Distributed Dynamic Spectrum Access in Multichannel Random Access Networks with Selfish Users

Dynamic spectrum allocation schemes enable users to share spectrum resources by exploiting the variations in spectrum demand over time and space. Performing dynamic spectrum allocation centrally can be prohibitively complex. Therefore distributed schemes in which users can access the available channels independently may be preferable to centralized allocation. However, in distributed dynamic spectrum access, the lack of central coordination makes it difficult to utilize the system resources efficiently. Furthermore, if some or all of the users decide to deviate selfishly from the commonly agreed access procedure, this may have a decisive effect on system performance. In this paper we investigate the effect of incomplete information and selfish behavior on system performance in wireless access systems. We extend previous work by studying a distributed multichannel wireless random access system. Using a game-theoretic approach, we analyze the behavior of users in the selfish system and derive the transmission strategies at the Nash equilibrium. Our results show that lack of information leads to substantial degredation in performance of cooperative systems. We also show that there is a large incentive for selfish behavior in such cooperative systems. Selfish behavior of all users, however, causes further performance degradation, particularly in high load settings.