Federated learning-enabled lightweight intrusion detection system for wireless sensor networks: A cybersecurity approach against DDoS attacks in smart city environments

Abstract

Background Wireless Sensor Networks (WSNs) are vital in applications such as healthcare, smart cities, and environmental monitoring, but are vulnerable to cyberattacks due to their resource-constrained nature. Traditional Intrusion Detection Systems (IDS) depend on centralized architectures, which increase communication overhead and privacy risks and create a single point of failure. Objective This paper proposes a novel Federated Learning-based Lightweight IDS (FL-LIDS) that utilizes optimized lightweight models to enable real-time, privacy-preserving DDoS attack detection in resource-constrained WSNs for smart city environments and presents a comprehensive comparative analysis of models to evaluate their effectiveness within the Federated Learning (FL) framework. Methods FL-LIDS utilizes the optimized lightweight deep learning models for intrusion detection, which provides effective anomaly recognition with minimal resource usage, making it suitable for resource-limited WSN environments. The lightweight methods are evaluated in terms of their efficiency on the TON-IoT dataset. Results The study demonstrates the effectiveness of various FL-LIDS in detecting and preventing DDoS attacks with high detection rates and minimal latency. Metrics used to examine performance include accuracy, F1-score, precision, and recall in emulated WSN scenarios. The lightweight deep learning architecture optimizes accuracy and computational cost, with the lightweight hybrid CNN+LSTM model achiev…