篇名 | A Detection System of Android Malware Based on SVM Algorithm |
---|---|
卷期 | 30:4 |
作者 | Lian-Fen Huang 、 Chao-Lin Ye 、 Chao Feng 、 Han-Bo Li 、 Ying-Min Zhang |
頁次 | 151-158 |
關鍵字 | Android 、 Malware 、 SVM 、 web interface 、 EI 、 MEDLINE 、 Scopus |
出刊日期 | 201908 |
DOI | 10.3966/199115992019083004014 |
In this paper, we propose a new detection system of android malware, a lightweight system combining static detection and Support Vector Machine algorithm (SVM). This system adopts static analysis to gather features of an application directly. These features are mapped to a vector space. An app corresponding a vector in the vector space, the SVM algorithm use these vectors to train a model which can indicate whether an android app has suspicious behaviors. To evaluate this system, we use a train set of 4500 benign applications and 6500 malicious applications to train model and acquire a True Positive Rate (TPR) of 97.96% and False Positive Rate (FPR) of 0.84% with a test set of 1371 benign applications and 829 malicious applications. Compared to traditional method for detecting android malware, this system can obviously improve the detection accuracy of malicious app and reduce the analysis time, while avoiding the high complexity of dynamic analysis. Moreover, this system provides a friendly web interface, allowing users to upload the app file and receive the analysis report.