University of Southern California Dissertations and Theses

Development of Lagrangian drifter for ocean monitoring and marine applications

(USC Thesis Other)

Development of Lagrangian drifter for ocean monitoring and marine applications

PDF

Download a page range

Download transcript

Copy asset link

Request this asset

Transcript (if available)

Content Development of Lagrangian Drifter for Ocean Monitoring and
Marine Applications
by
Supreeth Subbaraya
A Dissertation Presented to the
FACULTY OF THE GRADUATE SCHOOL
UNIVERSITY OF SOUTHERN CALIFORNIA
In Partial Fulllment of the
Requirements for the Degree
MASTER OF SCIENCE
(Electrical Engineering)
May 2014
Copyright 2014 Supreeth Subbaraya
DEDICATION
This Thesis is dedicated to my Parents, Family and Friends for their love and support.
ii
Acknowledgements
I would like to thank my mentor and advisor Prof. Gaurav Sukhatme, for encouraging
me to execute this project. I am grateful to him for encouraging me to try out my ideas,
conduct experiments and provide support through the Robotic Embedded System Lab
(RESL), USC. I would also like to thank Prof. David Caron, for agreeing to be part of
the thesis committee and helping me in understanding the dierent aspects of the Ocean.
I owe my gratitude to Prof. Urbashi Mitra for being a part of the thesis committee, for
her support and reading my thesis even though she was traveling.
I am immensely grateful to Carl Oberg, research sta, RESL for being a helping hand
in design and testing. I thank him for building the drifter mechanical parts using the
machinery available at his garage and coming out all the time for the deployments.
I thank Arvind Pereira, software engineer, Clover Network and an alumni of RESL for
being a great guide, giving ideas and support for design of the system. Special thanks to
Hordur K Heidarsson (RESL) for his help in setting up server software and web page. I
have to thank the entire aquatic group at RESL, Stephanie Kemna, Jnaneshwar Das and
Artem Molchanov for their constant guidance and help whenever I faced any problems.
Artem also helped me in a test and I thank him for that. I would also like to thank Joerg
Muller and Andreas Breitenmoser, PostDocs at RESL for their input towards my thesis.
A great thanks to all my lab mates at RESL, Megha Gupta, Christian Potthast, Harsh
Vathsangam, Max P
euger, David Kim, Karol Hausman, Chet Corcos, Mihir Daptardar
iii
and Ryan Williams for making my experience at RESL memorable one. Thanks to Megha
and Stephanie for suggesting changes required to this thesis.
A special thanks to Mas Dojiri, Manager, Environmental Monitoring Division, Los
Angeles and his sta for providing us the logistics and support required to conduct test
at Hyperion Treatment Plant. I have to thank Alyssa Gellene and all the other members of
Caron Lab, USC for their help especially to conduct tests on AHF rooftop. I am grateful
to Nalini Gujuluva and Valencia Teems, previous workers at RTH Business Center for
helping in acquiring the parts and components required for the project.
iv
Table of Contents
Dedication ii
Acknowledgements iii
List of Figures vii
Abstract xi
1 INTRODUCTION 1
1.1 Motivation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3
1.2 Related Work . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6
2 System Design 9
2.1 Design Considerations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9
2.2 Electrical Design . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10
2.2.1 SPOT Satellite GPS Messenger . . . . . . . . . . . . . . . . . . . . 11
2.2.2 XBee-Pro 900HP . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12
2.2.3 Microcontroller Board . . . . . . . . . . . . . . . . . . . . . . . . . 14
2.2.4 Integration . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17
2.2.5 Base Station . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 18
2.3 Software Design . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 20
2.3.1 Drifter Software . . . . . . . . . . . . . . . . . . . . . . . . . . . . 20
2.3.2 Base Station Software . . . . . . . . . . . . . . . . . . . . . . . . . 22
2.3.3 User Interfaces . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 25
2.4 Mechanical Design . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 30
2.4.1 Surface Float . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 30
2.4.2 Tether . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 32
2.4.3 Drogue . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 33
3 System Characterization 36
3.1 Power Consumption . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 36
3.2 Cost . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 41
3.3 Modularity . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 43
v
3.4 Current Following Capability . . . . . . . . . . . . . . . . . . . . . . . . . 44
3.5 Re-usability . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 44
3.6 Resolution . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 45
3.7 Data Delivery and Quality . . . . . . . . . . . . . . . . . . . . . . . . . . . 46
3.8 Future Enhancement . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 49
4 Field Tests 50
4.1 Initial Test . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 50
4.2 Radio Test . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 53
4.3 Test at Hyperion Sewage Treatment Plant . . . . . . . . . . . . . . . . . . 55
4.4 Test With Multiple Drifters . . . . . . . . . . . . . . . . . . . . . . . . . . 58
4.5 Final Test . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 60
5 Conclusion and Future Work 64
5.1 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 64
5.2 Future Work . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 65
5.2.1 Surface Float . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 65
5.2.2 Drogue Detachment Detection . . . . . . . . . . . . . . . . . . . . 66
5.2.3 Communication . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 66
5.2.4 Sensing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 67
5.2.5 Software . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 67
5.2.6 Similar Systems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 68
BIBLIOGRAPHY 69
vi
List of Figures
1.1 Typical Drifter Design. The surface
oat houses the electronics (position-
ing and communication) and provides buoyancy. The drogue provides the
drag required to move along with the currents. . . . . . . . . . . . . . . 2
1.2 Hyperion Treatment Plant (Green) Outfall Map. During Inspection or
repair, 300-350 million gallons of treated euent needs to be diverted
from the 5-mile pipe (red) to the shorter 1-Mile pipe (yellow). Such an
activity has to be monitored in order to track the outfall and nd out if
it reaches the Bay aecting beaches and public health. . . . . . . . . . . 3
1.3 ROMS Nowcast Output for a particular time of a day. The red marker
depicts the point of our interest. The spatial resolution of ROMS data
is a 3 km grid. The gures show that the ROMS Data is sparse and is
not of much help if we want to track a plume because we have only one
prediction of current direction in a grid of 3 km (Pictures from SCOOS
website: http://www.sccoos.org/data/roms-3km/) . . . . . . . . . . . . 4
1.4 Autonomous Underwater Vehicles used by USC Researchers. Slocum G1
Gliders in the picture can operate in ocean upto a month. Some variants
can operate longer. Ecomapper (Propelled AUV) cannot be used to track
the plume because of its low endurance (6-8 hrs) and cost involved. . . 5
2.1 XBee Range Test. The red marker shows the base station set up at
Santa Catalina Island, California. It was 75 m above sea level. It had a
computer, XBee radio and 6 dB antenna. The trajectory is of the boat
which had a computer, gps receiver and XBee radio. Both the units
were transmitting every 2 seconds. The XBee antenna was at a height of
approximately 1 m and had 2.5 dB gain. The yellow marker shows the
farthest point reached by the boat, which was 6 miles from base station. 13
2.2 XBee Range Test Characteristics. The range test was characterized by
counting the number of packets delivered in a minute (the best case is 30
packets, assuming transmission every 2 seconds). This picture is of the
transmission from boat to base station which is more important. There
are 5 points for every minute, which is why it looks sparse at some points.
The reduced number of packets delivered at farther points may be due to
the turning of the boat. The packets received were 1940/1997 (97%). . 14
vii
2.3 RESL-MSP430 Micro-controller Board Design Block Diagram. The board
was designed and populated at RESL, USC. The board has sucient
ports for interfacing SPOT device, radio, sensors, USB devices and has
logging capabilities. It operates from 3.3 to 36 V supply and has a current
consumption of 2 mA. Its form factor is 9.9 x 5.1 cm. . . . . . . . . . . 16
2.4 Picture of a populated RESL-MSP430 Board. Most of the components
were SMD components except for the headers. . . . . . . . . . . . . . . 17
2.5 Drifter Electronics Block Diagram. The GPIO lines control switching
whereas UART lines are for data communication. The system works on
a supply of 7.5 V. Switching circuits are available for conserving power. 18
2.6 Picture of a base station at Catalina Island. It is at an altitude of 73 m.
It was set up for a CODAR station and for a glider network. It has a
weather proof cabinet, power supply and Internet connection. . . . . . 19
2.7 Drifter Software Architecture. . . . . . . . . . . . . . . . . . . . . . . . 21
2.8 Work
ow of the Drifter Software. . . . . . . . . . . . . . . . . . . . . . 23
2.9 Block diagram of overall system software. Drifters broadcast their status
and data at periodic intervals. They can receive commands from a base
station or a monitoring station on a research vessel. . . . . . . . . . . . 24
2.10 The web interface, through which the user can interact with drifters. . 26
2.11 Drifter Communicator Application with a drifter selected. The user can
download data from the drifter, send a conguration le, get its recent
location when connected and change the update rate. . . . . . . . . . . 27
2.12 Drifter Communicator Application with a base station selected. The user
can download data from base station and upload conguration les. . . 27
2.13 Block diagram of the module to aid drifter recovery. This module receives
signals from the drifters using XBee and transmits the data to an Android
phone or tablet using bluetooth, to see the location of drifters. . . . . . 28
2.14 Screen-shot of the Android App for drifter recovery. The red marker al-
ways shows the boat location. The other markers show the latest location
of the drifters that connected to the recovery module. . . . . . . . . . . 29
2.15 Cross sectional view of cylindrical surface
oat. . . . . . . . . . . . . . 31
2.16 Top view of the surface
oat . . . . . . . . . . . . . . . . . . . . . . . . 32
2.17 Picture of the Surface Float . . . . . . . . . . . . . . . . . . . . . . . . . 33
2.18 Cross Sectional View of the Drogue. . . . . . . . . . . . . . . . . . . . . 34
2.19 Photo of a Completely Assembled Drifter . . . . . . . . . . . . . . . . . 35
viii
3.1 Current consumption plot for 30-minute data update. A magnied ver-
sion of the black box is shown in the gure. The ON time (20 minutes) in
the gure shows the current consumption when the SPOT Messenger is
on and the radio is transmitting periodically every minute. The SLEEP
time (10 minutes here) is when the micro-controller is in the sleep mode
and the XBee transmits every minute. The spikes during the ON and
SLEEP time represent XBee transmissions. . . . . . . . . . . . . . . . . 38
3.2 Current consumption plot for a 20-minute data update rate. . . . . . . 39
3.3 Current consumption plot for a 40-minute data update rate. . . . . . . 39
3.4 Current consumption plot for a 60-minute data update rate. . . . . . . 40
3.5 The current following capability of the drifter was tested by deploying a
drifter near the dock. The currents were fairly low in that region. The
drogue was at a depth of 1m. The drifter drifted only about 2 m over
a span of 40 minutes. This shows that the drifters were not aected by
surface winds or waves. . . . . . . . . . . . . . . . . . . . . . . . . . . . 45
3.6 Position data for the drifters. The top gure (red dots) is of the drifter
with a 20-minute update rate and the bottom gure (blue dots) is for a
30-minute update rate. We can see that some of the points in rst picture
are outliers. Second picture has more points near to the center. . . . . 48
4.1 Trajectory of the drifter during the rst test at Catalina Islands. The
green marker shows the starting point on Oct 7, 2013 at 12:54 PM and
the red marker shows the end point on Oct 8, 2013 at 8:19 AM. The north
of the picture is the actual north of the earth. The update rate for this
drifter was 30 min. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 51
4.2 Locations transmitted by the SPOT Satellite Messenger during the rst
test. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 51
4.3 Comparison between data obtained from the SPOT Tracker (green) and
the SD Card (red). . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 52
4.4 Picture of a drifter in water . . . . . . . . . . . . . . . . . . . . . . . . . 53
4.5 Radio test points. The drifter was left in the water for a couple of minutes
at these points. The drifters were transmitting their location every 5 sec. 54
4.6 Trajectory of the drifter deployed near the 1-mile pipe at Hyperion Sewage
Treatment Plant. The green marker shows the starting location and the
red marker shows the last location. The drifter was deployed on Nov 20,
2013 at 8:45 AM and recovered on Nov 22, 2013 at 8:15 AM. The drifter
remained close to the shore during the test showing that the outfall from
the 1-mile pipe may stay near the shore. . . . . . . . . . . . . . . . . . 56
4.7 Drifter trajectory using ROMS output for 1 m depth . . . . . . . . . . 57
4.8 Drifter trajectory using ROMS output for 10 m depth . . . . . . . . . . 57
ix
4.9 Trajectory of two drifters near Catalina Islands. One of them was drogued
at 5 m depth (green) and the other one at 1 m depth (red). They were
dropped o at the same location. Their nal separation was 400 m. . . 58
4.10 The top gure shows the drifter trajectory by combining both the SPOT
data and XBee data. The bottom gure is a comparison between the
SPOT data and the SPOT+XBee data. Red and blue lines are for the
drifter with drogue at 5 m. The green and magenta lines are for the
drifter with drogue at 1 m. . . . . . . . . . . . . . . . . . . . . . . . . . 59
4.11 Trajectory of three drifters near Catalina Islands. One of them was
drogued at 5 m depth (red), one at 3 m depth (green), and the last
one at 1 m (yellow). . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 61
4.12 Deployment of three drifters in the ocean . . . . . . . . . . . . . . . . . 61
4.13 Visualization of Data on Web Interface . . . . . . . . . . . . . . . . . . 62
4.14 Temperature data using the Hobo Pendant Sensor. The top gure shows
the data for the sensor attached to the
oat. The bottom plot shows the
data of the sensor attached to the drogue. Blue line represents the drifter
at 1 m depth, red at 3 m depth, and green at 5 m depth. . . . . . . . . 63
x
Abstract
This thesis describes the design and development of a Lagrangian Drifter for ocean mon-
itoring applications. The system is developed to move with the currents at a particular
depth, transmitting its location to the user. Through this design, the user can track
ocean currents and object or entities carried by them. This thesis introduces the sys-
tem by giving the motivation for the work and a survey of related work and present day
technologies available to design the system. The design considerations are laid out based
upon the application of the device. A description of the electrical, software and mechan-
ical design of the system is presented here. The system was characterized for its dened
design considerations. Finally, experiments and results are presented and concluded with
a discussion on future enhancements for the system.
xi
Chapter 1
INTRODUCTION
Lagrangian ocean drifters are used to study ocean circulation patterns. These drifters are
passive devices, which means they have no actuator, and they are typically carried with
the ocean currents at a particular depth. This property of the drifters makes them useful
for studies in marine biology and oceanography. The movement of euents, larvae and
other micro-organisms are interesting to marine biologists and environmental agencies.
The Oceanographers are interested in studying the complex
uid dynamics of the ocean
and its causes. Drifters are used in both of these elds to track the ocean currents to
explain the biological or oceanographic phenomena.
Figure 1.1 shows the schematic of a typical drifter design and its parts. A drifter
consists of a surface
oat which provides necessary buoyancy to hold the drogue at a
particular depth. Drogue is carried by the currents and is connected to surface
oat via
tether. The
oat acts as a surface expression for the drogue and can obtain the position
of the drifter and can transmit the location via a telemetry system.
1
Figure 1.1: Typical Drifter Design. The surface
oat houses the electronics (positioning
and communication) and provides buoyancy. The drogue provides the drag required to
move along with the currents.
This thesis describes the design and development of a low power, low unit cost La-
grangian oceanic drifter having positioning and telemetry capabilities. It explains the
design considerations taken into account. The Electrical, Mechanical and Software design
of the system are presented in this work. The system was characterized for its low power
constraints, data quality and current following capabilities. Various eld trials were car-
ried out to prove the working of system in real time scenarios. Finally, ideas for designing
similar systems which move along with the currents are discussed. The system described
2
has a GPS receiver, satellite transmitter, memory card, sensor interfaces and a 900 Mhz
radio giving opportunity for networked ocean monitoring capabilities.
1.1 Motivation
The motivation for the work comes from activities such as Euent Diversion (November
2006) [12] at Hyperion Sewage Treatment Plant, Los Angeles. The plant has a 5-Mile
outfall into Santa Monica Bay. Treated euent outfall is 300-350 million gallons per day.
Internal inspection of the pipe which happened after half a century, required the euent
to be diverted to a smaller 1-Mile Pipe. Both the pipes are shown in Figure 1.2. Even
though the possibility of plume from 5-Mile outfall reaching the bay is less, the outfall from
1-Mile pipe can reach to the bay probably resulting in Harmful Algae Blooms (HABs)
aecting the public health. To substantiate, the outfall needs to be tracked for a period of
1-1.5 months (or longer) to understand the movement of the water in the area of interest.
Figure 1.2: Hyperion Treatment Plant (Green) Outfall Map. During Inspection or repair,
300-350 million gallons of treated euent needs to be diverted from the 5-mile pipe (red)
to the shorter 1-Mile pipe (yellow). Such an activity has to be monitored in order to track
the outfall and nd out if it reaches the Bay aecting beaches and public health.
3
One of the methods to predict the location of plume would be to use Regional Ocean
Modeling System (ROMS) [4], a publicly available predictive ocean model for the Southern
California Bight (SCB). It is widely used in the oceanographic and modeling communities.
It provides nowcasts and forecasts for current directions by accumulating data from HF
Radar, Satellite Imagery, Mooring devices and various other sampling platforms in the
ocean. The predictions are for a 3 km grid.
Figure 1.3: ROMS Nowcast Output for a particular time of a day. The red marker depicts
the point of our interest. The spatial resolution of ROMS data is a 3 km grid. The gures
show that the ROMS Data is sparse and is not of much help if we want to track a plume
because we have only one prediction of current direction in a grid of 3 km (Pictures from
SCOOS website: http://www.sccoos.org/data/roms-3km/)
Figure 1.3 shows ROMS nowcast with a marker depicting the point of our interest; the
Hyperion Plant. One of the major concern about the model is its spatial resolution. We
have one prediction for a region of grid size 3 km around the point. This is very sparse
data to track the plume. To know whether the water stays in a particular region becomes
complicated. Another concern is the accuracy and precision of the model [5] because
of insucient sampling points. The 1-Mile pipe is at a depth of 6 m. But the ROMS
4
predictions are provided at depths 0, 10, 20,...400 m. So it becomes dicult to follow the
current at 6 m depth especially in areas with complex current motions.
Other options for tracking outfall are Autonomous Underwater Vehicles (AUV) (Fig-
ure 1.4) like underwater Glider or a propelled underwater vehicle. Propelled AUV's like
Ecomapper typically have a very low endurance of 6-8 hrs which defeats the requirement
of monitoring for several months continuously. Underwater gliders can be operated upto
a month, but the operational cost incurred would be a few thousand dollars and some
variants are not shallow water gliders. The use of moored devices which are static devices
would be inaccurate. Use of current meters would be expensive.
(a) Ecomapper (b) Slocum Gider
Figure 1.4: Autonomous Underwater Vehicles used by USC Researchers. Slocum G1
Gliders in the picture can operate in ocean upto a month. Some variants can operate
longer. Ecomapper (Propelled AUV) cannot be used to track the plume because of its
low endurance (6-8 hrs) and cost involved.
So in order to track the plume, we need a low cost device which sustains for more than
a month in the ocean, is drifted with the currents at a depth of up to 10 m, transmits
its location and is not aected by winds and waves. Lagrangian drifters have all these
properties to perform a monitoring task similar to the one at Hyperion.
5
1.2 Related Work
Surface current circulation patterns have been studied historically by using drift bottles,
ships, boats and other
oating objects. Floating objects with drogue attached have been
used to get drift measurements. But these were methods used in the pre-communication
era. Advent of communication technologies provided opportunities for better positioning
and location transmitters. Adding such communication devices to drifter devices, has
enabled more accurate tracking. Literature survey for positioning systems, location and
sensor data transmission systems and drifter mechanical design was done in this work.
For positioning, one of the early systems used was the ARGOS data collection and
location system on NIMBUS Satellites [2], [20], [11]. The system works based upon
the Doppler shift of the transmitted signal, which requires the satellite to pass over a
drifter to know its location. So it suers from resolution, accuracy (150 m - 1000 m)
and high cost to retrieve data. Other radio triangulation systems have problems with
accuracy and the cost involved in large number of base stations required for large area
operations [15], [16], [3]. With the increasing usage and reduced costs, Global Positioning
Systems (GPS) [1], [6], [7], [8], [19], [17], [18] have become popular for acquiring drifter
position.
For transmitting location and sensor data, transmission through ARGOS Satellites
were used in early days. Cellular Systems were proved to be better than satellite com-
munication systems in cost but still they suered from the problems of endurance and
6
energy consumption [1], [7]. Radio Based systems used earlier had power and cost con-
straints [6], [18], [18]. Iridium and Global Star Satellites have been used for the pur-
pose [21]. Transmission and retrieval of data from these satellites are costly if there is no
contract with the organizations maintaining it.
Various drifter mechanical designs were developed through the years. Holey Sock
drogues [2] and TRISTAR [9], [11] drogues were used for deep ocean studies whereas
a corner radar re
ector type drogue was used for near surface currents [1], [21]. Cru-
ciform shaped drifters were built for CODE Program [3], [6], [22]. Drifters were also
built for lakes, estuaries [6] and near shore zones [8]. Drifter technology has been devel-
oped, studied and improved upon during various research programs such as North Pacic
Experiment (NORPAX) [20], World Climate Research Program (WCRP) [2], Global At-
mosphere Research Program (GARP) and NOAA's Global Drifter Program [13]. The
drifters with the above mentioned technologies have been built since 1975. Some of the
drifters even have data logging [8], [17], [21] and sensing capabilities [21], [10].
This thesis work done here at Robotics and Embedded System Lab (RESL) is closely
related to the Microstar Drifters built by Pacic Gyre Inc [21]. A Microstar drifter has
a GlobalStar Simplex Satellite Transmitter and a GPS receiver whereas RESL drifters
have SPOT Satellite GPS messenger [29]. Both the systems use corner radar re
ector
drogue, have data logging capabilities and an endurance of 1 month. In both the cases
data can be retrieved from servers and have GUI's. Microstar Drifters sample every 10
min, whereas RESL drifters sample every 20 min. This duty cycle of RESL drifters is due
7
to the satellite messenger used which tries to send the data a couple of times for ecient
data delivery.
The RESL drifters have a 900 Mhz Radio on it which gives more advantages. The
radios send the drifter locations every 1 min which helps in ecient drifter recovery.
Even though the radios might not be useful in deep ocean for data, near shore monitoring
activities certainly have more data in addition to the data from satellite GPS messenger.
Data can be transmitted to a base station on land or a monitoring station on a vessel and
retrieved by querying the base station if it is in range. Another use of having a radio is
to perform networked ocean monitoring [24]. Sampling of the water being tracked can be
done using autonomous robots which could get the drifter locations by radio. A network
with multiple drifters, base stations and robots can be established. The radios are also
helpful for designing devices for recovery. Since the RESL drifter electronics are custom
made, many sensors can be interfaced to it whereas the Microstar drifters can have a
surface temperature sensor [2] [21].
8
Chapter 2
System Design
This section describes the hardware and software Design of the System.
2.1 Design Considerations
The following are the design features taken into account.
Power Eciency: The drifters should consume as little energy as possible. This
provides the system a longer lifespan, allowing it to stay in the ocean for many
months to years. It reduces the work involved in battery pack changing, and de-
ployments and recovery operations.
Low Cost: The drifters need to be inexpensive to be utilized in large numbers.
Another reason for desiring a low cost, is that we may lose them in the ocean.
Modularity: We should be able to attach dierent drogue congurations to select
dierent depths. The surface
oat should be easy to open to allow making electronics
9
and software changes. It should be small enough to be carried and deployed by one
person. Batteries should be easily replaceable.
Current Following Capability: The drifter should be least aected by winds,waves
and any other external eects.
Re-usability: Ecient deployment and recovery strategies have to be used to in-
crease re-usability and reduce costs. The system should survive in ocean conditions.
The user should be able to recongure drifter software parameters easily.
Resolution: The drifter has to sample at regular intervals so that any important
data of a phenomenon is not lost. This should also have a balance with battery life.
Data Delivery and Quality: Data should be delivered eciently to the user and
displayed in user friendly interfaces. The drifter should have very little positioning
errors and telemetry delays. Quality of data delivered should be usable. This also
requires the design to have little eect from wind and waves to have better data
accuracy. Data Telemetry should be near real-time leading to ecient tracking and
drifter recovery.
Future Enhancements: It should be easy to add extra sensors and have future
enhancement capabilities.
2.2 Electrical Design
A typical drifter electrical design should have the following electronic components present
in them.
Positioning System
10
Telemetry System
Computing and Control Device
Batteries
The component search was based upon the design considerations discussed above, pre-
vious related work, a survey of present technologies and resources available at Robotic
Embedded System Lab (RESL), USC.
2.2.1 SPOT Satellite GPS Messenger
A GPS device was the obvious choice because it is the best low cost positioning system
available today. It has good accuracy, is widely used and easy to interface. For a telemetry
system, radios were ruled out for their short range, cellular systems for their cost and
power. A satellite based system would be an ecient way to communicate data out in
the ocean but cost would be incurred on each message we send. We need good resolution
too. The search landed us to SPOT Satellite GPS Messenger [29] which is the world's
only satellite GPS messenger. It has a Ublox AMY-5M GPS Receiver (world's smallest
GPS receiver) and a GlobalStar STX-2 simplex modem. It can run on 3 AAA Lithium
batteries for approximately a week. The device costs $100 and the service charges $100
per year for any number of messages sent. It requires a 5 V power supply and the GPS
receiver and control points (on/o) have 3.3 V levels which means we can use present day
micro-controllers with it. It has a form factor of 9.4 x 6.6 x 2.5 cm. This was exactly the
device we were looking for and we selected it.
11
2.2.2 XBee-Pro 900HP
The idea of using a radio for recovery, as secondary telemetry system, and to allow for
a network for data retrieval and transfer and networked monitoring activities, led us to
look for a low cost, low power, modular, widely used radio. The XBee 900 MHz (902-928
MHz) Radio Modems [30] have been used extensively by embedded system designers and
electronics hobbyists. They work on Digimesh,point to point, point to multi point,peer
to peer networking protocols and can have RF data rate upto 200 Kbps. Supply voltage
required is 2.4 to 3.6 V, transmit current is 215 mA, receive current is 29 mA and 2.5 uA
sleep current. They have a very small size (2.2 x 3.3 x 0.4 cm). Communication with them
is through UART interface. These features again make them useful with micro-controllers.
Range of a Radio Modem depends on the line of sight, antenna gain and other external
conditions (e.g. humidity etc). The product sheet of XBee claims a range of 28 miles.
But in order to check, we performed a Range test at Santa Catalina Island, California.
The test setup had a base station consisting of a Gumtix Single Board Computer, 6 dB
gain antenna and a XBee unit. A boat having a computer, GPS receiver and a XBee unit
was planned to traverse in the ocean and transmit its location regularly. Since we wanted
to check the range from shore and the ocean, both the units were transmitting every 2
seconds. Received data was stored in the computers. The trajectory of the boat is shown
in the Figure 2.1. The red marker depicts the base station wherein the antenna was at
a height of 75 m above sea level. The XBee antenna (2.5 dB gain, 1/2 Wave whip) on
the boat was approximately at 1 m above sea surface. The baud rate was 115200 bps.
The farthest point reached by the boat was 6 miles from base station which is shown by
yellow marker. We could not go further because of less capabilities of the boat we used.
12
Figure 2.1: XBee Range Test. The red marker shows the base station set up at Santa
Catalina Island, California. It was 75 m above sea level. It had a computer, XBee radio
and 6 dB antenna. The trajectory is of the boat which had a computer, gps receiver and
XBee radio. Both the units were transmitting every 2 seconds. The XBee antenna was
at a height of approximately 1 m and had 2.5 dB gain. The yellow marker shows the
farthest point reached by the boat, which was 6 miles from base station.
Figure 2.2 is a scatter plot of the number of packets received per minute from boat.
Ideally 30 packets needs to be delivered in a minute taking 2 sec transmission cycle. We
can see that at most of the places the packet transmission and reception were ecient. An
accuracy of 97% packet delivery was achieved. Range and eciency would have changed
if it was high transmission rate (in msec), loop back of data and if large number of bytes
were transferred. But the drifters require very little data to be transmitted in a timely
manner (in seconds). So the test was according to our use case. The loss of packets from
farther points may be due to the turning of the boat. Points near to the shore had the
13
eect of not being in the line of sight. The transmission from 6 miles was impressive for
a $40 radio.
Figure 2.2: XBee Range Test Characteristics. The range test was characterized by count-
ing the number of packets delivered in a minute (the best case is 30 packets, assuming
transmission every 2 seconds). This picture is of the transmission from boat to base sta-
tion which is more important. There are 5 points for every minute, which is why it looks
sparse at some points. The reduced number of packets delivered at farther points may be
due to the turning of the boat. The packets received were 1940/1997 (97%).
2.2.3 Microcontroller Board
After selecting the SPOT device and XBee radios, a micro-controller board was required
to control and use them for the application. To receive GPS data and communicate with
XBee radios we needed 2 UART ports. To control the SPOT device we needed GPIO
peripheral. Most of the digital sensors today communicate using I2C/SPI protocol and
14
some of them output their data through UART. Some sensors give analog output requiring
ADC's to interface them with a micro-controller. Keeping in mind that more sensors could
be added to the drifter, a micro-controller with some I2C/SPI, UART, ADC ports needed
to be selected. Also in order to have more controlling capabilities, we required more GPIO
pins on the micro-controller. To log the data an SD Card needs to be used, which works
over SPI or SDIO protocol. The board also needed to be energy ecient, low cost and
small in size to satisfy design considerations.
Our design requirements suggested that we build a custom board rather than going
for a commercial-o-the-shelf board in market. The micro-controller chosen for the task
was Texas Instrument's MSP430F5438A. The complete specication can be found in the
datasheet [31]. It is a 16-bit ultra low power micro controller. It has 4 UART ports, 4
SPI/I2C ports, 12 ADC channels and numerous GPIO pins. It operates from 1.8 V to
3.3 V. It has other peripherals like Real Time Counter and Timer which are useful to us.
These features make it suitable for our application.
The board was built in house at RESL with the idea of it being used for designing
similar systems in the future. Figure 2.3 shows the block diagram of the board design.
The GPS port and SPOT control port are for interfacing SPOT device. It has a dedicated
port for radio and an SD Card slot. 4 SPI/I2C and 8 ADC ports are available for sensor
interfacing. A port for RS232 interface was added, which can also be congured for normal
UART operation by using a jumper. It also has a USB interface which makes it easy to
connect to other devices and computers. It has a 24 Mhz crystal for clock supply. A
JTAG connector helps in programming and debugging the board. A Motor control port,
15
general purpose pins and test points were added for future applications. An on-board
coin cell battery facilitates running of Real Time Clock even when external supply is cut
o.
Figure 2.3: RESL-MSP430 Micro-controller Board Design Block Diagram. The board
was designed and populated at RESL, USC. The board has sucient ports for interfacing
SPOT device, radio, sensors, USB devices and has logging capabilities. It operates from
3.3 to 36 V supply and has a current consumption of 2 mA. Its form factor is 9.9 x 5.1
cm.
The board has 3 power ports capable of giving a power supply of respectively 6-36
V, 5 V, 3.3 V.It has a 5 V switching regulator and 3.3 V linear regulator which can
be used in cascade or individually by shorting the jumpers. The active state current
consumption of the board is 2 mA, which is mainly due to the power regulators, whereas
micro-controller has A consumption. Digital Isolators are used to isolate the noise from
Radio Modem. The optocouplers are used for power and signal switching operations.
16
Operational Ampliers (op-amp) on the board amplies the analog signals and feed them
to the ADC. The dimensions of the board is 9.9 x 5.1 cm. The board was designed
(schematic and layout) using EAGLE PCB Design Software and was populated at RESL.
Most of the components are SMD components. The components are energy ecient and
low cost. Each board along with the components costs approximately $60. A soldered
board is shown in Figure 2.4
Figure 2.4: Picture of a populated RESL-MSP430 Board. Most of the components were
SMD components except for the headers.
2.2.4 Integration
Now that we have the dierent components, system integration was done as per the
block diagram in Figure 2.5. A push power switch turns on the whole system. Linear
regulators are used to supply the XBee and SPOT devices with their respective voltage,
17
Figure 2.5: Drifter Electronics Block Diagram. The GPIO lines control switching whereas
UART lines are for data communication. The system works on a supply of 7.5 V. Switching
circuits are available for conserving power.
which is controlled using a Solid state relay to turn on/o. This is mainly done to save
power when these devices are not in use. Alkaline D-Cell Batteries are used for providing
power. The power supply given here is 7.5 V i.e. 5 D-Cells. The SPOT Tracker was
modied according to our needs [28]. Batteries can be put in parallel to increase lifespan.
The total cost of the electronics for the drifter was $500, including damage costs incurred
during manufacturing and mount.
2.2.5 Base Station
The Base Station consists of an Intel Atom Computer connected to the Internet, a 6
dB gain antenna and an XBee unit. All the base stations are connected to the CINAPS
server [24]. There can also be monitoring stations (on a research vessel) or a user on land
or boat having a computer and an XBee. These units can send commands to or receive
18
data from drifters and base station. A picture of a base station set up at Catalina is shown
in Figure 2.6. It was set up as part of work done for communications for gliders [23]. We
installed our required components at that site. It has a weather cabinet, power source
and Internet connection.
Figure 2.6: Picture of a base station at Catalina Island. It is at an altitude of 73 m. It
was set up for a CODAR station and for a glider network. It has a weather proof cabinet,
power supply and Internet connection.
19
2.3 Software Design
2.3.1 Drifter Software
The drifter software architecture is shown in the Figure 2.7. The software was developed
in C language. The system used two of the four available UART ports for XBee and GPS.
The $GPRMC message from the GPS was parsed to obtain the location, time and date in
order to synchronize with world time. Various messages are exchanged via XBee UART.
The drifter broadcasts dierent messages namely,
DS MESSAGE : Data and status message. These messages are periodically broad-
cast (commonly every minute) .
SYSTEM START : This message is sent when the drifters are switched on.
Acknowledgement Messages : These messages are sent for various commands from
the base station.
The drifter receives commands from the base station. Command messages are
CHANGE UPDATE RATE : Change the location update rate using SPOT Tracker.
CHANGE MODE : Switches between SPOT-XBee Mode, SPOT only mode and
XBee Only mode
CHANGE TIME : Change system date and time if not in sync
SEND DATA : Command to send the data les from the drifter.
A typical drifter data message would have dierent elds and is shown below:
20
Figure 2.7: Drifter Software Architecture.
$DRIFT,ID:(drifterId),DS_MESSAGE,SystemTime,SystemDate,Mode,Dutycycle,latitude,
longitude,checksum,#\r\n
Here $DRIFT is the header of the packet. A unique id is assigned to each drifter
which is in the drifterID eld. DS Message is the type of the message. A checksum
is calculated to facilitate discarding packets if there was an erroneous transmission or
reception. The packet resembles the NMEA messages sent by the GPS. More messages
and network overhead are not introduced to the system to reduce power consumption.
A FAT File System was implemented based upon the FatFs module for small embedded
systems for SD Cards [32]. A seconds counter was implemented based upon a timer
21
to provide delays and time dependent operations. A Real Time Clock (RTC) driver
helps in maintaining accurate system time, which is synchronized to GPS and basestation
connected to the Internet. RTC has the aid of the coin cell battery to maintain the time
and date as much as possible. The clock driver selects the clock source.
A
ow chart showing the working of the drifter is shown in Figure 2.8. A drifter can
have 3 modes of operation. SPOT-XBee, SPOT Only and XBee Only. The
ow chart
shows the SPOT-XBee mode of operation. The other two modes are reduced versions of it.
System Initialization loads up the drivers and make the required peripherals ready. System
Parameters are either set to default or read from a conguration le. The BootId.txt le
helps in creating log les (both data and system) related to a particular boot of the
system. The system start signals are broadcast to dierent base stations or any other
type of stations (monitoring etc). Through an innite loop, the system is sending data
and status messages, receiving commands, performing location update and logging onto
the SD Card.
2.3.2 Base Station Software
An overview of the complete operation of the system is shown in Figure 2.9. Drifters
update their locations through the SPOT Satellite GPS messenger periodically. They
constantly broadcast their status and location in periodic intervals. These messages are
parsed by the base station and can also be heard in monitoring stations. XBee Message
Parsing script is one of the main components of the base station software. It facilitates
logging of the data and sending commands and acknowledgements to the drifters. The
base station maintains record of all the received and sent messages.
22
Figure 2.8: Work
ow of the Drifter Software.
Every time a message is sent through the SPOT messenger, the SPOT service sends an
e-mail to an address dedicated to drifters. The e-mail is parsed and the data is extracted
using the Gmail script. This e-mail script can be run on any computer with Internet
and need not be on the base station. The data collected is sent to another pipeline where
KML les are generated from them so that they can be viewed on Google Earth or Google
Maps. The CINAPS server at RESL fetches the KML Files from the base station. This
is displayed on a web interface. The user can request KML les to be generated for data
belonging to certain period of time.
23
Figure 2.9: Block diagram of overall system software. Drifters broadcast their status
and data at periodic intervals. They can receive commands from a base station or a
monitoring station on a research vessel.
The base station has a conguration le for its setup and le generation process. A
monitoring station or a user with an XBee can also speak to the drifters and base sta-
tion. Both the base station and monitoring stations can query the drifters for logged
les. Monitoring station can send conguration les for the base station and request to
download data from it. This results in a network being created for monitoring. Most of
the software was written in Python.
24
2.3.3 User Interfaces
One of the design considerations was to provide quality data to the user and allow
interaction with the drifters. To serve this purpose, dierent user interfaces were created
for visualization, interaction and recovery.
Web Interface
Figure 2.10 shows the web interface designed for the drifters. It is hosted on the
CINAPS server. The web page allows the user to get the data belonging to a particular
time interval. The user can view the drifter trajectories and points on Google maps. They
can download the KML les to view on Google Earth application or download the text
les containing the data. Whenever a user requests the data by entering the time interval,
scripts are run to generate the KML les and the text les.
The user can also change the update rate of the drifters. It varies from 20-400 minutes.
As soon as the user requests to change this value, a notication is sent to the base stations
to update. Upon an update, the acknowledgement sent by the drifters to the base station
is sent to the server which is notied to the user. Even though an acknowledgement is
not received the user can always send the update rate whenever required.
Drifter Communicator
Drifter Communication Software was designed to be used on a monitoring station on a
vessel or for an user on land or a boat. The software lists the devices forming the drifter
network. It provides an interface to send conguration les to the drifter. It also allows
to query the drifters to download on-board les. It displays the location of the drifters.
25
Figure 2.10: The web interface, through which the user can interact with drifters.
The software requires a laptop connected to an XBee modem. It also provides the status
info available. Finally, the user can also send the update rate required to a connected
drifter. Figure 2.11 shows the software with a particular drifter selected. We can always
add or remove devices.
Figure 2.12 shows a base station selected in the software. This is another feature which
allows the user to congure the base station remotely and allow for downloading the
data collected from the base station. The user can also request the data belonging to a
particular time-frame. Location of the base stations is also displayed. This software helps
the user in the eld doing tests or recovery or on a monitoring station at a particular
location. The software was built using the Qt framework.
26
Figure 2.11: Drifter Communicator Application with a drifter selected. The user can
download data from the drifter, send a conguration le, get its recent location when
connected and change the update rate.
Figure 2.12: Drifter Communicator Application with a base station selected. The user
can download data from base station and upload conguration les.
Recovery Module
Re usability of drifters mainly depends on ecient recovery. Since power consumption
is one of the design parameter, a drifter cannot constantly update its location. Delays
27
include acquiring position data from the GPS receiver, transmitting the location at pe-
riodic intervals, notifying the location to the boat units and travel time of the boats to
that location. This delay results in missing the drifters every time on a location update.
The boat used during recovery might not be high enough to to get a distant view in the
ocean.
Figure 2.13: Block diagram of the module to aid drifter recovery. This module receives
signals from the drifters using XBee and transmits the data to an Android phone or tablet
using bluetooth, to see the location of drifters.
The drifters have a radio which transmits the location every minute. Drifters may be
out of range from a base station, but a radio unit on a recovery boat can receive the
signal from the drifters once it is in the vicinity of its last location. A better thing would
be to see the location on a map. Laptops with XBee would require a stored map or wi
connectivity to download maps. It would also require a gps receiver to know its location.
A smartphone would be a good choice having on board gps and magnetometer and access
to data networks. A module was built with an XBee unit to receive the locations and a
blue-tooth module to send the location to a smart phone. Figure 2.13 shows the hardware
block diagram of the system.
28
Figure 2.14: Screen-shot of the Android App for drifter recovery. The red marker always
shows the boat location. The other markers show the latest location of the drifters that
connected to the recovery module.
An Android app was created to see the drifter location on Google Maps and the direction
for the boat movement. The app would constantly show the present location of the boat
and the latest location of the drifters which connected to the module. Figure 2.14 shows
a screenshot of the app.
29
2.4 Mechanical Design
The drifters need to have good current following capabilities. The drag area ratio i.e
the ratio of the area of the drogue by the area of the
oat needs to be larger than 40 for
ecient operation [9]. The drifters should not be moved by winds and waves and should
not rectify wave velocity eld [3] [6]. Also since we have a radio antenna which should
be above surface level, oscillations of the body due to wave action may lead to bad radio
communication. The drifters also need to be modular to be able to easily open them up ,
change batteries and the electronics. The Drifter Mechanical Design is closely related to
Microstar Drifter design as discussed in the related work section in introduction chapter
[1], [21]. The Drifters have three main mechanical components as listed below.
2.4.1 Surface Float
The surface
oat houses the drifter electronics and provides buoyancy stabilization.
Ideally, a spherical surface
oat is recommended to have the least eect from winds and
waves. But making a spherical
oat requires to building a mold which is time consuming.
During the testing phase, the
oat needs to be opened and closed quite often. A spherical

oat requires a proper closing mechanism which requires a lot of eort and testing. To
build a system within a month, we went for another option of having cylindrical surface

oats. Even though their working is not as great as spherical
oats, they are a compromise
between ideal case and time.
The body of the
oat is made of a PVC pipe of six inches diameter. The PVC pipe is
closed at one end and has a ring at the other end, to mount the electronics on a sheet of
ber glass laminate. This pipe also houses the batteries. The pipe is closed at the bottom
30
Figure 2.15: Cross sectional view of cylindrical surface
oat.
end using a PVC cap. This is modied to allow insertion of ballast material into this
cap. A threaded plug closes this ballast section. The other end of the pipe has another
PVC cap with a groove for an O-ring and a groove for the closing mechanism using a
nylon mono lament. The cap is modied for housing the antenna and protecting it. A
threaded plug covers a power switch at the top. Another threaded cap is provided on
31
Figure 2.16: Top view of the surface
oat
the body to interface with the sensors. Figure 2.15 and 2.16 show respectively the cross
sectional and top view of the
oat. Figure 2.17 shows a photo of the surface
oat.
2.4.2 Tether
The tether is made of a section of 1/8 inch type 316 stainless steel brous cable, with
loops formed at each end, using oval compression sleeves. The threaded plug to close the
ballast section of the
oat has an eyebolt attached, which provides a means of connecting
the tether to the
oat. Similarly an oval loop on the drogue connects to the tether. We
made tethers with lengths of 1 m, 3 m and 5 m. Since the tether is a detachable part,
the drifters can be drogued at any depth.
32
Figure 2.17: Picture of the Surface Float
2.4.3 Drogue
The drogue designed here is known as a corner radar re
ector type drogue. The drogue
consists of a section of 1 inch diameter PVC pipe with a threaded rod through the center.
There are plugs at each end to allow for bolting. The ends of the rod have an eyenut
to lock the drogue into place. The drogue is made up of rip stop nylon fabric that is
sewn from 2 square and 4 triangular pieces to form an octahedron. The octahedron is
constructed with a center tube section that slips over the PVC pipe. There are spars
33
Figure 2.18: Cross Sectional View of the Drogue.
made of polyester berglass that are inserted into a PVC block at the center and into
tubes sewn into the fabric that keeps the octahedron rigid. Figure 2.18 shows a cross
sectional view of the drogue, with measurements.
Figure 2.19 shows a completely assembled drifter. Foam had be put on the surface
oat
to provide better buoyancy.
34
Figure 2.19: Photo of a Completely Assembled Drifter
35
Chapter 3
System Characterization
The system was designed as per the design considerations listed in the previous chapter. In
this chapter we discuss the fulllment of the design considerations and any shortcomings
in the system.
3.1 Power Consumption
The system is intended to be out in the ocean for weeks to months. Power is an important
criterion to obtain such an endurance. Also data should be provided to the user at periodic
intervals so that they can use it for their studies. Low power usage saves cost on changing
batteries during deployments of several months.
The system has been characterized for its current consumption. A bench test was done
in the USC area that had GPS signal reception and satellite transmission. The radios
were transmitting to a nearby laptop acting as a base station. There are two dierences
between this test and the application scenario.
36
When the drifter is out in the ocean, it might have a lossy link with the base station
which makes it communicate more with the base station compared to the test done
here. So, the statistics shown in this test would have less current consumption than
a test in the ocean.
The ocean state might aect the time required to acquire a GPS x. The device
might oscillate due to the waves and might not have a clear view of the sky. The
radio antenna might also suer from oscillation and lead to losses in communication.
Humidity might also aect the range of communication to some extent.
The power consumption test ran for 6 hours. Figure 3.1 shows the current consumption
with 30 minutes of data update rate. SPOT Satellite GPS Messenger has a duty cycle
of 20 minutes wherein it rst acquires GPS x and then transmits the location to ensure
delivery. The ON time shown in the gure is when the SPOT Messenger is switched
on, and is performing the above mentioned operations. Also during this time, the radio
transmits periodically every minute. The SLEEP time is when the SPOT device is com-
pletely turned o and the micro-controller goes to sleep, only to wake up every minute
to transmit through the radios. Two cycles of ON and SLEEP time are shown in the
magnied image in the gure. The spikes during the ON/SLEEP time are due to radio
transmissions. During SLEEP time, except for XBee transmissions, the current consump-
tion is very small (in order of few milliamperes) resulting from the on-board regulators
and power supply circuitry. The large spikes, where the current consumption has crossed
500 mA, are during the ON time. These spikes occur during the satellite transmission. A
30-minute update rate is one cycle of ON (20) + SLEEP (10) time.
37
Figure 3.1: Current consumption plot for 30-minute data update. A magnied version
of the black box is shown in the gure. The ON time (20 minutes) in the gure shows
the current consumption when the SPOT Messenger is on and the radio is transmitting
periodically every minute. The SLEEP time (10 minutes here) is when the micro-controller
is in the sleep mode and the XBee transmits every minute. The spikes during the ON
and SLEEP time represent XBee transmissions.
Figures 3.2 , 3.3, 3.4 shows the current consumption for an update rate of 20, 40, and
60 minutes respectively. From these plots, the ON and SLEEP time can be clearly seen
for dierent update rates and the current consumed. The pattern in which the current is
consumed can also be seen from the gures.
38
Figure 3.2: Current consumption plot for a 20-minute data update rate.
Figure 3.3: Current consumption plot for a 40-minute data update rate.
39
Figure 3.4: Current consumption plot for a 60-minute data update rate.
Table 3.1: Average Current Consumption per hour
Update Rate (min) Avg Current Consumption (mA)
20 40.165
30 33.103
40 26.238
60 20.488
Table 3.1 shows the average current consumption per hour. Battery life is given by,
Battery Life =
Battery Capacity (mAh) 0:7
Device Consumption (mA)
40
0.7 is a constant taken into account for external factors which can aect battery life.
We use 5 Alkaline D-cell batteries which have a capacity of 20,000 mAh. Thus we have:
Battery Life =
20000 0:7
40:165
gives us approximately 15 days of life with a 20-minute update
rate. There is room to put more batteries in parallel, which would bump up the lifetime
to about a month. So the system can be out in the ocean for about a month on a single
battery pack.
A test was done to nd out the the actual lifetime of the drifters. The test was performed
by placing them on the rooftop of a building at USC. Similar to the current consumption
test, this is not the actual application scenario and the same dierences apply here as well.
It should also be noted here that there was no laptop acting as a base station because this
was a continuous test with drifters being left on the rooftop for days together. Since we
had built two drifters, one of them was congured to have a 20-minute update rate and
the other 30-minutes. The one with the 20-minute update rate lasted for approximately
14 days and the other lasted approximately for 23 days. This is quite close to the expected
battery life calculated above. Battery conguration might be the reason for a performance
worse than the expected one.
3.2 Cost
Drifters need to be cheap to enable deployment of a large number of them. A drifter may
be lost during operation in the ocean due to various factors such as a boat crashing into
it, people picking it up, and other external factors. We calculated the cost of construction
of each drifter. The cost calculated here is the unit cost and does not include the time
41
involved in building the drifters. Table 3.2 shows the unit cost of the various components
of the drifter. It also includes the overhead costs (damage costs) incurred during the
manufacturing process.
Table 3.2: Drifter Unit Cost
Component Cost ($)
Micro-controller board 60
Satellite transmitter and GPS 210
Radio modem with antennas 70
Batteries and miscellaneous 30
Surface
oat 400
Drogue and tether 100
Manufacturing process overhead 150
Total Cost 1020
Table 3.3: Other Costs
Component Cost ($)
Base Station setup 800
Recovery module 100
SPOT message yearly charge 100
Base Station maintenance per year 300-500
Drifter maintenance per year 500
42
Table 3.3 shows the costs involved in operating a drifter. Some of the costs in the table
are for an entire year of operation. That includes the battery costs (change every month
given the lifetime) and the travel cost involved for its deployment, battery changing, and
recovery.
The cost estimates here do not include the time involved in building the drifters. Time
is an important factor since it accounts for the hourly rate of the personnel involved in
the manufacturing process. Since two drifters were built initially, the component cost
listed in table 3.2 are calculated based upon the components acquired for these two
drifters. Building more drifters would reduce the component costs. Since the scale of
drifter construction is very small, we have not included the cost for time involved. In
order to build the drifter within a certain time frame (within a month), the surface
oat
was built to be cylindrical and was made out of PVC pipe. This resulted in a large cost
for the surface
oat which otherwise was supposed to be low. Given more time, it could
have been built with ber glass by creating molds for a spherical
oat which would cost
$100 - 150 as opposed to a $400
oat.
3.3 Modularity
Being able to follow the currents at dierent depths was one of the important design
considerations. To accomplish this, the surface
oat, tether, and drogue were designed as
separate modules. Dierent depths can be followed by changing the tether size. Dierent
kinds of drogues can also be attached easily. The surface
oat cap can be easily opened
by a single person to allow for accessing the electronics. It can easily be deployed by a
single person by using the power switch on top of the surface
oat. No other device is
43
required to launch it. The drifter operation can be congured for a mission in the lab by
just changing a le (cong.txt) and then sent for deployment. The device also provides
options for the user to change the update rate and select the telemetry device to be used.
3.4 Current Following Capability
A test was conducted at Catalina Islands to check whether winds and waves have any
eect on the drifters. A drifter was deployed close to the dock. The drifter was drogued
at 1 m. Even though currents are fairly low near the dock, wind and wave action were
present. Figure 3.5 shows the location data of the drifter. As the gure shows, the drifters
moved less than 2 m over a span of 40 minutes. This substantiated that the design was
not aected by the winds and the waves.
3.5 Re-usability
The user should be able to reuse the drifters for multiple deployments. This requires
ecient deployment and more importantly, recovery. To facilitate ecient deployment,
the system was designed in such a way that a single person could take it on a boat and
deploy it, by turning it on using a switch on top of it. The drifter is modular so it can be
easily assembled and disassembled. For recovery, a recovery module and an Android app
were built. Eorts were also made to provide good location data as much as possible.
44
Figure 3.5: The current following capability of the drifter was tested by deploying a drifter
near the dock. The currents were fairly low in that region. The drogue was at a depth of
1m. The drifter drifted only about 2 m over a span of 40 minutes. This shows that the
drifters were not aected by surface winds or waves.
3.6 Resolution
The current consumption and battery life tests discussed in Power Consumption section
of this chapter suggest that a 20-minute update rate over satellite provides the best
45
resolution for operation. Even though radios can provide better resolution data, they
become increasingly unreliable as the distance between the drifter and the base station
increases. So, we use them only as a means of secondary data.
3.7 Data Delivery and Quality
The drifter battery life test done on a building's rooftop gave insight into the data
delivery and accuracy. As mentioned earlier, the drifter with a 20-minute update rate
lasted for 14 days and the one with an update rate of 30-minute lasted for 23 days.
Table 3.4 shows the number of messages delivered during the test. The expected number
of messages are calculated by the number of hours the drifters were active, and the delays
caused by the system and the drift in the system clock. For example, the 20 minute
update rate drifter lasted for 13 days and 16 hours. We divide this (13 24 60 + 16 60
minutes) by 22 (20 + 2 minutes for delays and any clock drift) to get 894. The delivered
column shows the number of delivered messages by the SPOT tracker. Almost 90% of the
messages were delivered. The reasons for losses may be many. Since the SPOT device is
a simplex device, it has no way of knowing whether the message was delivered or not. So
it sends the signal for a xed number of times. The device might also have had diculties
in obtaining a x. But this possibility is very small. Data from the radios and the SD
card on the drifter helps in obtaining the missing data. Furthermore, the update rate
through these secondary means is high (1 min) which helps in providing better data to
the user.
46
Table 3.4: Messages Delivered
Update Rate Expected Delivered %
20 min 894 798 89.26
30 min 994 883 88.88
This test also helped in analyzing the position accuracy of the system. Figure 3.6
shows the location data for both the drifters that were tested. The drifter with a 20-
minute update rate has data (red dots) with more points near to the center and some
points o signicantly from center. On the other hand, the system with a 30-minute
update rate has data more spread out at the center but has fewer outliers (blue dots).
This may be entirely related to the GPS receiver on both of them.
Table 3.5 shows the mean latitude and longitude of the drifters that are the coordinates
on the AHF rooftop where the drifters were placed. The distance root mean square
(DRMS) error, which gives the 2D accuracy, is calculated after converting the coordinates
to Universal Transverse Mercator (UTM) to get the accuracy in meters. As the table
shows, one of them had 21 m accuracy and the other 7.7 m over the time span of their
operation. These errors are mainly due to the outliers in the data. But as the mean
suggests, most of the time they have transmitted data close to the point of their placement.
Another important factor is the delay in data telemetry. Comparing the time of GPS x
obtained in the log les on the SD Card of the drifters and the time of data delivery using
the SPOT Satellite Messenger, the delay between a x and transmission was found out
to be anywhere between 1 10 minutes. This means a location data may be 10 minutes
47
Figure 3.6: Position data for the drifters. The top gure (red dots) is of the drifter with a
20-minute update rate and the bottom gure (blue dots) is for a 30-minute update rate.
We can see that some of the points in rst picture are outliers. Second picture has more
points near to the center.
old in the worst case which may be a problem during recovery. But the use of radios
and the recovery module prevents this. Also, when a drifter is in view of a base station,
a better data update rate is obtained and this acts as secondary data. This provides
better tracking. The data is also logged in the SD Card on the drifters which helps in
48
Table 3.5: GPS Accuracy
Update Rate Mean Lat Mean Lon DRMS
(minutes) (meters)
20 34.0196462454 -118.284962289 21.192
30 34.019649932 -118.284967259 7.772
interpolating the missing data. The delay between the time when a x is obtained and
the time when it is stored on the SD Card is of the order of milliseconds, and transmission
over radio is of the order of seconds in the worst case. Data is also provided to the user
through a user friendly interface (web and standalone). Also the negligible eect of wind
and waves on the system was discussed in earlier sections, substantiating that negligible
position error would occur due to these factors.
3.8 Future Enhancement
The device can be enhanced by adding extra sensors. The board was designed to
provide both analog and digital interfaces for dierent sensors. USB and RS232 ports
provide interface for a large range of sensors (e.g.,
uorometers). Also the UART (3.3 V)
lines would facilitate serial communication. SPI/I2C ports help in interfacing digital
output sensors (accelerometers, gyroscopes, magnetometers, etc.) and data can be read
serially from the devices by using these protocols. Analog to digital converters help in
tapping a large number of analog signals of the world (temperature, pressure, etc.). The
surface
oat has a port to interface the sensors. Finally, dierent battery congurations
could be used based upon the mission.
49
Chapter 4
Field Tests
4.1 Initial Test
The rst test with the drifters was done at Catalina Islands. A drifter with a 30-minute
update rate was deployed at 12:54 PM on Oct 7, 2013. Its initial location is shown by a
green marker in Figure 4.1. The drifter was drogued at 1 m depth. The drifter started
to move towards the east direction until 10:00 PM. Then it changed its direction moving
west till about 2:15 AM in the morning and then again turned towards east, before being
picked up at 8:35 AM. The last reported location was at 8:19 AM (shown by the red
marker in Figure 4.1).
Figure 4.2 shows the sampling locations. It should be noted that even though the
update rate was set for 30 minutes, the time dierence between two markers would not
be exactly 30 minutes due to various delays including the transmission delay of the SPOT
Device. The maximum speed attained by the drifter was 0.536 m/s which is between the
third and fourth sampling locations. Most of the time the speed was close to 0.1 m/s.
50
Figure 4.1: Trajectory of the drifter during the rst test at Catalina Islands. The green
marker shows the starting point on Oct 7, 2013 at 12:54 PM and the red marker shows
the end point on Oct 8, 2013 at 8:19 AM. The north of the picture is the actual north of
the earth. The update rate for this drifter was 30 min.
Figure 4.2: Locations transmitted by the SPOT Satellite Messenger during the rst test.
51
Figure 4.3 shows the comparison between the data obtained from the SPOT Tracker
and the SD card. As we can see from the gure, the data is almost the same for both the
sources except at a few places. Few missed out points on the satellite data can be seen
in the SD Card data.
Figure 4.3: Comparison between data obtained from the SPOT Tracker (green) and the
SD Card (red).
One of the major failures of the test was radio communication. The drifters were
not able to communicate with the base station on the hilltop at Catalina Islands. The
dierence from the XBee test discussed before was that the antenna was placed inside the
drifter with a cap. Since the
oat was cylindrical, its oscillations could have aected the
communication. The phase of the signal could have been aected due to the oscillations
of the
oat. Also, the wave in front of the antenna might have aected the signal by
blocking it. These were some of the reasons investigated. Even though we had tested the
52
XBee radios before using a boat, we had not tested it out in the eld after incorporating
them into the drifters.
Figure 4.4: Picture of a drifter in water
4.2 Radio Test
We tested out more extensive tests of radio communication for the drifters because
of the radio failure during the rst test. These tests were done at Catalina Islands.
Figure 4.5 shows the points of the test. The drifters were deployed at each of these points
for a couple of minutes and they transmitted data every 5 seconds. Dierent antennas
were used initially but nally the 2.5 dB antenna was chosen because of its superior
performance. In order to reduce the oscillations, the drifter was lled with lead at the
bottom to give it better stability.
Table 4.1 shows the statistics of the test. The points were selected to be close to 1.5,
2, 3, and 4 miles respectively. Packets were transmitted every 5 seconds. The table
shows the number of packets transmitted, received (corrupted or uncorrupted), and the
53
Figure 4.5: Radio test points. The drifter was left in the water for a couple of minutes at
these points. The drifters were transmitting their location every 5 sec.
Table 4.1: Radio Test Statistics
Marker Distance (miles) Transmitted Received Corrupted %
Red 1.46 107 90 8 76.64
Pink 2.19 76 61 6 72.36
Green 3.34 177 133 16 66.10
Yellow 4.28 135 87 10 57.03
percentage of packets delivered uncorrupted. As we can see, the eciency decreases with
distance. The eciency is not the same as in the previous radio test conducted using a
boat because of dierent conditions. In this test, the antenna was only 3-5 inches above
54
water which aected its eciency. The waves, being good absorbers of radio signals, aect
the radio communication. One solution to this could be to increase the antenna height,
but that would increase the eect of wind on the drifters. The radio test have shown the
potential radio performance for the drifters. Given that a very low cost radio was used,
performance was great.
4.3 Test at Hyperion Sewage Treatment Plant
The motivation for this work came from an activity at Hyperion Sewage Treatment
Plant as discussed in Chapter 1. We tested our drifters exactly at the end of 1 mile pipe
coming out from the plant. Figure 4.6 shows the trajectory followed by the drifter. The
drifter was deployed next to the 1-mile outfall, which is shown by the green marker. The
drifter was drogued at 3 m depth. The base station was not set up here because it required
permissions and logistic support from the plant. The drifters were deployed on 20 Nov,
2013 at 8.45 AM. The drifter moved south, i.e away from the coast, till 11:00 AM. Then it
changed its course and started moving east until 7:30 PM in the evening. Then it moved
west for some time and north west till 3:10 AM on 21 Nov, 2013. Then it moved to the
east again till afternoon and to the west till 5:00 AM on 22 Nov, 2013. It was recovered
at 8:15 AM on 22 Nov, 2013 and the point of recovery is shown by the red marker in the
gure.
From the movements described above, it seemed that the drifter was moving away from
the coast from morning till noon and towards the coast till night. This may be due to
the upwelling activity in the ocean. The nearest point reached by the drifter from the
coast was 0.29 miles. It was at a depth of 7 m. This test over 2 days substantiated that
55
Figure 4.6: Trajectory of the drifter deployed near the 1-mile pipe at Hyperion Sewage
Treatment Plant. The green marker shows the starting location and the red marker shows
the last location. The drifter was deployed on Nov 20, 2013 at 8:45 AM and recovered on
Nov 22, 2013 at 8:15 AM. The drifter remained close to the shore during the test showing
that the outfall from the 1-mile pipe may stay near the shore.
the outfall from the 1-mile pipe may stay close to the coast, thus aecting it. One of the
problems faced during the test was the SPOT outage. The SPOT servers were down for
some time on Nov 21. But the presence of the SD Card on the drifter helped in retrieving
all the data.
Figure 4.7 shows the drifter trajectory using the ROMS data for 1 m depth. Figure 4.8
shows the output for ROMS data at 10 m depth. It clearly shows that the data is
completely opposite to the test done at Hyperion. Even though our drifters were drogued
at 3 m depth and ROMS provides data at 1 m and 10 m depths, it is nowhere close to
56
Figure 4.7: Drifter trajectory using ROMS output for 1 m depth
Figure 4.8: Drifter trajectory using ROMS output for 10 m depth
the data obtained by the test. This indicates that ROMS is not a good choice to study
57
the diversion activity at Hyperion because of the sparsity in the data and the grid size of
the test area.
4.4 Test With Multiple Drifters
Figure 4.9: Trajectory of two drifters near Catalina Islands. One of them was drogued at
5 m depth (green) and the other one at 1 m depth (red). They were dropped o at the
same location. Their nal separation was 400 m.
The direction of the current varies at dierent depths. In order to test this, two drifters
were deployed on March 6, 2014 at 11:12 AM at Catalina Islands. Figure 4.9 shows
the trajectory of the two drifters. The green trajectory is that of a drifter drogued at
5 m depth and the red one was drogued at 1 m depth. They were dropped at the same
location about a mile from the shore (shown by the green marker). The drifters moved
58
westwards during the course. For an hour their separation was less than 100 m. The test
continued till 1:38 PM. Their nal separation was close to 400 m. They had started to
move apart with time even though their drop-o location was the same. After about an
hour of deployment, both of them started moving fast at about 0.6 m/s.
Figure 4.10: The top gure shows the drifter trajectory by combining both the SPOT
data and XBee data. The bottom gure is a comparison between the SPOT data and the
SPOT+XBee data. Red and blue lines are for the drifter with drogue at 5 m. The green
and magenta lines are for the drifter with drogue at 1 m.
59
The radios were working during this test and we were able to obtain the data from them.
Figure 4.14(a) shows the trajectory obtained from combining data from SPOT and XBee.
It also compares the trajectory obtained just by using SPOT data and trajectory obtained
from combining SPOT and XBee data. Red (SPOT) and blue (SPOT + XBee) lines are
for the drifter drogued at 5 m. Green (SPOT) and magenta (SPOT + XBee) lines are for
the drifter drogued at 1 m . We can see that intermediate information can be obtained
using the XBee data. The location data was 10 points using SPOT device. Combining
SPOT with XBee data gave us 86 points. This was really more data than just the SPOT
tracker alone. We also tested the communication of the drifters to the boat. We were
able to receive signals from the drifter in the ocean to the boat for a distance of up to
about 700 m.
4.5 Final Test
A nal test of the drifters was conducted at Catalina Islands. Three drifters were
deployed at the same location at 11:15 AM and recovered back by 2:15 PM. Figure 4.11
shows the trajectory followed by the 3 drifters. The drifters were drogued at 5 m (red), 3 m
(green), and 1 m (yellow). Initially the separation between them was less than 100 m.
It was interesting to see that the separation between the drifters drogued at 3 m and
5 m increased with time. Another interesting observation of the test was that the drifter
drogued at 1 m moved very little compared to the other two. It remained very close to the
deployment location for 2 hours and then started moving in the nal hour. The dierent
patterns of current movement can be seen in the gure. The test here and the previous
one discussed above suggest that currents move in dierent directions at dierent depths.
60
Figure 4.11: Trajectory of three drifters near Catalina Islands. One of them was drogued
at 5 m depth (red), one at 3 m depth (green), and the last one at 1 m (yellow).
Figure 4.12: Deployment of three drifters in the ocean
61
Figure 4.13: Visualization of Data on Web Interface
Another part of this test was the use of the Hobo Pendant sensors that can measure
temperature. These small sensors were attached to the bottom of the
oat and to the
drogue making it two sensors per drifter. This was done to test the temperature variations
with depth. Figure 4.14 shows the comparison of the temperature data for the three
drifters. The initial dip in the temperature at the starting of the plot shows the transition
of the drifters from boat to water.
62
Figure 4.14: Temperature data using the Hobo Pendant Sensor. The top gure shows the
data for the sensor attached to the
oat. The bottom plot shows the data of the sensor
attached to the drogue. Blue line represents the drifter at 1 m depth, red at 3 m depth,
and green at 5 m depth.
63
Chapter 5
Conclusion and Future Work
5.1 Conclusion
This thesis described the design and development of a Lagrangian Drifter for ocean
monitoring applications. A thorough survey of related work and the present day tech-
nologies was conducted before designing the system. The design considerations were laid
out depending on the applications for the device and based on the environment it would
be operated in. We performed the electrical design in-house at RESL, USC by designing
and fabricating a micro-controller board and integrating the available positioning and
telemetry systems. We took into account both the operation of the drifters and the user
requirements while designing the software for the system. We developed various software
tools to aid the user and for ecient deployment and recovery. The mechanical design
caters to the robustness required in marine conditions. The system was characterized
64
in order to check whether it fullled the design considerations. We analyzed any short-
comings of the system and tried to address them. We were able to meet most of the
considerations.
We tested the system extensively out in the ocean. Nearly 5 campaigns in the ocean
showed the applicability of our system in practical situations. We conducted single drifter
tests to validate the feasibility of our system and identied the improvements required.
We gave an insight into the variations of currents in the ocean by conducting tests with
multiple drifters and also showed how the radio could be used to obtain more data. We
used software tools for data visualization, recovery, and processing. We presented various
results in the form of drifter trajectories and statistics for the tests conducted. We discuss
the future work in the next section.
5.2 Future Work
Some of the design considerations could not be met completely due to time and logistical
constraints. Some enhancements can also be made to the drifter design presented here.
We discuss these two aspects in this section.
5.2.1 Surface Float
As we discussed earlier, a surface
oat should not be aected by winds and waves and
should provide good buoyancy stabilization. Also, the oscillation of the
oat in water can
cause communication problems. The ideal design of a
oat is to make it spherical. We
designed a cylindrical
oat using PVC materials due to time constraints. In the future,
we intend to design a spherical
oat made out of berglass material. This would give us
65
all the advantages of a spherical
oat as well as reduce the cost of the
oat from $400 to
$100 150. Care has to be taken while designing the closing and opening mechanism of
the
oat due to the development phase considerations.
5.2.2 Drogue Detachment Detection
One of the features not addressed in this thesis is the detection of the drogue detachment
when the drifters are in water. The primary goal of this work was to design a system
and get its basic functionality. Normally submergence sensor is used to detect the drogue
detachment [1], [2], [21]. When a drogue detachment is detected, the drifters send an
alert signal indicating the detachment. With this the user can know that the system is no
longer following the current at the depth which it was intended to. So the user can ignore
the data coming from it for her study, or go and recover it. We also looked into using
an accelerometer for this purpose. Since we have a cylindrical
oat, it tips over when
a drogue is not attached to it and stays upright when attached. So if we use a triaxial
accelerometer, the g-force would be acting on X or Y axis of the accelerometer. This
could probably be used as a mechanism to detect drogue detachment in cylindrical
oats.
But this changes for spherical
oats because they stay upright even when the drogue is
detached. The only dierence here would be that it might have a pull from the drogue
when it is attached and no pull during detachment. So submergence sensors may be good
for spherical
oats.
5.2.3 Communication
One of the problems we faced with the radio communication was the oscillation of
the
oat. This resulted in continuous oscillation of the antenna resulting in bad line of
66
sight. Moreover, since the antennas were close to the water surface, the waves being good
absorbers of radio signal would aect communication. The oscillation can be reduced
by using spherical
oats. The problem from the waves can be solved by using adaptive
communication based upon wave parameters [25]. High gain antennas may also be used
for better communication. The SPOT Messenger worked well for us but we can send only
location data through them. Low cost satellite telemetry systems can be looked into to
send more data.
5.2.4 Sensing
As discussed previously, the drifter designed here can be enhanced with lots of sensors
since it has the interfaces to accommodate them. But one of the key things would be to
measure temperature. This will be useful for marine biologists for their research. Even
though we measured temperature at surface and the drogue depth using a Hobo Pendant
Sensor, a system of interfacing thermistor chains should be looked into. Thermistor chains
could be developed specically for the drifters discussed here so that temperature at dif-
ferent depths can be measured as the drifter moves along with the water. Accelerometers
can be used to provide wave data. Sensor packages could also be attached to the drogue
so that they move along with it and provide sensing at that depth.
5.2.5 Software
Most of the data in the base station and the servers is le-based. Eorts have to be
made to use database systems for the data. Currently, location is the only data available
to the user for visualization. Software needs to be developed for providing sensor data
and visualizing it. The present UI's can be made more user friendly.
67
5.2.6 Similar Systems
The test with multiple drifters showed how the currents at dierent depths vary in
their directions in the ocean. This suggests a system design that can change the depth of
the drogue adaptively based on some data [26], [27]. This would provide the capability of
proling a vertical section and also hovering over a dened bounding area. A winch-based
mechanism can be used to change the depth of the drogue. Since we have a degree of
motion now, methods of running cables between the sensors at the drogue and electronics
at surface
oat have to be investigated. Also a easy way of communication (optical,
acoustic etc) between sensors and electronics can be developed.
68
BIBLIOGRAPHY
[1] Ohlmann C, White P, Sybrandy A, Niiler P.,2005.GPS-cellular drifter technol-
ogy for coastal ocean observing systems. J Atmos Ocean Tech 22:1381-1388
[2] Lumpkin, R., and M. Pazos, 2006: Measuring surface currents with Surface Veloc-
ity Program drifters: The instrument, its data, and some recent results. Lagrangian
Analysis and Prediction of Coastal and Ocean Dynamics (LAPCOD), A. Gria et
al, Eds., 39-67.
[3] Davis, R. E.,1985: Drifter observations of coastal surface currents during CODE:
The method and descriptive view. J. Geophys.Res., 90, 4741-4755.
[4] A. Shchepetkin and J. McWilliams, The regional oceanic modelingsystem
(ROMS): a split-explicit, free-surface, topography-following-coordinate oceanic
model, Ocean Modeling,vol. 9, pp. 347-404, 2005. doi:10.1016/j.ocemod.2004.08.002.
[5] R. N. Smith, Y. Chao, B. H. Jones, D. A. Caron, P. P. Li,G. S Sukhatme,
"Trajectory design for autonomous underwater vehicles based on ocean model pre-
dictions for feature tracking, in Proc. 7th Int. Conf. Field and Service Robotics ,
Cambridge, MA, July 2009.
[6] George, R., and J. L. Largier, 1996:Description and performance of nescale
drifters for coastal and estuarine studies. J. Atmos.Oceanic Technol., 13, 1322-1326.
[7] Hitchcock, G. L., D. B. Olson, S. L. Cavendish, and E. C. Kanitz, 1996:
A GPS-tracked surface drifter with cellular telemetry capabilities. Mar. Technol.
Soc. J., 30, 44-50.
[8] Johnson, D., R. Stocker, R. Head, J. Imberger, and C. Pattiaratchi,
2003: A compact, low-cost GPS drifter for use in the oceanic nearshore zone, lakes,
and estuaries. J. Atmos. Oceanic Technol., 20, 1880-1884.
[9] Niiler, P. P., A. L. Sybrandy, K. Bi, P. M. Poulain, and D. Bitterman,
1995: Measurements of the water following capability of holey-sock and TRISTAR
drifters. Deep-Sea Res., 42, 1951-1964.
69
[10] Kennan, S., P. P. Niiler, and A. Sybrandy, 1998: Advances in drifting buoy
technology. International WOCE Newsletter,30, 7-10.
[11] Niiler, P. P., R. E. Davis, and H. J. White, 1987: Water-following charac-
teristics of a mixed layer drifter. Deep-Sea Res., 34, 1867-1881.
[12] Comprehensive Summary of The Hyperion Treatment Plant Euent Diversion of
November 2006.
[13] Niiler, P. P.,2003:A brief history of drifter technology. Autonomous and La-
grangian Platforms and SensorsWorkshop,Scripps Institution of Oceanography, La
Jolla, California.
[14] Sybrandy, A. L. and P. P. Niiler,1992: WOCE/TOGA Lagrangian drifter con-
struction manual. WOCE Rep. 63, SIO Ref. 91/6, 58 pp., Scripps Inst. of Oceanogr.,
La Jolla, Calif. California.
[15] Crawford, W. R.: 1988, The use of Loran-C drifters to locate eddies on the conti-
nental shelf,J. Atmos. Oceanic Technol. 5, 671-676.
[16] Woodward, M.J.; Crawford, W.R.; Drinkwater, K.F.; Hill, M.J.,"Lagrangian
Measurement Of Surface Currents Using Loran-c Drifters," OCEANS '91. Ocean
Technologies and Opportunities in the Pacic for the 90's. Proceedings. , vol.2, no.,
pp.1190,1196, 1-3 Oct 1991 doi: 10.1109/OCEANS.1991.628216
[17] MUZZI, R. W., AND M. J. MCCORMICK:1994. A new global positioning system
drifter buoy. J. Gt. Lakes Res. 3: 1-4.
[18] Johnson D, Stocker R, Head R, Imberger J, Pattiaratchi C:(2003) A com-
pact, low-cost GPS drifter for use in the oceanic nearshore zone, lakes, and estuaries.
J Atmos Ocean Technol 20:1880-1884.doi:10.1175/1520-0426(2003)
[19] Perez, J.C.; Bonner, J.; Kelly, F.J.; Fuller, C., "Development of a cheap,
GPS-based, radio-tracked, surface drifter for closed shallow-water bays," Cur-
rent Measurement Technology, 2003. Proceedings of the IEEE/OES Sev-
enth Working Conference on , vol., no., pp.66,69, 13-15 March 2003 doi:
10.1109/CCM.2003.1194285
[20] McNally, G. J., W. C. Patzert, J. A D Kirwan, and A. C. Vastano, 1983:
The near-surface circulation of the North Pacic using satellite tracked drifting buoys.
J. Geophys. Res., 88, 76347640.
[21] Pacific Gyre Microstar GPS Drifter,Online:http://www.pacicgyre.com/les/
microstar.pdf
[22] Davis, R. , "An inexpensive drifter for surface currents," Current Measurement,
Proceedings of the 1982 IEEE Second Working Conference on , vol.2, no., pp.89,93,
0-0 Jan. 1982 doi: 10.1109/CCM.1982.1158429
70
[23] Pereira, A., Heidarsson, H., Caron, D., Jones, B.and Sukhatme,G. (2009).
,"An implementation of a communication framework for the cost-eective operation
of slocum gliders in coastal regions. Proceedings of the 7th International Conference
on Field and Service Robotics, Cambridge, MA
[24] Smith, R. N., Das, J., Heidarsson, H., Pereira, A., Cetinic, I.,
Darjany, L.,et al (2010a). "USC CINAPS builds bridges : Observing and
monitoring the Southern California Bight",IEEE Robotics and Automation
Magazine, 17: 2030.
[25] Pereira, Arvind A.; Sukhatme, G. "Estimation of wave parameters from ac-
celerometry to aid AUV-shore communication" OCEANS 2010 IEEE - Sydney , vol.,
no., pp.1,10, 24-27 May 2010
[26] Ryan N. Smith and Matthew Dunbabin. "Controlled drift: An investigation into
the controllability of underwater vehicles with minimal actuation" In Proceedings of
the Australasian Conference on Robotics and Automation,July 2011.
[27] Matthew Dunbabin. "Optimal 4D path-planning in strongly tidal coastal environ-
ments: Application to AUVs and Proling Drifters" In Proceedings of the Workshop
on Robotics for Environmental Monitoring,RSS 2012
[28] J. Gonzlez, I. Masmitj, S. Gomriz, E. Molino, J. del Ro, A. Mnuel,
J. Busquets, A. Guerrero, F. Lpez, M. Carreras, D. Ribas, A. Carrera,
C. Candela, P. Ridao, J. Sousa, P. Calado, J. Pinto, A. Sousa, R. Martins,
D. Borrajo, A. Olaya, B. Garau, I. Gonzlez, S.Torres, K. Rajan, M. McCann,
J. Gilabert. 2012. "AUV Based Multi-vehicle Collaboration:Salinity Studies in
Mar Menor Coastal Lagoon ",In: Navigation, guidance and control of underwater
vehicles-NGCUV 2012, Porto
[29] SPOT Satellite GPS Messenger,Online:http://www.ndmespot.com/downloads/
SPOT2-SellSheet.pdf
[30] XBee-PRO 900HP Radio Modem,Online:https://www.digi.com/pdf/ds xbeepro900hp.pdf
[31] MSP430F5438A,Online:http://www.ti.com/lit/ds/symlink/msp430f5438a.pdf
[32] FatFs - Generic FAT File System Module,Online:http://elm-
chan.org/fsw//00index e.html
71

Abstract (if available)

Abstract This thesis describes the design and development of a Lagrangian Drifter for ocean monitoring applications. The system is developed to move with the currents at a particular depth, transmitting its location to the user. Through this design, the user can track ocean currents and object or entities carried by them. This thesis introduces the system by giving the motivation for the work and a survey of related work and present day technologies available to design the system. The design considerations are laid out based upon the application of the device. A description of the electrical, software and mechanical design of the system is presented here. The system was characterized for its defined design considerations. Finally, experiments and results are presented and concluded with a discussion on future enhancements for the system.

Linked assets

University of Southern California Dissertations and Theses

Conceptually similar

PDF

Data-driven robotic sampling for marine ecosystem monitoring

PDF

Informative path planning for environmental monitoring

PDF

A robotic system for benthic sampling along a transect

PDF

Risk-aware path planning for autonomous underwater vehicles

PDF

USC microgrid development conceptual plan

PDF

Compilation of data-driven macroprograms for a class of networked sensing applications

PDF

Navigation and guidance of an autonomous surface vehicle

PDF

Ultrasonic microelectromechanical system for microfluidics, cancer therapeutics and sensing applications

PDF

New Lagrangian methods for constrained convex programs and their applications

PDF

Large system analysis of multi-cell MIMO downlink: fairness scheduling and inter-cell cooperation

PDF

Communication and cooperation in underwater acoustic networks

PDF

Joint communication and sensing over state dependent channels

PDF

Multi-robot strategies for adaptive sampling with autonomous underwater vehicles

PDF

Synthetic and bio-inspired molecular communications

PDF

Advancing robot autonomy for long-horizon tasks

PDF

Novel variations of sparse representation techniques with applications

PDF

Distribution system reliability analysis for smart grid applications

PDF

Socially assistive and service robotics for older adults: methodologies for motivating exercise and following spatial language instructions in discourse

PDF

Integrated wireless piezoelectric ultrasonic transducer system for biomedical applications

PDF

Vision-based studies for structural health monitoring and condition assesment

Asset Metadata

Creator Subbaraya, Supreeth (author)

Core Title Development of Lagrangian drifter for ocean monitoring and marine applications

School Viterbi School of Engineering

Degree Master of Science

Degree Program Electrical Engineering

Publication Date 04/22/2014

Defense Date 03/24/2014

Publisher University of Southern California (original), University of Southern California. Libraries (digital)

Tag drifter,drogue,marine system,OAI-PMH Harvest,ocean monitoring

Format application/pdf (imt)

Language English

Contributor Electronically uploaded by the author (provenance)

Advisor Sukhatme, Gaurav S. (committee chair), Caron, David A. (committee member), Mitra, Urbashi (committee member)

Creator Email subbaray@usc.edu

Permanent Link (DOI) https://doi.org/10.25549/usctheses-c3-382246

Unique identifier UC11296975

Identifier etd-SubbarayaS-2390.pdf (filename),usctheses-c3-382246 (legacy record id)

Legacy Identifier etd-SubbarayaS-2390-0.pdf

Dmrecord 382246

Document Type Thesis

Format application/pdf (imt)

Rights Subbaraya, Supreeth

Type texts

Source University of Southern California (contributing entity), University of Southern California Dissertations and Theses (collection)

Access Conditions The author retains rights to his/her dissertation, thesis or other graduate work according to U.S. copyright law. Electronic access is being provided by the USC Libraries in agreement with the a...

Repository Name University of Southern California Digital Library

Repository Location USC Digital Library, University of Southern California, University Park Campus MC 2810, 3434 South Grand Avenue, 2nd Floor, Los Angeles, California 90089-2810, USA