Computer Science Colloquium
Friday, September 27
2:35pm in Wege Auditorium
Towards a Practical Secure Memory for Modern Deployments
Applications expect values stored in memory to be in the same state when they are fetched for correct execution. Unfortunately, memory is an imperfect device; physical device vulnerabilities allow for data in memory to be corrupted from software without needing access to the device. In this talk, I overview how hardware can implement secure memory, the challenges that doing so would entail, and proposed architectures that address these limitations. In particular, I argue that secure memory can incorporate lightweight extensions to the existing protocol logic that leverage application behavior to optimize for (1) performance, (2) complexity, and (3) scalability.
Samuel Thomas is a fifth year Ph.D. student at Brown University advised by R. Iris Bahar working on secure architecture, emerging technologies, software-hardware co-designed systems, and application deployments in the cloud. In addition, he works closely with graduate and undergraduate groups at University of Colorado Boulder, Colorado School of Mines, Boston University, and Technion. Prior to Brown, he received a B.S. from Davidson College where he worked on architecture-aware design of concurrent data structure libraries.