Classification Of Web Browsing And Video Services Based On Novel Feature Selection Algorithm

With different kinds of services based on the HTTP protocol such as web browsing, audio, video and the threats of anonymous traffic on the network being growing, traffic classification faces huge challenge. Meanwhile, with the development of multimedia communication, the appearance and fast spreading of various video applications(such as P2 P, IPTV, etc.) greatly facilitate the people's lives. However, different video applications with high bandwidth and low latency have different QoS requirements. Therefore, to identify flow accurately is crucial for effective network management and reasonable allocation of network resources. Nowadays, machine learning algorithm based on the statistical characteristics has been widely used in classification of traffic, but the emphasis and difficulty is how to discover the efficient and simple characteristics.This paper focus on six kinds of network multimedia applications: Skype audio, video streaming, network live TV, HTTP download, web browsing(text and images) and web browsing with video. We propose a new novel feature selection method which is based on coefficient of variation. Experimental results show that the proposed method can achieve higher accuracy of classification than existing methods. This paper also analyzes its computational complexity by comparison of feature selection method based on information gain and chi-square, which proved our method being more effective.The redundant features may be selected by the above method, so this paper proposes an improvement approach based on coefficient of variation by considering the correlation between characteristics. The experimental results show that this method can obtain simple and effective features.Moreover, from a QoS perspective, it's the first time for fine-grained classification of web browsing. Most of the literatures classify web browsing as one category for identification of web applications based on HTTP protocol. However, with the development of network applications, web that contains all kind of information becomes increasingly complicated. Therefore, web browsing being treated as a class is no longer appropriate. Fine-grained analysis of web browsing is feasible and necessary. In addition, the results demonstrated that it's reasonable for fine-grained classification of web browsing.