Bayesian inference approach for establishing efficient routes in wireless ad-hoc networks

Abstract

Mobile ad hoc networks (MANETs) are dynamic wireless networks that have no fixed infrastructures and do not require predefined configurations. In this infrastructure-less paradigm, nodes in addition of being hosts, they also act as relays and forward data packets for other nodes in the network. Due to limited resources in MANETs such as bandwidth and power, the performance of the routing protocol plays a significant role. A routing protocol in MATET should not introduce excessive control messages to the network in order to save network bandwidth and nodes power. In this paper, we propose a probabilistic approach based on Bayesian inference to enable efficient routing in MANETs. Nodes in the proposed approach utilize the broadcast nature of the wireless channel to observe the network topology by overhearing wireless transmissions at neighboring nodes in a distributed manner, and learn from these observations when taking packet forwarding decision on the IP network layer. Our simulation results show that our routing approach reduces the number of control message (routing overhead) by a ratio up to 20 % when the network size is 60 nodes, while maintaining similar average route establishment delay as compared to the ad-hoc on demand routing protocol.