Abstract: Internet traffic classification is one of the key foundations for research works and traffic engineering in Internet. The categories of Internet applications and the number of Internet flows increase fast last years. The technique challenges arc coped with development of traffic classification all the time. A research was carried out systematically in Internet traffic classification using machine learning in the paper. A mathematical description was given for the technology of traffic classification first. After surveying the technology of traffic classification based on supervised learning and unsupervised learning, we commented on the data preprocessing, classification model and performance evaluation etc. three aspects, and indicated the shortages now in these studies. Then, four common issues were summarized, which are data skew, label bottleneck, attribute changing and classify timely respectively. At the end, the prospect and our going work in this area were pointed out.

Key words: Traffic classification, Machine learning, Traffic flow, Network measurements