The Run-Time Infrastructure (RTI) is a software infrastructure implements the interface specification for High-Level Architecture (HLA), which provides efficient data exchange to logical groups of federates. At the same time, it supports a variety of time management policies and coordinates the exchange of events between federates. RTI should be designed delicately to meet simulations of several different types of system, and to keep high performance of HLA-based application systems. There is a great difference between real time platform simulation and discrete-event simulation, because they are in two separate fields. Compared with what happens in the simulation based on DIS, the effort to combine those two types of simulations and form a larger scale HLA-compliant simulation will result in whole performance loss. Further more, since HLA is designed toward standardization in the M&S community and to facilitate the reuse of M&S components, it's hard for RTI designer to strike a proper balance between increasing complexity of framework and reusability. Finally, in a distributed simulation environment, the system resource will be consumed rapidly when the system extends to a large scale. However, we should try to lower the network bottleneck and the system cost from the point of view of a designer. We have been investigating design and implementation alternatives version of RTI and relevant toolkit to be used with distributed virtual simulation, and our prototype of RTI, namely AST-RTI, complies with HLA Interface Specification (IF) version 1.3. The paper explores the possibility to improve the performance of RTI and reduce the system cost, and mainly discusses the layered hierarchical software, as well as the way to optimize the data structure and the network.
