Maximum Likelihood Based Adaptive Detector.

摘要

Adaptive, frequency domain, array signal detection is considered when there exist data blocks containing interference alone and signal plus interference that are disjoint in time and span several frequencies. The data block solely containing interference is used to form estimates of the interference covariance structure at each frequency. These estimates are substituted into generalized likelihood ratios to form constant false alarm rate detectors for deterministic and Gaussian signals that are limited in time to the signal plus interference data block. Central limit theorem based normal approximations are used to determine thresholds and the signal-to-interference ratio 201SIR202 required to achieve specified false alarm and detection probabilities. The SIR required for the proposed detectors is shown to be the SIR required for the ideal conventional detector with interference covariance estimation losses and intra-block correlation losses. The gain in adaptive processing 201i.e., the reduction in signal strength required to achieve specified false alarm and detection probabilities202 is seen to be the ratio of the array gain improvement 201AGI202 to the change in the estimation and correlation losses resulting from the shift from conventional beamforming 201dimension equals one202 to higher dimension processing. Thus, the AGI must exceed the change in the estimation and correlation losses before adaptive detection becomes attractive.