HTTP Flood DOS Attack Detection on Big Data Using Data Mining

Abstract

Nowadays, increasing use of Internet connection, security becomes a huge challenge for individuals as well as governments and organizations. Therefore, in the last decade, the world is moving towards green computing in the purpose either to store energy or to decrease operational costs. So, this new technology uses web servers to provide web applications to end user. Generally, these web servers become unavailable because of HTTP flood DoS attack. This paper applied data mining techniques such as WEKA and TANAGRA tools for data preprocessing in the aim to get a consistent data in one hand, and for classifying the traffic in normal behavior or anomaly using like J48, Random tree, SMO, Naïve bayes, IBK, and combined classifier in the other hand. For this aim, a well-known big data set is used which called http_csic_2010_full dataset. Results show that the two data mining tools are well close with all the classifiers, just MLP in WEKA, and CVM in TANAGRA they didn’t perform well in terms of detection accuracy.