UDP flow detection and analysis using dynamical timeout strategy based on sliding-window

With the development of network, the bandwidth is increasing enormously, more and more new applications such as audio, video, online games and p2p network occupied a great deal of traffic. And most of these applications use UDP as transport layer protocol, which directly increase the UDP traffic. In order to analyze the characteristic of UDP using applications or measure the UDP traffic, we need to detect and analyze the UDP flow. But traditional traffic detection and analyses is focus on the TCP protocol, the most of the flow measurement research is based on the elephant flow detection or sampled flow statistics. To solve these problems, this paper presents a Sliding-Window-based Dynamical Timeout Strategy (SWDT) for the UDP flows detection. With this approach we can deal all the UDP streams, and detect the UDP flows efficiently and accurately. Our approach can be used in both traffic analysis and network measurement, detecting UDP stream, to find out the UDP flows and maintain their information.