Abstract—With the huge amount of ubiquitous multimedia data transmitted in nowadays Internet, the use of packet sampling for traffic measurements has become widely employed for network operators. In this paper, we present an adaptive packet sampling technique from the classification perspective, the main sampling principle of which is to select as many packets with low occurrence rate as possible based on two useful features for multimedia traffic: Packet Size (PS) and Packet Inter Arrival Time (IAT). We build a model of the ideal packet sampling technique for classifying multimedia traffic, which adjusts adaptively the sampling probability of selecting packets according to PS and IAT predicted simultaneously by multi-output support vector regression, and define general indexes for evaluating the sampling performance of the proposed approach. We compare our approach with other sampling methods and evaluate their impact on the performance of traffic classification using two machine learning methods with real multimedia traffic data. The experimental results show that this approach has good sampling performance and is able to enhance the performance of the traffic classification methods.

Index Terms—Sampling, traffic classification, multimedia

Cite: Rongqiang Lin, Ou Li, Qing Li, and Kun Dai, "Exploiting Adaptive Packet-Sampling Measurements for Multimedia Traffic Classification," Journal of Communications, vol. 9, no. 12, pp. 971-979, 2014. Doi: 10.12720/jcm.9.12.971-979