The Effectiveness of Data Diversity As An Error Detection

Report
Authors:Knight, John, Department of Computer ScienceUniversity of Virginia Ammann, Paul, Department of Computer ScienceUniversity of Virginia Santos, Steven, Department of Computer ScienceUniversity of Virginia
Abstract:

We have proposed previously an approach to software fault tolerance based on diversity in the data space; a technique we term data diversity. In a fault-tolerant system employing data diversity, the same software is executed on several set of points in the data space, each of which should produce the same (or simply related) output. Error detection in data diversity is achieved by observing unexpected differences between the outputs of the software on the different data points. This approach to error detection can be employed in software testing as well as fault tolerance, and an important question in both applications is the error detection performance of data diversity. We report in this paper on a study in which several pieces of commercial software were obtained and into which faults were injected. These faulty programs were tested using data diversity for error detection. The results of this study show that data diversity can be an effective error detection technique.
Note: Abstract extracted from PDF file via OCR

Rights:
All rights reserved (no additional license for public reuse)
Language:
English
Source Citation:

Knight, John, Paul Ammann, and Steven Santos. "The Effectiveness of Data Diversity As An Error Detection." University of Virginia Dept. of Computer Science Tech Report (1991).

Publisher:
University of Virginia, Department of Computer Science
Published Date:
1991