Software-level TMR approach for on-board data processing in space applications

Start Page
147
End Page
152
Citation
21st IEEE International Symposium on Design and Diagnostics of Electronic Circuits and Systems, DDECS 2018 : 147-152 (2018-07-11)
Contribution to Conference
21st IEEE International Symposium on Design and Diagnostics of Electronic Circuits and Systems, DDECS 2018  
Publisher DOI
10.1109/DDECS.2018.00033
Scopus ID
2-s2.0-85050969704
Publisher
IEEE
Handling faults in computing systems is often expensive in terms of power, area and financial costs. In domains requiring high reliability in harsh environments, like the space domain, special highly reliable components are used, which may adversely impact the processing performance. In this paper, we propose the STROBES algorithm for fault handling in a multi-node embedded system which can be composed of standard commercial off-the-shelf components. In particular, it does not require underlying synchronization, but relies on embedded system's properties to derive bounds for communication and processing times. The algorithm can handle asynchronous behavior between the nodes up to user-defined bounds, in addition to a fault in the state or fail-stop failure of a single node. Theoretical analysis shows that this is sufficient for extended operating times. Experimental data show the efficient behavior of the STROBES algorithm for practical application with different state and time bounds.