TABLE OF CONTENTS This section describes a handful of applications of the theory presented in Sections 6.5 and 6.6. In subsequent sections a number of issues are considered that affect these applications, viz: choosing subdomains (Section 6.8.1), component independence (Section 6.8.2), proving components correct (Section 6.8.3), and validating experiments (Section 6.10).
The run time of systems constructed from components is the most straightforward application of the theory. For this case the issues of choosing subdomains and of component independence have the best resolution, the intuitive plausibility of the theory is greatest, and validating experiments are easiest to perform.
Run time was used in Sections 6.5 and 6.6 to present the theory and it will be the subject of the first validation experiments in Section 6.10.
Reliability is the basic quality parameter for all engineering artifacts, The inability to factor an accurate reliability for software components into the safety analysis of embedded systems is today a serious problem for agencies like the U.S. Nuclear Regulatory Commission, It was the reliability application that first suggested the subdomain solution [Hamlet (1996)] to the problem of input dependence.
For reliability, in the mathematical treatment the run-time combination operator + is replaced by multiplication of reliability probabilities, so equation (6.2) becomes:
| (6.4) |  |
where the < t i> vectors now represent reliability values rather than run times.
Reference [Hamlet et al. (2001)] presents a detailed discussion of the reliability application.
Compared with run time, the reliability...
