Abstract: Vast amounts of data request and distribution exist in large-scale military simulation, which are managed by the data exchanging bus. The architecture of bus has become more and more complex while the data interacting increasingly. And the management of bus has become an important issue in large-scale distributed simulation. Quality of data interaction is the key to running system smoothly, especially during the real-time, remote and distributed training. This paper carries out prototype system combining with interactive needs for model platform. Under the test for this mechanism in a variety of conditions and interaction request, the results show that mechanism can significantly improve the quality of service interactions in a complicated network environment and various link-types, and then provides a guarantee for model platform to run stably.

Key words: military simulation, data exchange, bus, adaptive perception, load balance

摘要： 大规模军事仿真中存在着海量的数据请求与分发，这些都由数据交互总线进行管理。随着数据交互的不断频繁，总线的架构也越来越复杂，对总线的管理也成为大型分布式仿真系统的重要课题之一。在进行实时、异地、分布的模拟训练时，仿真交互的质量成为系统顺利运行的关键。对总线中服务资源的自适应感知与负载均衡进行了设计与实现。针对总线服务交互请求的不同状态，进行了多种条件下的测试，结果显示该机制在网络环境复杂、链路情况多样的情况下，能够大幅提高服务交互质量，为模型平台的正常运行提供了保障。

关键词: 军事仿真, 数据交互, 总线, 自适应感知, 负载均衡