A time-space beamformer and deconvolution for rotating acoustic sources

Abstract

This paper introduces a practical time-domain beamforming framework for rotating source fields. We first reformulate classical frequency-domain delay-and-sum processing into a four-stage pipeline called \emph{Time--Space Beamformer} (TSB): (i) stationary-grid frequency-domain beamforming, (ii) inverse FFT to time-domain beamformer signals, (iii) spatial interpolation onto a rotating focus grid, and (iv) Welch spectral estimation for each rotating focus point. The result is a frequency-wise beamforming vector $\mathbf{b}(f)\in\mathbb{R}^{N}$ for $N$ rotating focus points. Based on this formulation, we propose an extended deconvolution method that constructs a frequency-dependent system matrix by evaluating synthetic point sources at all focus points and solves a non-negative sparse inverse problem. The method is evaluated on a fully reproducible synthetic rotating-source test case.