SUBSPACE-BASED ADAPTIVE GENERALIZED LIKELIHOOD RATIO DETECTION

Subspace-based adaptive detection performance is examined for the gene ralized likelihood ratio detector based on Wilks' Lambda statistic. Th e problem considered here is detecting the presence of one or more sig nals of known shape embedded in Gaussian distributed noise with unknow n covariance structure, The data is mapped into a subspace prior to de tection, The probability of false alarm is independent of the subspace transformation and depends only on subspace dimension, The probabilit y of detection depends on the subspace transformation through a nonada ptive signal-to-noise ratio (SNR) parameter, Subspace processing resul ts in an SNR loss that tends to decrease performance and a gain in sta tistical stability that tends to increase performance. It is shown tha t the statistical stability effect dominates the SNR loss for short da ta records, and subspace detectors can require substantially less SNR than full space detectors for equivalent performance. A method for des igning the subspace transformation to minimize the SNR loss is propose d and illustrated through simulations.