DESIGN AND IMPLEMENTATION OF INTRUSION DETECTION SYSTEM USING NEURAL NETWORK

• Main Author: DWI PRASETYO (NIM 23206067); Pembimbing: Dr. Ir. Hendrawan, KURNIAWAN
• Format: Theses
• Language: Indonesia
• Online Access: https://digilib.itb.ac.id/gdl/view/16255
• Institution: Institut Teknologi Bandung

With the rapid expansion of computer network during the past years, security has become a crucial issue for computer system especially for intrusion and computer network attack. Some methode and product have been proposed in the recent years for the development of Intrusion Detection System. <br /> <br /> One of the very well known IDS software is an opensource IDS system known as Snort. Although this software have been widely used as a solution in network security problem, still there are so many challenges to be faced of. <br /> <br /> Recently, most of IDS software using some signature to detect intrusion in the network. This methode was great but still not smart enough to detect some new intrusion types. To solve the problem, some soft computing technics have been developed as solution. <br /> <br /> One of the soft computing technique is neural network, which is the most common soft computing technique for IDS. <br /> <br /> This thesis aims to implement the feed-forward back propagation neural network , one of the supervised neural network technicque to detect anomaly or misuse behaviour on the network. This research also compare the system performance as the function of the neural network parameters like the number of hidden layer and its neuron, the value of the goal, and the usage of different training algorithm. <br /> <br /> The result shows that the designed system is capable of classifying record up to 98.75% for training data, 84.15% for validation data and 83% for test data.