An Approach Based in Petri Net for Requirement Analysis

Abstract

Embedded systems that have timing constraints are classified as real-time systems. In these systems, not only the logical results of computations are important, but also the time instant in which they are obtained. Hard real-time systems are those whose the respective timing constraints must be met at all cost, since violation might be catastrophic. Hence, time predictability is an essential issue (Barreto & Lima (2004)). In addition, the widespread expansion of mobile devices market has forced embedded systems companies to deal with several new challenges in order to provide complex systems in this market niche. In this context, energy consumption deserves special attention, since portable devices generally rely on constrained energy sources (e.g. battery) . As consequence, early estimation of the energy consumption can provide important insights to the designer about the battery lifetime as well as parts of the application that need optimization (Tavares et al. (2007)). Nowadays, UML (UML (2005)) is the most adopted modeling language for system design in the software engineering organizations and industry. The main reasons are: (i) its friendly and intuitive notations, (ii) availability of commercial and open source tools that support the UML notations and (iii) autonomy of particular programming languages and development processes. However, UML does not provide support for quantitative notations. Quantitative notations are especially important when modeling Embedded Real-Time Systems (ERTS). Hence, we consider UML in combination with MARTE (UML Profile for Modeling and Analysis of Real-Time and Embedded systems) as specification language for the design of ERTS. MARTE foster the construction of models that may be used to make quantitative predictions regarding real-time and embedded features of systems taking into account both hardware and software characteristics (MARTE (2005)). UML 2.0 is composed of several diagram types (e.g. activity, sequence, use case, class, timing and many others). Interaction Overview Diagram (IO) (UML (2005)) is adopted in this work due to its suitable characteristics for modeling requirements when dealing with ERTS, since UML-IO combine elements of activity diagrams with sequence diagrams to represent the embedded system behavior. Without loss of generalization, this work aims to depict the mapping process of UML-IO into a Time Petri Net with Energy constraints (ETPN) in order to estimate the energy consumption and execution time of ERTS. These estimates are performed in the early stages of the embedded system life cycle, serving as one instrument for design decision-making process. First, the