Analysis of Rate-adaptation Techniques in Wireless Local Area Networks

This work involves analytical modeling of the most widely implemented rate-adaptation algorithms deployed for throughput optimization in WLAN systems. We are developing an analytical framework to gain an insight into the performance of rate-adaptation techniques with a goal of optimizing them. Specifically we are analyzing two of the most widely implemented rate-adaptation algorithms, Auto Rate Fall-back (ARF) and Adaptive Auto Rate Fall-back (AARF), by modeling them as Semi-Markov processes. Having recognized the IEEE 802.11 WLAN to be the most widely deployed wireless system we have furthered the relevance of our analysis by capturing the protocol specific MAC mechanisms which impact the performance of rate-adaptation algorithms. Our model takes into account the physical and virtual channel sensing techniques as specified in the IEEE 802.11 standard utilized for collision avoidance. We also model the binary exponential back-off procedure which is used as the primary collision resolution mechanism. Hence under the statistical assumption made by us in order to keep the developed analysis traceable we precisely model the dynamics of evolution of ARF and AARF in IEEE 802.11 WLAN system.

OPNET plays a key role in our research. Using OPNET we simulate ARF and AARF deployed in an IEEE 802.11 network and gather relevant statistics. We then compare the simulation results with the ones obtained from the analysis. This provides us the validation we need as OPNET very closely replicates the IEEE 802.11 WLAN protocol. We use the OPNET Modeler and the wireless module to carry out the simulations.

We now provide a brief overview of our analytical framework which also includes a description of ARF and AARF. For a detailed description of our work so far you can download our IEEE Secon conference paper.

Auto Rate Fall-back (ARF)

• Start data transmission at the highest bit-rate.
• If f consecutive packet transmissions fail at the current bit-rate, set the current rate to the next lower bit-rate.
• If s consecutive packet transmissions succeed at the current bit-rate, set the current rate to the next higher bit-rate.
• Other wise continue at the same bit-rate.
• In the current implementation of ARF s = 10, f = 2.

Adaptive Auto Rate Fall-back (AARF)

• Derived from ARF.
• Implements a binary-exponential back-off procedure.
• Transmits a probe packet at the next higher bit-rate after encountering a threshold number of consecutive successful packet transmissions s at the current bit-rate.
• If the probe packet transmission fails, the number of consecutive successful packet transmissions required at the current bit-rate before the next probe packet is sent is doubled (up to a maximum value)
• If the probe packet transmission is successful, the current bit-rate is set to the bit-rate at which the probe was sent.
• If f consecutive packet transmissions fail at the current bit-rate, set the current bit-rate to the next lower bit-rate.
• In the current implementation s (initial) = 10, f = 2 and _s_(max)=50.