1. TGn Simulation Methodology Validation Proposal
Month 2004
doc.: IEEE /0185r0
TGn Simulation Methodology Validation Proposal
Jeff Gilbert Atheros Communications
J.Gilbert, Atheros Communications
J.Gilbert, Atheros Communications

2. Month 2004
doc.: IEEE /0185r0
Overview
The simulations methodology ad hoc is investigating methods to accurately, yet efficiently, model the behavior of proposals’ PHYs in MAC/system simulations.
Various proposals trade off accuracy of modeling of the PHY itself with accuracy of modeling of the PHY/MAC interface.
I.e. some proposals use link-level detailed PHY but use abstractions to capture its performance (e.g /0172 by Atheros/Mitsubishi)
Some proposals use approximation of PHY but are able to incorporate it more exactly in the MAC / System Simulations (e.g /0863 by Intel)
J.Gilbert, Atheros Communications
J.Gilbert, Atheros Communications

3. Month 2004
doc.: IEEE /0185r0
Necessity to Validate
If approximations of the PHY are made, it is critical that we have a methodology to validate the approximations.
Since proposals could have arbitrary PHYs, the proposer must show how they have accurately incorporated their PHY into the methodology.
Otherwise the validity of results will be immediately challenged during the proposal comparison process.
J.Gilbert, Atheros Communications
J.Gilbert, Atheros Communications

4. Month 2004
doc.: IEEE /0185r0
A Validation Option
One method to validate the PHY / MAC abstraction is to have proposers plot unidirectional, point-to-point MAC SAP throughput vs. SNR or Range for both:
Simulation using link-level PHY (w/MAC) simulation – incorporating true rate adaptation etc.
Simulation using proposed methodology with any approximations
By only running full link-level simulation for point-to-point scenario, computation is limited.
Latency and jitter curves could be included too.
J.Gilbert, Atheros Communications
J.Gilbert, Atheros Communications

5. Benefits of Validation Curves
Month 2004
doc.: IEEE /0185r0
Benefits of Validation Curves
Validate PHY abstraction
The closeness of the two curves validates both the methodology but also the application of the methodology. This will be one primary source of debate when discussing results
Accurate comparison of PHY / MAC properties
E.g. tradeoffs between ABL and fixed-rate systems, as well as other PHY and MAC properties can be accurately assessed in terms of PHY improvements, MAC overheads, and Doppler mitigation
Point-to-point would simplify vs. full usage model
J.Gilbert, Atheros Communications
J.Gilbert, Atheros Communications

6. Month 2004
doc.: IEEE /0185r0
Relationship to CC67
Comparison Criteria 67 currently looks at PHY-only performance in “real” channels.
However being PHY-only makes it more difficult to accurately include rate adaptation effects.
These validation curves could replace CC67 and thus keep the total required work from proposers similar.
J.Gilbert, Atheros Communications
J.Gilbert, Atheros Communications

7. Month 2004
doc.: IEEE /0185r0
Proposed text for CC67
Show curves for achieved point-to-point MAC SAP throughput as a function of total SNR. Plot the achieved throughput, versus SNR per receive antenna, for a 1500B packet source. SNR in this case is the ratio of the received power at a single receive antenna, to the input-referred receiver noise power. The received signal power is as measured in -10dB signal bandwidth at a single receive antenna, is summed over all transmit antennas and averaged over time and receive antennas. Note that when computing the SNR, the signal and noise bandwidth shall be identical. The throughput shall be computed using a minimum of 100 packet errors. This should be simulated for channel models B, D, and F. The simulations should all include the Doppler effect as specified in the text of the channel model document. All models should be run without the fluorescent effect but additionally model D should be run with and without the fluorescent effect. The shadowing variance should be 0. Simulations should use a link-level phy simulation with rate adaptation effects incorporating the channel dynamics present in the TGn channel model.
J.Gilbert, Atheros Communications
J.Gilbert, Atheros Communications