Cross-Entropy Optimized Cognitive Radio Policies

In this paper we consider cognitive processes and their impact on the performance of cognitive radio networks (CRN). We model the cognition cycle, the main control process of the cognitive radio (CR), during which CR sequentially senses and estimates the environment state, creates plans based on the knowledge (models) of itself and that of the environment, makes decisions in order to optimize certain objectives and then acts. The proposed framework is analyzed and the performance of the CRN is evaluated. We show the impact of the sensing rate and the system dynamics on the waiting times of secondary users. Then model-based analysis is used to solve control and decision making tasks, which actually gives the radio its “cognitive” ability. Particularly, we design an efficient strategy for accessing the vacant spectrum bands and managing the transmissionsampling trade-off. In order to cope with the high complexity of this problem the policy search uses the stochastic optimization method of cross-entropy. The developed cognition cycle model represents CRN ability to intelligently react to external environment and internal state changes and gives a good understanding of the cross-entropy optimized policies. Keywords-cognitive radio networks;dynamic spectrum access; state estimation; queueing analysis; cross-entropy