Federated Learning Based Energy-Aware Clustering and Sleep Scheduling for Faulty Node Detection in Wireless Sensor Network

Wireless Sensor Networks (WSNs) play a vital role in environmental monitoring, such as forest fire detection, where real-time, energy-efficient data acquisition is crucial. However, faulty node detection impacts network reliability, power consumption, and data accuracy. Traditional centralized approaches suffer from high power usage, communication overhead, and scalability issues.This paper proposes an efficient framework integrating Federated Learning (FL), Energy-Aware Clustering (LEACH), and Adaptive Sleep Scheduling to enhance fault detection and network performance. FL enables decentralized model training, ensuring data confidentiality while reducing transmission overhead. LEACH optimizes power consumption by dynamically selecting Cluster Heads (CHs) based on residual energy, minimizing communication load. Adaptive Sleep Scheduling extends network lifespan by efficiently managing node activity. Experimental results show that the proposed scheme improves fault detection accuracy, reduces power consumption, and lowers false alarms compared to centralized approaches, making it suitable for real-time anomaly detection in WSN applications.