Search

A major objective of the Project Manager, Remotely Monitored Battlefield Surveillance System (REMBASS), is the development of sensor systems capable of classifying targets. Existing classifier design procedures rely heavily on statistical techniques, such as multiple correlation analysis, which have been shown to be strong tools for this purpose. Seismic and acoustic signals are affected by a number of target and environmental variables, and since the REMBASS sensors are intended to operate satisfactorily for a large variety of targets and terrains, it is recognized that an adequate design will require a signature data base representative of the spectrum of conditions under which the system is to operate. This report presents a plan for assembling a data base for the development and testing of two types of seismic and acoustic classifying sensors: a sensor for classifying single targets, and a sensor for classifying single targets in a multiple-target environment. The plan also (a) defines the targets to be used in the data collection program, (b) defines the test site conditions to be used in the data collection program and develops a method for relating test site conditions to worldwide environments, (c) establishes a method for assembling a data base of realistic background noise signatures, and (d) specifies the test procedures for signature acquisition from the various target classes. The report includes maps showing predicted worldwide performance of seismic and acoustic sensors and the rational behind their formulation. (Author)