Mean-Shift and Local Outlier Factor-Based Ensemble Machine Learning Approach for Anomaly Detection in IoT Devices

IoT devices are rapidly being used in everyday life. However, many of these devices are susceptible as a result of insecure design, implementation, and setup. As a consequence, many networks already include susceptible IoT devices that are simple to infiltrate. This paper presents a Mean-Shift and Local Outlier Factor (LOF) based ensemble machine learning (ML) approach for anomaly detection in IoT devices. The UNSW-NB15 IoT-based network dataset is used for the evaluation of the proposed model. LOF is used with the Mean-Shift clustering approach for clustering and the ensemble of ML technique is used for the classification of usual or unusual activities. The proposed model is tested and evaluated with different performance measures like Precision (P), Recall (R), Accuracy (ACC), and F1-score. This research study has comprehensively evaluated the proposed model and compared it with different ML methods. The proposed model reported the best results for the detection of anomalies.