TSSG PhD student Runxin Wang recently presented a paper entitled ‘Training Traffic Classifiers with Arbitrary Packet Sets’, at the 1st IEEE Workshop on Traffic Identification and Classification for Advanced Network Services and Scenarios (TRICANS 2013) in Budapest, Hungary. The paper, co-authored by Dr Lei Shi and Dr Brendan Jennings addresses techniques for classifying traffic on the Internet, that is analysing data as it flows through the network in order to classify why type of application is generating that data. Internet Service Providers typically treat different types of traffic differently, for example prioritising voice and video communications over peer-to-peer traffic. Previous works have shown that analysing the first few packets in a flow is usually enough to correctly identify the application. However, some application developers have realised this and instead change the first few packets in order to emulate a different application type, thus defeating the classification algorithms. Runxin’s paper investigated the use of a random selection of packets at the start of a flow for identification purposes—this approach is shown to work with an acceptable degree of accuracy.