No-Regret Learning in Collaborative Spectrum Sensing with Malicious Nodes

In cognitive radio networks, spectrum sensing is a key component to detect spectrum holes (i.e., channels not used by any primary user). Collaborative spectrum sensing among the cognitive radio nodes is expected to improve the fidelity of primary user detection. However, malicious nodes can significantly impair the collaborative spectrum sensing by sending wrong reports to the fusion center. To overcome this problem, we propose in this paper no-regret learning to study the non-constructive secondary users caused by either evil-intention or altruistical incapability. We investigate learning scenarios under both perfect observation and partial monitoring and propose two algorithms, for which we also establish some convergence properties. Moreover, we analyze the case in which the nature is assumed to be a player to develop a game-theoretical point of view towards the no-regret learning algorithms. Illustrative examples and simulation results demonstrate that the proposed schemes can assist the users to figure out the malicious nodes in a distributed way.