MACHINE LEARNING BASED AUTOMATIC PERMISSION GRANTING AND MALWARE IDENTIFICATION

The mobile device have gained an imperative predominance in the daily routine of our lives, by keeping us connected to the real world seamlessly. Most of the mobile devices are built on android whose security mechanism is totally permission based controlling the applications from accessing the core details of the devices and the users. Even after understanding the permission system often the mobile user are ignorant about the common threat, due to the applications popularity and proceed with the installation process not aware of the targets of the application developer. The aim of the paper is to devise malware detection with the automatic permission granting employing the machine learning techniques. The different machine learning methods are engaged in the malware detection and analyzed. The results are observed to note down the approaches that aids better in enhancing the user awareness and reducing the malware threats, by detecting the malwares of the applications.

New Security Risk Value Estimate Method for Android Applications

Nowadays, mobile applications are the devices’ core, so their security is essential for the platform on which are installed. Therefore, to make this security strong, Android implements permission system and sandboxing mechanism to reduce the attacks contingency. Also, antivirus software is used to detect the already known malware based on their signature, but unfortunately, this kind of program cannot detect cloned or repackaged malware. Thus, using authorizations to estimate the security vulnerability will surely be very useful for avoiding applications that are more susceptible to be dangerous according to the risk values obtained. Consequently, this will provide systematic support that will make it easier for users to make appropriate decisions and therefore greatly improve the Android devices’ security. In this article, additionally to the risk measures that have been already proposed in previous studies based on functionalities such as permissions and function calls, we add a set of mathematical metrics describing the case of susceptible and innocent applications already known. Our risk assessment exploits applications already known as malware and safe samples without any threat.