A technique for testing command and control software
A technique for testing embedded-microprocessor command and control programs is described. The continuity inherent in functions computed by programs which monitor natural phenomena is exploited by a simple difference equation-based algorithm to predict a program's next output from its preceding ones. The predicted output is compared with the actual output while indexing through the program's domain. Outputs which cannot be predicted are flagged as potential errors. Data are presented which show that this technique can be a sensitive measure of a program's correctness.