基于交互优先算法的网格仿真实体静态调度策略

为了解决网格仿真中的静态任务调度问题，提出一种创新的基于交互优先算法的实体调度策略．该策略使用了一种兼顾通信优化和计算平衡的评价标准．调度策略分为聚类和映射两阶段进行．聚类过程使用一种交互优先算法，将交互密集的实体聚类到一个实体组中并且将被调度到同一处理器上运行；映射过程则使用一种计算优先的启发式算法，优先映射计算消耗大的聚类实体到处理能力强的处理器上．最后，将使用上述两种算法的调度策略同贪婪对分法进行对比实验，结果证明本文的方法能够较好地优化仿真中的通信和计算性能并且更加适合于网格仿真．

Interaction Priority Algorithm Based Entities Static Scheduling Strategy for Simulation over the Grids

In order to solve the problems of static tasks scheduling for simulation over the Grids, a novel interaction priority algorithm based scheduling strategy for entity level tasks was proposed. A criterion trade off communication optimization and load balance was used in the scheduling strategy. The scheduling strategy has two phases: aggregating and mapping. The aggregating phase is implemented by an interaction priority algorithm. The algorithm partitions entitied according to their interaction frequency, namely the entities with high interaction frequency are aggregated into one group and would be mapped onto the same processor to be simulated. On the contrary, the mapping phase used a computing priority algorithm. It'was a heuristic maps an aggregated entity with most computation cost onto a machine with most computation capacity. After the two algorithms are presented and analyzed, the scheduling strategy was tested and compared with the bisection greedy algorithm. The results showed that the proposed strategy satisfied communication optimization and load balance effect, and it was more adaptive for simulation tasks over the Grids.