Traffic identification using Bayes' classifier

Due to the exponential increase in Internet traffic, there has been a demand for increasing quality of service (QoS) offered by servers, routers and client machines whereby some types of traffic will be given priority. This can only be achieved by quickly identifying the type of traffic passing. In this paper, a new way of identifying type of traffic using a Bayes' classifier is investigated. The probabilities of different patterns in the data stream for every type of data were found with the help of pre-defined lookup tables containing corresponding byte values and packet size probabilities. Results show the capabilities of Bayes' classifier to identify different types of traffic.