A secure framework for the Internet of Things anomalies using machine learning

Abstract The Internet of Things (IoT) revolutionises modern technology, offering unprecedented opportunities for connectivity and automation. However, the increased adoption of IoT devices introduces substantial security vulnerabilities, necessitating effective anomaly detection frameworks. This Paper proposes a secure IoT anomaly detection framework by utilising four machine learning algorithms such as: Logistic Regression (LR), Linear Discriminant Analysis (LDA), Classification and Regression Trees (CART), and Gaussian Naive Bayes (GNB). By generating synthetic datasets with induced anomalies, the framework employs AWS IoT Core infrastructure and Python-based analysis to identify irregularities in device performance. The proposed framework achieved a high detection accuracy ranging from 91 to 98% across the tested algorithms, with CART showing the best performance. Key performance metrics, including precision, recall, and F1-score, confirmed the model's reliability in distinguishing between normal and anomalous IoT data. Experimental results demonstrate superior detection accuracy across all methods, validating the robustness of the proposed approach. This research offers a scalable solution for IoT security, paving the way for improved anomaly detection and mitigation strategies in connected environments. The integration of machine learning algorithms with IoT infrastructure allows for real-time monitoring and proactive anomaly detection in diverse IoT applications. The proposed framework enhances security measures and contributes to the overall reliability and efficiency of connected systems.