Performance Studies on LTE Advanced in the Easy-C Project 19.06.2008 Andreas Weber, Alcatel Lucent Bell Labs All Rights Reserved Alcatel-Lucent 2007
Agenda 1. Introduction 2. EASY C 3. LTE System Simulator 4. Results 5. Conclusions and Outlook 2 LTE System Simulation June 2008
Introduction EUTRAN (Evolved Universal Terrestrial Radio Access Network) also called LTE (Long Term Evolution) is the upcoming standard for packet switched based mobile communication LTE physical layer is based on OFDMA in the DL and SC-FDMA in the UL The scope of EASY C is beyond LTE -> LTE Advanced EASY C field trials are accompanied by system simulations Candidate algorithms shall be evaluated before the real system is implemented Accuracy of simulations can be evaluated by comparison with measurements 3 LTE System Simulation June 2008
EASY C Overview EASY C Project topics / objectives BMBF project 3 year project / start Q2/2007 Preparation of a new Standard: LTE Advanced Focus on improved spectral efficiency, cell border throughput, fairness, and latency Field trials with optimized MIMO algorithms Project partners: 4 LTE System Simulation June 2008
EASY C Field Trial Phasing Step 1: Basic LTE Release 8 system SU-MIMO MU-MIM0 in UL Step 2: Enhancements above Release 8 Remote Radio Heads Enhanced receivers Optimized codebooks Beam Forming MU-MIMO in DL Step 3: Collaborative MIMO Schemes Network MIMO Cooperative scheduling Interference coordination 5 LTE System Simulation June 2008
EASY C Test Campus Dresden Test Ues, partly for UL interference One to max. 3 sites will be equipped with enodebs Allows realistic tests without and with mobility up to 30 km/h AGW simulator 16 Mb/s IP connection to NodeB (S1) Surrounding cells are equipped with interferers Signalion SORBAS based + boosters Multi-cell scenario included for MIMO/ multi-cell and interference co-ordination 6 LTE System Simulation June 2008
EASY C System Simulation Approach Field tests shall be accompanied by system simulations Evaluation of candidate algorithms Evaluation of accuracy of simulation models System Simulations shall be 3GPP/NGMN compliant (TR 25.814, R1-070674) Full simulation of interference Wrap around Spatial channel model Full buffer simulation Results shall be realistic (channel estimation loss model,...) First phase: Calibration of simulators of different partners (1x2 in DL and UL) Second phase: Reference model results (2x2 in DL, 1x2 in UL) Spectral Efficiency User throughput CDF, fairness Cell border throughput Third phase: Simulation of algorithms Substitution of spatial channel model with ray tracing data channel measurements 7 LTE System Simulation June 2008
LTE System Simulator Objectives Evaluation of LTE system performance in UL and DL Antenna systems 2x2, 4x2, 4x4,... correlated antennas uncorrelated antennas mixture of correlated and uncorrelated antennas Algorithms Scheduler Link Adaptation Interference Coordination Combination of performance enhancing technologies Optimization of algorithms that are impacted by spatial channel behavior 8 LTE System Simulation June 2008
LTE System Simulator Reminder: DL LTE Channel Structure f Subcarrier BW [MHz] Nr. PRB one OFDM symbol first 1...3 OFDM Symbols reserved for L1-L2-Signaling Physical Resource Block (PRB) 1.4 5 6 25 14 OFDM Symbols x 12 Subcarrier 10 50 15 75 20 100 PRB 15 khz Resource Element Slot (0.5 ms) Slot (0.5 ms) Subframe (1 ms) t 9 LTE System Simulation June 2008
LTE System Simulator Detailed Features Features Spatial channel model (WiM, Winner Model) generates spatial fast fading Full simulation of interference, i.e. SCM is used for all channels Event driven simulation on resource element basis, i.e. per subcarrier (in frequency) and per OFDM symbol (in time), lower granularity possible Monte Carlo drops in order to get a quicker randomization of mobile positions (during drop path loss and shadowing is kept constant) Link to system interface based on MIESM (Mutual Information Effective SINR Mapping) Receiver is explicitly modeled (MMSE or MRC) 1x1, 1x2, 2x2, 4x2, 4x4 TX/RX antennas Single and multiple stream transmissions (e.g. PARC and SDMA) Switching between single stream and multiple stream transmission 10 LTE System Simulation June 2008
LTE System Simulator Detailed Features Features (continued) Frequency selective and diverse allocation Different schedulers CQI generation, CQI reporting delay, CQI reporting period, CQI filtering Ideal and realistic link adaptation Asynchronous, adaptive HARQ (DL) and synchronous HARQ (UL) with feedback delay Transport blocks consisting of an arbitrary number of PRBs BLER calculation on transport block basis (with chase combining and IR) Signaling overhead Pilot symbol patterns (for 1, 2, 3, and 4 antennas) Full and soft fractional frequency reuse Large number of measurement values 11 LTE System Simulation June 2008
LTE System Simulator Spatial Channel Model BS array Cluster n Subpath m nm,, AoD θ n, m, AoA δ n, AoA nm,, AoA N Ω MS θv v N δ n, AoD Ω BS θ MS θ n, m, AoD MS array broadside MS array θ BS BS array broadside MS direction of travel source: 3GPP TR 25.996 Example: urban macro: 6 paths with 20 subpaths each 12 LTE System Simulation June 2008
LTE System Simulator Fast Fading for OFDM OFDM receive signal OFDM Relative Receive Signal Level relative amplitude relative Amplitude Veh B 1.6 1.4 1.2 1 0.8 0.6 0.4 0.2 0 1.6 1.4 1.2 1 0.8 0.6 0.4 0.2 0 0 20 40 subcarrier 60 80 100 120 0 10 20 30 40 50 60 70 time [5ms] 80 90 100 13 LTE System Simulation June 2008
LTE System Simulator SINR over Frequency and Time 14 LTE System Simulation June 2008
LTE System Simulator Wrap Around Mirror 2 Mirror 1 Wrap Around avoids border effects Mirror 6 every BTS has six mirrors Mirror 3 every mobile claims to be in the middle of 2 rings of BTS sites Mirror 4 Mirror 5 15 LTE System Simulation June 2008
LTE System Simulator Wrap Around 16 LTE System Simulation June 2008
LTE System Simulator Wrap Around in case of Frequency Reuse Border effects with wrap around if reuse factor is not divisor of the number of cells Solution: Simulation with 21 sectors or restriction of evaluation to inner cells 17 LTE System Simulation June 2008
LTE System Simulator Connect all Mobile and Transceiver Antennas Channel Channel Channel Channel Channel Channel Channel Channel Channel Channel Channel Channel Example: 57 sectors, 10 mobiles per sector, TX and RX diversity: 57 * 2 * 57 * 10 * 2 = 129,960 channels 18 LTE System Simulation June 2008
Results Calibration: Geometry (exemplary) User Geometry CDF 1.0 0.9 0.8 0.7 Cumulative Probability 0.6 0.5 0.4 0.3 21 sectors, 500m ISD 21 sectors, 1732 m ISD 57 sectors, 500 m ISD 57 sectors, 1732 m ISD 0.2 0.1 0.0-10 -5 0 5 10 15 20 Wideband SINR [db] User Geometry = E[S]/(E[I] + R*N) S = Signal Level, I = Interference Level, R = Receiver Noise Figure 19 LTE System Simulation June 2008
Results Reference Simulations: Exemplary results DOWNLINK Antenna Configuration Inter Site Distance [m] Spectral Efficiency [bits/s/hz] 5-Percentile of UE Throughput [kbit/s] 1x2 1732 1.28 204 1x2 500 1.38 324 2x2 1732 1.37 255 2x2 500 1.46 345 UPLINK* Antenna Configuration Inter Site Distance [m] Spectral Efficiency [bits/s/hz] 5-Percentile of UE Throughput [kbit/s] 1x2 500 0.97 295 1x2 1732 0.85 57 wi thout IoT control -> high spectral effic iency, low edge user throughput 20 LTE System Simulation June 2008
Results Step 2 Candidate: Adaptive 4x2 SU-MIMO λ/2 UE velocity A high B low C Polarisation beams + Alamouti Polarisation beams + Closed-loop Tx diversity Polarisation beams + Spatial Multiplexing low high SINR 21 LTE System Simulation June 2008
Results Step 2 Candidate: Adaptive 4x2 SU-MIMO 600 Comparison of different Antenna Systems and Precoding Matrices, 500m ISD 500 500 1x1 Single Antenna TX 500 1x2 Single Antenna TX 500 2x2 CL TX Div & PSRC (36.211) optimized codebook Cell Border Throughput [kbit/s] 400 300 200 500 4x2 CL TX Div & PSRC (36.211) 500 4x2 Directional CL TX Div & PARC, 4 Beams, 4 Weights 500 4x2 Directional CL TX Div & PARC, 16 Beam, 8 Weights 1x2 1x1 2x2 36.211 codebook 4x2 100 0 1.0 1.1 1.2 1.3 1.4 1.5 1.6 1.7 1.8 1.9 2.0 Spectral Efficiency [bit/s/hz/sector] 22 LTE System Simulation June 2008
Results Step 2 Candidate: Adaptive 4x2 SU-MIMO Cell Border Throughput [kbit/s] 600 500 400 300 200 Comparison of different Antenna Systems and Precoding Matrices, 500m and 1732 m ISD 500 1x1 Single Antenna TX 500 1x2 Single Antenna TX 500 2x2 CL TX Div & PSRC (36.211) 500 4x2 CL TX Div & PSRC (36.211) 500 4x2 Directional CL TX Div & PARC, 4 Beams, 4 Weights 500 4x2 Directional CL TX Div & PARC, 16 Beam, 8 Weights 1732 1x1 Single Antenna TX 1732 1x2 Single Antenna TX 1732 2x2 CL TX Div & PSRC (36.211) 1732 4x2 CL TX Div & PSRC (36.211) 1732 4x2 Directional CL TX Div & PARC, 4 Beams, 4 Weights 1732 4x2 Directional CL TX Div & PARC, 16 Beam, 8 Weights 1x1 1x2 2x2 4x2 100 0 1.0 1.1 1.2 1.3 1.4 1.5 1.6 1.7 1.8 1.9 2.0 Spectral Efficiency [bit/s/hz/sector] 23 LTE System Simulation June 2008
Conclusions and Outlook Conclusions System simulations have been performed that show the benefits of candidate algorithms for LTE Advanced The results are based on an accurate simulator that includes models for the spatial channel behavior Parts of the receiver have to be modeled in the system simulator; a huge number of channels has to be simulated -> computing time saving programming is essential Outlook Many more sophisticated algorithms wait for their simulative evaluation Channel measurements allow the evaluation of the accuracy of the ray tracing data and spatial channel models System simulations will be based on ray tracing data and channel measurements > possibility to compare field test and simulation results 24 LTE System Simulation June 2008
www.alcatel-lucent.com 25 LTE System Simulation June 2008