Detection of mobile applications leaking sensitive data

Abstract

Smart phones are the most popular and used devices by people. They host numerous types of data which can be classified as public and private. It is a new phenomenon to develop some malicious applications to exfiltrate personal data from smart phones. Hence personal data on these devices such as short messages, contacts, photos, videos, GPS locations, etc. require a specialized security mechanism which protects them from being leaked by malicious applications. The main problem is the disclosure of sensitive information when mobile applications try to access them via android permissions. This situation tells us that these mobile applications will probably leak the sensitive data. Therefore, some solutions especially considering data leakage on smart phones are required. This paper has two goals which are detecting the applications leak sensitive data and analyzing benign applications whether they have malicious intents or not. In this study, a sensitive data leakage prevention system for Android systems was proposed. J48 classification algorithm was used to detect mobile applications leaking sensitive data. In addition, trusted mobile applications downloaded from Google Play Store were evaluated whether they have any similarity with malicious mobile applications or not by clustering with K-Means clustering algorithm. The result has shown that the proposed system can detect malign and benign applications in 98.6% accuracy.