Enhanced Lightweight Blockchain-Based Integrated Finite Element Neural Network for Secure Routing in Wireless Sensor Networks

Abstract

Ensuring secure and energy-efficient routing in wireless sensor networks (WSNs) remains a critical challenge due to limited resources and susceptibility to attacks. This paper introduces a novel model, PBFT-IFENN—a lightweight blockchain-based Practical Byzantine Fault Tolerance system integrated with a Finite Element Neural Network designed to address these issues comprehensively. The model leverages the Duck Swarm Algorithm (DSA) for dynamic cluster head selection, whereas data aggregation is optimized through an Integrated Finite Element Neural Network (I-FENN), incorporating Physics-Informed Neural Networks (PINNs) for reducing redundancy. Security is ensured using a lightweight PBFT consensus, and efficient routing is maintained via the Musical Chairs Routing Protocol (MCRP). Performance was evaluated using the NS-3 simulator and compared against recent benchmark protocols including UDTP-RPR, GTGAN, DRPL_SDN, DA-TD3, and ILEACH. Results demonstrate that PBFT-IFENN achieves superior outcomes, with a packet delivery ratio (PDR) of 95%, energy consumption of 1.12 J for 100 nodes, and throughput of 750 Kbps. These results significantly outperform all referenced methods, showcasing enhanced data reliability, reduced latency, improved network longevity, and robust security. The suggested method offers a thorough and scalable solution for secure and efficient communication in modern WSN deployments.