Decision Support Systems (DSS)
Collecting data from one sensor, or many sensors, is only the first step in the overall decision-making process, which might be inspection, monitoring, tracking, etc. Often, many other components, such as databases, simulation models, and mathematical optimization, must be combined to form a fully developed decision support system (DSS).
The final output of a DSS is a recommendation, interpretation, or prediction regarding the situation of interest, such as crop treatment, food safety, or water quality. Such systems can become quite complex, with many interacting components, and can be either embedded and real-time systems or off-line systems.
When multiple data streams are collected, one must decide how those data will be fused in the decision process. Separate data streams can be:
- Mixed mode (including both quantitative and qualitative data).
- Redundant (validating the data from another stream).
- Complementary (providing measures of multiple characteristics of an object).
- Supportive (helping verify some interpretation of another data stream).
Given the wealth of sensors available today and the volume of information they generate, sensor fusion research and development has become a growing and highly active community. The final result of the sensing process (which may include sensor fusion) is an accurate and reliable interpretation of the data regarding an object.
Once reliable information has been collected in the sensing step, it may need to be combined with information and knowledge from other sources. For example, a crop simulation model may be used to project a future crop condition (or value) based on current measurements. Or, data on a fermentation process may be compared to historical trends stored in a data base to assess whether they lie within acceptable limits.
More qualitative knowledge, in the form of expert rule bases, may also be applied. A DSS may also incorporate economic models or calculations to determine which courses of action are reasonable. Other factors that might need to be considered include operational cultures within the organization or the industry or current financial markets.