ContextPoint : An Architecture for Extrinsic Meta-Adaptation in Smart Environments

The establishment of mobile devices had a high impact on the use and development of software systems. It is expected that the ability to automatically adapt to changing environments will be a crucial property for future apps running on mobile devices. The problem with current approaches for self-adaptive systems is that developers must define the adaptive behaviour at design-time. In many cases, however, the developer cannot predict all situations at design-time, which should trigger adaptation at runtime. Furthermore, applications for mobile devices are usually optimized for a small set of use cases and have a narrow, welldefined scope. In order to support more complex tasks, the functionality of several apps has to be composed dynamically. Another problem arises from the ever increasing number of available applications. In this paper, we address these problems by proposing a novel infrastructure for selfadaptive systems in smart environments, namely ContextPoint. Our goal is to describe an architecture which supports unanticipated adaptation for single systems, as well as the automatic integration of actuators and sensors, situated in the environment, with services and data from both, mobile devices and the cloud. Therefore, a distributed adaptation technique is proposed, where adaptation logic and rules are provided by the environment itself. This decentralization simplifies the development of self-adaptive systems with dynamic adaptive adaptation processes (meta-adaptation) and, thus, the design and operation of systems with unanticipated adaptation. Furthermore, our approach provides means for describing context-dependent collaboration between varying systems enabling the design of ad-hoc system-of-systems. Keywords—Adaptation; Self-Adaptive; Meta-Adaptation; Architecture; Context-aware; Location-aware.