Statistical Testing of Real Time Software

In view of the difficulties with real-time software verification that were discussed in the earlier chapters of this book the reader might ask wether it can be easier, and therefore cheaper, to use probabilistic methods instead of systematic ones. As we will see later, however, it is normally quite costly to apply statistical testing for singular (i.e. not diverse) software system. This is mainly due to the large number of test runs that are required to achieve quantitative results which are acceptable for a system of any practical importance. This leads to the recommendation to use statistical tests only as a complement to systematic verification.