Autonomous Spectrum Assignment of White Space Devices

Abstract

White-space spectrum has temporal and spatial variations, and fragmentation, making the spectrum assignment for devices in this space challenging. In this paper, we propose an autonomous agent model for spectrum assignment of white space devices at a given location. Each white space device (WSD) acts autonomously out of self-interest, choosing a strategy from its bag of strategies. It obtains a payoff based on its choice and choices made by all other WSDs. Based on the payoffs received by different strategies, WSDs evolve their strategic profile over time. This has the effect of demographic changes in the population which is published as demographic profile by the Master. WSDs are expected to choose a strategy with a probability distribution based on this, for optimising network utilisation. In evaluation runs, network utilisation levels in such an approach are found to be high, and approaching optimal values computed in a centralised fashion.