Edge-Based Intelligent Software Architecture for Fault Detection in IoT Networks Using Mathematical Models and Machine Learning

The rapid growth of Internet of Things (IoT) systems has increased the need for reliable and timely fault detection to ensure system performance, safety, and availability. Many existing fault detection approaches rely on cloud-centric processing, which can introduce latency, increase bandwidth consumption, and depend on continuous network connectivity, limiting theireffectiveness in time-sensitive IoT applications. This study proposes an edge-based intelligent software architecture that integrates mathematical modeling with machine-learning techniques to enable accurate, interpretable, and low-latency fault detection closer to data sources. The proposed approach combines residual-based analytical models with data-driven anomaly detection and distributes fault detection tasks between edge and cloud layers to balance responsiveness and computational efficiency. The framework is evaluated through simulation under multiple fault scenarios and varying network conditions. The results show that the proposed hybrid approach improves fault detection effectiveness, reduces detection latency, and lowers communication overhead compared with cloud-only and single-method solutions, while maintaining stable performance as the number of connected IoT devices increases. Overall, the findings indicate that integrating mathematical models and machine learning within an edge-based framework provides a practical and scalable solution for reliable fault detection in modern IoT environments.