Research On Shared Mechanism Of Context Information In Ubiquitous Environment

It is inevitable that the distributed mobile computing will evolve to the ubiquitous computing gradually whose final goal is fusing the information space composed of computers with the physical space in which the people are working and living in.Therefore,the people can get information service transparently according to their personal needs whenever and wherever they are.However,the ubiquitous environment keeps changing all the time and we must adjust the systems according to these changes to satisfy the needs of the users and utilize the systems better.So the ubiquitous applications must be context-aware so they can adjust their actions according to the changes and the history of the contexts automatically.Distributivity of the ubiquitous environment determines that plenty of context resources are ingrated into the environment. As a communication platform for compoments in context-aware environments, the context-aware platform provides channels and approaches for context-aware applications'getting context information. In the large distributed context-aware platform, the problem should be faced that how to make good use of the widely distributed context resources and make more and more context-aware applications get the resources timely. Howerver, traditional context-aware systems focus on studying local problems of context-aware technology, such as modelling, fusion and reasoning for context information, or supporting mobility of compoments, rather than scalability of the context-aware platform, which has difficuty in adjusting to the large ubiquitous environment. Therefore, this paper, mainly studies scalability of context-aware platform in the large ubiquitous environment, focuses on solving scalability in architecture of context-aware platform and delivery of context information. The latter contains that location-sensitive composite subscription for context events and delivery of result streams of subscriptions for context events.The main contents of this paper are composed of three parts: the first one pays attention to build distributed P2P platform to provide timely and low traffic burden context data-shared approach for mobile applications in large ubiquitous environment; the second one studies how to reduce matching between context data and data receivers and save monitoring cost of sensors; the third one focuses on how to make full use of intersections among result streams of context subscriptions to avoid redundant comsuming of network bandwidth in common delivery pathes. Including:1.Based on the context-aware middleware Context Toolkit, this paper improves the centralized P2P architecture to a scalable distributed P2P architecture. Besides, we provide dynamic registration approach and three-step subscription approach: the former not only provides a lightweigh way for mobile applications'getting ambient context data, but also make sure freshness of received context data to some extent. The latter is able to help reduce traffic in remote delivery of context data.2.A location-sensitive composite subscription for context event mechanism is presented. It contains a location-sensitive composite subscription model and a dynamic binding approach: the model utilizes finite automata verification to make sure behavior consistancy between atom subscriptions and composite subscriptions, between applications and sensors; in order to adapt to dynamics in ubiquitous environments, the dynamic binding approach accompolishes auto mapping between event sources and subscriptions, which makes the matching between events and subscriptions translate into the matching between event source and subscriptions. Hence, the approach is able to reduce matching efficiently and improve system performance.3.In a large distributed context-aware platform, result data streams of different subscriptions often include the same contents. To achieve the goal that delivey of the same data content in common paths is only once to the full extent, this paper utilizes the approach of mutiple subscriptions grouping to a representive query to reduce traffice comsuming. Besides, considering the continunity of context data, this paper presents a list based on sequence in which context data sent. Every list node is tagged by data range and corresonding query set, which constructs local access for context data. Experiments show that the query grouping approach based on the list makes the maintaining cost be low and the ratio of traffic saved in common paths is satisfying.