A two-stage allocation optimization in static load balancing of HLA distributed simulation

The load balancing problem is becoming the main technical obstacle in HLA distributed simulation development with the increasing scale of distributed simulation system. The solution of task-federate mapping and federate-processor mapping used to be determined based on experiences and the entity is taken as a task granularity in simulations in order to simplify the model in previous studies. Neither the experience methods nor the previous studies could solve the huge communication overhead or load imbalance problem effectively in large scale distributed simulation. The static load balancing optimization of HLA distributed simulation is divided into task-federate mapping and federate-processor mapping, and a two-stage mapping allocation optimization method is proposed in this paper. Decreasing communication overhead of the system is the main mission in the task-federate mapping stage and the load balancing is studied in the federate allocation stage. The effectiveness of the proposed method, which developed a new direction for static load balancing distribution optimization in distributed simulations, is verified with experiments.