Linear Filter Theory for Circular Array Processing

G. Vettori

DOI: https://doi.org/10.1121/1.2143503

1966-11-01

The Journal of the Acoustical Society of America

Abstract:A transparent circular array in monochromatic plane waves can be regarded as a “linear spatial filter” in which the contour phase function f(φ−θ), due to a plane wave from direction θ, is the filter impulse response when the plane wave is represented by a delta function δ(φ−θ). The spatial frequency spectrum F(s) [the Fourier transform of f(φ)] is saddleshaped and limited to a bandwidth 2R/lgr;. The angle-spatial bandwidth product is then 4πR/λ, analogously to time-frequency bandwidth product. A spatial processing filter, of impulse response h(φ), can be put in “series” to synthesize a desired directly function g(φ) at the output, whose spectrum G(s) satisfied the above bandwidth limitation. Function h(φ) is then obtainable as the inverse transform of H(s)=G(s)/F(s). Optimization of H(s) for maximum signal-to-background ratio for a signal source in isotropic monochromatic reverberation [describable at the array input as white (in s) noise] leads to a matched filter with H(s)=F(s)*/|F(s)|2 that whitens the spectrum measured at the contour and gives a processing gain of 4πR/λ. Then g(φ) is a sin x/x function.