The Research And Implementation Of Data Distribution Management In HLA

High Level Architecture which is proposed by Modeling and Simulation Office of the United States is the latest simulation frame for reuse and interoperation of the complex large-scale simulation. High Level Architecture has five features:open, extensible, interactive, distributed and reusability. The framework of High Level Architecture is accepted by International Electrical and Engineers as the standard—IEEE 1516. Data Distribution Management, One of six services, develops to provide scalability and reusability. Data Distribution Management is responsible for limiting and controlling the data exchanged during a simulation, in order to decrease the network width and the required computer resource, to improve the efficiency of simulation system, to achieve the simulation system scalability requirements. It also aims at reducing the processing requirements of simulation hosts, or federates, by communicating updates regarding intractions and state information only to federates that require them.The contents and results of paper include:(1) Discusses the concepts and components of High Level Architecture, current research status of High Level Architecture and Data Distribution Management, the application of High Level Architecture; and also discusses the concepts and principle of Data Distribution Management, analyses Region-based and Grid-based Data Distribution Management methods in detail.(2) Analyses the principle of sort-based DDM. Based on the disadvantages of the overlapping information matrix with vast storage space and the overlapping computation with large number of multi-cycles, a new and improved sort-based DDM algorithm is proposed. The proposed method only needs one M*N matrix storing overlapping information. It reduces the number of multi-cycles and doesn't need an "AND" operation. The proposed method is proved by experiments to be better than the original method in the storage space and time performance. It is suitable for an extended number of regions in large-scale distributed simulation system.(3) Experiment results show that each DDM performance is affected by the attributes changing of the simulation application scenarios. Through setting parameter which effect DDM method executing performance, comparing DDM executing time and matching precision of region-based DDM, grid-based DDM, sort-based DDM, and analysis the Experiment results, we summarize different scene property parameters which affected different DDM methods. In the end, we proposed the adaptive selection strategy of DDM method. We can the statically or dynamical to select different DDM methods in the simulation application run before or simulation operation processing, in order to improve the overall performance of the simulation system operation.