A Load-Balancing Routing Protocol Based On Traffic Predition By A Neural Network Model In Ad Hoc Network

Ad Hoc network is defined as a collection of mobile platforms or nodes where each node is free to move about arbitrarily, the most prominent routing protocol are AODV and DSR. They use the shortest path as their routing criteria. This route selection philosophy result in routing load concentrate on just a few nodes,then lead to network congestion,and this congestion not only causes high transmission delay but also forces several nodes to be early expired by consuming most of their power to forward others'packets. So,it is important for the research on the load-balancing routing protocol.Proposes a new load-balancing routing protocol NNP-LBRP(Load-Balanced Routing Protocol Based On Neural Network Prediction Model),in which a traffic prediction method based on a neural network model is introduced for dynamically balancing the load of an Ad Hoc network during data transmitting. In this protocol, the traffic prediction model developed by using RBF neural network is used to predict the traffic load of a node at next time point in order to judge whether the node is suitable for an intermediate node of a valid route. According to prediction results, the route which was established and had optimal load in the route discovery will be updated appropriately before the network congestion occurs which results from the dynamically changed traffic of nodes. In addition, the selectively transfer of RREQ on the basis of the load status and the delay response of destination node also improve the network performance to some extent.Simulation was done on ns2-based platform for comparing the properties of NNP-LBRP,NRT-LBRP with conventional AODV protocol through three different experiments.and the results showed the good performance on the presented protocol.In the experiment of different packet sent rate, the packet delivery ratio increases approximately by 4%,10%,the average end-to-end delay reduces by 16%,27% and the normalized routing overhead decreases by 25%,26% on average, in comparison to that of NRT-LBRP,AODV protocol, respectively.