协同虚拟环境下的自适应分区方法

A New Self-Adaptive Approach for Optimizing both Speed and Quality in Partitioning Algorithm in CVE Systems

协同虚拟环境是一个为多用户提供协同操作的共享虚拟空间,为了满足大量用户同时参与系统进行协同并发操作,使协同虚拟环境具有可伸缩性的特点,通常采用基于混合模式的多服务器体系结构,将虚拟环境中的所有计算负载平均分配到多个服务器中,并使服务器之间的通信量最小化。文中定义了基于系统响应特性的分区评价函数,并提出了一种自适应分区方法,将虚拟环境划分成单元格,并给出聚合度的定义,通过选择聚合度大的单元格作为区域种子,使用区域生长法对虚拟环境中的用户进行划分。实验证明,该方法具有较好的执行效率,且分区评价函数值低于其它算法,更接近于最优分区策略。

Aim. Existing partitioning algorithms in CVE （Collaborative Virtual Environment） systems based on multiple-server architecture are, in our opinion, somewhat deficient either in speed or in quality. We now propose a new algorithm that strikes a proper balance between speed and quality. In the full paper, we explain in detail our new algorithm; in this abstract, we just add some pertinent remarks to listing the three topics of explanation. （1） region management model; the optimal partitioning method should assign the workload efficiently among different servers and minimize inter-server messages; （2） region quality evaluation function; a new quality evaluation function based on the system response characteristics is used to evaluate the partitioning method; （3） self-adaptive division into regions; in the self-adaptive partitioning approach, the aggregation degree function of the cell is defined; the virtual environment is divided into cells and those cells that have the larger aggregation degree values are chosen as seeds and our algorithm based on regional growths uses the seeds as growing points to partition the virtual environment. The experimental results, given in Fig. 4 and Table I in the full paper, show preliminarily that our self-adaptive partitioning approach does strike a proper balance between speed and quality.