| With the rapid development of mobile Internet, smart phones are increasingly infiltrated all aspects of people’s work and life. Due to the increasing users for smart phones dependence, more and more users private data stored in the handsets. Because the user’s private information related to personal interests, thus causing a large number of the attacker’s attention, with the privacy leakage of malware is also increasing. Android mobile phone as one of the two major operating systems of the current smart mobile terminal, the potential security threats have gradually increased. Facing the growing problem of Android privacy leakage, the researchers pay more and more attention on it. And, at the present stage, the combination of the malicious software and the mobile payment has become more and more closely,which makes the problem of privacy leakage become the focus of the current research.In summary,this stage needs to strengthen the research of the Android platform privacy leak detection technology.For the issue of privacy leakage, the researchers developed more and more sophisticated static or dynamic analysis tool to detect whether the privacy theft, privacy leaks and other issues exist. However, these tools rely on manual configure the list of sources (the API which access to user privacy) and sinks (the API which send data from the mobile phone). Such lists are too difficult to implement and maintain. Especially in the update of Android version, each time there are lots of API are created, moreover, some developers will use sources and sinks that little people know. Therefore, this paper presents a method which based on machine learning to identify of source and sink from Android API. In order to provide more detailed information, the method of this paper is focus on classification of the sources and sinks.Based on the API in Android 4.2 version, the method proposed in this paper has the characteristics of fast training speed and high recognition rate. And the method can be used to classify the newly released Android API.For the problem of privacy leakage detection, this paper proposed a method based on privacy leakage path to detect whether the privacy leak exist in the malicious software. This method is beginning with smali code which is obtained by reverse analysis. According to its function call graph and the accessibility of sources and sinks, to detect the presence of Android malicious software privacy breach path, so as to determine the existence of privacy leakage problem. In this paper, according to the specific problems of privacy leakage detection, the time complexity of Warshall algorithm in the detection of privacy leakage is optimized, which makes the calculation of the accessibility of sources and sinks faster.Through experiments, in case of large proportion in sensitive function, or in the case that the function call relation matrix is sparse, the method has better performance and higher detection rate. |
