Implementation Web Deployment of an Anomaly Detection System for IoT Environments Using the CRIPST-ML Methodology

Abstract

The Internet of Things (IoT) has transformed modern industry through real-time monitoring and automation, but it has also increased cyber risk in edge implementation with limited resources. The article presents the implementation and web deployment of a lightweight anomaly detection system for IoT environments, structure using the cross-Industry Standard Process for Machine Learning with Quality models (Isolation Forest, Single Class SVM and K-Means), the BoT-Iot dataset was used, and operational behavior was validated with real-time packet capture. The models are evaluated with internal clustering metrics (Silhouette, Calinski-Harabasz, Davies-Bouldin) and attack-driven behavior, and the selected artifacts are implemented through a Streamlit interface for interactive inference and monitoring. The results show that the proposed pipeline is feasible and reproducible for small-scale IoT deployments, while providing accessible, user-oriented security analytics.