Date Added: Mar 2012
Internet traffic classification into specific network applications is essential for managing network resources and from security point of view. Traditional classification techniques such as port based and payload based are having significant limitations. Hence, newer statistical approach is adopted for classifying internet traffic into specific applications. This paper reviews modern statistical classification techniques which may or may not use Machine Learning approach, but still classify traffic with great accuracy. Network traffic classification is one of the major challenging task in past few years. The tasks of network engineers include meeting application performance, meeting bandwidth requirement of customers, managing bandwidth consumption, apply security rules, fault diagnosis, performing accurate accounting for billing, etc.