M.A. Thesis:

Malware Detection using Step by Step Ensemble Classification

Android malware is growing at a fast pace and has become a serious threat to smartphone users who use the Android platform. They are progressing rapidly and endangering the privacy of users and their confidential information. Hence, we need effective and efficient techniques for identifying Android malware. This thesis proposes a strategy to detect Android malware using Android applications network traffic. This approach consists of two steps. The first step is to select the most appropriate traffic features that have the most impact in categorizing traffic into two types of malware and benign. For this purpose, we use a chi-square algorithm to select the effective features and select the characteristics that matter most in the categorization. In the second step, we present a new step-by-step model for detecting Android malware using Ensemble Classification. Then the selected features of the model were evaluated and compared with similar models. The results showed that these selective features can provide a model with proper performance. The model was also evaluated at four levels of security, and the results showed that our model could increase malware detection accuracy to over 95% and reduce the False Positive rate to less than 0.03%.

Keywords: Malware Detection, Machine Learning, Network Traffic, Ensemble Classification