面向多服务器架构的虚拟角色划分方法研究
发布时间：2018-08-05 15:03
【摘要】：随着计算机和网络技术的飞速发展，分布式虚拟环境技术的研究也逐渐升温。分布式虚拟环境技术就是把虚拟环境和网络技术结合在一起，使得地球上任何一个角落的用户，，都能够通过网络在虚拟世界中协同完成任务。目前，分布式虚拟现实技术已在经济建设、国防安全、文化教育和在线游戏等领域有着深入的研究和广泛的应用。在现今的分布式虚拟环境系统中，需要提供多个服务器组服务于全局的用户集合，服务器为用户提供通信和数据处理服务，而如何把大规模用户集合合理的分配给多个服务器组已成为分布式虚拟环境技术中一个关键的研究方向。 已有的算法大多是从遗传思想、算法效率以及服务器层架构设计等角度给出解决方案，都没有考虑到空间距离这一重要因素。因为在基于虚拟世界中，空间距离较近的用户之间发生通信的概率高，而如果能够把所有较集中的用户划分为不同聚类，然后聚类分配给服务器，那么能够保证整个系统拥有良好的网络质量。因此，空间距离角度的区域划分研究是十分必要的。目前针对空间距离角度来研究区域划分方法的工作尚未展开，在分布式虚拟环境系统中，由于在初始化阶段系统具有数据量大、系统整体等待时间短，而在运行阶段对网络质量要求又较高，所以我们在不同阶段提出了不同的解决方法。 首先在初始化阶段，本文提出了一种基于四叉树概念的区域划分方法。该方法是通过计算用户之间的欧氏距离和构建四叉树模型将所有用户划分到四叉树结果暂存，然后执行分配服务器过程得到算法结果。由于在分配服务器过程中，具有负载均衡特性的服务器集群能够为用户提供更优质稳定的网络服务，所以本文进一步分析了服务器集群的负载问题，提出了改进型分配服务器方法。 其次在运行阶段，本文设计了基于密度聚类的区域划分方法。该方法主要是改进了空间密度聚类算法中经典的DBSCAN算法。在该算法中改进了影响范围值（Eps）和最小用户阈值（MinPts）的计算方法，并且增加了裁剪步用于缩小待处理用户规模。 最后本文分析了以上两个方法的时间复杂度并给出了具体实例演示，模拟实验验证了其合理性与有效性。
[Abstract]:With the rapid development of computer and network technology, the research of distributed virtual environment technology has gradually increased. Distributed virtual environment technology is a combination of virtual environment and network technology, so that users in any corner of the earth can cooperate to complete the task in the virtual world through the network. At present, distributed virtual reality technology has been widely used in the fields of economic construction, national defense security, cultural education and online games. In today's distributed virtual environment system, it is necessary to provide multiple server groups to serve the global user set, and the server provides users with communication and data processing services. However, how to reasonably allocate large scale user sets to multiple server groups has become a key research direction in distributed virtual environment technology. Most of the existing algorithms provide solutions from the perspectives of genetic thought, algorithm efficiency and server layer architecture design, all of which do not take into account the important factor of spatial distance. Because in the virtual world, there is a high probability of communication between users who are close to each other, and if you can divide all the more concentrated users into different clusters and then cluster them to the server, Then can guarantee the whole system to have the good network quality. Therefore, it is necessary to study the regional division of spatial distance angle. At present, the work of studying the method of region division for the angle of space distance has not been carried out. In the distributed virtual environment system, because of the large amount of data in the initialization stage, the whole waiting time of the system is short. However, in the operation phase, the network quality requirements are higher, so we put forward different solutions in different stages. Firstly, in the initialization phase, a region partition method based on quadtree concept is proposed. By calculating the Euclidean distance between users and constructing a quadtree model, all users are partitioned into quadtree results for temporary storage, and then the algorithm results are obtained by executing the allocation server process. Because the server cluster with load balancing feature can provide users with better and stable network service in the process of server allocation, this paper further analyzes the load problem of server cluster. An improved allocation server method is proposed. Secondly, in the phase of operation, this paper designs a method of region division based on density clustering. This method mainly improves the classical DBSCAN algorithm in spatial density clustering algorithm. In this algorithm, the calculation method of the influence range value (Eps) and the minimum user threshold (MinPts) is improved, and the clipping step is added to reduce the size of the user to be processed. Finally, the time complexity of the two methods is analyzed and an example is given to demonstrate the rationality and validity of the two methods.
【学位授予单位】：杭州电子科技大学
【学位级别】：硕士
【学位授予年份】：2013
【分类号】：TP391.9;TP368.5
本文编号：2166155
[Abstract]:With the rapid development of computer and network technology, the research of distributed virtual environment technology has gradually increased. Distributed virtual environment technology is a combination of virtual environment and network technology, so that users in any corner of the earth can cooperate to complete the task in the virtual world through the network. At present, distributed virtual reality technology has been widely used in the fields of economic construction, national defense security, cultural education and online games. In today's distributed virtual environment system, it is necessary to provide multiple server groups to serve the global user set, and the server provides users with communication and data processing services. However, how to reasonably allocate large scale user sets to multiple server groups has become a key research direction in distributed virtual environment technology. Most of the existing algorithms provide solutions from the perspectives of genetic thought, algorithm efficiency and server layer architecture design, all of which do not take into account the important factor of spatial distance. Because in the virtual world, there is a high probability of communication between users who are close to each other, and if you can divide all the more concentrated users into different clusters and then cluster them to the server, Then can guarantee the whole system to have the good network quality. Therefore, it is necessary to study the regional division of spatial distance angle. At present, the work of studying the method of region division for the angle of space distance has not been carried out. In the distributed virtual environment system, because of the large amount of data in the initialization stage, the whole waiting time of the system is short. However, in the operation phase, the network quality requirements are higher, so we put forward different solutions in different stages. Firstly, in the initialization phase, a region partition method based on quadtree concept is proposed. By calculating the Euclidean distance between users and constructing a quadtree model, all users are partitioned into quadtree results for temporary storage, and then the algorithm results are obtained by executing the allocation server process. Because the server cluster with load balancing feature can provide users with better and stable network service in the process of server allocation, this paper further analyzes the load problem of server cluster. An improved allocation server method is proposed. Secondly, in the phase of operation, this paper designs a method of region division based on density clustering. This method mainly improves the classical DBSCAN algorithm in spatial density clustering algorithm. In this algorithm, the calculation method of the influence range value (Eps) and the minimum user threshold (MinPts) is improved, and the clipping step is added to reduce the size of the user to be processed. Finally, the time complexity of the two methods is analyzed and an example is given to demonstrate the rationality and validity of the two methods.
【学位授予单位】：杭州电子科技大学
【学位级别】：硕士
【学位授予年份】：2013
【分类号】：TP391.9;TP368.5
【参考文献】
相关期刊论文 前4条
1 卫宁;蔡鸿明;姜丽红;;DVE系统下基于区域密度划分算法研究[J];计算机仿真;2009年08期
2 杨孟洲,潘志庚,石教英;分布式虚拟现实系统体系结构[J];计算机应用研究;2000年07期
3 赵沁平;;虚拟现实综述[J];中国科学（F辑:信息科学）;2009年01期
4 潘志庚,cad.zju.edu.cn,姜晓红,张明敏,石教英;分布式虚拟环境综述[J];软件学报;2000年04期
相关硕士学位论文 前1条
1 谭阳;关于随机数生成算法的研究[D];湖南师范大学;2008年
本文编号：2166155
本文链接：https://www.wllwen.com/kejilunwen/jisuanjikexuelunwen/2166155.html
