基于全局共享内存的协同仿真运行支撑  

Runtime support for collaborative simulations using global shared memory

作　　者：刘雁兵[1] 孙宏波[1] 范文慧[1] 肖田元[1] 

机构地区：[1]清华大学自动化系,国家CIMS工程研究中心,北京100084

出　　处：《清华大学学报（自然科学版）》2014年第12期1573-1579,共7页Journal of Tsinghua University(Science and Technology)

基　　金：国家科技支撑计划项目(2012BAF15G00);中国博士后基金项目(2013M540097)

摘　　要：在基于高层体系结构(high level architecture,HLA)的协同仿真中,随着仿真对象数量的增加,联邦成员之间的交互也在增多。在这种情况下,通常的以太网通信往往不能满足仿真成员之间的通信需求。共享内存作为一种高效的进程间通信机制,已经被广泛应用于系统间的通信,然而分布式共享内存对于计算节点间的通信支持尚显不足。该文通过反射内存构建了分布式仿真系统的全局共享内存环境,开发了基于全局共享内存的协同仿真运行支撑环境TH_SHMRTI,设计了相应的RTI(runtime infrastructure)通信机制,能够支持异构操作系统、异构混合平台上联邦成员之间高效的通信。测试结果表明,该通信机制能有效提高RTI在异构平台下的通信性能。In HLA-based collaborative simulations,the interactions between federates increased with increasing number of simulation objects. The conventional Ethernet communications often then cannot meet the demands for communications between members of the simulation. Distributed shared memory is an efficient inter-processes communication mechanism that is widely used for communications among different systems. However,distributed shared memory cannot be easily adapted to computing environments composed of multiple computing nodes.This paper describes a global shared memory（GSM）environment for distributed simulation systems that uses a reflective memory card and the corresponding RTI communications method.A collaborative simulation runtime infrastructure is developed to support the communications among federates residing in different operating systems and heterogeneous computing nodes.Tests show that this mechanism promotes RTI communication performance with heterogeneous platforms.

关 键 词：高层体系结构 运行支撑环境 全局共享内存 反射内存卡 异构平台 

分 类 号：TP391.9[自动化与计算机技术—计算机应用技术]