An AODV Routing Protocol Optimization Strategy Based On Grey Prediction Model

Mobile Ad hoc network is a self-organization, multi-hop, no center wireless network. It is a quick and easy networking without infrastructure, without time and space constraints. Its properties make it very suitable for military applications or some emergency situations. Ad hoc network has a specificity routing algorithm due to its dynamic changes of network topology characteristics. Designing a suitable routing protocol for Ad hoc network becomes a research hotspot.This paper does a detailed description of Ad hoc network, then introduces the widely used routing protocols and points out advantages and disadvantages of AODV protocol. After that, it introduces the basic concepts of load balancing routing protocols, then does an in-depth analysis for random early detection and grey system theory, and applies to the feasibility of the design of routing protocols.However, AODV routing protocol itself in the process of route discovery does not take into account the current load of intermediate nodes, this may result in localized network congestion; AODV routing protocol relies solely on the number of hops as the routing metric, regardless of route load. Because of these, the link may become busy and the network delay may increase. According to the load balancing idea, we proposed a new routing protocol called Q_AODV. Q_AODV routing protocol calculates the traffic load through the interface queue of MAC layer. In the process of route discovery, intermediate nodes use access control based on random early detection theory. The received route request packet is dropped probability in order to reduce the pressure of the high-load nodes. Meanwhile, destination nodes use delayed response to select the best way according to the route load other than the number of hops as the routing metric.Q_AODV routing protocol uses the optimized route discovery strategy which makes data packets transfer along the light load paths, so the data stream can distribute evenly throughout the whole network and localized overload will never exist. However, Q_AODV routing protocol lacks the route maintenance strategy. If the process rate of nodes is less than the receive rate of nodes, there will exist link congestion and packet loss. So this paper makes a further optimization for AODV routing protocol. It adds flow prediction mechanism to Q_AODV routing protocol and proposed a new routing protocol called GM_AODV. Compared with Q_AODV routing protocol, GM_AODV routing protocol has a well route maintenance mechanism. In the process of route maintenance, nodes predict its load value based on grey system theory, and then the forthcoming congestion can be prevented by the route-discovery mechanism according to the nodes’load state in the future. So the transfer reliability will increase. Finally, simulation is done on NS2-Based platform for comparing the properties of AODV, Q_AODV and GM_AODV through different sending rate and different number of connections. The NS2simulation shows that compared with AODV, Q AODV reduces the average delay and packet loss rate while increasing the data forwarding capacity of nodes. And GM_AODV has a better performance than Q_AODV.This paper makes an optimization for AODV routing protocol from the angle of flow control. However, there are many factors that affect the protocol performance. In future study, we shall take these factors into account.