Adaptive CFAR Detection and Reduced-Rank Space-Time Adaptive Processing

摘要

A new approach to adaptive Constant False Alarm Rate (CFAR) detection and reducedrank space time adaptive processing is introduced. A general methodology is developed for the design and analysis of reduced rank detectors with known and unknown covariance. The detection and false alarm probabilities are explicitly calculated and a cross spectral metric is introduced for rank reduction. This metric is demonstrated to result in a low rank detector which outperforms the principal components technique. The cross spectral metric provides subspace dimensionality robustness and allows for the dimension of the detector to be reduced beneath the dimension of the noise subspace eigenstructure without significant performance loss.