Synchronization and management of shared state in HLA-based distributed simulation

The HLA runtime infrastructure can support a conservative simulation protocol for its time management service. However, the performance of conservative simulation protocols is very much dependent on lookahead that one can extract out of a simulation model. The most conservative value has to be taken in order to ensure the causality constraint. In this paper, we propose two algorithms, pullRO and pushRO, that allow one to replace some of the timestamp order (TSO) messages (possibly those causing zero lookahead values) with receive order (RO) messages. This removes the time constraint that these messages impose on the lower bound timestamp calculation, which in turn will improve the time advancement rate of federates. The algorithms still ensure the causality constraint and a middleware approach is used to preserve the semantics of the RTI APIs. The performance of the two algorithms is compared against a baseline model where no TSO messages are replaced.