Enhanced Recursive Feature Elimination for IoT Intrusion Detection Systems

Due to the rapidly growing reputation and utilization of IoT devices, The threat to our privacy and security is increasing exponentially, since it is a convenient platform for cyber-attacks. in 2017 the number of attacks utilizing/against IoT devices increased by 600%. As there are many areas of security that utilize machine learning (ML) technology. Based on previous research, using ML to detect intrusions appears to be realistic and workable. While several studies have been conducted using ML for intrusion detection, they are either not highly accurate, only compatible with specific types of attacks or devices, or require a long duration to classify the attack. In this paper, we present an Intrusion Detection system that utilizes a Recursive Feature Elimination wrapper with the Decision Tree classifier to remove the redundant features. Experimental results on the Bot-IoT and NSL-KDD data-set minimized the number of features to only 16 and 15 respectively while maintaining high accuracy, precision, recall, and f1-score in comparison to previous works.