|
Chris Jones, Senior Engineer
Ocean Acoustics Department
Applied Physics Laboratory
University of Washington
National Science Foundation
|
Research objectives This effort is a step towards improving multi-channel digital sonar technology for implementing new processing methods and imaging applications in the ocean. The target application for this instrument is measuring acoustic scatter from weak volume fluctuation in the deep-ocean water volume (e.g., turbulent flow, suspended biological matter). The effort is a significant improvement over commercially available instruments. This project provided a unique educational opportunity and training for several electrical engineer graduate students at the University of Washington with interdisciplinary study in engineering and oceanography. 
Figure 1 shows the final system on the bench. The array is shown attached to a cylindrical underwater housing (22 inches long, 7.5 inches dia.). The array and housing are capable of full ocean depth (4000 m) operation. The top-level processor is a CPU running Linux with full network connectivity. The external interface to the sonar is a cable with power and Internet. A graphical user interface has been developed separately (by Rutgers University) for ease of operation and data visualization. The lowest level system interface is a Linux SSH or Telnet terminal and data transfer FTP. This level of system control is critical for cabled observatory applications, enabling direct access by researchers and customization of the application program control and data processing. System overview Onboard signal and image processing is implemented with a cascading mixed-signal design utilizing sigma-delta A/Ds, FPGAs, DSPS, and finally a Linux CPU module. Software is implemented in C and fully reconfigurable over the network. The hardware is a combination of custom mixed signal PCBs and commercial PC/104 components. The custom electronics includes 96 channels of sigma-delta analog to digital conversion with a maximum digitization rate of 1.2 Mhz per channel. Eight custom ADC boards were designed and fabricated with each board processing 12 analog channels. Each A/D board has a reconfigurable FPGA and 128 Mbytes of memory for data control, buffering, and initial processing. Several data processing schemes are implemented to allow continuous data streaming to disk (using a fast form of data base-banding) or non-continuous "burst mode" operation were 10 pings can be stored in memory and downloaded to the CPU allowing the acoustic ping rate to be decoupled from the data telemetry rates. Flexibility in data acquisition, control, and processing is critical in the design of scientific experiments using multi-beam acoustics systems, mainly because data rates can be very high (96 x 1.2 Mbytes/sec). The approach here has been to provide three layers of processing (FPGA, DSP, and PC) with data buffered at each stage using RAM and a hard disk at the last stage (up to 120 GB presently). Data is accessed via TCP-IP (FTP, SSH, or custom socket interface). A primary design concern for this system has been reducing electronic system noise and increasing the signal-to-noise ratio (SNR) of the acoustic data. This is a major drawback of exiting commercial systems and a limiting factor when measuring acoustic scattering from weak targets in the water column, such a turbulent microstructure or small suspended particles. In addition to electronic noise, much effort has been better side-lobe performance for the array will increase signal to noise performance. Although actual signal to noise measurements can only be taken in-situ with real weak targets, estimates of electrical noise for the system indicate significant improvement over the SM2000 multi-beam system (the only comparable commercial system). Further analysis of SNR and calibration are currently being performed at APL-UW on the system with continuation funding. |
System testing and calibration In-water testing and calibration of the system was performed at the APL-UW acoustic test facility in Seattle. Calibrations include beam pattern analysis and sensitivity measurements. Figure 2 illustrates the measured beam patterns for steering narrow beams over a 180 degree sector in front of the array. In this test the beams are streered at 10 degree increments over a –60 to +60 angular sector to illustrate the range of beam widths and beam patterns. Since digital beamforming is performed, beams can be steered in arbitrary directions. All beams form simultaneously. The beams show good side-lobe response (–13 dB) with –3 dB beam widths ranging from 1.5 to 2.0 degrees for the center beams and increasing to a maximum of 4.0 degrees for the outer beams. This calibration was performed using 72 elements of the array, achieving the predicted results. A maximum of 96 channels can be recorded, which will reduce the beam widths and/or side-lobe levels. Figure 3 shows a higher angular resolution test of the beam patterns illustrating the side-lobes and beam shapes near the center of the azimuthal sector. Figure 4 illustrates the beams in polar coordinates. 
Publications and products D.R. Jackson, C.D. Jones, P.A. Rona and K.G. Bemis, "A method for acoustic measurement of black smoker flow fields", Geochem. Geophys. Geosyst., 4, 1095 (2003). "Water-colum measurements of hydrothermal vent flow and particulate concentration using multibeam sonar," C. D. Jones, J. Acoust. Soc. Amer., 114(4), October 2003.  Printable project summary |