Mobile Ad Hoc Networks On-demand Routing Protocol Research

In recent years, the common wireless network already could not satisfy the demands any more due to the needs for base stations. As a result, Mobile Ad Hoc Network technology came into being. This kind of network is distributed, and supports rapid-expansion and self-organization without the requirement of infrastructures.The Mobile Ad Hoc Network routing protocol could be divided into two categories: Table-Driven Routing Protocol and On-Demand Routing Protocol. The Table-Driven Routing Protocol prescribes that every node in the network should maintain a routing table which indicates how to reach any other nodes. Its Average End-to-End Delay is very short, but will consume massive resources to gain a considerable amount of redundant information. The On-Demand Routing Protocol would originate a routing requirement process by the source node merely when there is a communication demand. But this method increases the End-to-End Delay remarkably, and do not suit to the high real-time request situation.According to the background described above, this paper introduces the Mobile Agent to the Ad Hoc On-Demand Distance Vector Routing Protocol, and proposes a new algorithm -The Mobile Agent Based AODV On-Demand Routing Protocol (MA-AODV). The Mobile Agent is transmitted continually, collects information about the network topology, and updates the routing table of the node dynamically which is being visited. The improved algorithm looks as on-demand essentially, but the Mobile Agent is initiative.In the experiment process, the problems which manifested in the trace file, had been made a further improvement by purpose. Some key points with emphasis on the vanishing of Mobile Agent, the motion strategy of Agent, the judgment of information validity carried by Agent and the congestion control brought by the transmission of Agent, were discussed and given the final solutions through experiments.This algorithm enhances the cognition of node connectivity regarding the entire network. Therefore there is a remarkable reduction of the system's Average End-to-End Delay. It also reduces the initialization of RREQ engendered by source nodes, enhances the real-time capability enormously, and improves the network performance obviously.