Peer-to-Peer (P2P) networks have been widely applied in file sharing, streaming media, instant messaging and other fields, which have attracted large attention. At the same time P2P networks traffic worsens the congestion of a network significantly. In order to better manage and control P2P traffic, it is important to identify P2P traffic accurately. In this paper we propose a novel P2P identification scheme, based on the host and flow behaviour characteristics of P2P traffic. First we determine if a host takes part in a P2P application by matching its behaviour with some predefined host level behaviour rules. Subsequently, we refine the identification by comparing the statistical features of each flow in the host with several flow feature profiles. The experiments on real world network data prove that this method is quite efficient to identify P2P traffic. The classification accuracy achieves 93.9 % and 96.3 % in terms of flows and bytes respectively.

SUMMARYTraffic classification has recently gained much attention in both academic and industrial research communities. Many machine learning methods have been proposed to tackle this problem and have shown good results. However, when applied to traffic with out-of-sequence packets, the accuracy of existing machine learning approaches decreases dramatically. We observe the main reason is that the out-of-sequence packets change the spatial representation of feature vectors, which means the property of linear mapping relation among features used in machine learning approaches cannot hold any more. To address this problem, this paper proposes an Improved Dynamic Time Warping (IDTW) method, which can align two feature vectors using non-linear alignment. Experimental results on two real traces show that IDTW achieves better classification accuracy in out-of-sequence traffic classification, in comparison to existing machine learning approaches.

Communities are an important feature of real-world networks that can reveal the structure and dynamic characteristics of networks. Accordingly, the accurate detection and analysis of the community structure in large-scale IP networks is highly beneficial for their optimization and security management. This paper addresses this issue by proposing a novel community detection method based on the similarity of communication behavior between IP nodes, which is determined by analyzing the communication relationships and frequency of interactions between the nodes in the network. On this basis, the nodes are iteratively added to the community with the highest similarity to form the final community division result. The results of experiments involving both complex public network datasets and real-world IP network datasets demonstrate that the proposed method provides superior community detection performance compared to that of four existing state-of-the-art community detection methods in terms of modularity and normalized mutual information indicators.

No abstract

No abstract