Lenguaje de modelado de arquitectura para sistemas ciber-físicos adaptativos

Abstract

It is becoming increasingly common for systems to be dynamically adapted to changing contexts at run time. Such systems exhibit degrees of variability that depend on the fluctuations of the runtime in their contexts. Although dynamic variability has been addressed by long-established concepts in the field of system families, this approach is considered insufficient to meet the needs of distributed and dynamically adaptive systems such as IoT (Internet of Things)-based intelligent agriculture systems considered cyber-physical systems. These systems are required to improve the productivity and competitiveness of Colombian agricultural production. For dynamically adapted systems, approaches to support the variability of cyber-physical systems may not be just component specializations or restrictions on system variables since often the operating conditions of these systems are not predictable. In this study that seeks to establish the most appropriate techniques for managing dynamic variability in self-adapting systems, it has been necessary to address architecture modeling as a key practice in software engineering. It is considered that a modeling language that allows to represent such systems will be definitive for the production of systems with the level of quality and reliability that is required. All the results of this research are focused on enabling development companies, the tools that allow them to increase the productivity, flexibility and quality that cyber-physical systems require for agriculture. For the study it will be necessary to analyze the literature to create a language that improves the state of the art and to evaluate and perform at least one test case of a real system, this will allow to validate the proposed language. This work starts from the premise that that language should allow you to specify software, hardware, monitoring, and action components on the physical environment of each system and represent physical objects that are part of the system to be deployed. In order to test the novel nature of the proposed language, this study includes a comparison with other modeling languages proposed by the world's academic and industrial community. Finally, the study will result in a language of modeling the architecture of self-adapting systems, seen as a family or individually. The language is implemented in the VariaMos platform (http://variamos.dis.eafit.edu.co/#/) seeking to show that it can be used to represent, at the same time, reference architectures and individual system architectures, thanks to the mechanisms offered by the tool to represent and manipulate variability. The justification for the validity of the use of the proposed language is made from the presentation of a real case of a smart greenhouse. It is intended to be able to demonstrate that: (i) The modeling language has characteristics that give the possibility to represent elements such as physical and environment objects that other languages do not allow and extends the possibilities of designing cyber-physical systems (ii) The language has elements that expand the possibility of specifying in models, the components of variability that an architecture for a self-adapting cyber-physical system requires in its representation. (iii) The architecture modeling language for self-adapting cyber-physical systems while designed with IoT system design as a benchmark in a software product line domain, this language can be extended to design other cyber-physical systems.