引用本文: | 梁洪波,朱卫国,姚益平,等.一种面向大规模HLA仿真的并行区域匹配算法[J].国防科技大学学报,2013,35(3):84-91.[点击复制] |
LIANG Hongbo,ZHU Weiguo,YAO Yiping,et al.A parallel region matching algorithm for large scale HLA simulation[J].Journal of National University of Defense Technology,2013,35(3):84-91[点击复制] |
|
|
|
本文已被:浏览 7522次 下载 5449次 |
一种面向大规模HLA仿真的并行区域匹配算法 |
梁洪波1,2, 朱卫国1, 姚益平2, 刘步权2, 慕晓冬3, 杨勇2 |
(1.92117部队,北京 100072;2.国防科技大学 计算机学院, 湖南 长沙 410073;3.第二炮兵工程大学, 陕西 西安 710025)
|
摘要: |
HLA仿真中,数据分发管理实现基于值的过滤,可以有效减少盟员接收冗余数据的可能性和网络中的数据流量。大规模HLA仿真系统在仿真推进中需要大量的区域匹配计算以维护数据分发管理的正确性。现有的区域匹配算法大多需要对所有区域进行匹配计算,造成了大量计算资源的浪费;同时,主要基于串行匹配思想,难以充分发挥多核平台的并行计算优势。针对现有区域匹配算法的局限性,提出了一种面向大规模HLA仿真的并行区域匹配算法,该算法能够实现对一次仿真推进中多个改变区域的并行匹配计算,同时在匹配计算中采用基于移动相交的基本思想,利用区域范围移动前后的历史信息,将匹配限定在移动区间之内,减少了大量的无关计算。理论分析与实验结果表明该算法尤其适合基于多核计算平台构建大规模分布式仿真的应用需求。 |
关键词: 分布式交互仿真 运行支撑环境(RTI) 数据分发管理(DDM) 区域匹配 并行计算 |
DOI: |
投稿日期:2012-10-20 |
基金项目:国家自然科学基金资助项目(61170047) |
|
A parallel region matching algorithm for large scale HLA simulation |
LIANG Hongbo1,2, ZHU Weiguo1, YAO Yiping2, Liu Buquan2, MU Xiaodong3, YANG Yong2 |
(1.Unit 92117,Beijing 100072,China;2.College of Computer, National University of Defense Technology, Changsha 410073, China;3.Second Artillery Engineering University, Xi’an 710025, China)
|
Abstract: |
In HLA simulation, data distribution management (DDM) service provides data filtering capability, which can effectively reduce the transmission and reception of redundant data. During the advancing of large-scale HLA simulation system, much region matching calculation is needed. However, in most of the existing regional matching algorithms, all regions are calculated in every matching calculation process, which results in a waste of computing resources. Meanwhile, the current matching algorithms are mostly based on serial region matching, and the advantage of multi-core cannot be fully taken. Aiming at the limitation of existing algorithms, a parallel region matching algorithm for large-scale HLA simulation is proposed. By this algorithm, the matching process of multiple regions which changes in the same simulation advancing can be calculated in parallel, the advantage of parallel computing environment can be taken. During the process of matching calculation, a method which is based on intersecting information was used. By the history of moving regions, matching calculation only happened within a limited range, and the redundant calculation could be avoided. Theoretical analysis and experimental results showed that the algorithm is particularly suitable for large-scale distributed simulation applications which execute on multi-core computing platform. |
Keywords: distributed interactive simulation runtime infrastructure data distribution management region matching parallel computing |
|
|
|
|
|