No-Regret Learning in Collaborative Spectrum Sensing with Malicious Nodes

In cognitive radio network, spectrum sensing is a key component to detect spectrum holes (i.e., channels not used by any primary users). Collaborative spectrum sensing among the cognitive radio nodes is expected to improve fidelity of primary user detection. However, malicious nodes can significantly impair the collaborative spectrum sensing by sending the wrong reports to the fusion center. To overcome this problem, in this paper we propose non- regret learning algorithms to study the non-constructive secondary users caused either by evil-intention or altruistical incapability. Both perfect observation and partial monitoring are investigated, and two algorithms are proposed respectively. Some convergence properties are also shown. Moreover, we also analyze the case in which the nature is assumed to be a player. Illustration example and simulation results demonstrate the proposed schemes can automatically pick the malicious nodes in a distributed way.