Close
About
FAQ
Home
Collections
Login
USC Login
Register
0
Selected
Invert selection
Deselect all
Deselect all
Click here to refresh results
Click here to refresh results
USC
/
Digital Library
/
University of Southern California Dissertations and Theses
/
Data overload in unmanned aircraft systems: improving bandwidth utilization through wavelet compression
(USC Thesis Other)
Data overload in unmanned aircraft systems: improving bandwidth utilization through wavelet compression
PDF
Download
Share
Open document
Flip pages
Contact Us
Contact Us
Copy asset link
Request this asset
Transcript (if available)
Content
DATA OVERLOAD IN UNMANNED AIRCRAFT SYSTEMS:
IMPROVING BANDWIDTH UTILIZATION THROUGH
WAVELET COMPRESSION
by
Mary Elizabeth Parker
___________________________________________
A Thesis Presented to the
FACULTY OF THE USC GRADUATE SCHOOL
UNIVERSITY OF SOUTHERN CALIFORNIA
In Partial Fulfillment of the
Requirements for the Degree
MASTER OF SCIENCE
(GEOGRAPHIC INFORMATION SCIENCE AND
TECHNOLOGY)
May 2013
Copyright 2013 Mary Elizabeth Parker
ii
ACKNOWLEDGEMENTS
I would like to thank my committee, particularly the chair,
Jordan Hastings. Without his encouragement, guidance, and
persistent help this thesis would not have been possible.
I would like to express my deepest appreciation to the Intergraph
Corporation, especially Andrew Pursch and Chris Powell. Their
continued support and enthusiasm enhanced the quality of my
research and the caliber of this thesis.
iii
TABLE OF CONTENTS
Acknowledgements .............................................. ii
List of Figures ............................................... iv
Abstract ....................................................... v
Introduction ................................................... 1
Chapter One: Background ........................................ 3
Unmanned Aircraft Systems ................................... 3
Image Compression ......................................... 10
Chapter Two: Resources and Research Phases .................... 17
Chapter Three: Execution ...................................... 21
Chapter Four: Results ......................................... 23
Part 1 .................................................... 23
Part 2 .................................................... 28
Chapter Five: Future Research ................................. 31
Conclusion .................................................... 34
Bibliography .................................................. 35
APPENDIX A: Events Log from Solarwinds ........................ 42
Appendix B: Total bandwidth utilization ....................... 43
APPENDIX C: Raw data transfer bandwidth utilization ........... 44
APPENDIX D: Raw data transfer response time ................... 45
APPENDIX E: ECW data transfer bandwidth utilization ........... 46
APPENDIX F: ECW data transfer response time ................... 47
iv
LIST OF FIGURES
Figure 1: Concept of Operations for Unmanned Aerial System ..... 5
Figure 2: Example of Huffman Coding ........................... 12
Figure 3: Discrete Cosine Transform Process ................... 15
Figure 4: Discrete Wavelet Transform Process .................. 16
Figure 5: Network Diagram ..................................... 18
Figure 6: Comparison of MrSID raw data and ECW data ........... 25
Figure 7: Comparison of RPF raw data and ECW data ............. 25
Figure 8: Pixel data for RPF frame and after ECW conversion ... 26
Figure 9: Solarwinds analysis of bandwidth utilization during
raw data transfer ............................................. 28
Figure 10: Solarwinds analysis of bandwidth utilization during
ECW transfer .................................................. 29
Figure 11: Apollo Web Client .................................. 32
Figure 12: SharePoint® site with Apollo Web Client link ....... 33
v
ABSTRACT
Between 2008 and 2010, the number of unmanned aircraft
systems (UAS) in the military increased by 330% in support of
operations throughout the Middle East (Iraq, Afghanistan, Iran,
etc.). The Pentagon has developed numerous initiatives to enhance
the overall performance of UASs, demonstrating that reliance on
and deployment of these systems is expected to continue. Via
real-time aerial imagery, UASs provide commanders with continuous
intelligence-gathering in hostile territories, without placing
personnel in imminent danger; however the intelligence collected
is valuable only if it is accessible.
The data communications capabilities of UASs are severely
restricted due to the limitations of bandwidth in the
battlefield. Transmission of imagery, in raw form, consumes large
amounts of bandwidth. Increasing transmission bandwidth is not a
feasible solution in battlefield conditions. Reducing the size of
transmissions, imagery in this case, is the only realistic
approach.
This thesis demonstrates the use of wavelet compression on
UAS imagery to better support military combat operations, thereby
reducing the “fog of war” and saving lives. Specifically this
thesis studies ERDAS®’ Enhanced Compression Wavelet (ECW)
technology, which allows compression and decompression of imagery
without placing a large burden on processors and memory
(necessarily limited in UASs) and thereby economizing the use of
data communications networks. Tests using simulated battlefield
vi
equipment show that image compression of 93%, and a concomitant
decrease in bandwidth demands, is possible.
1
INTRODUCTION
Military technology advances steadily, at the same time
reducing the value of previous technologies. Human life, though,
never loses its value. The motivation for enhancing the
capabilities of unmanned aircraft systems (UAS) in combat
operations is simple: to reduce human loss. UASs provide the
military continuous access to intelligence without having to risk
the lives of troops being assigned reconnaissance missions. Also,
UASs provide continuous observation of areas of interest,
allowing both defensive and offensive measures to be taken
effectively and in a timely manner. Overall, the information
gathered through UASs reduces the “fog of war” by improving
situational awareness and preparedness.
Although the capabilities of both airborne imagery and data
communication have improved significantly in the military over
the past decade, the amount of data that can be transmitted
between the UAS’ ground control station (GCS) and external units
remains limited, especially in the battlefield. With tactical
communications equipment, the resources may be intermittent and
often must be distributed amongst multiple units operating within
a region. Entire communication links to and from the GCS cannot
be dedicated to UAS feeds without severely degrading command and
control of all functions supporting military operations. Because
transmission bandwidth cannot be increased, in order for the
benefits of UASs to be fully realized the amount of data they
transmit needs to be reduced. Image data compression is not
2
widely used by UASs presently, but could be, as this thesis
demonstrates.
The overarching goal of this research is to save lives by
improving bandwidth efficiency in the battlefield, providing
commanders with imagery of hostile territories in a timely
manner. UASs can certainly reduce the loss of both American
warfighters and noncombatants, as well as increase the overall
strength of the American military, both offensively and
defensively.
3
CHAPTER ONE: BACKGROUND
UNMANNED AIRCRAFT SYSTEMS
UASs have been operating in the Marine Corps for over 30
years but have only recently supported the complete integration
of the Marine Air Ground Task Force (MAGTF) (Bertagna, 2010). In
2008 the Marine Corps purchased almost 2000 UASs from the Army
and the Air Force in order to fill the urgent need identified in
support of operations in Iraq and Afghanistan (Defense Daily,
2008). UASs have proven successful in intelligence, surveillance,
reconnaissance, cargo movement, defense, and target
identification (Bertagna, 2010). The reduction in personnel and
resources required for a UAS to execute an intelligence mission
as compared to a ground force is compelling: the same tasks can
be accomplished with one vehicle and zero personnel entering
hostile territory. UASs also contribute to the readiness of a
fighting unit by reducing demands on personnel and equipment,
allowing the unit to maintain operational strength.
UASs have taken the lead as the preferred system for
surveillance tasks in the military in the past decade. The
systems involve remotely piloted aircraft together with
battlefield communications and control equipment that supply
commanders with real-time images, providing the capability to
“view developing situations in their geographic context, track
and visualize events as they unfold, and predict possible
outcomes” (Luccio, 2009). The MQ-1 Predator, the most popular UAS
available to the military, can reach altitudes of 25,000 feet,
4
and can stay airborne for approximately 40 hours (Luccio, 2009).
Daily missions have nearly tripled since the launch of operations
in Afghanistan (OEF) and Iraq (OIF)
1
with the systems being
deployed in support of combat troops on the ground, as well as
battle damage assessment, coalition operations, disaster relief,
counter-terrorism, and homeland defense, to name a few.
The Predator UAS is equipped with a targeting system, an X-
band
2
synthetic aperture radar, a variable aperture day camera, a
variable aperture infrared camera (for low light/night), as well
as two Hellfire missiles; it is supported by a ground control
station and satellite links (Figure 1). Predators are able to
detect and engage targets, and have been successful in
identifying and destroying targets of interest, as well as
identifying targets to other combat assets through optical
sensors and a laser designator (Luccio, 2009). The system is
controlled from the GCS via a line-of-sight or a satellite data
link, and is capable of full motion video or still frames.
Streaming video, real and near real time, is a valuable asset to
a commander, but the massive amounts of data needs to be
transmitted and analyzed quickly.
The GCS serves as the hub for information collection and
dissemination, downloading the data
1
Operations Enduring Freedom (OEF) is the official name for the war in
Afghanistan beginning in October 2001 and still ongoing. Operation Iraqi
Freedom (OIF) is the official name for the war in Iraq beginning in March 2003
and ending in December 2011.
2
X-band is a segment of the microwave radio region of the electromagnetic
spectrum and is set at 8.0 – 12.0 GHz for radars. The shorter wavelengths of
the X band allow for higher resolution imagery from high-resolution imaging
radars for target identification and discrimination.
5
collected by the UAS via line-of-sight data link or satellite
link and forwarding it to the end-user via satellite link
(McHale, 2010). The GCS is equipped with two satellites which
utilize the commercial Ku-band
3
. The GCS, which is located at the
UAS launch site, is designed to forward information and not store
information for security reasons. This ensures that in the event
the UAS is recovered by the adversary, the entire mission is not
disclosed.
With traditional intelligence operations, narrative reports
at the end of a mission are the primary means of communications.
With UASs, by contrast, communications needs are digital,
ongoing, and voluminous throughout the mission. Currently the
3
Ku Band is a segment of the microwave radio region of the electromagnetic
spectrum and is set at 12.0 – 18.0 GHz. This band is used for broadcasting
satellite services and supports the use of receiver antennas as small as 18
inches.
6
military uses commercial satellites for digital data transmission
purposes, but with steadily increasing demands, from commercial
as well as military sources, these satellites become saturated
quickly. Also, relying on extra-military services in battle
situations is perilous. In short, the data communications
bandwidth available is limited and it needs to be partitioned
among a number of communication links: voice and video-
teleconferencing capabilities, email, Web, etc. It is noteworthy
that commanders of current operations are not necessarily co-
located in combat zones, but rather may be anywhere in the world;
they depend on these communication links to make decisions at the
strategic, operational, and tactical levels.
The military’s reliance on UASs is expected to continue and
even accelerate. Between 2008 and 2010 the number of UASs
deployed in support of operations in the Middle East increased by
330% (Defense Industry Daily, 2010). As UASs continue to advance
their technology, it is essential that communication systems
advance too, so that data acquired by UASs can be disseminated to
military commanders and other decision-makers in a timely manner.
As this thesis demonstrates, commercial software, such as ERDAS’
Imagine™, provides an attractive alternative to building up data
communications architectures to provide additional bandwidth for
the massive UAS datasets.
The widespread deployment of UASs over the past decade has
provided many additional capabilities to the warfighter and has
significantly increased the amount of information available to
7
the battlefield commanders. In 2009 alone the Army generated 24
years worth of video from UASs (Defense Industry Daily, 2010).
The advancements in the quality of video and images collected by
UAS have resulted in massive amounts of data requiring advanced
transmission, storage, and retrieval capabilities. As UAS
missions rapidly grow so does the need for resources that reduce
the burden placed on existing systems. Although significant
effort has been devoted to enhancing UAS as remote-sensing
platforms, there has been a lack of research on improving the
interoperability between UAS and existing communications
platforms. Avoiding the degradation of services resulting from
data communications “overload” is essential to utilizing UASs to
their full potential.
The Department of Defense spent over $1 billion in 2010 on
UAS technology improvements, reflecting a reliance and desire to
continue using these systems to support military operations
(Keller, 2011). Research and development devoted to the
improvement of imagery transmission is essential in reducing the
burden of massive amounts of UAS data on tactical communication.
The U.S. Army Research Laboratory (ARL) is currently
conducting various research projects on so-called Command,
Control, Communication, Computers, Intelligence, Surveillance,
and Reconnaissance (C4ISR) systems to address the information
overload. The ARL objective is to develop theories for C4ISR data
processing, information extraction, and information integration
8
to undertake the rapidly increasing quantities of data
overwhelming commanders (U.S. Army Research Office, 2011).
In one initiative, ARL is researching High Performance
Computing (HPC) on large-scale mobile networks to provide
sufficient speed, fidelity, and security for UAS data
communications (U.S. Army Research Laboratory, 2011). Another
initiative is exploring less computationally burdensome methods
for viewing digital imagery, anticipating development of new
techniques for interactive identification of regions with
visibility discontinuities (U.S. Army Research Office, 2011).
These capabilities were showcased at a C4ISR network
modernization event in 2011 where simulated real-time data to a
tactical network and Command and Control (C2) systems were able
to interact with simulated HPC-generated entities.
An internal U.S. Marine Corps white paper by Neushal (2011)
describes specific communication requirements to support the
increase in UAS deployment and the challenges presented by UASs
in battlefield conditions. Major Neushal, an Amphibious
Communications Officer
4
, highlights a need for standardized
communication and system architectures on open UAS platforms
which are capable of high data rates, data preservation in non-
proprietary formats, and accessible through standard web based
tools. He identifies information exchange requirements (IER) that
4
Amphibious communications is relevant because it is the essence of tactical
communications: movement from ship to shore with limited communications
capabilities that can be installed and operated in a timely manner. The size of
the footprint depends solely on the amount of assets that can be transported
via landing craft.
9
are necessary for the transmission of information, i.e. network
load necessary to handle the throughput of still imagery, geodesy
data, digital terrain elevation data, and meta-data. He also
stresses the importance of taking a holistic approach because in
order for any of these IERs to be useful they must be integrated
with others.
10
IMAGE COMPRESSION
For UASs, an alternative to increasing transmission
bandwidth is reducing the volume of data transmitted, referred to
as “data payload”, through image compression, i.e. creates a
shorter encoding of images by reducing the amount of redundant
and/or irrelevant data in them. Image compression can be
accomplished through either technical or perceptual approaches,
or some combination.
A digital image is a rectangular array of dots, or pixels
(picture elements), arranged in m rows and n columns; thus the
product m × n represents the size in pixels of the image
(Salomon, 2008). A digital representation of color is the usual
attribute of each pixel, which is stored as a fixed-size code,
typically 1 to 3 bytes. In raw form, images are large: for
example, the average Smartphone is equipped with a 5-6 megapixel
camera and can produce in raw images as large as 18 megabytes
(Wall, 2010).
Redundancy arises because neighboring pixels display
spatial auto-correlation (Getis & Ord, 2010). In addition to
color, the brightness of neighboring pixels is correlated, even
if neighboring pixels have different colors, they generally will
be similar in brightness.
Because human vision is sensitive to small variations in
brightness but not small variations in color
5
, compressing
5
The human eye contains 6-7 million cones, which perceive colors, and
120 million rods, which perceive brightness, causing human vision to be
more sensitive to brightness.
11
information in the color components reduces the size of the image
by introducing distortions that are not noticeable to the eye
(Salomon, 2008). For example, converting pixel representations
from three color components, such as RGB (Red, Green, Blue; 8-
bits each), to one brightness component and two admixed color
components, such as YCbCr (luminance, Change in blue, Change in
red; 8-bits plus 2 x 5-bits), referred to as chroma subsampling,
achieves compression (to 18/24 bits = 133% in this example)
(Salomon, 2008). There are numerous methods for image
compression, but all remove redundancy based on the same
principle: given a pixel selected at random in an image it is
likely its neighbors will have the same or similar colors
(Salomon, 2008).
Lossless image compression reduces bits by eliminating
statistical redundancy, exploiting the redundancy in order to
represent the data more concisely without losing information.
Variable-length coding and run-length encoding are examples of
lossless compression.
Variable-length coding is an approach to compression that
encodes source symbols, here color pixels, to a variable number
of bits. Redundancy is reduced by assigning short codes to
common symbols and longer codes to rare symbols, resulting in a
low expected bit length (Salomon, 2008). An example of variable-
length encoding is Huffman coding, developed in 1951 by David A.
Huffman, a MIT information theory student. Huffman coding uses a
specific method of assigning binary codes to symbols and encoding
12
higher weighted (frequency of occurrence) symbols with fewer
bits. First a binary tree of nodes containing the symbols and the
weights is created, then the two nodes with the lowest
probability are combined to form an equivalent symbol that equals
the sum of the two symbols. The combining process is repeated
until only one symbol exists at which point the tree is read
backwards (right to left) and bits are assigned to the branches,
see Figure 2. Variable-length coding is primarily used in other
compression methods, such as JPEG, to encode data units in the
final stages of compression.
Run-length encoding (RLE) is a simple form of data
compression where sequences in which the same data value occurs
consecutively are replaced by a repeat count and a single data
value. RLE reduces the size of a repeating string of characters
(run) and is typically encoded into two bytes, the number of
characters in the run (run count) and the value of the character
in the run (run value) (Murray & VanRyper, 1996). For example, a
character run of 10 “B” characters would originally require 10
bytes but after RLE encoding would require only 2 bytes and be
stored as simply 10B (RLE packet). In this example 10 is the run
count and contains the number of repetitions and B is the run
value and contains the actual value in the run (Murray &
13
VanRyper, 1996). A new packet is created each time the run
character changes. For example, the character string
XXXXXyyyyqqBBBB would convert to 5X4y2q4B, reducing the string
from 15 bytes to 8 bytes. Typically an image is encoded in row-
major order (row by row) starting at the upper left corner and
scanning left to right across each line to the bottom right
corner, easily recognizing intra-column redundancy. Because small
variations in color differ numerically but are not visually
important, RLE is best suited for bi-level (black and white)
images, such as fax machines.
By contrast, lossy image compression reduces bits by
removing marginally important information, accepting loss of
information that is not detectable to the human eye. Orthogonal
image transform and sub-band image transform are examples of
lossy image compression.
In general, an image transform is a mathematical technique
that transforms original pixels into an easily compressible form
in either or both of two ways: (1) curtailing redundancy by
aggregating similar but not idenitcal pixels, effectively
reducing their number and (2) isolating “high-frequency” detail
(see below) in the image to identify less important parts
(Salomon, 2008). Compression occurs when the transformed pixels
are written to the output, at which point they are quantized.
Quantization is the process of converting values to a single
quantum value, such as rounding real numbers to the nearest
integer or reducing large numbers to small numbers by converting
14
them to an average integer. The most efficient method for doing
this is to replace the raw pixels (24 bits) by their indexes (8
bits) within a quantized array of pixels; the array is built-in
to both the encoder and decoder. Beyond simple indexing, image
transform is offered in two forms, orthogonal and sub-band.
Orthogonal transform converts pixels to a ranked set of
coefficients according to frequency, with the first coefficient
being most important (containing much of the data from the
original image) and the remaining coefficients being
progressively less important (containing the less important
details of the original data) (Salomon, 2008). The frequency of
an image is measured by the number of color changes along a row
and/or column. For example, white Christmas lights on an
evergreen tree would be considered high frequency; the green
background would be low frequency. Low frequencies correspond to
the basic image features, whereas high frequencies correspond to
details in the image, which beyond some cutoff, are less
important (Salomon, 2008). By isolating various frequencies,
pixels corresponding to high frequencies can be greatly modified
and pixels corresponding to low frequencies can be modified only
slightly or not at all, resulting in effective compression that
only loses unimportant details. The most popular orthogonal
transform is the discrete cosine transform (DCT) which takes
correlated input data and concentrates only the first few
transform coefficients, i.e. the important low frequency
components (Figure 3). The DCT uses real numbers as coefficients
15
to express the data points in terms of a sum of cosine functions
oscillating at different frequencies. Through use of cosine
functions, blocks of the image can be examined and the colors
averaged to create an image with far fewer total colors. DCT is
the compression method that is used in JPEG-93.
Sub-band transform, also known as wavelet transform,
decomposes an image into two orthogonal frequency bands,
separating the sharpness/contrast bands from the signal/noise
bands (Figure 4). This allows for each band to be independently
quantized. “The wavelet transform is a tool that cuts up data (or
functions or operators) into different frequency components, and
then studies each component with a resolution matched to its
scale” (Daubechies, 1992). Correlated values are converted to
ranked transform coefficients, and compressed by quantizing the
difference and encoding with variable-length codes (Salomon,
2008). Discrete wavelet transform (DWT) replaced discrete cosine
transform (DCT) in the upgraded JPEG-2000 due to its superior
compression performance and quality of images delivered. DCT
suffers from “blocking” artifacts that are introduced when the
image blocks are sub-divided, leading to very noticeable loss in
16
image quality upon reconstruction. DCT also only performs well
for compression ratios of 25:1 or lower, while DWT performs well
for ratios beyond 100:1 (Nanavati & Panigrahi, 2005).
This thesis examines wavelet compression, specifically
ERDAS’ Imagine technologies to compress and disseminate imagery
data thereby increasing bandwidth efficiency and reducing the
time it takes to view images. ERDAS’ Enhanced Compression Wavelet
(ECW) algorithm provides fast compression and decompression rates
without heavily burdening computer memory or processors, at the
same time maintaining high compression ratios and visually
lossless image quality (Intergraph Corporation, 2012a). ERDAS
claims that Imagine has the ability to process images at
>25MB/second, resulting in up to 95% compression depending on
file size.
17
CHAPTER TWO: RESOURCES AND RESEARCH PHASES
To test the capability of ERDAS’ Imagine software in
improving bandwidth efficiency and promoting timely dissemination
of UAS imagery, a simulation facility was set-up within the
Marine Corps Communication-Electronics School located on Marine
Corps Air Ground Combat Center (MCAGCC) at Twenty-nine Palms,
California. Computer devices (nodes) were linked together to
share simulated image data and transfer as a wide-area network
(WAN)
6
comprised of two Virtual Local Area Networks (VLAN)
7
(Figure 5). VLAN
1
included the server and a computer that managed
access to the centralized UAS image data and ERDAS software in
the network, representing the UAS ground-station. VLAN
2
included
the laptop retrieving the UAS image data, simulating a remote
user. A “Layer-3” switch
8
was used to connect the two VLANs to
the WAN, simulating data leaving the network and travelling
across the internet to reach a destination in a different
geographic location.
6
A WAN is a network that covers a large geographic area (links across
metropolitan, regional, or national boundaries).
7
A VLAN is a group of general network devices within a smaller geographic area
(home, office, school).
8
A Layer-3 switch provides routing capabilities and allows the VLAN to connect
to the WAN.
18
SUPERMICRO SUPERSERVER
CISCO CATALYST 2950
CISCO CATALYST 2950
CISCO CATALYST 3750
IBM THINKPAD
VLAN-1
VLAN-2
ISP MODEM
The ERDAS software was installed on a SuperMicro
SuperServer® 1026T-URF serving as the “start node” (representing
the system where the data originates and the transmission process
begins), running Microsoft Windows Server® 2008 R2 and supporting
two Intel 64-bit Xeon processors. The two Cisco Catalyst® 2950
12-port switches and one Cisco Catalyst 3750 switch installed
allowed communication between the different devices on the
network. The Catalyst 3750 is an Open Systems Interconnection
model Layer-3 switch and served as the distribution switch, local
area network, and the VLAN routing device for the entire network.
An IBM ThinkPad served as the “end node” (final destination) and
received information from the SuperServer. Finally Solarwinds®
Network Performance Monitor software was installed on the
Figure 5: Network Diagram.
19
SuperServer and used to monitor and record bandwidth utilization,
packet loss, latency, errors, and CPU load for all the nodes on
the network.
Bandwidth is the amount of data the network can transmit at
a given time and bandwidth utilization measures the percentage of
that bandwidth being consumed. Packet loss occurs when packets
(units of data) fail to reach their destination initially and
must be resent, resulting in degraded or slower performance.
Latency is the amount of time it takes for a packet to traverse
nodes over the network. For Solarwinds, errors are any event that
trigger a warning alert, such as sustained, high levels of
bandwidth utilization overburdening the network. CPU load is the
number of instructions being executed by the system.
The image data used for the research was acquired through
the Marine Unmanned Aerial Vehicle Squadron which provided 219 GB
worth of data to test. All data were viewable using FalconView®,
a Microsoft Windows® based mapping application that is used as a
moving map display within the UAS GCS (FalconView, 2012). The
simulated FalconView map data included three types of files: DTED
(digital terrain elevation data), RPF (raster product format),
and MrSID® (multiresolution seamless image database).
DTED is the military standard for medium-resolution terrain
datasets, which includes a matrix of terrain elevation values
described as the height above the Earth Gravitational Model 1996
(EGM96) geoid, and provides medium resolution, quantitative data.
RPF is a military standard for geospatial databases,
20
composed of rectangular arrays of pixel values that comprise
digital maps, images, and other geographic data for military
applications (National Digital Information Infrastructure and
Preservation Program, 2011). Although RPF supports imagery data
it is restricted to raster images of vector maps in the Marine
Corps. An example of RPF data compression is the National
Geospatial-Intelligence Agency's Compressed ARC Digitised Raster
Graphics (CADRG), which achieves a nominal compression of 55:1.
Because of their widespread use in maps, RPF data are
colloquially referred to as “map data.”
MrSID is a file format developed and patented by LizardTech
for encoding georeferenced raster imagery and optionally
compressing large raster image files, also through wavelet
compression, but restricted to be lossless, which limits its
compression features. Both RPF and MrSID divide image into zoom
files, which allows for quick retrieval without having to
decompress the entire file.
These three data types were chosen due to their commonality
in the military and because they comprised a vast majority of the
data delivered by the UAS through FalconView.
21
CHAPTER THREE: EXECUTION
Researching the performance of ERDAS’ compression and data
management software was conducted in two parts. Part 1 consisted
of first compressing map data then decompressing map data and
gauging visual loss. Part 2 consisted of transmitting both raw
and compressed map data across the network to monitor bandwidth
utilization.
Part 1 consisted of using ERDAS Imagine geospatial “data
authoring” software to compress the raw data representing a mix
of DTED, RPF, and MrSID imagery, converting it to ECW format.
Management of the data authoring was accomplished through the use
of a simple, user-friendly interface. Batch conversion of
multiple DTED, RPF, and MrSID files within a folder are
supported; however, all files in a batch must be of the same
format, resulting in three separate conversion cycles necessary
for the three file formats being tested. On average the
compression of the DTED and RPF files, approximately 1 MB each,
took less than 2 seconds but because of the large number of files
(67,648) being converted the process took over 10 hours. On
average the compression of MrSID files, approximately 133 MB
each, took about 5 minutes, and although significantly fewer
(624) it required an additional 13 hours. In total 68,272 files
were converted to ECW format.
Part 2 involved transmitting both the raw data and the ECW
data across the network, simulating the transfer of large
datasets containing the imagery of a single mission, and
22
determining the extent to which the burden on the network was
reduced through the use of ECW compressed files. Access to the
data from an end node located in a different VLAN was enabled
through Layer-3 switching, representing different geographically
located regions. Monitoring traffic reaching Layer-3 switching
allows for the evaluation of network performance as if the data
was travelling between distant networks.
To test the impact raw data transfer had on the network,
two separate transfers were conducted: the first transfer of raw
data involved the end node “pulling” all the raw map data (49.6
GB) across the network, and the second pulling a subset of the
raw map data similar in size to the compressed data (23.7 GB). To
test the impact of a large ECW data transfer on the network the
entire ECW folder (22 GB) was pulled in one transfer.
23
CHAPTER FOUR: RESULTS
PART 1
Using Imagine, the size of the test map data was reduced
from approximately 135 GB raw data to 22 GB ECW data, as shown in
Table 1.
DTED RPF MRSID
RAW DATA 1638.4 MB 49.9 GB 83.5 GB
ECW DATA 112.0 MB 16.2 GB 5.5 GB
% REDUCED 93% 67% 93%
The MrSID imagery being converted to ECW format
resulted in a 93% reduction in total file size for two reasons.
First, although both MrSID and ECW are wavelet compression
technologies, in the UAS application, MrSID is lossless
9
and ECW
is lossy compression. The MrSID file will lose only marginally
important information upon conversion to ECW, further reducing
the size of the file. Second, the Imagine software has proven to
achieve a compression percentage that increases as file size
decreases. For example, a 3.3 GB MrSID file showed a 16%
reduction to 2.8 GB ECW and a 650 MB MrSID file showed a 45%
reduction to 290 MB ECW (Pursch, 2013). This behavior coupled
with the fact that MrSID images coming to the GCS are only
slightly compressed, explains the large ECW compression
9
For automatic analytics and archival purposes, which are not of
concern to battlefield commanders.
Table 1: File sizes before ECW conversion and after. Note: DTED files are individual 1
degree cells, resulting in a substantially smaller file size than either RPF or MrSID.
24
percentages achieved during this research, where the MrSID files
averaged 5 MB each.
Although reducing the size of the imagery data has obvious
benefits to transmission, storage, and retrieval of data on a
network, they are meaningless if the imagery suffers significant
loss during compression. The assessment of losses was limited to
visual quality because it will be humans alone making battlefield
decisions based on the imagery. The process proved to be visually
lossless for the MrSID imagery files which were indistinguishable
with and without compression (Figure 6), but both the DTED and
RPF map files suffered significant visual degradation as a result
of compression (Figure 7).
The reason for the quality loss seen with RPF files is due
to these files being 8-bit with a color lookup table (CLUT)
(Pursch, 2012); this is a naïve transform technique. The CLUT is
a matrix of color data that is indexed by the pixel values in
order to portray these pixels as colors. Compression causes the
pixel values to change and those changed values no longer
correctly index to the color table, creating the result seen in
Figure 7. The pixel data values for an RPF frame before and after
ECW conversion can be seen in Figure 8 (Pursch, 2012). Notice how
the pixel data changes, for example, if the original pixel value
was 21 and that mapped to Blue in the CLUT, after compression 21
might become 16, which in the CLUT is Green. The result is a very
colorful, incorrect image.
25
Figure 6: Comparison of MrSID Raw Data (left) and ECW data (right).
Figure 7: Comparison of RPF Raw Data (left) and ECW data (right).
26
As mentioned above, although RPF can be used for continuous
imagery its use is limited to raster graphics (raster images of
vector maps) in the Marine Corps, preventing this research from
testing if ECW properly compresses continuous images coded in
RPF.
DTED is a 16-bit data representation, stored in a column
major format (Department of Defense, 2000). The ECW conversion of
DTED produced an image as severely degraded as the RPF file. DTED
can be thought of as a gray scale image where latitude and
longitude identify the pixel, and elevation is the pixel value
(Jacobs & Boss, 1992). Quantization results in a smoother
Figure 8: Pixel data for RPF frame (left) and after ECW conversion (right). (Courtesy
of Andrew Pursch, Intergraph)
27
distribution of pixel values, leading to poorer fidelity of the
compressed images.
Maintaining visual integrity is essential in the military
because lives depend on the decisions made on the basis of
intelligence imagery. Although Imagine does not properly compress
RPF and DTED files, for the reasons explained above, the MrSID
files comprise the bulk of UAS imagery in terms of raw size and
using the program to only convert these images would still
enhance the performance of the network.
28
PART 2
The first transfer of 49.6 GB of raw data caused the
bandwidth utilization to increase from less than 20% to 79%,
resulting in an alert from the system notifying that the network
was operating at a transmit utilization above the 75% threshold.
The bandwidth utilization remained above threshold throughout the
entire transfer of raw data, lasting 98 minutes (Figure 9).
Response time increased on multiple occasions throughout the
transfer reaching levels of 90 ms and 135 ms, identified as in
the 95th percentile, meaning only 5% of applications experienced
worse response time. The total average utilization remained over
85% throughout the transfer of the raw data and also achieved a
95th percentile rating. The network utilization reports collected
during the transfer can be found in Appendix A-F.
Even with a smaller folder of 23.7 GB raw data, the same
trends were identified on the network. The average utilization
increased to over 80% for the duration of the transfer and the
response time increased to 80 ms, triggering the same alerts and
Figure 9: Solarwinds analysis of bandwidth utilization during raw data transfer.
29
95th percentile rating. The transfer only took 51 minutes for a
smaller data transfer but the reduced size did not reduce the
burden on the network.
By contrast, the transfer of ECW-compressed data lasted 78
minutes (vs. 98+51 for the raw data). The bandwidth utilization
reduced to an average of 44% and never going above the 75%
threshold (Figure 10). The demand on the network decreased due to
the reduction in files sizes, allowing for smaller files to
travel the pipe quicker, which prevented “clogging” and reduced
the strain on the network to get the files to their destination.
Response time remained below 20 ms throughout the transfer except
on two occasions where it peaked at 140 ms and 80 ms. The
military network on which the transfer was being conducted
commonly receives patches and updates after working hours,
explaining the presence of these random spikes. The total
utilization never surpassed 54% during the transfer and stayed
below 40% throughout most of the process.
Figure 10: Solarwinds analysis of bandwidth utilization during ECW data transfer.
30
These tests determined that the ERDAS Imagine software
significantly reduced the size of the imagery data, by an average
of 80%, while delivering a visually lossless image for MrSID
imagery data. The burden on the network was also reduced with the
transfer of ECW-compressed data across the network compared to
raw data because of the reduction in the amount of bandwidth
being consumed. For the purposes of this research it was
determined that one test of the data transfer was sufficient. The
total utilization percentages varied between the three tests yet
remained steady within each individual test. It was determined
that this validated a stable network and that further testing
would produce the same results. Although the ECW data transfer
did reach levels of 65% utilization on one occasion, it did not
maintain a utilization percentage outside the 75% threshold for
the entire transfer as did the raw data transfer. Particularly in
a combat environment, a reduction of 35% in total network
utilization is of significant value.
31
CHAPTER FIVE: FUTURE RESEARCH
Further research is required to determine if ERDAS software
is the solution most appropriate and supportable for the military
to enhance network performance and allow for the timely transfer
of imagery data. First, it may be advantageous to use ECW in
place of MrSID in the UAS-to-GCS downlink, reducing imagery file
size and network utilization at the outset.
Second, the issue of compressing RPF and DTED data needs to
be addressed. Although these files are significantly smaller than
the MrSID imagery files, not being ECW-compatible prevents these
files from being handled transparently with ERDAS products used
to compress, store, or retrieve imagery and introduces
unnecessary complexity (along with network burden) whenever the
files are accessed. The ability of software to process all data
relevant to UASs is essential to improve the efficiency of the
network, promote dissemination, and reduce the burden on both
bandwidth and storage.
A third concern relates to the need for data management in
the Marine Corps. Apollo Data Manager (ADM) is a companion
product to Imagine, integral to the ERDAS software suite, that
allows users to comprehensively manage and deliver massive
amounts of file-based and Web-enabled data (ERDAS, 2009). The ADM
is capable of managing terabytes of data through an enterprise-
class system, which “catalogs information through metadata and
provides a user-friendly interface allowing users to easily
modify server configuration parameters” (ERDAS, 2012). The Apollo
32
Web client supplied with ADM delivers data quickly through Web
service interfaces, allowing customers to easily find and deliver
data through custom designed websites that support commonly used
GIS and CAD software packages (ERDAS, 2012). ERDAS itself
provides an “out-of-the-box” Web client (Figure 11).
However, Microsoft SharePoint®, a web portal that
centralizes information and applications, is inoperable with
Apollo Web client. SharePoint is the primary means of sharing
data in the Marine Corps and the ability to embed Apollo within
SharePoint will not only support timely dissemination but will
also prevent data from being retrieved from and stored in
multiple locations. Solving the authentication issue that arises
when Apollo Web client is embedded into SharePoint would allow
the user to retrieve imagery data while maintaining view of the
additional “Web parts” on the page, increasing situational
awareness and promoting speedy decision-making.
In addition to ECW compression tests, a follow-on test was
conducted importing the compressed data into ADM and publishing
Figure 11: Apollo Web Client.
33
it on the web through Web client and Microsoft SharePoint. The
goal was to establish a link between Apollo Web client and
Microsoft SharePoint to support accessibility and retrieval of
all stored map data.
A SharePoint site was established, as well as an additional
collection of sites to including Shared Document Libraries, Web
Parts and a Commander’s comprehensive “dashboard” to simulate the
basic design of SharePoint Portals used by the Marine Corps
(Figure 12). The tests determined that the ERDAS Imagine software
catalog file list was viewable through Microsoft SharePoint 2010;
however, authentication to the ERDAS FireLogin, a utility that
identifies users and passwords, failed while viewing the Apollo
Web Client through SharePoint, with the result that the actual
images were not displayable. The only option was to create a
link to the Apollo Web Client website on the SharePoint site for
use as a pass through to reach the Web client.
Figure 12: SharePoint® site with Apollo Web Client link.
34
CONCLUSION
This research provides a possible solution to the bandwidth
burdens placed on military battlefield networks, specifically due
to the increased reliance on UAS data for mission accomplishment.
The employment of UASs will continue to grow and the amount of
data collected will increase substantially as UAS technology
advances and these systems’ flight time and onboard storage
increase, but the number of communications nodes will not be
significantly expanded. The GCS data communications systems
deployed in support of UAS operations are designed to be tactical
and easily transported, and increasing the number of systems
defeats this purpose. Finding a solution to the bandwidth burden
placed on the network by the sudden abundance of UAS-collected
data allows for the UASs to collect as much intelligence as
possible and for military personnel to access that information in
a timely manner. Commercial software such as ERDAS’ Imagine and
Apollo provide a possible solution to improving the bandwidth
utilization of the limited battlefield networking resources
available to the military. In order for these commercial
solutions to be beneficial though, further research is required
to allow interoperability between other software, such as
SharePoint, already widely used within the military. Still,
finding a solution to bandwidth limitations to support the
benefits of UAS imagery data is critical to future military
operations. Lives depend on it.
35
BIBLIOGRAPHY
Army and Air Force Work UAS Issues. (2007, January). Helicopter
News, p. 1.
Marine Corps Begins Training on Shadow UAS. (2007, July). Defense
Daily.
Aviation History and UAS: On Integrating Unmanned Aircraft
Systems into the National Airspace System. (2009).
Intelligent Systems, Control and Automation:Science and
Engineering, pp. 9-28.
UAS Safety Assessment and Functional Requirements: On Integrating
Unmanned Aircraft Systems into the National Airspace
System. (2009). Intelligent Systems, Control and
Automation: Science and Engineering, pp. 63-108.
Army UAS Airspace Integration. (2010, June). Air Safety Week.
Army Improving UAS Effectiveness. (2011). 11th UAS Warfighter
Forum. Tucson, Arizona: M2 Presswire.
Raster Product Format. (2011, November 17). Retrieved June 16,
2012, from Sustainability of Digital Formats Planning for
Library of Congress Collections:
http://www.digitalpreservattion.gov/formats/fdd/fdd00298.sh
tml
Air Safety Week. (2010, January). UAS:Shifting Terrain. Air
Safety Week.
Alapati, S. (n.d.). Accelerate Replication on Low-Bandwidth
Links: SnapMirror Network Compression. Retrieved December
21, 2011, from NetApp Tech On Tap:
http://www.netapp.com/us/communities/tech-ontap/tot-
snapmirror-compression-0910.html
Barni, M. (2006). Documant and Image Compression. Boca Raton: CRC
Press.
Bergmann, K. (2002). Operational Use of UAS. Asia-Pacific Defence
Reporter.
Bertagna, J. (2010, September). UAS. Marine Corps Gazette, p. 38.
Bleicher, A. (2010, October). The UAV Data Glut. Retrieved
December 8, 2011, from IEEE Spectrum:
36
http://spectrum.ieee.org/robotics/military-robots/the-uav-
data-glut
Daubechies, I. (1992). Ten Lectures on Wavelets. SIAM.
de Jong, E. (2009). End-to-End UAV Messaging over Unreliable Data
Links. Retrieved December 8, 2011, from COTS Journal:
http://www.cotsjournalonline.com/articles/view/100940
Defense Daily. (2008, September). Marine Corps Buys 1,993 UAS to
Fill Urgent Warfighter Need. Defense Daily.
Defense Daily. (2009, November). Army Moves Toward UAS Universal
Ground Control Station. Defense Daily.
Defense Daily. (2009, September). WGS Block II Satellites to
Support UAS. Defense Daily.
Defense Daily. (2010, November). Army Could Help Marines
Weaponize Shadow UAS. Defense Daily.
Defense Daily. (2010, January). Hunter UAS Now Fielded With
Tactical Common Data Link. Defense Daily.
Defense Industry Daily. (2010, May 16). Too Much Information:
Taming the UAV Data Explosion. Retrieved December 8, 2011,
from Defense Industry Daily:
http://www.defenseindustrydaily.com/uav-data-volume-
solutions-06348/
Department of Defense. (2000). Performance Specification Digital
Terrain Elevation Data (DTED). Department of Defense.
Desjarlais Jr., O. F. (2008, September). UAS: Unheard, Unseen,
Unstoppable. Airman, p. 4.
ERDAS. (2008). ERDAS Solutions forDistributed Common Ground
Station (DCGS). Retrieved February 14, 2012, from
INTERGRAPH erdas:
http://erdas.com/Libraries/White_Papers/ERDAS_Solutions_for
_Distributed_Common_Ground_Station_DCGS.sflb.ashx
ERDAS. (2009). URL in the Sky. Retrieved January 31, 2012, from
INTERGRAPH erdas:
http://erdas.com/Libraries/CaseStudies/URL_in_the_Sky.sflb.
ashx
ERDAS. (2009). Who is Using ERDAS APOLLO? Retrieved February 14,
2012, from INTERGRAPH erdas:
37
http://erdas.com/Libraries/Brochures/Who_is_Using_ERDAS_APO
LLO.sflb.ashx
ERDAS. (2011). ERDAS ECW/JP2 SDK. Retrieved February 14, 2012,
from INTERGRAPH erdas:
http://erdas.com/Libraries/Brochures/ERDAS_ECW_JP2_SDK_Broc
hure.sflb.ashx
ERDAS. (2011). Image Delivery Options - ERDAS APOLLO 2012.
Retrieved February 14, 2012, from INTERGRAPH erdas:
http://erdas.com/Libraries/White_Papers/Image_Delivery_Opti
ons_-_ERDAS_APOLLO_2011.sflb.ashx
Erdas. (2012). ERDAS APOLLO Image Web Server 2011 Product
Description. Retrieved February 14 , 2012, from INTERGRAPH
erdas:
http://erdas.com/Libraries/Tech_Docs/ERDAS_APOLLO_Essential
s_-_Image_Web_Server_2011_Product_Description.sflb.ashx
ERDAS. (2012). From the Sensor to the Internet. Retrieved
February 14, 2012, from INTERGRAPG erdas:
http://erdas.com/Libraries/Tech_Docs/From_the_Sensor_to_the
_Internet.sflb.ashx
ERDAS. (2012). Image Exploitation for the Enterprise. Retrieved
February 14, 2012, from INTERGRAPH erdas:
http://erdas.com/Libraries/White_Papers/Image_Exploitation_
for_the_Enterprise.sflb.ashx
ERDAS. (2012). Revolutionize Your Geospatial Workflow with ERDAS
2011. Retrieved February 14, 2012, from INTERGRAPH erdas:
http://erdas.com/Libraries/White_Papers/Revolutionize_Your_
Geospatial_Workflow_with_ERDAS_2011_Software.sflb.ashx
ERDAS. (2012). The New Paradigm: Leveraging the "Information
Cloud" with Dynamic GIS. Retrieved February 14, 2012, from
INTERGRAPH erdas:
http://erdas.com/Libraries/White_Papers/The_New_Paradigm_Le
veraging_the_Information_Cloud_with_the_Dynamic_GIS.sflb.as
hx
ERDAS. (2012). Top 10 Reasons to Use... Retrieved January 31,
2012, from INTERGRAPH erdas:
http://erdas.com/Libraries/Brochures/Top_10_Reasons_to_Use_
ECW.sflb.ashx
FalconView. (2012, April). Frequently Asked Questions. Retrieved
June 16, 2012, from FalconView:
38
http://www.falconview.org/trac/falconview/wiki/frequentlyas
ked questions
Fulghum, D. A. (2011, December 5). Downed UAV Technology Already
Dated. Retrieved December 8, 2011, from Aviation Week:
http://www.aviationweek.com/aw/generic/story_channel.jsp?ch
annel=defense&id=news/awx/2011/12/05/awx_12_05_2011_p0-
401894.xml
Fulghum, D. A. (2011, August). UAS Combat. Aviation Week and
Space Technology, p. 72.
Getis, A., & Ord, J. K. (2010). The Analysis of Spatial
Association by Use of Distance Statistics. Perspectives on
Spatial Data Analysis, 127-145.
Helicopter News. (2011, August). Army UAS Office Aims at
Effectiveness, Efficiency. Helicopter News.
Hu, B. (2010, May 25). High Data Rate Links Based on Turbo-OFDM
for UAV System. Retrieved December 8, 2011, from China
Papers: http://www.china-papers.com/?p=119395
Intergraph Corporation. (2012). Erdas Apollo. Retrieved February
14, 2012, from INTERGRAPH erdas:
http://erdas.com/Libraries/Brochures/ERDAS_APOLLO_Brochure.
sflb.ashx
Intergraph Corporation. (2012a). ECW Format. Retrieved January
31, 2012a, from INTERGRAPH erdas:
http://erdas.com/products/ecw/ecw/details.aspx
Ivancic, W. a. (2011). NASA/TM —2011-216943 Delivery of Unmanned
Aerial Vehicle DATA. Hanover: NASA Center for Aerospace
Information.
Jacobs, E. W., & Boss, R. D. (1992). FRACTAL IMAGE COMPRESSION
USING ITERATED TRANSFORMS:APPLICATIONS TO DTED. IEEE
Communications Society.
Kasner J.H., B. B. (2011). Delivery Methods for LVSD Systems.
Retrieved May 28, 2012, from SPIE Digital Library:
http://scitation.aip.org.libproxy.usc.edu/getpdf/servlet/Ge
tPDFServlet?filetype=pdf&id=PSISDG00805300000180530D000001&
idtype=cvips&doi=10.1117/12.887158&prog=normal
Keller, J. (2011, October 21). U.S. defense spending for UAV data
links and ground-control stations reached billion-dollar
39
mark in 2010. Retrieved December 8, 2011, from sUAS News:
http://www.suasnews.com/2011/10/9376/u-s-defense-spending-
for-uav-data-links-and-ground-control-stations-reached-
billion-dollar-mark-in-2010/
Kenyon, H. (2011, June 24). Army Deploys More Small Drones in ISR
Surge. Defense Systems.
Learned, M. C. (2011, October 4). Unmanned Aerial Systems (UAS)
Integrated Operations in Support of Regional Command
Southwest (RC(SW)). Retrieved October 26, 2011, from MCCLL:
https://www.mccll.usmc.mil
Lina T., C. C. (2009). An enhanced Wavelet-based scheme for near
lossless satellite image compression. Retrieved May 28,
2012, from SPIE Digital Library:
http://scitation.aip.org.libproxy.usc.edu/getpdf/servlet/Ge
tPDFServlet?filetype=pdf&id=PSISDG00744300000174431S000001&
idtype=cvips&doi=10.1117/12.826111&prog=normal
Luccio, M. (2009). Remote Sensing by Remote Control: UAVs Become
Invaluable Assets. Imaging Notes.
Market Research Media. (2011, April). U.S. Military Unmanned
Aerial Vehicles (UAV) Market Forecast 2010-2015. Retrieved
October 26, 2011, from Market Research Media:
http://www.marketresearchmedia.com/2010/04/09/unmanned-
aerial-vehicles-uav-market/
McHale, J. (2010, June 18). Ground control stations for unmanned
aerial vehicles (UAVs) are becoming networking-hub cockpits
on the ground for U.S. unmanned forces. Retrieved March 7,
2013, from Military Aerospace:
http://www.militaryaerospace.com/articles/2010/06/ground-
control-stations.html
Meka, H. M. (2010, June 21). Efficient Simulation Architecture
for Routing and Replication in Mobile Peer to Peer Network
of UAVs. Retrieved December 2011, from IEEE Xplore:
http://ieeexplore.ieee.org/xpl/freeabs_all.jsp?arnumber=548
9675
Military Technology. (2005, June). Enabling Technologies for
Future-Generation UAS. Military Technology, p. 8.
Murray, J. D., & VanRyper, W. (1996). Run-Length Encoding (RLE).
In D. a. Murray, Encyclopedia of Graphics File Formats, 2nd
Edition. O'Reilly & Associates.
40
Nanavati, S. P., & Panigrahi, P. K. (2005). Wavelets:
Applications to Image Compression-I. Resonance, 52-61.
National Digital Information Infrastructure and Preservation
Program. (2011, October 27). Raster Product Format.
Retrieved February 21, 2013, from Sustainability of Digital
Formats Planning for Library of Congress Collections:
http://www.digitalpreservation.gov/formats/fdd/fdd000298.sh
tml
NetApp. (2009). NetApp SnapMirror. Retrieved December 21, 2011,
from NetApp IT Management Services:
www.huonit.com.au/_literature_51512/NetApp_Snap_Mirror
Neushul, J. (2011). Information Exchange Communications
Requirements for Direct Support (DS) and Intelligence,
Surveillance and Reconnaissance (ISR) by Unmanned Aircraft
System (UAS).
Newsweek. (2008, March 31). Up in the Sky, An Unblinking Eye: The
hundreds of drones cruising over Iraq and Afghanistan have
changed war forever. . Retrieved October 26, 2011, from The
Daily Beast:
http://www.thedailybeast.com/newsweek/2008/05/31/up-in-the-
sky-an-unblinking-eye.html
PCMAG. (n.d.). Color Lookup Table. Retrieved August 10, 2012,
from PC MAG:
http://www.pcmag.com/encyclopedia_term/0,1237,t=color+looku
p+table&i=39989,00.asp
Pike, J. (2011, June 15). Unmanned Aerial Vehicles (UAVs).
Retrieved October 26, 2011, from Federation of American
Scientists: http://www.fas.org/irp/program/collect/uav.htm
Polikar, R. (1999). The Story of Wavelets. CSCC'99 (pp. 5481-
5486). IMACS/IEEE.
Pursch, A. (2012, August 7). Intergraph Government Solutions. (M.
Parker, Interviewer)
Pursch, A. (2013, March 22). Intergraph Government Solutions. (M.
Parker, Interviewer)
Riverbed. (2009). NetApp SnapMirror. Retrieved December 21, 2011,
from Riverbed:
http://www.riverbed.com/docs/PerformanceBrief-Riverbed-
SnapMirror.pdf
41
Rosenberg, Z. (2011, June). Budget Squeeze for UAS. Flight
International, p. 20.
Salomon, D. (2008). A Concise Introduction to Data Compression.
London: Springer.
Smithsonian. (n.d.). Military Unmanned Aerial Vehicles (UAV) .
Retrieved October 26, 2011, from National Air and Space
Museum: http://www.nasm.si.edu/exhibitions/gal104/uav.cfm
The New York Times. (2011). U.S. Military Unmanned Aerial
Vehicles (UAVs). Retrieved 10 26, 2011, from About.com:
http://usmilitary.about.com/od/uavs/U_S_Military_Unmanned_A
erial_Vehicles_UAVs_.htm
The UAV. (n.d.). UAV Types. Retrieved October 26, 2011, from The
UAV: http://www.theuav.com/
Trimble, S. (2011, August). Services Combine for UAS Purchase.
Flight International, p. 18.
U.S. Army Research Laboratory. (2011). Army Review 2011. Adelphi:
U.S. Army Research Laboratory.
U.S. Army Research Office. (2011). ARO in Review 2010. Research
Triangle Park: U. S. Army Research Office.
Valdes, R. (n.d.). How the Predator UAV Works. Retrieved March 7,
2013, from How Stuff Works:
http://science.howstuffworks.com/predator6.htm
Wall, M. (2010, February 11). The Megapixel Myth: What Really
Makes a Good Cell Phone Camera? Retrieved February 20,
2013, from Tech News Daily:
http://www.technewsdaily.com/127-the-megapixel-myth-what-
really-makes-a-good-cell-phone-camera.html
Warwick, G. (2011, August). Harmonizing UAS. Aviation Week and
Space Technology, p. 57.
42
APPENDIX A
EVENTS LOG FROM SOLARWINDS
43
APPENDIX B
TOTAL BANDWIDTH UTILIZATION THROUGHOUT DATA TRANSFER FROM
SOLARWINDS
44
APPENDIX C
RAW DATA TRANSFER BANDWIDTH UTILIZATION FROM SOLARWINDS
45
APPENDIX D
RAW DATA TRANSFER RESPONSE TIME FROM SOLARWINDS
46
APPENDIX E
ECW DATA TRANSFER UTILIZATION FROM SOLARWINDS
47
APPENDIX F
ECW DATA TRANSFER RESPONSE TIME FROM SOLARWINDS
Abstract (if available)
Abstract
Between 2008 and 2010, the number of unmanned aircraft systems (UAS) in the military increased by 330% in support of operations throughout the Middle East (Iraq, Afghanistan, Iran, etc.). The Pentagon has developed numerous initiatives to enhance the overall performance of UASs, demonstrating that reliance on and deployment of these systems is expected to continue. Via real-time aerial imagery, UASs provide commanders with continuous intelligence-gathering in hostile territories, without placing personnel in imminent danger
Linked assets
University of Southern California Dissertations and Theses
Conceptually similar
PDF
Spatiotemporal patterns of salt and nutrient contamination in Los Angeles County's groundwater basins
PDF
Integrating spatial visualization to improve public health understanding and communication
PDF
Improving wetland determination utilizing unmanned aerial systems
PDF
Unmanned aerial systems for surveying and mapping: cost comparison of UAS versus traditional methods of data acquisition
PDF
Address points and a maser address file: improving efficiency in the city of Chino
PDF
Modeling patient access to point-of-care diagnostic resources in a healthcare small-world network in rural Isaan, Thailand
PDF
Spatial distribution of the Nile crocodile (Crocodylus niloticus) in the Mariarano River system, Northwestern Madagascar
PDF
Modeling open space acquisition in Boulder, Colorado
PDF
Crowdsourced maritime data: examining the feasibility of using under keel clearance data from AIS to identify hydrographic survey priorities
PDF
Analyzing earthquake casualty risk at census block level: a case study in the Lexington Central Business District, Kentucky
PDF
GeoBAT: crowdsourcing dynamic perception of safety data through the integration of mobile GIS and ecological momentary assessments
PDF
Beyond visual line of sight commercial unmanned aircraft operations: site suitability for landing zone locations
PDF
Generating trail conditions using user contributed data through a web application
PDF
GIS data curation and Web map application for La Brea Tar Pits fossil occurrences in Los Angeles, California
PDF
An application of aerial drones in high definition mapping for autonomous vehicles
PDF
Cartography for visualizing anthropogenic threats: a semiotic approach to communicating threat information in 3-D spatial models
PDF
Effect of spatial patterns on sampling design performance in a vegetation map accuracy assessment
PDF
Exploring urban change using historical maps: the industrialization of Long Island City (LIC), New York
PDF
Utilizing GIS and remote sensing to determine sheep grazing patterns for best practices in land management protocols
PDF
Evaluation of weights of evidence to predict gold occurrences in northern Minnesota's Archean greenstone belts
Asset Metadata
Creator
Parker, Mary Elizabeth
(author)
Core Title
Data overload in unmanned aircraft systems: improving bandwidth utilization through wavelet compression
School
College of Letters, Arts and Sciences
Degree
Master of Science
Degree Program
Geographic Information Science and Technology
Publication Date
04/26/2013
Defense Date
03/20/2013
Publisher
University of Southern California
(original),
University of Southern California. Libraries
(digital)
Tag
bandwidth utilization,ECW,image compression,Military,OAI-PMH Harvest,tactical,UAS,unmanned aircraft,wavelet compression
Language
English
Contributor
Electronically uploaded by the author
(provenance)
Advisor
Hastings, Jordan T. (
committee chair
), Chiang, Yao-Yi (
committee member
), Kemp, Karen K. (
committee member
), Mendoza, Karlo F. (
committee member
)
Creator Email
meparker@usc.edu,missparker1979@gmail.com
Permanent Link (DOI)
https://doi.org/10.25549/usctheses-c3-243817
Unique identifier
UC11288153
Identifier
etd-ParkerMary-1602.pdf (filename),usctheses-c3-243817 (legacy record id)
Legacy Identifier
etd-ParkerMary-1602.pdf
Dmrecord
243817
Document Type
Thesis
Rights
Parker, Mary Elizabeth
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
Tags
bandwidth utilization
ECW
image compression
tactical
UAS
unmanned aircraft
wavelet compression