Optimizing WSN Network Lifetime With Federated Learning–Based Routing

Abstract

Wireless sensor networks (WSNs) have become essential in applications such as environmental monitoring and smart infrastructure due to their ability to provide real-time data collection and analysis. However, WSNs face significant challenges related to limited battery life and the need for efficient energy management, which can impact their performance and longevity. Traditional routing protocols often fail to adapt to dynamically changing conditions and energy constraints inherent in WSNs, necessitating innovative approaches to enhance energy efficiency and network longevity. This paper introduces a federated learning–based adaptive routing (FLAR) model designed to address these issues by integrating federated learning with adaptive routing protocols. The primary aim of this research is to optimize energy utilization across the network and extend the operational lifespan of WSNs. The novelty of the proposed FLAR model lies in its unique combination of energy-aware participant selection (EaPS), adaptive model compression (AMC), and dynamic data sampling (DDS), which collectively enhance energy efficiency and adapt dynamically to changing network environments. The FLAR model was simulated and analyzed using Network Simulator 2 (NS2) under various network conditions and node densities. The results demonstrate that the FLAR model significantly outperforms traditional protocols by reducing energy consumption by up to 30% and enhancing network longevity by approximately 25%. Additionally, the proposed methodology improves packet delivery ratio and reduces latency, making it a robust solution for sustainable WSN deployment. Overall, the FLAR model offers a significant advancement in WSN technology by effectively managing energy resources and dynamically adapting to network changes.