Abstract: An FFT-like array architecture (500), for use on the Doppler filters of a radar system, includes a plurality of stages (505, 506) of weighted butterflies (501, 502, 503, 504), in which each butterfly is provided with four weighting multipliers (410-416). The weights (W1, W2, W3, W4) of the multipliers of the array are determined by an iterative process in which the input and output signals are selected, the input signals are applied to the array, and the actual output signals are compared with the desired output signals to produce error signals. The error signals are backpropagated through the array, to correct the weights. The input signals are again applied, and the corrected output signals are again compared with the desired output signals to produce new error signals, which are again backpropagated to correct the weights. This procedure is used iteratively until the array "learns" the weights which give the desired output signals.

Abstract: A monopulse radar system includes sum and difference receivers, and an auxiliary receive beam offset from the sum and difference beam axis. The received sum and difference signals are processed to produce first and second weighting signals which, when multiplied by the auxiliary signal, produce first and second weighted auxiliary signals which are approximately equal to the magnitude of, and are in a particular phase with, the received sum and difference signals, respectively. In the presence of a jamming signal in the main lobe of the monopulse receiver, the weighted auxiliary signals are summed with the sum and difference signals to produce adapted sum and difference signals, in which the jammer signal is suppressed by a null in the effective receive antenna pattern. The null perturbs the sum and difference patterns, so that the ratios of the sum and difference signals no longer accurately represent the off-axis angle of the target.