Research On Wireless Sensor Network Middleware For Unibqitous Computing

Recent years have witnessed the rapid progress of the theory and technology of the ubiquitous / pervasive computing,and the wireless sensor network.Sensing-based ubiquitous computing applications,that make use of the wireless sensor network as their supporting information collection and dissemination infrastructure,play more and more important roles in our lives.These applications facilitate the integration of and co-operation between the virtual world and the physical world.However, developing such kind of application is not a trivial task.On one hand,ubiquitous computing applications have high requirements on the flexibility,re-usability, reliability,etc.On the other hand,WSN is constrained by its limited resource, dynamic topology,low level and tedious embedded programming primitives. WSN-middleware oriented towards ubiquitous computing provides a potential solution to resolve the above contradiction.Although in traditional distributed and network environment,middleware has been well studied for many years,but the solutions can not be directly adopted in the new environment.The research on WSN middleware is still in its infancy.The interweave characteristics coming from both the physical world and the virtual world raise many challenging issues in the research.In this dissertation,we address the following key issues in the design of WSN middleware for of ubiquitous computing applications.The first issue is about providing system abstraction to the application programmer,to help shield the complexity and heterogeneity of the WSN hardware and the network,so as to make it possible for them to concentrate on the application logic without the need to have much knowledge on the lower implementation details.The second issue it about providing important and reusable middleware functions and services to ubiquitous applications,so as tO easy the development and deployment process,as well as to manage the resource and improve the predictability of the applications.The third issue is about providing domain dependent services to facilitate real world application development. To address the first issue,we propose the Ubiquitous Intelligent Object(UIO) model,which extends of the traditional Object Oriented(OO)concept to model and program the intelligent objects in the ubiquitous computing paradigm,integrating the characteristics from both the physical world and the virtual world.We define the abstract data model,abstract relationship model,abstract operation model,and abstraction ability model for UIO.We also discuss the implementation techniques for UIO and the methodologies for developing UIO-based applications in the ubiquitous intelligent environment(UIE).The UIO model shields the complexity of the lower infrastructure,and suits the development of both simple and complex ubiquitous applications.This conceptualization not only facilitates the study on the generic characteristics of UIO,but also facilitates the study on the collaborations between them.UIO model forms the foundation for efficiently designing and developing novel ubiquitous applications.In fact,UIO model as a universal intelligent object model can be used as programming and system abstraction for various ubiquitous computing middleware.For example,WSN,WSAN(Wireless sensor and actuator network).In this dissertation,we focus on the use of UIO in WSN middleware.To address the second issue,we propose middleware service for distributed event processing based on UIO collaboration.Events incIuding application events and system events are referred to network and system conditions that trigger system responses and operations in event-driven ubiquitous computing applications.Thus, event service is one of the basic services in WSN middleware.In this dissertation,we propose a WSN event and data service framework based on the Dempster-Shafer(DS) theory.Based on this framework,we develop a series of algorithms,namely the distributed event region detection algorithm(DERD),the distributed point event detection algorithm / distributed fault sensor detection algorithm(DFSD),distributed event boundary detection algorithms(DEBD)algorithm.The simulation results show that these algorithms perform better than the existing works in terms of accuracy and fault tolerance.At last,to address the third issue,we describe two domain specific services and applications that are designed and developed based on our proposed framework.We describe the model,the system architecture,and the algorithms for them.One application is the WSN-based Structural Health Monitoring,and another is the Ubiquitous Searching.The first one demonstrates the effectiveness of the collaborative distributed event processing,and the second demonstrates how the UIO abstraction model can help facilitate new and exciting application development.We have developed system prototypes for these applications,using Berkeley motes and our self-made TFAD901 nodes.We have also carried out simulations,emulations,as well as lab testing and in-field deployment to evaluate the effeteness of those systems. We describe and analyze the evaluation results.