Survey on Intrusion Detection Systems: Techniques and Datasets

Abstract

Intrusion Detection System (IDS) plays a crucial role in the context of securing modern computer networks from malicious attacks. Traditional IDS techniques, such as signaturebased systems, are only effective for known attacks and fail to detect unknown or zero-day attacks. In recent years, the rising trend of the use of data-driven approaches, such as machine learning and deep learning, has been noticed in the area of intrusion detection. This paper presents a comprehensive survey of recent machine learning-based and deep learning-based intrusion detection techniques, along with their approaches, benefits, and limitations. Further classification is performed on the basis of specific algorithms, such as Random Forest, Support Vector Machine (SVM), XGBoost, Convolutional Neural Network (CNN), Recurrent Neural Network (RNN), and autoencoders. The surveyed research articles are analyzed in terms of feature extraction methods, accuracy of detection, false alarm rates, computational complexity, and ability to detect unknown attacks. Comparison analysis shows the current trends in the research and the challenges being faced in this field. This survey will assist researchers and practitioners in having a clear understanding of the IDS techniques in intrusion detection and will also discuss the future research directions in this area.