Mobile Agents Aided Congestion Aware Multipath Routing in Mobile Ad Hoc Networks

Abstract

Multipath routing is an important consideration to support quality of service in communication networks. This paper proposes mobile agents aided congestion aware multipath routing protocol (MAMPR) which uses the multipath routing opportunities to transfer the packets between a source and a destination in mobile ad hoc network (MANET). Existing routing protocols proposed for MANET uses 'shortest route' as a metric to find routes. MAMPR finds multipath routes based on quality of service metric, 'congestion'. In MAMPR, intelligent mobile agents moving around the network, collect and disperse the network topology information based on congestion state. The congestion state is determined based on the interface queue occupancy and MAC layer drops by the static agent deployed at the link layer. This information helps in making congestion aware multipath routing decisions. MAMPR was implemented, tested and compared with ad hoc on-demand distance vector routing protocol (AODV) in network simulator (NS-2). Results show that MAMPR spatially distributes the traffic across the network, avoids congestion and achieves better end-to-end delay and network throughput with reduced control overhead.