Simulation Effort Required to Satisfy the n Comparison Criteria

Slides:



Advertisements
Similar presentations
Doc.: IEEE /209r2 Submission March 2004 Lanzl, Aware; Ketchum, QualcommSlide 1 Carrier Frequency / Symbol Clock Offset Proposal for TGn FRCC Colin.
Advertisements

Doc.: n-proposal-statistical-channel-error-model.ppt Submission Jan 2004 UCLA - STMicroelectronics, Inc.Slide 1 Proposal for Statistical.
Doc.: ax Submission July 2014 Slide 1 Proposed Calibration For MAC simulator Date: Authors:
Doc.: IEEE /1229r1 Submission November 2009 Alexander Maltsev, IntelSlide 1 Application of 60 GHz Channel Models for Comparison of TGad Proposals.
Doc.: IEEE / Submission March 2013 Juho Pirskanen, Renesas Mobile CorporationSlide 1 Discussion On Basic Technical Aspects for HEW Date:
Doc.: IEEE 11-04/0304r0 Submission March 2004 John S. Sadowsky, Intel PER Prediction for n MAC Simulation John S. Sadowsky (
Doc.: IEEE /1226r0 Submission Sep 2014 Slide 1 SLS Box 1&2 Calibration Results Date: Authors: Russell Huang (MediaTek)
Doc.: IEEE /0174r1 Submission February 2004 John Ketchum, et al, QualcommSlide 1 PHY Abstraction for System Simulation John Ketchum, Bjorn Bjerke,
Proposal for Statistical Channel Error Model
Spatial Discovery in 60 GHz
GI Overhead/Performance Impact on Open-Loop SU-MIMO
Closed versus Open Loop Comparisons
Comparisons of Simultaneous Downlink Transmissions
Comments on TGn FRCC Doc # r17
<month year> doc.: IEEE <030158r0> March 2004
Further Discussion on Beam Tracking for ay
PHY abstraction and performance for outdoor channel models
Closed versus Open Loop
John Ketchum, Bjorn A. Bjerke, and Irina Medvedev Qualcomm, Inc.
MU-MIMO STA scheduling strategy and Related PHY signaling
SLS Box5 Calibration Results and Discussions
OFDMA Performance Analysis
TGn Sync MAC Response to Reasons and Cures Relating to Link Adaptation
TGn Simulation Methodology Ad Hoc Overview
MU-MIMO channel access flow for 11ay
Multi-User MIMO Channel Measurements
Record and Playback PHY Abstraction for n MAC Simulations
OFDMA Performance Analysis
RBIR-based PHY Abstraction with Channel Estimation Error
SMSC March 2004 Session Speaker List
TGn Simulation Methodology Validation Proposal
Further Discussion on Beam Tracking for ay
Further Study of Time Varying Interference and PHY Abstraction
Comments on PHY Abstraction
Comments on PHY Abstraction
High-Throughput Enhancements for : Features and Performance
Low Latency and Low Medium Utilization
802.11ac Channel Modeling Authors: Jan 19, 2009 Month Year
Joint Processing MU-MIMO
Further Discussions on PHY Abstraction
Initial Distributed MU-MIMO Simulations
Update on “Channel Models for 60 GHz WLAN Systems” Document
TGn Simulation Methodology Validation Proposal
Detailed Responses to “Reasons and Cures” Comments on MCS Set
Carrier Frequency / Symbol Clock Offset Proposal for TGn FRCC
Marvell Semiconductor
Effect of CCA in residential scenario part 2
PHY Abstraction based on PER Prediction
Joint Processing MU-MIMO – Update
QUALCOMM Proposal Summary
September 2006 doc.: IEEE /1458r0 May 2007
Feedback-jamming ARQ mechanisms
Erik Lindskog Hemanth Sampath Ravi Narasimhan
OFDMA Performance Analysis
Joint Processing MU-MIMO – Update
TGn Simulation Methodology Ad Hoc Overview
PER Prediction for n MAC Simulation
Performance Comparison of Dynamic OFDM with n
Numerology for 11ax Date: Authors: March 2015 Month Year
PHY Abstractions Types For HEW System Level Simulations
HARQ Feasibility for EHT
PHY Performance Evaluation with 60 GHz WLAN Channel Models
Spatial Discovery in 60 GHz
A New Metric for Evaluating the Throughput of HEW
Comparisons of HARQ transmission schemes for 11be
False L-STF Detection Issue
DSC Calibration Result
TGn Simulation Methodology Special Committee March 2004 Report
Consideration on System Level Simulation
Consideration on Joint Transmission
Presentation transcript:

Simulation Effort Required to Satisfy the 802.11n Comparison Criteria March 2004 doc.: IEEE 802.11-04/307r0 March 2004 Simulation Effort Required to Satisfy the 802.11n Comparison Criteria Bjorn A. Bjerke, John Ketchum, and Irina Medvedev Qualcomm, Inc. bbjerke@qualcomm.com Bjorn Bjerke et al., Qualcomm, Inc Bjorn Bjerke et al. Qualcomm, Inc.

March 2004 Simulation Effort Significant simulation effort required to satisfy the current version of the 802.11n Comparison Criteria Two types of simulations: PHY layer sims: typically implemented in C System sims: typically implemented in NS2 or similar Bjorn Bjerke et al., Qualcomm, Inc

CCs Involving Simulations March 2004 doc.: IEEE 802.11-04/307r0 March 2004 CCs Involving Simulations PHY sims: CC59 CC67 CC67.1 CC67.2 System sims: CC15 CC18 CC19 CC20 (CC24) CC27 CC28 CC58 Bjorn Bjerke et al., Qualcomm, Inc Bjorn Bjerke et al. Qualcomm, Inc.

March 2004 PHY Sims: Assumptions NxN MIMO system  N*N possible system configurations PER vs SNR evaluated for 10 SNR values 100 packet errors collected per point CPU time per point: 1 hour  CPU time per curve: 10 hours Bjorn Bjerke et al., Qualcomm, Inc

PHY Sims: CPU Time Estimate March 2004 PHY Sims: CPU Time Estimate CC59 (PER vs SNR with fixed channel): 5 representative data rates Example: N=4  16 system configurations CPU time = 16*5*10 hours = 800 hours CC67 (PER vs SNR with fading channels): 16 system configurations 3 channel models (B, D, F) Channel D w/fluorescent lights and highest rate CPU time = (16*5*3+16)*10 hours = 2560 hours Bjorn Bjerke et al., Qualcomm, Inc

PHY Sims: CPU Time Estimate March 2004 PHY Sims: CPU Time Estimate CC67.1 (Rate vs SNR with fading channels): 16 system configurations 3 channel models (B, D, F) Channel D w/fluorescent lights CPU time = 16*4*10 hours = 640 hours CC67.2 (PER for 2 SNR values with frequency offset): Example: 5 representative freq. offset values 2 SNR values (1 hour each) CPU time = 16*5*2 hours = 160 hours Bjorn Bjerke et al., Qualcomm, Inc

PHY Sims: CPU Time Estimate March 2004 PHY Sims: CPU Time Estimate Total CPU time needed for PHY sims: 800+2560+640+160 hours = 4160 hours With 8 CPUs running continuously, this will take 4160/(24*8) days = 22 days Assume 100% overhead  44 days Bjorn Bjerke et al., Qualcomm, Inc

System Sims: Assumptions March 2004 System Sims: Assumptions Simulation Scenario 1 (baseline): 40 secs. of simulated time to reach steady state; < 1 hour of CPU time to simulate Assume CPU time goes as the square of the number of STAs Bjorn Bjerke et al., Qualcomm, Inc

System Sims: CPU Time Estimate March 2004 System Sims: CPU Time Estimate CC15: SS17: 1 AP, 1 STA, 1 channel model; < 1 hour SS18: 1 AP, 1 STA, 1 channel model; < 1 hour SS19: 1 AP, 2 STA, 1 channel model; < 1 hour CCs 18, 19, 20, 24 SS1:  1 AP, 11 STA --  1 hour SS4:  1 AP, 30 STA -- 7 hours SS6:  1 AP, 52 STA  -- 20 hours SS9:  1 AP, 6 STA --  1 hour SS11: 2 AP, 12 STA -- 2 hours Bjorn Bjerke et al., Qualcomm, Inc

System Sims: CPU Time Estimate March 2004 doc.: IEEE 802.11-04/307r0 March 2004 System Sims: CPU Time Estimate CCs 27, 28, 58 Need to run SS16 for each of the two channel models, B and D. SS16: 1 AP, 1 STA, repeated for various locations of the STA    Total: 4 hours Total CPU time for system sims: < 40 hours  Add 100% overhead: < 80 hours Bjorn Bjerke et al., Qualcomm, Inc Bjorn Bjerke et al. Qualcomm, Inc.

The Simulation Effort May Be Reduced Without Compromising Quality March 2004 The Simulation Effort May Be Reduced Without Compromising Quality PHY sims represent the majority share of the effort  focus on PHY sims: Limit the number of required system configurations to a set of 5 representative, e.g., 1xN, Nx1, N1xN1, N1xN2, and N2xN2 Let proposer choose between CC67 and CC67.1, i.e., make CC67 mandatory for fixed rate systems, and make CC67.1 mandatory for rate-adaptive systems Total CPU time required drops to 1100 hours and 500 hours, respectively (or 1300 hours if all versions of CC67 are mandatory) Bjorn Bjerke et al., Qualcomm, Inc