Abordagem para especificação das ocorrências de características transversais em software embarcado

Abstract

Embedded software engineering has evolved rapidly over the past few years, especially to deal with the increased complexity associated with a large number of functional requirements and crosscutting concerns associated with non-functional requirements. The process of identifying and specifying the selection of points where crosscutting concerns should be addressed has been widely discussed since the mid-1990s, especially in the context of Aspect-Oriented Software Development (AOSD). However, understanding these specifications is often difficult, since the syntax and semantics of the languages used, especially visual languages, do not allow you to easily deduce the meaning of the symbols used. Such a situation produces artifacts that are difficult to understand and maintain. Therefore, the process of locating, identifying, and specifying the points where the crosscutting concerns must be handled is a key in the embedded software design, as it can lead to problems in the development and, mainly, in the integration of the system’s software components. This master’s thesis proposes a graphical modeling technique for specifying the selection of Crosscutting Concerns Occurrence Points (CCOP) for embedded systems under the name of JSD (Join Point Specification Diagram). An empirical evaluation was carried out to assess the understanding of the CCOP selection specification indirectly through the evaluation of the cognitive effects and the perception properties of the notations. Five notations were evaluated, two graphical (JPDD and Theme/UML) and three textual (AspectJ, Tracematch, and AspectOCL), which were used to specify implicitly and explicitly the selection of CCOP (for 16 different points) in three different embedded software projects that represent real applications in the context of automation systems. This work also presents the same evaluation for the proposed approach, JSD, using the same embedded software projects. The purpose of the assessment is to verify the ability to understand the specifications using the proposed technique through the assessment of cognitive effects and perception properties. A quality model based on the conceptual framework “Physics of Notation” (PoN) is proposed. Nine metrics were used to quantify the perceptual properties in the specification of the CCOP selection, with some metrics created in the context of this work and others representing concepts within the principles of PoN. The results of the experiments show how the characteristics of each of the notations impact the understanding of the specifications created. Empirical evidence was found that the graphic specification of CCOP selections using JSD has high discrimination on the cognitive effectiveness of visual representations (according to the PoN framework) and, hence, it seems that the use of JSD to specify the selection of CCOP in embedded systems is more effective in understanding and interpreting its perceptual properties (syntax).