Applications of Belief Propagation in CSMA Wireless Networks

Abstract

The belief propagation (BP) algorithm is an efficient way to solve "inference" problems in graphical models, such as Bayesian networks and Markov random fields. The system-state probability distribution of CSMA wireless networks is a Markov random field. An interesting question is how BP can help the analysis and design of CSMA wireless networks. This paper explores three such applications. First, we show how BP can be used to compute the throughputs of different links in the network given their access intensities, defined as the mean packet transmission time divided by the mean backoff countdown time. Second, we propose an inverse-BP algorithm to solve the reverse problem: how to set the access intensities of different links to meet their target throughputs? Third, we introduce a BP-adaptive CSMA algorithm to find the link access intensities that can achieve optimal system utility. BP solves the three problems with exact results in tree networks. It may, however, lose accuracy in networks with a loopy contention graph. We show how a generalized version of BP, GBP, can be designed to solve the three problems with high accuracy for networks with a loopy contention graph. Importantly, we show how the BP and GBP algorithms in this paper can be implemented in a distributed manner, making them useful in practical CSMA network opera-tion.