摘要

协同虚拟地理环境作为一个新的研究方向，其多维多感知的虚拟环境更加符合人类空间认知特点，分布式的系统结构又为专家、政府人员以及民众等提供了交流、研讨甚至决策的理想平台，协同工作机制则保证了交流交互过程中的有序性。这些基本特性使得协同虚拟地理环境自然成为虚拟地理环境发展的重要方向。与一般虚拟环境所具有的沉浸感、交互性、信息集中和智能性四大特性相比，协同虚拟地理环境还具有分布式、共享性、多用户参与、协同性等特性。

多用户的协同机制，包括多用户之间的通信、数据、计算协同控制等是目前研究的焦点问题。目前国内外对协同虚拟地理环境的多用户参与协同所涉及的一些问题（如开放系统结构、分布资源协同、多用户协同控制机制以及网络瓶颈限制等）未进行系统分析和提出切实可行的解决方案。本着学科交叉的思想和原则，借助其他领域的技术和方法，本文首先建立协同虚拟地理环境的多Agent系统和多用户协同控制模型，并设计动态群组算法来求解协同任务，接着分别从数据协同、计算协同、实时协同环境几方面开展深入研究，最后结合一个应用案例建立原型系统并验证了上述理论和技术方案的可行和正确性。论文主要研究工作如下：

1）由于多用户可能处于各种异构的环境下，且具有动态加入和离开系统环境的特性，为了使协同交互控制更加灵活有效，在分析现有协同模型的基础上，本文建立协同虚拟地理环境的多Agent系统，并利用Agent之间信息交互集中和对等通信两种模式的优点，设计了混合协同控制模型，即部分Agent间可以直接通讯，部分Agent间通过转发进行通讯，从而既能够减轻服务器的负担，也能实现高效的资源管理和搜索，以适用复杂的应用环境。此外本文也设计一个动态群组算法来组织和控制不同的Agent协同完成任务。

2）针对异构数据源和异构地理信息处理环境下的数据共享问题，本文对协同虚拟地理环境内常见的数据类型和特点进行分析，进而基于混合协同控制模型，采用 Agent 技术，开展协同功能模型设计、数据访问并发控制、多服务器协同、以及数据搜索机制研究，建立虚拟地理环境数据协同服务，把整个系统内部可以共享的数据管理起来，逻辑上整合在一起形成一个虚拟数据库，提供给用户使用，实现异构数据的互连、互操作和协同工作。

3）针对集中服务器模式容易受到带宽、时间及流动性的限制，从而影响数据传输效率的问题，本文根据协同虚拟地理环境多用户参与特点，分析了协同虚拟地理环境主要数据类型的传输特点，针对异步同时协同模式对数据协同的要求，基于混合协同控制模型和对等服务思想，设计了两种数据传输模型，一是数据分发模型，解决地理场景等大数据量的静态数据分发问题；一是即时数据传输模型，通过建立多播树来处理协同工作中即时产生的交流交互动态数据，通过数据中转来减轻数据服务器负载和降低带宽限制的影响。

4）针对协同虚拟地理环境内的资源模型集成和开放系统结构问题，论文利用网格技术建立协同虚拟地理环境服务体系框架，并设计了其运行机制和逻辑任务管理流程，以实现异地分布的数据资源、地学应用模型、计算资源和信息共享，增强协同虚拟地理环境服务能力，提高系统的可扩展性。并在此服务框架下，以分布式计算和移动Agent计算为例，探讨了计算资源注册管理、模型服务部署、路由控制管理、运行流程等计算服务的关键问题，最后通过实验验证该方案的可行性。

5）针对协同虚拟地理场景的建立，论文首先从场景对象、场景建模技术、场景优化管理对三维地理场景建模进行总体讨论，并以分布式地形实时漫游为例，对地形数据预先处理、数据流程、简化模型等关键技术进行探讨，并利用数据分发模型，实现多用户参与的虚拟地形环境实时漫游和协同交互。同时也讨论了协同虚拟地理环境中的多用户交流交互模式，侧重混合协同控制模型和即时数据传输模型在多用户视频协同中的应用。

6）从技术实现上，论文以小流域协同坝系规划为研究背景，设计了基于多Agent系统的原型系统功能体系结构，同时分别针对地理场景绘制和协同、视频协同、网格服务、移动计算服务、分布式计算服务选择相应的技术实现方案，最后选择了Globus ToolKit、Voyager、Jini和ComputeFarm等技术和软件开发包，建立原型实验环境，分别对论文的理论模型和实现机制进行阐述说明，并进行初步实验和效率分析。

关键词：协同虚拟地理环境，多Agent系统，混合协同控制模型，动态群组算法，数据协同服务，网格服务，分布式计算服务，移动Agent计算服务

Abstract

As a new research direction, collaborative virtual geographic environment (CVGE) has some essential characteristics. Multidimensional and perceptive virtual environment accords with people’s cognition characteristic, distributed system structure can offer experts, officers and citizens a intercourse, discussion and decision-making platform, and cooperated mechanism can make sure the orderliness in intercourse process. All of these features make CVGE naturally become an important research direction of geographic science. Besides four basic features, namely immersion, interactivity, information intensity and intelligent objects, CVGE also owns other characteristics such as distribution, share, collaboratiion and multiuser.

Collaborative mechanism among multiuser, including communication, data, and computing collaboration, has been becoming one of research’s key issues. Now some key problems of CVGE such as open system structure, distributed resource collaboration, cooperative control mechanism for multiuser, and bottleneck restriction of network bandwidth, have not been discussed in detail and a practical feasible solution has not been fully implemented. According to interdisciplinary method, this paper firstly designs an agent-based hybrid cooperative control model and dynamic group algorithm for solving collaboration task. Based on MAS, CVGE system architecture and some key problems, such as dynamic group algorithm and concurrency control, are also discussed. Meanwhile data collaboration, computing collaboration and real time virtual environment collaboration are discussed in detail. Finally, a prototype system is developed and the experimental results prove that the scheme addressed in the paper is feasible and efficient. The major context of the paper is listed as follows:

1) CVGE users’ tools and platforms are heterogeneous, and they may intermittently join or leave system environment. In order to make interactive control more flexible and efficient, the paper firstly builds up CVGE multi agent system after analyzing existing cooperation models. Then the paper makes use of some merits of centralized and decentralized communication pattern of among agents and designed hybrid cooperative control model, which some agents can communicate directly, while others agents transfer information indirectly. This model can reduce server’s load, and implement resource managing and searching rapidly, and meet needs of complicated application. In addition, the paper also designs a dynamic group algorithm to organize and control different agents to finish collaborative task.

2）How to share different data source under heterogeneous geographic information handling environment for solving complex geographic problem is very important. Some typical CVGE data types and characteristics are firstly analyzed. Then the paper builds up data cooperative service based on agent technology and some key problems such as cooperative function model, data access concurrency control, multi server strategy and data search mechanism, are also discussed. All efforts aim to integrate dispersed data in system and form a logical virtual database, which offers a simple and convenient port to access.

3）Some traditional network structures, such as browser/server and client/server, emphasize the function of the server, which results in some restrictions on the bandwidth, performance in real-time and mobility. In order to improve data transmission efficiency, we analyse CVGE data types and transmitting features and design two kind data transmission models based on coordinative service ideas, which balance server load and reduce bandwidth restriction. One deals with handing out large amounts of static data geographic scene data, another one builds up a multicast tree to transfer real-time data for large-scale users in a collaborative group.

4）In order to integrate resource models and establish open system structure, we build up a CVGE service system framework based on Grid, and design its running mechanism and logic task management flow. Moreover, distributed computing service and mobile agent computing service are also implemented under our service framework. Some key problems, such as computing resource registering management, model service deploying, routing control management and running mechanism design, are also discussed in detail. Finally some experiments are implemented to prove the feasibility of our scheme.

5）As collaboration geographic scene is one of important components in CVGE system, several key problems such as scene objects, scene modeling technologies and scene optimizing management are discussed in detail. For real-time rendering of large-scale terrain under web-based distributed environment, some key technologies, such as data download model, data preprocess, data display stream, simplification algorithm, are also discussed to implement multiuser participation and collaborative intercourse. Meantime, interactive patterns for multiuser in CVGE are also discussed, and the paper focuses on how to apply hybrid cooperative control model and real-time data transfer model to video collaboration.

6）Taking collaboration dam planning as research background, the paper designs the function architecture of a prototype system based on MAS, and chooses technical implementation methods for geographic scene rendering and collaboration, video collaboration, Grid service, mobile agent computing service and distributed computing service. Some software toolkits including Globus ToolKit, Voyager, Jini and ComputeFarm, are used to establish prototype experiment environment, which can explain and implement our models and methods. Some experiments are implemented to show that the scheme addressed in the paper is efficient and feasible.

Keywords：Collaborative Virtual Geographic Environment; Multi Agent System; Hybrid Cooperative Control Model; Dynamic Group Algorithm; Data Collaboration Service; Grid Service; Distributed Computing Service; Mobile Agent Computing Service

摘 要

Abstract

摘要