Characterizing Temporal SNR Variation in 802.11 Networks

The analysis and design of wireless medium access control (MAC) protocols, coding schemes, and transmission algorithms can significantly benefit from an understanding of the channel quality variation. We attempt to represent channel quality variation using a finite-state birth-death Markov model. We outline a method to compute the parameters of the model based on measured traces obtained using common wireless chipsets. Using this Markov chain, we statistically evaluate the performance based on the channel quality, long-term correlations, and burst length distributions. Such a model significantly performs better than a traditional two-state Markov chain in characterizing 802.11 networks while maintaining the simplicity of a birth-death model. We interpret the variation of the model parameters across different locations and different times. A finite-state stationary model is amenable to analysis and can substantially benefit the design of efficient algorithms and make simulations for wireless network protocols faster.