Pervasive computing environments are commonly regarded as being made up of a multitude of autonomous elements collaborating to sense and respond to a user’s requirements and the context of the task-at-hand. The management of pervasive computing environments therefore requires runtime understanding of a user’s task requirements and operational context, the state of resources available to support the user’s tasks as well as the broader management policies that also govern those resources. At the same time, the dynamic, ad hoc nature of pervasive computing environments requires an integrated, context-aware and self-organizing approach to their management. Autonomic systems self-manage their configuration, performance and integrity based on high-level goals set by users and administrators, and are therefore are well suited to the management of pervasive computing environment.
In this project, KDEG addresses how management of pervasive computing must balance the need to hide the complexities of adaptive service operation, while at the same time providing suitable levers of control and windows of inspection through which users can maintain a sense of ownership over adaptive service behaviour. With a system offering highly adaptive services to users, the management task of enforcing responsibilities over resources is essentially one of constraining the range of adaptivity the system can exhibit in different situations. These constraints must themselves adapt to a heterogeneous, ever-changing set of resources, that will be difficult to determine a priori. The work in this project focuses on: seamless interoperability amidst a high level of organisation heterogeneity at development time and runtime; service composition as the primary adaptive mechanism; satisfy management goals using policy-based management and supporting the needs of users both as individuals and as collaborative communities. Experimentation has involved some use of sensor and actuation systems, but has been effectively accelerated by the development of a 3D, multi-user simulator of smart spaces.