Distributed exploration bandit for dynamic spectrum access

In this paper, we address the problem of allocating channels to wireless users in the multiuser-multichannel dynamic spectrum access (DSA) scenario. To solve this problem, we introduce a special form of multi-armed bandit (MAB), named exploration bandit, into DSA for the first time. Two distributed algorithms are proposed to maximize the channel rewards (e.g., transmission rates) of the users. Meanwhile, the proposed algorithms can minimize the error probing probability in the exploration phase to solve the potential collision problem when different users target at the same channel in the same time. Simulation results show that our scheme could achieve higher channel rewards than classic MAB approaches.