EFFICIENT AND OPTIMIZED DATA ROUTING IN WIRELESS SENSOR NETWORK USING NEURAL NETWORK

Abstract

 Wireless Sensor Networks (WSNs) are crucial in applications such as environmental monitoring and military surveillance, where energy efficiency is key. This study focuses on optimizing cluster head selection to enhance energy efficiency in WSNs using Backpropagation Feed-forward Neural Networks (BP-FNN) and artificial intelligence. Key parameters include node proximity, residual energy levels, and network topology. Dynamic adaptation based on transmitted sensor data ensures efficient resource utilization. Random weights and biases fine-tune the selection process, balancing energy consumption across the network. The BP-FNN model uses random weights and biases for activation, transferring the weighted sum to the hidden layer. The sigmoid activation function predicts cluster head selection probabilities, with a linear function processing the output. A Backpropagation-based training algorithm refines the model by propagating errors backward through the layers. Performance analysis considers packets sent to the cluster head, alive nodes, dead nodes, and total energy consumption. MATLAB simulations demonstrate the proposed protocol's superiority over LEACH, E-MODLEACH, and DEEC protocols in cluster head formation and energy efficiency, highlighting its potential for sustainable WSNs.