基于电量消耗的Android平台恶意软件检测

发布时间：2018-05-08 21:25

本文选题：移动终端 + 电量消耗　；参考：《清华大学学报(自然科学版)》2017年01期

【摘要】：根据Android应用在运行期的耗电时序波形与声波信号类似的特点,该文提出了一种基于Mel频谱倒谱系数(Mel frequency cepstral coefficients,MFCC)的恶意软件检测算法。首先计算耗电时序波形的MFCC,根据MFCC的分布构建Gauss混合模型(Gaussian mixture model,GMM)。然后采用GMM对电量消耗进行分析,通过对应用软件的分类处理识别恶意软件。实验证明:应用软件的功能与电量消耗关系密切,并且基于软件的电量消耗信息分析可以较准确地对移动终端的恶意软件进行检测。
[Abstract]:According to the similar characteristics of power consumption time series waveform and acoustic signal in Android application, a malware detection algorithm based on Mel frequency cepstral efficient coefficients of Mel spectrum cepstrum is proposed in this paper. Firstly, the MFCC of the power consumption time series waveform is calculated, and the Gauss mixed model is constructed according to the MFCC distribution. The Gaussian mixture model is constructed. Then GMM is used to analyze the consumption of electricity, and the malware is identified by the classification of the application software. The experiment proves that the function of the application software is closely related to the consumption of electricity, and the analysis of the consumption information based on the software can detect the malware of the mobile terminal more accurately.
【作者单位】：中国民航大学计算机科学与技术学院;
【基金】：国家科技重大专项基金资助项目(2012ZX03002002) 国家自然科学基金资助项目(60776807,61179045) 中国民航科技基金资助项目(MHRD201009,MHRD201205) 中央高校基本科研业务费专项资金项目(3122014D033)
【分类号】：TP309;TP316

【相似文献】