Processing of Bistatic SAR Data From Quasi-Stationary Configurations

Abstract

Standard synthetic aperture radar (SAR) processing algorithms use analytically derived transfer functions in the 2D frequency and range/Doppler domains. These rely on the assumption of hyperbolic range histories of monostatic SARs with straight flight paths. For bistatic SARs, the range histories are no longer hyperbolic, and simple analytic transforms do not exist. This paper offers two solutions for bistatic SAR data processing under the restriction of quasi-stationarity, i.e., sufficiently equal vectors of transmitter and receiver. 1) Moderately bistatic configurations can be handled satisfactorily by using hyperbolic range functions with a modified parameter, which is a solution already well known for the accommodation of curved orbits in the monostatic case. This velocity approach is shown to be of surprising range of validity even for pronounced bistatic situations. It is not to be confused with the monostatic flight path approximation, which is shown to be inapplicable for any practical case. 2) With increasing separation of transmitter and receiver, the equivalent approximation deteriorates. To cope with extreme bistatic configurations, a general approach named NuSAR is proposed, where the involved transfer functions are replaced by numerically computed ones. This paper describes how the transfer functions are computed from the given orbits and the shape of the Earth surface. In any of these two cases, the bistatic SAR data can be processed by standard SAR processors; only the conventional transfer functions need to be replaced. Neither are there time-domain prefocusing or post focusing steps required nor complicated mathematical expansions involved. The presented algorithms are also applicable to very high resolution wide-swath (or squinted) SARs on curved orbits.