안드로이드 플랫폼에서 정보 유출 탐지를 위한 지능형 접근제어 모델 설계

Abstract

As smartphones become increasingly popular, attacks on the open-source platform Android are also on the rise. A user’s personal information is stored on a smartphone and when the user activates an application which includes malicious code, that user’s information can be leaked, with accompanying damage. Therefore a method of detecting and controlling access to a user’s personal data within a smartphone is needed.

In this dissertation, categories of applications those operating on the device to detect theft of personal information using smartphone applications and to control access to prevent malicious behavior sold on the Android Market were classified based on package information. An ontology was constructed based on joint permissions of applications in each category as classified here and a security policy defined to detect use exceeding permissions.

The proposed technique was actualized on the Android platform and simulations and results were deduced. It was confirmed to have a lower error false positive rate than previous techniques based on experimental measurements of detection rate and error rate. Since the proposed technique compares application permissions and resource requirements, it has the advantage of being able to quickly detect and control access when new malicious behavior occurs.