Improving Discrimination Accuracy Rate Of DDoS Attacks and Flash Events

Abstract

Millions of people across the world are using internet for their day to day activities. People are highly dependent on internet as they are using internet resources for their work in every field. It connects billions of people across the world. Internet Security has become a big issue and with passage of time. Among many threats, the Distributed Denial-of-Service (DDoS) attack is the most frequent threats in the networks. Consequences of these attacks are more powerful when launched during flash events which are legitimate traffic and cause denial of service. This paper focuses on improving discrimination accuracy rate of DDoS Attacks and Flash events. Random forest is used for classification. Symmetric uncertainty is used for feature selection. NSL KDD data set used to evaluate performance of classifier. Weka is used for implementing algorithms.