期刊名称：International Journal of Advanced Research in Computer Engineering & Technology (IJARCET)
印刷版ISSN：2278-1323
出版年度：2016
卷号：5
期号：3
页码：0859-0862
出版社：Shri Pannalal Research Institute of Technolgy
摘要：The network consists of lots of heterogeneous data flowing through it. The network traffic should be classified for various networking as well as security issues, for instance network measuring, network monitoring or detection of some inobvious activity. For the identification of application protocols, the network traces are worked on to get the protocol message format specifications. This is important because currently most application protocols are proprietary on the internet that means they have no well documented protocol specifications. The proposed system implements a network trace based application protocol identifying system which gives the semantic information as result from protocol message formats with no previous knowledge of protocol specifications. In the proposed system, we find the statistical protocol message formats first by clustering n-grams with the relative semantics, and then classify the raw traces using the statistical formats. The system hopefully is capable in accurately and efficiently identifying the network traces of the target protocol. The system would hopefully work efficiently by the application of a Collapsed Variational Bayesian Inference algorithm for Latent Dirichlet Allocation as the application of Gibbs sampling was too time consuming due to its iterative nature in prior related work.
