PREDICTING THE SURVIVABILITY OF LINKS IN AN AD HOC NETWORK

Abstract

Ad hoc networks are composed of a collection of self-configuring mobile routers that do not rely on any pre-existing network infrastructure. Nodes within an ad hoc network are expected to be able to route data-packets for other nodes in the network in a peer-level multi-hopping networks, constructing a interconnecting structure for the mobile nodes. Routing schemes that are adaptations of static network routing protocols do not perform well for the dynamic, infrastructure-less and self-operated environment of ad hoc networks. Therefore, ad hoc networks require a novel routing scheme that provides efficient and high throughput communication among mobile nodes. The ideal ad hoc network routing scheme must be scalable and able to cope with constantly changing topology that results from node mobility, conserve precious transmission power by reducing transmission overhead and avoid packet collisions. Routing protocols for ad hoc networks developed in existing researches are either tabledriven (proactive) and demand-driven (reactive). We look into improving the performances of reactive protocols through caching and link survivability predictive strategies. Based on the On-Demand Multicast Routing Protocol (ODMRP), we enhanced the protocol to include route selection metrics that utilise delay time and hop count and implement and evaluate our modified protocol by comparing simulation results, using GloMoSim, with performance of the basic ODMRP (without enhancement). We proved that the modified protocol performs significantly better in Multicast traffic for most simulated scenarios and the performance is more consistent across the varied scenarios than the basic protocol. The modified protocol performs significantly better in Multicast traffic for most node speeds and Multicast group sizes. However, for Unicast traffic, the performance difference between the modified ODMRP and the basic ODMRP is insignificant for most node speeds.