SMINER: Detecting Unrestricted and Misimplemented Behaviors of Software Systems Based on Unit Test Cases

Despite the advances in automated vulnerability detection approaches, security vulnerabilities caused by design flaws in software systems are continuously appearing in real-world systems. Such security design flaws can bring unrestricted and misimplemented behaviors of a system and can lead to fatal vulnerabilities such as remote code execution or sensitive data leakage. Therefore, it is an essential task to discover unrestricted and misimplemented behaviors of a system. However, it is a daunting task for security experts to discover such vulnerabilities in advance because it is time-consuming and error-prone to analyze the whole code in detail. Also, most of the existing vulnerability detection approaches still focus on detecting memory corruption bugs because these bugs are the dominant root cause of software vulnerabilities. This paper proposes SMINER, a novel approach that discovers vulnerabilities caused by unrestricted and misimplemented behaviors. SMINER first collects unit test cases for the target system from the official repository. Next, preprocess the collected code fragments. SMINER uses pre-processed data to show the security policies that can occur on the target system and creates a test case for security policy testing. To demonstrate the effectiveness of SMINER, this paper evaluates SMINER against Robot Operating System (ROS), a real-world system used for intelligent robots in Amazon and controlling satellites in National Aeronautics and Space Administration (NASA). From the evaluation, we discovered two real-world vulnerabilities in ROS.     

MAPAS: a practical deep learning-based android malware detection system

A lot of malicious applications appears every day, threatening numerous users. Therefore, a surge of studies have been conducted to protect users from newly emerging malware by using machine learning algorithms. Albeit existing machine or deep learning-based Android malware detection approaches achieve high accuracy by using a combination of multiple features, it is not possible to employ them on our mobile devices due to the high cost for using them. In this paper, we propose MAPAS, a malware detection system, that achieves high accuracy and adaptable usages of computing resources. MAPAS analyzes behaviors of malicious applications based on API call graphs of them by using convolution neural networks (CNN). However, MAPAS does not use a classifier model generated by CNN, it only utilizes CNN for discovering common features of API call graphs of malware. For efficiently detecting malware, MAPAS employs a lightweight classifier that calculates a similarity between API call graphs used for malicious activities and API call graphs of applications that are going to be classified. To demonstrate the effectiveness and efficiency of MAPAS, we implement a prototype and thoroughly evaluate it. And, we compare MAPAS with a state-of-the-art Android malware detection approach, MaMaDroid. Our evaluation results demonstrate that MAPAS can classify applications 145.8% faster and uses memory around ten times lower than MaMaDroid. Also, MAPAS achieves higher accuracy (91.27%) than MaMaDroid (84.99%) for detecting unknown malware. In addition, MAPAS can generally detect any type of malware with high accuracy.