A Network Security Application for Detecting DDOS Attacks Using Data Mining Approach

Abstract

Distributed denial of service (DDoS) attacks is difficult for individuals and organisations to cope with on a consistent basis. Keeping a service running at all times is the job of the security engineer. Unusual activity can be detected and classified using an intrusion detection system (IDS). Attack packets will contain more packets than normal, but the inter arrival rate will be too short for attackers to deplete resources quickly. To keep the service's confidentiality, integrity, and availability, such IDS have to be regularly rationalized over newest prowler outbreak. Due to a lack of common data sets for recent DDoS attacks across multiple network levels, a new dataset was created in this article (SIDDoS, HTTP Flood). Multilayer Perceptron (MLP), Naive Bayes, and Random Forest are three well-known classification methods used in this study. Also, a new dataset was created for this study. A 97.63% accuracy rate was attained by the MLP in the experiments.