网格环境下基于HLA的自适应仿真研究

Research on Adaptive Simulation Based on HLA in Grid Environment

在面向服务的网格环境下进行基于HLA的分布仿真时,网格环境的变化会使模型服务无法完成联邦成员对象实例的访问,从而导仿真运行终止,故必须在仿真中引入很强的容错机制,使仿真能自适应地运行。为满足网格环境下分布仿真运行的这些要求,先将模型设计成具有自适应能力的网格服务,再对运行支撑的本地组件(LRC)和核心组件(CRC)作服务化的改造,在这两组件间建立了两条基于Http协议的链接。通过为HLA增加几个模型管理接口,在LRC和CRC以及自适应模型服务三者之间实现了信息的互通,这使得模型服务的自适应能力被提升为仿真运行的自适应能力。实验表明,在网格环境下进行的自适应仿真具有了容错能力。

When a distributed simulation based on HLA runs in the Grid Environment, environment changes of grid services may make the model service fail to finish the access of object instances of a federate and result in the simulation termination. A good fault-tolerated mechanism must be introduced into the simulation so that it can adaptively run. In order to satisfy this requirement of distributed simulation in the Grid Environment, the model was firstly designed as a grid service with adaptive capability. Then the local component （LRC） and the core component （CRC） of RT1 were modified using web service technology, and two links based on http protocol were set up between the LRC and the CRC. The two components of RT1 and the adaptive model-service could communicate with each other by adding some model managing interfaces to HLA, which made the adaptive ability of model service change into the adaptive ability of simulation. Experiment indicates that the adaptive simulation in the Grid Environment has fault-tolerated ability.