| Abstrakt: |
The basic characteristics of the model of the software system which belong to the working capacity and predictability, are described. Analyzed tools were used for the modeling of software systems. As a basis for the tools of modeling and analysis of the dynamic properties of software systems, the choice of interpretations and modifications of Petri nets is justified. They allow to follow structural similarity in the model system and also have an unambiguous mathematical description. The analysis of the properties of Petri nets allows to reveal obvious and hidden errors in functioning of PN-model of software system, that is not always possible in the application of simulation modeling. To such properties belongs liveliness, boundedness, reachability, preservation, conflictlessness, controllability. The T- and P-invariants, as well as the characteristics of the incidence matrix, allow to detect compliance with these characteristics. The matrix description of the PN-model was carried out with the help of the basic equation of the Petri net, definition identification of invariants and their analysis to reveal dynamic properties of the model. The rule of absence of hidden deadlocks and infinity loops in PN-model is formulated. The detection of hidden errors — deadlock and infinity loop are described. The solution of the deadlock elimination was proposed; its conformity to the dynamic properties was checked. The definition and correction of hidden errors on the example of not fully controlled PN-model is illustrated. The presented method of analysis of PN-models has practical application in step-by-step analysis of the software model. Phasing of the analysis of software model caused by the proliferation of component-oriented technology of software system development and the limiting the number of PN-model elements. It can be applied both independently and as a part of the combined approach to the simulation modeling of systems with parallelism. [ABSTRACT FROM AUTHOR] |
