A Machine Learning Based Cluster Optimization Technique for Vehicular Ad Hoc Network

Abstract

Many precious lives are lost world-wide due to road accidents. To counter this issue, one of the vital solutions is Vehicular ad hoc networks (VANETs). VANETs is recognized as a network aggregated of wirelessly linked vehicles. Apart from providing safety to the vehicles, VANETs help in management of traffic and also provide infotainment applications. Owing to the high speed of vehicles and varying network topology, efficient communication among the vehicular nodes is of extreme importance. Clustering is a technique among many others, which targets to improve communication proficiency in VANETs. In each cluster, there is one cluster head (CH) whose responsibility is to supervise the whole cluster. The cluster heads (CHs) are responsible for all the intra-cluster and inter-cluster communications. The effectiveness of a network is evaluated by number of CHs, load on each CH and lifetime of clusters. VANETs are a kind of mobile ad hoc networks (MANETs), however, clustering algorithms of MANETs are not applicable to VANETs. Therefore, a clustering algorithm is designed which can effectively work in high mobility nodes scenario of VANETs. The primary purpose of this research is to create an optimal algorithm for vehicular node clustering in VANETs. This technique will enhance the overall performance of the network by optimizing the clusters and boosting the life-time of clusters. A novel algorithm for VANETs clustering, based on Moth-Flame Optimization (MFO), titled CAMONET, is designed in this research work. CAMONET is based on nature inspired algorithm and creates optimized clusters for reliable and efficient transmission. MFO algorithm is an innovative bio-inspired optimization algorithm which is utilized to tackle complex optimization problems. Since clustering is regarded as an optimization problem so MFO algorithm is much suitable for it. ii CAMONET is evaluated experimentally with other well-known procedures like Multi Objective Particle Swarm Optimization (MOPSO), Clustering algorithm for VANETs based on Ant Colony Optimization (CACONET) and Comprehensive Learning Particle Swarm Optimization (CLPSO). Several experiments are conducted to measure the effectiveness of these procedures. The results are accomplished by modifying the values of number of nodes, the grid size and the transmission range of the nodes in the network. The direction, velocity and the range of transmission of the vehicular nodes are regarded as the notable factors for optimized clustering. The results signify that CAMONET delivers near optimal results which develops it into an efficient method to perform vehicular clustering with the objective of enhancing the network’s overall performance. The results depict the diversity and the flexibility of our technique, as for different network scenarios, our proposed algorithm is producing better results comparatively. The cumulative average results for all the grid sizes are 27.1% for CAMONET while 36.3% for CACONET, 54.9% for CLPSO and 58.7% for MOPSO. From these results, it is evident that the proposed technique is producing far superior results for all the traffic scenarios discussed in our experiments. Our algorithm covers the entire network and generates least number of clusters, consequently reducing the routing cost of the vehicular network. Smaller number of clusters ensures reduced packet delays and the number of hops is also diminished. As a result, a proficient vehicular network based on clusters can be envisioned.