Hybrid Auto ML Models for Intelligent Fault Detection in IoT Systems

The detection of anomalies in Smart IoT Sensor Networks (SISNs) plays a critical role in identifying unusual events or behaviors that may signal security vulnerabilities or operational inefficiencies. Traditional rule-based detection methods are often insufficient due to the dynamic, heterogeneous, and rapidly evolving nature of IoT ecosystems. In contrast, machine learning (ML) techniques have emerged as effective alternatives, enabling systems to automatically learn patterns from data and independently identify deviations without relying on rigid predefined rules. This journal article investigates a broad range of ML approaches applied to anomaly detection in SISNs, including supervised, unsupervised, and semi-supervised learning techniques. It discusses essential components such as data preprocessing, feature extraction, feature selection, and the choice of appropriate algorithms to enhance detection accuracy and computational efficiency. Real-world case studies are presented to illustrate the practical implementation of ML-based anomaly detection in IoT environments, demonstrating their effectiveness in identifying various types of anomalies, including cyber threats and sensor faults. Furthermore, the study evaluates detection performance using widely accepted metrics such as precision, recall, and F1-score to ensure comprehensive assessment. The paper concludes by highlighting current challenges, identifying potential research directions, and emphasizing the significant role of ML-driven anomaly detection in improving the security, reliability, and resilience of intelligent IoT sensor networks.