While this may sound straightforward, there is the additional constraint to also operate within limitations of memory and processing power available in fielded systems. MERC has now accumulated several years of experience in network selection for such systems.
l Applied Mechanics Mechanical Design Information Systems Reliability and Human Engineering Electronic Systems Applied Systems Logistics Services
AAR-44 Missile Warning System AAR-47 Missile Warning System Applied Systems Division © 1987-2010 Mercer Engineering Research Center 135 Osigian Blvd. • Warner Robins, GA 31088 info@merc-mercer.org • www.merc-mercer.org • 478.953.6800 The AI approach compares to tediously juggling millions of available data items using traditional programming (i.e., decision trees and loops). In particular, changes in one parameter to accommodate one scenario frequently deteriorate programmed previously scenarios. Standard programming becomes a difficult process, balancing one signal against another. In comparison, a classification network, after training, is programmed to the mathematically best case to all inputs. Fortuitously, relatively recent software developments in the field of as artificial intelligence (AI) can now attack the problem. MERC has received two USAF contracts for MWRs, both IR and UV, calling for study and analysis of the ability of AI techniques and algorithms to distinguish missiles from background signals. MERC analysis has utilized AI techniques to globally train from large quantities of recorded data from both missiles and background sources. After training, a network then has the detection intelligence. MERC has shown false alarm rates reduced by two-thirds on one system. Advanced Signal Analysis for USAF Missile Warning Receiver Systems It is a challenging proposition to passively detect a passively guided missile from an aircraft platform. A Missile Warning System (MWR) must detect missiles with very high probability, at the same time minimizing the number of false alarms. The MWR must function autonomously and must be able to signal for countermeasures in a very short time. Background clutter acts synergistically with missile signatures changing during flight to provide a demanding task for the programmers of these systems.