Grant: Discovery Grants Program - Individual

In this proposal, the PI will primarily focus on (1) developing evaluation frameworks to systematically assess security components and features, specifically in widely-used "trusted" security products---from security and privacy perspectives; and (2) develop new ways to improve system security and user privacy by hardening operating systems and CPU trusted execution environments (TEEs)---reducing the over-reliance on traditional third-party security solutions. Critical user-level applications may rely on OS/hardware for isolation, and on security solutions for a malware-free operating environment. Security research involving modern CPUs and operating systems has a long history. However, other components that are also heavily relied on to maintain system/application security have been much less explored---such as security products used for preventing common attacks, and security features (e.g., obfuscation, custom encryption) used by applications for self-defence purposes. These components will be the focus of our analysis. In addition to analyzing them individually, we will also highlight the consequences on the applications that depend on security products for their security and privacy guarantees - e.g., what happens when a security solution is maliciously terminated, or even worse, if the solution itself is infected by malware, specifically, malware posing advanced persistent threats (APTs). From the analysis of multiple products and applications, we will develop evaluation frameworks targeting specific solution classes. We will summarize the lessons learned from the security analyses of selected components and critical solutions, and use these lessons in the following ways: to design security components/solutions with failure/compromise in mind; to recover from a compromise and restore the root-of-trust (for both software and hardware root-of-trust, e.g., after the leakage of TEE private keys); to better utilize the existing OS-enabled features for application security and privacy; to develop and embed new privacy features in the OS to aid user privacy in modern tracking-infused application ecosystems; to develop and embed more security features into the OS to reduce reliance on third-party security solutions; and to improve OS-based security and privacy enforcement through the use of hardware-enabled TEEs. Overall, this proposal will revisit "trusted" security components that are critical for many other high-impact IT operations, uncover new security and privacy issues, and develop new solutions that will significantly improve system security. The PI is uniquely suited for this research program based on his past experience on hard system-level problems. The program will also train students in topics at the intersection of systems, security, and privacy---producing critical HQP for long-term socio-economic benefits. The outcome of this work will fundamentally change how application security and user privacy are currently treated.