A performance model for an agent-based routing protocol for MANETs

A Mobile Adhoc Network is formed with no fixed infrastructure. The nodes in such a network communicate by short range transmission radio links. The mobile nature of the nodes may make or break some of the radio links. The nodes act as routers as well as hosts. Because of the dynamic change in the topology the design of routing protocols is a challenging task.; In this thesis, we develop a performance model using both the routing table dynamics of the continuum model [16] and random walk theory. The protocol we use is the Modified Local Mobile Agent Routing Protocol (Modified LMAR). We need a performance model for our modified LMAR to tune the parameters of the protocol. As the agents move randomly to update the routing information, we use random walk theory in developing a performance model for our protocol. The assumptions we make for this model are tested. Extensive experiments were done using the NS-2 simulator for testing the assumptions in the performance model. Our simulation results show that the assumptions are probably good enough to use for our performance model.