Abstract: The performance of direction of arrival (DOA) estimation algorithms based on subspace projection deteriorates under nonideal scenario. A new method which synthesizes noise subspace projection with weighted signal subspace projection is proposed. The article analyses many subspace projection algorithms based on eigen decomposition and categorizes them according to spatial projection matrixes. The method obtains the spatial spectrum based on signal subspace projection weighted with the reciprocal of principal eigenvalues. Then it is multiplied on the spatial spectrum based on noise subspace projection. The synthetic spatial spectrum can be used to multitarget resolution. Especially it shows high resolution and robust performance under scenarios of low SNR, snapshot deficiency or targets with different powers. Theoretical and statistical analysis shows that its performance is better than MUSIC and SSMUSIC.

Key words: array signal processing, spatial spectrum, DOA estimation, subspace projection, robustness