Target detection via Kalman filtering

Abstract

A simple, time domain method is used to analyze moderate to high PRF radar signals. The quantities of interest are the signal's PRF, SNR, and time of arrival. The time of arrival problem is important because it can be used, with multiple sensors, to determine the position of the emitting target. An algorithm is described which will produce these values using Kalman filtering. Individual pulses in a pulsed type radar are measured against a threshold using a two sample detection scheme to provide some glitch rejection. Results of individual, time domain measurements of the signal parameter are smoothed with a Kalman filter. Integrating the pulse train envelope during the radar dwell time provides the energy centroid for a scan cycle. This centroid, time differenced with multiple sensors provides observables for an Extended Kalman Filter for emitter localization. The work here simulates all data. Tests of the algorithms developed were conducted on real, classified data in addition to the work presented here...