Research Of Framework Based On Personalized Service Discovery In Pervasive Computing Environments

In pervasive computing, the computing environment for a user is no longer a fixed computer,but a space that indcludes multiple heterogeneous devices that can change dynamically according to user's situation.Managing the space is an essential part of pervasive computing because application services in this environment need to be adaptive to the user's current situation..Now the mainstream Service Discovery Protocols include SLP(Service Location protocol),UPnP(Universal plug and play),Jini and so on.In a pervasive computing environment,handheld devices have limited resources and user interfaces,so the mainstream SDPs have their limits.In pervasive computing environment, service discovery enables devices and services to properly discover,configure,and communicate with each other. In order to provide more appropriate services for users, service discovery protocol can obtain the context information of users by the context analyzer, but existing personalized service discovery requires the user's context information as a query, in this way the privacy of users is not safe. In this paper, I proposed a personalized service discovery framework which aggregates services that contextually close to the user in advance,instead of attatching user context to a discovery query to find appropriate services.So it can protect the privacy of the user. At the same time,the framework can find the service extended to the entire domain,so it can find the most appropriate service to users.The main contents and innovations in this paper are summarized such as:(1)ANFIS(Adaptive-Network-based Fuzzy Inference System) is described firstly.The inference engine in the framework is designed based on ANFIS,it is an important componet of the framework.The inference engine have learning capability,so it can reflect the user's preference.(2) Personal virtual space in pervasive computing environment is proposed.In this scheme all of the objects in pavasive computing is seen as three types of componets,they are virtual objects, services and users.Virtual personal space can effectively manage these components and can include the services the virtual objects meeting the user's context. Personal virtual space is very sensitive to the user's state change and it also supports multi-user.(3)The framework based on personalized service discovery is proposed.It is built on SLP and combined virtual personal space and personalized inference engine. First,I introduce the four components of the framework,then,I describe the flow of the framework and the process of the personalization in detail.In this paper,I define a set of fuzzy inference rules,the inference engine apply it to the service vector which is trasmitted by the personal processor,if the result is greater than the threshold,the service will be included into the vitual space.(4) Finally,in this paper I performed simulation-based experiments.The simulator can emulate three types of management scheme:location-based scheme,quality-based scheme and personalized-based scheme.In order to compare to these schemes,I employed three metrics:discovery time,user satisfactory and user ratio.In the experiments 1 have proved the framework's practicability and validity. Last I have summarized and proposed the prospect of further research.,which lays a solid foundation for further research.The processor will send feedback to inference engine.Based on it the inference engine adjust the range of fuzzy variable.Through the feedback the engine can learn the user's preference.The QoS management in the framework can aggregate the QoS feedback,then recompute the QoS of the service.The component of the QoS management can improve the raliability of the service discovery.