Modelagem de sistemas com restrições temporais em redes de petri orientadas a objetos.

Abstract

Over the last years, with the crescent and rapid advance of distributed and parallel systems, timed models have become subject of several researches. Over different areas, different timed models have been proposed to model time constrained functionalities and promote system performance analysis in different situations. The proposed models are, in general, extensions of existing models, in which it was identified the need for coping with time parameters in a quantitative way. Essentially, this work concerns the definition of a timed extension for Redes de Petri Orientadas a Objetos (RPOO), a formalism that, like many others, has presented problems when applied to the modelling of real time systems, due to the lack of inherent means to explicitly cope with time information. RPOO is a formalism that integrates several features of both the OO and the Petri net languages in order to improve the production, simulation and validation of formal models. RPOO is a modelling language that can be seen either as an object-oriented extension for high-level Petri nets or as a way to give formal concurrent semantics for object-oriented models. Its original proposition does not provide inherent, quantitative time modeling in none of these two perspectives. Our timed extension integrates the time treatment in both the OO and the Petri net levels. As a result, the extension is more expressive than the original formalism proposition, in the sense that it makes easier the task of modeling certain classes of problems (the ones which present time constrained features). As a mean to validate the formal definitions, a simulator prototype supporting some features of the proposed extension was implemented. Also, we present a case study, in which the prototype was applied in the analysis of Mobile IP, a protocol whose modeling in original RPOO identified deficiencies of the formalism when it comes to time treatment. Several protocol performance measures — which could not be obtained in previous works with original RPOO — are presented as results of automatic simulation process with the developed simulator prototype.