Measures are important as an interface between the computer engineer and the application designer. We also needed tools to quantify those features of dependable performance which havs the greatest impact on the application.
Measures already existed for non-real-time computers; however, measures for embedded systems are fundamentally different from these. The key parameter in our case was system response time, which is a complex function of a number of hardware and software parameters.
Here is our assessment of measurement tools that were available at that moment and why we considered them unsuitable for real-time applications.
Our tactic was to identify generic attributes that were important to real-time dependability, assess the capability of the computer system in terms of these attributes, and using the information generated from these two steps, obtain the application-driven computer reliability.
A diagram that illustrates where our project lied in the tradeoff between generic and specific usage