Clar: A Novel Cluster Oriented Agent Based Routing Protocol for Manet

In this paper we propose a novel algorithm CLAR to build an agent based routing protocol using ACO technique in cluster based mobile ad-hoc networks. Clustering makes possible hierarchical routing in which paths are recorded between cluster-heads instead of nodes. This reduces routing overheads. Presence of an unstable and non-reliable cluster-head degrades the performance of the network since cluster-heads of the network take active role in routing messages between a source destination pair. The proposed algorithm chooses the most reliable and stable node as cluster-head depending on four criteria i.e.: stability, battery power, degree, and trust value of the node. It also uses a multiagent based routing algorithm to generate load balanced routes between source and destination in cluster-based network. Performance of CLAR is compared with other existing efficient algorithms such as AODV, CBRP, Wu & Li’s Algorithm, dominating set based clustering algorithms, agent-based algorithms named as SACO and MALBACO. Simulation results prove that the proposed protocol combines merits of both clustering approach and agent based approach and provides efficient packet delivery even in the presence of high load in the network