Integrated Avionics Information Processing Development

Intelligent processing of battlefield sensor data information provides substantial force-multiplying benefits. The timely and cost-effective leveraging of information provides a significant advantage over adversaries who do not possess comparable information processing capabilities. Among these capabilities is the ability to detect patterns and trends in sampled data. Automated discovery of causal or deterministic behavior in sampled data provides a model of the underlying system that produces the data. Such a model can then be used to interpolate or forecast the behavior of the system under a variety of operating conditions. The ability to accurately model sampled data has a wide variety of high-payoff avioncs, electronic warfare, and other military applications. This proposal presents two existing methodologies for finding causality or patterns in data. Pattern Theory provides a framework for analytically determining the degree of "pattern-ness" of an existing function. It has been extended to the Decomposed Function Cardinality (DFC) and Function Extrapolation by Re-composing Decompositions (FERD) methodologies, which can produce a minimal functional description from a sample data set [27]. The Abductive Information Modeler" is a numeric machine learning algorithm suited to modeling data of a continuous nature. It has been used to solve many complex problems in the defense and commercial arenas. Areas of commonality and complementary differences are described. An approach for integrating these technologies is presented. Finally, existing and potential future applications of FERD and AIM are presented.