An Efficient Cyber Assault Detection System using Feature Optimization for IoT-based Cyberspace

• Published in: Procedia computer science Vol. 235; pp. 757 - 766
• Main Authors: Dey, Arun Kumar, Gupta, Govind P., Sahu, Satya Prakash
• Format: Journal Article
• Language: English
• Published: Elsevier B.V 2024
• Subjects: Cyber Assaults; Ensemble learning; Feature Selection; Internet of things; Random Forest; Feature Selection; Internet of things; Ensemble learning; Cyber Assaults; Random Forest
• ISSN: 1877-0509; 1877-0509
• DOI: 10.1016/j.procs.2024.04.072

With the exponential growth in connected smart devices that interchange sensitive, crucial, and personal data over the Internet of Things (IoT)-based cyberspace, IoT becomes undefended to cyber assaults. Intrusion Detection System (IDS) is a vital segment of security tool to detect cyber assaults. However, IDS produce high False Positive Rate (FPR) because IDS improves accuracy at the same time as it increases FPR. Therefore, a more effective cyber assaults detection system is needed. Based on the aforementioned issues, this paper proposes an approach to recognizing cyber assaults in the IoT environment. This approach first performs various data pre-processing steps. Next, important features are optimized using modified Binary Pigeon Inspired Optimizer. After that, optimized feature subset is fed separately to the monolithic classifiers such as k-Nearest Neighbours (kNN), Support Vector Machine (SVM), Decision Tree (DT) and bagging-based ensemble model such as Random Forest (RF) for detection of cyber assaults. In this study, UNSW-NB15 is used to test the success of the system, and related existing methods are compared with these results. Based on the results of the experiment, the proposed system with RF is the most effective in terms of accuracy (99.41%), F1-score (99.33%), detection rate (99.09%), and precision (99.92%) with a minimum FPR (0.03%) for a subset of optimized features (4 independent features out of 42).