A Fast Neighbor Discovery Scheme Based On Mobility Prediction In Vehicular Networks

In vehicular ad hoc networks, the rapid moving of nodes leads to the frequent changes of network topology. A fast neighbor discovery algorithm becomes an important factor influencing the performance of network protocols. In traditional mobile ad hoc networks,there are a lot of protocols which can be used to improve the performance and efficiency of data transformation, such as routing protocol and hello protocol. Hello protocol is always combined with routing protocol to improve the performance of routing protocol since it can be used for neighbor detection and to maintain the neighbor table.If we have the most accurate neighbor table, we will get the most efficiency data transformation in VANET. This thesis proposed a novel mobility prediction hello protocol based on Kalman filter, named KFH( Kalman Filter-based Hello protocol).In KFH, the time domain was divided into slots of equal length. Each node had a predict model based on Kalman filter used to make mobility prediction according to the time and space features of the vehicles in the mobile network environment. When a node predicted its own position in the next slot using the predict model, the positions of its neighbors would also be predicted using the same model. If the prediction error was too big, a hello message contained the real position of the node would be broadcasted and neighbors which received the message would update the corresponding predict model with the latest data. In each slot, each node will calculate the distance between its neighbors and itself. If the distance between two mobile nodes is too long to transfer data, the neighbor information will be deleted. So it always keep a latest and accurate neighbor table of each node.We compares the three HELLO protocol: KFH, Autoregressive Hello protocol( ARH) and a widely used certain time interval-based HELLO protocol. Simulation results show that KFH achieves a higher efficient neighbor discovery and improves the performance of HELLO protocol. KFH reaches the lowest neighbor error rate(only about 2%) and the shortest neighbor discovery delay under the same network overhead.