Research On Narrowband Array Signal Processing Algorithms

Array signal processing refers to a sensor array composed of some sensors thatare arranged in a certain way and distributed in the different space, receiving radiatingsignals, then specific processing is operated on them. In this way, the interested targetsignals are enhanced while the unwanted interference and noise signals are suppressed,which can extract the useful parameters of signals. Having been widely applied inmany areas such as seismic exploration, wireless communications, biomedicalimaging and so on, spatial spectrum estimation and beamforming are two basicproblems in the array signal processing.This thesis mainly focuses some researchs on array signal processing algorithmsbased on far-field narrow-band signal. Firstly, the most basic model that consists ofsingle-source and two sensors is bulit, which contributes to deduce the receivedsignal’s mathematical model of uniform equidistant linear array, uniform equidistantcross array, uniform equidistant circular array, and on the basis of the uniformequidistant linear model, the basic theoretical algorithms of array signal processingare described. Secondly, in order to deal with the problem that the reduced-rankfourth-order cumulant matrix exists an estimation error induced by finite samplingsnapshots, three improved direction-finding algorithms are put forward. In theproposed algorithms, the reduced-rank fourth order cumulant (FOC) matrix isobtained via removing the redundant information encompassed in the primary FOCmatrix, while the effective aperture of the virtual array keeps unchanged. Then, theToepltiz structure of the reduced-rank FOC matrix is recovered with applying theToepltiz approximation, and finally, by using the MUSIC algorithm, the ESPRITalgorithm and the OPM, the direction of arrival signals can be estimated, respectively.By using the Toeplitz approximation, the proposed algorithms can not only improvedirection-finding accuracy with less antennas, but add computational complexityunconspicuously. Lastly, a modified decorrelating beamforming algorithm ispresented. In the proposed algorithm, the information of all sub-array covariancematrixes that need to be smoothed is found by analyzing the structure of thecovariance matrix of the received data. To avoid the massive calculation of sub-arraycovariance matrixes, the forward and backward sub-array covariance matrixes areconstituted by means of processing the covariance matrix of the received data. Theproposed algorithm can adjust the length of sub-array flexibly and reduce the calculation greatly, while guarantee the performance of beamforming.