A Scalable HLA RTI System Based on Multiple-FedServ Architecture

A scalable and high performance RTI (Runtime Infrastructure) system implements a two-layer architecture to supporting large-scale simulation is proposed in this article. The two-layer architecture, Multiple-FedServ, exploits both centralized and distributed way to manage a simulation. In the first layer, each FedServ is in charge of a number of federates and in the second layer, a simulation federation is then formed by all the FedServs. This paper describes how the messages are routed and synchronization performed in this two-layer architecture. An RTI system based on Multiple-FedServ architecture and follows the specification of IEEE 1516 standard is implemented. Performance evaluations of this RTI using standard HLA/RTI benchmarks are presented. We evaluate the latency, throughput and time advancement benchmarks under varied size of federates and varied size of FedServs. Issues such as message routing, multicasting and synchronization are specially addressed in this article. Results show that Multiple-FedServ architecture can scale well.