Self-learning ad-hoc routing protocol

Abstract

As the technologies for ad-hoc networks emerge, an ad-hoc routing protocol that can supply connectivity to mobile devices while utilizing a minimal amount of bandwidth as overhead becomes increasingly important. Minimal packet delay and the conservation of power and memory in wireless devices must also be considered. In this research, the self-learning ad-hoc routing protocol (SARP) is proposed. This protocol discovers information about the network topology while forwarding packets that it would have to process anyway. Due to the large number of routes learned through forwarding packets and the route discovery procedure, the overhead created by the routing protocol is reduced. The performance of SARP is compared with AODV in terms of data packet delivery ratio, average end-to-end delay, control overhead, average energy consumption and average number of routes stored in the routing table of each node at any given time. The simulation results show that SARP had up to a 34% reduction in the percentage of overhead generated by the routing protocol and up to a 26% increase in the data packet delivery ratio over AODV. The average energy consumed by SARP was comparable with that of AODV although the average end-to-end delay for SARP was slightly higher than AODV at high mobility rates. The average number of routes stored in the routing table of each node in SARP was always greater than in AODV.