An Automated Signature Generation Method for Zero-Day Polymorphic Worms Based on Multilayer Perceptron Model

Abstract

Polymorphic worms are considered as the most dangerous threats to the Internet security, and the danger lies in changing their payloads in every infection attempt to avoid the security systems. In this paper, we propose an accurate signature generation system for zero-day polymorphic worms. We have designed a novel Double-honeynet system, which is able to detect zero-day polymorphic worms that have not been seen before. To generate signatures for polymorphic worms we have two steps. The first step is the polymorphic worms sample collection which is done by the Double-honeynet system. The second step is the signature generation for the collected samples which is done by k-means clustering algorithm and a Multilayer Perceptron Model. The system collects different types of polymorphic worms, we used the k-means clustering algorithm to separate each type into a cluster. The Multilayer Perceptron Model is used to generate signatures for each cluster. The main goal for this system is to reduce the false positives and false negatives.