Real-time anomaly detection in distributed IOT systems: a comprehensive review and comparative analysis
Pavlo Pustelnyk, Ye. V. Levus · Vìsnik Nacìonalʹnogo unìversitetu Lʹvìvsʹka polìtehnìka Serìâ Ìnformacìjnì sistemi ta merežì · 2025
The rapid expansion of the Internet of Things (IoT) has resulted in a substantial increase of diverse data from distributed devices. This extensive data stream makes it increasingly important to implement robust and efficient real-time anomaly detection techniques that can promptly alert about issues before they could escalate into critical system failures. Anomaly detection in data is essential in today’s interconnected landscape, as it facilitates the early identification of deviations from established baseline behavior that may indicate system malfunctions, security vulnerabilities, or operational inefficiencies. By promptly identifying these deviations, organizations can reduce downtime, optimize performance, and safeguard critical assets. This article provides a comprehensive review and comparative analysis of modern methods for detecting anomalies in distributed IoT systems. It examines a wide range of techniques, including traditional statistical approaches, distance-based methods, machine learning models, deep learning architectures, and explainable AI frameworks. Each category is evaluated with respect to detection accuracy, computational efficiency, and interpretability. Real-world examples – ranging from predictive maintenance in industrial IoT and energy management in smart grids to fraud detection in financial networks – demonstrate the broad practical applications of these techniques. The review further identifies current challenges and promising future research directions, including active learning-based approaches, which offer potential solutions to improve adaptability and reduce the reliance on large labeled datasets. The insights from this review provide a strong foundation for future research aimed at developing hybrid anomaly detection models that integrate advanced techniques to further enhance system adaptability and security in distributed IoT environments.