Research On Performance And Security Detection Of Android Application Software

In this current information age,the mobile phone has become an indispensable part of people's life.Android mobile phone occupies most of the market,and the number of applications has been growing wildly.Due to the lack of supervision in the market,there is a variety of problem software,especially the performance and the security issues.How to detect problem software effectively is important,which can enhance users' experience and protect the security of users' information and property.This paper studies the performance data acquisition and security detection of application software,and develops a system to detect the performance and security based on Android and machine learning.The main research of this paper is as follows:(1)According to the analysis of existing detection methods for application software and the needs of users,the overall framework of Android application software performance and security detection system are designed,which is divided into two modules including the performance detection and the security detection.(2)In the performance detection module,various performance indexes are designed,including fluency,CPU usage,memory usage,battery and network traffic.A new fluency evaluation standard is proposed by studying the UI rendering mechanism of Android.Based on the research of Linux system files and Android system processes,this paper studies the calculation methods of CPU usage,memory usage and other indexes.(3)In the security detection module,the features of the application software are extracted,and the machine learning algorithms are used to study and establish the security detection model.First,the feature information in the application software is extracted based on the analysis of the security mechanism of Android and the difference between benign and malicious software.Next,the feature selection algorithms are used to reduce the dimensions of the original features.Finally,aiming at the problem that the SVM algorithm is not good for training detection model,the AdaBoost-SVM algorithm is used to improve the accuracy of the detection model,which is based on the ensemble learning method.(4)After studying the performance detection method of application software and constructing the security detection model,this paper designs and implements a APP tool(Androidet)to detect the performance and security of the application software by using the modular method.The innovations of this paper are as follows:(1)In the performance detection module,in order to overcome the shortcomings of the traditional FPS method for measuring the smoothness of Android application,a new fluency evaluation parameter FD is proposed,and a scoring model of fluency is designed.(2)In the feature selection stage of the security detection model,combined with the imbalance and the correlation of the samples,this paper improves the Relief algorithm,which improves the accuracy of the security detection model effectively.