Channel exploration for aggregation in cognitive radio system

In this paper we focus on channel exploration problem in the cognitive radio (CR) system, where the exploration process consumes time. The explored channels can be aggregated by CR user for its transmission. CR user adopts a synchronous slotted structure with primary users. Usually, the channel exploration problem is formulated as an optimal stopping problem. However, most previous related works are based on the assumption that channel state changes randomly across slots. In the system, channel state contains channel availability and link quality on the channel. When channel state’s transition across slots follows a Markov process, the problem becomes different. Then we introduce a two-dimension Partially Observable Markov Decision Process framework into the optimal stopping problem. We concentrate on the myopic rule of the new problem. By exploring the structure property of the myopic rule, we can achieve the optimal performance under the myopic rule with a lower computation complexity. To further reduce the computation complexity for a practical application, we then propose a greedy approach. The simulation results show CR user can obtain a near-optimal performance by the greedy approach. The validity of our proposed approaches is also verified by simulation.