Vanet framework for optimal target selection in handoff using machine learning

Abstract

The impact of mobile networking has been far-reaching and has transformed many technologies for the better. Among the more pioneering technologies of recent times, autonomous vehicles (AV) are being touted as the future of transportation. AVs are featured prominently due to their ability to provide services across the implementations concerning smart-city applications. Since the newlinenodes (AV) involved are highly mobile, it is important to keep the device connected and adapt the dynamic topology change. The network should be capable of switching AV between access points (AP s) dynamic and must take place without much delay. Handoff is the process where a node switches between APs depending on the changes in the network topology. Various studies have claims that the use of Machine Learning or Deep Learning improves the efficiency of decision making in mobile networks. In this research work a dynamic handoff framework, inspired by the traditional Indian game of Kho-Kho is proposed. It selects the most optimal AP for newlinecommunication in a mobile environment. Initially the nodes within an AP are newlinecategorized as sticking and steering nodes. Identifying the nodes list for executing newlinethe Kho-Kho based handoff for the steering nodes. The framework makes use of Artificial Neural Networks (ANN) to perform the operation of handoff and selects the most optimal AP once the handoff trigger occurs. ANN is designed for multiple parameters, including signal strength, noise, and direction to make its decisions regarding handoff. The proposed Kho-Kho model has been newlineimplemented and analyzed using the NS3 simulator. Simulation results have indicated a higher performance of the proposed Kho-Kho approach when compared to the existing standard implemented for vehicular ad hoc newlinecommunications in IEEE 802.11p.