Cross-layer Dynamic Rate Adaptations for Green Cognitive Radio Networks

We investigate cross-layer rate adaptation techniques for a secondary user (SU) equipped with a finite buffer in green cognitive radio networks. We assume that the activity statistics of the licensed primary users (PUs) channels are independent and identically distributed, and a SU detects their states using a spectrum sensing method. We formulate the problem as an infinite-horizon partially observable Markov decision process (POMDP). The policy is obtained using maximum-likelihood heuristic policy (MLHP) technique. We assume that the transition probabilities of the PUs and the fading channel between the SU's transmitter and receiver are known, but the exact PU's states are hidden. By tracking belief of the hidden states, the SU takes decision on the rate and corresponding power to minimize energy consumption with constraints on delay and bit error rate. Numerical results are given to show the performance of the proposed MLHP, which is found to perform very close to fully observable optimal policy. We also show pointer to choose design parameter so that the scheduler becomes the most energy-efficient for given quality of service requirements of the application.