DMCA
Experimental Evaluation of a COTS System for Space Applications (2002)
| Venue: | in Proc. Int. Conference on Dependable Systems and Networks (DSN-2002 |
| Citations: | 14 - 2 self |
BibTeX
@INPROCEEDINGS{Madeira02experimentalevaluation,
author = {Henrique Madeira and Raphael R. Some and F. Moreira and D. Costa and David Rennels},
title = {Experimental Evaluation of a COTS System for Space Applications},
booktitle = {in Proc. Int. Conference on Dependable Systems and Networks (DSN-2002},
year = {2002},
pages = {325--330},
publisher = {IEEE CS Press}
}
Share
 |
 |
 |
 |
||
OpenURL
Abstract
The use of COTS-based systems in space missions for scientific data processing is very attractive, as their ratio of performance to power consumption of commercial components can be an order of magnitude greater than that of radiation hardened components, and the price differential is even higher. A major problem, however, is that COTS components are susceptible to Single Event Upsets (SEU), which are transient errors caused by space radiation. This means that their actual use in space missions must be preceded by careful study on the impact of faults on system behavior and the identification of weak points that should be strengthened with specific software fault tolerance techniques. This paper evaluates the impact of transient errors in the operating system of a COTS-based system (CETIA board with two PowerPC 750 processors running LynxOS) and quantifies their effects at both the OS and at the application level. The results provide a picture of the impact of faults on LynxOS key features (process scheduling and the most frequent system calls), data integrity, error propagation (among application processes, from applications to the OS, and from OS to the application processes), application termination, and correctness of application results. The study has been conducted using a Software-Implemented Fault Injection tool (Xception) and both realistic programs and synthetic workloads (to focus on specific OS features) have been used. Results show that the impact of errors is very dependent on the application and on the LynxOS module in which they occur. A variable but significant percentage of wrong application results have been observed, which demonstrate that additional fault tolerance techniques are required to guarantee result correctness in the presence of faults.
Keyphrases
experimental evaluation space application cot system space mission transient error cots-based system data integrity single event upset result correctness careful study weak point wrong application result process scheduling application process major problem commercial component space radiation application level application termination significant percentage lynxos module price differential system behavior additional fault tolerance technique software-implemented fault injection tool power consumption lynxos key feature cot component error propagation frequent system specific software fault tolerance technique operating system cetia board synthetic workload actual use scientific data processing application result specific o feature realistic program
