Real-time Malware URL detection system based on Multi-layer Perceptron Neural Networks

Abstract

With the rapid proliferation of Internet technologies, mobile devices, and web applications, attackers now use the Web as a vector for introducing malware into enterprise networks. This change in attack vector has forced many organizations to subscribe to blacklisting services of malware Uniform Resource Locators (URLs) which are provided by a range of techniques including manual submission of suspected malware URLs and honeypots. However, the blacklist approach to detect malware URLs is no longer sufficient as many new malware URLs are not blacklisted immediately they are launched on the Internet. To address this problem, there is a need for real-time malware URL detection system that will be able to detect malware URL on the fly. The few previous studies that addressed this problem was unable to achieved high level of accuracy and none of the study has proposed Artificial Neural Network (ANN) based on detection system. Consequently, this study proposed a real-time detection system that is based on Multi-Layer Perceptron (MLP) Neural Networks. With the application of WEKA data mining tool, the performance of the proposed detection system was tested using an existing dataset comprises of malware URLs and benign URLs. The proposed detection system in this paper outperformed previous studies with accuracy of 97.33%.