Energy-Efficient Channel Switching in Cognitive Radio Networks: A Reinforcement Learning Approach

In this paper, we investigate energy-efficient channel switching for secondary users (SUs) in cognitive radio networks. Unlike existing schemes where SUs adopt the same channel switching strategies regardless of which channel they currently stay at, our scheme allows SUs to adapt their channel switching strategies to the primary users' (PUs') behaviors on the current channels and apply different channel switching strategies on different channels. Considering the unknown PUs' behaviors, we formulate a reinforcement learning problem which allows SUs to learn channel switching schemes by interacting with the environment. Through simulations, we demonstrate the effectiveness of the learned channel switching scheme.