Optimizing Voting Rule for Cooperative Spectrum Sensing Through Learning Automata

In cooperative spectrum sensing (CSS), an optimal decision fusion rule can effectively improve the performance of the secondary system. Considering that a priori knowledge about the characteristics of primary users and secondary users (SUs) is usually unknown in practice, we propose a learning automata-based algorithm to derive the optimal voting rule for decision fusion in CSS. The proposed algorithm can work with any detection method used at SUs. Its convergence and robustness are shown here. Its stability is also analyzed by using noncooperative game theory. The condition that the obtained optimal voting rule remains unchanged is derived based on the concept of subgame perfect Nash equilibrium. It is shown that the proposed algorithm generally does not need to run many times and can work at a low cost in practice. Computer simulations validate the effectiveness of the proposed scheme.