Intel Validation of TGn Simulation Scenarios January 2002 doc.: IEEE 802.11-02/xxxr0 November 2003 Intel Validation of TGn Simulation Scenarios Adrian P Stephens adrian.p.stephens@intel.com Dmitry Akhmetov Sergey Shtin Intel Corporation Adrian Stephens, Intel John Doe, His Company

January 2002 doc.: IEEE 802.11-02/xxxr0 November 2003 Purpose of the work To show that the simulation scenarios defined in 11-03-802 are implementable in a realistic protocol Adrian Stephens, Intel John Doe, His Company

Simulation Methodology November 2003 Simulation Methodology Adrian Stephens, Intel

November 2003 Methodology Framework: Opnet V9 with substantially modified 802.11 components PHY Model: TGn channel models (with shadowing) Collect SNIR per sub-carrier per interference region Adaptive Bit Loading MAC Model: DCF channel access RTS/CTS training exchange Aggregate transmission using ABL-trained packets Adrian Stephens, Intel

Methodology - 2 Only interested in the QoS (UDP) flows November 2003 Methodology - 2 Only interested in the QoS (UDP) flows Don’t have EDCA or HCCA yet Assume HCCA good enough to isolate UDP based on polled TXOPs from loss of service by applied TCP/IP flows Measure throughput, delay, number of packets lost and delayed “too long” Adrian Stephens, Intel

November 2003 PHY Performance (40MHz) Peer-to-peer saturated link. Only distance is changed during the simulation. Output signal power 17 dBm. For every point 3000 data packets were used for data collection. Allocated bandwidth: 40 MHz. Modulation: BPSK. For E&V channel model the shadowing gain was turned off. Averaged BER value was collected during whole simulation (including control frames). 3 conditions were evaluated per channel model: 1 Tx antenna X 1 Rx antenna. 1 Tx antenna X 2 Rx antennas. The same shadowing gain was used for both receiving antennas. Antenna with strongest input level was selected for the future analysis. 1 Tx antenna X 2 Rx antennas. Each antenna has own shadowing gain. Antenna with strongest input level was selected for the future analysis. Adrian Stephens, Intel

November 2003 PHY Performance Adrian Stephens, Intel

MAC Results Conditions Expect to be similar to 2x2 40MHz results November 2003 MAC Results Conditions 1x1 Antenna configuration Channel width 80MHz to give 4x raw speed Because we haven’t completed our MIMO model yet Expect to be similar to 2x2 40MHz results Adrian Stephens, Intel

Scenario 1 – Aggregate throughput November 2003 Scenario 1 – Aggregate throughput Adrian Stephens, Intel

November 2003 Scenario 1 – Delay Adrian Stephens, Intel

Scenario 1 – Delay (individual) November 2003 Scenario 1 – Delay (individual) Adrian Stephens, Intel

November 2003 Scenario 4 – Enterprise Adrian Stephens, Intel

Scenario 4 – QoS Performance November 2003 Scenario 4 – QoS Performance Rx Tx Delay limit ms Peak ETE delay Total received % packet late Max allowed STA0 All STA 30 100 0.721116769 21586 2% 0.2% 5% 1% STA25 0.036222164 461 1 MSDU STA26 0.037672439 STA27 0.037807343 STA28 0.039754651 460 STA29 0.040706760 2 MSDU STA30 0.055721106 457 1.5% Adrian Stephens, Intel

Incomplete Simulation scenarios not implemented: Simulation 2 November 2003 Incomplete Simulation scenarios not implemented: 9 & 11 Don’t expect any problems Simulation 2 Needs EDCA Using DCF means high Video Rate traffic causes UDP voice to exceed its delay limits Need to look at the delayed PLR vs limits (so far only “eyeballed” them, except for scenario 4 Adrian Stephens, Intel

November 2003 Conclusion All scenarios except 9,11 and 2 have been simulated as specified including only UDP traffic All scenarios appear to meet the QoS limits Recommend we recommend acceptance of these scenarios to TGn in Albuquerque Any changed models/scenarios will need re-validation Adrian Stephens, Intel