Fuzzy tuned gossip algorithms in mobile ad hoc networks

Abstract

Mobile ad hoc networks (MANET) are modeled as agents that form communities without infrastructure, for a random period of time and with usually cooperative behavior. The nodes of MANET often carry information to disseminate. The dynamics of information delivery, mostly referred as average consensus, is a common problem in these networks. The gossip protocols are designed to implement this task. The standard algorithms that are used in these protocols exploit the network describing matrix, aka Laplacian, and exchange information to all node neighbors. In static networks the problem can be considered as an output feedback problem but in the case of MANET the problem is getting complicated due to the continuous change of network topology. In this paper the fuzzy reasoning approach is proposed to tune and leverage the gossip protocol. Illustrative simulations are included to demonstrate the application of the method and to present comparative results in various cases.