基于测量的UDP流特性分析

Measurement-Based Characterization of UDP Flowing

随着互联网的快速发展,网络测量分析成为认识和评估网络发展的重要手段之一.本文首先设计并搭建校内的数据采集平台;结合美国骨干网数据集,着重分析比较UDP流量基本特性,包括端口号、包长、字节等.分析结果表明UDP流更倾向于使用随机端口,且流数量有大幅度的增加,尤其是携带少量数据包和字节数的短流.我们推断这些现象和近年来P2P广泛应用和P2P协议的特性演变相关.

Internets continual and rapid evolvement in scope and its complexity impede our understanding on it, especially with the recent rising of P2P file-sharing and video streaming applications, we evaluate the amount of UDP traffic, in terms of flows, packets, and bytes. According to the traces collected from the US backbone links and our campus, we find that the UDP usage become more popular, especially in terms of flows. Our analysis infers that UDP flows tend to use high ports randomly, carrying few packets with little payload. That is consistent to the increasing usage of P2P applications.