Effect of the Channel and Channel Model on Dynamic Blind Spectrum Sensing Performance

Blind spectrum sensing using Bayesian sequential testing is examined to determine the effects of channel and channel model on detection performance. The theoretical framework is developed for Monte Carlo determination of type I and type II error probabilities for Bayesian sequential testing of Gaussian interference in a Rayleigh channel. It is shown that, while detection performance in a Rayleigh channel is degraded in comparison to that in a Gaussian channel, the choice of channel model has little impact on detection performance. This important result is extended to a previously-proposed method for dynamic spectrum sensing using pseudo two dimensional hidden Markov modeling, and new results given for Rayleigh channel performance.