frame structure, DM-RS- and CSI-RS-centric reference signal design without any always-on signal, and mechanism to ensure forward compatibility. The next chapter will begin our investigation of the 5G NR air interface.

References

1. E. Dahlman, S. Parkvall and J. Skold, 4G: LTE/LTE-Advanced for Mobile Broadband, Amsterdam, The Netherlands: Elsevier Press, 2011.
2. S. Sesia, I. Toufk and M. Baker (eds), LTE: The UMTS Long Term Evolution, section edition, NY, NY: John Wiley and Sons Publishing, 2011.
3. 3GPP, "TS 36.211: Evolved Universal Terrestrial Radio Access (E-UTRA); Physical channels and modulation," June 2018. [Online]. Available: https://portal.3gpp.org/desktopmodules/ Specifcations/SpecifcationDetails.aspx?specifcationId=2425.
4. ITU-R, "Report M.2370: IMT Traffc estimates for the years 2020 to 2030," July 2015.

[Online]. Available: https://www.itu.int/pub/r-rep-m.2370.

5. R. V. Nee and R. Prasad, OFDM wireless multimedia communications, Boston, Massachusetts: Artech House Publisher, 2000.
6. C. C. Yin, T. Luo and G. X. Yue, Multi-carrier Broadband Wireless Communication Technology, Beijing, China: Beijing University of Posts and Telecommunications Press, 2004.
7. Motorola, "R1-060385: Cubic metric in 3GPP-LTE," 3GPP TSG-RAN WG1 Meeting #44, Denver, CO, USA, February, 2006. [Online]. Available: https://www.3gpp.org/ftp/tsg_ran/ WG1_RL1/TSGR1_44/Docs.
8. 3GPP, "TR 25.814, Physical layer aspects for evolved Universal Terrestrial Radio Access (UTRA) (Release 7)," September 2006. [Online]. Available: https://portal.3gpp.org/desktopmodules/Specifcations/SpecifcationDetails.aspx?specifcationId=1247.
9. 3GPP, "TS 36.104: Evolved Universal Terrestrial Radio Access (E-UTRA); Base Station (BS) radio transmission and reception," March 2010. [Online]. Available: https://portal.3gpp.org/ desktopmodules/Specifcations/SpecifcationDetails.aspx?specifcationId=2412.
10. 3GPP, "TR 25.913: Requirements for Evolved UTRA (E-UTRA) and Evolved UTRAN (E-UTRAN) (Release 8),"December 2008. [Online]. Available: https://portal.3gpp.org/desktopmodules/Specifcations/SpecifcationDetails.aspx?specifcationId=1342.
11. 3GPP, "TS 36.331: Radio Resource Control (RRC) Protocol Specifcation (Release 8)," September 2009. [Online]. Available: https://portal.3gpp.org/desktopmodules/Specifcations/ SpecifcationDetails.aspx?specifcationId=2440.
12. 3GPP, "TS 36.213: Evolved Universal Terrestrial Radio Access (E-UTRA); Physical Layer Procedures (Release 15)," June 2008. [Online]. Available: https://portal.3gpp.org/desktopmodules/Specifcations/SpecifcationDetails.aspx?specifcationId=2427.
13. 3GPP, "TS 36.214: Evolved Universal Terrestrial Radio Access (E-UTRA); Physical layer; Measurements (Release 15)," June 2018. [Online]. Available: https://portal.3gpp.org/desktopmodules/Specifcations/SpecifcationDetails.aspx?specifcationId=2428.
14. Samsung and Nokia Networks, "RP-141644: New SID Proposal: Study on Elevation Beamforming/Full-Dimension (FD) MIMO for LTE," 3GPP TSG-RAN #65, Edinburgh, UK, September, 2014. [Online]. Available: https://www.3gpp.org/ftp/tsg_ran/TSG_RAN/ TSGR_65/Docs/.
15. Samsung, "RP-160623: New WID Proposal: Enhancements on Full-Dimension (FD) MIMO for LTE," 3GPP TSG RAN Meeting#71, Göteborg, Sweden, March, 2016. [Online]. Available: https://www.3gpp.org/ftp/tsg_ran/TSG_RAN/TSGR_71/Docs/.
16. 3GPP, "TR 36.872: Small cell enhancements for E-UTRA and E-UTRAN, Physical layer aspects (Release 12)," December 2013. [Online]. Available: https://portal.3gpp.org/desktopmodules/Specifcations/SpecifcationDetails.aspx?specifcationId=2573.
17. D. C. Chu, "Polyphase Codes with Good Periodic Correlation Properties," IEEE Trans. On Information Theory, vol. 18, no. 4, pp. 531–532, 1972.
18. Panasonic and NTT DoCoMo, "R1-073626: Reference signal generation method for E-TTRA uplink," 3GPP TSG-RAN WG1 Meeting #50, Athens, Greece, August, 2007. [Online]. Available: https://www.3gpp.org/ftp/tsg_ran/WG1_RL1/TSGR1_50/Docs/.
19. 3GPP, "TS 36.212: Evolved Universal Terrestrial Radio Access (E-UTRA); Multiplexing and channel coding (Release 15)," June 2018. [Online]. Available: https://portal.3gpp.org/desktopmodules/Specifcations/SpecifcationDetails.aspx?specifcationId=2426.
20. Motorola, "R1-080072: TBS and MCS Table Generation and Signaling for E-UTRA," 3GPP TSG RAN1 Meeting#51bis, Seville, Spain, January, 2008. [Online]. Available: https:// www.3gpp.org/ftp/tsg_ran/WG1_RL1/TSGR1_51b/Docs/.
21. Ericsson, "R1-080556: Outcome of ad hoc discussions on TB size signaling," 3GPP TSG RAN1 Meeting#51bis, Seville, Spain, January, 2008. [Online]. Available: https://www.3gpp.

org/ftp/tsg_ran/WG1_RL1/TSGR1_51b/Docs/.

22. Huawei and HiSilicon, "R1-105247: A/N coding schemes for large payload using DFT-SOFDM," 3GPP TSG RAN1 #62bis, Xi'an, China, October, 2010. [Online]. Available: https:// www.3gpp.org/ftp/tsg_ran/WG1_RL1/TSGR1_62b/Docs.
23. Ericsson, Panasonic and Motorola, "R1-082091: MCS and TBS Tables for PUSCH," 3GPP TSG RAN1 #53, Kansas City, MO, USA, May, 2008. [Online]. Available: https://www.3gpp. org/ftp/tsg_ran/WG1_RL1/TSGR1_55/Docs.
24. Ericsson et al., "R1-073871: Maximum number of hybrid ARQ processes," 3GPP TSG RAN1 #50, Athens, Greece, August, 2007. [Online]. Available: https://www.3gpp.org/ftp/tsg_ran/ WG1_RL1/TSGR1_50/Docs.
25. Huawei et al., "R1-081124, Way forward for TDD HARQ process," 3GPP TSG RAN1 #52, Sorrento, Italy, February, 2008. [Online]. Available: https://www.3gpp.org/ftp/tsg_ran/ WG1_RL1/TSGR1_52/Docs.
26. 3GPP, "TR 36.913: Requirements for further advancements for Evolved Universal Terrestrial Radio Access (E-UTRA) (LTE-Advanced) (Release 8)," March 2009. [Online]. Available: https://portal.3gpp.org/desktopmodules/Specifcations/SpecifcationDetails.aspx?specifcati onId=2585.
27. GSA, "Evolution from LTE to 5G," April 2018. [Online]. Available: https://gsacom.com/ paper/evolution-lte-5g/.
28. Huawei, "R1-072321: P-SCH sequences," 3GPP TSG RAN1 #49, Kobe, Japan, May, 2007.

[Online]. Available: https://www.3gpp.org/ftp/tsg_ran/WG1_RL1/TSGR1_49/Docs.

29. Texas Instruments et al., "R1-074143: Way Forward for Secondary SCH Mapping and Scrambling," 3GPP TSG RAN1 #50, Shanghai, China, October, 2007. [Online]. Available: https://www.3gpp.org/ftp/tsg_ran/WG1_RL1/TSGR1_50/Docs.
30. 3GPP, "TS 36.300: Evolved Universal Terrestrial Radio Access (E-UTRA) and Evolved Universal Terrestrial Radio Access Network (E-UTRAN); Overall description; Stage 2 (Release 15)," June, 2007. [Online]. Available: https://portal.3gpp.org/desktopmodules/ Specifcations/SpecifcationDetails.aspx?specifcationId=2430.
31. S. Wicker, Error Control Systems for Digital Communication and Storage, Englewood Cliffs: Prentice Hall, 1995.

Chapter 3

5G Fundamental Air Interface Design

Now that we have a good grounding in the fundamental of LTE design, it is now natural to understand the new 5G air interface by comparing and contrasting the designs between the two systems. We will begin of discourse with the NR-specifc designs for the carrier, frame structure, physical channels, and reference signals. Then the global candidate spectrum for 5G-NR is introduced which demonstrates the new characteristic with wider frequency range and bandwidth impacting NR design. With respect to C-band as NR deployment typical spectrum, the coverage issue is identifed and the mechanism of UL/DL decoupling scheme (i.e., LTE/NR spectrum sharing) to solve the coverage issue is mainly introduced. Afterwards, NR physical layer technologies including waveform, polar/LDPC codes, MIMO, and mMTC are described, and in all cases the enhancements of NR made relative to LTE are made apparent.

3.1 5G-Nr Design Of Carrier And Channels 3.1.1 Numerology For The Carrier

NR spans much wider frequency range compared to LTE, which currently defnes two frequency ranges (FR) as FR1 and FR2. The corresponding frequency ranges for FR1 and FR2 are 450–6000 MHz and 24,250–52,600 MHz, respectively, and a set of operating bands are defned for FR1 and FR2 [1]. For the spectrum beyond 3 GHz, there is larger spectrum bandwidth available; see Sect. 3.2.1, which can be utilized to fulfll the high data rate requirement of IMT-2020 [2]. For each operating band, a UE or base station can support a number of carriers which is dependent on the bandwidth of a carrier and the UE capability. The carrier bandwidth is related to base station and UE processing capability. The supported carrier bandwidths for FR1 and FR2 are summarized in Table 3.1. However, from UE perspective, the

Table 3.1 Carrier bandwidth

supported in NR Frequency range

Carrier bandwidth

(MHz)a

FR1 5, 10, 15, 20, 25,

30, 40, 50, 60, 70, 80, 90, 100

FR2 50, 100, 200, 400

aCarrier bandwidth is dependent on the subcarrier spacing and operating band [1]

supported transmission bandwidth (i.e., UE channel bandwidth) may be smaller than the carrier bandwidth due to the restriction of UE capability. In this case, the network can confgure a part of contiguous spectrum being equal to or smaller than UE channel bandwidth from the carrier for the UE, which is also called as bandwidth part (BWP). A UE can be confgured with up to four BWPs in the downlink and uplink, but there is only one BWP being active at a given time. The UE is not expected to receive or transmit outside an active BWP, which is benefcial for UE power saving because it does not have to transmit or receive on the entire system bandwidth.

Since CP-OFDM-based waveform is applied for both downlink and uplink transmission, see Sect. 3.4.1; the design of numerology for a carrier is similar to LTE that includes subcarrier spacing (SCS) and CP. The key factor to determine the SCS is the impact of Doppler shift which is related to the carrier frequency and mobility. The frequency range of LTE operating bands is within that of NR. The 15 kHz SCS design has been well verifed in the practical networks. Hence 15 kHz SCS is supported in NR as well. Considering the mobility requirement supporting up to 500 km/h and the wider frequency range, having only 15 kHz SCS is not enough and multiple larger SCS values with 2μ × 15 kHz, where μ = 0, 1, 2, 3, 4 were introduced. Larger SCS results in the shorter time duration of OFDM symbol which is helpful for short TTI transmission especially for the latency-sensitive service, e.g., remote control. Such design of SCS also has the beneft of reducing implementation complexity in terms of the number of subcarriers for a certain carrier bandwidth. The utilization of SCS is dependent on the frequency range. Of course, it is not necessary to support all the SCS values for each and every frequency range.

The CP length was determined to well mitigate the impact of the delay spread and have a reasonable overhead. In the case of 15 kHz SCS of LTE, the ratio of CP length over the duration of one OFDM symbol is 144/2048 = 7.03% for the OFDM symbols other than the frst one in each slot. The CP length for the frst OFDM symbol in each slot is slightly longer as 160/2048 = 7.81% due to the restriction of 0.5 ms slot and to help with the settling of the automatic gain adjustment. This CP length has been shown to be a good trade-off between the mitigation of delay spread and the CP overhead, and is, therefore, reused for NR in the case of 15 kHz SCS and normal CP. As the duration of one OFDM symbol is equal to the reciprocal of SCS, the CP length for other SCS will be proportionally reduced by 2μ, μ = 0, 1, 2, 3, 4 which keeps the same ratio of CP overhead as 15 kHz. This can also enable the OFDM symbols of different SCS within one slot to align, which is benefcial for the coexistence of carriers with different SCS especially for the TDD networks because of the need of synchronization. It is noted that the CP length of the 1st and 7 ∗ 2μth OFDM symbols within one sub-frame is slightly longer, and the detail of frame structure can be found in Sect. 3.1.2. Extended CP is only supported for 60 kHz SCS, and the length of extended CP is scaled from that of LTE. The 60 kHz SCS is envisioned to be applicable for URLLC service because of the resulting shorter TTI. When the URLLC service using 60 kHz SCS is deployed in sub-3 GHz, the normal CP may not be able to mitigate the inter-symbol interference in case of highdelay-spread scenario, and the extended CP is useful in this scenario.

Due to the proportional reduction of CP for larger SCS, it is clear that the CP will not be proper for scenarios with large delay spread. The larger SCS is, thus, typically utilized for high frequency which has smaller delay spread [3]. There is the frequency range restriction for different SCS as shown in Table 3.2. The SCS and CP for one BWP can be obtained from the higher layer parameters subcarrierSpacing and cyclicPrefx, respectively. In one carrier, multiple numerologies can be confgured.

Given a certain SCS, a resource block is defned as 12 consecutive subcarriers in the frequency domain. It is different from LTE which always has 7 or 6 OFDM symbols within one resource block. Here there is no restriction on the number of symbols within one resource block to facilitate the multiple short TTI transmission per SCS. The smallest resource unit within the resource block is called resource element (RE) indicated by (k, l), where k is the subcarrier index in the frequency domain and l represents the OFDM symbol index relative to a starting position in the time domain.

Two kinds of resource blocks are defned: common resource block (CRB) and physical resource block (PRB). The CRB is defned from the system perspective, where the CRBs are numbered from 0 and upwards in the frequency domain for subcarrier spacing confguration μ, i.e., n k CRB / 12 , where k is the subcarrier index relative to a frequency starting position (i.e., reference point) of a carrier

common for all SCS. The PRB is defned within one BWP which is a set of contiguous common resource blocks. The relation between the CRB and RRB is n n CRB PRB NBWP i start , , where N i start BWP, is the starting position of BWP i, i = 0, 1, 2, 3 expressed by CRB.

The actual number of resource blocks per carrier depends on the carrier bandwidth, SCS, and frequency band [1].

3.1.2 Frame Structure

One radio frame with 10 ms duration consists of 10 sub-frames and each sub-frame is of 1 ms duration. Each radio frame is divided into two equally sized half-frames of fve sub-frames, which are half-frame 0 consisting of sub-frame 0–4 and halfframe 1 consisting of sub-frames 5–9. The number of slots within one sub-frame is 2μ, μ = 0, 1, 2, 3, 4 which is dependent on SCS. In one slot, there are always 14 OFDM symbols regardless of the SCS confguration in the case with normal CP. The relationship among the number of OFDM symbol, slot, sub-frame, and radio frame is demonstrated in Fig. 3.1.

NR has very fexible frame structure compared to LTE, which has only two frame structures which are for FDD and TDD, respectively. The slots in one radio frame can be fexibly confgured for downlink or uplink transmission. The 14 OFDM symbols in one slot can be classifed as "downlink," "fexible," or "uplink." A slot format includes downlink symbols, uplink symbols, and fexible symbols. The downlink and uplink symbols can only be used for downlink and uplink transmission, respectively; however the fexible symbols can be used for downlink transmission, uplink transmission, GP, or reserved resources. The slots can be classifed into fve

Fig. 3.2 Slot Types

3.1.2.1 Cell-Specifc Higher Layer Confguration

different types: as downlink only, uplink only, downlink dominant, uplink dominant, and full fexible (Fig. 3.2). For type 3 or type 4, the downlink PDSCH scheduling and the corresponding ACK/NACK or the UL grant and PUSCH transmission are contained in one slot, which can be considered as a type of self-contained transmission to reduce latency. All the OFDM symbols in type 5 slot are fexible and can be reserved for future to achieve the forward compatibility.

The fexible frame structure is benefcial to adapt the service traffc in downlink and uplink, which can be found in Sect. 3.2.1. The signaling of slot formats is needed for UE to obtain the frame structure that includes cell-specifc higher layer confguration, UE-specifc higher layer confguration, UE-group DCI, and UE-specifc DCI as below: The cell-specifc higher layer parameter TDD-UL-DL-ConfgurationCommon is used by the UE to set the slot format per slot over a number of slots indicated by the parameter. The higher layer parameter includes [4]: {

• Reference SCS - Pattern 1
• DL/UL transmission periodicity (P1) - Number of downlink slots (X1): Number of consecutive full DL slots at the beginning of each DL-UL pattern
• Number of downlink OFDM symbols (x1): Number of consecutive DL symbols in the beginning of the slot following the last full DL slot
• Number of uplink slots (Y1): Number of consecutive full UL slots at the end of each DL-UL pattern
• Number of uplink OFDM symbols (y1): Number of consecutive UL symbols at the end of the slot preceding the frst full UL slot } Additionally, a second pattern, i.e., pattern 2, can be confgured in the TDD-ULDL-ConfgurationCommon, which has a different confguration; for example, the second pattern has more uplink slot confguration for uplink capacity and coverage.

In this case, it is known as dual-periodicity slot confguration (Fig. 3.3). The confguration of the second pattern is optional. It is noted that the remaining slots and symbols not addressed by TDD-UL-DL-ConfgurationCommon during the DL/UL transmission period are default to be "fexible" and can be further confgured by other signaling.

3.1.2.2 Ue-Specifc Higher Layer Confguration

The UE-specifc higher layer parameter TDD-UL-DL-ConfgDedicated can be additionally confgured to override only the fexible symbols provided by TDD-UL-DLConfgurationCommon. The TDD-UL-DL-ConfgDedicated provides a set of slot

confgurations, where each slot confguration includes the slot index and the symbol confguration for the indicated slot. The symbols in the indicated slot can be confgured for all downlink, all uplink, or a number of downlink frst symbols in the slot and a number of uplink last symbols in the slot [5].

3.1.2.3 Group Common Pdcch

In case the symbols in a set of consecutive slots in time domain indicated to a UE as fexible by higher layer parameters TDD-UL-DL-ConfgurationCommon and TDDUL-DL-ConfgDedicated, the layer 1 signaling DCI format 2_0 can be used to provide the slot format(s) for the set of consecutive slots. There are a total number of 56 slot formats defned in the specifcation [5].

DCI format 2_0 is a group common PDCCH which can be detected by a group of UEs. Whether a UE needs to detect DCI format 2_0 is confgured by the network. When a UE is confgured by higher layer signaling with parameter SlotFormatIndicator, the UE is provided with a slot format indicator (SFI-RNTI) and the payload size of DCI format 2_0. In addition, the location of SFI-index feld for a cell in DCI format 2_0 is provided. The SFI-index represents a slot format combination. The mapping between a list of slot format combinations and the corresponding SFI-index is confgured by the parameter SlotFormatIndicator. The slot format combination refers to the one or more slot formats that occur in the set of consecutive slots in time domain.

3.1.2.4 Dl/Ul Dynamic Scheduling

If an SFI-index feld value in DCI format 2_0 indicates the set of symbols of the slot as fexible, the set of symbols can be dynamically scheduled for downlink or uplink transmission, which depends on the received DCI format. When the UE receives a DCI format (e.g., DCI format 1_0, 1_1, 0_1 [6]) for scheduling downlink transmission PDSCH or CSI-RS in the set of symbols of the slot, the set of symbols will be used for downlink transmission. In case the UE receives a DCI format (e.g., DCI format 0_0, 0_1 [6]) for scheduling PUSCH, PUCCH, or SRS transmission in the set of symbols, it will be used for uplink transmission. If there is not any scheduling information for the set of symbols, the UE does not transmit or receive in these resources.

In summary, there are four levels of signaling to confgure the frame structure which provides much fexibility as shown in Fig. 3.4. However, considering the potential complexity and the cost of cross-link interference mitigation in the practical system, it may be enough to use only a few frame structure. Regarding the signaling, it is not mandatory for the UE to receive all the four levels of signaling. The UE can receive one or more of these signaling to obtain the frame structure.

1. Cell-specific higher layer configuration

3.1.3 Physical Layer Channels

After the initial access procedure (see Sect. 4.1.1 ), gNB starts to perform normal communication with the UE. The information to be transmitted in both the downlink and uplink is conveyed on different physical channels. The physical channels correspond to a set of resource elements carrying the information originating from higher layers. According to the transmission direction and the characteristics of the carried information of physical channel, a set of downlink and uplink physical channls are defined [ 7 ]. In this section, the functions of different physical channels in the DL and UL transmission are introduced. The physical random-access channel does not convey the information from higher layer, and transmits the preamble to build RRC connection, uplink synchronization, etc., which is discussed in Sect. 4.1.1.

3.1.3.1 Physical Broadcast Channel (Pbch)

The PBCH is detected by UE and is used to acquire the essential system information during the initial access described in the section of IAM. The system information from higher layer is divided into MIB (Master Information Block) and a number of SIB (System Information Block). The MIB is always carried on the PBCH with 80 ms periodicity and makes repetitions within 80 ms. The MIB includes the following information [ 4 ]: System Frame Number (SFN): the 6 most significant bit (MSB) of the 10-bit SFN Subcarrier spacing for SIB1, Msg.2/4 for initial access and broadcast SI messages

• SSB Subcarrier Offset: the frequency domain offset between SSB and the overall resource block grid in a number of subcarriers
• The time position of the frst DM-RS for the downlink or uplink - The resource confguration for SIB1 detection, e.g., a common control resource set, a common search space, and necessary PDCCH parameters for SIB1
• The cell is barred or not
• Intra-frequency reselection is allowed or not In addition to the information from the MIB transmitted in the PBCH, there are additionally 8 bits of information from physical layer as 4 least signifcant bit (LSB) of 10-bit SFN, 1 bit half-frame indication, and 3 MSB of SS/PBCH block index in case with 64 candidate SS/PBCH blocks (otherwise there is one bit as MSB of PRB offset and two reserved bits). The total payload size of PBCH is 56 bits including 24-bit CRC, which is encoded to 864 bits with very low coding rate to guarantee the reliability of PBCH detection.

The PBCH is transmitted with PSS and SSS together as the SS/PBCH block (SSB). The same subcarrier spacing and cyclic prefx are applied for PSS, SSS, and PBCH. An SSB consists of 4 consecutive OFDM symbols in the time domain and 240 consecutive subcarriers forming 20 SSB resource blocks in the frequency domain (Fig. 3.5). For each SSB, the PSS and SSS are located at the frst and the third OFDM symbols, respectively, and the length of PSS and SSS sequence is 127 symbols mapped onto the center resource elements of each located RB; PBCH and the associated DM-RS are mapped onto at the second and the fourth OFDM symbols and partial of the third OFDM symbol. There is one DM-RS resource element, with every four resource elements giving a PBCH DM-RS density of 1/4. For the third OFDM symbol within one SSB, PBCH and the associated DM-RS are mapped onto the 48 resource elements of both sides.

In NR, beam-based access mechanism was adopted for cell coverage. Multiple SSBs covering different beam directions are used for performing beam sweeping. More SSBs are needed in the high-frequency scenario. The maximum number of candidate SSBs within a half frame is 4/8/64 corresponding to the sub-3 GHz, 3–6 GHz, and above 6 GHz scenario, respectively. The transmission time instances of candidate SSBs, i.e., SSB pattern, within one half frame are predefned which is dependent on the subcarrier spacing and the applicable frequency range [5]. As SSB is used to determine the sub-frame timing of the associated PRACH resources, etc., the candidate SSBs need to be identifed. The candidate SSBs in a half frame are indexed from 0 to L − 1, where L = 4, 8, 64. The index of candidate SSB is implicitly associated with its corresponding PBCH DM-RS sequence when L = 4 and 8. In case L = 64, if there are 64 DM-RS sequences to associate with the SSB index, it will signifcantly increase the SSB blind detection complexity. To help with the blind detection, the index of candidate SSB is determined by obtaining the three MSB from the PBCH and three LSB from the PBCH DM-RS sequences. Once the UE successfully detects one SSB, the UE will obtain the essential system information from PBCH, frame timing according to the SFN, and index of detected SSB. For the initial-access UE, the UE can assume that the periodicity of SSBs in a half frame is 20 ms.

3

The channel raster and synchronization raster in NR are decoupled in order to have a sparse synchronization raster for reducing search complexity. In this case, the SB resource block may not be aligned with the common resource block, and there is a subcarrier offset between them. The subcarrier offset is from subcarrier 0 in the common resource block with the subcarrier spacing provided by MIB to subcarrier 0 of the SSB, where this common resource block overlaps with the subcarrier 0 of the first resource block of SSB as shown in Fig. 3.6. As the system bandwidth is not signaled in PBCH as in LTE, the UE cannot obtain the frequency position of common resource blocks according to the system bandwidth. Hence, a subcarrier offset is signaled in the PBCH for the UE to determine the frequency location of common reference point A and the common resource block arrangement. This is different from LTE in which PSS/SSS/PBCH is always located at the center of the system bandwidth, where the channel and synchronization have the same raster.

3.1.3.2 Physical Shared Data Channel (Pdsch)

There is only one transmission scheme defned for the PDSCH transmission, which is based on DM-RS similar to TM9 in LTE. From the UE perspective, up to eightlayer transmission is supported. There is no open-loop transmit diversity scheme as SFBC used in LTE defned for PDSCH transmission, which means that NR could be less robust in certain environments (such as high speeds). Hence, link adaptation in NR plays a crucial role in the network's choice of transmission parameters such as spatial precoder, frequency domain granularity of spatial precoder, and time-domain granularity of spatial precoder.

The maximum number of code word per UE for the PDSCH transmission is 2 and there is a one-to-one mapping between transport block and code word. Similar to LTE, there are the corresponding MCS, HARQ-ACK, and CQI for each code word. However, the code word to layer mapping is slightly different from LTE discussed in Sect. 3.4.3.1.1. There is one code word if the number of transmission layer is less than or equal to 4; otherwise the number of code word is 2. Such code word to layer mapping has the beneft of reducing the signaling overhead of the MCS, HARQ, and CQI when the rank is no more than 4, but it cannot enable to use the successive interference cancellation at the UE side in case of single code word transmission. The relationship between the number of code word and layer is a trade-off of performance and signaling overhead. The general transmission structure of NR PDSCH is shown in Fig. 3.7. As NR only supports DM-RS-based

transmission scheme, the precoding operation is transparent and the data of each layer is directly mapped onto the DM-RS ports.

The channel coding scheme used for PDSCH transmission is LDPC; see Sect.

3.4.2.1. The scrambling and modulation operation is similar to LTE, and the modulation schemes for PDSCH are QPSK, 16QAM, 64QAM, and 256QAM.

In NR, since low latency is a very important requirement [2, 8], shorter TTI is a key feature supported to fulfll the requirement. To enable the shorter TTI, the resource allocation for the PDSCH is more fexible especially in the time domain. The number of OFDM symbols allocated for PDSCH can be {3, 4,…,14} or {2,4,7} which is dependent on PDSCH mapping type. PDSCH mapping type A is used for starting the PDSCH in the frst three symbols of a slot with a duration of three symbols or more until the end of a slot. PDSCH mapping type B is used for starting the PDSCH anywhere in the slot, with a duration of 2, 4, or 7 OFDM symbols. In addition, the slot aggregation in time domain for PDSCH transmission is supported to improve the coverage. In this case, the same symbol allocation is used across a number of consecutive slots. The number of aggregated slots can be 2, 4, or 8, which is confgured by higher signaling, but there is only one downlink control channel for the slot aggregation to reduce the signaling overhead. The frequency domain resource allocation supports two types, and it is relative to a bandwidth part. Type 0 allocates resources with a bitmap of resource block groups (RBG), where each RBG is a set of consecutive virtual resource blocks. Type 1 allocates resources as a set of contiguous non-interleaved or interleaved virtual resource blocks. The network can also indicate to the UE how many physical resource blocks (PRB) are bundled with the same precoder, which constitutes precoding resource block groups (PRG) (see Sect. 3.4.3.1.2).

On the downlink, the UE receives a DL DCI over PDCCH (see Sect. 3.1.3.3).

DCI format 1_1 offers the most scheduling fexibility, while DCI format 1_0 is more robust and can be used for fallback. A DCI scheduling PDSCH can also indicate a rate-matching pattern that allows reserving resource elements where PDSCH and DM-RS cannot be mapped. Such resource reservation signaling allows for example rate matching around entire OFDM symbols or PRBs, or rate matching around LTE common reference signals (CRS), or rate matching around ZP CSI-RS resources. This signaling also allows dynamic mapping of the PDSCH in REs that are semi-statically allocated for PDCCH, when the network decided not to use those resources for PDCCH in a certain slot.

3.1.3.3 Physical Downlink Control Channel (Pdcch)

The PDCCH carries downlink control information (DCI) for PDSCH scheduling, PUSCH scheduling, or some group control information, e.g., power control information for PUSCH/PUCCH/SRS and slot format confguration. The defned set of DCI formats are shown in Table 3.3 [6].

Similar to LTE, to minimize the PDSCH decoding latency, the PDCCH is normally located at the beginning 1/2/3 OFDM symbols of a slot in time domain.

However PDCCH does not span the entire carrier bandwidth in the frequency domain as in LTE. The rationale is that the UE channel bandwidth may be smaller than carrier bandwidth, as well as the resource granularity of the PDCCH spanning the entire carrier bandwidth is rough which could result in increasing the resource overhead especially for larger bandwidth, e.g., 100 MHz. Hence, a number of resource blocks in the frequency domain are confgured by a higher layer for the PDCCH. The multiplexing of the PDCCH and PDSCH in one slot is TDM-like but not pure TDM. In NR, the PDSCH resource mapping is rate matched around the control resource set(s) when the PDSCH is overlapped with the confgured control resource sets.

The resource unit assigned for PDCCH is known as a control resource set (CORESET). A control resource set consists of NRB CORESET resource blocks in the frequency domain and Nsymb CORESET symbols in the time domain, where the resource blocks are confgured by a bitmap. These two parameters are confgured by the higher layer parameter ControlResourceSet IE [4]. The assigned resource blocks are in the form of a number of resource block group (RBG) consisting of six consecutive resource blocks each. Up to three control resource sets can be confgured for one UE to reduce the PDCCH blocking probability.

Given the confgured PDCCH resources, the PDCCH is mapped onto these resources and transmitted. A PDCCH is formed by aggregating a number of control channel elements (CCEs), which depends on the aggregation level of that particular PDCCH. The aggregation level may be 1, 2, 4, 8, or 16. One CCE consists of six resource element groups (REGs) where a REG equals one resource block during one OFDM symbol. There is one resource element for PDCCH DM-RS every 4 resource elements in one REG, and therefore the number of available resource elements of one CCE is 48. The REGs within a control resource set, are numbered in increasing order in a time-frst manner, starting with 0 for the frst OFDM symbol and the lowest-numbered resource block in the control resource set as demonstrated in Fig. 3.8.

The PDCCH transmission is also based on DM-RS as PDSCH, and the transparent precoding cycling can be applied to achieve transmit diversity gain. More precoders for one PDCCH are benefcial for performance from diversity perspective, but it impacts the channel estimation performance because channel interpolation cannot be performed across the resources using different precoders. To balance the precoding gain and channel estimation performance, the concept of REG bundle was introduced. An REG bundle consists of L consecutive REGs, and the same precoding is used within a REG bundle. One CCE consists of 6/L REG bundles. The two modes of CCE to REG mapping in one control resource set are non-interleaving and interleaving. The non-interleaving mapping is such that the REG bundles of one CCE are consecutive and located together. For the interleaving mode, the REG bundles of one CCE are interleaved and distributed to achieve frequency diversity gain and randomize the inter-cell interference [7]. The REG bundle size depends on the CCE to REG mapping mode, which is summarized in Table 3.4. The bundle size enables the REGs with the same resource block index spanning over different OFDM symbols to belong to one REG bundle as illustrated in Fig. 3.9. The bundle size in the case of interleaving mode is confgurable.

At the receiver side, PDCCH is blindly detected by UE. To reduce the blind detection complexity, the number of candidate PDCCHs for blind detection is restricted while reducing the PDCCH blocking probability. A set of PDCCH candidates for a UE to monitor is defned as the PDCCH search space set. A search space set can be a common search space set or UE-specifc search space set. The common search space normally conveys some common control information for many UEs, e.g., SIB, paging, RACH response, slot format confguration, and TPC command.

Hence, the supported aggregation levels in the common search space are high as 4/8/16 to guarantee the performance. DCI formats 0_0, 1_0, 2_0, 2_1, 2_2, and 2_3 can be signaled in the common search space. DCI formats 0_1 and 1_1 for UE-specifc scheduling information are signaled in the UE-specifc search space. The aggregation levels of UE-specifc search space are 1/2/4/8/16.

The PDCCH blind detection procedure is similar to LTE, but the difference is some related blind detection parameters are confgured by the network for fexibility. The number of search space sets for UE to monitor is confgured by higher layer signaling, and each search space set is associated with one control resource set. In addition, the information related to each confgured search space set is also signaled, e.g., common or UE-specifc search space set and the number of PDCCH candidates for each aggregation level [9].

3.1.3.4 Physical Uplink Shared Data Channel (Pusch)

For PUSCH, both DFT-S-OFDM- and CP-OFDM-based transmissions are supported (see Sect. 3.4.1). Two transmission schemes are supported for the PUSCH: codebook-based transmission and non-codebook-based transmission. The difference of these two schemes is whether the precoding operation is transparent or not.

For codebook-based transmission scheme, the UE is confgured to use one or more SRS resources for SRS transmission. More than one SRS resource is better for The PUCCH conveys uplink control information (UCI) including scheduling request (SR), HARQ-ACK, and periodic CSI. Based on the control content and payload size of PUCCH, multiple PUCCH formats are supported as shown in Table 3.5.

The PUCCH has a short-duration transmission with 1 or 2 OFDM symbols and a long-duration transmission with 4–14 OFDM symbols within one slot. Within one slot, intra-slot frequency hopping can be confgured to take advantage of the frequency diversity gain in case the number of OFDM symbols is more than 1. If it is confgured, the number of symbols in the frst hop is the foor of half of the length of OFDM symbols. The support of duration less than 4 symbols PUCCH is for lowlatency transmission. The number of symbols for PUCCH is confgured by higher layer signaling based on the requirement of both latency and coverage. The PUCCH formats with one or two symbols are normally applied in the DL dominant slot, which is transmitted in the last one or two symbols of one slot. Other PUCCH formats can be used in the UL dominant slot.

3 5G Fundamental Air Interface Design higher frequency scenario because beam management is easier. Based on the transmitted SRS, the network selects the preferred SRS resource, transmission rank, and precoder corresponding to the preferred SRS resource. Then the network signals them in the uplink grant DCI format 0_1. The signaled information is used by the UE for PUSCH precoding. The codebooks used for DFT-S-OFDM and CP-OFDM are defned in [7]. In the case of DFT-S-OFDM, only rank 1 codebook was defned for both two and four antenna ports. The reason is that DFT-S-OFDM is normally used for the coverage-limited scenario due to low cubic metric. In this scenario, rank 1 transmission is the typical case. For non-codebook-based transmission scheme, the UE frst determines a number of precoders based on the measurement from CSI-RS in downlink. Then a number of SRS is precoded by these precoders and transmitted by the UE. The network selects the preferred precoded SRS. After the selection, the network signals the index of the preferred precoded SRS to UE. The UE uses the precoder corresponding to the signaled precoded SRS for the PUSCH transmission, but the precoding is transparent. The non-codebook-based transmission scheme is more applicable for TDD operation with channel reciprocity.

PUSCH transmission is based on DM-RS and up to four-layer transmission for one UE is supported. The number of transmission layer is determined by the network regardless of codebook- or non-codebook-based transmission scheme. The same code word to layer mapping as downlink is used, and, therefore, there is only one code word for PUSCH. However, from the network perspective, it can support up to 12-layer transmission in the form of MU-MIMO because of DM-RS capacity. In NR, downlink and uplink use the same DM-RS structure in order to keep the orthogonality of downlink and uplink reference signal, which can ease the crosslink interference mitigation in the case of fexible duplex.

3.1.3.5 Physical Uplink Control Channel (Pucch)

3.1.3.5.1 Pucch Format 0

The transmission of PUCCH format 0 is constrained to one resource block in the frequency domain. The 1- or 2-bit information on PUCCH format 0 is transmitted by performing sequence selection. The used sequences are the cyclic shifts of one low PAPR base sequence [7]. One cyclic shift is selected from a set of 2 or 4 cyclic shifts according to the 1- or 2-bit information to be transmitted. The selected cyclic shift is used to generate PUCCH sequence and then transmitted. For SR, a positive SR uses a predefned cyclic shift and the sequence corresponding to the predefned cyclic shift is transmitted, and nothing is transmitted for negative SR. In case there is simultaneous transmission of HARQ-ACK and positive SR, a different set of 2 or 4 cyclic shift values are used for selection. It should be noted that for PUCCH format 0 there is no DM-RS sequence for coherent detection. The network can detect the received sequence to obtain the information.

3.1.3.5.2 Pucch Format 1

The one or two information bits are modulated as one BPSK or QPSK symbol, respectively. The modulation symbol is spread by a sequence with length 12, and then is block-wise spread with an orthogonal code covering (OCC). Two-dimensional spreading is used, which is similar to the PUCCH format 1a/1b in LTE. The multiplexing of PUCCH data and DM-RS is interlaced in the time domain. The length of the PUCCH data depends on the number of OFDM symbols for PUCCH format 1, which is equal to ⌊L/2⌋. The remaining L − ⌊L/2⌋ symbols are used for DM-RS. The multiplexing capacity is determined by the number of available cyclic shifts and OCC. The transmission structure of PUCCH format 1 with 14 OFDM symbols is illustrated in Fig. 3.10.

3.1.3.5.3 Pucch Format 2

The PUCCH format 2 is for moderate payload size UCI transmission due to the restriction of the number of OFDM symbols. As the typical scenario of PUCCH format 2 with 1 or 2 OFDM symbols is not the coverage-limited case, PUCCH format 2 transmission is similar to OFDM-based PUSCH. The coded information bits after channel coding are scrambled and QPSK modulated. The modulation symbols are then mapped onto the resource elements of the physical resource blocks confgured for PUCCH format 2. In each of the symbols confgured for PUCCH format 2, DM-RS and PUCCH UCI information are FDM-based multiplexed, and there is one resource element for DM-RS every three resource elements as shown in Fig. 3.11. The number of resource block and symbol for PUCCH format 2 is confgured by higher layer signaling.

3.1.3.5.4 Pucch Formats 3 And 4

PUCCH format 3 is for large payload size of UCI. Both PUCCH format 3 and 4 transmissions are based on DFT-S-OFDM which has low PAPR to enhance coverage, which is similar to LTE PUCCH formats 4 and 5, respectively. The difference is that the number of symbols, ranging from 4 to 14, is confgured. The number of subcarrier assigned for PUCCH format 3 shall fulfll the requirement of 2 3 5 α2 α α 3 5 .

as PUSCH to reduce the implementation complexity. PUCCH format 4 is, however, constrained in one resource block, and block-wise spreading across the frequency domain within one resource block is used. The length of block-wise spreading is 2 or 4. The PUCCH format 4 of different UEs can be multiplexed by using different orthogonal sequences. The orthogonal sequence is confgured by the higher layer signaling. To preserve the single carrier property, the multiplexing of PUCCH data and DM-RS is TDM as in Table 3.6 [7].

3.1.4 Physical Layer (Phy) Reference Signals

For a wireless communication system, reference signal (aka pilot signal) is one of the key elements of the system design. Reference signal in general carries multiple fundamental functionalities to ensure proper and highly effcient PHY layer performances. Such functionalities include synchronization in time, frequency, and phase between the transmitters and receivers, conveying channel characteristics (long term and short term) for transmission property determination at the transmitter side and channel estimation and feedback at the receiver side; access link identifcation; and quality measurement. Though designing a single reference (for downlink) to fulfll all of these functionalities is possible, as in the case of the frst release of LTE, it will largely limit the performance and fexibility of the system. Therefore, fabricating a set of small number of reference signals to jointly fulfll these functionalities is a better choice. While on the surface NR may appear to have quite some similarity in terms of reference signal design to that of LTE, learning from the advantages and shortcomings of LTE, NR reference signal adopted a new design framework from the beginning and took into account additional design requirements that are unique for 5G networks. In the following, NR reference signal design framework and considerations are frst elaborated followed by design details for each type of reference signal including demodulation reference signal (DM-RS), channel state information reference signal (CSI-RS), sounding reference signal (SRS), and phase tracking reference signal (PT-RS). Quasi-co-location (QCL) and transmission confguration indicator (TCI) are then described as the linkage between different reference signals.

3.1.4.1 Reference Signal Design Framework And Considerations

As of the design framework, one of the fundamental changes of NR is to remove cell common reference signal (CRS) as in LTE system [10]. In LTE, CRS carries several important functionalities including cell identifcation, time and frequency synchronization, RRM measurement, data and control channel demodulation, and CSI measurement. It is (almost) the single reference signal for downlink in LTE Release-8 and is to transmit in every downlink sub-frame with certain time and frequency density to meet the minimum requirements of the most stringent functionality. CRS transmission is always on regardless of the presence and absence of data transmission except for the case with cell on/off and license-assisted access (LAA) introduced in its later releases. An always-on signal imposes persistent interference and overhead even when there is no data traffc and degrades the system performance signifcantly especially when the network deployment is dense [11]. In addition to interference and overhead issue, always-on CRS also largely limits the fexibility of the system design and forward compatibility which is a problem that became obvious for LTE during the discussions of small cell enhancements and new carrier type in 3GPP Release-12. Therefore, eliminating always-on CRS was adopted as a basic design assumption for NR.

The functionalities carried by CRS thus should be distributed among other reference signals or newly designed NR reference signals. Figure 3.12 shows the overview of LTE functionalities and the corresponding reference signals. As shown in Fig. 3.12, in LTE, UE acquires coarse time/frequency synchronization and cell identifcation through the detection of PSS/SSS (SS). Cell identifcation is also partially carried out through CRS as PBCH demodulation reference signal. CRS also carries the functionalities of digital AGC, fne time/frequency synchronization, and RRM measurement. CRS provides data demodulation reference signal for transmission modes 1–6 and control demodulation reference signal for all of the transmission modes (except for EPDCCH demodulation). In CSI measurement, CRS is used for deriving the signal part for transmission modes 1–8 and the interference part for transmission modes 1–9. CSI-RS is utilized in transmission modes 9 and 10 for measuring signal quality. DM-RS is the data demodulation reference signal for transmission modes 7–10 and EPDCCH control channel demodulation reference signal. DRS consists of SS, windowed CRS, and optionally CSI-RS and carries the functionalities of cell discovery and RRM measurement.

Fig. 3.12 LTE reference signal set and functionalities In LTE, QCL is defned with respect to delay spread, Doppler spread, Doppler shift, average gain, and average delay. LTE defnes several QCL assumptions among reference signals: - CRS ports of serving cell are assumed to be quasi-co-located with respect to delay spread, Doppler spread, Doppler shift, average gain, and average delay.

• CSI-RS, DM-RS ports, and their tied CRS ports are assumed to be quasi-colocated with respect to Doppler spread, Doppler shift, delay spread, and average delay.
• SS and CRS ports are assumed to be quasi-co-located with respect to Doppler shift and average gain.

The long-term channel characteristics derived from CRS can be utilized to facilitate the reception of the target reference signals when they can be assumed to be quasi-co-located.

With the removal of CRS, all of the functionalities (including cell identifcation (partial), fne time/frequency tracking, RRM measurements, digital AGC, CSI acquisition (partial), control channel demodulation) and long-term channel property estimation for QCL assumptions carried by it should be distributed to other reference signals or newly designed NR reference signals.

In addition, NR design should support frequency of up to 100 GHz. To combat the signifcantly higher path loss at these high-frequency bands, hybrid (analog/ digital) beamforming as a good trade-off between complexity and performance needs to be supported. In addition to the RS functionalities for the low-frequency band, NR design should consider the additional RS design to facilitate the analog beam acquisition, tracking, and feedback. Another issue for communication system in the high carrier frequency band is the phase noise. Compared to low-frequency band, the phase noise is considerably larger in the high-frequency band. Performance degradation caused by phase noise could be mitigated by increasing the subcarrier spacing to some extent. However, larger subcarrier spacing means shorter symbol length and larger CP length overhead. Thus with the possible maximum subcarrier spacing limitation, a specifc RS to fulfll this purpose may be introduced.

The overall reference signal design framework is summarized as follows (Fig. 3.13): - SSB enhanced for T/F synchronization

• CSI-RS confgurations to multiple functionalities such as fne T/F synchronization and main QCL assumption source (as TRS), RRM (jointly with SSB), CSI acquisition, discovery, and beam management
• DM-RS for demodulation of PDSCH/PUSCH and PDCCH - PT-RS is introduced for phase noise compensation - Uplink SRS One key aspect that is worth mentioning is that all of the above reference signals are UE specifcally confgurable and hence effectively no cell common reference signal exists in NR.

3.1.4.2 Demodulation Reference Signal

As only one transmission scheme based on DM-RS is defned in NR, its design needs to consider different scenarios and various and sometimes conficting requirements including:

• Good channel estimation performance: as the reference signal for demodulation,

good channel estimation performance is a most important criteria for DM-RS design. This demands suffcient signal density in time and frequency matching the underlined radio channel characteristics and the confgured numerology.

• DM-RS time and frequency pattern design need to support a large range of carrier frequency from below 1 GHz to up to 100 GHz and various velocities of up to 500 km/h while also considering that numerology may generally scale with carrier frequency.
• Total number of orthogonal DM-RS ports and its multiplexing scheme(s) with relatively small DM-RS overhead while ensuring good demodulation performance to support the large number of data layers for massive SU/MU-MIMO transmissions.
• In addition to MBB traffc, NR also needs to support URLLC type of services.

DM-RS design should then also enable very small channel estimation and demodulation processing time at the receiver.

• Flexible and confgurable numerology and frame structure are key properties of the NR design. DM-RS design needs to ft into these vast number of possible confgurations. One implementation of the fexible frame structure is to enable fexible duplexing of downlink and uplink transmission which may introduce severe cross-link interference. To mitigate the impact of such interference to channel estimation performance, common downlink and uplink DM-RS design that allows confguration of orthogonal downlink and uplink DM-RS transmissions is desirable.

In the following, overall design of NR DMRS is described followed by details of type 1 and type 2 DM-RS confgurations.

3.1.4.2.1 Overall Design Of Nr Dm-Rs

For DM-RS time and frequency pattern, two types (type 1 and type 2) of confgurations are introduced in NR. Type 1 DM-RS supports up to 4 orthogonal DM-RS ports when 1 symbol is confgured for DM-RS transmission and up to 8 orthogonal DM-RS ports when 2 symbols are confgured. Type 2 DM-RS supports up to 6 orthogonal DM-RS ports when 1 symbol is confgured for DM-RS transmission and up to 12 orthogonal DM-RS ports when 2 symbols are confgured. These orthogonal DM-RS ports are multiplexed in time domain and frequency domain by OCC. Both types of DM-RS confgurations are confgurable for downlink and for uplink and it can be confgured such that the DM-RSs for downlink and uplink are orthogonal to each other.

Two 16-bit confgurable DM-RS scrambling IDs are supported. The confguration is by RRC and in addition the scrambling ID is dynamically selected and indicated by DCI. Before RRC confguring the 16-bit DM-RS scrambling IDs, cell ID is used for DM-RS scrambling.

When mapping to symbol locations of a PDSCH/PUSCH transmission within a slot, front-loaded DM-RS symbol(s) only or front-loaded DM-RS plus additional DM-RS symbol(s) can be confgured. The additional DM-RS when present should be the exact copy of the front-loaded DM-RS for the PDSCH/PUSCH transmission, i.e., the same number of symbols, antenna ports, and sequence.

With front-loaded-only DM-RS, channel estimation can only rely on these 1 or 2

symbols in the early part of data transmission duration in order to speed up demodulation and reduce overall latency. However, without additional DM-RS symbol to enable time domain interpretation/fltering, the channel estimation and hence overall performance will degrade for scenarios with just moderate mobility.

For PDSCH/PUSCH mapping type A, the front-loaded DM-RS starts from the third or fourth symbols of each slot (or each hop if frequency hopping is supported). For PDSCH/PUSCH mapping type B, the front-loaded DM-RS starts from the frst symbol of the transmission duration. Number of additional DM-RS can be 1, 2, or 3 per network confgurations. The location of each additional DM-RS depends on the duration (i.e., number of OFDM symbols) of the PDSCH/PUSCH transmission and follows a set of general rules for better channel estimation performance. These rules include no more than 2 OFDM symbols for PDSCH/PUSCH after the last DM-RS, 2 to 4 symbols between neighboring DM-RSs, and DM-RSs almost evenly distributed in time. Figures 3.14 and 3.15 show some examples.

Two enhancements to reduce DM-RS PAPR are introduced in Release-16. For CP-OFDM PDSCH and PUSCH, code division multiplexing (CDM) group-dependent sequence generation is used for DM-RS to solve the high PAPR issue observed for Release-15 design where same sequences are used for different CDM groups. In addition, for PUSCH/PUCCH DM-RS for pi/2 modulation, new DM-RS sequences are specifed in Release-16 to reduce the PAPR to the same level as for data symbols. In these cases, computer-generated sequence is used for lengths 6, 12, 18, and 24. For lengths 12, 18, and 24, BPSK sequences are adopted. For length 6, 8PSK sequences are selected. For sequences with length 30 or larger, DMRS for π/2 BPSK modulation for PUCCH is generated based on gold sequence followed by π/2 BPSK modulation followed by transform precoding.

3.1.4.2.2 Dm-Rs Type 1 Confguration

For DM-RS type 1 confguration, as shown in Fig. 3.16, comb of every other tone in frequency and a cyclic shift in frequency is allocated to a DM-RS antenna port. With 2 combs and 2 cyclic shifts, 4 orthogonal DM-RS ports are supported when 1 OFDM symbol is confgured. When 2 OFDM symbols are confgured for DM-RS type 1 confguration, time domain OCC of size 2 is further used to generate orthogonal DM-RS ports which gives a total of 8 orthogonal ports.

DM-RS type 1 can be confgured for CP-OFDM PDSCH and PUSCH. For DFTs-OFDM PUSCH, only type 1 DM-RS is used. Type 1 DM-RS with a specifc confguration is also used before RRC as default DM-RS pattern.

For DM-RS type 2 confguration, as shown in Fig. 3.17, frequency domain OCC of size 2 over adjacent 2 REs and FDM are used to support 6 orthogonal DM-RS ports when 1 OFDM symbol is confgured. When 2 OFDM symbols are confgured for DM-RS type 2 confguration, time domain OCC of size 2 is further used to generate orthogonal DM-RS ports which gives a total of 12 orthogonal ports. With more orthogonal ports, type 2 confguration can potentially provide high system throughput for massive MIMO where a large number of data streams of MU-MIMO is desired. DM-RS type 2 can only be confgured for CP-OFDM PDSCH and PUSCH by RRC confguration.

3.1.4.3 Csi-Rs

CSI reference signal (CSI-RS) for NR generally takes the LTE CSI-RS as starting point, but is expended further to support, in addition to CSI acquisition, beam management, time and frequency tracking, and RRM measurement. In order to support these variety of functionalities which naturally come with different performance target and design requirements, NR CSI-RS overall design needs to have a lot more fexibility and confgurability in terms of number of antenna ports, time and frequency resource pattern, density in frequency, periodicity in time, etc. In the following, we frst describe the overall design of CSI-RS followed by specifcs of CSI-RS for each functionality.

3.1.4.3.1 General Design Of Csi-Rs

Overall, NR CSI-RS was designed to support a variety of number of antenna ports from single port to up to 32 orthogonal ports, different frequency domain density of {1/2, 1, 3} REs per PRB, and confgurable periodicity of {4, 5, 8, 10, 16, 20, 32, 40, 64, 80, 160, 320, 640} slots as well as aperiodic (i.e., trigger based) transmission.

Note that here the periodicity in time is in the unit of slot and hence depends on the numerology of the carrier/BWP.

The numbers of CSI-RS antenna ports supported are {1, 2, 4, 8, 12, 16, 24, and 32}. Extensive discussions and considerations had been taken into account to design the time and frequency resource mapping and CDM multiplexing of these antenna ports as shown in Table 3.7 [7]. In general, a CSI-RS resource pattern is aggregated with 2- or 4- or 8-port CSI-RS patterns and multiple density {1/2, 1, 3} and CDM of 2, 4, or 8 is supported. More specifcally:

• For 1-port CSI-RS, comblike resource mapping in frequency domain without CDM is used. For density (ρ) of 3 REs/PRB, comb-4 is used and for density of 1 or 0.5 REs/PRB, comb-12 is used within a PRB while every other PRB is mapped for density of 0.5 REs/PRB. See rows 1 and 2 of the table.
• For 2-port CSI-RS, frequency domain CDM of 2 (FD-CDM2) over 2 consecutive REs is used. For density of 0.5 REs/PRB, every other PRB is mapped. See row 3 of the table.
• For 4-port CSI-RS, FDM over 2 neighboring pairs of 2 consecutive REs with FD-CDM2 (as given in row 4) and TDM over 2 consecutive OFDM symbols each with FD-CDM2 of 2 REs of the same location (as given in row 5) are supported.
• For 8-port CSI-RS, FDM over 4 pairs of 2 consecutive REs with FD-CDM2 (as given in row 6) or TDM over 2 consecutive OFDM symbols each on the same location having FDM over 2 pairs of FD-CDM2 of 2 REs (as given in row 7) or FDM over 2 sets of 4 REs each of which in term uses CDM4 from a combination of FD2 (over 2 consecutive REs) and TD2 (over 2 consecutive OFDM symbols) (as shown in row 8) is supported.
• For 12-port CSI-RS, FDM over 6 pairs of 2 consecutive REs with FD-CDM2 (as given in row 9) or FDM over 3 sets of 4 REs each of which in term uses CDM4 from a combination of FD2 (over 2 consecutive REs) and TD2 (over 2 consecutive OFDM symbols) (as shown in row 10) is supported.
• For 16-port CSI-RS, TDM over 2 consecutive OFDM symbols each on the same location having FDM over 4 pairs of FD-CDM2 of 2 REs (as given in row 11) or FDM over 4 sets of 4 REs each of which in term uses CDM4 from a combination of FD2 (over 2 consecutive REs) and TD2 (over 2 consecutive OFDM symbols) (as shown in row 12) is supported.
• For 24-port CSI-RS, TDM over 2 sets of 2 consecutive OFDM symbols (i.e., total 4 OFDM symbols) each on the same location having FDM over 3 pairs of FD-CDM2 of 2 REs (as given in row 13) or TDM of 2 sets of 12 ports where each set uses FDM over 3 sets of 4 REs each of which in term uses CDM4 from a combination of FD2 (over 2 consecutive REs) and TD2 (over 2 consecutive OFDM symbols) and these two sets of 12 ports is aligned in frequency location (as given in row 14), or FDM over 3 sets of 8 REs each of which in term uses CDM8 from a combination of FD2 (over 2 consecutive REs) and TD4 (over 4 consecutive OFDM symbols) (as shown in row 15) is supported.
• For 32-port CSI-RS, TDM over 2 sets of 2 consecutive OFDM symbols (i.e., total 4 OFDM symbols) each on the same location having FDM over 4 pairs of FD-CDM2 of 2 REs (as given in row 16) or TDM of 2 sets of 16 ports where each set uses FDM over 4 sets of 4 REs each of which in term uses CDM4 from a combination of FD2 (over 2 consecutive REs) and TD2 (over 2 consecutive OFDM symbols) and these two sets of 16 ports is aligned in frequency location (as given in row 17), or FDM over 4 sets of 8 REs each of which in term uses CDM8 from a combination of FD2 (over 2 consecutive REs) and TD4 (over 4 consecutive OFDM symbols) (as shown in row 18) is supported. Among the above time and frequency pattern and multiplexing confgurations, a subset of them may be used for different functionalities.

3.1.4.3.2 Csi-Rs For Csi Acquisition

The prominent usage of CSI-RS, as indicated by its name, is for UE to preform measurements for CSI report. CSI-RS-related resource confgurations for CSI reporting are described below. A very-high-level description is given here. In general, UE needs to perform channel measurement(s) for reporting CSI over some NZP (nonzero power) CSI-RS resource(s). For certain cases, UE also needs to perform interference measurement(s) over additional resource(s) in order to derive proper CSI report(s) and these additional resource(s) can be CSI-IM (channel state information-interference measurement) resources or they can also be NZP CSI-RS resources or a combination of both.

The resources for CSI reporting, named CSI resource settings, are confgured via higher layer signaling. Each CSI resource setting CSI-ResourceConfg contains a confguration of a list of CSI resource sets (given by higher layer parameter csi-RSResourceSetList), where the list is comprised of references to either or both of NZP CSI-RS resource set(s) and SS/PBCH block set(s) or the list is comprised of references to CSI-IM resource set(s). Each NZP CSI-RS resource set consists of a number of NZP CSI-RS resource(s). The UE can be confgured with one or more CSI-IM resource set confguration(s) as indicated by the higher layer parameter CSI-IMResourceSet. Each CSI-IM resource set consists of a number of CSI-IM resource(s).

A CSI-IM resource confguration is basically a set of RE resources with time and frequency pattern of (2, 2) or (4, 1). The UE is to measure interference level on CSI-IM resource but not any specifc signal. On the other hand, when NZP CSI-RS resource is confgured for interference measurement, the UE is to measure interference level associated with specifc signals transmitted there.

It is worth to clarify here what is a ZP (zero power) CSI-RS resource in NR. UE can be confgured with one or more ZP CSI-RS resource set confgurations, with each ZP CSI-RS resource set consisting of a number of ZP CSI-RS resources. A ZP CSI-RS resource confguration consists of a set of RE resources with time and frequency pattern as of a CSI-RS resource and, with corresponding confguration and indication from the network, the UE will then assume that the RE resources associated with the ZP CSI-RS resource(s) are not mapped for PDSCH transmission.

3.1.4.3.3 Csi-Rs For Beam Management

For downlink beam management as well as uplink beam management when beam correspondence is supported by the UE, CSI-RS resource set is confgured for the UE to select preferred downlink transmit beam(s) (i.e., spatial domain transmission flter) via L1 RSRP reporting.

Downlink receiving beam sweeping can be implemented by gNB that confgures and transmits a CSI-RS resource set with the higher layer parameter repetition set to "on," and the UE may assume that the CSI-RS resources within the set are transmitted with the same downlink transmission beam and hence try different receiving beam. Downlink transmission beam sweeping can be implemented by gNB that confgures and transmits a CSI-RS resource set with the higher layer parameter repetition set to "off," and the UE shall not assume that the CSI-RS resources within the set are transmitted with the same downlink transmission beam. The UE may use the same receiving beam to receive and measure these CSI-RS resources and then report the best CSI-RS resources which correspond to the best transmission beams at the gNB side. With the combination of multiple CSI-RS resource sets with higher layer parameter repetition set to "on" or "off" and proper confguration of UE reporting, different beam management procedures can be implemented. Note that for CSI-RS resource set used for beam management purpose, it should have only the same number (1 or 2) of ports for all the CSI-RS resources within the set.

3.1.4.3.4 Csi-Rs For Time And Frequency Tracking

Tracking functionalities in NR system (as in LTE) include coarse time/frequency tracking, fne time/frequency tracking, and delay/Doppler spread tracking. In addition to the absence of CRS, NR supports a large variety of system bandwidth, subcarrier spacing, and carrier frequency. In addition to meeting the tracking performance requirements, signaling overhead, confguration fexibility, and performance robustness should also be considered. In order to satisfy tracking performance, based on the evaluation for various applicable scenarios, the reference signal should meet certain minimum design criteria, for example, time and frequency density and spacing, periodicity, and bandwidth.

Based on the above considerations, instead of introducing a separate type of reference signal for tracking purposes, NR defnes a set of CSI-RS resources for tracking which are labeled with higher layer parameter trs-Info as tracking reference signal. To guarantee UE baseband performance, though CSI-RS for tracking is UE specifcally confgurable, a UE in RRC connected mode is expected to receive higher layer confguration of at least one such type of CSI-RS resource set. The CSI-RS for tracking when confgured should satisfy the following restriction: - 2 or 4 CSI-RS resources of 1 antenna port in the CSI-RS resource set: For FR1 and FR2, 4 CSI-RS resources can be confgured within 2 consecutive slots with 2 CSI-RS resources per slot. In addition, for FR2, 2 CSI-RS resources can be confgured within 1 slot.

• The time-domain locations of the two CSI-RS resources in a slot are spaced by 4 OFDM symbols.
• Each CSI-RS resource is of a single port with frequency density of 3, i.e., mapping to every 4th tone.
• The bandwidth of the CSI-RS for tracking is confgurable but should be at least of 52 PRB or equal to the number of resource blocks of the associated BWP.
• The periodicity of the periodic CSI-RS for tracking is confgured as {10, 20, 40, or 80} ms. Periodic CSI-RS for tracking should be always available for a UE in connected mode. In addition, aperiodic CSI-RS for tracking is also supported in association with a periodic CSI-RS for tracking when confgured. Due to its design purpose and property, periodic CSI-RS for tracking is used by the UE to estimate long-term properties of its observed channels including timing, carrier frequency, Doppler, etc. It is then natural for the UE to use these long-term properties to derive channel estimator for receiving subsequent control and data channels. Therefore, periodic CSI-RS for tracking serves the main source reference signal for QCL assumptions at the UE.

3.1.4.3.5 Csi-Rs For Mobility Measurement

CSI-RS is also used for mobility measurement when it is confgured with the higher layer parameter CSI-RS-Resource-Mobility either independently or jointly with an associated SSB indicated by the higher layer parameter associatedSSB where the UE obtains the timing of the CSI-RS resource from the timing of the corresponding serving cell. And only CSI-RS resource with one antenna port can be confgured for mobility measurement.

3.1.4.4 Sounding Reference Signal

Sounding reference signal (SRS) is supported in NR transmitting from the UE at the uplink. Compared to that of the LTE, several new aspects are considered in NR to augment the SRS design.

The most prominent use case for uplink SRS is to obtain downlink CSI information in TDD system and enable outstanding downlink MIMO performance by utilizing channel reciprocity. Since uplink SRS is a UE-specifc signal, with potentially large number of active UEs in the cell as well as mobility of these UEs, SRS capacity may become a bottleneck. In addition, UE transmission power is generally much lower than that of a gNB, and SRS can be power limited and received at a relatively low SINR level resulting in poor estimation quality of CSI. Therefore, increasing SRS capacity as well as coverage performance is a critical issue to solve.

Another issue is the unbalanced UE capability for downlink reception versus uplink transmission. In general, most UEs can receive with more antennas at a number of downlink aggregated carriers but may only be able to transmit with a subset of these antennas at a smaller number of uplink carrier(s) at the same time which then hinders the gNB's capability to obtain full CSI knowledge of the downlink channels. In order to address these issues without dramatically increasing the complexity and cost of the UEs, switching SRS transmission from antenna to antenna and from carrier to carrier is specifed in NR.

SRS can of course also be used to obtain uplink CSI and enable uplink MIMO schemes. In addition to codebook-based uplink MIMO as in LTE, NR also introduced non-codebook-based MIMO scheme for uplink where potential uplink precoding choice(s) (determined by the UE) is conveyed through precoded SRS(s). In addition, to support FR2 with beamforming, uplink beam management using SRS is needed at least for UE without the capability of beam correspondence.

In the following, general SRS design is described followed by SRS specifcs for different use cases.

3.1.4.4.1 General Design Of Srs

In NR, an SRS resource with 1, 2, or 4 antenna ports is supported which can be mapped to 1, 2, or 4 consecutive OFDM symbols. Comb transmission on every 2 or 4 REs in frequency domain is supported. In addition, cyclic shift is supported with the maximum number of cyclic shifts equal to 12 (or 8) when comb of size 4 (or 2) is used. SRS sequence ID is confgured by higher layer parameter. Up to 6 OFDM symbols from the end of a slot may be used for SRS transmission which is a signifcant increase from that of LTE. A SRS resource may be confgured for periodic, semi-persistent, aperiodic SRS transmission. In frequency domain, SRS allocation is aligned with 4 PRB grids. Frequency hopping is supported as in the case of LTE. With same design approach, NR SRS bandwidth and hopping confguration are designed to cover a larger span of values compared to that of LTE.

For a given SRS resource, the UE is confgured with repetition factor 1, 2, or 4 by higher layer parameter to improve coverage. When frequency hopping within an SRS resource in each slot is not confgured, i.e., the number of repetition is equal to the number of SRS symbols, each of the antenna ports of the SRS resource in each slot is mapped in all the SRS symbols to the same set of subcarriers in the same set of PRBs. When frequency hopping within an SRS resource in each slot is confgured without repetition, each of the antenna ports of the SRS resource in each slot is mapped to different sets of subcarriers in each OFDM symbol, where the same transmission comb value is assumed for different sets of subcarriers. When both frequency hopping and repetition within an SRS resource in each slot are confgured, i.e., with 4 SRS symbols and repetition of 2, each of the antenna ports of the SRS resource in each slot is mapped to the same set of subcarriers within each pair of 2 adjacent OFDM symbols, and frequency hopping is across the two pairs.

A UE may be confgured with single symbol periodic or semi-persistent SRS resource with inter-slot hopping, where the SRS resource occupies the same symbol location in each slot. A UE may be confgured with 2- or 4-symbol periodic or semipersistent SRS resource with intra-slot and inter-slot hopping, where the N-symbol SRS resource occupies the same symbol location(s) in each slot. For the case of 4 symbols, when frequency hopping is confgured with repetition 2, intra-slot and inter-slot hopping is supported with each of the antenna ports of the SRS resource mapped to different sets of subcarriers across two pairs of 2 adjacent OFDM symbol(s) of the resource in each slot. Each of the antenna ports of the SRS resource is mapped to the same set of subcarriers within each pair of 2 adjacent OFDM symbols of the resource in each slot. For the case where the number of SRS symbols is equal to repetition factor, when frequency hopping is confgured, inter-slot frequency hopping is supported with each of the antenna ports of the SRS resource mapped to the same set of subcarriers in adjacent OFDM symbol(s) of the resource in each slot.

The UE can be confgured with one or more SRS resource sets. For each SRS resource set, a UE may be confgured with a number of SRS resources. The use case (such as beam management, codebook-based uplink MIMO, and non-codebookbased uplink MIMO, and antenna switching which actually is for general downlink CSI acquisition) for a SRS resource set is confgured by the higher layer parameter.

3.1.4.4.2 Srs For Dl Csi Acquisition

For downlink CSI acquisition, the usage of the SRS resource set is confgured as "antenna switching" to cover not only the cases where SRS transmission antenna switching happens but also the case where the numbers of transmit and receive antennas are the same and no antenna switching happens. As stated above, in the case that UE supports a smaller number of uplink carrier than downlink carrier, SRS switching between component carriers is also supported in NR.

3.1.4.4.2.1 Srs Antenna Switching

For SRS antenna switching, the following cases are supported: 1 (or 2) TX (transmission) to 2 (or 4) RX (receiving) antenna switching denoted as "1T2R," "2T4R," "1T4R," and "1T4R/2T4R" where a UE supports both "1T4R" and "2T4R" switching. In addition, "T=R" where the numbers of transmission and receiving are equal is also supported.

To support antenna switching, an SRS resource set is confgured with two (for "1T2R" or "2T4R") or four (for "1T4R") SRS resources transmitted in different symbols. Each SRS resource consists of one (for "1T2R" or "1T4R") or two (for "2T4R") antenna port(s) and the SRS port(s) of each SRS resource is associated with different UE antenna port(s). In addition, for the case of "1T4R," two aperiodic SRS resource sets with a total of four SRS resources transmitted in different symbols of two different slots can be confgured. The SRS port of each SRS resource in given two sets is associated with a different UE antenna port. The two sets are each confgured with two SRS resources, or one set is confgured with one SRS resource and the other set is confgured with three SRS resources.

Due to hardware limitation, the UE is confgured with a guard period of a number of symbols in between the SRS resources of a set for antenna switching. During these symbols, the UE does not transmit any other signal, in case the SRS resources of a set are transmitted in the same slot. The number of symbols depends on the subcarrier spacing.

3.1.4.4.2.2 Srs Carrier Switching

For a UE with the capability of supporting more downlink carriers than uplink carrier(s), in TDD system, a carrier with slot formats comprised of DL and UL symbols may not be confgured for PUSCH/PUCCH transmission. In such a case, normally no uplink transmission including that of SRS is expected. In order to obtain downlink CSI and take advantages of TDD reciprocity via SRS, the UE can be confgured with SRS resource(s) in this carrier. Since the UE may not have RF capability to transmit SRS on carrier(s) without PUSCH/PUCCH and at the same time to transmit on carrier(s) confgured with PUSCH/PUCCH, the UE needs to borrow some RF capability from a carrier confgured with PUSCH/PUCCH (denoted as switch-from carrier) and to switch the associated RF chain for transmission of SRS on this carrier (denoted as switch-to carrier). The switch-from carrier is confgured by the network according to the UE's reported capability. During SRS transmission on switching-to carrier including any interruption due to uplink and downlink UE RF retuning time as reported by the UE, the UE temporarily suspends the uplink transmission on the switching-from carrier.

In the case of collision between SRS transmission (including any interruption due to uplink and downlink RF retuning) on the switch-to carrier and uplink or downlink transmission(s) on the switch-from carrier, priority rules are defned for collision handling.

Both periodic/semi-persistent and aperiodic SRS are supported on the switch-to carrier. For triggering aperiodic SRS transmission, in addition to DL and UL DCI, group-based DCI format is also introduced.

3.1.4.4.3 Srs For Codebook- And Non-Codebook-Based Uplink Mimo

In NR uplink, UE can be confgured by the network for codebook-based transmission or non-codebook-based transmission.

For codebook-based uplink PUSCH transmission, one or two SRS resources can be confgured for the UE. When multiple SRS resources are confgured, an SRS resource indicator feld in uplink DCI is used to indicate the selected SRS resource whose antenna ports are used by the network to derive the precoding for uplink transmission.

For non-codebook-based uplink PUSCH transmission, multiple (up to 4) SRS resources can be confgured for the UE. The UE can calculate the precoder used for the transmission of precoded SRS based on the measurement of an associated NZP CSI-RS resource. The network determines the preferred precoder for PUSCH transmissions based on the reception of these precoded SRS transmissions from the UE and indicates to the UE via the wideband SRS resource indicator either in DCI for dynamic scheduling or in RRC for confgured grant.

3.1.4.4.4 Srs For Ul Beam Management

In the case that beam correspondence is not supported by the UE, the UE will not be able to determine its transmission beam for uplink based on downlink beam management outcomes. SRS with higher layer parameter usage set to "BeamManagement" will then be used for uplink beam management. To enable beam sweeping, when the higher layer parameter usage for SRS is set to "BeamManagement," only one SRS resource in each of the multiple SRS sets can be transmitted at a given time instant. The SRS resources in different SRS resource sets can be transmitted simultaneously. SRS resource indicator is used to indicate the selected beam from the network.

3.1.4.5 Phase Tracking Reference Signal

Phase tracking reference signal (PT-RS) is introduced in NR for high-frequency band (FR2) to compensate phase noise for both downlink and uplink data transmissions (PDSCH/PUSCH). When PT-RS is confgured and dynamically indicated (implicitly via DCI) by the network, the UE shall assume PT-RS being present only in the resource blocks used for the PDSCH/PUSCH with corresponding time/frequency density and location determined by several factors of the associated data transmission format. Specifcs of PT-RS design are different for PDSCH, PUSCH with CP-OFDM, and PUSCH with DFT-s-OFDM.

3.1.4.5.1 Pt-Rs For Pdsch

The presence of PT-RS is confgurable by the network. In addition, the time and frequency density of the PT-RS are dependent on the scheduled MCS and bandwidth of its associated PDSCH transmission for good trade-off of PT-RS overhead and PDSCH demodulation performance as shown in Table 3.8.

The thresholds in the tables are confgured by higher layer signaling. For the network to derive the values of these thresholds to determine the density and location of the PT-RS inside PDSCH transmission resource blocks, a UE reports its preferred MCS and bandwidth thresholds based on the UE capability at a given carrier frequency, for each subcarrier spacing applicable to data channel at this carrier frequency, assuming the MCS table with the maximum modulation order as it is reported to support.

The DL DM-RS port(s) associated with PT-RS port are assumed to be quasi-colocated. For PDSCH transmission of a single code word, the PT-RS antenna port is associated with the lowest indexed DM-RS antenna port for the PDSCH.

For PDSCH transmission of two code words, the PT-RS antenna port is associated with the lowest indexed DM-RS antenna port assigned for the code word with the higher MCS. If the MCS indices of the two code words are the same, the PT-RS antenna port is associated with the lowest indexed DM-RS antenna port assigned for code word 0.

3.1.4.5.2 Pt-Rs For Pusch Of Cp-Ofdm

PT-RS for PUSCH of CP-OFDM is generally similar to that of PDSCH which of course uses the same CP-OFDM waveform with the difference described in the following.

For uplink MIMO transmission, depending on the implementation architecture, antennas and RF chains at the UE may differ in terms of supporting of coherent transmission from these antennas and RF chains. If a UE has reported the capability of supporting full-coherent UL transmission, the UE shall expect the number of UL

PT-RS ports to be confgured as one if UL-PTRS is confgured. For partial-coherent and non-coherent codebook-based UL transmission, the actual number of UL PT-RS port(s) is determined based on TPMI and/or TRI in DCI. The maximum number of confgured PT-RS ports is given by the higher layer parameter and the UE is not expected to be confgured with a larger number of UL PT-RS ports than it has been reported as needed.

For codebook- or non-codebook-based UL transmission, the association between UL PT-RS port(s) and DM-RS port(s) is signaled by DCI when needed. For noncodebook-based UL transmission, the actual number of UL PT-RS port(s) to transmit is determined based on SRI(s). A UE may be confgured with the PT-RS port index for each confgured SRS resource by the higher layer parameter.

3.1.4.5.3 Pt-Rs For Pusch Of Dft-S-Ofdm

For PUSCH of DFT-s-OFDM waveform, DFT transform precoding is enabled.

When PT-RS is confgured for PUSCH transmission of DFT-s-OFDM, PT-RS samples are inserted into data transmission before DFT transform as shown in Fig. 3.18.

PT-RS before DFT transform is divided into a number of (2, 4, or 8) PT-RS groups and each PT-RS group consists of a number of (2 or 4) samples. An orthogonal sequence of the same size is applied within a PT-RS group.

The group pattern of PT-RS depends on the scheduled bandwidth according to Table 3.9 [12] and the thresholds are confgured by the network.

A scaling factor is applied according to the scheduled modulation order per [12] (Table 3.10).

Note that since PUSCH of DFT-s-OFDM only supports 1-layer transmission, its associated PT-RS has also only 1 port and the mapping is straightforward.

3.1.4.6 Quasi-Co-Location And Transmission Confguration Indicator

As discussed in Sect. 3.1.4.1, there is no common and always-on CRS-type reference signal in NR and the necessary functionalities are distributed into a set of downlink reference signals. Therefore, these distributed functionalities need to be integrated properly at the receiver to facilitate good performance of data and control channel demodulation, measurements, etc. Quasi-co-location between antenna ports is used as the venue for such integration.

The defnitions of antenna port and quasi-co-location are given in [7] as follows: "An antenna port is defned such that the channel over which a symbol on the antenna port is conveyed can be inferred from the channel over which another symbol on the same antenna port is conveyed.

Two antenna ports are said to be quasi co-located if the large-scale properties of the channel over which a symbol on one antenna port is conveyed can be inferred from the channel over which a symbol on the other antenna port is conveyed. The large-scale properties include one or more of delay spread, Doppler spread, Doppler shift, average gain, average delay, and spatial Rx parameters." Via quasi-co-location relationship between antenna ports of a source reference signal and a target reference signal with respect to a set of large-scale channel properties, the UE receiver can utilize channel properties estimated/derived from the source reference signal for channel estimation and measurement performed on the target reference signal.

Quasi-co-location relationship between reference signals is confgured and signaled through TCI (transmission confguration indicator) state(s). The UE shall not assume that two antenna ports are quasi-co-located with respect to any QCL type unless specifed otherwise. A TCI state contains information for confguring quasico-location relationship including one or two downlink reference signals as the source reference signal and the QCL types associated with the one or two source reference signals. QCL types are defned in the following to categorize a subset of QCL parameters: - "QCL-TypeA": {Doppler shift, Doppler spread, average delay, delay spread} - "QCL-TypeB": {Doppler shift, Doppler spread} - "QCL-TypeC": {Doppler shift, average delay} - "QCL-TypeD": {Spatial Rx parameter} As different reference signals are designed and optimized for different purpose(s), a single type of reference signal is not feasible to derive all QCL parameters of a channel between the TRP and the UE with suffcient accuracy. For example, CSI-RS for tracking is specifcally designed to obtain and maintain fne time and frequency synchronization and Doppler estimation for a channel on certain spatial (or beamforming) direction and should be used to derive parameters in QCL-TypeA in most of the situations. However, CSI-RS for tracking may not be a good option to obtain spatial Rx parameter via beam sweeping which can result in very large overhead and long delay. On the other hand, CSI-RS for beam management (i.e., CSI-RS resource confgured with higher layer parameter repetition) is generally used to derive QCL-TypeD for FR2. Therefore, it needs both CSI-RS for tracking and CSI-RS for beam management for the UE to obtain a full set of quasi-co-location parameters. For the case of two source DL RSs, the QCL types should not be the same, regardless of whether the references are to the same DL RS or different DL RSs.

The possible confgurations of quasi-co-location relationship through TCI states are given in Table 3.11. Note that QCL-TypeD may not always be applicable.

In the absence of CSI-RS confguration, for example before RRC confguration, the UE may assume PDSCH/PDCCH DM-RS and SS/PBCH block to be quasi-colocated with respect to Doppler shift, Doppler spread, average delay, delay spread, and, when applicable, spatial Rx parameters.

3.2 5G-Nr Spectrum And Band Defnition 3.2.1 5G Spectrum And Duplexing 3.2.1.1 Imt-2020 Candidate Spectrum

The IMT spectra identifed in the ITU's World Radiocommunication Conferences (WRC) 2015 and 2019 (which are below 6 GHz and above 24 GHz, respectively) are applicable for 5G deployments. 3GPP defnes frequency bands for the 5G-NR

interface according to the guidance both from ITU and from the regional regulators, with prioritization according to the operators' commercial 5G plan. According to [13], three frequency ranges are identifed for 5G deployments for both eMBB and IoT applications, including the new frequency ranges of 3–5 GHz and 24–40 GHz, as well as the existing LTE bands below 3 GHz.

As illustrated in Fig. 3.19, generally, a triple-layer concept can be applied to the spectral resources based on different service requirements. An "over-sailing layer" below 2 GHz is expected to remain the essential layer for extending the 5G mobile broadband coverage both to wide areas and to deep indoor environments. This is especially important for mMTC and URLLC applications. On the other hand, the "coverage and capacity layer" spanning from 2 to 6 GHz can be used for striking a

compromise between capacity and coverage. However, compared to the range below 2 GHz, these bands suffer from a higher penetration loss and propagation attenuation. The "super data layer" above 6 GHz can be invoked for use cases requiring extremely high data rates but relaxed coverage. Given this triple-layer concept, the eMBB, mMTC, and URLLC services that require different coverage and rate capability can be accommodated in the appropriate layer. However, a service-based single-layer operation would complicate the 5G deployments and it is ineffcient in delivering services that simultaneously require both good coverage and high data rate as well as low latency. To accommodate these diverse services, the employment of joint multiple spectral layers becomes a "must" for a meritorious 5G network.

3.2.1.1.1 C-Band (3300–4200 Mhz And 4400–5000 Mhz)

Spectrum availability for IMT in the 3300–4200 and 4400–5000 MHz ranges is increasing globally (Fig. 3.20). The 3400–3600 MHz frequency band is allocated to mobile service on a co-primary basis in almost all countries throughout the world. Administrations will make available different portions of the 3300–4200 and 4400–5000 MHz ranges at different times, incrementally building large contiguous blocks.

The 3GPP 5G-NR specifcation supports 3300–3800 MHz from the start, using a TDD access scheme. In line with the release plans from many countries, the 3300–3800 MHz band will be the primary 5G band with greatest potential for global harmonization over time: it is recommended that at least 100 MHz of contiguous bandwidth from this band be allocated to each 5G network. In order to take advantage of the harmonized technical specifcation—the 3GPP 5G-NR specifcation for 3300–3800 MHz band—regulators are recommended to adopt a frequency arrangement with an aligned lower block edge of usable spectrum and harmonized technical regulatory conditions, at least in countries of the same region. The 5G-NR ecosystem of 3300–3800 MHz is expected to be commercially ready in 2018 in all the three regions [14].

3.2.1.1.2 Mm-Wave Bands

The World Radiocommunication Conference 2015 (WRC-15) paved the way for the future development of IMT on higher frequency bands by identifying several frequencies for study within the 24.25–86 GHz range (Fig. 3.21) for possible identifcation for IMT under Agenda Item 1.13 of WRC-19. The 24.25–27.5 and 37–43.5 GHz bands are prioritized within the ongoing ITU-R work in preparation for WRC-19; all regions and countries are recommended to support the identifcation of these two bands for IMT during WRC-19 and should aim to harmonize technical conditions for use of these frequencies in 5G. The frequency band of 27.5–29.5 GHz, though not included in the WRC-19 Agenda Item 1.13, is considered for 5G in the USA, South Korea, and Japan.

The 24.25–29.5 and 37–43.5 GHz ranges are the most promising frequencies for the early deployment of 5G millimeter wave systems, and several leading markets are considering portions of these two ranges for early deployments (Fig. 3.22), and the two ranges are also being specifed in 3GPP Release-15 based on a TDD access scheme. It is recommended that at least 400 MHz of contiguous spectrum per network from these ranges be assigned for the early deployment of 5G.

3.2.1.1.3 Sub-6 Ghz Frequency Bands For 5G

Sub-3 GHz candidate frequency bands are given in Fig. 3.23.

The S-band (2496–2690 MHz) is another 5G candidate band that may have early commercial deployment. Currently, there are LTE TDD networks deployed in this band partially in both China and the USA, while EU deploys LTE FDD in the two edges of this band instead. It is highly likely that those regions may deploy 5G with TDD duplex mode on the left spectrum of that band, with maximized ecosystem sharing.

The L-band (1427–1518 MHz) is one 5G candidate band that has the potential to be allocated to mobile in most countries in the world. CEPT and CITEL regions have adopted the SDL (Supplemental Downlink) scheme for this band. The requirement for stand-alone operation in the band (both UL and DL transmissions) has emerged in some other regions. In the case of stand-alone 5G systems, a TDD access scheme is a potentially appropriate option, which can accommodate traffc asymmetry in the UL/DL directions with good potential for economies of scale. The same 5G-NR equipment can serve both the TDD and SDL markets. In addition, the SDL band can be paired with an individual SUL band (as described in Sect. 3.2.2) as well, following the traditional way of FDD operation.

The 700 MHz band has already been harmonized for mobile in most countries.

Europe plans to use this band for 5G. Over the long term, the other frequencies of UHF band (470–694/698 MHz) could also be used for mobile, while the USA has already started the process of transferring the band from broadcasting to mobile service.

3.2.1.2 5G Duplexing Mechanisms 3.2.1.2.1 5G Candidate Band Types And Duplex Modes

Duplexing is another key factor affecting the network operation. There are two typical spectrum types available for IMT systems, i.e., paired spectrum and unpaired spectrum. At present, FDD and TDD are two dominant duplex modes, which are adopted to paired spectrum and unpaired spectrum, respectively, as introduced in the following:

• FDD over paired spectrum: FDD is more mature in 2G/3G/4G telecommunication systems, whose spectra are mainly located in low frequency range below 3 GHz. Due to the limited frequency resource in low frequency range, generally the bandwidth of a FDD band is quite limited. In addition, common RF flter design for FDD band is needed for the convenience of ecosystem sharing; thus a fxed duplex distance between DL and UL spectra is defned for each FDD band, which is also the obstacle in expanding the bandwidth of a FDD band.
• TDD over unpaired spectrum: With the increasing telecommunication traffc load, TDD attracts more attention due to more available wideband unpaired spectrum in middle and high frequency range. LTE TDD successfully operates the wide-coverage public networks for eMBB service in 2.6 GHz (Band 41) and 3.5 GHz (Band 42). Moreover, the DL and UL channel reciprocity in TDD system can well accommodate the optimized multi-antenna operation with effcient sounding design instead of redundant feedback of channel state information, which brings a signifcant throughput gain especially for multiple-user MIMO mechanisms. The channel statement information feedback overhead required for multi-antenna mechanism is un-neglectable for a spectrum with very large DL transmission bandwidth; thus TDD operation with UL sounding is a must for the multi-antenna solutions in C-band and mm-wave bands. This is also one of the

main reasons that the current 5G new spectrum in C-band and mm-wave chooses TDD mode.

• SDL over unpaired spectrum: LTE also defnes a supplementary DL (SDL) band type, which also fts to unpaired spectrum. However, SDL band type can only operate jointly with a normal FDD or TDD band which has UL transmission resources for feedback. A SDL carrier is aggregated to a normal FDD or TDD primary carrier, following the normal access, scheduling, and HARQ processes, which is quite in line with a normal FDD DL carrier.

In addition to the traditional band types, 5G introduces a new supplementary UL (SUL) type.

• SUL over unpaired spectrum: SUL band is combined with a normal 5G TDD (or SDL and FDD) band, to provide the UL control signal and feedback transmission as well as UL data transmission. The original introduction of SUL band type is to compensate the UL coverage shortage of the 5G TDD spectrum in C-band and mm-wave bands, where SUL bands are generally overlapped with the UL spectrum region of the existing LTE FDD band, which provides the possibility for an operator to reuse their spared UL frequency resources in LTE network for 5G-NR UL transmission for cell-edge users. The detailed concept, specifcations, and benefts of SUL band combinations are introduced in Sect. 3.3.

3.2.1.2.2 Flexible Duplex: Convergence Of Fdd And Tdd

In the future, more and more operators will own multiple spectrum bands, and most likely both FDD and TDD bands. Consequently, there are, currently, many LTE networks operating in both FDD and TDD bands jointly to accommodate the eMBB and IoT services to end users. It is foreseen that convergence of TDD and FDD will be the evolution trend of mobile communication systems.

3.2.1.2.2.1 Joint Operation Of Fdd And Tdd

There are several FDD/TDD joint operation mechanisms specifed in 3GPP, as below:

• Common network with multi-RATs of FDD and TDD: One public cellular network contains two radio access layers in a FDD band and a TDD band individually, with the authentication, access, and mobility control by a common core network. Which radio access layer is selected by a user is dependent on the coverage and the available radio resource units of the two layers. Switching between FDD and TDD layers is generally triggered by the inter-RAT handover based on the UE reference signal receiving power (RSRP) measurement of the FDD cell and TDD cell, respectively. In such a joint FDD/TDD multi-RAT operation, the FDD or TDD radio access layer provides independent scheduling and transmission procedure.
• FDD/TDD carrier aggregation: LTE-Advanced system introduced FDD/TDD carrier aggregation mechanism since Release-12, which enables the physical layer scheduling of parallel data transmission in both FDD and TDD carriers for a single UE, which requires either co-site FDD/TDD eNodeB (eNB) or ideal backhaul between FDD and TDD eNBs. Either a FDD carrier or a TDD carrier can be the anchor carrier to provide the basic mobility functions such as handover and cell reselection, while it can activate and deactivate one or more (up to four in DL and two in UL) secondary component carriers for parallel data transmission. In LTE-Advanced system, each component carrier has its own scheduling control signaling and HARQ process, while the HARQ feedback of the SCCs can be transmitted only on the UL carrier of PCC. Accordingly, 3GPP defnes detailed specifcations on how to deal with HARQ feedback timing and potential confiction cases for different TDD DL/UL confgurations. 5G-NR goes further by enabling the SCC scheduling via PCC DL control signaling. However, 3GPP Release-15 only specifes the cross-carrier scheduling for the component carriers with the same numerology, while 5G-NR TDD carriers in mm-wave band and C-band generally have different numerologies from that of low-frequency FDD band.
• FDD/TDD dual connectivity (DC): FDD/TDD DC provides the capability of parallel transmit multiple streams over FDD and TDD to a single UE semi-statically confgured at higher layer. The higher layer (RLC/PDCH or upward) parallel transmission makes it applicable to those deployment scenarios without ideal almost zero-latency backhaul, such as inter-site FDD/TDD stream aggregation for a single UE.

3.2.1.2.2.2 Synchronization For Tdd Band

Network synchronization is a basic requirement for the TDD cellular systems. It requires multiple cells to synchronize the TDD frame structure, including DL and UL switching point, as well as DL/UL confgurations.

There are multiple types of synchronization requirements as the following:

• Intra-frequency inter-cell synchronization
• Figure 3.24 describes the inter-cell interference, while (a) is the given network topology with (b) for the scenario with two neighboring cells that are confgured with two different TDD DL/UL confgurations, and (c) is for the scenario with two neighboring cells that are asynchronous in DL/UL switching points. It can be seen that the period with cell 1 schedules DL transmission while cell 2 schedules the UL transmission at the same time, and the high transmit power from cell 1 base station (BS) ends to a strong interference at the receiver of the cell 1 BS, and thus may block the cell 1 BS reception during that time period.
• To avoid such inter-cell BS-to-BS interference, it is recommended for the operator to deploy a TDD network with frame synchronization. Guard period

between DL and UL transmission is usually chosen corresponding to the interference-protection distance from an aggregator BS to the victim BS.

• Intra-band inter-frequency inter-cell synchronization
• When one operator deploys TDD system on two or more carriers in the same band, network synchronization is also necessary. Figure 3.25 describes the inter-frequency BS-to-BS interference, for both neighboring carrier and two carriers with a certain frequency distance. Since the BS-to-BS interference comes from the same network and also the same BS, the intra-band transmission signals will fall into the RF flter of the same receiver flter range and thus block the BS receiver totally.
• Intra-band inter-operator synchronization with regional regulation coordination
• For different operators who deploy TDD systems on their own spectrum located in the same band, the similar inter-frequency BS-to-BS interference is suffered from the victim TDD network that with the UL time widow overlaps with the aggressor TDD network with the DL transmission. There was a proposal to introduce a guard band between the two neighboring TDD networks in the same frequency band; however for cellular network with macro BSs, a quite big guard band is required in addition to the inter-operator special isolation to avoid the BS receiver blocking and basic receiving performance of the victim TDD network. According to the regulation evaluation in both China MIIT [15] and EU ECC [16], a guard band of 5–10 MHz is needed for two 2.6 GHz LTE TDD systems with an operating bandwidth of 20 MHz individually, and more than 25 MHz guard band is required for two 3.5 GHz NR TDD systems with an operating bandwidth of 100 MHz individually as shown in Fig. 3.26. Such big guard band is a big waste of the cherished spectrum resources, and thus is unacceptable for any regional regulator as well as operators.

To reach the intra-band inter-operator TDD synchronization, the regional regulator generally specifes a common TDD DL/UL frame structure according to the DL/ UL traffc load ratio statistics from multiple operators. Few countries also ask operators to do coordination if they have to deploy TDD network in the same band.

• China: For the frst 2.6 GHz TDD network in the world, China operators contributed many efforts to coordinate the network. Finally, under the guideline of MIIT (Ministry of Industry and Information Technology) of China, the synchronization among operators' networks at the band of 2.6 GHz was implemented based on the same frame structure TDD confguration 2 and the same DL/UL traffc ratio of 4:1. China MIIT is actively organizing MNOs and relevant stakeholders to negotiate a single frame structure for synchronization of 5G networks in 3.5 GHz band.
• Japan: Public open hearing of potential operators for Japan 3.4–3.6 GHz band was held by MIC (Ministry of Internal Affairs and Communications) on 23rd January 2014. All operators presented a clear position of TDD and advocated the necessity of operator's consensus for collaboration for realizing TDD synchronization including the DL/UL confguration ideally, in order to achieve No Guard Band for effcient usage of spectrum resources. All the operators have the same opinion of DL heavy frame confguration, by referring to the heavy data traffc in the downlink side. Japan Ministry of Internal Affairs and Communications (MIC) issued the guideline for the introduction of 4G for comments in September 2014, which included that 3480–3600 MHz should be assigned for 3 operators (40 MHz per operator) for TDD use, and licensees are obliged to agree with each other in advance about the matters for TDD synchronized operation, where frame structure confguration 2 is also chosen.

Fig. 3.27 UK 3.4 GHz band plan based on fnal auction results

Fig. 3.28 TDD frame structure with DL/UL ratio of 4:1. (a) LTE-TDD frame structure of confguration 2, 5 ms single DL/UL switching period. (b) 5G-NR frame structure of 2.5 ms single DL/UL

switching period

• On 6 April 2018, MIC allocated the remaining spectrum of 3.5 GHz to 2 operators, and the two licenses are to synchronize existing 3.5 GHz TDD network.
• UK: On 11 July 2017, UK Ofcom issued auction regulations [17] for the award of the 2.3 and 3.4 GHz spectrum bands and also updated Information Memorandum [18] alongside the regulations. The updated Information Memorandum sets out the conditions of the licenses that will be issued for the 2.3 and 3.4 GHz bands, respectively. The licenses are technology neutral and based on Time Division Duplex (TDD) mode. Licensees will be required to synchronize their networks in order to avoid interference to one another, so traffc alignment and the "Preferred Frame Structure confguration 2" as in Fig. 3.28 (a) for transmission with the limits of the Permissive Transmission Mask are mandated to implement the synchronization. Time slots must have the duration of 1 millisecond. TD-LTE frame confguration 2 (DL/UL ratio 4:1) is compatible with this frame structure. Other details of the preferred frame structure can be found in paragraph 12 in the Information Memorandum.
• In early 2019, Ofcom conducted the auction, and the 3.4 GHz band plan based on fnal auction results1 is shown in Fig. 3.27.
• As described above, almost all the commercial LTE TDD systems globally adopt TDD confguration 2, i.e., 5 ms period with "DSUDD" pattern and the DL/UL ratio of around 4:1 as in Fig. 3.28(a). For the TDD band with only 5G-NR deployment, the same DL/UL ratio of 4:1 can be chosen as well, but with 2.5 ms period with "DDDSU" pattern due to wider subcarrier spacing and shorter slot length, as shown in Fig. 3.28(b). As described in Sect. 3.1.2, 3GPP specifes a very fexible frame structure for 5G-NR; therefore other TDD DL/UL confgurations can also be adopted according to regional regulation coordination among operators, while both the traffc statistics and the implementation complexity should be taken into account.
• Intra-band LTE/NR synchronization For both 2.6 GHz Band 41 and 3.5 GHz Band 42, there are countries, e.g., China and Japan, that have planned to deploy both LTE TDD and 5G-NR systems within the same band. Same as the intra-band inter-operator TDD synchronization, the regulators mandate the same DL/UL switching period and switching points. LTE and 5G-NR usually have different numerologies, i.e., 15 kHz subcarrier spacing for LTE and 30 kHz subcarrier spacing for NR. For LTE TDD, the frame confguration 2 is the most widely used frame structure, i.e., 3:1 DL/UL ratio with 5 ms DL/UL switching period. Then, 5G-NR is recommended to operate with a TDD DL/UL switching period of 5 ms with the pattern of "DDDDDDDSUU" and the DL/UL ratio of 8:2 to attain synchronization with LTE. In addition, the slot format confguration in NR is very fexible which can match all the confgurations of special subframe in LTE. The only modifcation is to adjust the starting point of the frame as shown in Fig. 3.29.

It is worth noting that network synchronization is not an exclusive feature of communication systems that work in Time Division Duplex (TDD) mode, while it could also be applied to the communication systems working in Frequency Division Duplex (FDD) mode if performance gain is expected by utilizing techniques such as interference cancellation (IC). The difference lies in the requirement of synchronization accuracy precision. GSM, UMTS, WCDMA, and LTE-FDD mobile technologies require only frequency synchronization with accuracy within 50 parts per billion (ppb) at the radio interface. CDMA2000, TD-SCDMA, and LTE-TDD services have the same frequency requirements as the other earlier 2G/3G network, but also specify requirements for phase and time. Different with FDD systems where frequency synchronization is generally suffcient, it is essential that the time and phase reference in TDD system is traceable to Coordinated Universal Time (UTC).

Without the common UTC time reference cell sites cannot operate as intended. In the LTE-TDD system, two phase accuracy granularities are specifed, 1.5μs and 5μs, corresponding to two cell radius sizes differentiated by 3 km. Table 3.12 presents a summary of the synchronization requirements of different mobile network modes.

The mechanisms to guarantee the TDD network synchronization are mature since 3G TD-SCDMA system, and already widely used in LTE TDD systems globally. Currently, the main solutions include:

• Type 1: distributed synchronization scheme based on GNSS systems. GNSS signal receivers are directly deployed on the terminal and BSs; each BS acquires satellite time signals (GPS, Beidou, Glaness, etc.) directly to achieve the time synchronization between different BSs and to ensure the maximum deviation of any two of the base stations no more than 3 us. Generally macro BSs are located in open areas, which makes it easy to install GPS antenna with good satellite signal receiving performance. However, for those BSs deployed indoor or outdoor but surrounded by tall objects that easily block GPS signals, it will be diffcult to receive GPS signal correctly.
• Type 2: centralized synchronization scheme based on IEEE1588v2 system maintained by the cellular network backhaul. IEEE 1588v2 specifes accurate time transfer protocol, which can achieve sub-microsecond precision time synchronization like the current GPS. The clock synchronization information of the main time source is transmitted through the 1588v2 protocol packet on the transmission network. The BSs can obtain time information from transmission network through the 1588v2 interface to achieve synchronization accuracy of ns level. It requires all nodes of the bearer network to support PTP, and additionally the clock synchronization quality is affected by network QoS. The above two types of synchronization solutions are quite complemented to each other. Both of them are widely used in the commercial LTE networks already. It is up to each operator to choose their own network synchronization solutions.

3.2.1.2.2.3 Dynamic Tdd And Flexible Duplex

Traditional TDD network has a static TDD DL/UL confguration, which is usually decided by the statistical UL/DL traffc load ratio among multiple operators in a specifc country or region. However, in the practical telecommunication networks, the DL traffc constitutes a large portion of the entire tele-traffc, as shown in the practical network traffc statistics in Fig. 3.30(a). With the popularity of video streaming increasing, it is forecast that the proportion of DL content will grow even further in the future as in Fig. 3.30(b) [19]; hence it is natural that more resources should be allocated to the DL. Therefore, a smaller proportion of the resources is left for the UL, which will further affect the UL coverage performance.

Telecommunication industry never gives up on improving the total spectrum utilization effciency [20]. If the network can confgure the DL/UL radio resource ratio according to the actual DL/UL traffc ratio geographically and also timely, assuming that it can resolve the serious DL-to-UL interference in the wide-area TDD network, the spectrum utilization can be improved a lot. There are two steps in the duplex evolving road map as below; additionally there were academia papers proposing full duplex [21, 22] as the fnal duplex evolution step, which is however far immature due to many implementation challenges.

Dynamic Tdd

LTE-Advanced system introduced the dynamic confguration of TDD DL/UL ratio

(Fig. 3.31) from 3GPP Release-12 and onwards, called as enhanced interference management traffc adaption (eIMTA) [23]. In the theoretical analyses and simulation [24], eIMTA can be applied to an isolated area such as indoor hotspot scenario like football game with a very special DL/UL traffc statistics. One drawback of eIMTA is that the newly introduced interference between DL and UL signals will

reduce signal-to-interference plus noise ratio (SINR) when the inter-cell interference is present, and thus cause the communication quality degradation over that of the conventional TDD system. As interpreted in the following fgure, the interference suffered by Cell 0 base station receiver increases since the DL signals from neighbor base stations with higher transmit power than UL signals cause strong interference.

The commercial LTE TDD networks are mainly wide-coverage cellular network; thus eIMTA has not yet been deployed in practical systems due to its severe intercarrier and intra-carrier interference of the normal transceiver implementation in Macro eNB.

Flexible Duplex

• Flexible confguration on DL/UL ratio: 5G standardization takes into account the future trend of FDD/TDD convergence from day 1, which defnes a very fexible physical layer design as well as the symmetric DL/UL air interface. 5G-NR inherits the dynamic TDD DL/UL ratio from LTE eIMTA, while it goes further with both static cell-specifc DL/UL confgurations and DL/UL switching period, as well as the additional potential semi-static or dynamic UE-specifc DL/UL confgurations. As shown in Fig. 3.32, 3GPP Release-15 specifes up to 62 slot confgurations with each OFDM symbol marked as DL, UL, or unknown direction, while the unknown-direction symbols can be confgured specifcally for each UE receiver. Among those 62 slot confgurations, there are 4 typical confgurations as DL-only slot, DL-dominant slot, UL-only slot, and UL-dominant slot [5]. Within one 5G-NR radio frame, each slot can have its own DL/UL confgurations out of the total 62 candidates. Theoretically, 5G-NR can support a large number of candidate DL/UL confgurations for the frame structure seen from UE side, while keeping the basic cell-level TDD synchronization on those basic DL or UL symbols which does not necessarily occupy the whole time of a frame.

Fig. 3.32 5G-NR slot DL/UL confgurations in 3GPP Release-15

• Other application scenarios of fexible duplex: Theoretically, such fexible slot direction can be confgured for paired spectrum as well. Due to the more and more serious imbalance between DL and UL traffc load, the symmetric DL/UL bandwidths in FDD band end to more and more spared UL radio resources when FDD DL resources are fully occupied by the eMBB traffc. Accordingly, there is a potential solution to make use of the spared UL for the low-power DL transmission of pico cells, as shown in Fig. 3.33(a). The fexible duplex mode can also be applied to some new FDD DL spectrum and SDL band as in Fig. 3.33(b), where there is possibility to introduce SRS in the DL carriers to enable effcient multiantenna mechanism based on the DL/UL channel reciprocity. For fexible duplex application in FDD DL or UL spectrum or SDL bands, the main limit would be regional spectrum regulation restrictions. Flexible duplex is also applicable to access-integrated wireless backhaul, D2D, etc., as in Fig. 3.34. In reality, it is very costly for operators to widely deploy highspeed and low-latency backhaul (e.g., fber), considering the cost and diffculty to fnd suitable sites, while the traditional wireless backhauling or relay requires additional spectrum to avoid the interference to and from the access link [25, 26], which is very low effcient. With fexible duplex, the same resource may be allocated to the backhaul link and access link, with MU-MIMO type of advanced receiver to mitigate the inter-link interference. The similar mechanism is for in-band D2D system as well.

• Flexible duplex interference mitigation mechanisms: For cellular network with fexible duplex confguration, the main issue is still the DL-to-UL interference among cells. To reach a good inter-cell inter-direction interference mitigation performance with an acceptable advanced receiver implementation complexity, 5G-NR air interface is designed to have a symmetrical DL/UL transmission format, multi-access, as well as similar reference signals for channel estimation and demodulation. A. Multi-Access Aspect
• In LTE, DFT-S-OFDMA was adopted as the multiple-access scheme for UL due to low peak-to-average power ratio (PAPR) for better coverage, while OFDMA is adopted for DL for more effcient wideband frequency-domain selective scheduling. Accordingly, the subcarrier mapping schemes are different for DL and UL signals, leading to half-subcarrier offset between the two kinds of signals, as interpreted in Fig. 3.35. Such design works well in conventional systems but cannot support new applications in 5G. In order to support fexible duplex, symmetrical design is proposed for multiple-access schemes with both DL and UL signals adopting OFDMA scheme. Subcarrier mapping scheme should also align with each other, so as to avoid inter-carrier interference between DL and UL signals. Based on such symmetry, current processes to receive two DL (UL) signals can be reused to receive DL and UL signals simultaneously.

B. Reference Signal Aspect

In LTE, the reference signal (RS) patterns were designed quite differently for DL and UL. In order to well support simultaneous reception of DL and UL signals, it is preferred to avoid interference between data and DM RSs and ensure DL and UL DM RSs to be orthogonal with each other, so that the receiver can estimate DL and UL fading channels for both signals accurately to ensure decoding afterwards. In 5G, the DL/UL symmetric air interface enables the orthogonal resources among reference signals or between reference signal and data transmissions in different transmission directions. By such design, the inter-direction interference will be seen from the advanced receiver as the orthogonal streams in multiuser MIMO of the actual scheduled signal. Virtual MIMO advanced receiver is well implemented in LTE network already, which can be the basis of the 5G fexible duplex advanced receiver.

3.2.2 3Gpp 5G-Nr Band Defnition 3.2.2.1 3Gpp Rel.15 5G-Nr Band Defnition

The 5G-NR candidate spectrums specifed in 3GPP Release-15 are classifed into two frequency ranges. FR1 represents the sub-6 GHz spectrum, and ranges from 450 MHz to 6 GHz, while FR2 is for millimeter wave, from 24.25 GHz to 52.6 GHz.

FR1 consists of the traditional LTE bands as well as the new bands identifed for IMT-2020 in WRC'15, which are listed in Table 3.13 [27]. In addition to the duplexing mode FDD, TDD, and SDL, there are six SUL bands with the new duplexing mode defned in Release-15. Among FR1 bands, NR bands n7, n38, n41, n77, n78, and n79 mandate UE capable of supporting four receiving antenna ports to maximize the massive MIMO UE experience [27, 28].

3GPP Release-15 only specifes four-millimeter-wave bands within FR2 as listed in Table 3.14 [29]. All of the Release-15 FR2 bands adopt TDD mode for the convenience of effciently utilizing the DL/UL channel reciprocity for multi-antenna transmission.

3.2.2.2 3Gpp 5G-Nr Band Combination

Due to the large path loss and penetration loss of middle- and high-frequency 5G new bands, C-band (n77, n78, n79) and millimeter-wave bands (n257, n258, n260, n261) have very limited coverage and thus limit the cell-edge user experience throughput of the UEs that access the operated NR base station. Therefore, typical

network operation will combine the new wideband middle- or high-frequency 5G band with the low-frequency band, while the latter band will serve as an anchor carrier to provide a continuous coverage and seamless mobility. Different operators may choose different band combination(s) for their 5G network deployment according to their own spectrum.

Operating band

Uplink (UL) operating band Downlink (DL) operating band Duplex

fUL_low - fUL_high fDL_low - fDL_high mode

n257 26,500 MHz–29,500 MHz 26,500 MHz–29,500 MHz TDD n258 24,250 MHz–27,500 MHz 24,250 MHz–27,500 MHz TDD n260 37,000 MHz–40,000 MHz 37,000 MHz–40,000 MHz TDD n261 27,500 MHz–28,350 MHz 27,500 MHz–28,350 MHz TDD

3.2.2.2.1 5G-Nr Band Combination Mechanisms

3GPP specifed two band combination mechanisms for LTE-Advanced: carrier aggregation (CA) since Release-10 and dual connectivity (DC) since Release-12. For 5G-NR, 3GPP defnes two additional band combination mechanisms, which are LTE/NR DC with typically NR non-stand-alone operation and NR band combination of the normal TDD (or FDD/SDL) band with a SUL band. In total, there are three potential band combination mechanisms for 5G-NR stand-alone deployment and two potential band combination mechanisms for 5G-NR non-stand-alone deployment, described as below: For 5G-NR stand-alone deployment, the following three band combination mechanisms can be adopted. However, only NR-NR DL CA and SUL specifcations are completed within Release-15, while NR-NR DC only specifes few mm-wave and C-band combinations partially. - NR-NR CA: An operated network aggregates multiple NR component carriers (CC) to serve a single UE in order to increase the potential scheduled bandwidth and therefore increase the experience throughput (Fig. 3.36). The data transmission in the secondary CC (SCC) can be scheduled by the control signaling of the primary CC (PCC), and the HARQ feedback of the SCC can be transmitted over the PCC as well. There are three types of NR-NR CA, i.e., intra-band CA with continuous CC, intra-band CA with noncontinuous CA, and inter-band CA, as shown in Fig. 3.36(a), (b), and (c), respectively.

• NR-NR DC: Similar to NR-NR CA, the data streams of single UE can be transmitted over multiple CCs simultaneously to reach a higher bit rate. However, NR-NR DC forces a self-contained data scheduling, transmission, and HARQ procedure within a CC, and requires longer activating and deactivating period for a CC.
• NR band combination for SUL [30]: This band combination combines the NR SUL CC with the normal NR CC(s) that contains DL, which can be one or multiple TDD CC(s), FDD CC(s), or SDL CC(s). The typical operation combination for SUL is to combine with a TDD CC. Different from NR-NR CA, the combined normal NR CC and SUL CC forms a single cell, with all the UL time-frequency resources in the normal CC and SUL CC forming a single pool for a single-cell scheduling, as well as sharing the same HARQ process. Such a mechanism has the beneft of supporting both simultaneous UL carrier switching and good balance between DL capacity and UL coverage, as described in Sect. 3.3.

CC1 CC2 … CCn CC1 CC2 … CCn

(a) Intra-band contiguous-CC CA (b) Intra-band non-contiguous-CC CA

Fig. 3.36 NR-NR CA. (a) Intra-band contiguous-CC CA. (b) Intra-band noncontiguous-CC CA. (c) Inter-band CA

For 5G-NR non-stand-alone deployment, the following two band combination mechanisms were specifed within Release-15. - LTE/NR DC: One network can allow single-UE data transmission of multiple streams over LTE and NR in parallel. The typical operation mode is called EN-DC, i.e., the non-stand-alone NR deployment with LTE as the anchor network to provide the basic coverage and mobility layer, accessing the EPC.

• LTE-FDD and NR SUL band combination: One special case of the LTE/NR DC operation mode is to confgure the NR cell as NR band combination with SUL, while the SUL carrier can be either orthogonal from LTE UL carrier or shared with LTE UL carrier. More details will be introduced in Sect. 3.3.

3.2.2.2.2 5G Band Combination Defnition

3GPP Release-15 only completed the specifcations of some prioritized 5G band combinations for LTE/NR DC and NR band combination for SUL, NR-NR CA, and NR-NR DC, as shown in Tables 3.15, 3.16, 3.17, and 3.18 respectively.

For the aforementioned band combination mechanisms, the coexistence of multiple bands may suffer from severe receiver desensitization due to spurious emissions, and thus the strict RF requirements as well as the corresponding solutions for each

band combinations individually have to be defned. Spurious emissions are emissions which are caused by unwanted transmitter effects such as harmonics emission and intermodulation products. Receiver desensitization is specifed in terms of maximum sensitivity deduction (MSD) in UE specifcations [31].

The intermodulation is the unwanted emission within the downlink transmission bandwidth of the FDD band, generated by the dual-uplink transmissions at different frequencies via the nonlinear transceivers at UE side. Low-order intermediation causes severe interference for some band combinations of LTE/NR DC, CA, and LTE/NR SUL. For those band combinations, single UL transmission at UE side is recommended to avoid the FDD DL performance degradation due to intermodulation.

Harmonic emission is generated by the active uplink in a lower frequency band within a specifed frequency range, whose transmitter harmonics from the UE side fall within the downlink transmission bandwidth assigned in a higher band at the UE receiver. Harmonic can be mitigated by harmonic rejection flter at lower band or avoided by limiting the spectrum resources of the UL transmission in the lowfrequency band, and coordination with the DL scheduled resource allocation in the high-frequency band.

Unlike the usual harmonic issue which is concerned more with UE self-desensitization problem when low-frequency band UL harmonics falls onto high-frequency band DL carriers, some band combinations may be subjected to low-band desensitization when high-band UL carrier is located at 3rd or 5th harmonic of low-band DL carrier, due to the known harmonic mixing problem.

3.2.2.2.3.1 Intermodulation Avoidance With Single Ul Transmission

The intermodulation is the generation of signals in its nonlinear elements caused by the presence of two or more signals at different frequencies reaching the transmitter via the antenna, as shown in Fig. 3.37. For a UE confgured with some band combinations, intermodulation emission may occur, which is regardless of the band combination mechanism including LTE/NR DC, LTE/NR UL sharing, and inter-band CA or DC either in LTE or in NR system. Low-order intermodulation can be so severe that it can cause serious damage for the low-frequency band DL reception. For example, for Band 3 (FDD: DL/1805–1880 MHz, UL/1710–1785 MHz) and Band 42 (TDD: DL and UL/3400–3600 MHz) LTE-CA [33], the LTE DL sensitivity would decrease by 29.8 dB due to the second-order intermodulation interference. Such serious intermodulation issue and the challenge to UE implementation were acknowledged by 3GPP industries for a long time and were hotly discussed during Release-15 5G-NR standardization period.