The Design And Research Of The Security Evaluation Platform For Android Apps

With the development of Mobile Internet, people’s lives have changed. In the past, the main function of phones is calling, sending messages. But now, people can chat online, shopping and search information through the smart phones. What is more, smart phone can control the smart devices at home remotely because of the development of smart HEA. The smart phones of Android system occupy most of the smart phone market, and Android system has a huge number of users. Users need to install the application on their own Android phone in order to get more functions. But there are some risks to install these applications. A malware can bring benefits loss to mobile users.The main purpose of this paper is to design and implement a security evaluation platform for Android application with high detection speed and high recognition accuracy. The static analysis has a fast running speed and it brings low burden to the server, so this paper chooses the static analysis as the main research method to analysis the APK files of the Android application. Decompile the APK files and get the configuration file which includes the information of permissions. According to the permission check mechanism of Android system, Android application need to explicitly declare the permissions in the configuration file when they request for resources. In this paper, whether the application has malicious behaviors depends on its permissions.In this paper, a certain number of malicious apps and benign apps are collected. And then decompile these apps and get the permission information to compare. There are many difference between malicious apps and benign apps. This proves the permission information can be the features to classify malicious apps and benign apps. Using chi-square test, TF-IDF algorithm and Relief algorithm to select the features to recognize malware.In this paper, we choose BP neural network to identify the malicious software. Because of its strong nonlinear fitting ability, self-learning and adaptive ability, BP neural network is a research hotspot that a lot of scholars research. But BP neural network also has some problems. BP neural network has a slow convergence speed and it is easy to fall into local minimum value. In order to solve these problems, this paper used the cuckoo search algorithm to optimize the BP neural network. The cuckoo search algorithm use Levy flight to search step, and it can effectively solve the problem of local minima. And using cuckoo search algorithm to initialize the weights of BP neural network can increase the convergence speed.Finally, this paper also designed and implemented a security evaluation platform for Android apps to recognize the malwares.