基于最小距离分类器的Android恶意软件检测方案

针对Android手机应用商店对大规模软件的安全性检测问题,提出了一套轻量级恶意软件检测方案。该方案首先分析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,通过与其他方案对比,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。

Android malware detection based on minimum-distance classifier

Considering the problem of large-scale software malware detection, this paper proposed a lightweight malware detection method. Firstly we analyzed plenty of Android software permission frequencies information. Secondly. selecting some representative permissions by permission frequencies to reduce redundancy. Finally using minimum-distance classifier to classify the software. The experiments result prove 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.