A New Subspace Decomposition Based Array Manifold Estimation Algorithm

The manifold vector (MV) of an assembled array usually differs from its theoretical value. The deviation raises the sidelobe levels of the beampattern and degrades the performance of high-resolution array processing algorithms, hence limiting practical application of the array. A good approximation to the MV can be obtained on the basis of a properly chosen parametric array error model, with a partially estimated MV for bootstrapping. MV estimation algorithms such as definition-based method (DB) and least square method (LS) are computationally complex, and their accuracy is subject to the number of snapshots available for the estimation. We propose a new MV estimation algorithm that exploits the equivalence between the signal subspace of the covariance matrix and the space spanned by the MV to estimate the magnitude-phase response of an array. We can derive the MV by combining the magnitude-phase response with the estimated direction of arrival (DOA). Simulation results demonstrate that the proposed method is twice as accurate as DB and LS, and its computational complexity is reduced by an order of magnitude.