Research On Data Distribution Management Of High Level Architecture

HLA is an IEEE standardized distributed simulation architecture. It provides the specification of a common technical architecture for use across all classes of simulations. It provides the structural basis for simulation interoperability. The primary goal of Data Distribution Management (DDM), is one of six services offered by the HLA, is to control the volume of unnecessary messages exchanged during a simulation, thereby decreasing the workload on the simulation hosts. Various DDM methods accomplish this goal in different ways, but the key to efficient DDM is to limit the messages being sent to include only those messages with state update intended for federates that require them.Data distributed management service is analyzed as a main content in the thesis. The fitness, merits and demerits are summarized in the class-based DDM, the region-based DDM, the grid-based DDM and the mobile Agent-based DDM. The result shows that one of the main factors causing the disadvantages of distributed interactive simulation in cooperation and expansibility is excessively dependence on center-nodes in simulation system, once the center-nodes are down, it connects directly with the continuous operation.To overcome the excessively dependence on center-nodes and make the DDM be more effective and of scalability, and in the light of the P2P network's advantage of economy, scalability and fault tolerance, a peer to peer (P2P) based DDM is proposed, with which the super-nodes are appointed based on the small world phenomena to carry out subscribing or publishing region and realization of further filter mechanism, reduce the load of RTI, improve the fault tolerance, thereby, to achieve the goal of coordinative system. The P2P-based DDM includes the definition of the model, the description and analysis of algorithm. By comparison of the data filtering efficiency and the results of region match among the P2P-based DDM, the region-based DDM and the class-based DDM, simulation test is carried out in NS2. Conclusions for the test can be drawn as follow: under the same environment, the P2P based DDM has better stability and fault tolerance than region-based DDM and class-based DDM. And the P2P-based DDM is more adaptable to the large-scale and complex distributed simulation system.