Acoustic signaling and sound sources
for RAFOS navigation Signaling
Motivation
In this companion webpage we give a brief overview of the acoustic
signaling system used to provide navigation for SOFAR and RAFOS float
applications. In the early SOFAR float days the floats transmitted short CW
(continuous wave) pulses at 250 or 260 Hz. These signals, each about 1.2
seconds long, were sent exactly on the minute mark for 90 minutes every six
hours. The digital recorder at the listening stations, using a hydrophone
at the depth of the SOFAR channel (1200 meters deep), would record the
acoustic intensity in this narrow listening band. In the subsequent
processing on a computer, we basically searched for a pattern in which the
signal intensity increases regularly every minute. By noting when this
occurs, we know the arrival time to the second. By knowing the previous
location of the float, we can determine how many minutes to add to obtain
the total travel time. Figure 1 shows an example of this early signaling
system. The bright band in the center against the irregular gray background
shows the incoming signal. The scale here is one minute across days and 24
hours from top to bottom. The float was deployed at about 930am (Rossby and
Webb, 1970). This method worked well, but it didn't take advantage of the
fact the a signal transmitted in the ocean preserves its phase structure
(or pattern) for several minutes. In other words, it looked for the
presence of acoustic energy at a given frequency, but it didn't take
advantage of the shape of the signal.
In the mid-1970s Doug Webb proposed that we instead transmit a single long
signal, 80 seconds long, in which the frequency increases linearly 1.523 Hz
from start to end. At the receiver, we now listen not only in a certain
frequency band for the signal, but for one that has a specific phase
structure. The key to doing so is to have a replica of the 80 second signal
for comparison. By storing and comparing the most recent 80 seconds of
listening against the reference signal, we determine at what time the two
agree best within a given period of listening. This is illustrated in
Figure 2 which shows the ideal signal and several recorded signals. To
demonstrate for yourself how the signal detection method works, print the
figure and make a transparent copy on a copier machine. Then put the
transparency over the print and slide it slowly along the signal. You will
notice that there is only one position when the entire signal lines up
everywhere. At all other positions only short parts of the signal will line
up whereas other sections of the signal are out of phase. It is this best
line-up of the signal everywhere we look for. The detection scheme can be
further simplified if we keep only information on whether the signal is
positive or negative (all amplitude information in figure 2 is removed so
all that is left would be a square wave switching positive or negative
according to the zero crossings). The advantage of doing this is we only
need to work with a single bit of new information at each 0.1 second time
step, 0 for negative or 1 for positive. The detection technique becomes
what is known as an exclusive-or plus summation. The time step at which
this correlation differs most from random is considered to be the time of
arrival. The method works very well, and is easy to implement in small,
energy-efficient microprocessors. The latter development was a major factor
leading to the RAFOS float concept in which each float does its own
listening. Just a few moored SOFAR sound sources provide the navigation for
as many floats as desired.
Sound sources
The sound source used for all early RAFOS work consisted of the standard
SOFAR float, fabricated by Webb Research Corporation, Figure 3, at left. These are
moored at sound channel depth, typically around 1200 m depth. They are
typically programmed to send a signal once, twice or thrice per day
depending upon the requirements of program. Starting in 1993 we also have
used a source fabricated by Sparton of Canada, Figure 4, at right. This source is
more efficient and can insonify larger areas, and thanks to its dipole
radiation pattern it puts most of its energy in the horizontal plane at
sound channel depth such that less energy is lost towards the surface. This
source has been used in two major studies of the circulation in the
northern North Atlantic (see the NAC and ACCE webpages). It, too, has been
very successful. The electronics for the Sparton source come from Webb
Research corp.
Recently, we have developed a new source that has some similarity to the
Sparton source, but instead of using a piezoelectric ring around the
outside of the pipe to excite it, we suspend a monopole in the center of
the pipe. In both cases the active element excites pressure fluctuations in
the pipe, but it is only those at the right frequency, set by the length of
the pipe (which thus must be carefully trimmed to achieve resonance at the
desired frequency) that will excite standing waves in the pipe and thus
generate the desired acoustic output. This source was developed to meet the
needs for regional applications for RAFOS float studies at a lower cost.
Figure 5, at left, shows schematically the transducer and Figure 6, at right, shows the complete
system on deck awaiting deployment. This source is undergoing its first
tests and we plan to deploy it in its first prototype application in Spring
2001. When operational, this source will join the others in providing
acoustic navigation for various underwater applications.