An efficient network intrusion detection model for IoT security using K-NN classifier and feature selection

• Published in: Multimedia tools and applications Vol. 82; no. 15; pp. 23615 - 23633
• Main Authors: Mohy-eddine, Mouaad, Guezzaz, Azidine, Benkirane, Said, Azrour, Mourade
• Format: Journal Article
• Language: English
• Published: New York Springer US 01.06.2023; Springer Nature B.V
• Subjects: Actuators; Classifiers; Computer Communication Networks; Computer Science; Cybersecurity; Data Structures and Information Theory; Datasets; False alarms; Feature selection; Genetic algorithms; Internet of Things; Intrusion detection systems; K-nearest neighbors algorithm; Multimedia Information Systems; Performance evaluation; Principal components analysis; Special Purpose and Application-Based Systems; Statistical tests; K-NN; Network security; Feature selection; Bot-IoT; Intrusion detection; IoT
• ISSN: 1380-7501; 1573-7721
• DOI: 10.1007/s11042-023-14795-2

The Internet of Things (IoT) interconnects billions of sensors and actuators to serve a meaningful purpose. However, it is always vulnerable to various menaces. Thus, IoT security represents a big concern in the research field. Various tools were developed to mitigate these security issues. So, Intrusion detection systems (IDS) have gained much attention in the research community due to their critical role in maintaining network security. In this work, we integrate a network IDS (NIDS) to enhance IoT security. This paper presents a network intrusion detection model for IoT environments using a K-Nearest Neighbors (K-NN) classifier and feature selection. We built the NIDS using the K-NN algorithm to improve the IDS accuracy (ACC) and detection rate (DR). Furthermore, the principal component analysis (PCA), univariate statistical test, and genetic algorithm (GA) are used for feature selection separately to improve the data quality and select the ten best performing features. The performance evaluation of our model is performed on the Bot-IoT dataset. After applying the feature selection, the models have shown promising results regarding ACC, DR, false alarm rate (FAR), and predicting time. Our proposed model provided 99.99% ACC and maintained its superior performance for the ten selected features. Furthermore, we calculated the prediction time, as we consider it critical in building IDS for IoT, and by applying feature selection, we reduced it significantly from 51,182.22 s to under a minute. This novel model presents many advantages and reliable performances compared with previous models relying on the same dataset.