作者投稿和查稿 主编审稿 专家审稿 编委审稿 远程编辑

计算机工程 ›› 2018, Vol. 44 ›› Issue (7): 150-155. doi: 10.19678/j.issn.1000-3428.0048008

• 安全技术 • 上一篇    下一篇

基于上下文信息的Android恶意行为检测方法

卢正军,方勇,刘亮,张文杰,左政   

  1. 四川大学 电子信息学院,成都 610065
  • 收稿日期:2017-07-19 出版日期:2018-07-15 发布日期:2018-07-15
  • 作者简介:卢正军(1991—),男,硕士研究生,主研方向为Android安全、恶意代码检测;方勇,教授;刘亮,讲师、博士;张文杰,硕士研究生;左政,博士研究生。

Android Malicious Behavior Detection Method Based on Context Information

LU Zhengjun,FANG Yong,LIU Liang,ZHANG Wenjie,ZUO Zheng   

  1. College of Electronic Information,Sichuan University,Chengdu 610065,China
  • Received:2017-07-19 Online:2018-07-15 Published:2018-07-15

摘要:

针对现有Android恶意软件检测方法存在的局限性和常见Android恶意软件的特点,提出一种基于上下文信息的Android恶意行为检测方法。从方法调用图中提取敏感应用程序编程接口,分析其行为的激活事件和条件因子,生成能够有效描述恶意软件行为的语境特征。在此基 础上,通过对比正常应用程序和恶意软件的特征来判断其是否为恶意行为。对266个Android恶意应用样本进行实验,结果表明,该检测方法的精确率为92.86%,召回率为95.21%。

关键词: 恶意行为, 权限, 激活事件, 上下文信息, 静态检测

Abstract:

In view of the shortcomings of existing Android malware detection methods and the characteristics of common Android malware,a Android malicious behavior detection method based on context information is proposed.The sensitive Application Programming Interface (API) is extracted from the method call graph,the activate events and the conditional factor of its behavior are analyzed,and then the contextual features that can effectively describe the malware behavior are generated.On this basis,the malicious behavior is judged by comparing the normal application and the malware features.Experimental results on 266 Android malicious application samples show that the accuracy rate of the detection method is 92.86%and the recall rate is 95.21%.

Key words: malicious behavior, permission, activate events, context information, static detection

中图分类号: