AIRCC's International Journal of Computer Science and Information Technology

Wernhuar Tarng ¹, Cheng-Kang Chou ¹, Kuo-Liang Ou ¹

Affiliations
1 Graduate Institute of Computer Science, National Hsinchu University of Education, TW

Abstract

A P2P botnet virus detection system based on data-mining algorithms is proposed in this study to detect the infected computers quickly using Bayes Classifier and Neural Network (NN) Classifier. The system can detect P2P botnet viruses in the early stage of infection and report to network managers to avoid further infection. The system adopts real-time flow identification techniques to detect traffic flows produced by P2P application programs and botnet viruses by comparing with the known flow patterns in the database. After trained by adjusting the system parameters using test samples, the experimental results show that the accuracy of Bayes Classifier is 95.78% and that of NN Classifier is 98.71% in detecting P2P botnet viruses and suspected flows to achieve the goal of infection control in a short time.

Keywords

Data Mining, Bayes Classifier, Neural Network, P2P Botnet, Virus Detection Systems.

Full Text