A Decision Tree Clubbed Nearest Neighboring Classification Approach for Detecting Intrusion in IoT Networks
Navjot Kaur, Gauri Mathur, Navneet Malik, Suruchi Talwani, Jimmy Singla · 2023
The influence of the Internet of Things on large industrial systems and our daily lives is growing dramatically. Sadly, this caught the eye of hackers, who turned IoT into a target of nefarious activity, potentially paving the way to an attack on the end devices. To prevent these hackers from launching attacks, numerous Intrusion Detection Systems (IDS) have already been created by various researchers. However, these models undergo poor generalization abilities and a low accuracy rate, which hinder intrusion detection abilities in realtime scenarios. Therefore, tis research study has presented an effective and efficient IDS that is based on a Decision Tree (DT) and K-nearest neighbor (KNN). The primary objective of the suggested approach is to decrease False Alarm Rates (FAR) while raising the accuracy of attack detection rates. To do so, we used two standard datasets i.e., KDD99 and NSLKDD in our work, upon which pre-processing technique is implemented. After this, a hybrid feature selection strategy is employed to choose the most crucial features in the model as well as resolve dataset dimensionality issues. Finally, hybrid classifiers (DT+KNN) are used for identifying the attacks and categorizing them. Results indicated that the proposed DT+KNN model is able to achieve an accuracy of 99.983 and 100% on the KDD99 and NSLKDD datasets. Also, the FAR was 0% and 0.0333% in the KDD99 and NSLKDD datasets, respectively.