Energy Aware-Adaptive Walrus Optimization Algorithm (EA-AWaOA) Based Routing And Deep Q Learning Based Load Balancing In Vehicular Fog Computing

Abstract

The emerging vehicular applications demand for a lot more computing and communication capacity to excel in their compute-intensive and latency-sensitive tasks. Fog computing, which focuses on moving computing resources to the edge of networks, complements cloud computing by solving the latency constraints and reducing ingress traffic to the cloud. However, efficient load management in vehicular fog computing is challenging due to the dynamic nature of the Vehicular Ad-Hoc Network (VANET).  In this paper introduced the Energy Aware-Adaptive Walrus Optimization Algorithm (EA- AWaOA) for optimal selection of cluster heads and routes by the residual energy, inter-vehicle distance, closeness and vehicle velocity in fog-based VANET. Initially, it performs clustering by AWaOA while considering the mobility of fog devices. Secondly, a capacity-based load-distribution mechanism by Deep Q Learning (DQL) efficiently balances the load among the vehicular nodes. DQL is introduced for estimating the capacity load through the queue status obtained from each ground vehicular node, a load balancing scheme based on DQL and a reward control function for rapid convergence of Q-learning. This strategy minimizes the correlation among consecutive samples and improves learning stability. It proposes a mechanism for identifying a vehicle‘s departure time from the cluster, which allows predicting the future position of the vehicle within the dynamic network. The simulation results in the Network Simulator 2 (NS2) environment showed that as the balanced network energy consumption, reduced network delay, and improved network utilization in the vehicular fog computing.