Cognitive radio channel behavior prediction using the hidden Markov model

One of the most important issue in cognitive radio (CR) is prediction of the channel behavior. Different prediction methods have been developed for understanding channel usage pattern. Dynamic behavior of channel activity requires intelligent prediction method. Hidden Markov model (HMM), as an intelligent state predictor, has a great advantage in cognitive channel prediction based on its hidden functionality. In this paper we aim to apply HMM as Cognitive Radio channel status predictor. Specifically, we propose two channel frame structure approaches, called AP-I and AP-II, and apply them as training observations. We develop a hidden markov model for predicting the channel activity using the extracted data of two proposed frame structures. The results shows that AP-I´s predictions are more accurate when the channel SNR is high, furthermore; prediction is acceptable when channel traffic is unbalanced (equivalently High Traffic or Low Traffic). The results of applying AP-II indicate that ability to switch to more available space on channels in comparison to AP-I. Briefly speaking, applying AP-II yields better channel prediction and can increase CR performance. The output results of applying AP-II indicate 83% prediction accuracy.