标签: aerospace

As one of my full time job responsibilities, selling modeling products, I often encounter a very interesting question. Is the model cycle accurate ? Does it 100% reproduce the behavior of the actual target. Designers often think that this as a very important question. For years, my standard response has been "Teraptor Player CA is cycle aware", accuracy of the system can be made to be as close to the actual system as required.  However, in a recent meeting, my response was spontaneously different, and it was actually a question,  and therein I realized lies an important warning for software developers, especially those creating software for aerospace or automotive applications. Well, that question is -- what is the target system or board behavior ? Does it exhibit the same behavior every time ? Do multiple identical boards all exhbit the same behavior with respect to time ?  Any one who administers a network knows it takes some effort to keep all the systems synchronized with respect to time.  The reality is that there is almost no such thing as a "standard temporal behavior" of a target system. It changes with temperature, pressure, aging, battery voltage and yes, even the history of temperature changes. In other words, a perfectly working automotive software application in the lab, may experience "timing related failures" when being driven out there in the middle of a desert, hot or icy, 10 years after it was manufactured. Modeling time then becomes very important, not just in terms of making it accurate when compared to the real target, but in terms of being able to make it inaccurate -- vary the relative logical  time between the components, so that they are in fact closer or farther away from the "real target". Testing software with different timing models  ensures that the software operates as designed on different hardware platforms - and makes it independent of the underyling hardware platform, temporally speaking.