To participate or not in spectrum auctions with entry fee: Bayesian game theoretic approach

In this paper, competition among multiple secondary users (SUs) for spectrum access is modeled as a simultaneous repeated auction. Upon participation in an auction, a SU is charged with an entry fee. However, participation does not ensure an access to the channel. This tradeoff leads it to decide either for or against entering the auction. We consider no cooperation among the SUs, and model this situation as a Bayesian game. A modification of the standard regret testing procedure is proposed to fit our system model. Our proposed procedure converges to Nash equilibrium (NE) of the game. Since this procedure is computationally expensive, we propose a less expensive learning based procedure for the decision taking. We present computer simulation results to compare the average profits and bidding efficiencies over time for the proposed procedure. We also compare their bidding efficiencies to another procedure in the literature, based on second highest bid prediction.