Upstream Data Fusion: History, Technical Overview, and Applications to Critical Challenges

215 INTRODUCTION Sensor and Data Fusion The term “sensor and data fusion” refers to techniques that combine data from multiple sources to produce new information and inferences and achieve more complete, clear, precise, accurate, and timely estimates pstream data fusion (UDF) refers to the processing, exploitation, and fusion of sensor data as closely to the raw sensor data feed as possible. Upstream processing minimizes information loss that can result from data reduction methods that current legacy systems use to process sensor data; in addition, upstream processing enhances the ability to exploit complementary attributes of different data sources. Since the early 2000s, APL has led a team that pioneered development of UDF techniques. The most mature application is the Air Force Dynamic Time Critical Warfighting Capability program, which fuses a variety of sensor inputs to detect, locate, classify, and report on a specific set of high-value, time-sensitive relocatable ground targets in a tactically actionable time frame. During the late 2000s, APL began expanding the application of UDF techniques to new domains such as space, maritime, and irregular warfare, demonstrating significant improvements in detection versus false-alarm performance, tracking and classif ication accuracy, reporting latency, and production of actionable intelligence from previously unused or corrupted data. This article introduces the concept, principles, and applicability of UDF, providing a historical account of its development, details on the primary technical elements, and an overview of the challenges to which APL is applying this technology. Upstream Data Fusion: History, Technical Overview, and Applications to Critical Challenges