基于最小距离分类器的Android恶意软件检测方案  

Android malware detection based on minimum-distance classifier

作　　者：何文才[1,2] 闫翔宇[1] 刘培鹤[2] 刘畅[1] 

机构地区：[1]西安电子科技大学通信工程学院,西安710071 [2]北京电子科技学院通信工程系,北京100070

出　　处：《计算机应用研究》2017年第7期2184-2188,共5页Application Research of Computers

摘　　要：针对Android手机恶意软件数目增多,应用商店对大批量软件的安全性检测难度增大的问题,提出了一种轻量级恶意软件检测方案。该方案首先分析了大量恶意软件和正常软件样本的权限信息,再对权限频率特征去冗余,最后利用最小距离分类器进行软件分类。实验结果表明该方案的可行性,通过与同级别方案对比,在方案复杂度和检测效果上表现出优越性,可以应用于大规模恶意软件的初步检测。Considering the problem of the increasing number of android malware software and the increasing difficulty of large-scale software security detection, this paper proposed a lightweight malware detection method.Firstly, it analyzed plenty of Android software permission frequencies information.Secondly, it utilized permission frequencies to reduce redundancy.Finally it used minimum-distance classifier to classify the software.The experiments result proves that it is feasible.By comparison, it performs better in complexity and detection accuracy than other related work, which can be applied to preliminary analysis in large-scale software security detection.

关 键 词：最小距离分类器 安卓 权限频率 恶意软件检测 数据挖掘 

分 类 号：TP309.5[自动化与计算机技术—计算机系统结构]