Radial Basis Neural Network–Based Bayesian Fuzzy Clustering Fostered Object Tracking in Wireless Sensor Networks

Abstract

Efficient object tracking in wireless sensor networks (WSNs) remains challenging due to high energy consumption, node failures, and tracking inaccuracies. This paper introduces a novel radial basis neural network–dependent Bayesian fuzzy clustering object tracking protocol (RBNN-BFC-OT-WSN) that uniquely integrates three innovations: (i) Bayesian fuzzy clustering for dynamic and probabilistic cluster head selection to achieve balanced energy usage; (ii) radial basis neural network for high-precision, real-time tracking of nonlinear target movements; (iii) marine predator algorithm for adaptive optimization of RBNN weight parameters, enhancing convergence speed and prediction accuracy. Unlike existing GA-LEACH-WSN, VA-PSO-WSN, and LBR-GSO-WSN models, this approach combines clustering intelligence with neural learning and metaheuristic optimization in a unified framework. Simulation results using MATLAB and NS-2 demonstrate that RBNN-BFC-OT-WSN achieves 12.6%–29.1% lower delay, 18.4%–32.8% less energy consumption, and 53.6%–68.2% higher delivery ratio than baselines, while maintaining robust performance under high mobility and dense node deployments. This integrated framework offers a scalable, energy-efficient, and accurate solution for WSN-based surveillance applications.