Improved range superresolution via bandwidth extrapolation

Linear predictive bandwidth extrapolation (BWE) techniques are used for obtaining improved range superresolution. It is noted that conventional superresolution techniques have performed well in achieving point-target location estimates, but they cannot readily provide target signals with linear superposition properties. Several superresolution algorithms are described. They are utilized in computing the linear prediction filter coefficients needed for BWE prediction. These algorithms include both sidelobe canceller (SLC) techniques and singular value decomposition (SVD) techniques. They are based upon the accumulation of a sample covariance matrix (SCM) in the spectral domain via subaperture averaging over a modest number of pulse repetition intervals (PRIs). The objective is to achieve target resolution at reasonably low signal-to-noise ratio (SNR). The filter coefficients are then utilized for BWE prediction over the entire expanded bandwidth. Simulation examples demonstrate that a factor of four improvement in range resolution is readily achieved.<>