Full-Dimension MIMO: Status and Challenges in Design and - PowerPoint PPT Presentation

Full-Dimension MIMO: Status and Challenges in Design and Implementation Gary Xu, Yang Li, Young-Han Nam and Taeyoung Kim and Ji-Yun Seol Charlie Zhang DMC R&D Center, Samsung Electronics Co., Ltd. Samsung Research America (Dallas) May 27, 2014 1

Outline Current Status of FD-MIMO 1 Challenges of FD-MIMO 2 2

Background of Full-Dimension MIMO • Theory Behind: Massive MIMO* – Spatial resolution increases as number of eNB antennas – Narrow beam transmission with little MU interference • Active Antenna Array (AAA) – 2D vs. 1D AAA 3 *Marzetta , “Non - cooperative cellular wireless with unlimited numbers of base station antennas,” IEEE TWireless Nov. 2010

Full-Dimension MIMO (FD-MIMO) FD-MIMO simultaneously supports elevation & azimuth beamforming and > 10 UEs Elevation MU-MIMO FD-MIMO beamforming eNB Azimuth beamforming 10x 2-dimentional AAA & ~100 antennas Capacity 3-5x MU-MIMO with 10s of UEs 3D-spatial channel model 4 Rel-10 FD-MIMO FD-MIMO 32-64 Tx 100 Tx

FD-MIMO 2D AAS Form Factor Examples 0.5m 0.25m 0.5m λ /2 λ /2 λ /2 λ /2 2 λ 0.25m λ /2 λ /2 0.5m Eg.3: 8x8 array with cross-pol. 1m Digital beamforming 64 elements. Small Cell λ /2 Eg.2: 8x8 array with full digital beamforming across 64 elements Urban Micro 0.5m Eg. 1: 8x4 array with each element Eg. 1: 8x8 array with full digital Eg.4: 1x8 array. 4 antennas with analog beamforming beamforming across 64 elements Digital elevation λ =12cm Urban Macro beamforming @ 2.5GHz FD-MIMO antenna panel form factor is well within practical range

Industry Status and 3GPP roadmap 3GPP development Expected start of Elevation Beamforming Complete 3D channel Start EB/FD-MIMO (EB)/FD-MIMO SI model (SI) work item (WI) Complete channel& Start 3D channel EB/FD-MIMO SI WID Complete baseline calibration model study item (SI) Completion (Dec 2016) 2016 2014 2015 2013 PoC for Small Cell PoC for Macro Cell 1 st FD-MIMO Prototype: 32 antenna LTE base-station 6

3-Dimension (3D) Channel Model In SCM, channel is a composite response of cluster/subclusters to Tx/Rx antennas: • Number of cluster/subclusters • Delay of clusters • Power of clusters • Phases (due to e.g. reflection) • Angle of Departure/Arrival (AoD/AoA) Note: Spatial channel model (SCM) AoA/AoD critically determines channel correlations 2D model assumes all clusters zero elevation angles and cannot describe elevation differences. 3D model captures elevation angles and thus clusters can be distinguished in elevation domain 7

Statistics in 3GPP 3D Channel Model 1 1 Urban Macro Urban Macro 0.9 0.9 Urban Micro Urban Micro 0.8 0.8 0.7 0.7 0.6 0.6 cdf cdf 0.5 0.5 0.4 0.4 0.3 0.3 0.2 0.2 0.1 0.1 2 Rx antennas (+ pol.) 0 0 70 75 80 85 90 95 100 105 110 115 120 0 5 10 15 20 25 30 35 40 45 50 Elevation angle of depature (degree) Channel condition number (dB) • Elevation angle (w.r.t. zenith) has a range of 30-deg for UMa and 50-deg for UMi • 80% channels have condition number > 5dB Note : see “ 3GPP TR 36.873 ” for more details of 3D channel model and UE distribution. 8

System-Level Simulator (SLS) Evaluation Simulation Setup: • 3D ITU, UMa Overhead: 20% • 57 sectors with K=10 UEs per sector • Center frequency 2GHz, bandwidth 10MHz Ideal SRS estimation • UE speed 3km/h or 30km/h, uniformly 4 ms scheduling delay distributed Normalized by # of DL subframes • 40 drops, 4s per drop Baseline: SU-MIMO with rank1 • UE: 2 Rx, 1Tx 9

SLS Simulation Results (Up to 4 UE MU-MIMO) Average cell throughput (bps/Hz) Cell-edge throughput (bps/Hz) 0.3 10 83% 58% 8 0.2 6 0.298 99% 127% 8.04 4 0.163 0.1 5.09 0.082 2 2.24 0 0 3 km/h 4Tx 32Tx 32Tx 4Tx 32Tx 32Tx 4x2 32x2: 2 UE 32x2: 4 UE 4x2 32x2: 2 UE 32x2: 4 UE Average cell throughput (bps/Hz) Cell-edge throughput (bps/Hz) 0.2 7 6 0.16 68% 220% 5 0.12 4 91% 0.17 18% 6.26 0.08 3 3.73 2 0.04 0.053 0.045 1.95 1 30 km/h 0 0 4Tx 32Tx 32Tx 4x2 32x2: 2 UE 32x2: 4 UE 10 4Tx 32Tx 32Tx 4x2 32x2: 2 UE 32x2: 4 UE

Outline Current Status of FD-MIMO 1 Challenges of FD-MIMO 2 11

FD-MIMO Framework in LTE/LTE-A 12

Antenna Virtualization & CQI Prediction Issue: (1) How to generate wide beam from a large array (2) CQI (channel quality indicator) mismatch UE-specific beamforming Cell-wide beamforming by antenna virtualization • Wide-beam (ant. virtualization) for control signal (coverage) • Narrow-beam (precoding) for data signal • UE CQI* is measured based on wide-beam *CQI is a UE feedback value and is essential for eNB to decide transmission scheme, code rate, modulation for each UE. 13

Antenna Virtualization & CQI Prediction (2) CQI prediction Synthesized antenna virtual pattern (32 ant.) 0 12 180 30 330 120 10 60 300 𝜍 𝑞𝑠𝑓𝑒𝑗𝑑𝑢 = |𝒊 𝑙 𝒙 𝑙 | 2 60 8 |𝒊 𝑙 𝒙 0 | 2 𝜍 𝑛𝑓𝑏𝑡𝑣𝑠𝑓𝑒 90 270 0 6 -60 4 120 240 Mean of error: 0.0941 -120 2 150 210 -180 180 Var. of error: 0.0424 0 14 180 120 60 0 -60 -120 -180

FD-MIMO in TDD: Antenna Calibration 𝑠 1 LNA eNB transceiver Ant 1 𝑢 1 1 PA 𝑠 2 LNA eNB H transceiver Ant 2 2 𝑢 2 PA Reciprocal … … 𝑠 𝑁 LNA eNB Downlink transmission Ant M transceiver M 𝑢 𝑁 PA Uplink sounding Calibration requirement: Challenges: Joint Tx/Rx calibration 𝑢 1 − 𝑠 1 =•••= 𝑢 𝑁 − 𝑠 M • Inherent error in calibration circuit • Complexity grows with antennas Independent Tx/Rx calibration • Prefer independent Tx/Rx calibration 𝑢 1 =•••= 𝑢 𝑁, 𝑠 1 =•••= 𝑠 M 15

Front-haul Complexity Number of sectors 3 System bandwidth (MHz) 20 Sampling rate (Msps) 30.72 Bit width per I/Q-branch 16 Number of TX antennas (paths) 32 ~96 CPRI throughput (Gbps) CPRI Possible Solutions: • Front-haul (CPRI) compression • New baseband architectures 16

FD-MIMO in FDD: Exploit Uplink Correlation Issue: CSI (channel state information) acquisition  CSI acquired by training & feedback in FDD LTE/LTE-A  Pilot & feedback bits proportional to # of Tx antennas Possible to use uplink channel for downlink precoding? Uplink Downlink  Uplink & downlink channels are correlated  FD-MIMO can measure uplink better *Sana Salous and Hulya Gokalp , “Medium - and Large-Scale Characterization of UMTS-Allocated Frequency Division Duplex Channels”, IEEE TVT. 17

FD-MIMO in FDD: Exploit Uplink Correlation (2) • Duplex distance: 45 MHz. Channel condition: NLOS. Downlink: 2300 MHz; uplink: 2250MHz Angle between eigenvector of downlink & uplink WCS Band 50 45 Angle between Downlink and Uplink Beams 40 35 30 25 20 15 10 5 0 100 200 300 400 500 600 700 800 900 1000 Block Index 𝐐𝐬 𝐁𝐨𝐡𝐦𝐟 < 𝟒𝟏 𝟏 = 𝟘𝟘. 𝟕% 𝐃𝐩𝐬𝐬𝐟𝐦𝐛𝐮𝐣𝐩𝐨 𝐃𝐩𝐟𝐠𝐠𝐣𝐝𝐣𝐟𝐨𝐮 = 0. 0.995 PMI * : Highly correlated CQI : Highly correlated 18 *PMI (Precoding Matrix Indicator): quantized channel direction.

Other Challenges in FD-MIMO Feedback and codebook design in FDD • How to reduce overhead by exploiting channel correlation in azimuth and elevation domain? • Possible to combine with uplink measurement to provide better accuracy? Uplink sounding in TDD • How to accurately estimate a large number of channels? • How to reduce channel estimation complexity? Scheduling & precoding complexity • How to optimally schedule ~10 MU-MIMO UEs without exponentially increasing complexity? 19

Summary FD-MIMO simultaneously supports • Full-dimension MIMO is a promising technology to elevation & azimuth beamforming and > 10 UEs MU-MIMO Elevation significantly improve cellular capacity (by x3-5) FD-MIMO beamforming eNB • Challenges ahead include system design and implementation Azimuth beamforming • “ The Next Big Thing is Here ” in wireless industry 20

THANK YOU! 21