A novel use of Gram-Schmidt for detection and estimation

A novel approach to the detection and estimation of directional signals incident upon a linear array of sensors uses the diverse dimensions of the array polynomials associated with the statistically independent beams output by the adaptive Gram-Schmidt processor. Zero patterns generated by these polynomials provide the information necessary to detect the number of incident signals and estimate their angles of arrival. L² norms based on phase differences between zeros in observed pairs of beams are critical in the detection process. Monte-Carlo results compare the performance of the proposed method with that of the Root MUSIC algorithm for one signal and two closely spaced signals in white stationary noise. These results cover varying SNR, number of snapshots, and signal spacing for the two signals. Single trial results for multiple strong jammers are presented