This study will examine how distributed simulations are created and managed online so that they can be shared, used, augmented, coordinated, and enhanced in a highly decentralized way. The fundamental elements of this research will be to 1) study new approaches to effectively manage operational simulations and data assimilation over a distributed set of clients, 2) implement the technologies that will enable this distributed approach, and 3) evaluate these technologies using real-world experiments. The PI's envisage operational simulations to include scientific, environmental, public infrastructure, and other computer simulations that operate continuously and online, assessing the current state of a system and predicting future states. Distributed data assimilation is defined as the ability of a simulation to incorporate new sources of data dynamically while the simulation is operational. Examples include air traffic control, financial systems, and weather and ocean forecasting. The ability to easily combine simulations within and across such applications areas, asynchronously and remotely, is a goal of the proposed activity and would represent a major advance in knowledge networking and distributed intelligence.
