A Q-Learning based sensing task selection scheme for cognitive radio networks

More than an adaptive system, the cognitive radio system should be a kind of intelligent system. In the paper, a Q-learning algorithm of the intelligent control theory is adopted to solve the sensing task selection problem among cognitive radio users in the distributed cognitive radio networks. In the proposed scheme, each cognitive radio user selects its sensing task through times of interaction with the environment and self-learning by means of its embedded Q-learning module. The scheme works without any CSI or estimation of primary traffic. According to the simulation results, the proposed scheme can improve the sensing efficiency and attain the convergence in a short time, so it may be regarded as a good attempt for the future intelligent cognitive radio systems.