Scott Ranville and Paul E. Black, Automated Testing Requirements-Automotive Perspective, The Second International Workshop on Automated Program Analysis, Testing and Verification, Toronto, Canada (May 2001).

Abstract:

    With the ever-increasing complexity of embedded software applications, and the emergence of more and more safety critical applications, thorough validation and verification of the code is needed. To address this need, many embedded software development groups are using models and doing up-front engineering before testing on the final product. Using the old style of testing late in the development cycle resulted in very long and expensive release cycles. Ford estimated that 60% of work tasks were to correct requirements or design defects that had been released to downstream developers. With today's increasing need to get to market quickly with a safe product, this old style of testing is not adequate. Ford also used randomly generated unit test vectors, due to the lack of a commercially available tool, which only had approximately 75% coverage. Because of the need for safe systems, this level of testing is insufficient.

    This paper presents requirements for model checking and unit test generation tools so that the tools are practical in a large production environment that is typical in the automotive industry.

Get the paper in RTF (61k), or ASCII text (29k).

This page's URL is /~black/Papers/autoTestReqsWAPATV.html

Updated Mon Apr 18 14:05:22 2005

Go to Black's papers or NIST home page.