2007-08: Achieving Highly Reliable Embedded Software: An empirical evaluation of different approaches
The following technical report is available from http://aib.informatik.rwth-aachen.de: Achieving Highly Reliable Embedded Software: An empirical evaluation of different approaches Falk Salewski and Stefan Kowalewski AIB 2007-08 Designing highly reliable embedded software is a challenge and several approaches are known to improve the reliability of this software. However, all approaches have their advantages and disadvantages which makes empirical evaluations investigating their potentials necessary. In this paper, different approaches of software reliability improvement for embedded systems were compared on basis of experiments conducted at our institute. The first approach is an instance of N-version programming based on forced diversity. Two fundamentally diverse hardware platforms (microcontroller and CPLD/FPGA) were used to force diversity. Another experiment was conducted in which participants designed their software on one hardware platform only. The second half of this experiment was used for review and testing. Based on our experiments, the potentials of our application of N-version programming, review and testing are compared with respect to different fault categories (specification, implementation, application) identified during evaluation.
participants (1)
-
Peter Schneider-Kamp