Distributed Simulation Run Automation Capability for the Navy PRA Testbed

摘要

The Navy Probability of Raid Annihilation (PRA) Testbed implements HLA federated simulations of ship combat system elements against independent, reactive threat raids in a common environment to formulate an overall combat system assessment. The PRA Testbed is a cornerstone of the U.S. Navy Ship Self Defense Test & Evaluation (T&E) Enterprise. The LPD 17 ship class is the first to implement a PRA Testbed baseline as a formal component of ship class OT&E. Products and lessons from the LPD 17 baseline are being transitioned to multiple ship classes including DDG 1000, LHA 6, LCS, and CVN 21. The PRA Testbed LPD 17 baseline is deployed as a secure, geographically distributed, time managed federation with three nodes: the U.S. Naval Research Laboratory (NRL) Washington, DC; Johns Hopkins University (JHU) Applied Physics Laboratory (APL) Laurel, MD; and Naval Air Warfare Center (NAWC) Weapons Division China Lake. Federation execution is controlled centrally from the NRL node. The PRA Testbed LPD 17 baseline federation is required to conduct nearly 2000 simulation test events during the course of LPD 17 PRA Assessment. Each scenario execution involves computation intensive calculations that progress much slower than real time. The development team is using Run Automation software, dubbed OneButtonStart to automate batch runs of the federation from a single control node. OneButtonStart is an important part of the LPD 17 testing program because it dramatically reduces manning requirements and takes advantage of all available computer time to minimize schedule. Yet, there are numerous challenges to implementing automated control of a classified, non- realtime federation that includes both legacy and new software. In this paper, we describe the Run Automation capability and how it will be used in the PRA Testbed LPD 17 baseline. We also share lessons learned in automating runs for a secure, geographically distributed, time-managed simulation with multiple nodes.