A negotiation game for multichannel access in cognitive radio networks

We consider the problem of efficient opportunistic spectrum access in cognitive radio networks where there are multiple secondary users trying to share access to multiple channels. In our formulation, each user has a potentially different valuation of each channel and wishes to pick a channel in such a way as to maximize its benefit without interfering with other users. There is a fundamental tradeoff in this problem -- while information about other secondary users is useful in making a good channel sensing/access decision, the communication cost of gathering this information must be taken into account. We formulate the problem as a multi-round negotiation game in which the users try to gather "just-enough-information" to make their decisions. The channel valuations are modeled as independently uniformly distributed random variables between 0 and 1. We propose a threshold-based channel sensing policy based on observations from a previous work. For a two-user two-channel setting, we calculate optimal thresholds, and obtain the corresponding performance for cases with no information exchange, partial information exchange, and full information exchange. We then show how the optimal amount of information exchange varies with the cost of negotiation.