A variational approach to waveform design for synthetic-aperture imaging (Q2466977)

The idea of a synthetic aperture radar is to attach an antenna to an airplane and continuously transmit and receive radar signals during the flight. By sophisticated signal processing, it is possible to reconstruct a more detailed image of the overflown area than would be possible with a stationary antenna of the same size. Effectively the aperture of the antenna can be much larger than both the antenna and the airplane themselves, which makes it possible to reconstruct a highly resolved image. This paper deals with two problems for synthetic-aperture radars: finding the optimal filter for a filtered backprojection for the image formation process and to derive an optimal waveform for the transmitted signal. The radar signals are modelled with the scalar Helmholtz equation, which means that no polarization effects are considered. The scattered field is computed with the Born approximation. The first step is the construction of the optimal filtered backprojection (which is an integral operator in both space, time and frequency variables). The optimal filter is chosen so that it yields an image with minimal mean square error for a known signal when compared with the true image. From the derivation it follows that the error estimate of the filter depends on the transmitted waveform. The second step is thus to construct the optimal waveform for the transmitted signal. The object here is to choose a waveform that minimizes the mean square of the error. The analysis splits up to a noise-dominated and low-noise case. The techniques in this paper are purely analytical and involve optimization and approximation methods and microlocal analysis.