| Popis: |
This thesis contributes to the research domain of Ubiquitous and Context-aware Computing. It presents a novel middleware (entitled Context Provisioning Middleware with Support for Evolving Awareness; C- ProMiSE) that applies a consumer-producer role model as architectural basis. The middleware aims at supporting diverse applications and services to easily and coherently acquire relevant context information. A mediat- ing Context Broker facilitates the coordination between distributed Context Provider, Context Source and Context Consumer components. The chosen design principles support' self-management capabilities and modular extendibility during run-time. Communication is based on the Representational State Transfer (REST) approach. Context is represented in ContextML, an XML-based modelling schema, enabling a structured generally applicable basis for various context domains, e.g. spatial, temporal, device-specific and user-centric prop- erties. This combination of context management and context representation model allows for gradual and distributed context processing and reasoning. Context is abstracted in various layers from primitive data up to high-level interpretation, e.g. users' activities. Specific emphasis is put on probabilistic reasoning of data originating from physical, virtual and logical sensors. In addition to the conceptualisation, a specific prototype implementation is presented and utilised as ex- perimentation testbed. Its functional evaluation covers field tests and context emulation. With regard to quantitative evaluation, the C-ProMiSE performance is finally assessed by applying both black-box tests of specific prototype components and Discrete Event Simulation at system level. Focus of the experimen- tation is to estimate the responsive context provisioning behaviour realistically. The obtained results lend weight to the argument that a consumer-producer-broker based framework can serve as basis to realise a distributed multi-domain context provisioning middleware that does not only scale physically but also functionally with regard to context processing capabilities. Furthermore, the con- text management, context representation and context processing concepts allow for supporting a variety of emerging and evolving context-aware applications and services. |
