Composition and Self-Adaptation of Service-Based Systems with Feature Models

Abstract

The adoption of mechanisms for reusing software in pervasive systems has not yet become standard practice. This is because the use of pre-existing software requires the selection, composition and adaptation of prefabricated software parts, as well as the management of some complex problems such as guaranteeing high levels of efficiency and safety in critical domains. In addition to the wide variety of services, pervasive systems are composed of many networked heterogeneous devices with embedded software. In this work, we promote the safe reuse of services in service-based systems using two complementary technologies, Service-Oriented Architecture and Software Product Lines. In order to do this, we extend both the service discovery and composition processes defined in the DAMASCo framework, which currently does not deal with the service variability that constitutes pervasive systems. We use feature models to represent the variability and to self-adapt the services during the composition in a safe way taking context changes into consideration. We illustrate our proposal with a case study related to the driving domain of an Intelligent Transportation System, handling the context information of the environment.