With the development of networking technologies and computer graphics, distributed virtual environment (DVE) has become an important research area in computer science. In a DVE system, clients at different places can ...With the development of networking technologies and computer graphics, distributed virtual environment (DVE) has become an important research area in computer science. In a DVE system, clients at different places can simultaneously explore a virtual world and interact with each other. How to build an efficient DVE system that can support large number of concurrent users is a challenge. In the paper, we discuss the way of partitioning the DVE world to balance the workload among different servers as well as minimize the inter-server communication. We present the existent mathematics mode and algorithms at first, and then give our algorithm, partitioning algorithm based on competition. Ex- periments are carried out to illustrate the effectiveness of our proposed algorithm.展开更多
构建了一种能有效描述大规模虚拟环境的场景图和基于该场景图的场景多叉树,以此为基础提出了基于改进的动态二叉树神经网络(Dynamical Binary-tree Based Neural Network,DBTNN)进行场景调度的新方法,并给出了神经网络挖掘装配过程中视...构建了一种能有效描述大规模虚拟环境的场景图和基于该场景图的场景多叉树,以此为基础提出了基于改进的动态二叉树神经网络(Dynamical Binary-tree Based Neural Network,DBTNN)进行场景调度的新方法,并给出了神经网络挖掘装配过程中视点变化的规律。同时,通过该网络输出预测下一步或下几步的视点状态信息,使场景调度具有一定的容错性,提高了场景调度的实时性并使调度更加智能化。最后给出了不同场景的测试结果,表明该算法对于大规模复杂场景有非常好的优化效果。展开更多
文摘With the development of networking technologies and computer graphics, distributed virtual environment (DVE) has become an important research area in computer science. In a DVE system, clients at different places can simultaneously explore a virtual world and interact with each other. How to build an efficient DVE system that can support large number of concurrent users is a challenge. In the paper, we discuss the way of partitioning the DVE world to balance the workload among different servers as well as minimize the inter-server communication. We present the existent mathematics mode and algorithms at first, and then give our algorithm, partitioning algorithm based on competition. Ex- periments are carried out to illustrate the effectiveness of our proposed algorithm.
文摘构建了一种能有效描述大规模虚拟环境的场景图和基于该场景图的场景多叉树,以此为基础提出了基于改进的动态二叉树神经网络(Dynamical Binary-tree Based Neural Network,DBTNN)进行场景调度的新方法,并给出了神经网络挖掘装配过程中视点变化的规律。同时,通过该网络输出预测下一步或下几步的视点状态信息,使场景调度具有一定的容错性,提高了场景调度的实时性并使调度更加智能化。最后给出了不同场景的测试结果,表明该算法对于大规模复杂场景有非常好的优化效果。
