Feature ArticleWideband Sonar Mine Imaging In an Operational Environment
By R.J. Wyber
Midspar Systems Pty Ltd.
Oyster Bay, Australia
Midspar Systems Pty Ltd. recently carried out work on mine imaging in collaboration with the Australian Defence Science and Technology Organisation (DSTO). The aim of the project was to integrate advances in wideband technology in order to develop a system that would be capable of detecting, localizing and classifying all types of mines, including stealth mines, both on the seafloor and buried.
A fundamental part of the project was to understand the nature of the acoustic signals reflected from various mine types, and these were measured over a wide frequency band. From the data, it proved possible to easily model the impulse response returned from the mine. Using the impulse response measured in a controlled environment, various imaging algorithms were evaluated and adapted to the unique characteristic of the sonar signal. The sonar developed from this program, known as Safeport, has demonstrated a unique capability of detecting and classifying a stealth mine using only its fiberglass shell.
The final stage of the program was to develop a system capable of taking sonar measurements and overcoming the complexities of an operational environment on the floor of Sydney Harbour. While this paper primarily addresses the imaging capability of the sonar, a complete operational system was also designed that provides the initial mine detection and navigation to map the position of the imaged mines.
The impulse response from the Manta practice mine as the target rotates.
Signals Measuring Impulse Response
Three different sonars were used in the project, covering bands from one to 10 kilohertz, 22 to 66 kilohertz and 100 to 200 kilohertz.
To obtain a high signal-to-noise ratio, it is best to transmit a long broadband test signal. This may be compressed to a band-limited impulse by multiplying the frequency spectrum using a matched filter, which has the conjugate phase of the transmitted signal and an amplitude that combines with the system response to give the desired window for the band-limited impulse. For the measurements, a Kaiser-Bessel window was used, which allowed a trade-off between leakage of the impulse outside the main lobe and the width of the impulse to be optimized for the target.
While a linear frequency-modulated pulse is commonly used for sonar measurements, a linear period-modulated (LPM) pulse has the highly desirable property of being Doppler tolerant. This means that full processing gain is achieved and the impulse response is formed even in the presence of a Doppler shift in the received signal. The only effect of Doppler is to generate a time displacement in the impulse position that must be compensated for prior to combining multiple impulses to form an image.
This article presents results from the 100 to 200 kilohertz sonar, measured using the LPM pulse.
Target Impulse Response Features
When the wavelength is much smaller than the dimension of the target, the impulse can be well represented as the sum of the rigid body reflection and the structural response. The structural response is caused by mechanical waves being excited in the target and then reradiating as they propagate.
For this project, a Manta practice mine was used to simulate a stealth mine. At high frequencies, the impulse response of this mine is dominated by the rigid component. A very simple model allows the impulse response to be estimated as I(t)=k∂2A(t)/∂t2, where A(t) is the cross-sectional area of the target insonified at time t, ∂ denotes differentiation and k is a constant dependent on the medium.
At high frequencies, the impulse response is a sum of discrete impulses associated with specular facets and discontinuities.
Measuring Impulse Response
The system first underwent trials in a free-field test facility at Australia's Woronora Dam at a range of 60 meters, with the test sonar and Manta mine target mounted on separate barges. For these measurements, a pair of reference hydrophones were attached to the target to allow compensation for range variations between the barges and to measure the target aspect as it was rotated about its vertical axis.
In measuring the impulse response, a sonar with sufficient beam width was used to ensure that the entire target was insonified. A pair of hydrophones with vertical separation were used to receive the reflected signal. Due to the wide bandwidth of the sonars, a very high range resolution was achieved, and generally, there was only one impulse reflected from the target in each range cell.
As the Manta mine rotates about its vertical axis, the impulse response from the circular rim remains fixed in time. There is a central circular plate on the Manta from which the reflections would ideally also remain fixed in time; however, because this plate has a small offset from the true center, the position of the impulse from this plate has a sinusoidal variation in its position as the mine is rotated. The remaining impulses from the mine are from a number of point re'flectors on the Manta mine. As the target rotates 360°, the position of the impulses from these point reflectors trace out a single cycle of a sinusoid with an amplitude determined by the distance of the reflector from the axis of rotation.
The impulses tracing the sinusoids with the largest amplitude are the reflections from the reference hydrophones. These are suspended below the mine and extend slightly beyond the rim. The remaining impulses within the body of the mine are from the lifting eyes and internal ribs.
In general, only one impulse response is present in any range cell. When this occurs, the phase difference between the signals received at the two hydrophones is a measure of the height of the reflector on the target. This phase is represented by color in the measured impulse response, with the dark blue of the impulse reflected from the reference hydrophone being the lowest point and the red of the central plate being the highest point on the target.
The amplitude (a) and phase (b) of the Manta mine image formed at sea in Sydney Harbor.
Imaging of the Manta Practice Mine
By applying an inverse Radon transform to the impulse response measured by one of the hydrophones, a 2D image of the target is formed. In this image, all of the target highlights are projected onto a horizontal plane. The circular rim and central plate with its lugs are evident, and the lifting eyes and slings may also be seen. The remaining features are from internal ribs, which are visible in the impulse response due to the acoustic wave penetrating the outer shell of the mine.
If the phase of the impulse response measured between the hydrophones is used to form a set of impulses separated by height, a 2D image can be formed from each of these. The images formed at each height may then be combined to form a 3D image of the mine.
If a threshold for the reflected signal is set to reject the signals from the fiberglass skin of the mine, the internal ribs may be seen as four vertical triangular features distributed around the interior of the shell. The outer rim, the lifting eyes and the central plate are also shown at their respective heights.
If the threshold for the impulse response is lowered, the reflections from the outer fiberglass skin are included in the image. These shield the internal features, and the outer case of the mine is imaged. The two positioning hydrophones may also be seen suspended below the mine.
Operational Environment Trials
To evaluate the imaging algorithms in an operational environment, the Manta was deployed on the seafloor at the HMAS Penguin naval base in Sydney Harbour. This is a fairly harsh environment with a sandy bottom on which sea grass is growing, a significant swell comes through the heads and there is a strong velocity profile due to freshwater flowing from upstream. The depth of the water was six meters.
The tomographic measurements were made by circumnavigating the targets at a range of 50 meters. The sonar consisted of a single transmitter with a vertical beam width of 4° and a horizontal beam width of 30°. This allowed the beam to be maintained on the target as the vessel followed a circular track. Because of the wide bandwidth of the sonars, the reverberation was sufficiently reduced so that the targets could be detected with a single pulse, despite the wide beam width. Two receiving hydrophones, separated vertically by one meter, were used.
To form an image of the mine, it is necessary to align the impulse responses to compensate for time fluctuations caused by both the propagation through the water and the movement of the sonar. For the measurements of the Manta mine, this was done by aligning the impulses reflected from the rim of the target. While this alignment technique is not generally applicable, the alignment accuracy required to form these images is an order of magnitude less than that required to provide high side-lobe suppression with a synthetic aperture sonar.
It is anticipated that proven methods of aligning the data using the reverberation could be readily adapted for tomographic imaging. This was not investigated for these measurements, however, as a suitable multisensor receiving array was not available.
Due to the circumnavigation of the target, which insonifies all aspects of the mine, the outline is well-defined in the amplitude response. The plate in the center of the Manta mine is also visible. The remaining two points in the image are from fiberglass poles used to mount hydrophones on the target. The reverberation from the seafloor is also evident, both external to the mine and within the body of the mine where the signal penetrates the fiberglass shell without detecting it.
This practice mine is nominally a stealth mine, and the structure detected is from the rigid part of the mine at the rim, the center plate and the poles. This raises the issue as to how a stealth mine composed entirely from fiberglass could be detected. The solution to this was somewhat surprising and is an outstanding example of why experiments at sea are necessary.
When a 3D image was formed using only the phase of the reflected signal between the receiving hydrophones, it was expected to show the height of the dominant reflectors above the seafloor, as was observed in the test facility. Instead, it appears that the height of the sea grass on the ocean floor is imaged both inside and outside the mine. In the interior of the mine, this height appears raised by about 20 centimeters near the edge and 10 centimeters at the center. This is attributed to a disturbance of the sound wave as it propagates through the fiberglass shell of the mine. Refraction does not explain the amount by which the seafloor is raised. Modeling the flexural waves in the shell does, however, predict a coupling mechanism that not only predicts the observed increase in the apparent height of the seafloor, but also predicts the observed decrease in this height toward the center of the mine. This is potentially a mechanism that will allow true stealth mines to be detected and classified.
The Safeport sonar system was implemented in the frequency bands one to 10 kilohertz, 22 to 66 kilohertz and 100 to 200 kilohertz. In the lower of the frequency bands, where a parametric sonar was used, the successive pings received on a single hydrophone were correlated despite the movement of the hydrophone between pings. This allowed acoustic navigation to be used for positioning information and a complete operational system capable of detecting, imaging and mapping both proud and buried mines was implemented.
For a short time this system was available in Sydney Harbour. It was deployed operationally in the search for a World War II minisubmarine reported to be buried off the coast. At the higher frequencies, the system was proven using supplementary information to determine the position of the sonar.
Though the project was terminated by DSTO prior to completion of a high-frequency operational system capable of detecting stealth mines, a receiving array could be added to achieve this goal. This would allow the implementation of acoustic navigation methods using multiple receiving sensors.
The author gratefully acknowledges the contribution to the measurement of the data by the Maritime Operations Division of DSTO as well as John Shaw, Neil Tavener, Mark Savage, Chris Halliday, Ross Susic, Jane Cleary and Tony White. The author also acknowledges Brian Ferguson's contribution through ongoing discussions throughout the program. This work was completed as part of the DSTO's Capability and Technology Demonstrator program.
Ron Wyber received his B.Sc. in mathematics and his B.E. (Hons. 1) and Ph.D. in electrical engineering, all from Sydney University. After finishing his studies in 1974, he worked with the Australian Defence Science and Technology Organisation until 1987. He now operates a research and development company for acoustic and sonar systems, and his current work is primarily land-based acoustic systems with military applications.