Title: Intrusion detection system using statistical query tree with hierarchical clustering approach

Authors: P.V.N. Rajeswari; M. Shashi

Addresses: Department of CSE, PBR Visvodaya Institute of Technology and Science (PBR VITS), Kavali, Andhra Pradesh 524201, India ' Department of Computer Science and Systems Engineering, Andhra University, Visakhapatnam, Andhra Pradesh 530003, India

Abstract: The internet has become a major part of everyone's life. When no proper protection is provided, intruders misuse the access provided by the internet, leading to an increased risk of sensitive data leakage. To have a trade-off between scalability and precision, this research introduces a novel two-stage screening framework for intrusion detection systems (IDS) to identify the attacks and their types. The first stage aims to identify suspicious internet protocol (IP) addresses based on the abrupt deviation from the normal activity pattern. The second screening stage aims to analyse the packets received from suspicious IP addresses by applying a recently developed single-phase statistical hierarchical clustering (SHiC) algorithm designed for clustering and outlier detection. The data packets are classified as outliers based on their higher statistic distance to the existing components or clusters identified. The complete IDS framework is developed and applied to two benchmark datasets and compared with the results produced by several outlier detection algorithms. The proposed framework is found to be consistently more accurate in detecting attacks.

Keywords: statistical query tree; intrusion detection system; IDS; outlier; statistical hierarchical clustering; SHiC; cyber-attack; CICIDS-2017.

DOI: 10.1504/IJDMMM.2024.138822

International Journal of Data Mining, Modelling and Management, 2024 Vol.16 No.2, pp.176 - 195

Received: 28 Oct 2022
Accepted: 26 Jul 2023
Published online: 31 May 2024 *

Full-text access for editors Full-text access for subscribers Purchase this article Comment on this article