A novel deconvolution beamforming algorithm for virtual phased arrays

Beamforming techniques using phased microphone arrays are one of the most common tools for localizing and quantifying noise sources. However, the use of such devices can result in a series of well-known disadvantages regarding, for instance, their very high cost or transducer mismatch. Virtual Phased Arrays
(VPAs) have been proposed as an alternative solution to prevent these difficulties provided the sound field is time stationary. Several frequency domain beamforming techniques can be adapted to only use the relative phase between a fixed and a moving transducer. Therefore the results traditionally obtained using large arrays can be emulated by applying beamforming algorithms to data acquired from only two sensors. This paper presents a novel beamforming algorithm which uses a deconvolution approach to strongly reduce the presence of side lobes. A series of synthetic noise sources with negative source strength are introduced in order to maximize the dynamic range of the beamforming deconvolved map. This iterative sidelobe cleaner algorithm (ISCA) does not require the of use of the covariance matrix of the array, hence it can also be applied to a VPA. The performance of ISCA is compared throughout several simulations with conventional deconvolution algorithms such as DAMAS and NNLS. The results support the robustness and accuracy of the proposed approach, providing clear localization maps in all the conditions evaluated.