Optimal Feature Selection to Improve Vehicular Network Lifetime.

Abstract

The evolution of the Internet of Things (IoT) leads to the ascent of the need to develop a protocol for Low power and Lossy Networks (LLNs). The IETF ROLL working group then proposed an IPv6 routing protocol called RPL in 2012. RPL is in demand because of its adaptability to topology changes and its capacity to identify and evade loops. Although RPL in the recent past was only used for IoT networks. But, contemporary studies show that its applicability can be extended to vehicular networks also. Thus, the domain of the Internet of Vehicles (IoV) for RPL is of significant interest among researchers. Since the network becomes dynamic when RPL is deployed for vehicular networks, the heterogeneous network suffers from extreme packet loss, high latency and repeated transmissions. This reduces the lifetime of the network. The idea behind this article is to simulate such a dynamic environment using RPL and identify the principal features affecting the network lifetime. The network setup is simulated using the Cooja simulator, a dataset is created with multiple network parameters and consequently, the features are selected using the Machine Learning (ML) technique. It is inferred from the experiment that increasing PDR and reducing EC will improve the overall network lifetime of the network.