TWC: Small: STRUCT: Enabling Secure and Trustworthy Compartments in Mobile Applications

社会对移动的技术的依赖迅速增加，因为我们委托移动的应用程序与越来越多的私人信息和能力。现有的安全研究遵循一种常见的威胁模型，将应用程序视为单一实体，仅捕获应用程序之间的攻击面。然而，最近的研究表明，应用程序内部攻击正在迅速出现，因为具有冲突利益的复杂实体通常包含在单个应用程序中，以允许丰富的功能和快速开发。 这个项目被称为“应用程序隔离测试”，系统地研究了应用程序隔离，将其作为一种新颖而通用的方法来减轻应用程序的关键但未解决的内部威胁。它通过解决四个具有挑战性和有趣的研究问题将这种方法应用于主要的移动的平台：（1）推导设计应用程序内安全机制的原则和模型;（2）构建编译器工具链，用于自动和安全地划分应用程序;（3）构建开放平台的系统级执行机制;（4）构建封闭平台的应用程序级系统不可知的执行机制。 这些挑战的解决方案共同构成了应用内安全隔离和策略实施的设计和实现基础，这是目前不存在的，但需求很高。在促进移动的安全研究和教育的新方向以及在安全敏感场景中增加社会对移动的技术的采用方面，移动通信技术具有更广泛的影响。

Society's dependence on mobile technologies rapidly increases as we entrust mobile applications with more and more private information and capabilities. Existing security research follows a common threat model that treats apps as monolithic entities and only captures attack surface between apps. However, recent research reveals that app internal attacks are emerging quickly as complex entities with conflicting interests are commonly included inside a single app to allow for rich features and fast development. This project, known as STRUCT, systematically investigates app compartmentalization as a novel and general approach to mitigating the critical yet unaddressed internal threats of apps. It applies this approach to major mobile platforms via solving four challenging and interesting research problems: (1) Deriving principles and models for designing intra-app security mechanisms; (2) Building compiler toolchains for automatically and securely compartmentalizing apps; (3) Building system-level enforcement mechanisms for open platforms; (4) Building app-level system-agnostic enforcement mechanisms for closed platforms. Solutions to these challenges together form a foundation to the design and implementation of intra-app security isolation and policy enforcement, which is currently nonexistent but in high demand. STRUCT has its broader impact in fostering a new direction in mobile security research and education as well as increasing society's adoption of mobile technology in security-sensitive scenarios.