19 th ComNets-Workshop Analytical Validation of the IMT- Advanced Compliant openwns LTE Simulator Dipl.-Ing. Maciej Mühleisen ComNets Research Group RWTH Aachen University, Germany ComNets-Workshop, 11.3.211 Maciej Mühleisen, ComNets, RWTH Aachen University
Outline Motivation IMT-A Channel Model & Scenario CIR Calculation for Indoor Hotspot Scenario LTE Layer 2 Calibration Assumptions Results Summary, Conclusion & Outlook 2
Motivation World Radio Conference 27 (WRC-7) has identified new spectrum for mobile radio communication ITU-R controls the allocation of this spectrum Issued the IMT- Advanced process to evaluate candidate systems 3
C.D.F. [%] Motivation Self evaluation: Done by bodies behind candidate Radio Access Technologies (RATs) (3GPP for LTE-A, IEEE for WiMAX-A) External evaluation: Done by others ex. WINNER+ Problem: How to verify used simulators => calibration 1 Large-scale channel model calibration Calibration results for the Indoor Hotspot scenario LTE R8 basic configuration calibration 9 Org 1 8 7 6 5 Org 2 Org 3 Org 4 Org 5 Org 6 4 3 Org 7 Org 8 Average 2 1-1 1 2 3 4 5 6 Wideband SINR [db] From: FINAL EVALUATION REPORT FROM WINNER+ ON THE IMT-ADVANCED PROPOSAL IN DOCUMENTS IMT-ADV/6, IMT-ADV/8 AND IMT-ADV/9 4
IMT-A A Channel Model & Scenario Done once per simulation run: 2 user terminals; uniformly random placed Each terminal chooses channel condition on both BS links Random, distance dependent Line of sight (LoS) or non line of sight (NLoS) Calculate pathloss PL Draw log-normal shadowing value X Associate to BS with lowest attenuation (X + PL) on link Base Station (BS) User Terminal (UT) 5
IMT-A A Channel Model & Scenario Done once per simulation run: 2 user terminals; uniformly random placed Each terminal chooses channel condition on both BS links Random, distance dependent Line of sight (LoS) or non line of sight (NLoS) Calculate pathloss PL Draw log-normal shadowing value X Associate to BS with lowest attenuation (X + PL) on link 1 d=1m.8 Base Station (BS) User Terminal (UT) P(c = LoS).6.4.2 18m 37m 1 2 3 4 5 6 7 8 9 Distance d [m] 6
IMT-A A Channel Model & Scenario Done once per simulation run: 2 user terminals; uniformly random placed Each terminal chooses channel condition on both BS links Random, distance dependent Line of sight (LoS) or non line of sight (NLoS) Calculate pathloss PL Draw log-normal shadowing value X Associate to BS with lowest attenuation (X + PL) on link 1.8 d=1m,los Base Station (BS) User Terminal (UT) P(c = LoS).6.4.2 18m 37m 1 2 3 4 5 6 7 8 9 Distance d [m] 7
IMT-A A Channel Model & Scenario Done once per simulation run: 2 user terminals; uniformly random placed Each terminal chooses channel condition on both BS links Random, distance dependent Line of sight (LoS) or non line of sight (NLoS) Calculate pathloss PL Draw log-normal shadowing value X Associate to BS with lowest attenuation (X + PL) on link 1.8 d=1m,los,pl=65db Base Station (BS) User Terminal (UT) P(c = LoS).6.4.2 18m 37m 1 2 3 4 5 6 7 8 9 Distance d [m] 8
IMT-A A Channel Model & Scenario Done once per simulation run: 2 user terminals; uniformly random placed Each terminal chooses channel condition on both BS links Random, distance dependent Line of sight (LoS) or non line of sight (NLoS) Calculate pathloss PL Draw log-normal shadowing value X Associate to BS with lowest attenuation (X + PL) on link 1 d=1m,los,pl=65db d=75m.8 Base Station (BS) User Terminal (UT) P(c = LoS).6.4.2 18m 37m 1 2 3 4 5 6 7 8 9 Distance d [m] 9
IMT-A A Channel Model & Scenario Done once per simulation run: 2 user terminals; uniformly random placed Each terminal chooses channel condition on both BS links Random, distance dependent Line of sight (LoS) or non line of sight (NLoS) Calculate pathloss PL Draw log-normal shadowing value X Associate to BS with lowest attenuation (X + PL) on link 1 d=1m,los,pl=65db d=75m,nlos.8 Base Station (BS) User Terminal (UT) P(c = LoS).6.4.2 18m 37m 1 2 3 4 5 6 7 8 9 Distance d [m] 1
IMT-A A Channel Model & Scenario Done once per simulation run: 2 user terminals; uniformly random placed Each terminal chooses channel condition on both BS links Random, distance dependent Line of sight (LoS) or non line of sight (NLoS) Calculate pathloss PL Draw log-normal shadowing value X Associate to BS with lowest attenuation (X + PL) on link 1 d=1m,los,pl=65db d=75m,nlos,pl=13db.8 Base Station (BS) User Terminal (UT) P(c = LoS).6.4.2 18m 37m 1 2 3 4 5 6 7 8 9 Distance d [m] 11
IMT-A A Channel Model & Scenario Done once per simulation run: 2 user terminals; uniformly random placed Each terminal chooses channel condition on both BS links Random, distance dependent Line of sight (LoS) or non line of sight (NLoS) Calculate pathloss PL Draw log-normal shadowing value X Associate to BS with lowest attenuation (X + PL) on link.14 X 1 X 2 d=1m,los,pl=65db+x 1.12 d=75m,nlos,pl=13db+x 2.1 p(x).8.6.4 Base Station (BS) User Terminal (UT).2 5 6 7 8 9 1 11 12 PL [db] 12
IMT-A A Channel Model & Scenario Done once per simulation run: 2 user terminals; uniformly random placed Each terminal chooses channel condition on both BS links Random, distance dependent Line of sight (LoS) or non line of sight (NLoS) Calculate pathloss PL Draw log-normal shadowing value X Associate to BS with lowest attenuation (X + PL) on link Base Station (BS) User Terminal (UT) 13
IMT-A A Channel Model & Scenario Done once per simulation run: 2 user terminals; uniformly random placed Each terminal chooses channel condition on both BS links Random, distance dependent Line of sight (LoS) or non line of sight (NLoS) Calculate pathloss PL Draw log-normal shadowing value X Associate to BS with lowest attenuation (X + PL) on link Base Station (BS) User Terminal (UT) 14
IMT-A A Channel Model & Scenario Done once per simulation run: 2 user terminals; uniformly random placed Each terminal chooses channel condition on both BS links Random, distance dependent Line of sight (LoS) or non line of sight (NLoS) Calculate pathloss PL Draw log-normal shadowing value X Associate to BS with lowest attenuation (X + PL) on link 9 UTs 11 UTs Base Station (BS) User Terminal (UT) 15
CIR Calculation for Indoor Hotspot Scenario p(cir s=1,c 1,c 2 ) ~ N(μ 2,c2 -μ 1,c1, δ 1,c12 +δ 2,c22 ) {x>} / P(s=1) [25m, 15m].15.125.1 to left BS, LoS to left BS, NLoS to right BS, LoS to right BS, NLoS C I.8.6 s=1,los,los p(x).75 p(x).4.5.25.2 2 4 6 8 1 12 Path loss PL [db] 5 1 15 2 25 3 35 4 CIR [db] 16
CIR Calculation for Indoor Hotspot Scenario p(cir s=1,c 1,c 2 ) ~ N(μ 2,c2 -μ 1,c1, δ 1,c12 +δ 2,c22 ) {x>} / P(s=1) [25m, 15m].15.125.1 to left BS, LoS to left BS, NLoS to right BS, LoS to right BS, NLoS C I.8.6 s=1,los,los s=1,nlos,los p(x).75 p(x).4.5.25.2 2 4 6 8 1 12 Path loss PL [db] 5 1 15 2 25 3 35 4 CIR [db] 17
CIR Calculation for Indoor Hotspot Scenario p(cir s=1,c 1,c 2 ) ~ N(μ 2,c2 -μ 1,c1, δ 1,c12 +δ 2,c22 ) {x>} / P(s=1) [25m, 15m].15.125.1 to left BS, LoS to left BS, NLoS to right BS, LoS to right BS, NLoS C I.8.6 s=1,los,los s=1,nlos,los s=1,los,nlos p(x).75 p(x).4.5.25.2 2 4 6 8 1 12 Path loss PL [db] 5 1 15 2 25 3 35 4 CIR [db] 18
CIR Calculation for Indoor Hotspot Scenario p(cir s=1,c 1,c 2 ) ~ N(μ 2,c2 -μ 1,c1, δ 1,c12 +δ 2,c22 ) {x>} / P(s=1) [25m, 15m].15.125.1 to left BS, LoS to left BS, NLoS to right BS, LoS to right BS, NLoS C I.8.6 s=1,los,los s=1,nlos,los s=1,los,nlos s=1,nlos,nlos p(x).75 p(x).4.5.25.2 2 4 6 8 1 12 Path loss PL [db] 5 1 15 2 25 3 35 4 CIR [db] 19
CIR Calculation for Indoor Hotspot Scenario p(cir s=2,c 1,c 2 ) ~ N(μ 1,c1 -μ 2,c2, δ 1,c12 +δ 2,c22 ) {x>} / P(s=2) [25m, 15m].15.125.1 to left BS, LoS to left BS, NLoS to right BS, LoS to right BS, NLoS I C.8.6 s=1,los,los s=1,nlos,los s=1,los,nlos s=1,nlos,nlos s=2,los,los p(x).75 p(x).4.5.25.2 2 4 6 8 1 12 Path loss PL [db] 5 1 15 2 25 3 35 4 CIR [db] 2
CIR Calculation for Indoor Hotspot Scenario p(cir s=2,c 1,c 2 ) ~ N(μ 1,c1 -μ 2,c2, δ 1,c12 +δ 2,c22 ) {x>} / P(s=2) [25m, 15m].15.125.1 to left BS, LoS to left BS, NLoS to right BS, LoS to right BS, NLoS I C.8.6 s=1,los,los s=1,nlos,los s=1,los,nlos s=1,nlos,nlos s=2,los,los s=2,nlos,los p(x).75 p(x).4.5.25.2 2 4 6 8 1 12 Path loss PL [db] 5 1 15 2 25 3 35 4 CIR [db] 21
CIR Calculation for Indoor Hotspot Scenario p(cir s=2,c 1,c 2 ) ~ N(μ 1,c1 -μ 2,c2, δ 1,c12 +δ 2,c22 ) {x>} / P(s=2) [25m, 15m] p(x).15.125.1.75 to left BS, LoS to left BS, NLoS to right BS, LoS to right BS, NLoS I C p(x).8.6.4 s=1,los,los s=1,nlos,los s=1,los,nlos s=1,nlos,nlos s=2,los,los s=2,nlos,los s=2,los,nlos.5.25.2 2 4 6 8 1 12 Path loss PL [db] 5 1 15 2 25 3 35 4 CIR [db] 22
CIR Calculation for Indoor Hotspot Scenario p(cir s=2,c 1,c 2 ) ~ N(μ 1,c1 -μ 2,c2, δ 1,c12 +δ 2,c22 ) {x>} / P(s=2) [25m, 15m] p(x).15.125.1.75 to left BS, LoS to left BS, NLoS to right BS, LoS to right BS, NLoS I C p(x).8.6.4 s=1,los,los s=1,nlos,los s=1,los,nlos s=1,nlos,nlos s=2,los,los s=2,nlos,los s=2,los,nlos s=2,nlos,nlos.5.25.2 2 4 6 8 1 12 Path loss PL [db] 5 1 15 2 25 3 35 4 CIR [db] 23
CIR Calculation for Indoor Hotspot Scenario p(cir x,y) = Σ P(s)P(c 1 )P(c 2 )p(cir s,c 1,c 2 ) "s,c 1,c 2 P(s)P(c 1 )P(c 2 )p(x).35.3.25.2.15.1.5 s=1,los,los s=1,nlos,los s=1,los,nlos s=1,nlos,nlos p(x).8.7.6.5.4.3.2.1 CIR at x = 25m, y = 15m 5 1 15 2 25 3 35 4 CIR [db] 1 2 3 4 5 CIR [db] 24
CIR Calculation for Indoor Hotspot Scenario Rate probability P(r): Probability for each MCS P(r x,y) = Σ P(s)P(c 1 )P(c 2 )P(r s,c 1,c 2 ) "s,c 1,c 2 P(s)P(c 1 )P(c 2 )p(x).35.3.25.2.15.1.5 s=1,los,los s=1,nlos,los s=1,los,nlos s=1,nlos,nlos 5 1 15 2 25 3 35 4 CIR [db] MCS 1 MCS 2 MCS 3 MCS 4 MCS 5 MCS 6 [1] J. Olmos et al., Link Abstraction Models Based on Mutual Information for LTE Downlink, COST 21, Tech. Rep. TD(1)1152, June 21. 25
C.D.F. [%] CIR Calculation for Indoor Hotspot Scenario Integrate over the area and normalize to obtain scenario CIR & rate distribution No solution for integral is known => sum up, use symmetry 1 [m,m] [6m,m].8 Base Station (BS) P(CIR > x).6.4 1 db Handover Margin 1 9 8 7 6 Org 1 Org 2 Org 3 Org 4 Org 5.2 Analytic Result openwns Result for WINNER+ Calibration 1 2 3 4 5 6 CIR [db] 5 4 Org 6 Org 7 Org 8 3 Average 2 1-1 1 2 3 4 5 6 Wideband SINR [db] 26
CIR Calculation for Indoor Hotspot Scenario Integrate over the area and normalize to obtain scenario CIR & rate distribution No solution for integral is known => sum up, use symmetry P(x).5.45.4.35.3.25.2.15.1.5 Analytic Simulated (5 drops) Simulated (5 drops) 1 2 3 4 5 6 7 8 9 1 11 12 13 MCS Index [m,m] Base Station (BS) [6m,m] 27
LTE Layer 2 Calibration Assumptions Data Link Layer (DLL) constrains from ITU-R M.2135: Packets are scheduled with an appropriate packet scheduler(s) [ ]. Channel quality feedback delay, feedback errors, PDU (protocol data unit) errors and real channel estimation effects inclusive of channel estimation error are modelled and packets are retransmitted as necessary. The overhead channels (i.e., the overhead due to feedback and control channels) should be realistically modelled. 2 MHz FDD 1 Resource Blocks Dowlink: 1 frames = 1 superframe Broadcast Control Channel (BCH) every 1 frames => Parts of center 6 resource blocks Specified in [3GPP TR 36.814] 28
LTE Layer 2 Calibration Assumptions Data Link Layer (DLL) constrains from ITU-R M.2135: Packets are scheduled with an appropriate packet scheduler(s) [ ]. Channel quality feedback delay, feedback errors, PDU (protocol data unit) errors and real channel estimation effects inclusive of channel estimation error are modelled and packets are retransmitted as necessary. The overhead channels (i.e., the overhead due to feedback and control channels) should be realistically modelled. 2 MHz FDD 1 Resource Blocks Dowlink: 1 frames = 1 superframe Specified in [3GPP TR 36.814] Broadcast Control Channel (BCH) every 1 frames => Parts of center 6 resource blocks Downlink Control Channel (DLCCH) every frame =>3 of 14 symbols 29
LTE Layer 2 Calibration Assumptions Data Link Layer (DLL) constrains from ITU-R M.2135: Packets are scheduled with an appropriate packet scheduler(s) [ ]. Channel quality feedback delay, feedback errors, PDU (protocol data unit) errors and real channel estimation effects inclusive of channel estimation error are modelled and packets are retransmitted as necessary. The overhead channels (i.e., the overhead due to feedback and control channels) should be realistically modelled. 2 MHz FDD 1 Resource Blocks Dowlink: 1 frames = 1 superframe Specified in [3GPP TR 36.814] Broadcast Control Channel (BCH) every 1 frames => Parts of center 6 resource blocks Downlink Control Channel (DLCCH) every frame =>3 of 14 symbols Pilot tones: Reduce bit per symbol 3
LTE Layer 2 Calibration Assumptions Data Link Layer (DLL) constrains from ITU-R M.2135: Packets are scheduled with an appropriate packet scheduler(s) [ ]. Channel quality feedback delay, feedback errors, PDU (protocol data unit) errors and real channel estimation effects inclusive of channel estimation error are modelled and packets are retransmitted as necessary. The overhead channels (i.e., the overhead due to feedback and control channels) should be realistically modelled. 2 MHz FDD 1 Resource Blocks Dowlink: 1 frames = 1 superframe Specified in [3GPP TR 36.814] 31
LTE Layer 2 Calibration Assumptions Data Link Layer (DLL) constrains from ITU-R M.2135: Packets are scheduled with an appropriate packet scheduler(s) [ ]. Channel quality feedback delay, feedback errors, PDU (protocol data unit) errors and real channel estimation effects inclusive of channel estimation error are modelled and packets are retransmitted as necessary. The overhead channels (i.e., the overhead due to feedback and control channels) should be realistically modelled. 2 MHz FDD 1 Resource Blocks 9 associated UTs Dowlink: RoundRobin, full bandwidth allocation 1 frames = 1 superframe Specified in [3GPP TR 36.814] 32
LTE Layer 2 Calibration Assumptions Data Link Layer (DLL) constrains from ITU-R M.2135: Packets are scheduled with an appropriate packet scheduler(s) [ ]. Channel quality feedback delay, feedback errors, PDU (protocol data unit) errors and real channel estimation effects inclusive of channel estimation error are modelled and packets are retransmitted as necessary. The overhead channels (i.e., the overhead due to feedback and control channels) should be realistically modelled. 2 MHz FDD 1 Resource Blocks 9 associated UTs Dowlink: RoundRobin, full bandwidth allocation 1 frames = 1 superframe The number of associated UTs a significantly influences the L2 throughput T = r/a Specified in [3GPP TR 36.814] 33
Results P(x < X) 1.9.8.7.6.5.4.3.2.1 Analytic T = r/a a ~ Bernoulli(.5, 2) P(T) = P(a)P(r) 2 4 6 8 1 Throughput [bps] x 1 6 34
Results 1.9.8.7 Analytic Simulated (5 drops) T = r/a a ~ Bernoulli(.5, 2) P(T) = P(a)P(r) P(x < X).6.5.4.3.2.1 2 4 6 8 1 Throughput [bps] x 1 6 35
Results 1.9.8.7 Analytic Simulated (5 drops) Simulated (5 drops) T = r/a a ~ Bernoulli(.5, 2) P(T) = P(a)P(r) P(x < X).6.5.4.3.2.1 2 4 6 8 1 Throughput [bps] x 1 6 36
Results 1.9.8.7 T = r/a a ~ Bernoulli(.5, 2) P(T) = P(a)P(r) Analytic Simulated (1 drops) P(x < X).6.5.4.3.2.1 2 4 6 8 1 Throughput [bps] x 1 6 37
Results P(x < X) 1.9.8.7.6.5.4.3.2.1 T = r/a a ~ Bernoulli(.5, 2) P(T) = P(a)P(r) Analytic Simulated (1 drops) 5-Percentile 2 4 6 8 1 Throughput [bps] x 1 6 Cell Spectral Efficiency [bit/s/hz/ Cell] Cell Edge User Spectral Efficiency [bit/s/hz] Calc. 2.269 < 3.57 <.1 38 Sim. 2.265 < 3.57 <.1
Summary, Conclusion & Outlook Summary & Conclusion: Layer 2 simulator calibration results have been validated openwns creates correct results under the given assumptions See [1] for more Outlook: More than two cells Sectorized antennas Small scale fading MIMO Uplink => done Other partners should validate their simulators [1] M. Mühleisen et al., Analytical Evaluation of an IMT-Advanced Compliant LTE System Level Simulator, European Wireless, April 211. 39
Thank you for your attention! Maciej Mühleisen mue@comnets.rwth-aachen.de Maciej Mühleisen, ComNets, RWTH Aachen University 4