摘要
组播是大规模分布交互仿真中的主要技术之一,仿真中数据的可靠传输要求使可靠组播成为其中的重要问题。在多对多可靠组播RMMS系统研究的基础上,改进TCP单播备份机制,提出了可靠单播恢复丢失的组播备份报文的报文备份机制,消除了TCP备份带来的瓶颈,提高了可靠组播的性能。介绍了可靠组播集成到BHRTI的设计。实验结果表明,集成可靠组播的BHRTI在时延、吞吐量、多对多传输方面具有良好的性能。
Multicast is one of the most important technologies in large scale distributed simulation. Due to the requirement of reliable data transmission, reliable multicast becomes a significant problem to be solved. A new method for packet backup was proposed, instead of the primary "TCP unicast" in RMMS. In the new method, reliable unicast was used to recover the multicast packets which were lost during backup period. The improvement eliminated the bottle-neck owning to TCP connections and improved the performance of RMMS. The design of integrating the improved RMMS system into BH RTI was described. Experiments and result analysis indicate that BH RTI with reliable multicast integrated performs well in latency, throughput and many-to-many transmission aspects.
出处
《系统仿真学报》
CAS
CSCD
北大核心
2009年第16期5065-5071,共7页
Journal of System Simulation
基金
国家自然科学基金(60603084)
国家863计划项目(2006AA01Z331)
关键词
多对多
可靠组播
报文备份
分布交互仿真
BHRTI
many-to-many
reliable multicast
packet backup
distributed interactive simulation
BH RTI