Packet Analyzer With Multiple Artificial Neural Network Back-Propagation Training

Lim , Siew Yuen (2008) Packet Analyzer With Multiple Artificial Neural Network Back-Propagation Training. Project Report. UTeM, Melaka,Malaysia. (Submitted)

	PDF (24 Pages) Packet_Analyzer_With_Multiple_Artifical_Neural_Network_Back-_Propagation_Training_Lim_Siew_Yuen_QA76.87.L55_2008_-_24_Pages.pdf - Submitted Version Download (1MB)
	PDF (Full Text) Packet_Analyzer_With_Multiple_Artifical_Neural_Network_Back-_Propagation_Training_Lim_Siew_Yuen_QA76.87.L55_2008.pdf - Submitted Version Restricted to Registered users only Download (9MB)

Abstract

The packet analyzer with multiple artificial neural network back-propagation training is basically a system that integrated with the feed-forward back­ propagation neural network to analyze the packet from the network. This system is proposed to overcome the intrusion that expands widely nowadays. This system using a system solution, which is the output of the training and testing process using MATLAB, to integrated with the simple system developed 'USing Visual Basic 6.0. The system can automatically identified and categorizes the packet captured into two categories, that is either normal or misuse packet. Neural networks are widely considered as an effective approach to classifY patterns. In this project, the packet analyzer system will receive data packet from the network and the neural network will classify it into normal packet or misuse packet. For any occurrence of the possibility misuse packet, the system will alert the system administrator. Since neural network have a high learning rate, it will recognize most of the characteristics of attacks packet and by this, it can recognize, and block the packet before the misuse packet can get through the network. The advantage of using neural networks over statistics resides in having a simple way to express nonlinear relationships between variables, and in learning about relationships automatically. With few exceptions, behavior of most other users is also predictable. Neural networks are still a computationally intensive technique, and are not widely used in the packet analyzer community.

Actions (login required)

View Item