Channel state prediction for cognitive radios with stochastically varying primary user traffic density

In this study, a new algorithm for cognitive radios is proposed to predict the state of the future observation periods in channels where the traffic density of the primary user changes stochastically with time. Markov modulated Poisson process has been used to model the primary user (PU) traffic. According to the proposed method, transition probabilities are obtained using previously taken decisions and the state of the channel is decided as busy or idle for the next observation period based on these probabilities. Performance of the proposed method is compared against correlation based prediction methods. Two metrics called system utility and PU disturbance ratio, respectively, have been used for performance evaluation. According to simulations carried out for varying lengths of the history window, performance of the proposed algorithm is observed to be higher as compared to other techniques.