Intrusion Detection System for IoT Using Machine Learning

The exponential growth in the number of the interconnected devices and the rapid development in 5G networks has encouraged the evolution of progress and expansion of the Internet of Things (IoT) domain. Although, the convergence of IoT and 5G has brought numerous benefits and convenience to our daily lives, it also presented a new landscape of potential threats and security. The incoming threats has increased attack surface for cybercriminals that leads sometimes to catastrophic damages. It is urgently needed to recognize the potential threats it poses and take proactive measures to mitigate risks and ensure the security and privacy of both individuals and organizations. The massive generated data from IoT devices harden the traditional analysis techniques. Machine Learning-based techniques contributed significantly in developing Intrusion Detection Systems (IDSs) for monitoring IoT devices data. This paper proposes an enhanced Machine Learning-based model that enhanced IDS for IoT accuracy compared to other traditional methods. The proposed approach compared three different machine learning algorithms; Random Forest, Decision Tree and Gradient Boost using Bot-IoT dataset exploiting the Feature Engineering and Synthetic Minority Oversampling Technique (SMOTE) that contributed significantly in solving the imbalance problem as well as the feature selection process for high detection accuracy. The results from balanced dataset show great detection accuracy and F1-score.