A Decoupled Federate Architecture for Distributed Simulation Cloning

Distributed simulation cloning technology is designed to perform “what-if” analysis of existing High Level Architecture (HLA) based distributed simulations. The technology aims to enable the examination of alternative scenarios concurrently within the same simulation execution session. State saving and recovery are necessary for cloning a federate at runtime. However it is very difficult to have a generic state manipulation mechanism for any existing federate, as these can be developed independently and freely. The correctness of replicating a running federate significantly depends on the Runtime Infrastructure (RTI) software. The distributed simulation also needs fault tolerance to provide robustness at runtime. This paper proposes a decoupled federate architecture to address the above issues. A normal federate is decoupled into two processes, which execute the simulation model (virtual federate) and the local RTI component (physical federate) respectively. The decoupled approach interlinks the two processes together via Inter-Process Communication. The virtual federate interacts with the RTI through the standard RTI service interface supported by a customized library. The decoupled architecture ensures the correct replication of federates and facilitates fault tolerance at the RTI level. At the same time, it provides user transparency and reusability to existing federate codes. Benchmark experiments have been performed to study the extra overhead incurred by the decoupled federate architecture against the normal federate. The encouraging experimental results indicate that the proposed approach has a performance close to the normal one in terms of latency and time synchronization.