Learning Nash Equilibria in Distributed Channel Selection for Frequency-Agile Radios

Abstract

Wireless communication networks are evolving towards self- configuring, autonomous and distributed multiagent systems in which nodes are deployed randomly and have to adapt to the en- vironment in which they operate. A cognitive network is a self- organising system that relies on the ability of its autonomous nodes to support communication in an adaptive and distributed manner. In this paper we address the distributed channel selec- tion problem, which is a crucial component of many cognitive networks scenario, in the context of frequency-agile radios that are able to operate in multiple frequency bands simultaneously. We formulate the problem as an N-player stochastic game with incomplete information. We prove that by adopting a simple reinforcement scheme, namely learning automata, nodes will converge to a Nash equilibrium, under the assumption of sym- metric interference between the players