Broadband frequency invariant beamforming method with low computational cost

This paper presents a method for reducing the computational complexity required for farfield broadband beamforming. First, the propagating wave received by a linear or planar array of sensors is sampled efficiently on the multidimensional frequency plane in spatio-temporal sense so that the so-called non-physical area where no spectrum exists is filled with aliasing component. Next, the derived time-space series is upsampled and processed by a multidimensional filter to derive the target beampattern. Through some considerations and examples, it is shown that this method has some restrictions on the beamwidth and/or maximum beam center angle, but it can reduce the operations to about half the number of conventional one.