A LSTM-FCNN based multi-class intrusion detection using scalable framework. (April 2022)

• Record Type: Journal Article
• Authors: Sahu, Santosh Kumar; Mohapatra, Durga Prasad; Rout, Jitendra Kumar; Sahoo, Kshira Sagar; Pham, Quoc-Viet; Dao, Nhu-Ngoc
• Is Part Of: Computers & electrical engineering. Volume 99(2022)
• Publication Date: 2022-04
• Subjects: LSTM -- FCN -- Deep learning -- Intrusion detection -- Multi-class classification -- Scalable framework -- Deep Neural Network; Computer engineering -- Periodicals; Electrical engineering -- Periodicals; Electrical engineering -- Data processing -- Periodicals; Ordinateurs -- Conception et construction -- Périodiques; Électrotechnique -- Périodiques; Électrotechnique -- Informatique -- Périodiques; Computer engineering; Electrical engineering; Electrical engineering -- Data processing; Periodicals; Electronic journals; 621.302854
• DOI: 10.1016/j.compeleceng.2022.107720
• Languages: English
• ISSNs: 0045-7906

Abstract

Abstract: Machine learning methods are widely used to implement intrusion detection models for detecting and classifying intrusions in a network or a system. However, many challenges arise since hackers continuously change the attacking patterns by discovering new system vulnerabilities. The degree of malicious attempts increases rapidly; as a result, conventional approaches fail to process voluminous data. So, a sophisticated detection approach with scalable solutions is required to tackle the problem. A deep learning model is proposed to address the intrusion classification problem effectively. The LSTM (Long Short-Term Memory) and FCN (Fully Connected Network) deep learning approaches classify the benign and malicious connections on intrusion datasets. The objective is to classify multi-class attack patterns more accurately. The proposed deep learning model provides a better classification result in two-class and five-class problems. It achieves an accuracy of 98.52%, 98.94%, 99.03%, 99.36%, 100%, and 99.64% using KDDCup99, NSLKDD, GureKDD, KDDCorrected, Kyoto, NITRIDS dataset respectively. Graphical abstract: Highlights: Conventional methods fail to process vast data resulted out of increased malicious attempts. An effort was made to provide a detection approach that can tackle scalability. Proposed deep learning addresses the intrusion classification problem effectively. The proposed LSTM-FCN model performs well on imbalanced datasets for multiclass problems. Attain improved accuracy on various datasets such as NSLKDD, GureKDD, Kyoto, etc.