Uma abordagem dirigida por modelos para a configuração de aquisição de contexto intermediada por middleware

Abstract

Context-aware mobile (CAM) applications retrieve contextual information from the environment in which they run in order to achieve a specific goal (e.g., interface adaptation, content recommendation). Features such as device heterogeneity, scarce resources and sensors diversity improve the complexity of development to this kind of software. Recent way to deal with such development issues is the adoption of two combined approaches: (i) middleware platforms and (ii) principles of MDE paradigm (e.g., Model-Driven Engineering). These approaches aim at reducing the total development time of CAM applications by using code generation from higher-level models, and increasing the potential applications through the use of services provide by middleware platforms. Thus, this dissertation proposes an approach to generate code that configures the acquisition of context in Android mobile applications. Based on MDE, we created the DSL (Domain-Specific Language)ContextRuleML, which allows the developer to model the contextual rules of a CAM application in a higher-level notation. A configuration tool, named CRiTICAL, uses the models produced with the DSL to generate an initial structure of a CAM application that uses an existing middleware platform to acquisition of context called LoCCAM. The tool also generates the configuration and the automated installation of LoCCAM and its components based on generated model. Two evaluations were made. First, a usability evaluation was realized by 14 volunteers. Second, a performance evaluation compares response time and memory used by the code generated by CRITiCAL against code written by a specialist in development with LoCCAM, without using our solution.