Abstract:

We propose a spectrum allocation algorithm based on the auction theory and Gaussian process regression in cognitive wireless networks. Based on vickreyclarkegroves(VCG ) auction model, the algorithm takes the requirement for communication quality of cognitive users into consideration, and formulates a more effective utility function. During the auction, cognitive users predict the bids of other cognitive users by utilizing historical auction data to optimize their bidding strategy. The auctioneer assigns spectrum according to all the bids of cognitive users. Theoretical analysis and simulation results prove the effectiveness of the proposed algorithm, and verify that it can enhance spectrum utilization and the earnings of cognitive users.

Key words: auction model;spectrum allocation;Gaussian process regression;VCG mechanism