Enhancing ML-based anomaly detection in data management for security through integration of IoT, cloud, and edge computing

The widespread adoption of cloud computing, edge computing, and the Internet of Things across various domains has enhanced data management and automation capabilities. However, large-scale IoT implementations face considerable challenges, including concerns about data quality, security risks, and the need to identify anomalies. This study introduces a multi-tiered machine learning-based approach to detect anomalies, specifically targeting security threats, performance irregularities, and sensor malfunctions within IoT-Edge-Cloud ecosystems. The proposed system improves detection precision, response times, and overall security resilience by incorporating XAI for transparent decision-making and a layered security approach to mitigate threats. The framework operates on several levels: (i) the IoT Layer, where secure microcontrollers collect and transmit sensor data; (ii) the Edge/Fog Layer, which conducts pre-processing and real-time filtering to minimize cloud reliance; (iii) the Cloud Layer, where ML-based anomaly detection algorithms, such as the Isolation Forest and Local Outlier Factor, examine data; and (iv) the Smart Single-Page Application architecture that integrates IoT-Edge-Cloud ecosystem, which offers low-latency visualization, secure data transmission, and interactive anomaly monitoring. Furthermore, XAI techniques improve interpretability by identifying key factors that influence anomaly classification and increase transparency for security analysts. A case study in IoT-Healthcare settings validated the efficacy of the proposed system in identifying network intrusions, sensor failures, and operational anomalies, achieving an AUROC score of 1.00 using an isolated forest. Comparative assessments demonstrate that this approach surpasses existing anomaly detection solutions by enhancing detection accuracy, decreasing latency through edge processing, and improving explainability with AI integration. The study concludes by exploring the challenges and advantages of combining IoT, cloud and edge computing for secure and scalable anomaly detection, thus providing insight into optimal database management and security strategies for IoT–cloud interactions.