Impact of complex wireless environments on rate adaptation algorithms

Due to the popularity of mobile computing devices, 802.11-based wireless networks have become common place. A critical but unspecified component of 802.11 is transmission rate adaptation. Rate adaptation is difficult to do because it requires accurately estimating channel conditions while experiencing distinctly different wireless phenomena such as multi-path fast fading, nodal mobility, environmental mobility and hidden terminal interference. Rate adaptation algorithms use various methods to estimate channel conditions but these methods can be inaccurate in complex wireless environments causing incorrect adaptations. In this paper, we investigate the impact of inaccurate channel estimation methods by studying two rate adaptation algorithms, CHARM and SampleRate. We emulate high fidelity versions of both algorithms, using QualNet and evaluate their performance under environmental mobility and hidden terminal interference conditions. We found that the performance of both algorithms degrade dramatically due to inaccurate estimations caused by stale channel information. We propose two optimization techniques, REGS SINR and IntRec, to counteract these inaccuracies and implement them for CHARM and SampleRate. The proposed optimization techniques increase the performance dramatically for CHARM, up to 340% for environmental mobility and more than 290% for hidden terminal interference. We saw similar improvements of 220% and 88% for SampleRate.