Abstract

Abstract: NetworkTraffic classification is the need of today’s emerging and rapidly growingcomputer network for network traffic analysis, network management, securitymonitoring, flow detection, QoS and lawful interception. It is possible toapply machine learning techniques to classify traffic based on flow statisticalfeature. Supervised and unsupervised classification algorithms have beenapplied to classify traffic. Conventional methods for classification includeport based prediction and payload based deep inspection methods. In recentnetwork environment the usual methods undergoes from some troubles like dynamicports and encrypted applications. The nearest neighbor (NN) method havingadvantages, such as no need of training procedure, no risk of over fitting ofparameters, and able to handle a large number of classes. But, the performanceof NN classifier method affected if the size of training data is small. Thispaper conducts a survey on the various network traffic classificationtechniques, also focuses on a new non parametric traffic classification method,which makes use of correlation information for the purpose of classification.

Keywords: Trafficclassification, network operations, security, nearest neighbor