A POMDP model for opportunistic spectrum access under ordered channel assignment policy

In this paper, we consider channel access problem in cognitive radio networks. We assume ordered channel assignment policy in the primary network. Searching channels sequentially from the beginning, the first available channel is assigned when a request arrives from a primary user, leaving idle channels in the upper part of the spectrum. As secondary users have limited sensing capabilities, ordered channel assignment makes finding available channels much easier. When a channel is sensed to be idle, the user can send over that channel and all subsequent channels. Modeling channel occupancy in primary network as a discrete time Markov chain, secondary users use their prior knowledge of state transition probabilities to derive their channel selection action. Under ordered channel assignment, several channel selection strategies are investigated. The optimal strategies based on POMDP framework are derived. For lower complexity, three greedy policies are also proposed. The proposed ordered channel assignment policy results in efficient 1-D Markov model with significant reduction in the size of the state space compared to N-D models deduced in previous works. When secondary network consists of multiple users, users can only send at the sensed channel in each time slot. This problem is also investigated in this paper. Furthermore, partially ordered channel assignment is considered for primary network and a sensing strategy is proposed for this situation.