A Recommendation System for Optimizing Security Protocols in Resource-Constrained IoT Devices

Source abstract

The Internet of Things (IoT) offers benefits in smart homes, healthcare, and industry but introduces security challenges for devices with limited processing power, memory, and energy. This study proposes a framework combining anomaly detection with adaptive security protocol recommendations to balance protection and resource use. Data preprocessing includes cleaning, normalization, outlier detection, and the Synthetic Minority Oversampling Technique (SMOTE) for class balancing. A Random Forest classifier identifies normal and anomalous device behavior, while a feedback-based recommendation system selects security protocols according to observed behavior and available device resources. The framework was evaluated using the Smart LED Light dataset and achieved 97.52% accuracy, 97.48% precision, 97.56% recall, and a 97.52% F1-score. These results indicate effective anomaly detection and support the proposed integration of classification with resource-aware security protocol recommendations for constrained IoT environments.