Load-aware self-organizing spectrum access for small cell networks

This paper investigates the problem of self-organizing spectrum access for small cell networks, in which different cells have different loads. Since there is no central controller and information exchange, we formulate the problem as a graphical game. We prove that the game is an ordinal potential game which has at least one pure strategy Nash equilibrium (NE). More importantly, the potential function of the game is the aggregate weighted interference. A multi-agent learning algorithm is proposed to achieve NE of the game. The algorithm is fully distributed and self-organizing. Simulation results show that the algorithm converges not only in statistic scenarios, in which only large-scale path loss is considered, but also in dynamic scenarios, in which channel fading is considered.