基于Agent的分布仿真中层次时间管理机制的研究与实现

Study and Implementation of a Hierarchical Time Management Mechanism in Agent-Based Distributed Simulation

基于Agent的分布仿真是当前国内外研究的热点,而时间管理是其中的关键技术,也是影响系统性能的主要因素。在大规模的基于Agent的分布仿真中,数目巨大的 Agent给时间管理带来了挑战,本文提出了一种层次的时间管理机制,它采用了局部仿真引擎和全局仿真引擎来进行时间管理,通过增加结点内的计算开销来提高时间管理中的并发性、减少网络传送开销,从而提高仿真的性能。最后对该机制进行了性能评测。

Agent-based distributed simulation is an active research area currently. Among this area, time management is not only one of the most important technologies but also the key factor affecting system performance. However, there are a great number of Agents which challenges the efficiency of time management. In this paper, a hierarchical time management mechanism is proposed and it is composed of Local Simulation Engines and Global Simulation Engine. It improves parallelism in time management and reduces network communication overhead by increasing local computation overhead in each node, which helps to enhance the performance of simulation. Lastly, the performance of this mechanism is analyzed and tested.