A Novel Dynamic Link Prediction Protocol for Frequent Link Disconnections in Vehicular Ad Hoc Networks

Abstract

– Vehicular Ad hoc Networks (VANETs) face challenges in maintaining communication links due to their large network sizes and rapidly changing topologies. Frequent link disconnections can impact the performance of vehicular applications, which are crucial for Intelligent Transport Systems (ITS). The objective of the research is to develop a dynamic link prediction protocol (NDLP) that can predict when a link is likely to become unavailable. By predicting link disconnections in advance, the protocol aims to reroute data packets through alternative paths to ensure uninterrupted communication. In this paper, a novel dynamic link prediction protocol (NDLP) is proposed to determine the duration of availability of the current path. This protocol predicts the duration of current path availability, aiming to pre-emptively predict connection breakdowns and reroute data packets via alternate paths. The proposed methodology involves the use of Newton's divided difference interpolation to assess the presence of active links to adjacent nodes. This technique employs historical data or real-time measurements to predict the future state of links. The primary focus is on predicting link disconnections before they occur and pre-emptively rerouting packets using an alternative path. The estimation of the time of link breakage and the ability to select the best route before the link breakage is analysed. Simulation results have proven the effectiveness of NDLP protocol with its counterpart protocols in terms of delay, packet delivery ratio and throughput.