The research of statistical testing fault detection method based on embedded software

The purpose of this research is to locate software failure by establishing a improved method of fault location based on Statistical predicate assessment model quickly and accurately. In our research, we using TTCN-3 language for modeling the entire testing system, evaluating reconstructing the needed TTCN-3 test set and then combining the implementation of the testing cases after memorization to build a fault locating method based on predicate evaluating model, building the TTCN-3 abstract model of aiming at the field of real-time embedded, applying the class coherence of LCMO metric into TTCN-3 model coherence analysis, and applying this metric project into Trex for it´s extension, building the statistics evaluation model aiming at predicate as well as modeling by the way of using testing method of parametric and non-parametric. The results indicate that it´s more efficient and accurate to locate procedural errors.