High Frequency Performance of Arc Arrays Using Adaptive Beamforming

A baffled arc array (BAA) comprises a circular array of sensors surrounding a cylindrical metal baffle, and beamforming is carried out by processing sensors on an arc centred on beamsteer. It is widely used in sonar for passive underwater surveillance. An important factor that limits the performance of the array at high frequencies is the appearance of grating lobes in the beamformer response. This occurs when sensor separation is greater than half a wavelength, and leads to masking of weak signals by the grating lobes of strong signals. In the case of the BAA, the grating lobes are imperfect as a consequence of the circular geometry; there is a mismatch between array responses to signals arriving from the beamsteer and grating lobe directions. Adaptive beamforming is sensitive to this mismatch, and suppress the grating lobes, thus extending the frequency range of the BAA beyond its design frequency. The purpose of this paper is to demonstrate the effectiveness of adaptive beamforming for suppressing grating lobe effects in the BAA. INTRODUCTION Beamforming of sensor array data is used in radar, sonar and communications (Van Trees 2002). In passive sonar (Barger 1997) it is used to increase signal-to-noise ratio (SNR) of underwater acoustic signals and to indicate signal bearings. As a consequence of the discrete sampling of the wavefield by the array, grating lobes appear in the beamformer response when sensor separation is greater than half a wavelength. In the case of a uniform linear array (Van Trees 2002), the array response is identical for signals arriving from beamsteer and grating lobe directions. For a curved array, however, the array responses are distinguishable. Grating lobes still appear in the beamformer response, but are imperfect in the sense that they are not at the same level and shape as the main-lobe. An example of a curved array that is widely used in sonar is the baffled arc array (BAA). This is used to provide 360 coverage of bearings in passive underwater surveillance. It is similar to a uniform circular array (Van Trees 2002, Davis 1983) except that the sensors surround a cylindrical metal baffle, and only sensors on an arc centred on beamsteer are processed. The primary purpose of this paper is to demonstrate the effectiveness of adaptive beamformers at suppressing grating lobe effects in BAAs. Grating lobe suppression is a consequence of the sensitivity of the algorithms to mismatch between the array responses to signals arriving from the beamsteer and grating lobe directions, and is closely related to sidelobe suppression. It improves the detection of weak signals masked by strong interferers, and leads to more effective array operation when sensor separation on the arc is greater than half a wavelength. BAFFLED ARC ARRAY (BAA) Let Q sensors be equi-spaced around a cylindrical metal baffle. In sonar, the sensors are often physically offset (displaced) from the baffle surface, to isolate them from baffle vibrations and improve sensor coupling to the water. Acoustic signals are assumed to arrive with zero-elevation; i.e. perpendicular to the cylinder axis. In practice, each sensor (or stave) is a linear array of omnidirectional hydrophones, aligned with the cylinder axis and having phone outputs summed with zero time-delay. Thus at each stave, the phone outputs add coherently for the signal, while ambient noise from non-zero elevations is suppressed. Only M < Q staves on a cylindrical arc centred on beamsteer are processed. As the array is steered to provide 360 coverage of azimuths at zero-elevation, the arc swings around with it and selects new staves for processing. Figure 1 schematically shows the m stave and the coordinate system for the processing.