An agent-based model of the risk-based spectrum auction in the Cognitive Radio Networks

In this paper, we propose an agent-based model for spectrum trading in the shared use model of dynamic spectrum access. Spectrum trading is employed using the single-unit sealed-bid first-price auction, which takes into the account risk due to the imperfect spectrum sensing. Bidding strategies of the bidder are controlled by the reinforcement learning algorithm. We consider cooperative energy-based spectrum sensing as a spectrum sensing mechanism. Two different decision fusion strategies, which provide different levels of risk are discussed. The results demonstrate that in risky environment, total revenue and total payoff of the auctioneer and bidder respectively is higher, than in the case of system with lower level of risk. On the other hand, normalized revenue and payoff per a single auction round is higher in the case with lower level of risk. Moreover, the results have shown that the optimum sensing time for maximizing revenue and payoff is different.