A New Method For Generation Of Optimum Matched Filter From Replica Data For Pulse Compression

Abstract

The conventional matched filter approach uses the replica of thc transmitted, frequency modulated signal to generate a reference function for the range compression. Phase and amplitude errors occuring in the total transceiver chain before digitalization will be contained in the reference function, which is the complex conjugate of the time reversed replica. The impulse response function in this case has the best achievable signal to noise ratio but high sidelobes or paired echoes will appear for severe phase and amplitude errors. We present here a new method for the generation of the reference function from the replica, which provides both, matched filtering and deconvolution of the amplitude and phase errors. The final impulse response function will be like the ideal one irrespective of the amount of errors present. Several simulation results are shown which are significant for the improvement of the image quality and for relaxation in the amount of allowed amplitude and phase errors in the S A R design for range processing. The proposed method can also be extended to azimuth processing of signals received a t high squint angles.