Cross-Correlation: Statistics, Templating, and Doctrine

Abstract : This is one of a series of algorithm analysis reports on work performed covering selected algorithms in existing Intelligence and Electronic Warfare (IEW) systems. It focuses on cross-correlation algorithms in four fusion systems, especially on those dealing with information from Signals Intelligence (SIGINT) reports. These algorithms decide if one battlefield entity (e.g. a radar) belongs to or contains a higher or lower echelon entity (e.g. an artillery battery). The decision tests mix some statistics (mainly Chi-square tests) with simple in-bounds, linear, and exponential tests based on doctrinal templates (deployment radii and Table of Organization and Equipment (TO/E) equipment lists). The mathematics of such hybrid tests is analyzed, illustrated by the tests implemented in BETA(Battlefield Exploitation and Target Acquisition) and TEMPRO(Template Assisted Intelligence Fusion Program) (MAGIS(Marine Air/Ground Intelligence System) and TCAC were also surveyed), and focusing on the underlying mathematical assumptions and military template structure. Effects of previous processing (e.g. by the sensor systems) and incoming message formats on the data distribution seen by the tests, and thus on the outcome of the tests, is also discussed.