IPTV流量分类与特征优化技术与研究.pdfVIP

P2P IPTV IPTV IPTV IPTV IPTV KISS KISS KISS IPTV IPTV IPTVKISS i ABSTRACT With the continuous development of Internet technology, P2P based internet protocol television (IPTV) rapidly become popularity, which also accordingly brings many potential security risks, so it’s imperative to effectively monitor and manage the IPTV traffic. The accurate classification of a variety of IPTV traffic is the prerequisite for its effective supervision. At present, statistics-based IPTV traffic classification with extremely accuracy and high expansibility has become the mainstream of this domain. However, this method also has some shortcomings to be sloved, such as, overabundance of dependent statistical features, and not enough flexibility of classifier. In this paper, we first have a systematic study on the current most outstanding IPTV traffic classifier: KISS. After drawing a detailed theoretical analysis on KISS, we put forward some improvements on it through re-selecting its statistic packets and re-setting its parameters. Experimental results show that the improved KISS has got a faster classify speed and a greater ability to handle real-time network traffic. For the shortcomings of current IPTV traffic classifier, we propose an important features optimal screening algorithm, which first screens the important features from defined features before the classifier executes machine learning and automatic classification, so as to rese