Doppler-only tracking with the recursive Gauss-Newton filter

Abstract

This paper investigates target tracking using Doppler only, observed from multiple receivers in a multi-static configuration. The tracking filter selected for this study is the recursive Gauss-Newton filter based on the LevenbergMarquardt method. FM Band Commensal Radar (radar that uses another system's EM emissions) produces a Range versus Doppler plot, where the Doppler resolution is very precise due to the long integration time, but range resolution is at best poor due to the narrow broadcast channel bandwidth. Thus, tracking that relies mostly on Doppler seems most sensible for this type of radar.

A multi-mode real-time terrain parameter estimation method for wheeled motion control of mobile robots

“…The RGN algorithm is a recursive nonlinear minimum least square (NLS) method [25,30] which is established on GaussNewton (GN) algorithm and has been applied in power system estimation [31] and target tracking [32]. We use it as the outer layer algorithm because its cost function is the sum of squared estimation error that is related with the history measurements, utilizing all the measurements and producing a global minimum least square solution.…”

Evaluating commensal sensors for detecting objects of interest in the Low Earth Orbit

“…While strong Doppler-shifts provide increased insight into the object being analyzed [12], they also limit the effective integration time thus reducing the ability for many radar algorithms to detect the presence of the object. In satellite communications, Doppler-shifts are compensated for by adjusting the received signal (Fig.…”