This repository contains an implementation of the IEEE 802.11ad A-BFT beamforming training medium access protocol [1]. We developed a discrete-event time-driven simulator in Python that implements this medium access protocol in order to study the performance of beamforming training in the A-BFT - especially the average time required for a station to complete its beamforming training. Our simulator closely follows the IEEE 802.11ad standard's specifications for each independently transmitting station.
We model a Responder Sector Sweep (RSS) as a single frame transmission occupying one time slot. According to the IEEE standard, a station succeeds its RSS attempt only if it receives a feedback frame from the AP before the end of the ongoing time slot. If no collisions occurs, the RSS is successful. We abstracted the feedback mechanism and instead kept count of the number of colliding stations in each time slot to determine a RSS failure. We consider the transmission channel ideal so that RSS failures are due to collisions only.
All protocol parameters can be adjusted: the number of slots per period, the maximum number of successive attempts, the idle backoff window and simulation time in periods.
We also log a set of statistics including but not limited to:
- Average number of periods before a station succeeds its RSS
- Average number of periods until a station becomes idle
- Success rate in a period
- Probability of success of a RSS
- Probability a station is idle
- Probability distribution of success and idle.
The above results can be logged to a file using the --tofile option.
A working installation of Python 2 with the following modules are required:
- numpy
- collections
python abft-simulator.py --help
python abft-simulator.py 20 8 4 2 --time 5000 is a 5000 periods simulation run with 20 stations, 8 time slots per period, 4 maximum allowed consecutive attempts and an idle backoff window of 2.
[1] "IEEE Standards 802.11ad-2012: Enhancements for Very High Throughput in the 60 GHz Band", 2012.