Automatic Reactive Adaptation of Pervasive Applications

Pervasive computing envisions seamless and distraction-free support for everyday tasks through distributed applications that leverage the resources of the users´ environment. Due to the mobility of users and devices, applications need to adapt continuously to their changing execution environment. Therefore, developers need a suitable framework in order to efficiently create adaptive applications. In this paper, we present and evaluate our approach to adapting a pervasive computing application to changes during its execution. This work is based on the minimal component system PCOM [2] and on an algorithm to fully automate the initial configuration of a component-based application [11] which we have presented in earlier work. The contribution of this paper is threefold. First, we describe a number of modifications to the component model that are required to enable fully automatic adaptation. Secondly, we propose a simple yet powerful cost model to capture the complexity of specific adaptations. Thirdly, we describe an online optimization heuristic that extends our distributed configuration algorithm in order to choose to a low-cost configuration whenever the current configuration of a pervasive application requires adaptation.