A Hybrid Three-layered Approach for Intrusion Detection using Machine Learning Methods

kground and Objectives: Intrusion Detection Systems (IDS) are crucial for safeguarding computer networks. However, they face challenges such as detecting subtle intrusions and novel attack patterns. While signature-based and anomaly-based IDS have been widely used, hybrid approaches offer a promising solution by combining their strengths. This study aims to develop a robust hybrid IDS that effectively addresses these challenges.Methods: We propose a three-layered hybrid IDS that leverages machine learning techniques. The first layer utilizes a signature-based approach to identify known intrusions. The second layer employs an anomaly-based approach with unsupervised learning to detect unknown intrusions. The third layer utilizes supervised learning to classify intrusions based on training data. We evaluated the proposed system on the NSL-KDD dataset.Results: Experimental results demonstrate the effectiveness of our proposed hybrid IDS in accurately detecting intrusions. Comparisons with recent studies using the same dataset show that our system outperforms existing approaches in terms of detection accuracy and robustness.Conclusion: Our research presents a novel hybrid IDS that effectively addresses the limitations of traditional IDS methods. By combining signature-based, anomaly-based, and supervised learning techniques, our system can accurately detect both known and unknown intrusions. The promising results obtained from our experiments highlight the potential of this approach in enhancing network security.

کلیدواژه ها:

Intrusion Detection Systems ، Network Security ، Machine Learning ، NSL-KDD Data Set