Trusted ILLIAC

This project uses reconfigurable hardware to implement application-aware error and attack detectors. This is part of my PhD thesis at the University of Illinois. I was the lead student in this project.

Motivation

The vision of Trusted Illiac is to create a demonstrably trusted platform to support the paradigm of “On-demand computing” or “Utility computing”. In this paradigm, the system configures itself transparently based on the needs of the application to provide optimal levels of performance and dependability to the application. This calls for novel solutions that are application-specific, and can be tuned to the needs and characteristics of the application. Further, it is important that the solutions be implemented in hardware to ensure low detection-latency and minimal overhead.

To this end, we propose the use of reconfigurable hardware to execute application-specific checks in parallel with the main program. The checks are derived by special-purpose compilers and configured onto the reconfigurable hardware at application load-time.

Project Summary

The project leverages the Reliability and Security Engine (RSE), which is a reconfigurable hardware framework that closely interfaces with the processor’s pipeline to provide application-aware checking. The RSE framework and the checks it executes are shown in the figure.

Hardware framework for Application-aware Checking

The left side of the figure shows the security checks executed by the framework including the Information Flow Signatures (IFS) technique, while the right side shows the reliability checks executed by the framework including the Critical Variable Recomputation (CVR) technique. Both classes of checks were derived using the common compiler framework described in detectors.
Read more about this project (Video)

Project Publications

Collaborators : William Healey, Peter Klemperer , Paul Dabrowski, Galen Lyle, Shelley Chen, Zbigniew Kalbarczyk, and Ravishankar Iyer