Botnet Attacks Detection in Internet of Medical Things Network

With the rapid applications of the Internet of Things (IoT) in the areas of Healthcare, the threats to IoT in healthcare is increasing significantly. Attackers always try to compromise the less secured devices connected to IoT and form a large botnet of compromised devices. The IoT botnet-based attacks are very severe and cause significant impacts on healthcare applications. If the attacker manipulates medical sensor readings and generates false data, it might cause havoc during diagnosis. So, it is an utmost security concern to protect IoT networks from vulnerabilities and attacks. This paper analyzes the effectiveness of Machine Learning (ML) and Deep Learning (DL) methods on the CICIoMT2024 IoT healthcare dataset. We undertake an experimental evaluation of this dataset, employing a spectrum of ML and DL methods. Evaluation of these models encompasses a range of performance metrics such as Accuracy, Precision, Recall, F1-score, Area Under the Curve, Matthews Correlation Coefficient, and Kappa score, thereby facilitating a thorough understanding of their efficacy in detecting intrusions within IoT networks. Our findings reveals that advaced ML models such as XGBoost and RF outperforms DL models with an accuracy of 99.85% and 99.87% respectively. This paper reveals valuable insights and will help future research in the same domain.