Development of a system model and least mean square (LMS) filter for the Naval Postgraduate School (NPS) Infrared Search and Target Designation (IRSTD) system.

Abstract

A system model and a least mean square (LMS) filter for the Naval Postgraduate School (NPS) Infrared Search and Target Designation (IRSTD) system were developed. The system model was developed and run on the NPS IBM 3033,4381 mainframe computer network. The model simulated the effects of the optics and electronic processing equipment of the IRSTD system, and produced output data representative of the detector outputs of the system. The outputs of the IRSTD model were used to develop a digital filter based on the principle of least mean square optimization between an actual IRSTD detector output and a power series expansion representing a detector output containing both background clutter and a model target signal. It was determined that the raised cosine function served as the best model for IRSTD point and near-point targets, (0.1 mrad by 0.1 mrad to 1.5 mrad by 1.5 mrad), and a set of trial LMS filters were zenerated based on this model. After filtering both simulated and real data, consisting of simulated and real target signals embedded in simulated and real backgrounds, it was determined that an LMS filter generated from a raised cosine with a half-amplitude width of 0.9 mrad was optimal for point and near-point targets. The signal-to-noise ratios of all target and background combinations increased by a factor of approximately 30 for the simulated backgrounds, and approximately six for the real backgrounds, upon filtering the detector outputs with the optimal LMS filter. It is believed that this filter should be incorporated into the NPS IRSTD system as an initial signal processing filter, and that the filtered outputs are appropriate for use as inputs to target detection and acquisition routines.