IOTs Traffics Detection and Analysis Using Machine Learning for Cybersecurity Application
Abdallah M M Altrad · 2023
The data generation and transmission over the internet daily by users’ applications are increasing tremendously. Also, the amounts of traffic data transmission produced by IoTs over different networks are massive. The normal networks of companies, campuses, and others connect standard devices such as servers, routers, and switches. However, smart networking and the Internet of Things (IoTs) communication connect devices such as cameras, smart lighting systems, Alexa, and outlets. These devices communicate over networks via protocols. However, they are not robust in their software development due to security reasons, digital vulnerabilities and threats exist and open doors for hackers. Thus, the feature extract technique was applied to detect and analyze IoT’s benign and attack traffic features from a recent and large-scale dataset called CICIoT2023. The dataset contains a diversity of traffic data types generated by the IoT lab-connected 105 smart devices. Finally, the selected features were applied to machine learning algorithms to understand the IoT’s traffic behaviors for security applications and analytics. The applied algorithms showed high performance. The highest F score results were 0.979 for the decision tree, 0.973 for KNN, 0.704 for Naive Bayes, 0.939 for Random Forest, and 0.902 for MLP.