Research On Adaptation In Pervisive Computing Environment

Users can access services and share information at anytime and anywhere in pervasive environment. According to different layer's requirements, several complete and mature research fields in pervasive computing come into being. Context-aware computing is one of famous research fields.Context-aware computing is one of key issues in current research. In dynamic, heterogeneous and distributed environment, the inner complexities in context-aware computing present challenges for the software infrastructure of context-aware computing. Context-aware computing has wide applications,such as, Guides system for travel, Assistant software for offices and intelligent community,which improves research on context-aware computing.Current research on context-aware mainly focus on context acquirement, context modeling, and context inference. However, they have some limitations. For example, current context-aware middleware do not pay enough attention on context conflict problem. Moreover, they lack an approach to deal with service selection problem which have multiple objectives and global limitationconditions. So, this paper's major works are based on the new appearing context-aware applications, present related approaches to deal with some limitations of current context-aware systems. The major works of this paper are listed as follows.Design a context-aware middleware and implement a prototype system. It not only supports the context acquirement and context modeling, but also provides context conflict detection and service selection so that it facilitates the development of context-aware applications.Present a logic-based approach for context conflict detection. The advantages are as follows. This approach gives a clear classification of context conflict so that basic context conflict can be calculated from the view of semantics of rules as well as from the view of business logic. Using inferring axioms, it combines two views together, and complete context conflict can be detected. Furthermore, this approach is based on first order logic so that it is helpful for inference and has scalability to some extent.Present a genetic algorithms based approach for services selection. This approach overcomes the two limitations of local optimization. That is, local optimization method can not guarantee the global optimization solution and local optimization method can not solve the problem that contains the global limit conditions. Moreover, this genetic algorithms based approach can provide a set of candidate solutions to meet user's needs. Experiments show that this genetic algorithms based approach can achieve higher Quality of Services than typical localoptimization methods.Base on works of this paper, several future works may be carried out. Design methods to determine precision of the context conflict detection. Present more accurate approaches to resolve context conflict.