Generation of models for reliability growth

The author considers computerized reliability growth model generation based on the development of Petri nets. The proposed method makes it possible to show and investigate the main aspects of debugging using the following techniques: time-variable stochastic mechanism of manifestation and removal of already existing errors and new errors introduced in the debugging process; time and probability characteristics of interaction between debugged and debugging systems; the reliability growth of software and the debugging system; and time delay between the first manifestation of error and its removal. The practical realization of this method in the SIMONA-PH system provides the possibility of generating a model closer to reality, and of solving the problems of prediction and reliability growth management. The probability of debugging aim realization is considered as the debugging completion criterion. The debugging criterion of efficiency allows one to analyze the level of realization of debugging aims depending on debugging tools, performance characteristics, and time-resources allocated for debugging.<
>