Handling vast amounts of complex information from databases, sensors and simulations as well as keeping track of its location, requires database support for information storage and retrieval as an important part of the infrastructure for information fusion.
The decision-making affects the control of physical processes in real-time decision support applications. The fusion of information must not only be semantically correct and consistent in the database, it must also satisfy time constraints. If simulation is used to predict the behavior of the system in response to a specific parameter choice or action, the simulation must return the result within a bounded time, especially if the simulation has hardware-in-the-loop that depends on the simulation outcome. With hardware-in-the-loop, robustness and recoverability of the simulated parts become important. The cost of restarting failed simulations may otherwise become prohibitive.
Since information sources are developed and maintained by people in different groups and organizations distributed physically and logically, on heterogeneous hardware and software, issues concerning the distribution of data and simulations become relevant. This problem can be compared to the data warehousing challenge in combining different database sources.
The information fusion process involves monitoring and responding to system state and events, such as complex sequences of events or specific data states in the database that trigger the computation of refined information. An infrastructure that supports such reactive behavior, using rules to specify active functionality, allows these processes to be described independently of specific application code.