Phase-Dependent RCS Measurements in the Presence of Outliers
Lorant A. Muth, Timothy Conn
Coherent radar cross section measurements on a target moving along the line-of-sight in free space will trace a circle centered on the origin of the complex (I,Q) plane. The presence of additional complex signals (for example, background, clutter, average of the target-mount interaction, etc.)that do not depend on target position will translate the origin of the circle to some complex point(Io,Qo). A number of techniques have been used successfully to analyze phase-dependent I-Q data. The presence of outliers, however, can degrade the results, that is, significant errors can be introduced in the determination of all 3 parameters, Hence, we seek to increase the robustness of our analysis technique by first eliminating or reducing the influence of outliers. This is especially important at sub-wavelength translations at VHF, where spectral techniques are not applicable and only a limited arc of data is available. A combination of a robust and efficient Least-Median Square (LMS)[2,3] and an Orthogonal Distance Regression (ODR)[4-6] algorithm is used 1)to eliminate outliers, and then 2)to separate the target and background signals. We analyze data obtained as an Arrow III target moves relative to its supporting pylon. To demonstrate the effectiveness of the technique, we introduce rf interference signals into S band data and show that uncontaminated parameters can be recovered with acceptable uncertainties [7,8].