Compensating for Motion Errors in UWB SAR Data

NASA’s Jet Propulsion Laboratory

A method of processing data acquired by ultra-wide-band (UWB) syntheticaperture radar (SAR) provides for suppression of those errors that are caused by the undesired relative motion of the radar platform and the targets. This method involves, among other things, processing of data in the wave-number or frequency domain and the application of motion compensation as a function of the positions of a target relative to the radar platform.

Motion Compensation to first order is effected for a data strip, then refined to second order over small overlapping constituent patches. A motion-compensated image of the large patch is then constructed as a mosaic of the images of the smaller patches.

The need for this method arises because of two sources of complication in UWB SAR that are not present in narrow- band SAR. One source of complication is that the prior commonly used SAR motion-compensation algorithm depends upon the Doppler shift of a well-defined radar-signal frequency, but the signal frequency in UWB radar is not well defined. The other complication is that the prior motion-compensation algorithm depends on azimuthal narrowness of the radar beam, but in UWB SAR, the requirement to make the range and azimuth resolutions approximately equal translates to a requirement that the radar beam be azimuthally wide (typical width of the order of a radian).

In a wide-beam SAR system, the key dilemma in properly compensating for motion is that one needs to track the location of each target on an SAR strip map, but the locations of the targets are not known from the outset. The present method addresses this dilemma. The method involves two stages of processing (see figure).

In the first stage, a strip parallel to the flight track is processed and targets are motion compensated, assuming that they are located in the antenna fan beam plane. Following the first-order motion compensation, a frequency-domain SAR processing algorithm is applied. The motion compensation for targets in off-boresight directions is not correct, but the motion compensation for the target(s) in the nominal center of the beam is correct. This stage of processing ensures that target impulse responses are located correctly geometrically, albeit insufficiently focused.

In the second stage of processing, the data strip is divided into overlapping small patches and it is pretended that the target(s) in each small patch lie at the center of the patch. The data processed in stage 1 are reprocessed within each small patch to refine the motion compensation. This reprocessing includes second-order motion compensation that takes the form of frequency and phase shifts applied to the partially motion-compensated UWB SAR data. Inasmuch as the motion compensation is perfect only at the center of each small patch, the smaller the patches, the better the motion-compensation performance. For the sake of computational efficiency, the two-stage processing algorithm has been formulated such that the reprocessing in small patches is much less computationally demanding than is the processing of the wide area patch, such that it is computationally affordable to reprocess many small patches.

This work was done by Soren Madsen of Caltech for NASA’s Jet Propulsion Laboratory. For further information, access the Technical Support Package (TSP) free on-line at www.nasatech.com/tsp under the Information Sciences category. NPO-21096