DESIGN AND IMPLEMENTION OF HYBRID NETWORK INTRUSION DETECTION SYSTEM WITH SNORT AND DEEP NEURAL NETWORK FOR HOME AREA NETWORK

• Main Author: Ivan Thenu, Willard
• Format: Theses
• Language: Indonesia
• Online Access: https://digilib.itb.ac.id/gdl/view/47060
• Institution: Institut Teknologi Bandung

The speed of internet that continues to increase compared to its prices per bandwidth that continue to decrease, the internet become one of the main needs for public to seek information, entertainment and others. Furthermore, the installation of internet facilities in the home area or commonly called the Home Area Network (HAN) become very common. HAN itself has several security holes that need to be reviewed to prevents cyber attacks, such as the biggest DDoS attack in the history of small-scale networks. One solution to reviewing the security side of home network is to implement a Network Intrusion Detection System (NIDS). However, the implementation of NIDS at home network is very rarely done because of the cost and human resource needed to operate it. NIDS itself has two detection methods that are often used, namely Misuse Detection and Anomaly Detection. Each method has a weakness in detecting intrusions that occurred in the network. Therefore, in this thesis the author designs and implements a Network Intrusion Detection System (NIDS) to cover weaknesses in Misuse and Anomaly Detection methods. Artificial Intelligence with Deep Neural Network (DNN) algoritm used in anomaly detection has an accuracy rate of 94.45% with very low false positive rate in classifying anomalous NetFlow. In this research, Raspberry Pi is used to reduce production costs. In addition, Security information and Event Management (SIEM) and automated system are used to facilitate users in the NIDS operation process. An NIDS prototype was produced in this thesis and is expected to be implemented in Home Area Network (HAN) or home-scale network with the aim of increasing the security factor on the network.