Adding to the continual quest for a scalable routing approach to deal with continual changing mobile ad hoc networks (MANETs), this paper describes a proxy-assisted routing scheme, called PART, that uses broadcast zones and proxy nodes to limit routing overhead. This contrasts with other on-demand approaches that “re-invoke a [costly] route discovery procedure from the beginning” when collisions or routes break. PART distinguishes between these failures and uses a proxy node to store some of the route information in the hope that just a re-send is necessary. Thus, PART is a hop-by-hop routing protocol. If nodes have moved, then a unicast transmission between proxy nodes helps determine the correct route; PART attempts to be resilient to route failures in a dynamic mobile network.
The paper contrasts PART with other approaches including ad hoc on-demand distance vector (AODV) and dynamic source routing (DSR). The PART protocol is then discussed, and the authors show how its routing table is constructed and maintained. A number of algorithms detail how the PART protocol deals with route discovery, proxy node communication, limiting of the broadcast zone, repairing routes at the proxy node, and detection of proxy failure conditions.
Of particular interest are the simulation results; since PART is meant for a MANET, the experiments deal with the effects of node movement, including source node, proxy node, and destination node movement. The PART protocol is designed to tolerate node movements, in terms of throughput, delay, and overhead. The authors’ results show that “PART reduces the average end-to-end delay up to 4% lower than AODV and 25% lower than DSR,” but there is little difference in other areas like packet loss. Other tests were performed over a large-scale network to test for average delay, packet loss rate, normalized routing load measurements, and throughput measurements. The overall simulation results of PART show that it “significantly reduces the normalized routing load by almost 55% and the packet losses by almost 30%, and increases throughput almost 70% if compared to the traditional routing protocols.”