Computer Science · Computer Networks and Communications
Advancing IoT Anomaly Detection through Dynamic Learning
A. V. Kalpana, T. C. Jermin Jeaunita, R. Anandh, T. Chandrasekar, A. Thilagavathy
See on the mapAbstract
In our interconnected era, the internet of things (IoT) has transformed numerous sectors, notably manufacturing, by facilitating real-time data gathering and analysis. Anomaly detection in manufacturing processes has emerged as a vital strategy to ensure quality, optimize operations, and reduce costs. Anomaly detection is crucial across various domains, including fault and fraud detection, aiming to identify unusual patterns deviating from expected behaviors. By harnessing machine learning algorithms, anomalies are detected through the differentiation between normal and abnormal data patterns. This study primarily delves into machine learning methodologies for detecting anomalies in IoT settings and the associated challenges with existing datasets. We scrutinize prevalent machine learning techniques comprising of both supervised and unsupervised for IoT anomaly detection, along with their inherent challenges. Moreover, we propose and explore potential avenues for future research, particularly focusing on dataset-related considerations.