HDOMLM: Hybrid Dual Optimized Machine Learning Model for Cluster Head Selection in MANET

The selection of cluster heads with efficient energy awareness is a crucial concern in mobile ad hoc networks (MANETs). MANETs' energy efficiency is enhanced when a suitable cluster head (CH) is selected. Clusters are established in Mobile Ad hoc Networks (MANETs) to enhance communication among nodes. Efficient clustering is necessary to provide the rapid and precise transmission of information in the middle of nodes. In this work, we proposed a hybrid dual optimization of the machine learning model (HDOMLM). In this protocol, we are performing two different optimizations for clustering. Initial clustering is performed using Particle Swarm Optimization (PSO) based on node location and mobility as objection function cost. Once the initial cluster is formed, the selection metric is evaluated from the available network topology for the execution period. Delta difference, average distance, related energy, related mobility, and transmission delay are the five selection metrics used as features in the Optimized Machine Learning Model (O-MLM). By using O-MLM, we can classify that the node belongs to a normal node or cluster head node. To evaluate the effectiveness of our proposed HDOMLM, we are performing the simulation in MATLAB tool. Different performance metrics such as Alive nodes, Residual energy, Energy Tax, Average End-to-End Delay, Number of Successful packets received as BS, and Total processing time. The objective of network lifetime improvement is achieved in the proposed HDOMLM by 200 rounds and residual energy of 56% increased than earlier works.