Continued development of broadband passive bearing estimation

One of the major challenges in open-ocean deployment of unmanned underwater vehicles (UUVs) or unmanned surface vehicles (USVs) is collision avoidance with approaching surface shipping. Existing methods assume that the underwater robot carries a radar and/or has access via radio to commercial Automatic Identification System (AIS) information. However, radars are expensive and both radars and radio can only operate above the water. Moreover, AIS status is not always broadcast by passing shipping. An alternative solution may be found through the use of underwater acoustics whereby a sonar system determines the direction of arrival for a surface ship based upon the ship´s radiated noise. The effectiveness of this method lies in the use of as much of the radiated frequency spectrum as is practically accessible, hence a broadband sonar is preferred. We refer to our device as the Passive Bearing Detector (PBD). Ultra-short baseline (USBL) is a well-known interferometric method for estimating the direction of arrival for known narrowband acoustic signals. In this paper, we discuss a method that supports direction of arrival estimation for any wideband source of acoustic energy without a priori knowledge of the specific temporal or spectral characteristics of the signal. In support of this, we build an array of several listening hydrophones that are spaced at a pre-determined distance in a known geometry. The spacing among hydrophones is dictated by the upper limits of the frequency band, while the available bandwidth then is limited by the mismatch between array aperture and source wavelength. Our broadband interferometric algorithm provides estimates of azimuthal arrival angle, but the accuracy of the algorithm for such a small and compact array is highly dependent on variable channel conditions. We compare the performance of conventional and adaptive beamforming techniques applied to PBD data recorded at sea. We demonstrate that our approach is far less computationally intensive, but appears to perform as well as or better than the beamformers. An extensive sea trial was conducted at an acoustic test range in Northern Europe. We review performance of our system and further demonstrate the qualities of a bearings-only tracker and the (future) ability to autonomously extract snippets of time series, conduct time-series analyses, compress the results, and provide the result for telemetry via our integrated acoustic communications.