Link State Estimator for VANETs Using Neural Networks

Abstract

In Vehicular Ad-hoc NETworks (VANETs), it is important to consider the quality of the path used to forward data packets. Because of the fluctuating conditions of VANETs, stringent requirements have been imposed on routing protocols and thus complicating the entire process of packet delivery. To determine which path is the best, a routing protocol relies on a path assessment mechanism. In this paper, the problem of link quality estimation in VANET networks is addressed. Based on the information gathered from the packet decoding errors at the physical layer, a novel link quality estimator is proposed. The proposed link quality estimator named LSENN for Link State estimation based on Neural Networks, has been tested under realistic physical layer and mobility models for reactivity, accuracy and stability evaluation.