See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/277990263

Trans-Interface Optical Communication (TIOC)

Article · January 2009

CITATIONS READS
0 3,653

1 author:

Kendall Carder
SRI International
189 PUBLICATIONS 19,399 CITATIONS

SEE PROFILE

Some of the authors of this publication are also working on these related projects:

Satellite remote sensing coastal environment View project

Autonomous observation systems and vehicles View project

All content following this page was uploaded by Kendall Carder on 11 April 2016.

The user has requested enhancement of the downloaded file.

 DISTRIBUTION STATEMENT A: Distribution approved for public release; distribution is unlimited.

Trans-Interface Optical Communication (TIOC)

Kendall L. Carder
University of South Florida, College of Marine Science
140 7th Avenue South, St. Petersburg, FL 33701
phone: 727-553-3952 fax: 727-553-3918 email: [email protected]

Award Number: N00014-07-1-0841

LONG-TERM GOALS

The goal of our work is to develop a system for optical communications through the sea-air interface
and to provide a predictive performance model for the system for different environmental conditions,
source power levels, receiver range, etc.

OBJECTIVES

The scientific objectives of the project involve understanding the spectral and temporal effects of water
constituents and surface characteristics on trans-interface communication. This involves the
development of a system that utilizes several, separated, redundant, sources to reduce the data drop-out
rate due to wave refraction. Engineering objectives of the project are the design and development of an
intelligent optical receiver that can identify the positions of redundant sources within video imagery
and control a micro electro-mechanical system (MEMS) device to direct radiance from only those
sources to a high-speed sensor for demodulation. This spatial filtering should reduce interference from
surface reflection of skylight.

APPROACH

Wave refraction of light rays passing through the sea-air interface causes a perturbation in the path of
the ray relative to its position and direction for a flat sea surface. While gravity waves range in slopes
up to about 15o (e.g. Lighthill 1980), these slopes can be augmented near the crests by parasitic
capillary waves (Martin 2004). According to the Cox-Munk (Cox and Munk 1954) wave-slope
relationship to wind speed, a wind of 10m/s results in a mean wave slope of about 0.19. Wu (1990)
suggests a more rapid increase of slope with wind speed, however. Past experience in trying to avoid
sun glint when performing remote sensing operations indicates that slopes of 0.6 are not unusual for
individual waves.

For capillary waves, the disruption in directionality of an optical beam or ray that results in
information dropouts can be significant. This is corroborated by an experiment made in Bayboro
Harbor (St. Petersburg, FL) where the continuity of signal from a single LED was contrasted against
that from up to four LEDs, where four LEDs were separated by four feet at the corners of a square
frame approximately 1 foot below the interface. Generating capillary waves with 2 air jets provides
enough capillary wave activity to provide significant signal drop-outs. Redundant, spatially separated
sources are used with digital signal processing (DSP) of video imagery to identify the positions of

1
 Form Approved
Report Documentation Page OMB No. 0704-0188

Public reporting burden for the collection of information is estimated to average 1 hour per response, including the time for reviewing instructions, searching existing data sources, gathering and
maintaining the data needed, and completing and reviewing the collection of information. Send comments regarding this burden estimate or any other aspect of this collection of information,
including suggestions for reducing this burden, to Washington Headquarters Services, Directorate for Information Operations and Reports, 1215 Jefferson Davis Highway, Suite 1204, Arlington
VA 22202-4302. Respondents should be aware that notwithstanding any other provision of law, no person shall be subject to a penalty for failing to comply with a collection of information if it
does not display a currently valid OMB control number.

1. REPORT DATE 3. DATES COVERED

2. REPORT TYPE
2009 00-00-2009 to 00-00-2009
4. TITLE AND SUBTITLE 5a. CONTRACT NUMBER
Trans-Interface Optical Communication (TIOC) 5b. GRANT NUMBER

5c. PROGRAM ELEMENT NUMBER

6. AUTHOR(S) 5d. PROJECT NUMBER

5e. TASK NUMBER

5f. WORK UNIT NUMBER

7. PERFORMING ORGANIZATION NAME(S) AND ADDRESS(ES) 8. PERFORMING ORGANIZATION

REPORT NUMBER
University of South Florida,College of Marine Science,140 7th Avenue
South,St. Petersburg,FL,33701
9. SPONSORING/MONITORING AGENCY NAME(S) AND ADDRESS(ES) 10. SPONSOR/MONITOR’S ACRONYM(S)

11. SPONSOR/MONITOR’S REPORT

NUMBER(S)

12. DISTRIBUTION/AVAILABILITY STATEMENT

Approved for public release; distribution unlimited
13. SUPPLEMENTARY NOTES

14. ABSTRACT

15. SUBJECT TERMS

16. SECURITY CLASSIFICATION OF: 17. LIMITATION OF 18. NUMBER 19a. NAME OF
ABSTRACT OF PAGES RESPONSIBLE PERSON
a. REPORT b. ABSTRACT c. THIS PAGE Same as 10
unclassified unclassified unclassified Report (SAR)

Standard Form 298 (Rev. 8-98)

Prescribed by ANSI Std Z39-18
redundant sources within the imagery (areas of interest, AOIs). Radiance from these dispersed sources
it then collected onto a common intensified receiver to reduce interference from surface reflection of
skylight while increasing the signal for high-rate data reception.

BACKGROUND

In this concept for optical communication through the air-sea interface, multiple emitters are placed
beneath the water’s surface and imaged from above. These emitters appear as small points of light in a
video image of the area. In a project funded by Blackbird Technologies, we have demonstrated the
ability to process video in real-time and identify our emitters within the video. For this work, we
designed optical emitters using LEDs that transmit data using on/off keying. Since standard video
rates are used, data bandwidth is limited to 5 baud and only short messages containing emitter ID
information is transmitted.

Utilizing digital signal processing (DSP) hardware, we developed a system that processes NTSC video,
identifies sources within the video matching search criteria, decodes data transmitted by those sources,
and outputs the input video signal with graphical overlays highlighting the detected signals and
displaying received data. Fig. 1 shows the prototype electronics (left) and a screen capture from a
video clip (right). The screen capture shows several sources of light imaged through a narrow band-
pass filter. For the case shown, only one point of light is actually a device (ID 399) communicating
data. The other points are street and car lamps with energy in the pass band of the optical filter.
Without our DSP approach, these other points would not only decrease S/N but could be identified as
spurious targets. We have demonstrated the ability to maintain detection of signal sources under a
wide range of environmental conditions, have demonstrated tracking of moving signal sources, and
proven our ability to distinguish our devices from sources that may appear similar in nature.

This extension of the basic system involves the use of a spatial modulator to optically direct only the
portions of the image containing signal sources to high-speed detectors for data reception. This
concept is illustrated below in Fig. 2. Any camera outputting NTSC video is used to image the scene of
interest. The video signal is processed by the DSP electronics to determine the coordinates of signal
sources. These coordinates are used to control an active modulator to direct only the portions of the
image containing signal sources to a high-speed detector, reducing background noise and raising the
signal-to-noise ratio.

MEMS 2-D imaging arrays are commercially available and widely used in TV and data projection
systems [eg. Texas Instruments Digital Mirror Device (DMD)]. The DMD is designed to tilt each of
several million mirrors into 1 of 2 positions according to a video image received by the coupled
electronics module. In the first (default) position the mirrors are turned off and are all coplanar.
Incoming radiation is simply reflected at the same angle as the incoming beam (relative to the surface
normal).

2
Fig. 1: Prototype electronics used to process video signals to locate and decode optical signals, and a
representative screen capture showing the detection of a signal source placed near many noise
sources.

Beam- 2-D Active

splitter Array
PMT Electronics

Camera PC Data
Scene (Filter) Imaging
Optics

Fig. 2: Actively-steered high-speed, high sensitivity detector. 2-D Active Array (Eg. TI DLP) is used
to actively project only areas of the scene of interest onto a high speed,
highly-sensitive PMT detector.

In the second position, the mirrors are tilted by approximately 7 degrees. Depending on the orientation
of the DMD, the reflected beam can either be at a larger or smaller angle (by 14 degrees) than the beam
reflecting off the mirrors in the first position. By turning the appropriate mirrors on or off, light from
specific portions of the image can be separated from other regions. It is proposed that such a 2D DMD
array be utilized within a custom optical system to allow simultaneous targeting of each of the
individual sources that comprise the underwater transmitter. Only those mirror elements
corresponding to the locations of the signals will be energized to redirect the signals to the high-speed
detector. As wave action and other factors interfere with the reception of an individual source, the
corresponding mirror elements will be turned off and only light from the other valid sources will

3
continue to the high-speed detector. In the best case, all individual sources will be present and their
combined energy will be directed to the detector for maximum signal to noise. Note that by rejecting
residual background image noise due to reflected skylight or sun glint, the gain on the high-speed
detector can be increased.

WORK COMPLETED

Scientific A single-wavelength (525 nm), 4-source, LED array was constructed and deployed in both
day and night field tests. Analysis of the data acquired during these deployments indicates that the
basic hypothesis is correct and that our envisioned approach is viable (Carder et al. 2007). Redundant
sources increased transmission efficiency in both daytime and nocturnal field tests. We utilized only
narrow-band-pass and polarization filters on the camera. Compared to losses with one LED, data drop-
outs for the redundant sources were reduced by factors of 6-14 (see Fig. 3).

LED Transmission Efficiency (Day)

0.9

0.8

0.7

0.6
Efficiency

0.5

0.4

0.3

1+2+3+4
1+3+4
2+3+4
1+2+4
1+2+3

0.2
1+2
1+3
1+4
2+3
2+4
3+4

0.1
1
2
3
4

0
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15

LED Group

Fig. 3. Statistical analysis of the various permutations of the four, redundant, LED sources. Transmission
efficiency increased to 95% utilizing all four sources versus 30% utilizing a single LED.

When completed, spatial filtering is expected to increase the signal-to-noise, and thus range and/or data
rate, by a factor equivalent to the ratio of the total image area divided by the sum of the LED-
neighborhood areas of the images (e.g. > 20X).

Engineering Engineering activities have focused on construction and testing of prototype hardware
to demonstrate proof-of-concept. We completed construction of the prototype receiving hardware as
shown below in Fig. 4. The main lens views the target and produces an image on a beam splitter.
One path from this beam splitter is directed towards a video camera for image display and processing.
In the second path, an enlarging lens relays the primary image onto the spatial modulator (DMD). To
compensate for the different sizes of the camera sensor and the spatial modulator, the relayed image is

4
magnified by a factor of 1.6. Light reflecting off the spatial modulator is collected and focused onto
the data detector for subsequent analysis.

Figure 4: A photo of the prototype optical receiving hardware.

Not shown in the figure are the electronics used to process images and control the spatial modulator
(refer to Fig. 2). In our concept, optical transmitters will appear as small dots in the images. These
transmitters will modulate data at a high rate (> 100 Khz) within a much slower modulating envelope
(< 10 Hz) that can be detected with a standard video camera. Image processing will detect the slow
modulation and activate only those elements of the spatial modulator upon which the modulating
signal is incident. The active elements reflect the incident light to the data (high-speed) detector for
analysis. The goal of this approach is to increase the system’s signal-to-noise ratio by filtering out
ambient scene light and ultimately detecting only that light received from the close neighborhood of
the transmitter(s).

To increase gain, we replaced the photodiode initially chosen for the data detector with a
photomultiplier tube (PMT). These detectors offer excellent sensitivity and frequency response at the
cost of complexity.

Spatial Filter This experiment demonstrates the effectiveness of using a Digital Micro-Mirror Device
(DMD), from Texas Instruments, as an active, digital, spatial filter, allowing optical communication in
conditions not otherwise possible.
The DMD is an array of tiny mirrors, each corresponding to one pixel, which tilt +- 12 degrees along a
45 degree axis (lower-left to upper-right). Using an optical geometry very similar to that used in video
projectors, we were able to use the DMD to reflect light from specific parts of the image onto a Photo-
Multiplier Tube (PMT) for signal detection, or reflect light away from the PMT. This is called a
“mask” because it masks-off most of the image, only passing the parts of the image which correspond
to the LED transmitting the signal. The light that is reflected away from the PMT is called the “reject”

5
light. This light can be viewed with a camera, and we call that the “reject image”. By observing the
reject image, we can verify that the mask corresponds to the correct part of the image because the hole
in the mask should make the LED invisible. We can also invert the mask to allow our camera to see
the light that should be reaching the PMT. All of the grayscale images below were captured from the
reject image camera.

TIOC test results from the optics lab are shown in Figure 5. Equipment was in a light-tight box.
Fluorescent room lights were on. Incandescent lamps were off. The LED emitted 850 nm, high-power,
light, but it was operated using a large, current-limiting resistor to reduce output power. No optical
band-pass filter was used, allowing more background light as “noise” in the image. Imaging was
performed with a 3”, 400mm spherical mirror directly onto the DMD. The data were transmitted from
a standard computer, RS-232, serial port at 115200N81 (115,200 bits/second, no parity bit, 8 data bits,
one stop bit) through an IrDA encoder chip, which converts each bit into the presence (0) or absence
(1) of 1uS pulses. After reception by the voltage-output PMT, the signal passed through a single-pole
100 kHz band-pass filter with a gain of 6. After this, the signal was passed through a Schmitt trigger
and a 555 chip to ensure that the incoming pulses were of the proper width. They were then passed into
an IrDA decoder chip, which output RS-232 signals to the serial port of the receiving PC.

The lower image of Figure 5 shows the received signal on the oscilloscope. The unfiltered PMT signal
(pink) is off the screen due to the ambient light (noise). Note that the ambient light is not simply a DC
offset, or else it would be filtered-out by the 100 kHz high-pass analog filter (yellow), leaving our data
signal. The oscilloscope is set to trigger on the yellow trace, on any positive pulse shorter than 3.8 uS
and as tall as the yellow arrow on the left. One may observe that near the trigger point there appears to
be a tendency for a corresponding square wave on the RS-232 output (green is darker on the upper line
and darker on the lower line), however noise is causing too many false square waves to differentiate
the data signal from the noise. Despite adjusting the oscilloscope viewing and triggering parameters,
and adjusting the PMT gain, no usable data could be extracted from the noise.

In the two images in Figure 5 the image is focused upon the DMD. Without using the DMD as a
spatial filter, all light energy in this image reached the PMT. Although the image of the LED appears
obvious, it comprises only a fraction of the total light energy in the image, resulting in a signal-to-noise
ratio much less than 1. The resulting signal is indecipherable.

6
 LED “off” LED “on”

Figure 5. Source image with LED “off” (top left) and “on” (top right); oscilloscope traces for
various test configurations (bottom).

7
 LED “off” and masked by the DMD LED “on” and masked by the DMD

LED “off” through the DMD mask LED “on” through the DMD mask

Figure 6. Mask effects on the images rejected by the DMD.

In the upper two images of Figure 6, we can see the “reject” image of the mask. The black square over
the LED corresponds to the DMD mirrors that are tilted toward the PMT and away from the camera.
The remaining mirrors are tilted toward the camera. This allows us to view the image on the actual
DMD and verify the alignment of the active digital spatial filter (mirrors tilted toward the PMT) with
the signal source (LED). The lower two images show an inverted filter. This allows us to view the
spatially filtered light that should be striking the PMT. The other light areas in the image (not pointed-
to by the arrow) are noise, possibly due to glare off the edges of the mirrors or a smudge on an optic.

The effectiveness of this technique is confirmed in the oscilloscope image (Fig. 7) showing a high
signal-to-noise ratio (approximately 3:1 in the raw PMT signal, and about 7:1 after the 100kHz high-
pass filter). The IrDA-encoded pulses are easily decoded and converted back into RS-232 data, as
used in a standard computer serial port. This is shown in the green trace. The character received here
is the RS-232, ASCII encoded letter “f”, as when the user on the transmitting computer types an f on
the keyboard in HyperTerminal. In this encoding, the ASCII code for “f” is 01100110, and RS-232
specifies a start bit (0) and stop bit (1), where 1 is -7V and 0 is +7V. IrDA encoding (yellow) is
similar, but it uses a single pulse of 1.6uS for each transmitted 0 and the absence of a pulse implies a 1.
There is a slight time delay in the decoding from IrDA (yellow) to RS-232 (green).

8
 Figure 7: Oscilloscope showing the reception of the letter "f". Pink = raw PMT signal. Yellow =
output of 100kHz high-pass filter with a gain of 6. Green = RS-232 output.

This experiment showed the active digital spatial filter to be very effective at increasing the signal to
noise ratio. It dropped the unfiltered bias in the image due to background light (noise) from more than
5V to about 0.2 volts after spatial filtration, or more than 25X. This spatial filtration reduced
background light enough to permit the system to transceive data nearly flawlessly at the high data rate
of 112 kBaud. Data rates were mostly limited by the slew rate of the trans-impedance amplifier built
into the PMT. Additional signal-to-noise improvements will surely be seen with the use of an optical,
narrow band-pass filter, matched to the wavelength of the transmitting LED. For use outside, both the
band-pass filter and a “full-power” LED would be used to obtain more range under daylight
conditions.

CONCLUSIONS

Redundant LED sources at the corners of a 1-m square used for optical communication from below-
water to above water reduced the data drop-out rate by 6- to 14-fold because wave facets rarely
directed the radiance from all four LEDs away from the receiver.

Laboratory tests of elements of optical communication in a light-rich environment (e.g. reflected

skylight) were also performed. Because narrow-band-pass (NBP) optical filters are well-understood in
terms of suppressing out-of-band background light, NBPs were not used in the laboratory. Instead,
tests were of a spatial-filter to restrict the field of view of a high-gain receiver to the neighborhood
about the light source. This approach reduced the background light 25-fold, enabling the PMT to be

9
 operated at a much higher gain. The resulting signal-to-noise ratio was about 3:1. After passing the
spatially filtered PMT signal through a high-pass electronic filter, the signal-to-noise ratio was
increased to 7:1. This was sufficient to allow optical communication at 112 kBaud. In transitioning
outside during daylight, the LED emitter will be returned to full power, and a narrow band-pass filter
will be added to contend with the brighter environment.

IMPACT/APPLICATIONS

There are many scenarios where communication between a submerged asset and an airborne asset
would be very desirable. However, communication through the air/sea interface is, at best,
problematic, and any high-bandwidth communication strategy remains elusive. Tests of our basic
approach (redundant, separated sources with a spatial filter of the image field of view) have shown that
our hypotheses were sound. Data dropouts were reduced by factors from 6X to 14X, and background
light rejection was about 25X. This provides enough signal to noise to permit optical data transmission
rates up to 112 kBaud. Additional, brighter LED sources will further reduce data drop-outs.

RELATED PROJECTS

• USF College of Marine Science Center for Underwater Observability and Optical
Communication – Utilization of the ROSEBUD ROV and the TOPO-13 air platform

• Eric Kaltenbacher, SRI International – Optical tagging and 3D underwater imaging.

REFERENCES

Carder, K.L. and D.K. Costello, 2007. Trans-Interface Optical Communication (TIOC), ONR Annual
Report, Office of Naval Research, Arlington, VA, 7p.

Cox, C. and W. Munk, 1954. Statistics of the sea surface derived from sun glitter. J. Marine Res. 13:
198-227.

Lighthill, J. 1980. Waves in Fluids. Cambridge, Cambridge University Press.

Martin, S. 2004. An introduction to Ocean Remeote Sensing. Cambridge, Cambridge University Press.

Wu, J. 1990. Mean square slopes of wind-disturbed water surface, their magnitude, directionality, and
composition. Radio Sci. 25: 37-48.

10

View publication stats