Performance of Digital Matched Filter Correlator With Unknown Interference

The S/N loss due to amplitude quantization is examined in connection with a digital matched filter correlator for a binary coded signal. A minimax binary quantizing concept is described, which adds an optimized noise dither prior to hard clipping. The design philosophy is minimax; i.e., the loss from an ideal analog correlator is minimized in the presence of the worst interference constrained to a specified average power. The resulting worst performance degradation is 4.8 dB, and occurs for constant amplitude interference. The extension of the minimax approach to a multilevel randomized quantizing concept is sketched. With four-level quantization, an upper bound on the performance degradation is 2.3 dB. The effect of finite time estimation of the interference power is determined, with the conclusion that a short-term power averaging can be used with only a small additional loss. Computer simulation results are in accordance with theoretical predictions.