REMOTE OPERATING SYSTEM IDENTIFICATION USING ARTIFICIAL NEURAL NETWORKS

Abstract

The need to determine the identity of an Operating System is valuable in areas such as network security, internet modeling, and end to end application design. Various techniques have been used to solve the problem of remote operating system identification such as rule-based tools (e.g. network mapper). However, these techniques fail to detect the operating system with high accuracy. In this study, multilayer perceptron model that utilizes back propagation neural network based classifier was developed for accurately fingerprinting operating system of remote host. This network was simulated using Neural Designer simulator. The result of this study was compared with Network mapper and machine learning technique using accuracy. The result comparison proved that Back Propagation neural network based classifier is far more accurate than rule-based tools on packet traces for fingerprinting.