Design and Evaluation of Fault-Tolerant Systems

Computer Engineering Research Center
The University of Texas at Austin

Research projects include:

Fault injection in COTS applications

Verifying the equivalence of two Boolean functions is a the heart of all verification. Since even this problem is very complex in general, we have developed a variety of solutions to verifying large functions. These include probabilistic verification, indexed Binary Decision Diagrams, functional partitioning and the use of multiple filters.

Application-Level Checks for Concurrent Error Detection

There has been a lot of work in hardware techniques for detecting errors concurrently with normal operation. This research aims to develop application level checks in software to detect errors due to transient and permanent hardware faults. The checks are incorporated automatically into user code, and are executed as part of the application. Control flow checks as well as data checks based on extensions of the well-known technique of Algorithm-Based Fault Tolerance are studied.

Evaluation of System Reliability

We have developed a hierarchical technique for evaluating system reliability.

Back to: