摘要
针对日益泛滥的虚假安卓APP,分析了安卓APP的多方面特征,设计并实现了一种基于静态分析的虚假安卓APP分析与检测系统。系统通过反编译安卓APK文件,分析并提取出包的信息和代码特征作为分类的特征向量,采用多种机器学习算法进行分类,并对其虚假程度进行分析与检测。通过实验对比不同机器学习算法在虚假安卓APP分类与检测的准确率,分析了不同机器学习算法的局限性。实验结果表明,该系统能够高效率、高准确率地检测虚假安卓APP。
Aiming at the proliferation of false android APP,various characteristics extracts packet information of the android APP are analyzed,the analysis and detection system of fraud android APP base n static analysis is designed and implemented.By decompiling android APK files,the system analyzes and extracts packet information and code features as the feature vectors of classification.A variety of machinelearning algorithms are used to classify and analyze the false degree.Experiment and comparison are done on the accuracy of different machine-learning algorithms in false android APP classification and detection,and the limitations of different machine-learning algorithms also analyzed.The experiment results indicate that this system could efficiently and accurately detect false android APP.
作者
齐林
刘功申
孟魁
蔡逆水
QI Lin;LIU Gong-shen;MENG Kui;CAI Ni-shui(School of Electronic Information and Electrical Engineering, Shanghai Jiaotong University, Shanghai 200240, China;State Engineering Laboratory for Mobile Internet System and Application Security,Shanghai 201315,China)
出处
《通信技术》
2017年第12期2840-2845,共6页
Communications Technology