Trimming Approach of Robust Clustering for Smartphone Behavioral Analysis

Nowadays, smartphones get increasingly popular which also attracted hackers. With the increasing capabilities of such phones, more and more malicious softwares targeting these devices have been developed. Malwares can seriously damage an infected device within seconds. In this paper, we propose to use the trimming approaches for automatic clustering (trimmed k-means, Tclust) of smartphone's applications. They aim to identify homogenous groups of applications exhibiting similar behavior and allow to handle a proportion of contaminating data to guarantee the robustness of clustering. Then, a clustering-based detection technique is applied to compute an anomaly score for each application, leading to discover the most dangerous among them. Initial experiments results prove the efficiency and the accuracy of the used clustering methods in detecting abnormal smartphones applications and that with a low false alerts rate.