Automated software diversity for hardware fault detection

  • Authors:
  • Gerald Gaiswinkler;Andreas Gerstinger

  • Affiliations:
  • Elektrobit Austria GmbH, Vienna, Austria;Institute of Computer Technology, Vienna University of Technology, Vienna, Austria

  • Venue:
  • ETFA'09 Proceedings of the 14th IEEE international conference on Emerging technologies & factory automation
  • Year:
  • 2009

Quantified Score

Hi-index 0.00

Visualization

Abstract

Software in dependable systems must be able to tolerate or detect faults in the underlying infrastructure, such as the hardware. This paper presents a cost efficient automated method how register faults in the microprocessor can be detected during execution. This is done with the help of using compiler options to generate diverse binaries. The efficacy of this approach has been analyzed with the help of a CPU emulator, which was modified exactly for this purpose. The promising results show, that by using this approach, it is possible to automatically detect the vast majority of the injected register faults. In our simulations, two diverse versions have - despite of experiencing the same fault during execution - never delivered the same incorrect result, so we could detect all injected faults.