PCA算法在P2P加密流量识别中的研究与应用

传统的应用层协议识别方法均从改进匹配算法的角度来提高识别率,但是随着P2P协议的发展,其特征呈现多维化的趋势,算法复杂度也随之提高。鉴于此,在对P2P流量的多维特征进行分析并提取后,采用主成分分析(PCA)算法将提取到的特征降维处理,并通过实验证明了该方法在网络流量识别上的可行性和有效性。

Research and application of identifying encrypted P2P traffic with PCA

The bandwidth of Internet Service Provider can be fully utilized by popular P2P broadband tools such as PPLive, but in order to escape from supervision, dynamic ports and encrypted communications data are idiomatically employed by them, in the event, a lot of bandwidth is occupied by encrypted flow. At present, improving matching algorithm is commonly used by traditional application layer protocol identification methods to raise identification rate, but with the development of P2P protocol, features are multi-dimensional, and algorithms are more complex too, in view of this, with the pretreatment by analyzing and extracting multi-dimensional features of P2P traffic, PCA (Principal Component Analysis) algorithm is adopted to reduce the dimensions of features, and this strategy is approved feasible and effective through experiments.