【作者】 陈泽强；

【导师】 狄黎平； 陈能成；

【作者基本信息】 武汉大学 ， 地图学与地理信息系统， 2012， 博士

【摘要】 传感网(Sensor Web)是分布式网络环境下传感器的协同观测、信息融合和在线服务的新方法。开放地理信息联盟(Open Geospatial Consortium, OGC)制定了传感网实现(Sensor Web Enablement)标准,提出了传感器、观测资源和传感器网络服务等信息资源共享与互操作的标准规范。由于传感器多样、数据多样和服务多样,如何方便地动态地获取传感网数据,有效地处理数据,以及组合传感网多服务,是传感网应用和研究面临的问题。针对这些问题,本文提出了传感网服务整合框架和描述模型,研究了整合框架中若干关键技术。本文主要研究内容和成果包括以下几点：1.简述了传感网整合的研究背景,国内外研究现状,存在的问题和研究的意义。2.介绍了现有整合技术基础,地理空间数据整合技术和传感网整合基础。简述了地理空间数据整合技术中的整合机制、整合流程和分布式整合方法。3.提出了传感网服务整合框架和描述模型。研究了框架层次、协议、功能和流程,定义了描述模型的表现形式,说明了描述模型整合方法。研究了框架整合功能的关键支撑技术：传感网服务柔性实现技术,分布式传感网数据处理技术和传感网多服务组合技术。4.传感网服务柔性实现技术主要解决传感网服务可配置易扩展等问题,实现多源数据动态获取与访问,以及柔性数据与规划服务。柔性服务是指在目标服务框架不变的情况下,针对具体应用、具体任务和具体处理等情况方便开发具体服务实例。研制了基于工厂模式方法的柔性传感器观测服务(SOS)和传感器规划服务(SPS)实现,以及基于抽象重写方法的柔性网络处理服务(WPS)实现。引入虚拟传感器概念,即虚拟传感器是具有数据提供能力的非实体传感器实体,充实传感器概念,扩大传感器应用范围,实现多源观测数据统一访问。整合柔性SOS和SPS,提出了柔性数据和规划服务框架。5.分布式传感网处理技术主要解决传感网在分布式环境下有效处理的问题。提出了WPS耦合分布式处理环境的架构,并以Apache Hadoop为例,阐述了WPS和分布式处理环境的具体耦合方法。6.基于工作流的传感网多服务组合技术解决了传感网多服务一站式整合问题。提出了基于工作流的传感网多服务组合框架。分析了数据在传感网多服务间的流动方式和工作流在执行引擎中的实现。7.应用本文提出的传感网服务整合框架,实现了传感网下视频变化检测和目标跟踪系统,以及动态作物生长监测系统。研究了每个应用系统的整合框架,基于对象模型的整合框架实现,以及应用结果。前两个应用实现了柔性服务,多源数据的动态统一访问,后一应用实现了WPS和分布式处理环境的耦合。结果表明,本文提出的传感网服务整合框架、模型以及若干技术是可行的。更多还原

【Abstract】 Sensor Web is a new method for collaborative observation, information fusion, and online service of sensors in a distributed network environment. Open Geospatial Consortium (OGC) has developed Sensor Web Enablement for the specifications of information sharing and interoperation between sensors, observation source, and sensor network services. For the diverse sensors, observation data, and services, Sensor Web suffers problems that how to conveniently and dynamically obtain sensor data, effectively process data, and combine multiple services. To address these problems, a unified integration-and-processing framework and a describe model of service, as well as key technologies are proposed. The major research activities of this paper include the following:1. Research background, and review of research on Sensor Web at both home and abroad are introduced. After analyzing the existing problems, the significance of this study is put forward.2. This paper introduces the existing basis of integration and processing technology, including geographic spatial data integration and processing technology, and the Sensor Web specifications. The integrated mechanism, process, and distributed integration and processing method of geographic spatial data integration and processing technology are detailed.3. The integration-and-processing framework and the describe model of Sensor Web are proposed. The layers, protocol, function, and workflow of the integration-and-processing framework are studied. The formalization of object-oriented description model for Sensor Web is defined. The integration-and-processing method of description model is explained. The key technologies of integration-and-processing framework for Sensor Web are studied; the technologies are flexible Sensor Web service technologies, processing technologies of distributed Sensor Web data, and combining of multiple Sensor Web services. 4. The technologies of flexible Sensor Web service mainly design configurable and extendable Sensor Web services to realize the access of multiple source-data and flexible data and planning service. Flexible Sensor Web service means it is easy to develop service instance in any application/task/process and the framework itself is not changed. Develop SOS and SPS based on abstract factory pattern, and WPS with abstract rewrite method. Introduce the concept of virtual sensor. Virtual sensor is a non-entity sensor which can provide data to enrich the scope of sensor, to expand the range of applications and to help the unified integration of multi-source observations. Integrate flexible SOS and SPS to form the framework of flexible data and sensor planning service.5. Distributed sensor data processing technology can solve the problem of efficient processing in distributed Sensor Web environment. A framework of WPS coupling with a distributed processing environment is proposed. Take Apache Hadoop as an example, and illustrate the method of coupling WPS Hadoop.6. Workflow-based multi-sensor web services composition technology can solve the integration of one-stop multi-sensor web services. The combining framework of multiple Sensor Web services based on workflow is proposed. Analyze the realization of data flow in the multi-sensor web service, and the execution of workflow in execution engine, as well as the ability to integrate multi-sensor web services under different conditions.7. In order to verify the integration-and-processing framework and models of Sensor Web services, a video change detection and object tracking under Sensor Web, and dynamic crop growth monitoring experiment were designed. And these experiments are analyzed from three aspects:the design of integration framework, the implement of framework based on object model and results. The first two experiments verify flexible services, dynamic uniform access for multi-source data and the object model in integration. The last experiment verifies the coupling between WPS and distributed processing environments can achieve efficient computing. The results show that the proposed integration-and-processing framework, model and key technologies for the Sensor Web are feasible.更多还原