Wideband Sonar - Plankton, Communications and Picket Fences

Phil Atkins(1), Claire Bongiovanni(1), Tim Collins(1), Jon Davies(2), Shaun Dunn(2), Ken Foote(3), Tor Knutsen (4) & Tom Mortensen(5)

(1)- University of Birmingham, UK

(2) - DERA, UK

(3) - Woods Hole, USA

(4) - IMR, Norway

(5) - Reson, Denmark

 

Introduction

A broadband sonar covering the frequency range 25 kHz to 3.2 MHz has been developed. The results obtained from the highest frequency band (1.6 MHz to 3.2 MHz) have been particularly encouraging and show that it is possible to detect individual plankton at ranges of up to a few metres. By examining the echo strength, target extent and scattering complexity it is possible to make simple classifiers to distinguish between diverse species such as copepods and euphausiids. It is possible that such a sonar could be engineered into a relatively small package for use on an ROV or variable-depth glider. The advantages of using a wideband signal are numerous. The most important is that a processing gain of tens of dB may be implemented, thus allowing the operator to detect small targets at ranges far beyond those achievable by a simple narrow-band CW system. The use of a broadband system implies that a good range resolution is achievable. For example, a pulse length of perhaps 0.5 m in the water may be readily compressed to a range resolution of less than 1 mm. Such a fine range resolution permits the use of time-domain target strength classification techniques. When the dimensions of the target are large in wavelengths then the scattered energy is frequency dependent. Thus a narrowband system may fail to detect a target if the operating frequency corresponds to a null in the scattering function. The frequency diversity characteristics of a broadband system (greater than one octave) ensure that such detection failures do not occur.

The next generation of plankton sonar will almost certainly be engineered to fit on an ROV, AUV, drifter or glider. Such an application may well require an acoustic communication system to telemeter the data back to the host vessel. With this application partly in mind, a broad-band communication system has been developed to transmit 20  k bits per second at ranges of up to 5 km. This is optimised for a shallow-water channel subject to impulsive noise and where both the source and receiver may be manoeuvring in an uncontrolled manner. The conventional approach to this problem is to use an adaptive equaliser updated from the results of the decision required in any digital communication system (A Decision Feedback Equaliser, or DFE). Thus the output of the equaliser tends towards the desired digital waveform and this is achieved by cancelling all but the strongest of the multi-paths. The equaliser is normally trained by transmitting a known training sequence prior to the data packet. Experience has shown that the noise characteristics of many shallow water channels are far from Gaussian and are dominated by impulsive noise sources. These impulsive noise sources tend to confuse a standard DFE leading to a high bit-error rate. A communication system has been developed based on a continuous training sequence. This improves the efficiency of the link, as data is transmitted continuously rather than in packets interspersed with training data. A second major advantage is that the characteristics of the channel may be continuously monitored and the best of the multi-paths selected. The recovery of the system following bursts of impulsive noise is both rapid and robust. Extensive shallow-water field trials have shown that an error rate following equalisation of about 3% can be expected. Thus a suitable data encryption and error correction strategy can be devised. A novel error-tolerant video compression algorithm has been developed for underwater applications and has been successfully demonstrated by attaching a transmitter pod to a small ROV.

The equaliser coefficients of the communication link receiver have a direct link to the channel characteristics. These coefficients are updated in real time and show high sensitivity to changes in the channel introduced by movements of the transmitter, receiver or significant scatterers. Work has started on implementing a multi-static sonar system using transmission waveforms optimised for low-Doppler targets such as whales and small marine mammals. Such a sonar falls into the category of a 'picket fence' or a 'field anomaly sonar' and is based on the analysis of variations in the channel characteristics over a period of time. The technology required is that of multiple communication link pressure vessels tethered at mid-water depth and cabled back to a shore observation station.

Click here to see slide show

 

 

This page is maintained and was last updated 23-Nov-00 by Pip Sumsion (sumsionp@dfo-mpo.gc.ca).