Project: Memory-access pattern secure software systems on Intel SGX
- Faculty: Dr. Yuzhe (Richard) Tang [webpage]
- "Scaling Memory Access-Pattern Secure Computations via Dynamic Program Partitioning", Ju Chen, Cheng Xu, Kai Li, Yuzhe (Richard) Tang. [Tech. report 2018]
- "Strongly Secure and Efficient Data Shuffle on Hardware Enclaves", Ju Chen, Yuzhe (Richard) Tang, Hao Zhou. SysTex 2017 at ACM SOSP (Workshop paper), [PDF (ACM)], [PDF (Full paper on arXiv)]
- The work is to build a data shuffling capability on Intel SGX platform in a unique way that is both strongly secure (or oblivious memory access) and efficient. The key idea to lower performance overhead without sacrificing strong security is the notion of cache-miss obliviousness. Upon implementation, the cache-miss obliviousness is realized by leveraging Intel TSX features.
- "Towards Building Practical And Secure Multi-Party Databases", Yuzhe Tang, Wenqing Zhuang. IEEE SecDev 2016. [PDF]
- The work is to build a database system on top of multiple private-data sources. A user can submit a regular SQL query to our database which evaluates the query among data sources while preserving the privacy of each data source. Comparing existing work on applied multi-party computations (MPC), our work is distinct in building real MPC-based systems.
Broader Impact Activities
- Public lecture: "Get Your Head In The Clouds! Cloud Computing's Risks and Rewards", at The Museum of Science and Technology at Syracuse, June 2017, [link]
- Public lecture: "Blockchain: Applications, Security Promises and Internals", at CSIAC, Dec. 2017, [webinar], [slides]