Direction-of-Arrival Estimation For Wideband Signals Via Envelope Alignment

In practical array signal processing systems,the signals to be processed may come from either circular or noncircular sources,or both.It has been shown that the noncircularity of signal can be exploited via data augmentation to enhance the performance of array signal processing in beamforming and direction-of-arrival(DOA)estimation.However,the current methods apply only to the narrowband case and cannot be directly extended to the wideband case.This thesis is focused on the development of data augmentation based DOA estimators for circular and/or noncircular wideband source signals,via envelope alignment.The main work and contributions include the following two aspects.1)Based on the envelope alignment technique,two types linearly constrained minimum power wideband widely linear beamformers are developed for the cases of unknown signal quotient and known signal quotient,respectively.Two widely linear beam scanning based time-domain methods,called respectively the first type widely linear minimum power method(WL-MP1)and the second type widely linear minimum power method(WL-MP2),are then presented.Because of the extended virtual aperture due to the involved data augmentation in the widely linear processing framework,both methods have superior resolution capabilities to the traditional linear beam scanning methods in the presence of noncircular source signals with relatively large noncircularity rates.2)By envelope alignment in addition to data augmentation,the augmented effective noise subspace is constructed for each scanning angle via eigendecomposition of the aligned and augmented array output covariance matrix according to the effective rank of its noise free version.Two time-domain DOA estimation methods based on rank deficient scanning and norm null scanning are then proposed,respectively for the cases of coexisted noncircular and circular wideband source signals and partially noncircular or circular wideband source signals only.In the presence of noncircular source signals with relatively large noncircularity rates,the envelope aligned rank deficient method outperforms the conventional subspace based and sparse representation based non-augmentation methods in terms of resolvability and DOA estimation accuracy.