Machine Learning-Based IoT Device Fingerprinting for Enhanced Network Intrusion Detection

Abstract

Even though the IoT devices have brought about improvements in the networks, they have also caused new challenges in detecting the threats at the network level. Because IoT devices are so far and wide, traditional intrusion detection systems are not likely to identify most of their threats. An approach to IoT devices fingerprinting based on machine learning is proposed here to make network-level intrusion detection more accurate and efficient. The system identifies unique characteristics of devices in network traffic which It uses to categorize IoT devices and identify nefarious actions. This occurs by analyzing characteristics such as the communication method, types of traffic it carries and the supported protocols. traffic is classified into various categories by using Random Forest (RF), Support Vector Machine (SVM) and k-Nearest Neighbors (k-NN) algorithms. The evaluation of the performances reveals that the system suggested in this project improves the correct detection, reduces the occurrence of false alarms and improves the overall performance of IDS used in IoT networks. Applying device fingerprinting and machine learning helps to establish a strong protection in the IoT ecosystem, according to the research.