Levenberg–Marquart logistic deep neural learning based energy efficient and load balanced routing in MANET

Abstract

As the advent of new technologies grows, the deployment of mobile ad hoc networks (MANET) becomes increasingly popular in many application areas. In addition, all the nodes in MANET are battery operated and the node mobility affects the path stability and creates excessive traffic leads to higher utilization of energy, data loss which degrades the performance of routing. So, in this paper we propose Levenberg–Marquardt logistic deep neural learning based energy efficient and load balanced routing (LLDNL-EELBR) which is a machine learning method to deeply analyze the mobile nodes to calculate residual load and energy and it also uses logistic activation function to select the mobile node having higher residual energy and residual load to route the data packet. Experimental evaluations of three methods (LLDNL-EELBR, multipath battery and mobility-aware routing scheme (MBMA-OLSR) and opportunistic routing with gradient forwarding for MANETs (ORGMA)) were done and the result reveals that LLDNL-EELBR method is able to increase the through put and minimizes the delay and energy consumption in MANET when compared to works under consideration.