期刊名称：Journal of Theoretical and Applied Information Technology
印刷版ISSN：1992-8645
电子版ISSN：1817-3195
出版年度：2017
卷号：95
期号：15
页码：3684
出版社：Journal of Theoretical and Applied
摘要：Distributed Denial of Service attack (DDoS) is a structured network attack coming from various sources and fused to form a large packet stream. DDoS attacks aiming to disrupt the services available in the target tissue by flooding the target bandwidth or processing capacity of the system by making the target network server becomes overloaded. Network packet classification is one method of network defense system in the organization of the Internet in order to avoid DDoS attacks. Network packet classification can be carried out either by utilizing the method of Artificial Neural Network (ANN). The proposed work of network traffic packet classification applying variation of hidden layer with Quasi-Newton method training function and statistical network traffic packet feature extraction have the result that ANN with two hidden layers outperformed than ANN with single or three hidden layers. ANN with two hidden layers gives overall consistent mse and convergence speed, also higher correct classification percentage at 99.04%. Quasi-Newton method (trainlm) is qualified and suit for classification task based on value of regression both in the training and validation phase.
