Desarrollo de sistemas domoticos utilizando un enfoque dirigido por modelos

This Thesis continues the research line in the development of reactive systems initiated by the DSIE research group of the Technical University of Cartagena ten years ago and starts from the results of a previous Thesis, in which a component metamodel, called V3Studio, was designed to collect the requirements of reactive systems. This metamodel is intended (1) to be a point of confluence for reactive systems and (2) to obtain platform independent models. It has been previously used in the domains of robotics, computer vision and wireless sensor and actuator networks.;The present Thesis completes previous contributions proposing a new methodology and an associated environment to provide full support for the cycle of software development for home automation systems. The development of home automation systems involves areas such as information technology, automation, network management and microprocessors programming. At present, the design process in home automation is similar to the one employed in other reactive systems. In all of them a domain expert, with a wide background in the implementation platform, is required. Furthermore in most of cases the design is started from scratch and the effort to obtain the programming code is high, dealing to custom solutions which are rarely reused. These and many other problems raise the need for a new approach to improve the development process.;To solve the problems associated with the traditional development process in home automation this Thesis proposes (1) a domain specific language for the home automation domain to collect application requisites with a high abstraction level; (2) a metamodel to support the language and restrictions to guarantee well formed models and (3) a methodology which combines the newly Model-Driven Engineering paradigm together with the domain specific language. To achieve these goals a code generation approach for the KNX/EIB platform has been proposed and the required transformation rules have been defined. These rules are used to transform models from the highest abstraction level to the intermediate component model and, later, to platform models in order to produce executable code. Finally a case study example has been proposed to validate the language and the methodology.