Traffic identification using artificial neural network [Internet traffic]

Author

Ali, Ali A. ; Tervo, R.

Author_Institution

Dept. of Electr. & Comput. Eng., New Brunswick Univ., Fredericton, NB, Canada

Volume

1

fYear

2001

fDate

2001

Firstpage

667

Abstract

The paper investigates the use of artificial neural networks (ANN) to unconventionally classify Internet traffic. Structurally and functionally, the classifier used is a feedforward multilayer layer perceptron (FFMLP) network trained using backpropagation. The inputs are random samples of bits from a bit stream (i.e. all the inputs are either 1 or 0). The data was collected and pre-processed, then used to train, test and evaluate the classifier. Despite the lower capability to identify certain data types, the algorithm has shown that it has very good features as a classifier. SMTP, TELNET, FTP, HTTP, IP TELEPHONY and UDP data types were used in the investigation

Keywords

Internet; backpropagation; feedforward neural nets; identification; multilayer perceptrons; telecommunication computing; telecommunication traffic; FTP; HTTP; IP TELEPHONY; Internet traffic identification; SMTP; TELNET; UDP data; artificial neural network; backpropagation; feedforward multilayer layer perceptron; random bit stream samples; Artificial neural networks; Biological neural networks; Central nervous system; Internet; Multi-layer neural network; Multilayer perceptrons; Neurons; Niobium; Protocols; Telecommunication traffic;

fLanguage

English

Publisher

ieee

Conference_Titel

Electrical and Computer Engineering, 2001. Canadian Conference on

Conference_Location

Toronto, Ont.

ISSN

0840-7789

Print_ISBN

0-7803-6715-4

Type

conf

DOI

10.1109/CCECE.2001.933764

Filename

933764