Array signal processing robust to pointing errors

The objective of this thesis is to design computationally efficient DOA (directionof-arrival) estimation algorithms and beamformers robust to pointing errors, by harnessing the antenna geometrical information and received signals. Initially, two fast root-MUSIC-type DOA estimation algorithms are developed, which can be applied in arbitrary arrays. Instead of computing all roots, the first proposed iterative algorithm calculates the wanted roots only. The second IDFT-based method obtains the DOAs by scanning a few circles in parallel and thus the rooting is avoided. Both proposed algorithms, with less computational burden, have the asymptotically similar performance to the extended root-MUSIC. The second main contribution in this thesis is concerned with the matched direction beamformer (MDB), without using the interference subspace. The manifold vector of the desired signal is modeled as a vector lying in a known linear subspace, but the associated linear combination vector is otherwise unknown due to pointing errors. This vector can be found by computing the principal eigenvector of a certain rank-one matrix. Then a MDB is constructed which is robust to both pointing errors and overestimation of the signal subspace dimension. Finally, an interference cancellation beamformer robust to pointing errors is considered. By means of vector space projections, much of the pointing error can be eliminated. A one-step power estimation is derived by using the theory of covariance fitting. Then an estimate-and-subtract interference canceller beamformer is proposed, in which the power inversion problem is avoided and the interferences can be cancelled completely.