Abstract—Video traffic represents a large fraction of Internet traffic. For efficient service provisioning and controlling traffic based on various policies, measuring detailed user behaviors, such as when, at what rate, how much, and between whom video data are transmitted, is critical. To collect such information, this paper deals with video stream identification, since source or destination address in a packet header may not be reliable for identifying true source and destination. An unsupervised learning algorithm is proposed to perform stream identification. The algorithm overcomes two shortcomings of the existing clusterers. Experimental results show that the rate of correctly grouped classes achieved by the algorithm is 94%.
Index Terms—Stream identification, video traffic, decay rate, unsupervised learning
The authors are with the Department of Computer Science and Engineering, Fukuoka Institute of Technology, 3-30-1 Wajiro-Higashi, Higashi-ku, Fukuoka, 811-0295 Japan (e-mail: email@example.com).
Cite: Kazumasa Oida and Naoto Nakayama, "Video Stream Identification for Traffic Engineering," International Journal of Future Computer and Communication vol. 2, no. 4 pp. 275-280, 2013.