UMTS Radio Access Techniques for IMT-Advanced

Wireless Signal Processing & Networking Workshop at Tohoku University UMTS Radio Access Techniques for IMT-Advanced M. M. Sawahashi,, Y. Y. Kishiyama,, and H. H. Taoka Musashi Institute of of Technology Radio Access Development Department, NTT DoCoMo, Inc. February 6, 6, 2008 High-Speed Packet Access Based on W-CDMA: HSDPA and HSUPA 3G Long-Term Evolution: Evolved UTRA and UTRAN Broadband Packet Radio Access for IMT-Advanced MIMO Experiments in Broadband Packet Radio Access

2 High-Speed Packet Access Based on W-CDMA: HSDPA and HSUPA

Evolution of W-CDMA in 3GPP R99 W-CDMA R4 R5 HSDPA R6 Enhanced Uplink MBMS R7 MIMO ConCon HSDPA: Enhanced downlink Support for packet-data service Shortened delay High data rate (peak data rate of 14.4 Mbps) High frequency efficiency HSUPA: Enhanced uplink Support for packet-data service Shortened delay High data rate (peak data rate of 5.76 Mbps) High frequency efficiency MBMS (Multimedia Broadcast Multicast Service) Efficiently achieves point-tomultipoint service Broadcast service such as mobile TV HSDPA-MIMO Multi-stream downlink Increase in peak data rate Continuous Connectivity Improved gaming support Shortened setup delays 3

Outline of HSDPA Specification Most specifications on HSDPA were completed as of Release 5 Specifications were finalized as of Release 6 Basic views on designing HSDPA air interface Use functions of W-CDMA air interface specified as Release 99/4 smooth migration from R99 W-CDMA Minimize increase in complexity Development and implementation with low cost Low power consuming UE Objectives of HSDPA High peak data rate up to 14 Mbps High cell throughput (aggregated throughput per cell) and user throughput Focus on streaming (video streaming), interactive (gaming, E-mail, FTP (File Transfer Protocol) download,...), background services Do not support conversation 4

Key Techniques in HSDPA High-Speed Downlink Shared Channel (HS-DSCH) Achieves efficient radio resource usage using shared channel among UEs (sets of user equipment) Enhancement of Downlink Shared Channel (DSCH) HS-DSCH Introduction of short 2-msec transmission time interval (TTI) Scheduling Fast channel-dependent scheduling at Node B according to channel conditions of each UE Node B scheduler also considers QoS requirement of each UE Adaptive rate control (AMC: Adaptive Modulation and Coding) Achieve adaptive rate control according to instantaneous channel conditions of each UE Achieve adaptive rate control by changing modulation scheme and channel coding rate (transport block size) Hybrid ARQ Hybrid ARQ with soft-combining with short round trip time Use incremental redundancy with adaptive and asynchronous operations in Stop & Wait (S&W) protocol 5

Outline of HSUPA Specification Specifications on HSUPA were finalized as of Release 6 Basic views on designing HSUPA air interface Use functions of W-CDMA air interface specified as Release 99/4 Support urban, suburban, and rural regions, from-low-to-high mobility environments (optimize air interface under low mobility) Large performance enhancement with minimum increase in complexity of network and UE implementation Objectives of HSUPA Increase coverage area Increase user throughput and cell throughput Shorten delay (connection delay and transmission delay) Focus on streaming (video streaming), interactive (gaming, E-mail, FTP (File Transfer Protocol) upload,...), and background services 6

Key Techniques in HSUPA Enhancement of DPCH (Dedicated Physical Channel) E-DCH (Enhanced Dedicated Channel) Add short transmission time interval (TTI) of 2-msec Introduce Spreading Factor (SF) of 2 to provide lower PAPR (3.84- Mbps (raw data rate) is provided by 2 x SF of 2) Use transmission power control and soft-handover similar to W- CDMA case Scheduling Permit scheduled UE with data to use as high a data rate as possible without providing excessive interference to other UEs Grant based scheduling: Node B scheduler sends scheduling grant to assign maximum allowed transmission power for E-DCH to maintain the interference level at target value Hybrid ARQ Hybrid ARQ with soft-combining with short round trip time Use incremental redundancy with synchronous and non-adaptive operations 7

8 3G Long-Term Evolution: Evolved UTRA and UTRAN (Super 3G)

Migration to Radio Access Network for IMT-Advanced (4G) Mid-term 3G RAN evolution based on W-CDMA: HSDPA (High-Speed Downlink Packet Access) HSUPA (High-Speed Uplink Packet Access) MBMS (Multimedia Broadcast Multicast Service), etc. Evolved UTRA and UTRAN (Super 3G) Super 3G system will provide support for full IP capabilities Smooth introduction of IMT-Advanced (4G) system Now 3G 3G long-term evolution 3G Super 3G 3G Super 3G 4G (IMT- Advanced) Super 3G 4G UTRA: UMTS Terrestrial Radio Access 9

Evolved UTRA and UTRAN Evolved UTRA and UTRAN represent long-term evolution (LTE) of technology to maintain continuous growth in mobile communications industry Competitive technology even in 4G (IMT-Advanced) era RAN evolution to enable smooth migration to 4G (IMT-Advanced) Evolved UTRA and UTRAN Commercial deployment: Accomplished successfully Specifications and system development: Ongoing Launch 3G Mid-term evolution Long-term evolution 4G (IMT-Advance 2G 1G 1980s 1990s 2000s 2010s 2020s 10

11 Standardization of E-UTRA and UTRAN in 3GPP Study Item (SI) investigation SI investigation started in RAN in December 2004 Requirements were specified as TR25.913 in June 2005 Physical layer issues addressed by WG 1 (Working group 1) were specified as TR25.814 in September 2006 Work Item (WI) investigation WI investigation started from June 2006 Major WI specifications in WG1 were completed by September 2007 WI specifications were completed as Release 8 in December 2007, but minor changes in specifications are ongoing

Requirements for Evolved UTRA and UTRAN Spectrum Support of scalable bandwidths, i.e., 1.4, 3, 5, 10, 15, and 20 MHz Packet-switching (PS) mode only VoIP capability in PS domain Latency Short C-plane latency (transition time) - Idle to active: Less than 100 msec - Dormant to active: Less than 50 msec U-plane latency - Latency in RAN is less than 5 msec one way Peak data rate DL: 100 Mbps, UL: 50 Mbps User throughput (relative to Rel. 6 HSDPA, HSUPA) Cell edge user throughput: 2-3 times (DL), 2-3 times (UL) Average user throughput: 3-4 times (DL), 2-3 times (UL) Spectrum efficiency (relative to Rel. 6 HSDPA, HSUPA) 3-4 times (DL), 2-3 times (UL) 12

Frame Structure in Evolved UTRA Radio frame = 10 ms 1 sub-frame = 2 slot = 1 ms transmission time interval (TTI) 1 slot = 0.5 ms Slot 1 Slot 2 Slot 3 Slot 4 Slot 19 Slot 20 Symbol 1 Symbol 2 Symbol 7 CP Copy Effective data Short sub-frame length Adopted 1-msec sub-frame length to achieve short round trip delay (RTD) Common frame structure between FDD and TDD Inserted cyclic prefix (CP) at each FFT block to avoid inter-block interference in both DL and UL Defined sub-frame with long CP to provide MBMS (Multimedia Broadcast Multicast Service) with single-frequency network (SFN) in DL 13

OFDM-Based Downlink Radio Access Robust against multipath interference (MPI) Flexibly accommodates different spectrum arrangements High affinity to advanced techniques Frequency domain channel-dependent scheduling MIMO multiplexing/diversity High quality reception using soft-combining for MBMS signal Resource block (RB) Localized transmission Frequency Sub-frame (Different colors represent different users) Time Distributed transmission 14

Single-Carrier-Based Uplink Radio Access Single-carrier based FDMA access in uplink Low PAPR (Peak-to-average power ratio) achieves wide area coverage using limited transmission power Localized and/or distributed FDMA establishes intra-cell orthogonality among simultaneous transmitting users in frequency domain Employs frequency domain equalizer with cyclic prefix to suppress MPI Coded data symbol DFT Sub-carrier mapping IFFT Addition of CP Transmit signal Subframe Resource unit Frequency scheduling Freq. DFT-Spread OFDM Time Frequency hopping Contiguous resource units are assigned to UE to maintain the single-carrier property 15

Radio Access Comparisons Between W-CDMA and E-UTRA W-CDMA Downlink Partially orthogonal Achieves orthogonality within a cell using Walsh-Hadamard OVSF sequences However, multipath interference impairs performance (note that mandatory function is coherent Rake receiver) Uplink Non-orthogonal Non-orthogonal transmission using user-specific scrambled sequence at asynchronous-timing reception Evolved UTRA Downlink Orthogonal Achieves orthogonality within a cell in frequency domain for the delayed paths within cyclic prefix duration Uplink Orthogonal Achieves orthogonality within a cell in frequency and/or code domain in SC-FDMA associated with frequency-domain equalizer 16

Radio Access Features in Evolved UTRA Inter-Node B (cell) synchronization Air interface supports inter-cell asynchronous mode as baseline However, utilize merits of synchronized operation for MBMS with SFN and inter-cell interference coordination, etc. Cell search: Process to search for the best cell with the minimum path loss DoCoMo proposed SCH (Synchronization Channel) and Primary BCH (Broadcast Channel) structures appropriate for fast cell search in scalable transmission bandwidth from 1.4 to 20 MHz Reference signal (RS) Used for channel estimation and channel-quality measurement Orthogonal RSs between transmitter antennas in MIMO Data modulation DL: QPSK, 16QAM, 64QAM UL: QPSK, 16QAM, 64QAM (Optional) Channel coding Turbo code (Data channel) and convolutional code (Control signals) Hybrid ARQ with packet combining Incremental redundancy (IR) 17

Link Adaptation TPC (Transmission power control) AMC (Adaptive Modulation and Coding) QPSK (2 bit/symbol) Q Use 1 Low transmission power for user experiencing good conditions User 2 High transmission power for user experiencing poor conditions Applied to constant-rate service such as real-time traffic and control channel Satisfies the constant required received SNR for channel variation Affects other-cell interference Q 64QAM (6 bit/symbol) I User 1 High data rate transmission User 2 Low data rate transmission Applied particularly to variable-rate services such as non-real-time traffic Rate adaptation according to channel conditions Achieves higher capacity, i.e., cell throughput, than TPC 18

19 MBMS Using Soft-Combining in SFN Efficient MBMS with soft-combining in single frequency network (SFN): Received paths from multiple cell sites within a cyclic prefix are combined without yielding mutual interference BS #1 User #1 BS #3 From BS #1 Received power From BS #2 Cyclic prefix length BS #2 User #2 From BS #3 Soft-combining Time

Inter-cell Interference Mitigation Inter-cell Interference Mitigation Randomization using cell-specific scrambling code associated with channel coding and repetition to achieve one-cell frequency reuse Interference cancellation: Interference canceller, MMSE receiver etc. Interference coordination: Autonomous scheduling to establish fractional frequency reuse (fast inter-node B communication should be avoided) Inter-cell interference coordination Introduction of partial multi-cell frequency reuse near cell edge One-cell frequency reuse for user near cell site Multi-cell frequency reuse for a UE near cell edge using FDMA to reduce other-cell interference Optimize balance between cell throughput and user throughput at cell edge 20

21 MIMO Channel Techniques MIMO multiplexing (MIMO SDM) Baseline is 2-by-2 MIMO in downlink and 1-by-2 SIMO in uplink Single-user MIMO and multiuser MIMO (single-user MIMO in uplink is FFS) Major techniques applied Pre-coding Single/double coded streams (code words) Rank adaptation (selection of MIMO mode) based on channel measurement MIMO diversity (transmit diversity) Supports transmit diversity, which is suitable for each physical channel property DL: PVS (Precoding vector switching) for SCH, SFBC (Space Frequency Block Code) based scheme for control channels, and CDD (Cyclic Delay Diversity) for MBMS UL: TSTD (Time-Switched Transmit Diversity) for RACH and ASTD (Antenna Selection Transmit Diversity) for data channel (Option) Adaptive beam-forming Adaptive beam-forming is effective in increasing coverage area particularly for large cells

22 Broadband Packet Radio Access for IMT-Advanced

23 Requirements for IMT-Advanced 4G systems are real broadband wireless networks Packet-based radio access networks (RANs) provide equal or greater peak data rates than those for wired networks while maintaining equivalent QoS Technical requirements Reduced network cost (cost per bit) Higher capacity (frequency efficiency) Ex. frequency efficiency of 4 times that of Evolved UTRA Wider area coverage Equal or wider than that in 3G Simpler deployment and network operation Better service provisioning Higher user data rate Ex. peak data rate of greater than 1 Gbps with wider bandwidth Very low latency (connection and transmission delays) Efficient unified support of all traffic types with precise QoS control Complement to 3G (legacy) systems Flexible spectrum usage Ex. scalable transmission bandwidth up to 100 MHz Handover with 3G system and backward compatibility with existing 3G and legacy systems

24 Migration to IMT-Advanced W-CDMA, HSDPA/HSUPA 3GPP standardization: from 1998, Commercial service launch: W- CDMA 2001, HSDPA 2006 Peak data rate wide area: 1 Mbps, local area: 10 Mbps Mobility focus Long-term 3G evolution (Evolved UTRA and UTRAN) 3GPP standardization: from 2004, Commercial service launch: targeting 2009 2010 Peak data rate wide area: 10 Mbps, local area: 100 Mbps Supports only packet domain Further improvement in throughput particularly at cell edge IMT-Advanced (4G) Peak data rate wide area: 100 Mbps, local area: 1 Gbps Extend 3G LTE capability and performance New radio interfaces

Duplex Scheme Should support both paired and unpaired bands to achieve global air interface TDD Merit: Pair-band is unnecessary Demerit: Increase in RTD (round trip delay) and decrease in coverage area Slots for UL Repetition is possible only over UL slot assignment duration Limits coverage Slots for DL Retransmission using the next DL slot assignment duration Increase in RTD FDD Merit: Short RTD and provides wide area coverage Demerit: Pair-band is necessary Restrictions on pair-band assignment can be relaxed using asymmetric spectrum assignment according to the traffic demand between DL and UL Compared to TDD, FDD is more promising based on system requirements such as a short delay and wide area coverage, although TDD is more advantageous from the spectrum allocation viewpoint. 25

Key Radio Access Techniques for IMT-Advanced (1) Multi-access Supports Evolved UTRA and achieves backward compatibility with 3G systems Accommodates frequency spectrum with various bandwidths scalable transmission bandwidths Employs adaptive multi-access scheme according to deployment environment, QoS requirements, etc. (2) Intra- and inter-cell site (BS) orthogonality Intra-cell site orthogonality: Orthogonal multi-access in frequency/code domains within the same cell site similar to E-UTRA Inter-cell site (quasi) orthogonality: Inter-cell site fast radio resource management (i.e., Interference management) (3) Adaptive multi-antenna transmission/reception Mode (rank) selection adapts according to deployment environment and QoS using the same antenna configuration: MIMO multiplexing, MIMO diversity, pre-coding, and beam forming (4) Enhanced site diversity to extend area coverage Remote-BS using optical fiber, relay BS using radio, etc. 26

Adaptive Multi-access Scheme Multi-Carrier (MC)/Single-Carrier (SC) hybrid (Adaptive multi-access control) Universal switching of MC/SC based access using frequency domain multiplexing/de-multiplexing Optimizations of PAPR (coverage) and achievable peak data rate according to inter-site distance, cell structure, and QoS requirements SC generation Switch DFT Coded data symbols Subcarrier mapping Pulseshaping filter IFFT CP insertion MC generation S/P [Ref] R. Dinis et al., A Multiple Access Scheme for the Uplink of Broadband Wireless Access, IEEE Globecom, Dec. 2004. 27

28 Adaptive Radio Access Concept Control of multi-access schemes adapts according to deployment environment and required QoS Cell distance, i.e., wide area or local area For wide area coverage, low PAPR feature is necessary SC based multiple access is appropriate QoS on peak data rate and packet error rate (PER) To provide high data rate, robustness against MPI is necessary MC based multiple access is appropriate Wide area Local area (Heavy traffic in small area)

Inter-cell Site Orthogonality One-cell frequency reuse Baseline is one-cell frequency reuse to achieve high system capacity Intra-cell site (Intra-BS) orthogonality Achieves intra-cell site orthogonal multi-access (multiplexing) in both links as well as in E-UTRA Inter-cell site orthogonality Although inter-cell interference coordination (ICIC) is adopted in E-UTRA, it only introduces fractional frequency reuse at cell edge, which is semistatically controlled with slow control speed Thus, inter-cell site orthogonality will be established in IMT-Advanced to achieve high frequency efficiency and high data rate at cell edge Intra-cell site Inter-cell site DL W-CDMA E-UTRA IMT-Advanced (Partially) orthogonal Orthogonal Orthogonal UL Non-orthogonal Orthogonal Orthogonal DL Non-orthogonal Non-orthogonal (Quasi)-orthogonal UL Non-orthogonal Non-orthogonal (Quasi)-orthogonal 29

Adaptive Inter-cell Interference Management Achieve inter-cell site orthogonality through adaptive inter-cell interference management Centralized inter-cell interference management among remote-bss using optical fiber belonging to the same control BS achieves complete inter-cell site orthogonality Autonomous inter-cell interference management among independent cell sites using control signals via backhaul and/or air achieves inter-cell site quasi-orthogonality through slow adaptive control of partial multi-cell frequency reuse at the cell edge Autonomous inter-cell interference control Centralized inter-cell interference control 30

31 MIMO Channel Transmissions for IMT-Advanced (1) MIMO channel techniques in IMT-Advanced are more important than those in 3G systems (HSDPA, Evolved UTRA, etc.) High-order MIMO channel transmissions Larger number of antennas Number of transmitter/receiver antennas, which must be supported, is 4 in IMT-Advanced 2 in E-UTRA May be possible due to higher carrier frequency and focus on microcell with short inter-site distance Maximum number of antennas is e.g., 8-12. Adaptive MIMO channel transmission Mode selection according to different requirements/targets Adaptive rank control: Adaptive control of transmitted streams according to channel conditions Adaptive rate control through modulation and coding rates (PARC)

32 MIMO Channel Transmissions for IMT-Advanced (2) MIMO multiplexing (SDM, multi-layer transmission) Increase in the peak data rate, i.e., frequency efficiency beyond 10 bit/second/hz High capacity through SDM among multiple UEs MIMO diversity (transmit diversity) Improvement in received quality and coverage area Key question: Achievable diversity gain is offset by increasing overhead of orthogonal reference signals according to increasing number of transmitter antennas. What is the optimum number of transmitter antennas? Adaptive beam forming Increase in coverage and achievable data rate at cell edge High capacity through SDM among UEs Key problem: Increase in coverage of common/shared control channels.

33 MIMO Experiments in Broadband Packet Radio Access for IMT-Advanced

34 Target Peak Data Rates for IMT-Advanced Target peak data rate is one of the most important requirements in radio access systems Targets data rates specified in standardization or forum ITU-R Recommendation M.1645 Peak data rate of 100-Mbps in new mobile access under high mobility Peak data rate of 1-Gbps in new nomadic/local area wireless access under low mobility IST-2003-507581 in WINNER D7.1 v1.0 System Requirements (2004.07.16) Peak spectral efficiency in connected sites of 10 b/s/hz/site in wide area deployments for heavy traffic loads Peak spectral efficiency in isolated (non-contiguous) sites of 25 b/s/hz/site

35 Series of Experimental Demonstrations for IMT-Advanced Radio Access by DoCoMo Experimental demonstrations of target peak data rates for IMT- Advanced May 2003: Achieved 100 Mbps transmission in field experiments at the speed of 30 km/h in downtown Yokosuka Peak data rate of 135 (300) Mbps using 16QAM (64QAM) modulation and Turbo code with R = 1/2 (3/4) Aug. 2004: Achieved 1 Gbps transmission with 4-by-4 MIMO SDM in laboratory experiments using fading simulators (10 bit/second/hz) May 2005: Achieved 1 Gbps transmission in field experiments at the speed of 30 km/h in downtown Yokosuka Peak data rata of 1.028 Gbps using 16QAM modulation and Turbo code with R = 8/9 Dec. 2005: Achieved 2.5 Gbps transmission with 6-by-6 MIMO SDM in field experiments at the speed of 10-30 km/h in YRP district (25 bit/second/hz) Peak data rata of 2.556 Gbps using 64QAM modulation and Turbo code with R = 8/9

36 5-Gbps Packet Transmission (Frequency efficiency of 50 bit/second/hz)

Features of Experimental Configuration 37 (1) OFDM radio access with 100-MHz transmission bandwidth (2) Efficient modulation and channel coding scheme 64QAM modulation Turbo code with coding rate of R = 8/9 Multiple codewords (3) 12-by-12 MIMO multiplexing (4) MLD-based signal detection QRM-MLD [1] with ASESS [2] (adaptive selection surviving symbol replica candidates based on maximum reliability) LLR (log-likelihood ratio) generation appropriate for QRM- MLD [1] K. J. Kim, et al., IEEE Trans. on Wireless Commun., vol. 4, no. 2, pp. 710-721, March, 2005. [2] K. Higuchi, et al., in Proc. IEEE Globecom 2004, Nov. 2004.

Major Parameters for Field Experiments NTT DoCoMo Radio Proprietary access Carrier frequency Channel bandwidth Sub-frame length Number of sub-carriers OFDM symbol duration Data modulation Channel coding / decoding Number of antennas Information bit rate OFDM symbol timing detection Channel estimation Signal detection OFDM 4.635 GHz 101.4 MHz 0.5 msec 1536 (65.919 khz sub-carrier separation) Effective data 15.170 μsec + CP 2.067 μsec (2048 + 279 samples) 64QAM Turbo coding (R = 8/9, K = 4) / Max-Log-MAP decoding 12-by-12 MIMO 4.92 Gbps Pilot symbol-based symbol timing detection Pilot symbol-based two-dimensional MMSE channel estimation QRM-MLD with ASESS 38

Effect of Antenna Separation Between Receiver Antennas Throughput (Gbps) 5 4 3 2 1 Rx antenna separation d = 10 cm (1.5λ) d = 20 cm (3.1λ) d = 40 cm (6.2λ) Simulation Field experiments 0 24 25 26 27 28 29 30 Average total received SNR per receiver antenna (db) 12-by-12 MIMO multiplexing 64QAM, R = 8/9 (Max: 4.92 Gbps) Antenna separation: D = 70 cm Average speed: v = 10 km/h When d is reduced from 40 cm to 10 cm, the loss in the required received SNR is only 0.5 db. Achieved 4.9-Gbps throughput at average received SNR of approximately 28.5 (29) db when d is 40 (10) cm. Loss in the required average received SNR compared to simulation is approximately 1 db. 39

Effect of Moving Speed of Mobile Station (Throughput Performance) Throughput (Gbps) 5 4 3 2 1 v = 10 km/h v = 20 km/h v = 30 km/h Simulation Field experiments 0 24 25 26 27 28 29 30 Average total received SNR per receiver antenna (db) 12-by-12 MIMO multiplexing 64QAM, R = 8/9 (Max: 4.92 Gbps) Antenna separation Transmitter: D = 70 cm Receiver: d = 20 cm Even when v is 30 km/h, the loss in the required received SNR is only 1 db. Achieved 4.9-Gbps throughput at average received SNR of approximately 29.5 db when v is 30 km/h. 40

41 Conclusion Evolved UTRA and UTRAN, which adopts competitive technology even in 4G era, will achieve smooth migration to 4G systems WI specifications were completed last December IMT-Advanced systems will be real broadband packet-based radio access networks (RANs) that provide equal or greater peak data rates than those for wired networks while maintaining equivalent QoS (delay and quality) Adaptive control of multi-access schemes and major radio parameters, intra- and inter-cell site interface management, higher-order MIMO schemes etc. are promising. Field experiments demonstrated the feasibility of peak throughput of 5 Gbps, i.e., corresponding frequency efficiency of 50 bit/second/hz, which is close to the ultimate frequency efficiency in cellular environments