Wireless Routing Protocol Based on Auto -- Learning Algorithm

Given the highly dynamic nature of wireless ad hoc network, routing is one of the most challenging tasks. Among the routing protocols that are in the process of normalisation is AODV (Ad hoc on demand Distance Vector). AODV is a reactive source based on routing protocol whose performance is very sensitive to several fixed parameters, such as. ACTIVE_ROUTE_TIMEOUT, HELLO_INTERVAL, MY_ROUTE_TIMEOUT, NET_DIAMETER, RREQ_RETRIES, .... In this paper we propose an adaptive version of AODV (A²ODV) where the protocol regularly adapts its routing decision to the state of the behaviour and the state of the network. In A²ODV each node maintains an estimation of the reliability as well as an expected residual uptime of its interfaces. A route is characterized by two metrics: (1) reliability defined as the product of its constituent links; and (2) Route residual lifetime estimated by the residual lifetime of its weakest link. Route selection is a function of these two metrics. Simulation shows that A²ODV clearly outperforms AODV with respect to packet delivery ratio and overhead traffic.