Beam space signal processing for directional transmission phased arrays

Abstract

Beam space (BS) processing is a spatial signal processing technique using beam output data. For example, the BS beamformer applies weights to a set of beam outputs, which are then summed to form a new output. In this way, advanced optimum spatial signal processing algorithms can be applied when the element outputs are not accessible. However, existing BS processing algorithms are based on a model that assumes a passive receiving system or for active systems that the transmission is omni-directional and can be ignored. When the transmission is directional as is typical for phased arrays that electronically scan over a given sector, such methods are mismatched and result in significant performance degradation. The first part of this thesis presents a new formulation of BS processing for the scenario where relatively narrow beams are directionally transmitted and received and then scanned over a given sector of interest. New formulae are developed for this case and the performance of the new formulae is analysed. The second part of this thesis is focused on the properties of directional transmission BS processing. When beams are formed in a sector of interest, problems related to the region outside the sector of interest are investigated, including analysing the output in the direction-of-arrival (DOA) of an interference lying outside the sector of interest, removing the high response in the region outside the sector of interest and mitigating a spurious output peak caused by the interference. Additionally, phased array errors cause the array response to be different from that being assumed and can seriously degrade the performance of the BS beamformer, a robust BS beamformer is developed to improve the tolerance to errors. Cramér–Rao Bounds (CRB) for DOA estimation for the directional transmission BS are derived and compared with the omni-directional element space (ES) and BS cases. The performance of the optimum BS beamformer for a non-stationary scatterer is evaluated. The third part of this thesis deals with BS processing for coherent signals. The commonly used subarray algorithms for removing coherence in the ES processing cannot be applied to the BS problem directly. A method of reconstructing the ES signal subspace is developed for the omni-directional transmission BS case, and then existing methods, such as MUSIC, in ES processing can be applied. For the directional transmission BS case, a method is proposed to reconstruct a matrix which is a summation of weighted self-outer products of ES signal steering vectors, and this matrix allows the DOAs of coherent signals to be estimated regardless of coherence. Finally, the developed algorithms are investigated by carrying out spatial processing on real experimental data containing stationary targets.