Power allocation in cognitive radio networks by the reinforcement learning scheme with the help of Shapley value of games

In this paper, we present an algorithm based on the reinforcement learning scheme with the help of Shapley value of game for power allocation in the cognitive radio networks. The goal is to optimize the achievable transmission rates for secondary users and simultaneously to maximize their usefulness in the coalition. A performance measure is formed as a weighted linear function of the probability of the idle channel amongst N cooperating secondary users. Then, the problem is formulated as a semi-Markov decision process with an average cost criterion and reinforcement learning algorithm is developed to an approximate optimal control policy. The proposed scheme is driven by an estimated dynamic model of cognitive radio network learning simultaneously with the use of the Shapley value of games, to form the best coalition. The simulations are provided to compare the effectiveness of the proposed method against other methods under a variety of traffic conditions with some well-known policies