Detection of Coercive Parsing Attack in XML Requests using Machine Learning Techniques

V. Punitha ⁺ and C. Mala

National Institute of Technology, Tiruchirappalli, India
vpunitha21@gmail.com, mala@nitt.edu

Abstract

The enriched cloud technology enables wider usage of web services in commercial applications. The increased bandwidth facilitates the continuous availability of web services. Legitimate access to these services is intentionally blocked by denial of service attacks. The impact of distributed denial of service attacks varies from interrupting the convenience of using the services to primary failure at cloud servers. Besides, the attacks are targeted towards application layer protocols recently. In order to eradicate this type of victim, this paper proposes two classification models based on machine learning techniques. The XML requests are pruned and the features that discriminate the coercive parsing attack are computed. The proposed SVM based classifier and the classification model with unsupervised learning categorize the attacks using constructed features. The simulated results emphasize that the detection rate of the proposed SVM based classifier, is significantly higher than the proposed unsupervised classifier..

Keywords: Distributed Denial of Service attack, Application layer attack, Coercive parsing attack,
Machine learning techniques

+: Corresponding author: V. Punitha
Research Scholar, Department of Computer Science and Engineering,
National Institute of Technology, Tiruchirappalli, India

IT Convergence Practice (INPRA), Vol. 6, No. 3, pp. 9-17, September 2018 [pdf]