DOLPHIN would be transiting from San Diego in late January and the possibility existed
for additional SUBSAT tests. A test plan (Appendix B) was devised for rendezvous with the
DOLPHIN off San Clemen te Island with NOSC's Box L boat as the surface vessel. The
DOLPHIN would proceed to make a series of straight-line passes designed to pass directly
beneath the Box L boat. NOSC civilian personnel aboard both boats would originate and
receive the slow-scan transmissions at various points along each pass.
Equipment installation on the Box L and DOLPHIN was completed by 21 January.
The DOLPHIN setup was very similar to that on SEACLIFF, with camera, monitor, recorder,
and Robot 400 being brought aboard and interfaced with each other and the DOLPHIN'S
ATM504A UQC. Since the Box L had no UQC, an ATM504A was installed there in addition
to the slow-scan equipment. Two SP23LT conical transducers, one with baffle and one with-
out, were taken aboard. Using each transducer at depths of 20 and 200 ft would allow four
12
different receiving conditions to be tested. A 4- to 6-kHz noise source was brought aboard
Box L to serve as a beacon for DOLPHIN to align itself on prior to beginning a run. In
addition, a 9-kHz acoustic pinger system was installed aboard Box L and DOLPHIN. Prior
synchronization of these units with WWV would, in principle, allow continuous measurement
of slant range between these two platforms.
Box L left San Diego on 24 January for Wilson Cove, San Clemente Island, where it
spent the night. DOLPHIN left San Diego with NOSC SUBSAT personnel aboard later that
day for its rendezvous with Box L the next morning off San Clemente Island. Box L left
Wilson Cove at 0600, 25 January and rendezvoused with DOLPHIN off the southeast coast of
San Clemente Island. DOLPHIN dived in 2620 ft of water at 0825.
At 0923 DOLPHIN was at one-half test depth, the desired operating depth, and
began her first run. The actual test depths of DOLPHIN are classified and therefore all depths
are referred to here as fractions of test depth. At 1000 it became evident to DOLPHIN
personnel that DOLPHIN was not on the right course, and she requested Box L to put the
broadband noise source in the water. At 1020 a second run was begun and two slow-scan
transmissions were made. Since the slant range to DOLPHIN was opening rather than closing,
this run was aborted after the second transmission. At 1045 DOLPHIN turned to begin its
third run. Four video transmissions were made and again indications were that the range was
opening rather than closing, since the transmission which was supposed to have occurred at a
horizontal range of ft actually occurred at a horizontal range in excess of 5000 ft. Two
more runs were attempted, the second of these at one-quarter test depth. Although the CPA
of these runs came closer to directly below the Box L, it was evident from voice range marks
and the variation of the received signal strength that even these runs occurred at a consider-
able horizontal offset. The test ended at 1310.
A total of 13 slow-scan transmissions were made and received during the DOLPHIN
tests. All transmissions were received with the baffled hydrophone at a depth of 200 ft, since
there were not enough close-in runs to compare different hydrophones and depths. Although
a TV camera was provided aboard the DOLPHIN, all transmissions were from pre-recorded
tape (see Procedure B, Appendix A).
Post-exercise debriefing of personnel aboard DOLPHIN revealed that neither the 9-kHz
pinger nor the 4- to 6-kHz noise source could be used as intended. Throughout the exercise
no valid range reading could be obtained from the 9-kHz pinger, and the only range information
available was from intermittent and relatively inaccurate oral time marks. The broadband
noise source apparently suffered from multipath, and this — coupled with other noise sources
in the vicinity (i.e., a fishing boat) — led to grossly inaccurate bearings on the first three
passes. The last two passes apparently began on the proper headings, but from the measured
ranges near the end of the runs it was apparent that DOLPHIN did not pass directly beneath
Box L. Thus, just as in the SEACLIFF experiment, most acoustic transmissions from
DOLPHIN were at angles far from the vertical, that is, far from the main lobe of the sending
and receiving transducers.
3. RESULTS
3.1 EXAMPLES OF SUBSAT VIDEO
A total of 26 video transmissions, each consisting of a number of pictures, were sent
and received during the DOLPHIN and SEACLIFF operations. Although each picture was
13
"received" in the sense that observing the monitor indicated the presence of a slow-scan pic-
ture, most of the pictures suffered from distortion and were severely degraded below labor-
atory quality (see fig. 4). The primary source of distortion was loss or displacement of hori-
zontal lines. This effect has been seen previously in both the CUTLINK experiments (Ref. 3)
and amateur radio transmission (Ref. 7). Both of these references attribute line loss or dis-
placement to multipath arrivals that affect the line sync signal. In our experiments the severity
of line sync problems varied widely. In a few worst-case examples, half the lines were missing,
resulting in a picture highly compressed in the vertical direction. In the best pictures (e.g.,
fig. 8b) only a single line displacement occurred out of 120 transmitted lines. The LOFAR-
GRAM analysis of the next section also shows the effects of multipath on the received power
spectrum.
Commenting on his CUTLINK results, Mr. Guthals of Ball Brothers Research Corp-
oration noted that in their slow-scan images "distortion was displayed in the form of ghost
images, waviness, line sync irregularities and granulation" (Ref. 3). It is interesting to note
that aside from "waviness," all these effects were noted during our experiments. Next to line
sync irregularities, graininess was the most persistent distortion observed in our pictures.
Graininess is generally attributed to a low signal-to-noise ratio. The LOFARGRAM analysis
below indicates that in our case the "noise" is probably signal multipafhs. Just as in photo-
graphy, the effect of graininess is to lower resolution. Unlike line sync irregularities, which
produce an abrupt and total loss of information (as far as the eye is concerned), graininess
causes a gradual deterioration as the signal-to-noise ratio degrades. Ghost images, also caused
by multipath in time synchronization with the main signal, were barely in evidence in some
of the pictures and proved the least objectionable of the various distortions.
Figures 5a, 5b, and 5c are SUBSAT pictures received aboard Box L during the second
transmission of the third pass while DOLPHIN was at one-half test depth. Figure 5a, the gray
scale, shows almost no sync problems but is considerably more grainy than the original (see
fig. 4a) as are 5b and 5c. The girl, fig. 5b, shows numerous line sync problems due to an
early sync trigger. Indeed, in the line tear near the middle of the picture the actual sync
pulse appears as a block line segment about one-quarter the way across the screen. Early
sync triggers are the major line sync irregularities in figs. 5a and 5b, though some jitter due to
late sync triggering is evident upon close inspection.
(a) Gray scale (b) Girl (c) Test pattern
Figure 5. Pictures from DOLPHIN at one-half test depth.
14
Figures 6a, b, and c are the received video recorded during DOLPHIN's fifth pass
when she was at one-quarter test depth. Note the generally improved picture quality, espec-
ially the decreased graininess. It is not clear whether the decreased depth or a fortuitous
decrease, in lateral range accounted for the superior imagery. Some ghosting, evidenced by
faint vertical lines near the left edge of figure 6b, is apparent.
(a) Gray scale (b) Girl (c) Test pattern
Figure 6. Pictures from DOLPHIN at one-quarter test depth.
Figure 7 is gray scale received aboard MAXINE D and transmitted during SEACLIFF's
fourth transmission, when she was on the bottom at 3755 ft. Although this picture exhibits
sync problems and granulation, it is the best video obtained from so great a depth.
Figure 7. Gray scale from SEACLIFF at 3755 ft.
Figures 8a and b show live camera video sent from SEACLIFF from a depth of 2000
ft according to procedure C of Appendix A. Figure 8a is the picture as sent from the
SEACLIFF. It shows a portion of the SEACLIFF's sonar screen and electronics console and
was produced from the tape aboard the SEACLIFF, which recorded the locally generated
live camera video. Figure 8b is this same scene as received aboard MAXINE D. This picture
15
(a) SEACLIFF. (b) MAXINE D.
Figure 8. Live camera video from SEACLIFF at 2000 feet.
exhibits only one line sync problem plus very slight granulation and ghosting. Comparison
with the locally recorded frame, fig. 8a, shows that very little visual information has been
lost. Fig. 8b represents the highest quality acoustically transmitted picture observed during
the SEACLIFF or DOLPHIN experiments.
3.2 LOFARGR AM ANALYSIS
All cassette tapes of received video during both experiments, as well as the tapes of
live camera video were subjected to LOFARGRAM analysis. Both problems with the hard-
ware, such as excessive tape recorder flutter, and problems with acoustic propagation can
often be detected by such an analysis.
Figure 9a is a LOFARGRAM of Procedure D, the GREY SCALE transmission by
SEACLIFF from a 1000-ft depth. This figure is the analysis of the locally recorded signal
aboard SEACLIFF, i.e., the video prior to entering the acoustic path.
Since GREY SCALE is a sync pulse (1200 Hz) plus four other discrete frequencies,
it might be expected that the LOFARGRAM would consist of the five discrete lines corres-
ponding to these frequencies. What is actually observed is five frequency bands; each band
is split into equally spaced components modulated by an envelope function which decays
outwards from the center of each band. A little thought reveals that the splitting is due to
the analysis of a series of identical slow-scan lines, i.e., the splitting occurs at the line rate of
1 5 Hz. The envelope function is a (sinc)^ function having a width inversely dependent on the
duration of the individual frequency burst. Note the envelope of the 5-msec sync tone is
much wider than that associated with any of the 16-msec GREY SCALE tones.
Figure 9b is a LOFARGRAM of the same transmission represented in fig. 9a but
as received and recorded aboard the MAXINE D. Figure 9b shows obvious amplitude varia-
tion throughout in the form of many white, nearly horizontal interference bands, which are
best seen by viewing the figure edge-on from the side. These interference regions are char-
acteristic of "selective fading," which is caused by destructive multipath interference. Al-
though the selective fading was most evident in fig. 9b, because of the simple and repetitive
nature of its video content, it was noted on most of the DOLPHIN and SEACLIFF LOFAR-
GRAMS, indicating that multipath effects were a main factor in degrading picture quality.
16
•;
17
3.3 SURFACE-BOTTOM MULTIPATH
The presence of a 9-kHz pinger aboard the Box L during the DOLPHIN experiment
allowed measurement of the normal-incidence bottom reflection coefficient. This in turn
allows the calculation of the relative magnitude of the direct-path slow-scan signal to higher
order multipaths that reflect off the surface and bottom. This calculation will be presented
in this section and will show that the direct-to-multipath ratio can degrade rapidly as the
direct path departs from vertical.
During the period when the 9-kHz pinging was in operation, the geometry was as in
fig. 10a. Pings were transmitted from the 200-ft-deep baffled conical transducer and received
by another conical transducer near the surface and recorded. Figure 10a shows the direct
path, while figs. 10b, 10c, and lOd show multipaths with single, double, and triple bottom
reflections.
(a) PINGER GEOMETRY
SHOWING DIRECT PATH
) / ) I ) 7 V 1 1 1 1 / 1 //
(bl FIRST MULTIPATH
!c> SECOND MULTIPATH (d) THIRD MULTIPATH
Figure 10. 9-kHz pinger geometry.
A segment of tape containing in excess of 20 pings was played back onto a storage
scope. The horizontal sweep was triggered on the direct-path arrival. Figure 1 1 is the
resulting storage-scope record. The three pulses, from left to right are the returns due to one,
two, and three bottom reflections. The spacing between the center pulse and pulses on either
side corresponds to twice the 2620-ft water depth. After subtracting the background, we find
the amplitudes of the three pings to be in the ratio 4.38:1.13:0.348. The decibel difference
between the first and second is 1 1.77 dB and 10.22 dB between the second and third.
The difference in amplitude between the first-second and second-third pulses is due
to a surface reflection loss, a round trip from surface to bottom to surface, and a bottom
reflection loss. Using the known depth and an absorption coefficient of 0.8 dB/kiloyard,
we calculate a combined surface-bottom loss of 2.97 dB from the difference between the first
and second pings and 5.28 dB between the second and third pings, yielding an average of
4.13 dB for the combined normal-incidence bottom and surface reflection losses. Consider-
ing the glassy seas experienced during the DOLPHIN experiment, this figure agrees well with
coastal type bottom loss measurements (Ref. 6).
By means of the transducer directivity response and the measured bottom loss, the
transmission loss for the direct path or any multipath can be computed for the SUBSAT
geometry. In particular, we will calculate the signal-to-noise ratio between the direct path
18
j 1 ... - —4MMM â–
'*• — * • • ■- f*'
v*;": ' .-■•-.••; ^- ■;/-;;■.'
Figure 1 1. Storage scope record at 9 kHz. Pings 0.5 sec per division.
Figure shows direct path and three lowest order multipaths.
and the multipath arising from a single reflection off the surface and bottom. The calcula-
tion was done for a combined surface bottom loss of 4.13 dB (assumed independent of angle)
at 10 kHz for a slow-scan source near the bottom transmitting to a near-surface receiver.
Although these conditions do not apply directly to either experiment, they serve to illustrate
the effect of rapidly increasing multipath with horizontal offset. Figure 12 is the result of
this calculation and shows the rapid dropoff of the signal-to-noise ratio as the direct path
departs from the vertical (i.e., as the horizontal offset increases). In order to maintain a
signal-to-multipath level sufficient to yield reasonable picture quality (10 dB, for example)
it is necessary to keep the vertical angle less than about 30 deg. Control over horizontal
offset was not possible during the SEACLIFF and DOLPHIN experiments and leads us to
conclude that multipath contributed to the observed degradation in picture quality.
19
20 30 40
, ANGLE FROM VERTICAL, deg
DIRECT PATH
MULTIPATH
Figure 12. Signal to multipath ratio as a function of angle from the vertical.
4. CONCLUSION
The FY 77 SUBSAT experiments presented herein have shown the feasibility of trans-
mitting 7200-baud acoustic slow-scan television using off-the-shelf equipment to generate and
receive the slow-scan video in conjunction with existing UQC gear. The cost of the slow-
scan equipment was approximately $ 1000 per transmitting or receiving installation. Equip-
ment installation can be accomplished within an hour by simple cable interconnection. No
modification of the UQC is necessary nor is its usual voice function disabled. In 26 separate
acoustic video transmissions during the SEACLIFF and DOLPHIN tests no hardware failures
were encountered.
All of the acoustic video transmitted from depth was received on the surface; received
in the sense that from the marking of the video monitor it was clear that a video trans-
mission was occurring. However, most of the received transmissions were noticeably degraded
compared to the transmitted picture quality, the most persistent problems being line sync
failures and granulation. Evidence gathered during the experiments suggests that the observed
image degradations were due to a low signal-to-multipath ratio, which in turn was caused by
a large horizontal offset between sending and receiving platforms.
The FY 1978 SUBSAT program, already underway, will be directed at obtaining
additional data under more carefully controlled conditions. In particular it will attempt to
show what is currently predicted from the FY 1977 experiments: that useful acoustic slow-
scan imagery can be transmitted reliably, provided that undue horizontal offsets between
sending and receiving platforms can be avoided. Such a demonstration would validate the
SUBSAT hardware approach and indicate that a video imagery capability can be added to
existing UQCs at modest cost by using off-the-shelf units.
20
REFERENCES
1. D. W. Burrows, "Cableless Underwater Television Link, Design and Test Results,"
paper presented at Oceanology International '69, Brighton England, Feb 16, 1969.
2. "Acoustic Transmission of Underwater TV," J. Acoust. Soc. Am., vol. 45, no. 4, 1 969,
p. 1060.
3. D. L. Guthals, "Results of Television Transmitted Acoustically via a Water Medium,"
J. Acoust. Soc. Am. , vol. 47, no. l,part 1, 1970, p. 124.
4. Industrial Research, vol. 12, no. 12, Dec 1970, p. 59.
5. W. L. Konrad, "Underwater Acoustic Television Transmission via the Parametric Source,
J. Acoust. Soc. Am. , vol. 60, supp. 1, Fall 1976, p. 599.
6. R. J. Urick, Principles of Underwater Sound for Engineers, McGraw-Hill, New York,
N.Y., 1967, p. 118.
7. C. MacDonald, "Line-to-Line Jitter," CQ. Sept 1974, p. 50.
21
APPENDIX A-SLOW-SCAN TV TEST OPERATING INSTRUCTIONS
TO SEND VOICE AT ANY TIME
1 . Turn TRANSMIT SELECT on Robot to VOICE.
2. Use UQC as usual.
TO SEND VIDEO FROM TAPE *
1. Make sure cassette recorder is on and TEST TAPE is rewound and ready. If not
turn POWER (T 1) on then STOP (Til) then EJECT (T 12) and insert TEST
TAPE. Close lid and hit REWIND (T 7) until tape is rewound. Hit STOP.
2. On cassette recorder make sure CRO2 (T 3) is down and DOLBY (T 2) is up.
3. Adjust OUTPUT (T 5) on cassette recorder to match red index.
4. On Robot make sure DISPLAY (R2), POWER (R 4) and MEMORY INPUT (R 5)
toggles are up and MEMORY INPUT (R 7) switch is at TAPE.
5 . Make sure WIDTH (RIO), RECEIVE CONTRAST (R 9) and RECEIVE
BRIGHTNESS (R 8) on Robot are pointing to red index.
6. Turn monitor on and align to red indices on all front panel controls.
7. You are now ready to send video. To send video hit UQC mike button (push to
talk button must be held on for duration of video transmission) turn Robot
TRANSMIT SELECT (R 1) to TAPE and depress PLAY (T 9) on tape recorder.
Pictures will appear on monitor in a few seconds.
8. Terminate video by pressing STOP (T 1 1) on cassette recorder and turning
TRANSMIT SELECT (R 1) on Robot to VOICE (UQC normal operation is now
enabled).
9. Rewind tape by pressing REWIND (T 7) on cassette recorder and STOP when
rewound.
TO SEND VIDEO FROM CAMERA WHILE RECORDING **
1. Make sure a RECORD TAPE is in cassette recorder and recorder is stopped. If
not depress STOP (Til) then EJECT (T 12) and insert RECORD TAPE. Do not
rewind RECORD TAPES. Use RECORD TAPES in indicated order as needed.
2. Make sure on cassette recorder POWER (T 1) and CRO2 (T 3) are down and
DOLBY (T 2) is up.
3. Align LEFT RECORD LEVEL (T 4) slide control on cassette recorder to red
index.
4. On camera make sure POWER is on, ALC is ON and SYNC is LINE.
5. Make sure monitor is on and all front panel controls are aligned to red indices.
"Steps 1 through 9 are required for initial tape transmissions. For each subsequent tape transmission only
steps 7 through 9 are required provided that only voice transmissions have occurred since last tape trans-
mission.
*Steps 1 through 12 are required for initial transmissions from camera. For each subsequent transmission
from camera only steps 8 through 12 are required provided that only voice transmissions have occurred
since last camera transmission.
23
6. On Robot adjust SNATCH CONTRAST (R 13) and SNATCH BRIGHTNESS
(R 12) and WIDTH (R 10) to red indices.
7. Make sure POWER (R 4) and MEMORY INPUT (R 5) toggles are up, DISPLAY
(R 2) toggle down and MEMORY INPUT (R 7) on CAMERA.
8. Point camera at scene and adjust its LENS for sharpest image on monitor. Adjust
Robot SNATCH BRIGHTNESS (R 12) for good overall brightness on monitor.
Adjust SNATCH CONTRAST (R 13) for most pleasing picture.
9. You are now ready to send and record camera video. To do this depress RECORD
(T 6) and PLAY (T 9) simultaneously on cassette recorder, turn TRANSMIT
SELECT (R 1) to MEMORY and hit UQC mike button (push to talk button must
be held on for duration of video transmission).
10. If necessary adjust LEFT RECORD LEVEL (T 4) to indicate 50% of VU meter
(T 13).
1 1 . Terminate video by hitting STOP (T 1 1 ) on cassette recorder and changing
TRANSMIT SELECT (R 1) on Robot to VOICE. (UQC normal operation is now
enabled.)
12. Do not rewind RECORD TAPE.
D. TO SEND GRAY SCALE WHILE RECORDING*
1 . Make sure a RECORD TAPE is in cassette recorder and recorder is stopped. If
not depress STOP (Til) then EJECT (T 12) and insert RECORD TAPE. Do
not rewind RECORD TAPES. Use RECORD TAPES in indicated order as needed.
2. Make sure on cassette recorder POWER (T 1) and CRO2 (T 3) are down and
DOLBY (T 2) is up.
3. Align LEFT RECORD LEVEL (T 4) slide control on cassette recorder to red index.
4. On Robot make sure DISPLAY (R 2), POWER (R 4) and MEMORY INPUT (R 5)
toggles are up.
5. Make sure monitor is on and all front panel controls are aligned to red indices.
6. You are now ready to send and record gray scale video. To do this depress
RECORD (T 6) and PLAY (T 9) simultaneously on cassette recorder, turn
TRANSMIT SELECT (R 1) to MEMORY and hit UQC mike button (push to talk
button must be held on for duration of video transmission).
7. If necessary adjust LEFT RECORD LEVEL (T 4) to indicate 50% of VU meter
(T 13).
8. Terminate video by hitting STOP (T 1 1) on cassette recorder and changing
TRANSMIT SELECT (R 1) on Robot to VOICE. (UQC normal operation is now
enabled.)
9. Do not rewind RECORD TAPE.
"Steps 1 through 9 are required for initial transmissions of gray scale. For each subsequent transmission
of gray scale only steps 6 through 9 are required provided that only voice transmissions have occurred
since last gray scale transmission.
24
APPENDIX B - TEST PLAN FOR SUBSAT DOLPHIN OPS - 19 JAN 1976
1. PURPOSE - The purpose of this test is to investigate what conditions if any,
are suitable for transmitting slow-scan video over a standard UQC from a submerged platform
to a near-surface transducer. Transducer depth, d, transducer baffling, and horizontal offset
will be varied to ascertain what conditions are necessary for good video quality.
2. EQUIPMENT — Video will be transmitted by the submerged platform
(DOLPHIN) and received by the surface support craft (BOX-L). NUC Code 65 1 1 will provide
and interconnect its equipment to the UQC's aboard DOLPHIN and BOX-L. Each installa-
tion weighs approximately 35 lb, occupies 2 cu ft and draws 0.7 A of 1 15 VAC. NUC Code
2563 is requested to provide aboard BOX-L: (1) UQC with TIPE option, (2) UQC compatible
transducer with removable baffle capable of being deployed at either 20 ft or 200 ft, (3) a
secondary method (besides UQC TIPE) for allowing DOLPHIN to ascertain slant range to
BOX-L, and (4) a method for allowing DOLPHIN to obtain bearing of BOX-L.
3. MEASUREMENT RUNS (See fig. B-l) — Each measurement run will take place
with DOLPHIN at one-half test depth. For each run DOLPHIN will start at a horizontal
range somewhat in excess of 2000 ft. It will obtain bearing to BOX-L and come about to
heading so as to pass vertically beneath BOX-L. It will then proceed at 2 kt on this heading
until it has completed sending five video transmissions. Video transmissions will be initiated
at horizontal ranges of 2000 ft, 1000 ft, minimum range, 1000 ft and 2000 ft. As each of
these points is passed, the video operator will announce over UQC the run and horizontal
range. BOX-L and DOLPHIN will then terminate any acoustic transmissions. BOX-L will
then signal Morse "R" (short-long-short). Upon receipt of "R" video operator aboard