• Version
• Download 25
• File Size 376.80 KB
• File Count 1
• Create Date 6 March 2026
• Last Updated 6 March 2026

A Survey Paper on a Double Layered Approach for detecting network intrusions

Balasani Avinash1, Garisapati Sai Uday2, Banoth Arun 3, Ramisetti Vallabharayudu4
1Assistant Professor of Department of CSE (AI & ML) of ACE Engineering College, India.
2,3,4 Students of Department CSE (AI & ML) Of ACE Engineering College, India.

ABSTRACT
This project presents a Double-Layered Hybrid Approach (DLHA) for Network Intrusion Detection. The system combines a signature-based method with machine learning techniques to accurately identify both known attacks and rare anomalies. During the project, we applied Principal Component Analysis (PCA) to extract the most significant features from the NSL-KDD dataset, capturing the common characteristics of different attack categories. In the first layer, a Naive Bayes classifier detects Denial-of-Service (DoS) and Probe attacks. In the second layer, a Support Vector Machine (SVM) separates rare attacks from normal traffic. The approach was trained, tested, and evaluated to measure detection rate, accuracy, and false positive reduction, showing improved performance compared to single
classifiers.
Keywords: Network Intrusion Detection, Hybrid Approach, Machine Learning, Principal Component Analysis (PCA), Naive Bayes, Support Vector Machine (SVM), NSL-KDD Dataset.