Q-Learning Based Bidding Algorithm for Spectrum Auction in Cognitive Radio

Cognitive radio has been put forward to make efficient use of scarce radio frequency spectrum. Once available frequency bands have been detected using spectrum sensing algorithms, spectrum auction can be employed to allocate the detected available frequency bands to secondary users. In this paper, a bidding algorithm based on Q-learning for secondary users is proposed. Secondary users employ the proposed algorithm to learn from their competitors and automatically place better bids for available frequency bands. Simulation result shows the proposed algorithm is effective. This work is a part of the efforts toward building a cognitive radio network testbed.