Location-Oriented Evolutionary Games for Price-Elastic Spectrum Sharing

For a spectrum sharing system using economic approaches, conventional models without geographic considerations are oversimplified. In this paper, we develop a model where geographic information, including licensed areas of primary users (PUs) and locations of secondary users (SUs), plays an important role in the spectrum sharing system. We consider a multi-price policy and the pricing power of non-cooperative PUs in multiple geographic areas. Meanwhile, the value assessment of a channel is price-related and the demand from the SUs is price-elastic. To maximize the payoffs of the PUs, we propose a unique quota transaction process. By applying an evolutionary procedure defined as replicator dynamics, we prove the existence and uniqueness of the evolutionary stable strategy quota vector of each PU, which leads to the optimal payoff for each PU selling channels without reserve. In the scenario of selling channels with reserve, we predict the channel prices for the PUs leading to the optimal supplies of the PUs and hence the optimal payoffs. Furthermore, we introduce a grouping mechanism to simplify the process. In our simulation, the effectiveness of the learning processes designed for the two scenarios is verified and our spectrum sharing scheme is shown efficient in utilizing the frequency resources.