An evolutionary game theoretic approach for cooperative spectrum sensing

Many spectrum sensing techniques have been proposed to allow a secondary user (SU) to utilize a primary user's (PU) spectrum through opportunistic access. However, few of them have considered the tradeoff between accuracy and energy consumption by taking into account the selfishness of the (SUs) in a distributed network. In this work, we consider spectrum sensing as a game where the payoff is the throughput of each SU/player. Each SU chooses between two actions, parallel individual sensing and sequential cooperative sensing techniques. Using those techniques, each SU will distributively decide the existence of the PU. Due to the repetitive nature of our game, we model it using evolutionary game (EG) theory which provides a suitable model that describes the behavioral evolution of the actions taken by the SUs. We address our problem in two cases, when the players are homogeneous and heterogeneous respectively. For the sake of stability, we find the equilibria that lead to evolutionary stable strategies (ESS) by proving that our system is evolutionary asymptotically stable, in both cases, under certain conditions on the sensing time and the false alarm probability.