USING A HELIO BASED PROTOCOL IN A BATTLEFIELD SENSOR NETWORK WITH
DIRECTIONAL ANTENNAS AND ENHANCED SECURITY
Except where reference is made to the work of others, the work described in this dissertation is my
own or was done in collaboration with my advisory committee. This dissertation does not include
proprietary or classi ed information.
Fred L. Strickland
Certi cate of Approval:
Lloyd Stephen Riggs
Professor
Electrical and Computer Engineering
Yu Wang, Chair
Assistant Professor
Computer Science and Software Engineering
Min-Te Sun
Assistant Professor
Computer Science and Software Engineering
Chwan-Hwa "John" Wu
Professor
Electrical and Computer Engineering
George T. Flowers
Interim Dean
Graduate School
USING A HELIO BASED PROTOCOL IN A BATTLEFIELD SENSOR NETWORK WITH
DIRECTIONAL ANTENNAS AND ENHANCED SECURITY
Fred L. Strickland
A Dissertation
Submitted to
the Graduate Faculty of
Auburn University
in Partial Ful llment of the
Requirements for the
Degree of
Doctor of Philosophy
Auburn, Alabama
August 9, 2008
USING A HELIO BASED PROTOCOL IN A BATTLEFIELD SENSOR NETWORK WITH
DIRECTIONAL ANTENNAS AND ENHANCED SECURITY
Fred L. Strickland
Permission is granted to Auburn University to make copies of this dissertation at its
discretion, upon the request of individuals or institutions and at
their expense. The author reserves all publication rights.
Signature of Author
Date of Graduation
iii
VITA
Fred Lon Strickland, son of Alonzo Abe and Myrtle Lorene (Fields) Strickland, was born on April
23, 1952 in Gilroy, California. He graduated with a Bachelor of Arts degree in Religion from Stetson
University in 1974. Afterwards, he joined the United States Air Force. During his 21-year military
career, Fred completed a Masters of Arts Management and Supervision: Administration (MBA)
degree from Central Michigan University in 1977, an Associates of Science degree in Communications
Technology from the Community College of the Air Force in 1984, and a Bachelor of Science degree in
Computer Science from University of Maryland in 1987. While working on these degrees, he worked
in various Air Force communications positions. After retiring from the Air Force, he completed
the Masters of Computer Information Science from Troy State University Montgomery in 2001 and
was inducted into Gamma Beta Phi Honor Society. Since receiving the MBA, Fred has taught
at numerous colleges and universities. He entered the doctoral program in Computer Science and
Software Engineering at Auburn University in 2002. After completing the required course work, Fred
was inducted into the now defunct Alpha Theta Chi [Regional] Honor Society, into the Delta Epsilon
Iota Academic Honor Society, into Golden Key International Honour Society, and into Upsilon Pi
Epsilon (International Honor Society for the Computing and Informaton Disciplines). Fred was
nominated and accepted to the 2007-2008 Preparing Future Faculty (PFF) program. Fred married
Sharyn Lynn Fischer, born to John Robert and Lois Jeanne (Wesley) Metzen and later became
the daughter of Donald and Lois Jeanne (Wesley) Fischer, on December 23, 1974 and has a son,
James Allen Strickland, born on December 31, 1984. Fred?s research interests lie in bridging the gap
between wireless computing and communications technology.
iv
DISSERTATION ABSTRACT
USING A HELIO BASED PROTOCOL IN A BATTLEFIELD SENSOR NETWORK WITH
DIRECTIONAL ANTENNAS AND ENHANCED SECURITY
Fred L. Strickland
Doctor of Philosophy, August 9, 2008
(M.S., Troy University Montgomery, 2001)
(B.S., Univeristy of Maryland{University College, 1987)
(A.S., Community College of the Air Force, 1984)
(M.A., Central Michigan University, 1977)
(B.A., Stetson University, 1974)
(A.A., Florida Junior College at Jacksonville, 1972)
111 Typed Pages
Directed by Yu Wang
The problem is that current battle eld sensors are deployed close to each other, are draining
batteries at a fast pace, are using protocols that result in late and coarse grain information, and
put the human operator at risk. In labs and in graduate research, the problem solving approach
is to address a small segment of an area. The other issues are assumed to be solved or are left to
someone else to address. As a result, when someone attempts to pull together the various solutions,
mismatches and con icts arise.
This dissertation project used an integrated approach of addressing at the same time some of
the software, hardware, and security issues for supporting a wireless environment. For example,
the public safety world needs the ability to locate a person or an asset with a minimal amount of
infrastructure and overhead. Current technology does a poor job of pinpointing the exact location
and tends to provide information in two-dimensions. (Closely related to this is the issue of locating
users that are using access points.) Another example is a sensor network whereby most solutions
result in energy-demanding approaches.
The narrow focus of this dissertation project has been the battle eld sensor network. In additional
to the normal challenges of establishing and maintaining the network, there are the requirements
to operate undetected. The current approach is to use minimal transmitter power, which requires
v
the close spacing of the sensors. But this approach has some drawbacks. For example, when there
are many sensors in an area, the likelihood of a chance discovery increases. Closely related to
this drawback are two other problems: Greater network congestion and greater likelihood of signal
interception.
Our novel concept could solve many problems and have enhanced security. Our proposed ap-
proach makes the following assumptions:
1. The base station or the net control station (NCS) has the following capabilities:
(a) Able to communicate with many nodes at the same time on several channels.
(b) Very robust with high speed processing for handling large volume of information.
(c) Unlimited power. (No batteries used.)
(d) Directional antenna with active sectors.
2. The nodes have the following capabilities:
(a) Function as a sensor.
(b) Transmit straight to the NCS.
(c) Able to change frequencies and channels.
(d) Receive control messages.
Simulation and real world experience show that these and other problems could be solved in a
simple energy-saving way.
\All truth passes through three stages. First, it is ridiculed, second it is violently opposed, and
third, it is accepted as self-evident." { German philosopher Arthur Schopenhauer (1788-1860)[1]
vi
ACKNOWLEDGMENTS
I would like to express my deep gratitude to Dr. Alvin S. Lim, who helped put me on this
journey. I wish to express my very deep gratitude to my advisor, Dr. Yu Wang for her encour-
agement, insight, and support. She has helped to keep me on track. Thanks are also due to all
members of my advisory committee{I received hardware insights from the Department of Electrical
and Computer Engineering representatives and I received software insights from the Department of
Computer Science and Software Engineering representatives. Special thanks are due to members of
the Department of Computer Science and Software Engineering for all the useful advice, encourage-
ment, prayers, and support I have received throughout my career as an Auburn graduate student.
Finally, I wish to express my deepest thanks to my wife, Sharyn F. Strickland, for her help and her
encouragement and to my son, James A. Strickland for being proud of his father.
vii
Style manual or journal used The plainnat option of the NatBib package that is part of the
LATEX2e package. Citation usage and bibliography follows the IEEE style with some adjustments
based on the Graduate School guidance and on van Leunen?s A Handbook for Scholars.
Computer software used The document preparation package LATEX2e using the book document
class with selected options.
viii
TABLE OF CONTENTS
LIST OF TABLES . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xii
LIST OF FIGURES . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xiii
1 INTRODUCTION . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1
1.1 Location Examples . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1
1.2 Collection Examples . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2
1.3 Battle eld Sensor Networks . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3
1.4 White Hats vs. Black Hats . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4
1.4.1 Small Parts . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4
1.4.2 Focus . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5
1.5 The Helio Family . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5
2 LITERATURE REVIEW . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6
2.1 Hardware Viewpoint . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6
2.1.1 E orts to Improve Signal Strength . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7
2.1.2 Smart Antenna Systems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8
2.1.3 Obtaining Information from Antennas . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9
2.2 Software Viewpoint . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17
2.2.1 E orts in Using Directional Antennas without any Protocols . . . . . . . . . . . . . . . 17
2.2.2 E orts in Using Directional Antennas in Support of Protocol Based Communications . . 18
2.2.3 E orts in Using Location Awareness in Support of Protocol Based Communications . . 21
2.2.4 E orts in Developing Protocols to Use Directional Antennas in Ad Hoc Environments . 22
2.3 Sensor Protocols . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 29
2.4 Security Viewpoint . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 30
2.4.1 Wireless Security . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 31
2.4.2 Some Hardware Solutions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 32
3 THE HEAPINGS CONCEPT . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 33
3.1 HEAPINGS High Points . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 34
3.2 HEAPINGS: A Description . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 35
3.3 Flooding and Service Messages . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 38
3.4 No Flooding Needed . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 38
3.5 Large Message Volume . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 39
3.6 Frequency Selection . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 40
ix
3.7 How HEAPINGS Functions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 42
3.7.1 Establishing the Network . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 42
3.7.2 Sensors Working{An Example . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 42
3.8 Comparing Other Protocols . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 43
3.8.1 Flooding and Service Messages . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 43
3.8.2 Handling Large Volume of Messages . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 45
3.9 Answering Critics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 45
3.9.1 This is Not a Frequency Hopper Scheme . . . . . . . . . . . . . . . . . . . . . . . . . . . 45
3.9.2 High Power and Power Control is Not Needed . . . . . . . . . . . . . . . . . . . . . . . . 46
3.9.3 Bit Rate Issue . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 46
3.9.4 Idle Listening vs. Sleeping . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 46
3.9.5 Why Single Hop is Better than Multihop . . . . . . . . . . . . . . . . . . . . . . . . . . 47
3.9.6 Software De ned Radios . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 47
3.9.7 Border and Pipeline . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 47
3.9.8 Old Concept . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 47
3.9.9 Antenna Design . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 48
3.10 Mathematical Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 49
3.10.1 Roll Call . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 49
3.10.2 Aggregation Or Not . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 51
4 SIMULATION WORK . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 53
4.1 Ns2 Simulation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 53
4.2 OPNET Simulation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 55
4.2.1 Details on the Created Wireless Simulation . . . . . . . . . . . . . . . . . . . . . . . . . 56
4.2.2 Running the Wireless Simulation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 58
4.3 Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 61
5 EXPERIMENTAL WORK . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 62
5.1 Introduction to the Draft Journal Submission . . . . . . . . . . . . . . . . . . . . . . . . . 62
5.2 Working Background . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 63
5.3 What Should Be Understood? . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 64
5.3.1 Receiver sensitivity . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 65
5.3.2 Antenna Gain . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 65
5.3.3 Antenna Height . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 65
5.3.4 Transmitter Power . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 66
5.3.5 The Result . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 66
5.3.6 Friis Equation or Formula . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 67
5.3.7 Hata Equation or Formula . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 68
5.4 The Approach . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 69
5.4.1 Auburn University As An Example . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 70
5.5 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 71
x
6 CONCLUSIONS: ANTICIPATED BENEFITS . . . . . . . . . . . . . . . . . . . . . . . . . 72
6.1 General Comments . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 72
6.2 Security Comments . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 72
6.2.1 Physcial Security . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 72
6.2.2 Mission Security . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 73
6.2.3 Communications Security . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 73
6.3 The Future . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 74
BIBLIOGRAPHY . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 75
APPENDICES . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 83
APPENDIX A: A LISTING OF WRITTEN PAPERS . . . . . . . . . . . . . . . . . . . . . . . 84
APPENDIX B: COMMERCIAL AND INFORMAL INFORMATION . . . . . . . . . . . . . . 85
.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 85
.2 What has been done without radios? . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 85
.2.1 Fire ghters without radios . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 85
.2.2 Reconnaissance without radios . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 86
.3 What has been done with radios? . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 86
.3.1 Fire ghters with radios . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 86
.3.2 Reconnaissance with radios . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 87
.4 What has been done with the Global Positioning System? . . . . . . . . . . . . . . . . . . 87
.4.1 Fire ghters with Global Positioning Systems . . . . . . . . . . . . . . . . . . . . . . . . 89
.4.2 Reconnaissance with Global Positioning Systems . . . . . . . . . . . . . . . . . . . . . . 89
.5 What has been done with the Radio Frequency Identi cation? . . . . . . . . . . . . . . . . 89
.5.1 Fire ghters with the Radio Frequency Identi cation . . . . . . . . . . . . . . . . . . . . 90
.5.2 Reconnaissance with the Radio Frequency Identi cation . . . . . . . . . . . . . . . . . . 90
.6 What has been done with Local Positioning Systems? . . . . . . . . . . . . . . . . . . . . 90
.6.1 Fire ghters with Local Positioning Systems . . . . . . . . . . . . . . . . . . . . . . . . . 91
.6.2 Reconnaissance with Local Positioning Systems . . . . . . . . . . . . . . . . . . . . . . . 92
.7 What has been done with Locator Transmitters and Beacons? . . . . . . . . . . . . . . . . 92
.7.1 Fire ghters with Locator Transmitters and Beacons . . . . . . . . . . . . . . . . . . . . 92
.7.2 Reconnaissance with Locator Transmitters and Beacons . . . . . . . . . . . . . . . . . . 93
.8 What has been done with Radar? . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 93
.8.1 Fire ghters with Radar . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 94
.8.2 Reconnaissance with Radar . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 95
.9 What has been done with Enhanced 911 Service? . . . . . . . . . . . . . . . . . . . . . . . 95
.9.1 Fire ghters with Enhanced 911 Service . . . . . . . . . . . . . . . . . . . . . . . . . . . . 95
.9.2 Reconnaissance with Enhanced 911 Service . . . . . . . . . . . . . . . . . . . . . . . . . 95
.10 What has been done to support the uniformed services? . . . . . . . . . . . . . . . . . . . 96
.10.1 Fire ghters with support from uniformed services systems . . . . . . . . . . . . . . . . . 96
.10.2 Reconnaissance with support from uniformed services systems . . . . . . . . . . . . . . . 96
.11 What has been done with the Sensors? . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 97
xi
LIST OF TABLES
3.1 Example NCS Table Entries . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 36
3.2 Selected United States ISM Frequencies . . . . . . . . . . . . . . . . . . . . . . . . . . . . 37
3.3 US ISM Frequencies (-86 dBm sensitivity and 20 dBm output) . . . . . . . . . . . . . . . 41
3.4 Comparing Regular and HEAPINGS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 50
3.5 Comparing Aggregation and HEAPINGS . . . . . . . . . . . . . . . . . . . . . . . . . . . 51
4.1 The Four Scenarios . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 58
5.1 De nitions of Terms in the Distance Equation . . . . . . . . . . . . . . . . . . . . . . . . . 64
5.2 Working Receive Noise Figures . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 66
5.3 De nitions of Terms in Friis Equation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 67
5.4 De nitions of Terms in Hata Equations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 68
xii
LIST OF FIGURES
2.1 Drawing of Troposcatter Network . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7
2.2 Smart antenna equippped channel element [2] . . . . . . . . . . . . . . . . . . . . . . . . . 9
2.3 Tracking of mobiles wit multiple beams[3] . . . . . . . . . . . . . . . . . . . . . . . . . . . 10
2.4 Example for an indoor database [4] . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11
2.5 Tree structure of the visibility relations [4] . . . . . . . . . . . . . . . . . . . . . . . . . . . 12
2.6 Discovering and mapping the best routes through a building . . . . . . . . . . . . . . . . . 13
2.7 the HIPERLAN/2 network [5] . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14
2.8 Ground radar tracking an aircraft . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15
2.9 Model for FDTD calculations (left) and a sketch of measurement setup in the reverberation
chamber [6] . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17
3.1 Communications Rings . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 37
4.1 NS2 Prior to Executing Simulation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 53
4.2 Screen Shot of NS2 at the Maximum Load . . . . . . . . . . . . . . . . . . . . . . . . . . . 54
4.3 Jammer Trajectory Behind the Nodes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 57
4.4 Jammer Trajectory Between the Nodes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 57
4.5 Jammer Trajectory Between the Nodes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 59
4.6 Jammer Trajectory Between the Nodes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 60
4.7 Jammer Trajectory Between the Nodes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 60
4.8 Jammer Trajectory Between the Nodes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 60
5.1 Sample Form . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 69
5.2 Modi ed Sample Form . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 70
5.3 Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 71
1 Scott Health and Safety?s Pak-Alert SE . . . . . . . . . . . . . . . . . . . . . . . . . . . . 85
2 MSA?s ICM 2000 and ICM 2000 Plus . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 86
3 CAT?s Aircraft Wireless Intercom System Repeater . . . . . . . . . . . . . . . . . . . . . . 86
4 Sierra Wireless MP 880W GPS and MP 881W GPS Black Box . . . . . . . . . . . . . . . 88
5 RFID Tag (Texas Instruments 13.56 MHz encapsulated Transponder) . . . . . . . . . . . 90
6 A Local Positioning System . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 91
7 Artex ELT 200 operates on 121.5 and 243 MHz . . . . . . . . . . . . . . . . . . . . . . . . 93
8 COSPAS-SARSAT System Overview . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 94
9 View of numerous aircraft on a radar scope. . . . . . . . . . . . . . . . . . . . . . . . . . . 95
10 US Coast Guard Rescue 21 Advanced Communications System (Logo) . . . . . . . . . . . 96
11 \Smart Dust" Chip . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 98
xiii
Chapter 1
INTRODUCTION
When the purpose or function of any wireless network is reduced to the key points, then we are
looking at (1) locating resources and (2) collecting information. The wireless network attempts to
locate people, resources, other computers, or access points. And the wireless network attempts to
collect data or sensor information and route it to the requestor.
1.1 Location Examples
For some concrete examples of the  rst key point, consider public safety individuals, airborne search
and rescue, and locating users on an access point. Each of these are attempting to do resource
location with minimal overhead or no infrastructure.
Individuals working in the public safety world have enough challenges without having to deal with
the problem of imprecise location information. A glaring and tragic example from recent history is
\9-11." On 11 September 2001, the buried  re ghters? air tanks made \man-down" chirping sounds.
However, the rescuers could not determine exactly where the person was located and how deep in
the rumble. The haunting sounds continued until the batteries expired.
The United States Department of Homeland Security (DHS) used its SAFECOM Program to
explore this problem area and other related issues. In March 2004, DHS published Statement
of Requirements for Public Safety Wireless Communications and Interoperability [7]. The DHS
envisioned that future  re ghters would wear a vest that would measure each  re ghter?s \pulse
rate, breathing rate, body temperature, outside temperature, and three-axis gyro and accelerometer
data" plus measure the available air supply in the personal oxygen tank [7]. In addition, the vests
would transmit geo-location, sensor data, and the  re ghter identi cation (ID). Today such a vest
does not exist and there are no reliable means to precisely locate a  re ghter or other public safety
person inside a structure.
1
The foregoing is from the viewpoint of supporting public safety workers, but miners have lost
their lives from coal mining explosion and from cave-ins. It is tragic when the miners survive a
cave-in, but die when the rescuers have no idea where to look to  nd them. A recent tragic example
in this century was the Crandall Canyon mine near Huntington, Utah. In August 2007, 6 miners
died from most likely from a lack of air and 3 rescue workers died from attempting to dig a rescue
tunnel. (The rescue approach is to drill tunnels until one of these breaks through to the correct
spot.) After this cave-in, the rescue tunnel drilling was called o . If the 6 trapped miners were still
alive, then in time they would die from the lack of food, water, and air. Wayne Rash, the Senior
Analyst for eWEEK Labs wrote an article asking whether RFID could have helped to locate the
miners[8].
The Air Force Auxiliary/Civil Air Patrol (AFAUX/CAP) \ ies 95 percent of all federal inland
[search and rescue] missions" [9]. Flying a gird search pattern and keeping track of the airborne
resources are major challenges. This could result in two worst case situations. The  rst is that an
area is notcovered and the victim is missed. The second is that the aircraft could  y into each other
and drop out of the sky.
Locating a person who is moving from access point to access point is another challenge. How
does the environment perform tracking when the only device the person is carrying is a portable
computer?
1.2 Collection Examples
For an example of the second key point, consider civil (non-military) wireless sensor networks and
battle eld sensor networks. Both collect information where it is not possible to run landlines. Or
the location is not safe for a person to be. Another usage might be when the object is moving as in
the case of wildlife or of a military target.
These tend to be in industrial, scienti c and medical (ISM) radio bands with a number operating
in the 902-928 MHz ISM sub band. To avoid using high power and shortening the battery life, the
sensors tend to be place very close to each other. But this approach has several drawbacks.
2
First, more nodes are needed in order to relay messages from other nodes. Second, overlapping
data collection coverage may result. Compounding the last drawback is that some routing protocols
do not scale up very well with a large number of nodes. One protocol attempts to solve this drawback
by having certain nodes preprocess or aggregate the data and these make contact with other nodes
for relaying messages. The result of this approach is that these special nodes must be carefully
deployed to be in the correct spot or every node is given this ability and a scheme must be provided
that will determine which nodes will be \special" and which nodes will be \regular." The solution
thus becomes more complex.
1.3 Battle eld Sensor Networks
Battle eld sensor networks have additional challenges. Whereas civil sensors could be placed with
great precision and special units could be added to the sensor  eld, battle eld sensors might need
to be air dropped and this would make placement very general. Hence, every unit must have the
same ability. Civil sensors may not be bothered by wildlife or by a person exploring the wilderness.
With battle eld sensors, the enemy is looking for anything that is not natural to the area. If close
spacing is used, then the chances are increased that a battle eld sensor might be noticed. Will the
enemy hear the transmissions? Depending upon the location, civil users might be able to swap out
batteries. For battle eld sensors, this might not be an option. Some civil sensors could be con gured
to sample the general environment and route this information to the base station1. For a rapidly
changing combat environment, the information needs to be current and sent immediately to the
home base. Will the battle eld sensors be able to transmit the information to home base in a timely
manner? Civil sensors do not have a threat from a \foreign power" trying to take over the network.
For battle eld sensors, this is a very real threat. Civil sensors relay for each other. For battle eld
sensors, the extra transmissions could make it easier for the enemy to discover the network and to
map it.
There is another drawback that is frequently lost in the discussions on battle eld sensor networks.
The information must be passed to the command center. How is this done? Extrapolating the
1It is customary in formal writing to use a shorter form for an expression that appears numerous times. Unfortu-
nately, the shortened form of \base station" is \BS," which is also an American English expletive (\bad language").
So I will use the complete spelling in order to avoid invoking this meaning in the reader?s mind.
3
discussion, it would seem that the commander center must be a sensor node hop away or there
must be a \bread crumb" line of something (sensors or generic nodes) providing an \Information
Highway" to the commander center. The  rst approach sets up the command center to be attacked
by a missile or to be overrun by a swarm of foot soldiers. The second approach provides a series of
single point failures or an easy way for the enemy to  nd the command center. Having duplicate
\bread crumb" routes does not solve the problem, because these provide more means for locating
the command center.
1.4 White Hats vs. Black Hats
The rapid and \break-neck" pace in the development of wireless technology has created new capa-
bilities and opportunities. On one side, the \white hat" computer community tries to exploit these
for the bene t of mankind. But on the other side, the \black hats" try to  nd new ways to make
life di cult for everyone by such acts as denial of service attacks, blocking messages, and changing
messages.
Since there is no \gatekeeper" or inspector to make sure the players are legitimate and that
everyone plays by the rules, it would seem to be a wiser course of action to incorporate safety and
security features during the development stage. But safety and security are rarely considered during
the \drawing board" stage. As a result, many  xes and patches are needed to address weaknesses
in a deployed system.
1.4.1 Small Parts
In labs and in graduate research, the normal problem solving approach is to address a small segment
of an area. The other issues are assumed to be solved or are left to someone else to address. As
a result, when someone attempts to pull together the various solutions, mismatches and con icts
arise.
4
1.4.2 Focus
With the limitations of time and of resources, the dissertation problem area was narrowed to address
the unique challenges of battle eld sensors. Rather than attempt to build a harden sensor, the focus
was on devloping a workable protocol.
Previous e orts took a \table top" study approach to identify the needed features for such a
protocol. The proposed protocol has three parts:
 Directional antennas
 Software to determine range and direction
 Security
1.5 The Helio Family
The approach is a heliocentric (\helio" is the shorten form) protocol suite. For over the past year
or more, the concepts and issues have been explored. At this writing, there are four helio-based
protocols.
HERBSUDAI2 (HEliocentric Robust Base Station Using Directional Antennas and Inferences)
could be used for tracking people or resources.3
HAP (Heliocentric AP Protocol) and HAPDA (Helicoentric AP Protocol using Directional
Antennas) could be used for routing tra c in a network.
HEAPINGS (HEliocentric Ad hoc Protocol using Inferences and Nominal resources to provide
a Global view with Security) could be used in a combat or battle eld sensor environment.45
2Pronounced as \Herbs You Doll."
3The IEEE 7th BIBE (BIBE07) accepted 65 papers out of 500-plus submissions for inclusion in a printed pro-
ceedings. Noteworthy papers that could not be included were rendered as poster session papers and these were to
be published on the conference website. In checking before, during and after the conference dates of 14 through 17
October 2007, none of the poster session papers were placed on the conference website.
4Until 25 June 2007, this was an unpublished and unpresented concept. On that day, I presented a paper at the
2007 World Congress in Computer Science, Computer Engineering, and Applied Computing (WORLDCOMP ?07)[10].
The proceedings were published in the Fall.
5A second paper was presented at WORLDCOMP ?08[11].
5
Chapter 2
LITERATURE REVIEW
To provide a background of the helip protocols, the work of others will be reviewed.
The literature review is divided along three lines. The  rst part is from the viewpoint of
engineers{in a single word \hardware." The second part is from the viewpoint of computer sciencists{
in a single word \software." The third part covers security and privacy issues.
The interest in hardware (especially radios and antennas) has been present since the dawn of
radio communications, but the computer scientists? interest has only been since the 1980s. And of
course, the interest in security has grown as the result of many bad experiences with viruses and
other problems.
Commercial interests and topics are covered as extra material in Appendix B.
2.1 Hardware Viewpoint
Many papers have been written on radio communications and on antennas. Most of the papers
included in this literature review are from personal communications conferences and symposiums
such as the International Conferences on Personal Wireless Communications (better known as PWC
yy) and the Institute of Electrical and Electronics Engineers (IEEE) International Symposium on
Personal, Indoor and Mobile Radio Communications (better known as PIMRC yyyy). Other papers
are from sources that address antenna applications and theory. There are three general areas:
 E orts to Improve Signal Strength
 Smart Antenna Systems
 Obtaining Information from Antennas
Some papers could be placed into two areas and the decision is based on which topic is covered
in greater depth.
6
Software De ned Radios (SDRs)[12] are the newest frontier in communications hardware work. A
working premise of this PhD e ort is that the nodes will have the ability to generate any wavelength
(a radio frequency). By assuming the usage of SDRs, this enables using a totally di erent approach.
Instead of using software to overcome the limitations of the hardware, the advanced hardware will
enable a simpler protocol to be used.
2.1.1 E orts to Improve Signal Strength
One approach is to use transmit diversity, which involves using several antennas. A good example of
this concept is the defunct northern tier (the unpopulated areas of northern Canada) Distant Early
Warning communications sites (better known as \DEW line" sites) that used large troposphere
scatter (\troposcatter") communications systems. These sites used two large antennas in order to
improve the chances of the distant stations receiving a complete signal. Northern Lights disrupt
HF communications by interfering with the skip or bounce behavior in the upper layers of the
atmosphere. The troposphere level in the lower atmosphere does not seem to be bothered as much.
See Figure 2.1 to view a diagram of a troposcatter network (courtsey of US Army.1).
Figure 2.1: Drawing of Troposcatter Network
1Source is from the Army?s history website. http://www.history.army.mil/books/Vietnam/Comm-El/Photos/
pg08.jpg
7
Vishwakarma and Shanmugan took this concept and applied it to a base station with the intent
to see if this could \improve the forward link transmission performance (throughput, range, capacity
and quality) without increasing the complexity of the mobile station [13]." They only did analysis
with a simulator of two diversity schemes, but the results did look promising. Elsewhere Chheda
explored this concept by analyzing three di erent diversity schemes [14].
Another concept is to use di erent polarization. It is common knowledge that both ends of the
communications link should use the same polarization. Jeppesen et al. used this approach in order
to reduce interference and improve performance [15].
2.1.2 Smart Antenna Systems
Smart antennas are antennas that can develop beams in one or more desired directions. Much work
has been done with these systems. These have resulted in a number of bene ts.
Traditional antennas work well with narrow signals, but do not work well with spread spectrum
techniques. Adaptive arrays (\Smart Antenna Systems") are used to \maintain orthogonal signal
constellation in code division multiple-access signaling (DS-CDMA)" by means of tracking the de-
sired signal and generating null patterns toward unwanted signals [16]. Currently, researchers are
attempting to solve challenges such as when a signal bandwidth goes beyond a threshold, then the
performance falls o [16].
Smart antennas can improve the capacity of a cellular tower while performing load balancing.
Feuerstein et al. discovered through simulations and  eld trials that a \bolted on" smart antenna
resulted in greater than 50 percent improvement;  exible sectorization resulted in improvements up
to 75 percent; and integrated as a subsystem resulted in improvements ranging from 100 percent to
200 percent [2]. Figure 2.2 (courtesy to Feuerstein et al.) shows a block diagram of how the smart
antenna might be integrated into a radio.
Rambabu and Rajagopal explored using a \pilot beam" with a smart antenna system [3]. The
result is that each cellular phone user would be dead center of a directional beam as illustrated
by Figure 2.3 (courtesy to K. Rambabu and R. Rajagopal). The idea of using smart antennas in
a cellular phone network has been explored for some time by many researchers. Rambabu and
Rajagopal?s approach is to use a \pilot beam" to augment sectors that have a \spike" of heavy
8
Figure 2.2: Smart antenna equippped channel element [2]
tra c.
Smart antennas could be used on both ends of the communications link. Czylwik explored us-
ing smart antennas in space division multiple access (SDMA), time division duplex (TDD), and
frequency division duplex (FDD) [17]. He found that frequency reuse is improved, co-channel in-
terference is reduced, and less power is needed by the mobiles. Czylwik?s focus was not on these
bene ts, but rather he spent the rest of his paper on how to reduce the e ects of fast fading.
2.1.3 Obtaining Information from Antennas
2.1.3.1 Direction of Arrival
Engineers are aware that signals reach an antenna from di erent directions. For example, Dandekar
et al. used the direction of arrival (DOA) as one tool for analyzing their smart antenna data [18].
AlMidfa et al. used polarization diversity to improve the estimation of the DOA [19].
DOA could be used with  eld strength to determine the distance to or from a station. This is
called propagation modeling and one could do this for an area. Hoppe et al. found a novel way to
reduce the processing workload by preprocessing some of the data ahead of time [4]. They took a
9
Figure 2.3: Tracking of mobiles wit multiple beams[3]
building (see Figure 2.4 (courtesy to Hoppe et al.).) and create a database based on the planes,
corners, and its material. Then they preprocessed the data so that they could know ahead of
time how a signal would travel through the building and the amount of degradation that would be
experienced.
The preprocessing assumes that every signal that comes into the building at a certain point
would take the same path through the building. Figure 2.5 (courtesy to Hoppe et al.) illustrates
this idea.
Although preprocessing does speed up the raw work, the approach has some draw backs. First,
creating the database for one building takes time. Second, deploying the database of more than one
building would be a problem since laptops do not have unlimited storage. Third, buildings experience
change over time{aging and structure changes such as a room addition or a major renovation.
Fourth, the model does not address things in a room{adding or removing items would change the
10
Figure 2.4: Example for an indoor database [4]
signal path and behavior. Fifth, collecting changes and publishing to the users would take time and
it would be hard to keep current. Sixth, should temporary structures be documented or not? In
short, attempting to provide a  re department or public service agency with this information for a
whole city would be impossible. Keeping the information on a major  xed system (such as a super
computer) means that the user needs to link-up and stay connected while traveling to a service call.
Cellular towers are not yet providing full coverage to every spot in a city. Creating a new network
of cells independent of the public cellular system would be very expensive and may not be possible
due to legal and other reasons.
Instead of using ray tracing, with or without data preprocessing, other researchers are looking
at models that create a smaller database or that make assumptions about the environment. Tingley
and Phlavan used a statistical model of the space-time radio propagation to estimate behaviors
[20]. Their early results indicated that this tracks nicely and the system could be used to determine
11
Figure 2.5: Tree structure of the visibility relations [4]
geo-location.
Curry et al. found that the angles of arrival (AOA) would vary greatly and this could give rise
to the belief that the source could be located anywhere. But when one looks for clusters where the
AOA varies slightly from each other, then these would have the shortest route to the transmitter.
With this insight, Curry et al. attempted to develop algorithms that could quickly determine the
correct route between the transmitter and the receiver. See Figure 2.6 (courtesy to Curry et al.).
2.1.3.2 Wired for Geo-location
Both the IEEE 802.11 standard and the European Telecommunications Standards Institute (ETSI)
HIgh PERformance Radio Local Area Networks (HIPERLAN) standards addressed geo-location re-
quirements in future wireless Local Area Networks (LAN). Xinrong et al. explained that the HIPER-
LAN/2 \geo-location system architectures can be roughly grouped into two categories, mobile-based
and network-based" and that both approaches need \more than three geo-location Base Stations
(GBS) ... [in order] ... to geometrically locate [a mobile terminal] using multiple [Time of Arrival
12
Figure 2.6: Discovering and mapping the best routes through a building
and Time Di erence of Arrival] measurements [5]. (This is a di erent approach from the commercial
examples that are mentioned in Appendix B.)
Although this approach is independent of the Global Positioning System (GPS) and is able
to extract location information from the signal, it falls short for  re ghters, AFAUX/CAP, and
battle eld deployed sensors. The HIPERLAN/2 network operates in 5 GHz band with 30 meters
range typical for an indoor environment and 150 meters possible inside a large room (such as a
gymnasium). The information is extracted from access points in the room. See Figure 2.7 (courtesy
to Xinrong et al.) for how the HIPERLAN/2 network is con gured. In the normal mode (Centralized
Mode), access points manage all transmissions, including mobile to mobile transmissions. If power
is lost to the access points, then geo-location information cannot be obtained. Also the mobile user
must be accepted as a valid participant in the network.
2.1.3.3 Radars
Radio Detection and Ranging (RADAR or radar) is a concept that dates back to at least the 1940s.
Although research is still being done, the concept has matured to the point that commercial and
non-commercial systems have been deployed. The concept is that information can be extracted from
the signal to determine a target?s range and position. A strong narrow, beam is sent out. When it
hits an object, part of the energy is bounced back to the source where a sensitive receiver processes
13
Figure 2.7: the HIPERLAN/2 network [5]
the signal[21].
There are di erent types used. A slow pulse rate radar can reach out to a great distance. A
fast pulse rate radar can track close-in targets. A Doppler radar can track a large number of small
objects, such as rain drops. Fast clocks and equations are used to obtain distance and direction. See
Figure 2.8 (courtsey of Air University, Air Force) for an illustration.
One example of ongoing research is V. A. Kryachko?s 2003 paper concerning a better equation
for determining the direction  nding characteristic of a pencil-beam antenna [22]. Another example
is Bagdasaryan et al. 2003 conference paper concerning a better way to correct for background
interferences when an adaptive antenna array is used to determine the direction of a signal [23]. A
third example is Kukobko et al. 2003 conference paper about calculations for reducing the impact
that an aircraft radome (the cover or shield that protects the radar while in  ight) has on the
direction- nding function [24].
There is a class of radars known as subsurface radiolocation. These are portable and can penetrate
the ground in order to  nd objects. The approach takes four stages in order to generate useful
information. This is still a new area and papers are being published. For example, Grinev et
al. explored a di erent solution to the inverse problem set [25].
14
Figure 2.8: Ground radar tracking an aircraft
Radars do not require a \wired" environment. Radars can function in a stand-alone con guration
and do not need support from GPS. But there are drawbacks:
 The site needs to support a high power transmitter.
 The receivers are very sensitive.
 The antennas are huge (large, high gain parabolic antennas).
 The target must be useful. (Can re ect the signal back.)
 A network would require close placement2.
Hence radars would not be very helpful for supporting resource locating or for supporting bat-
tle eld sensors.
2For example, Doppler weather radars are placed 100 or more miles apart. Great details are gathered about
the tops of storm systems, but not about the lower levels. Dr _Kelvin Droegemeier of the Center for Analysis and
Prediction of Storms, Sarkeys Energy Center, University of Oklahoma, Norman OK 73019, would like to see Doppler
dishes placed about 20 miles apart [26]. In such a Doppler radar rich world, more information could be provided
about weather systems and tornado formation should be easier to track.
15
2.1.3.4 Pulse Signal Propagation
Much work is being done with pulse signal propagation. Pazynin and Sirenko explored pulse signal
propagation in the urban environment by using the  nite-di erence method [27]. Their main interest
is with pulsed emissions and they saw that these lessons could be applied to communications systems
such as a cellular telephone network. They determined that their approach would use a small
database. This is in opposition to Hoppe et al. approach, which uses a huge database.
2.1.3.5 Distances in Free Space (Real World, not in a Lab)
Rosengren et al. o ered a di erent approach for measuring antenna radiation e ciency [6]. The
traditional measurement method involves using free space or special chambers with the result that
the multi-path component is eliminated. Rosengren et al. see this approach as being expensive,
labor-intensive, and time consuming with questionable results. Their premise is that measurements
need to be performed in a test environment that is similar to the real world. Figure 2.9 (courtesy
to Rosengren et al.) shows how the test environment is con gured with the transmit antenna in
one corner and the receiving antenna is in the middle near the far side. Any object added to the
room would cause the transmitted signal to take multiple paths to the receiver. In the case of the
cylinder, it is  lled with material that is similar to the density and material that exists inside the
human skull.
They found that the measurements are \... almost the same in the whole ... chamber ... if the
receiving antenna is [half a wavelength distance] away from the wall." Rosengren et al. generated
results for three distances on two di erent occasions with the result that the measured values ranged
from 89 to 98 percent of the free space values. Although Rosengren et al. main purpose was to show
a cheaper, faster way for determining radiation e ciencies, they did work with distances.
Davis et al. took measurements in several hospital corridors and found that within one meter, the
 eld strength fell o as expected, but beyond that radius, the decline was slower than expected3 [28].
They wrote that the separation requirements may need to be increased between the various medical
devices that use the radio spectrum. Trueman et al. worked with one hospital corridor and used
3Italics are mine.
16
Figure 2.9: Model for FDTD calculations (left) and a sketch of measurement setup in the reverber-
ation chamber [6]
geometrical optics. They noticed that walls of various construction material had di erent re ection
values with plaster and wire construction having the greatest re ection values [29].
2.2 Software Viewpoint
2.2.1 E orts in Using Directional Antennas without any Protocols
For the longest time, only engineers worked with antennas. But that began to change in the 1980s
when the Association for Computing Machinery (ACM) published a paper [30] on multi-beam an-
tennas for satellites. This paper listed two bene ts of directional antennas: \Available power can
be concentrated in areas actually in use at a given time" and nulls could be created as a block
against interference. The rest of the paper addressed dish antennas and the special issues of satellite
communications. There was no e orts to explore these identi ed bene ts.
The ACM Special Interest Group (SIG) on O ce Information Systems existed from 1983 to 1988.
In the  nal year before it became part of the Information Systems SIG, an interesting paper was
published. Pahlava wrote about radio frequency systems and infrared systems for o ces [31]. He
noted that two or more cells could be connected with \a high-power, narrow-beam4, optical signal."
4Italics are mine.
17
The author did not explore this idea. Instead, he explored improving signals with di erent diversity
techniques.
In the late 1980s and early 1990s, cellular networks were the \hot new technology." IEEE Trans-
actions on Vehicular Technology explored using directional antennas whereas ACM looked at power
management schemes. Rosberg was typical of the ACM writers of the period when he acknowledged
that \more e cient spatial reuse techniques employ[ing] directional and smart antennas [would]
enable better frequency sharing" [32]. But no work was done in this area.
In the same issue that the Rosberg?s article appeared, van Rooyen wrote about a di erent ap-
proach in which a model was created to show that directional antennas, Maximum Ratio Com-
bining (MRC) diversity, and a combination of these two resulted in a great improvement over
omni-directional antennas in DS-CDMA tra c [33]. However, there was no attempt to address any
protocols.
In the only 1998 ACM paper that addressed any directional aspects, Gabber and Wool only
looked at using GPS [34]. This was to determine the location of the customer?s rented broadcasting
satellite receiver unit. Although this was not a true directional antenna usage, what is noteworthy
is the attempt to present an algorithm for how this might be done.
2.2.2 E orts in Using Directional Antennas in Support of Protocol Based Communications
In 1989, Pronios wrote about using directional receive antennas as a means to improve the slotted
ALOHA access protocol [35]. This simple change enabled transmitters to have greater successes in
delivering messages and to begin to have a regional approach to tra c handling. Simulation studies
indicated that \the throughput improvement was substantial for heavy tra c conditions, and the
gains were further ampli ed if the number of available receivers ... [were] increased."
In 1992, there was more interest in improving the performance of the slotted ALOHA packet
system. Ward?s and Compton?s  rst paper was about using an adaptive array antenna [36]. The
only change that was made to the protocol was to add a preamble, which gave the adaptive antenna
enough time to notice the signal and generate an antenna pattern toward the sending station. In
their second paper, they expanded the project from a single beam to a multi-beam system [37].
Again, the ALOHA protocol was not changed. The acquisition of signal stage was changed. The
18
ALOHA receive slots were made a bit wider. If the  rst packet is completely received, then a second
packet from a di erent station could be received. If this second packet came in too quick, then
collision would result and both packets would be rejected.
Sugihara et al. explored how an adaptive array could improve the performance of slotted non-
persistent carrier sense multiple access (CSMA). The results suggested that this \... attains higher
throughput and lower values for average number of transmissions and schedulings than the con-
ventional slotted nonperistent CSMA mode" [38]. The authors acknowledged that their approach
creates (actually, worsens) the hidden terminal or station problem, but they believed that this is
something that the network could live with. Also the authors only considered the situation of the
base station using the adaptive array; there was no consideration of having both ends using adaptive
array antennas.
Shad et al. explored using a smart antenna on an indoor base station [39]. Their main interest
was in the use of static space division multiple access (SDMA) and time division multiple access
(TDMA) capacity of an indoor system and to explore performing dynamic slot assignments. The
smart antenna was the means for obtaining the high quality communications. After  ve years of
work, they provided more information about Dynamic Slot Allocation (DSA) [40]. This paper
con rmed that DSA could result in better system capacity. There was no attempt to design a
protocol or to improve a current protocol.
Sass provided information about the \tactical Internet" [41]. The US Army?s Mobile Subscriber
Equipment (MSE) provides circuit switched voice telephone and facsimile, and packet formatted
computer data. The MSE uses Internet Protocol (IP) routers and interfaces with other IP networks.
Directional antennas are top-mounted on 15- to 30-meter tall masts and are manually aimed. The
non-MSE  xed stations in the net may use directional antennas too. The mobile platforms use
omni-directional antennas. Directional antennas are used to improve the signal strength (9 dBi gain
at 225 to 400 MHz, 20 dBi gain at 1350 to 1850 MHz, and 37 dBi gain at 15 GHz). There are some
changes in the protocols, but these were designed to improve the Quality of Service (QoS) and the
speed of delivery in a combat environment. These changes did not factor in the bene t of using
directional antennas.
One of the results of the military?s need for  exible communications is the software de ned radio
19
architecture. Razavilar et al. explored how this architecture with smart antennas could solve some
communications problems [42]. Currently, the base station requires the mobiles to reduce power in
order to combat co-channel interference (CCI) and the far-near problem. By directing a beam5 to
the user or a group of users in the same general area, nulls are \deployed" to block any possible
interference. The base station could support more users. The side-bene t is that this increases
the received signal strength and thus the user may use a less potent receiver antenna and a lower
transmit power setting. Their algorithms only dealt with beam forming.
Gomes et al. looked at Greedy Randomized Adaptive Search Procedure (GRASP) as a means to
solve frequency assignment problems within a cellular system [43]. GRASP looks at the most loaded
antenna and tries to add more users. If that is not possible, then it looks at the antenna with the
next large amount of users. This reduces CCI and adjacent channel interference. It depends upon
distance to solve co-site, co-cell, and far-site interference. The authors did not consider directional
antennas in their work.
Kobayashi and Nakagaw took a di erent tact by using a sectored approach instead of using a
CSMA/CA approach for all directions [44]. In a computer simulation, an eight-sector arrangement
enabled a base station to enjoy a threefold improvement over the omni-directional CSMA/CA and
transmission delays were reduced.
Yi et al. looked at capacity improvements with directional antennas [45]. There was no e ort to
put forth a new protocol or to revise an existing protocol. Instead they looked at antenna patterns.
They pointed out that bene ts can be achieved by narrowing the antenna beam, but a point is
reached when no further narrowing would produce any more bene ts.
Cheng et al. wrote about how an electronically steerable parasitic array radiator antenna could
be used in both omni-directional and directional mode in order to support ad hoc networks [46].
This antenna is able to develop high gain in both modes.
At MobiHOC 2001, Ramanathan presented some answers to nagging questions such as how to
use directional antennas and what are the implications upon ad hoc networking [47]. In a mili-
5This is the main energy or signal coming from the antenna. Directional antennas can be positioned to point
in a selected direction. The other directions receive very little energy. The receiving behavior is similar; only the
directional part receives a strong signal and the other directions would receive very little. The word \null" is some
times used to describe these weak or non-received signals.
20
tary environment, directional antennas mounted on vehicles resulted in better immunity to signal
jamming and interceptions. On the other hand, directional antennas may not be possible with lap-
tops, personal digital assistant (PDAs) and other small devices. Instead of creating a new protocol
or changing a current protocol, the author merely used current protocols such as CSMA/CA (in
two versions: aggressive and conservative), Hello, and link-state routing. He replaced the omni-
directional antennas with directional antennas and ran some network simulations. He concluded
that performance improvements could be realized: More throughput, less power used, and better
neighbor discovery. He did not attempt to solve any problems, but pointed out some areas where
future work could be done. (Korakis et al. attempted to expand on Ramanathan?s work and they no-
ticed that this protocol must limit the range of the directional antenna, otherwise there is a coverage
mismatch [48].)
Takai et al. proposed replacing both the physical layer carrier sensing and the medium access
control sub layer (MAC) virtual carrier sensing with Directional Virtual Carrier Sensing (DVCS) [49].
This works with the current protocols by supporting \both directional transmission and reception
based on the radio reciprocity, which allows directional transmissions of all frames without incurring
unnecessary collisions." The MAC protocols do not need to be modi ed to work with directional
antennas when DVCS is used. In brief, DVCS \...allows the MAC protocol to determine direction-
speci c channel availability." The authors concluded that DVCS could improve MAC protocols by
using directional antennas to work with a subset of nodes while letting the other nodes continue to
function in an omni-directional con guration.
2.2.3 E orts in Using Location Awareness in Support of Protocol Based Communications
Ko and Vaidya noted that wireless networks are burdened with extra message tra c in order to
determine valid routes [50]. Their approach, called Location-Aided Routing (LAR), would use GPS
location information for route discovery. If the sender does not know the location of the desired
station, then it  oods out a route request message. On the other hand, if the sender knows the
general zone or area, then the route request message is  ooded just to that area. If the sender knows
that the desired node is in a certain direction, then the route request message is  ooded to every
node on that \compass heading." Since this protocol does not use directional antennas, every node
21
within range of the transmitting node would hear the route request message. LAR would begin to
have some bene ts when the nodes are some distance apart and if the deployment area is sizeable.
Otherwise, LAR would have the same behavior as a regular ad-hoc network.
Instead of using GPS, Nasipuri and Li proposed a scheme whereby certain sensor nodes would
function as a local \GPS" for the other sensor nodes in the network [51]. A node listens to three or
more beacon stations in order to determine its position in relation to these stations. The proposal is
interesting in that the approach is similar to the space based GPS system. The drawback is that the
beacons seem to be transmitting frequently, which would enhance the enemy?s ability to detected
the sensor network plus drain these beacon nodes? batteries quickly.
2.2.4 E orts in Developing Protocols to Use Directional Antennas in Ad Hoc Environments
2.2.4.1 Simple Tone Sense (STS)
In 1992, Yum and Hung proposed a simple tone sense (STS) protocol for a multi-hop packet radio
networks with multiple directional antennas [52]. This protocol called for the tra c exchanging
stations to transmit tones on the unneeded antennas. The intent is that other stations would hear
and would know to keep silent. This tries to manage the hidden station problem and to reduce
con icts. They envision that the directional antennas are able to provide 360-degree coverage.
There are some problems with their approach. One of the stated assumptions is that the network
consists of all  xed stations; so it would not work with moving users. Another problem is that it
takes energy to transmit tones. So a node that is doing this would deplete its batteries in a short
amount of time. And still another problem is that those nodes that are aligned in a parallel fashion
to the two exchanging nodes would hear the tones on their own directional antennas (perpendicular
to this pair?s signals), but would not attempt to exchange tra c despite the fact that all of these
pairs are using directional antennas and thus would not harm each other?s e orts.
is that this approach prevents exchanges that would not con ict from taking place. Hence, the
over all throughput is reduced. In short, in order to gain a con ict-free environment, the tra c
volume is reduced.
22
2.2.4.2 Modi ed Dynamic slot Assignment (DSA) MAC Protocol
In 1996, Horne er and Plassmann explored using directed (adaptive) antennas in an European
wireless cellular network with a modi ed dynamic slot assignment (DSA) MAC protocol [53]. The
regular version of DSA uses one physical channel with several tra c channels multiplexed to that
channel. Reservation information and ACKs are sent from the base station to the clients via pig-
gyback messages. With the modi ed DSA, the base station could handle several clients if they are
within the coverage area of the directional antenna beam. If this is not the case, then the modi ed
DSA has a provision For handling this situation. It would use several transmissions in order to
service all the clients in the di erent sectors. The modi ed DSA does not use a broadcast approach
and does not piggyback acknowledgments. Instead a special signaling burst is used. The authors
admit that there are some e ciency issues that require more work. What is noteworthy is that this
appears to be the  rst documented attempt to develop a protocol for a mobile environment that
uses directional antennas. This is an evolutionary step toward having a protocol to handle an ad
hoc mobile network.
In 1997 Sakr and Todd wrote about \reverse link throughput and capacity" [54]. They used the
same environment as described by Sugihara [38]. That is, the ALOHA base station uses a beam-
forming antenna and the clients do not. What is noteworthy about this paper is that the authors
devoted some space to a modi ed carrier-sense protocol. This proposed protocol is able to use smart
antennas to reduce CCI and thereby has a more aggressive frequency re-use plan. The protocol
uses time division duplexing (TDD) and the Request to Send/Clear to Send (RTS/CTS) reservation
mechanism. The protocol has the ability to handle busy periods: \...when the channel becomes busy,
this transition de nes the start of an uncertainty interval of duration Tr seconds ...[during which]
any station generating a packet is free to commence transmitting." The base station works through
the received packets to resolve them and creates starting times for the clients to use. Other clients
not involved are carrier-sensed inhibited from transmitting. Then the base station manages the
network. When a number of packets have been received, the base station switches from a directional
antenna to an omni-antenna in order to broadcast an ACK message. This protocol has provisions
for handling NAK packets and for handling the hidden station problem.
23
There is one weak area. When the nodes are just in communications range through the usage
of directional antennas, but are not within omni-directional antenna range, some messages will be
missed. Hence, the omni-directional antenna ACK broadcasts will not be heard. One means to over
come this weakness is to use high power{this could be a problem if the station is on batteries.
2.2.4.3 Directional MAC (D-MAC)
Ko et al. appear to be the  rst ones to look at designing a protocol for using directional antennas
within an ad hoc network [55]. The proposed Directional MAC (D-MAC) comes in two versions.
Both versions use omni-directional antennas for exchanging data and both use a directional antenna
for transmitting ACKs. The other nodes not involved in the current exchange may use a di erent
directional antenna for communications in another direction. This sets the stage for having two or
more exchanges occurring at the same time.
Scheme-1 has the requesting stations using directional antennas for transmitting RTS messages
and the sink stations using omni-directional antennas for sending out CTS messages. This message
is modi ed to provide information about the locations of the two stations, so that other stations can
determine whether it is safe to transmit in another direction. Directional antennas are used for the
actual exchange of data.
Scheme-2 uses omni-directional antennas for transmitting the RTS messages. But as more links
are established, the RTS messages are transmitted on directional antennas. This version has the
bene t of avoiding con icts from stations that are not aware of a far station transmitting.
The authors used the ns-2 simulator and discovered that both versions performed better than a
regular omni-directional antenna network. The biggest problem with the regular MAC approach is
that a single transmission \locks up" a large area. The proposed D-MAC protocol does not have
this weakness. (Choudhury et al. attempted to go further with this work. They noticed that the
authors did not consider the extra gain that directional antennas provide [56]. Korakis et al. saw this
weakness from a di erent viewpoint; that is, this protocol must limit the range of the directional
antenna, otherwise there is a mismatch in coverage [48]. None of these critics o ered the suggestion
of increasing the power level when an omni-directional antenna is being used.)
Nitin Vaidya?s approach[57] (written as DMAC instead of D-MAC) has all nodes listening on
24
omni-directional antennas. When a station wishes to send tra c, it will use a directional antenna
to transmit a RTS message. The receiving station would reply with a CTS on a directional antenna.
This approach has the weakness that stations not within the directional antenna coverage will not
know what is about to transpire.
2.2.4.4 Nasipuri et al. MAC Protocol
Nasipuri et al. perceived that the previous approach has some weaknesses in the matters of location
and tracking [58]. Their MAC protocol uses RTS/CTS messages with directional antennas. Instead
of 90-degree coverage per antenna, the nodes have six or more antennas with conical radiation
pattern. All the nodes are always oriented to the north. Against this backdrop, the MAC protocol
is modi ed to obtain the direction of any node based upon clues provided by the transmission from
the distant node. Since all the nodes are always in motion, direction (not distance) is the only
useful piece of information. The narrow beam pattern reduces the likelihood of con icts. Simulation
studies revealed that the average throughput was two to three times better than the regular or
omni-directional antenna approach.
In another paper, Nasipuri et al. considered creating routes with a directional On-Demand Rout-
ing scheme [59]. Working with the selected MAC protocol,  ooding would be reduced. That is,\with
directional antennas, [the authors] proposed that a new route discovery [message] is propagated from
S in the direction of D with the help of directional transmissions of a query packet by a restricted
set of nodes in the network." Others6 noticed that Nasipuri et al. overlooked the gain or greater
range that directional antennas bring to a network. Another problem area is the need to  nd north.
All the nodes must be working o of the same grid scheme or coordinate system. So is north from
GPS or from certain nodes? What enhancements do the nodes need to have in order to use this
information?
6Choudhury et al. pointed out that the authors failed to consider the extra gain that directional antennas provide
[56]. Korakis et al. saw this weakness from a di erent viewpoint; that is, this protocol must limit the range of the
directional antenna, otherwise there is a coverage mismatch [48].
25
2.2.4.5 Bandyopadhyay et al. MAC Protocol
Bandyopadhyay et al. looked at using electronically steerable passive array radiator antennas as a
means to support a MAC protocol [60]. This antenna would only transmit to cover a 60-degree slice
and the antenna would develop null patterns against unwanted signals. There is no need to  nd
north. The MAC protocol uses omni-directional RTS packets. Instead of noting that a direction is
busy, the details of the session are stored at each neighbor node. The other station would transmit
a CTS packet on an omni-directional antenna. Both messages would contain the duration. The two
stations would switch to directional antennas. Other nodes would use the stored information plus
the knowledge of where the two stations are located to determine if another proposed link could be
established without destroying the current communications. Despite not having a fully  eshed out
routing scheme, the authors had several simulation runs with very encouraging results.
At MobiHOC 2001, Bandyopadhyay et al. presented a short paper that provided more informa-
tion about the adaptive MAC protocol and the directional routing scheme [61]. Each node tracks
information such as the angle and the Signal to Interference and Noise Ratio (SINR) about its
neighbors. Thus a node would have a snapshot of the directions of the various communications and
this would help it to select the best possible direction when communicating with another node. The
link-state routing protocol can capture most of the network health through monitoring and a minor
amount of service message tra c. When the network is new,  ooding is used to pass a RTS packet
to a desired station. The CTS packet comes back in the same fashion. Directional antennas are
used for the actual exchange. As the network matures (more knowledge is gained), more use will be
made of directional antennas. Link-state updates are periodically exchanged between two nodes; this
approach over time will provide new information to all nodes without the need for massive  ooding.
The aging scheme is similar to DSDV. Choudhury et al. thinks this protocol is a bit complex [56].
Even if it was not so complex, updates to the network would take some time to reach every node.
A rapidly changing topography would have problems with updates.
26
2.2.4.6 Directional Antenna Based MAC Protocol with Power Control (DMACP)
The next natural development is to marry directional antennas with power control and create a new
protocol. Nasipuri et al. did that and they named their protocol Directional antenna based MAC
protocol with Power control (DMACP) [62]. They modi ed the regular MAC protocol so that it
would obtain directional information on the involved nodes and that it would determine what is the
least amount of transmitter power for contacting a certain node. The authors proposed that RTS
packets and CTS packets are sent on the full power setting and that during the exchanges the nodes
would determine what power level should be used for the actual data transmissions. They concluded
that DMACP could improve power conservation and tra c throughput.
2.2.4.7 Multi-Hop RTS MAC (MMAC)
Choudhury et al. compared two directional MAC protocols [56]. The basic Directional MAC
(DMAC) is based on elements from Ko et al. [55] and Takai et al. [49]. The second one is called
Multi-Hop RTS MAC (MMAC). The authors viewed DMAC as being weak in addressing the hid-
den terminal problem plus being deaf to other stations. On the other hand, MMAC attempted to
use the greater transmission range with better frequency reuse than DMAC. The DMACs two prob-
lems are present here too, but the authors believe that their approach could compensate for these
weaknesses. The MMAC de nes neighbors either as Direction-Omni (DO) or Direction-Direction
(DD). The multi-hop RTS packet is used to tell two distant nodes to direct their antenna beams
toward each other. The rest of the MMAC protocol uses neighbor discovery, link characterization,
proactive routing, and a position information module. Channel reservation is the key to the success
of MMAC. The authors conclusions are mixed. On one hand, they stated that MMAC has better
results than regular MAC and DMAC. But on the other hand, they stated that \the performance
... clearly depends on the topology and  ow pattern in the network [with] more aligned topologies
degrading the performance...." They admitted that more work is needed.
27
2.2.4.8 Receiver-Oriented Multiple Access (ROMA)
Bao and Luna-Aceves have a di erent approach that uses a distributed receiver-oriented multiple ac-
cess (ROMA) channel access scheduling protocol [63]. This can support multiple beams and multiple
sessions. Instead of using on-demand schemes or signal scanning in order to resolve communication
targets, ROMA \determines a number of links for activation in every time slot using only two-hop
topology information." Using a neighbor-aware contention resolution (NCR) algorithm and direc-
tional antennas, ROMA o ers four major bene ts. First, both transmitting and receiving can be
done with directional antennas. Second, storing local two-hop information requires less memory
than storing global topology. Third, ROMA divides the nodes into equal groups of transmitters and
receivers for each time slot. (ROMA has a tool for pairing stations for the best network through-
put.) Fourth, ROMA can generate antenna use schedules. ROMA is better than other scheduling
approaches, but this has considerable overhead.
2.2.4.9 Roy et al. MAC Protocol
At MobiHoc 2003, Roy et al. presented a MAC protocol that would strive for load balancing via
\maximally zone disjoint routes" [64]. That is, the routes would be selected that would have the
smallest impact on other stations. In order to do this, each node must keep track of the nearby nodes
and what exchanges are occurring. This information is shared with other nodes. Each intermediate
node determines what is the best route from its location. Despite this overhead, the simulations
indicated that this approach is far better than reactive routing with regular MAC protocols. They
do not know if this would work during periods of high mobility or with a large network.
2.2.4.10 Korakis et al. MAC Protocol
Korakis et al. attempted to address all the shortcomings of the current directional MAC protocols
by using only directional antennas [48]. To establish contact, a RTS packet is transmitted on one
\compass" heading; then it moves to the next \compass" heading and repeat the same RTS packet.
This continues until all \compass" headings are covered. Any station not addressed by the message
would update its local table and not transmit in the direction of the RTS requesting station. The
28
addressed station would reply at the end of the last RTS transmission, but its reply would be only
in one direction. At the end of the RTS and CTS exchange, every node within range knows what
is about to take place. The information is stored in a table and this is used to make decisions
about transmitting or not transmitting and in which direction. Using directional antennas instead
of omni-directional antennas should reduce the hidden station problem due to the greater coverage.
2.3 Sensor Protocols
The main di erence between ad hoc networks and sensors networks is in the matter of movement.
Ad hoc networks tend to have movement whereas sensor networks tend not to have any movement.
This is not a hard requirement since an ad hoc network could be stationary. In section 2.2.2 above,
there was a mention of Nasipuri and Li work on a location scheme for sensor networks.[51] Another
trait of sensor networks is the need to stretch out the lifetime of batteries.
For this discussion, we assume that sensor nodes are  xed, must use low power, and must survive
for a long time. Within these limitations, researches have explored various issues.
Some researchers have explored using sleep procedures whereas some nodes are inactive and thus
extending battery life. Some researchers have considered using a schedule for waking up nodes. The
various e orts do achieve the goal of extending the sensor network life time, but the drawback is
that messages may be missed or delayed. Sampling changes in air temperature is one thing, but
detecting a column of tanks is another matter.
Some o ered solutions are interesting, but carry a huge overhead. Simple can be better. There is
no need to add control behaviors to the nodes and thus avoid the issues that Sinopoli et al. explored[65]
in a Defense Advanced Research Projects Agency (DARPA) funded research project.
Simple can still be better. As Chong and Kumar pointed out, a centralized network has poten-
tially the best performance[66] and avoids the di culties of trying to design a workable decentralized
algorithm for a self-organizing sensor network or other \smart nodes" scheme.
Some think that SPIN (Sensor Protocols for Information via Negotiation) in some form could save
energy and make the sensor network better[67]. Still others think that directed di usion protocol is
the way to go[68], [69]. But the nodes have to be aware of the tasking applications. This involves
29
some broadcasting and this means extra messages being exchanged between nodes. The nodes need
to be smart enough to select the best path, have enough memory to store the data, and be able to
do some pre-processing of the received data.
This is a \hot" topic since research indicates that as a data-centric protocol, directed di usion
would use less energy than other standard routing protocols such as  ooding7 and multicast. There
are other protocols or approaches that depend upon the nodes having knowledge of their environ-
ment, their data requirements, and of their neighbors. Such requirements adds additional overhead.
Alvarez et al. o ered a middleware solution in order to have reliable communications, but that in-
volves more work to achieve this[70]. They do acknowledge that directional antennas have some
advantages such as less energy consumption, less fading area, and less channel interference. How-
ever, they never used this information in the balance of their paper. Weslsh and Mainland suggested
abstract regions or interfaces as a means of making it easier to build sensor network applications[71].
2.4 Security Viewpoint
Security has become an important issue. Whether it is a person having \fun" reading private commu-
nications or a determined adversary that wants to do great damage, defenses are needed. Microsoft
spends considerable e ort deploying security patches for their operating systems and for their soft-
ware packages. Anti-virus companies are releasing updates that attempt to combat the latest \bug."
In April 2005, the House Homeland Security Committee?s Economic Security, Infrastructure Protec-
tion and Cybersecurity Subcommittee approved H.R. 285, the DHS Cybersecurity Enhancement Act
of 2005, which \create[d] a new cybersecurity czar [72]." For the past few years, Auburn University
has o ered an Information Assurance Option for graduate computer science students to add to their
transcript.[73] George Washington University has courses in information security management[74].
Tanenbaum divides security problems into four areas: secrecy, authentication, nonrepudiation,
and integrity control [75]. The following paragraphs brie y explain these areas:
 Encrypting a message is an example of ensuring secrecy for a message. Using the Advanced
7This is one node passing a message to all nearby nodes. Each of those nodes would repeat the message to all
its nearby neighbors. The message is not transmitted to the node that gave it the message. In a wired network, the
nodes upstream would not repeat the message to nodes upstream of it. However, downstream nodes in a topography
that has multiple links would receive the same message several times.
30
Encryption Standard (AES) would provide a long period of protection for any message.
 Authentication relates to the problem of determining that the sender is who we think it is.
 Nonrepudiation relates to the problem of a sender claiming that a sent message is false{he/she
did not send out the message.
 Integrity control relates to sent messages{Did a message really come from a certain sender?
Best practices point to using security tools from start to  nish and be inside a closed system.
Mobile users can gain some measure of the ideal by deploying with an encryption key. The challenge
for both  xed users and mobile users is supplying the next encryption key. A currier service on a
regular route can handle the  xed user, but this may not be possible for a mobile user deep behind
enemy lines.
Another issue is that encryption schemes require more central processor unit (CPU) work{This is
another drain on the batteries. If a node or a user is compromised, then things become \interesting."
Closely related to security is the matter of survival. Sensor nodes need to avoid detection. If a
node is destroyed, then the network needs to be robust enough to continue.
Sensors networks need to deal with bogus nodes that are trying to take over the network or are
trying to be members of the network. Is there a way to use these bogus nodes to help out the sensor
network?
2.4.1 Wireless Security
Wired communications can only be exploited by physically taping into a cable. Hence, physical
security results in communications security. This is not the case with wireless communications;
anyone with a receiver can capture and record the transmission. One might expect that encryption
would improve the situation.
Some security approaches have been rendered useless. Wired Equivalent Privacy (WEP) is a
case in point. It has a small key space. A number of keys are used frequently. Given a reasonable
amount of time, the key and the plaintext message can be obtained. With this information, false
messages could be generated. The security task is made more di cult when some devices have a
default setting with security disabled.
31
David B. Johnson looked at the problems and the current solutions[76]. Fixed stations are fairly
easy to handle. Slowly moving stations are possible to support, but the matter of large numbers of
fast moving stations is very challenging to handle. A location registry is one possible approach, but
registration authentication is needed in order to prevent evil nodes from creating problems for the
network.
2.4.2 Some Hardware Solutions
Ultra-Wideband (UWB) systems have the promise of high data rates, robustness to interference,
and low power usage. Racherla et al. wrote that UWB systems have \high transmission security low
prime power requirements" and that the UWB technology is able \to perform precision geo-location
which can aid in ad-hoc or mesh networking where the operations of the mobile hosts bene t by
knowing the location of the other hosts"[77]. Devices are able to function at the background noise
level and are able to exist without causing problems to other systems. Transmissions are very brief
and take place at widely space intervals. The UWB technology supports distributed sensor networks,
ad-hoc networks, and geo-location.
32
Chapter 3
THE HEAPINGS CONCEPT
Of the helio protocols listed in the  rst chapter, HEAPINGS is the most mature. Analysis
points to a number of bene ts:
 Could pass information in near real time
 Could hid the nodes locations
 Could avoid network congestion
 Could extend the life time of all nodes? batteries
 Could avoid close spacing of nodes
 Could avoid the need for a \bread crump" trail to the base station
 Could avoid having bogus nodes impact the network
 Would let the nodes operate with low power
 Would use the base station?s greater power and abilities to obtain the big picture
 Would avoid using message  ooding
 Would avoid using aggregation with its reduced precision and timeliness
 Would provide geo-location information without using GPS
A conference paper appeared in the WORLDCOMP ?07 proceedings[10]. The high points of the
paper follows1.
1In spots, I have expanded the topic and I have removed all citations. WORLDCOMP ?07 had a seven-page
limitation and so some information had to be cut out.
33
3.1 HEAPINGS High Points
We found that nearly every scheme or protocol used the same three approaches:
1. Nodes are independent and robust.
2. Flooding is during part of the network?s life.
3. Use only one frequency.
These three approaches may actually create problems. We argue that there should be centralized
control with nominal nodes, no  ooding, and use the best frequency band. Not all researchers are
using the previously mentioned approaches. For example, the work on tiered architecture breaks
with the  rst approach. (In some ways, our work is an extension of this area. Although our concept
was created prior to discovering papers in this area.)
Tiered networks do not use robust, full-featured nodes. Researchers in this area use di erent
terms and di erent assumptions, but there are some common elements. The micronodes (the sen-
sors), macronodes (the gateways), and preprocessing of data are used. One research group calls the
upper tier the \coordination units" and they assumed these to be \not energy constrained." The
Tenet Architecture2 has the same assumptions, but calls the upper tier nodes \masters." Another
research group suggested having multi-tiers and multi-modal networks.
Again, those nodes at the lowest levels have less ability. These collect information and forward
to the next level. The next level contains nodes that are more robust. Above these nodes might be
connections to the Internet. Just like the other networks,  ooding is used and only one frequency is
used.
This is a good start, but we think more could be done. Our approach does not require the low
level nodes to have knowledge of their neighbors3.
2See http://enl.usc.edu/projects/tenet for more information.
3There are real world systems deployed to monitor wildlife. One research group is exploring how to determine a
location, but so far their e orts requires the low level nodes to develop a list of nearby nodes.
34
3.2 HEAPINGS: A Description
Our novel concept could solve many problems and have enhanced security. Our proposed approach
makes the following assumptions:
1. The base station or the net control station (NCS)4 has the following capabilities:
(a) Able to communicate with many nodes at the same time on several channels.5
(b) Very robust with high speed processing for handling large volume of information.
(c) Unlimited power. (No batteries used.)
(d) Directional antenna with active sectors.
2. The nodes have the following capabilities:
(a) Function as a sensor.
(b) Transmit straight to the NCS.
(c) Able to change frequencies and channels.
(d) Receive control messages.
The network center (\HEliocentric") is the NCS and each node is subordinate to it. There is
no node-to-node communication. (This is known as a star topography or as a directed net. The
ALOHA net is an early example.)
The NCS and the nodes form the network through an Ad hoc means. That is, the nodes have
enough information to contact the NCS. After a quick roll call, the network is formed and the NCS
has a table of all active nodes. (The nodes do not have a pre-deployment list nor do they maintain
a routing table{Both actions are not needed.)
4Normally writers use the \BS" expression for \base station." I followed that custom when I submitted my
conference papers. I was not comfortable doing this. In the quoting of my conference paper within this document, I
will use \NCS" instead of \BS." Again, I wish to avoid invoking the wrong image in the reader?s mind.
5The terms \channel" and \frequency" are normally synonymous. For this writing, the word \frequency" will refer
to big jumps like from 915 MHz to 40 MHz and the word \channel" will refer to small frequency changes like from
915 MHz to 915.025 MHz.
35
Table 3.1: Example NCS Table Entries
NodeID Distance Direction
001 10 units 000
002 15 units 010
003 20 units 330
.... ... ...
This Protocol would use clues or Inferences to determine additional information. The NCS would
use signal strength to determine the distance to a node and would use the directional antenna to
determine the direction of the node in reference to itself and thus avoid the need for GPS. This
information is stored in a table. As a result of the roll call and these clues, this table contains the
node ID plus the distance and the direction from the NCS. See Table 3.1 for example entries.
The nodes have Nominal ability. The nodes collect data and use the radio to contact the NCS.
The nodes have the software and the hardware pieces for these two activities. The nodes do not
need the ability to determine its location or to manage a routing table or complex algorithms for
handling lost contact. Furthermore, the nodes do not need to process or aggregate the data of other
nodes. In short, the nodes collect data, transmit straight to the NCS, and comply with any NCS?
instructions.
The NCS would use the transmitted data plus the clues from the nodes? transmissions to deter-
mine a Global view of the environment. For example, several nodes might transmit a short message
(moving object). The NCS would process these messages by including the nodes? locations and the
time of receipt to determine additional insights such as the speed and compass heading of the object.
Security is achieved, because the NCS has a list of valid nodes. This prevents bogus nodes from
joining the network. Security is enhanced by selecting frequencies that will propagate out to the
nodes and no further. This would make it di cult for the enemy to intercept communications. If
the nodes have the ability to use directional antennas, then this would go further to reduce the
chance of interceptions. If a node is discovered, the enemy cannot determine where the next node
is located.
Consider Fig. 3.1. The NCS (the black dot) transmits to the nodes (the three rings). Nodes on
the outer most ring would hear the message, although the signal is a bit weaker. Beyond this, the
36
Table 3.2: Selected United States ISM Frequencies
Frequency Distance
40.68 MHz 111 km/69 miles
915.00 MHz 5 km/3 miles
2450.00 MHz 2 km/1 miles
5800.00 MHz 0.78 km/0.49 miles (2565.7 ft)
signal is too weak. So the enemy (the small circles) would not hear anything.
No received signal
y
  
  
&%
?$
d dd
d
Figure 3.1: Communications Rings
With the Friis equations, one can determine where the signal is too weak. If the cuto point is
-86 dBm and the power is 20 dBm, then Table 3.2 would show where the signal is too weak.6
The fourth frequency is used for many wireless devices. Notice the distance is barely 3/4 of a
kilometer. The third frequency was used by older wireless devices. Notice the distance is doubled
to a bit over 2 kilometers. The second frequency is used by some sensors and is very close to the
cellular telephone frequencies. Notice the distance is 5 kilometers. Eons ago, the  rst frequency
was used for cordless telephones, but now the third and fourth frequencies are used. In practice,
distances are less since whip and internal antennas are poor radiating devices.
Expanding the idea of the HEAPINGS base station having a list of valid nodes, if a new node
tries to enter the network or a known node appears to be in a di erent location, then the base station
can tag or mark the node as questionable and will watch the node closely.
Encoding the transmissions for increased security is an option. Of course, this will add to the
energy drain and be an additional overhead. There may be situations where these trade-o s are
needed. Otherwise the security that comes from being far apart and using a frequency that goes
6These  gures assume no line losses, zero antenna gain, and a  at earth. -86 dBm threshold and 20 dBm output
power are typical of 802.11 devices.
37
only so far may be good enough for the mission.
3.3 Flooding and Service Messages
Protocols may be proactive (nodes keep their own and other?s tables current) or reactive (sender
uses a route discovery message prior to sending data), control messages are  ooded out. In static
networks, these messages would reduce since caching would \replay" a previously used route. In
dynamic networks, control messages are numerous and could reach the point where every message
is a control message. Many protocols ignore the fact that many nodes are within radio range of
the sender. For example, a third ring node (see Fig. 3.1) hears a message at least three times (the
sender, the nearest  rst ring node, and the nearest second ring node). If other nodes are within
range, then more \repeats" are heard.
No current protocol di erentiates among the various line-of-sight (LOS) frequencies; all frequen-
cies are regarded as the same with manufacturers preferring the higher ISM frequencies (5.8 GHz
as \better" than 2.4 GHz). So large numbers of nodes must be used along with either a network
protocol or a multi-hop scheme.
All nodes are draining energy at nearly the same pace. To keep the network alive longer, energy
saving tricks are used (like sleeping and reduced power). When used, some nodes will have a longer
life, but the trade-o is delayed or missed messages.
3.4 No Flooding: Only Node-to-NCS and NCS-to-Nodes Communications
In contrast, the HEAPINGS NCS has the routing table and not the nodes. Flooding and relaying are
not used. All nodes are one hop away. Only nodes with data would transmit and the whole network
would live longer. Changing frequencies is the tool for keeping all the nodes in direct contact with
the NCS. The area could be a large room (then could use 5.8 GHz), or several  oors (then could use
2.4 GHz), or several city blocks (then could use 915 MHz), or the size of a city (then could use 40
MHz). This is possible, because the NCS is using directional antennas and these radio bands have
these behaviors. Changing the transmit power level is not necessary.
The nodes transmit only to the NCS and the NCS transmits directly to each node. There is
38
no need to have any nodes relaying for another node. This simple change avoids  ooding and its
tendency to have a \blizzard" of messages.
When a node has data, it requests permission to send (RTS). This avoids con icts, corrupted
messages, and congestion.
When two or more nodes have data, they RTS and the NCS would only acknowledge (ACK) the
node with the highest ID. This node transmits its data while the others would move to the next
open channel. On the second channel, the highest ID would receive the ACK. This is repeated until
each node receives an open channel. With a powerful NCS, many messages could be received at
nearly the same time. The only delay is when the nodes change channels. The worst case would be
when a node tries every channel and none are open. The node would start from the  rst channel
and would repeat the RTS with a note that this is a second attempt. The NCS would ACK and
accept its data. This likelihood is very remote and we will explain later on.
After the data is passed, the nodes would return to channel 1.7 This scheme has an e ciency of
100% channel usage and almost no delay.8
Some may see this as a broadcasting approach, which is true. However, all wireless schemes
are broadcasting to some degree. Any supporter that thinks his/her scheme is not broadcasting is
ignoring the nature of the wireless environment.
3.5 Large Message Volume
All protocols use a  rst-come- rst-served approach until there are con icts. The Slotted Aloha
protocol[75] and others solved this with assigned, but limited slots. Other protocols use contention
schemes[75], but there are con icts and some nodes would wait before retrying. So more nodes are
handled and tra c leveling is done. The problems are delayed data and possible unserviced nodes.
HEAPINGS messages are received fairly quickly. The channel number provides insights to the
age of the data and this is included in the processing. Since there is almost always an open channel,
7If experience reveals that the data follow a \machine gun" pattern where the data comes several times, then the
protocol could be modi ed whereby the nodes would delay for a period of time before returning to channel 1.
8Tanebaum[75] discussed the Binary Countdown Collision-Free Protocol. He noted that the nodes would wait
until the  rst node was  nished, then the base station would accept tra c from the node with the next highest ID.
He pointed out that this would have an e ciency of 100 %. Our approach has the same e ciency, but with greater
throughput.
39
there is no problem of new data overwriting old data.
Would the NCS be overloaded? Reality would mitigate the message volume. For example, the
largest army may have about 14,000 tanks. It is doubtful that all 14,000 tanks would be in the
same theater. If a typical tank is 3.4 meters wide and if the sensors are in a valley with a 40-meter
wide opening, then only 10 or 11 tanks could travel abreast. This would reduce the amount of data
collected and transmitted during any time period. If all 14,000 tanks passed through this valley,
there would be from 1,272 to 1,400 messages, but transmitted at staggered times.
Since HEAPINGS rarely uses control messages, most transmissions are data messages. Real
messages are not being held up by control or services messages. In short, with fast processors, with
short messages, and with irregular and real time transmissions, the NCS could handle many nodes.
3.6 Frequency Selection
All the nodes would use \Frequency A" until the NCS directs a change to \Frequency B." The
NCS would select a frequency that would ensure all nodes are in direct contact with itself. This
avoids the need for relays. Also, using the highest working frequency would bring in an element of
security since the signal would be too weak for anyone to use beyond a certain distance. (A di erent
approach might be to assign each node a dedicated channel. But the problem is that this would
not scale up very well. So having all the nodes on the same channel would be better. The nearby
channels would be needed during heavy tra c periods.)
To make this clearer, consider again Fig. 3.1. The black dot is the transmitter. Stations on the
three rings would hear the same message at the same time.9 Stations on the outer most ring would
hear the message, although the signal is a bit weaker. Beyond this ring, the signal is too weak to be
useful. So the enemy (the small circles) would not be aware of a transmitter in the center, because
the weak signal would be covered up by background noise.
To make the point stronger, look at Table 3.3. This is listing of all United States approved LOS
ISM frequencies. Recall that we can determine how far a signal travels before it becomes too weak
9Light travels about 300,000,000 meters (about 186,000 miles) per second. The equatorial radius of the earth is
about 6,000 kilometers (about 4,000 miles) and the circumference is about 40,000 kilometers (about 25,000 miles).
Since we are dealing with distances much less than the earth?s circumference{such as 1 degree or about 111 kilometers
(about 69 miles){We can regard a message as arriving at all spots at the same time.
40
Table 3.3: US ISM Frequencies (-86 dBm sensitivity and 20 dBm output)
Frequency Distance Expressed in feet
(MHz) (kilometers/miles) when below a mile
40.68 111 / 69
915.00 5 / 3
2450.00 2 / 1
5800.00 0.78 / 0.49 2565.7
24125.00 0.19 / 0.12 616.8316
61250.00 0.07 / 0.05 242.9561
122500.00 0.037 / 0.023 121.4781
245000.00 0.0185 / 0.0115 60.73903
by using the Friis10 suite of equations. If -86 dBm is used as the cuto point, then our Table 3.3
shows the rough spot where the signal is too weak to be useful for a 20 dBm transmitter.11
The NCS selects a frequency that would ensure communications with all of the nodes. With
stationary sensors, the selected frequency would work for the duration of the network. If the nodes
are mobile as in the case of the military?s  ying sensor platforms, then a frequency change plan
would be used. For example, the NCS might detect that the nodes are moving away when the signal
level begins to drop. When a decision point is reached, the NCS would direct all nodes to change to
the new frequency. (The frequency list is pre-loaded in all of the nodes. The NCS would direct the
nodes to change to the next frequency in the list.)
HEAPINGS is not a frequency hopper. HEAPINGS does not need high power nor need power
control; all the nodes may use the same low power setting. The NCS uses a directional antenna
for receiving weak signals. Some amateur HF stations use  ve watts or less, high gain directional
antennas, and small radios to talk around the world.12
10Sometimes this is misspelled as \Friss." For more information on Harld T. Friis, see the IEEE website biography
section.
11These  gures assume no line losses, zero antenna gain, and a  at earth. -86 dBm threshold and 20 dBm output
power are typical of IEEE 802.11 standard devices. Also, there was no consideration whether or not any radios exist
that could transmit on any of these frequencies.
12http://www.wb8nut.com/qrp.html has information about a 500-milliwatt radio and a 12-volt powered 750-
milliwatt radio.
41
3.7 How HEAPINGS Functions
3.7.1 Establishing the Network
Sensors are deployed in an area with random13 spacing. This could be anywhere from 5 kilometers
(3 miles) to the horizon. These could be placed far apart. The NCS could be on a mountain ridge.
The nodes are set to use 915 MHz as the starting frequency. The NCS transmits a roll call message.
It only hears from those nodes that are at the maximum distance of about 6 kilometers (4 miles)
away or less.14 The number of nodes, N, is known. Reaching a small number tells the NCS that a
frequency change is needed. The NCS commands these nodes to use a lower frequency such as 40
MHz.
At the same time, those nodes that did not hear the NCS would wait a bit, and then switch
to the lower frequency. When the NCS is on the new frequency, it would send out a new roll call
message. Both the previously found nodes and the previously \missing" nodes would reply. The
NCS would acknowledge each node and add an entry in its table{the node ID, distance (based on
signal strength), and bearing (based from the base station?s directional antenna).
3.7.2 Sensors Working{An Example
A northeastern sensor detects a tank. It transmits a simple message, \Tank" and its node ID. The
NCS records that a tank was sighted at a certain time.15 From the information about the node?s
direction and distance, the NCS can provide addition data about this tank sighting. If nodes west of
this node start transmitting messages on this tank, then the NCS can look at the time-of-receipt and
these nodes? locations to determine how fast the tank is moving and its direction of travel. If several
tanks are present, then these would generate several messages. Thus the NCS could determine that
more than one tank is moving through the area.
13A random placement is more likely than a precise placement. This makes for an interesting academic study, but
not very useful for the real world.
14Based on 20 dBm output and 3 dBi antenna gain.
15The time is based on the NCS?s internal clock. The nodes do not use a clock.
42
3.8 Comparing Other Protocols with HEAPINGS
The following paragraphs provide more insights on HEAPINGS by comparing and contrasting with
other protocols.
3.8.1 Flooding and Service Messages
Many protocols expect the nodes to be very robust (that is, having many abilities). Some protocols
are proactive and the nodes keep their own and other?s routing tables current. Other protocols are
reactive and will send out a route discovery message prior to sending out data. Either way, control
messages are used and these are  ooded out. In a static environment, these messages would become
fewer since caching techniques would \replay" a previously used route. In a dynamic environment,
control messages would be numerous and could reach the point where all the messages are control
messages. To solve this problem, the solution might be to slow things down or to reduce the network
membership.
As mentioned earlier, nearly all current protocols do not take advantage of the fact that many
nodes are within direct communications range of a sending node. Again, a node that is located on
the third ring (see Fig. 3.1) from the sending node would hear a message at least three times (the
sender, the nearest node on the  rst ring, and the nearest node on the second ring node). And again
if there are other nodes on the  rst and second rings that are within radio range of this example
third ring node, then four or more repeats of the sender?s message could be heard|a total of at
least seven transmissions. If the outer most nodes do not realize that there are no more nodes, then
this message would be repeated several more times. Until the message has worked its way to the
network edge, no new message could be introduced. Some protocols attempt to reduce the amount
of  ooding, but none can completely eliminate it.
In contrast, HEAPINGS has the routing table at the NCS. It does not send out table update
messages. Flooding is not needed, because each node is able to hear the NCS. Changing frequencies
is the tool to ensure that all nodes are able to stay in direct contact with the NCS and thus avoid
using relays.
In other protocols, all LOS frequencies are regarded as the same with manufacturers preferring
43
ISM frequencies and they perceive 5.8 GHz as the best one to use. There is no frequency management,
or selection guidance. To handle communications beyond direct node contact, either a network
protocol is used or a multi-hop approach is used. This adds additional overhead.
There is a continum from one end that uses simple hardware and complex protocols to the other
end that uses complex hardware and simple protocols. As radio technology continues to advance,
the wiser decision may be to rethink the protocols and use those that best take advantage of the
latest hardware.
Sensor networks tend to be stable, but this is not a requirement for HEAPINGS since a LOS
frequency is able to cover an area. So a dynamic environment is not a problem. The area could be
the size of a large room (would use 5.8 GHz), or could be several  oors (would use 2.4 GHz), or
could be several city blocks (would use 915 MHz), or could be the size of a \city-state" (would use
40 MHz). If some nodes started to move beyond the coverage area, then the NCS would command
all nodes to change to a lower frequency. Since these are all LOS frequencies without a sky wave
component, then all nodes would still stay in direct contact. Rapid movement within the area is not
a problem as direct communications is always possible.
In other protocols, energy saving schemes are used in order to keep nodes alive for a longer period
of time. Sleep schemes, reduced power, and other novel ideas are used. With  ooding being a part
of these protocols, all the nodes are draining energy at nearly the same pace. When these schemes
are used, some nodes will have a longer life, but there are trade-o s; messages are either delayed or
lost.
HEAPINGS nodes do not relay messages. Only those nodes with data would transmit. The
result is that the entire network would live longer than the life time of a typical node. So a sleep
scheme is not needed and is not desirable. To express this point in di erent terms, in the traditional
network, any event happening at the farthest distance would require every node from this distance
to the base station to be involved in relaying the message. In time, if every event happen at the
farthest edge, then all the nodes would die together. In the HEAPINGS network, only the outer
nodes are involved. When this outer ring of nodes die, then the next ring of nodes would step in.
These may preceive the events in a weak fashion, but at least there are nodes alive that can collect
data and send to the base station. That cannot be said of the traditional network, which at this
44
point in time the entire network would be dead.
3.8.2 Handling Large Volume of Messages
Another di erence is how high tra c volume would be handled. All protocols would use  rst-come-
 rst-served approach until there are con icts. The Slotted Aloha protocol[75] and others would solve
the con icts with assigned slots. There is a limit on the number of available slots. Have more users
than slots and there are problems. Other protocols might use a contention scheme[75]. There are
still con icts, but some of the nodes would wait for a period of time before retrying. The bene t is
that more nodes may be handled and tra c leveling is done. The problem is that tra c is delayed
and it is possible for a node to never receive service.
HEAPINGS has a di erent approach. When a node has data, it performances an RTS. The NCS
will ACK and the node would transmit. If the node does not receive an ACK and it does hear the
NCS, then it would change to a di erent channel. This repeats until an ACK is received. However, if
the node is on an open channel and it does not hear from the NCS, then the node has a transmitter
problem and it would stop attempting to make contact.
This approach has a number of bene ts. Messages are received fairly quickly. The order of trying
new channels provides insights to the age of the information. If the node is on the  fth channel,
then the message has about four or  ve units of time delay (aging). The NCS would take that into
account when processing the information. Since there is almost always an open channel to the NCS,
the issue of new data overwriting old data is not a problem.
3.9 Answering Critics
In seeking feedback on this concept, a number of issues were raised. These are valid and need to be
addressed. Only the negative comments are covered. The positive ones were omitted.
3.9.1 This is Not a Frequency Hopper Scheme
A frequency hopper scheme is an attempt to keep the other side from eaves dropping on communi-
cations. The dwell time and the list of frequencies are only known to the net members. The other
45
side would have to guess the dwell time and the list of frequencies. HEAPINGS uses one frequency
until it is time to change to another frequency. It does not hop around.
However, that does not prevent the user from adding a hop scheme that uses nearby frequencies.
Of course, any such enhancements would involve overhead and higher energy draining.
3.9.2 High Power and Power Control is Not Needed
All the nodes are assumed to be in the same con guration and using the same power level. The
power level is set at a low level. By using the proper frequency and the proper antenna, weak signals
can be received and processed by the NCS. If need be, the NCS could increase the power level. But
this should not be necessary.
Closely related to this is the matter of power control. The deployed nodes use the lowest power
and the NCS selects a reasonable power level. Since the lowest setting is used, this should enhance
the battary life of the nodes. The NCS has access to unlimited power.
Some amateur HF stations use  ve watts or less to talk around the world. They achieve this by
using high gain directional antennas and small radios.16
3.9.3 Bit Rate Issue
The higher frequencies are favored, because of the belief that these support greater bandwidth (more
information) in a unit of time. Combat environment sensors would be sending short messages with
a small amount of information.
3.9.4 Idle Listening vs. Sleeping
There seems to be at least two understandings of the term \sleeping." Some see this as complete
shut down while others see only certain functions are shut down. One reviewer from the  rst group
wrote that idle listening uses more energy than sleeping. Putting nodes to sleep could be done, but
the premise is that the sensors need to be awake in order to collect information. Being asleep goes
against this requirement.
16http://www.wb8nut.com/qrp.html has information about a radio that has a standard output of 500 milliwatts.
Another version puts out 750 milliwatts and uses a 12-volt battery.
46
If the second group preceives the transmitter as draining energy even when not transmitting, then
putting the transmitter to \sleep" would let the node continue to collect data while saving energy
from a sleeping transmitter. It is not clear what are the assumptions for turning the transmitter
back on again.
In a sense, this is a question that is more in uenced by hardware than by software. If the circuit
is designed where no energy is consumed when the transmitter is not being keyed, then this becomes
a non-issue for the second group and HEAPINGs. This becomes a discussion between the  rst group
and the \all others."
3.9.5 Why Single Hop is Better than Multihop
One reviewer wanted it spelled out clearly why single hop is better than multihop. Multihop involves
many nodes in order to relay one message. Single hop involves just one node. So if the nodes in the
single hop network and in the multiho network are using the same amount of energy, than a message
sent in a multihop network would costs more energy than in a single hop network.
3.9.6 Software De ned Radios
One reviewer was concerned about the size of the sensor node. Today, companies are making and
selling radios that can operate from 30 MHz to 512 MHz and these are no bigger than a \brick."17
Size does not matter as things are becoming smaller all the time.
3.9.7 Border and Pipeline
One reviewer suggested reviewing computer networking issues such as border and pipeline. This is
a sensor network, not the upper level Internet router.
3.9.8 Old Concept
Three reviewers reacted to the submitted conference paper being a concept paper. All protocols
started o as a concept. The writers may nor may not include any simulation work. Then they or
others will take the concept one step further by running experiments and tests.
17See Harris Radio website: www.rfcomm.harris.com/products/tactical-radio-communications/RF-5800m-hh.pdf
47
One reviewer wrote that this is \similar to the traditional wireless LAN communication case."
Furthermore, the reviewer saw the HEAPINGS concept as \neither new nor challenging." Also the
reviewer thought that \granting too much powers to a single NCS seems not a good idea, because
the compromise of NCS [would] paralyze the whole network." A second reviewer did not believe that
a centralized protocol is feasible! The  rst reviewer did agree that \the idea of frequency selection
is more reasonable" but \this small  x could not change the fact that the architecture is not ideal
in a combat environment."
The above mentioned second reviewer did not like the selection of the cited work that had
been done by others and did not like the level of coverage. Since all sensors networks must route
information from point to point, the focus was on protocols that have a routing component. It
would be ideal to attempt to correct this imbalance, but conferences have page limits. As a result
not everything could be included.
What is interesting is that HEAPINGS and tier sensor networks share a number of common
traits. These three reviewers objected to certain HEAPINGS features that also exist in tier sensor
networks. Progress comes by marrying old ideas with new approaches. Also some of the old ideas
like ALOHA were of a centralized nature.
As HEAPINGS matures, consideration might be given to handling the loss of a NCS. Maybe
the way things are done with gateway routers could be considered. Today combat sensors route
information to a central location. HEAPINGS would do the same thing, but without  ooding and
relays.
3.9.9 Antenna Design
One person wrote "widely varying frequencies put considerable burden on the antenna designer."
This is a valid hardware concern. HF radios could be used with an antenna tuner in order to operate
on any frequency between 2 and 30 MHz. The Harris company makes the FALCON III AN/PRC-
152 handheld radio that operates on any frequency between 30 and 512 with the same antenna[78].
This radio normally uses 5 watts and 10 watts to hit a satellite.
Another concern was about the feasiblity of a directional antenna being able to cover a large
part of the radio spectrum. As an example, When the AN/PRC-152 is mounted in a vehicle rack,
48
it is called the VRC-110. It can use the following directional antennas:
 RF-387-AT001 (30108 MHz) Dipole Whip (3.3 m)
 RF-387-AT002 (30108 MHz) Dipole Whip w/GPS Antenna (3.3 m)
 RF-398-03 (30108 MHz) | Dipole Whip (0.3 m)
 RF-9070 (100-400 MHz) Biconical (6.1 m) Mast
 RF-3080-AT001 UHF STACOM (240-400 MHz) Manpack or Base Station Antenna Set
Today, the directional antennas do not cover the same range as the whip antenna. Tomorrow,
that could change.
So the march of technology has made the matter of having a radio cover a large part of the radio
spectrum a non-issue. Today it is 30 to 512 MHz, tomorrow it will be even more. And the same
thing is true for antennas; 300 MHz of radio spectrum can be covered by one directional antenna.
Tomorrow, it will be even more.
3.10 Analysis From a Mathematical Viewpoint
Since the popular simulation packages have weaknesses[79, 80, 81] and may not correctly simulate
HEAPINGS, we decided to use some mathematical approaches before attempting to create any
simulations.
3.10.1 Roll Call
In a regular sensor environment, the NCS would transmit the roll call message. The nearby nodes
would receive and relay the message to the next \ring" of nodes. These nodes in turn would repeat
the message. This would continue until the extreme edge is reached. Then the nodes would transmit
their answers. The e ort of relaying from node to node would continue until the requesting node is
reached.
On the other hand, the HEAPING NCS would transmit the request. All of the nodes would hear
this message. Then the nodes would transmit their answers straight to the NCS.
49
Table 3.4: Comparing Regular and HEAPINGS
Roll Call Actions Regular HEAPINGS
Nodes Relaying 462 0
All Nodes Replying 5,566 484
Total 6,028 484
If the nodes are deployed in m rings with n nodes on each ring, then the total is mn. Assume
that each node will only relay for two nodes (a node on an inner ring and a node on an outer ring)
and transmitting uses one energy unit.
In a regular network, n nodes on m 1 rings would transmit the message for a total of n(m 1)
times. Then each node replies and after relays, the total is nm(m + 1)=2 (ring m?s messages used
mn energy units to reach the NCS; ring m 1?s messages used (m 1)n units to do the same, and
so on). The grand total is (n(m 1) +nm(m+ 1)=2) energy units, which is O(nm2).
HEAPING nodes do not repeat messages so no energy units are used. In replying, each node
transmits back for a cost of mn units, which is O(nm), one order of magnitude lower than using
 ooding. In an example of 22 rings with 22 nodes on each ring (see Table 3.4), the improvement is
about 92%.
In the matter of network life time, the regular network would require the nodes to relay. The
ring closest to the NCS would have the greatest energy drain. Every time the NCS transmits, these
nodes would be transmitting. Every time a distant node replies, these would be involved. If the
outer most ring was involved in every exchange, then the entire network would die together. There
are other schemes that attempt to reduce the amount of  ooding. For those that really can achieve
this, the network would last a bit longer. However, none of these schemes can avoid using  ooding
during some stage of the network?s existence. The network?s life time is fairly close to the average
life time of a node. If nodes are put to sleep, then the network would last longer, but at the expense
of timely message delivery. The trade-o is long network life time and slow message delivery versus
short network life time and quick message delivery.
In the HEAPINGS approach, the nodes? life time would be variable. If the outer most ring was
transmitting frequently, then these would die  rst. The next ring of nodes would become the new
50
Table 3.5: Comparing Aggregation and HEAPINGS
Regular Aggregation HEAPINGS
Nodes used 10240 2303 1024
Savings { 7937 9216
Time used 10 10 1
 rst line of data collection. The rest of the nodes would still have energy available. The issue of
trade-o s does not exist, because other nodes are not needed to relay messages.
3.10.2 Aggregation Or Not
Aggregation?s intent is to \substantially reduces the communication overhead." Consider a m-ring
network with 2m nodes on each ring. Assume two nodes feed to one node (this would be halved each
time as in 2m, 2(m 1), 2(m 2), ..., 2, 1). The total is 2(m+1)  1 nodes. The saving is (m 2)2m + 1
nodes ((m2m (2(m+1) 1)). Accounting for processing time, let?s assume 1 time unit is used. Then
10 time units are used.
Left out of the foregoing is the matter of irregular timing of events. This is aggregation major
weakness. How long should the nodes on the next inner rings wait? How will these determine that
something is one actor (a tank moving) in di erent places or a single event in various places (people
walking)? If a delay is added for collecting the information, then this increases the delivery time.
When would the data become too stale to be useful?
In contrast, HEAPINGS used 2m nodes and 1 time unit. Although aggregation would save energy
over a regular sensor network, it cannot surpass HEAPINGS? speed and savings.
As a concrete example (see Table 3.5, consider a ten-ring sensor network with 1,024 nodes on
each ring. Assume that two nodes feed information to one node. So the  ow would be 1,024 to 512,
then to 256, then to 128, then to 64 nodes, then to 32 nodes, then to 16 nodes, then to 8 nodes,
then to 4 nodes, then to 2 nodes, and  nally to 1 node. This is a total of 2,303 nodes. This is a
savings of 7937 nodes (10,240 - 2,303) or 77.5% (1 - 2303/10240).
However, we need to account for processing time. To make the math easy, we will assume that
one time unit is used. So in the ten-ring example, 10 time units are used.
51
In contrast, HEAPINGS would only use 1,024 nodes and less than 1 unit of processing. Although
aggregation would save energy and time over a traditional sensor network, it cannot surpass the
savings obtained by the HEAPINGS approach. HEAPINGS saving is 90% (1 - 1024/10240).
Again, repeating the problems with aggregation. How long should the nodes on the next inner
rings wait? How will these determine that something is one actor (a tank moving) in di erent places
or a single event in various places (people walking)? If a delay is added in order to collect this
information, then that would increase the delivery time. At what point would the report become
too stale to be useful?
Once an event is detected, HEAPINGS nodes would transmit small messages straight to the
NCS. By processing the pieces, the NCS could provide a Global view or \the big picture" without
aggregation \coarse" reporting and its built-in delays.
52
Chapter 4
SIMULATION WORK
The following describes in some detail the simulation work that was done on HEAPINGS and
appears in the WorldComp ?08 proceedings[11].
4.1 Ns2 Simulation
Using the example values from the Marc Greis? Tutorial[82] on the ns2 simulation program as a
starting point, the  rst simulation e ort was created with four nodes plus a sink node. The packet
size was set as 500 bytes, the interval was set as 0.005, and the capacity was set as 1 Mb. This works
out to 200 packets per second per link and the used bandwidth is 0.8 megabits per second per link.
Together the four links have a total of 3.2 Mb per second. The base station has a capacity that is
equal to the total capacity of all of the nodes. See Fig. 4.1 for the net con guration. The link from
node 0 to node 5 is used as a sink or receiving station.
Figure 4.1: NS2 Prior to Executing Simulation
Using droptail handling for the nodes and a stochastic queuing approach for the base station,
53
some packets are dropped. How much? A minor amount. Visually, the worst case might be  ve
packets with maybe three packets from one source. The lost packets could be quickly retransmitted.
See Fig. 4.2.
Figure 4.2: Screen Shot of NS2 at the Maximum Load
The next simulation e ort was to make the base station?s capacity considerably greater than that
of the total of all of the nodes. One would think that increasing the base station?s capacity would
provide the situation of no dropped packets. But ns2 simulator does not agree. Experimenting with
capacity up to 100 Mb revealed that a few packets still would be dropped.
The third simulation e ort varied the base station?s delay from 10 ms to 5 ms and keep the
capacity equal to the total of all of the nodes. This approached resulted in better behavior. The
worst case packet drop situation improved visually from 5 to 3. Increasing the capacity from 4 Mb
to 10 Mb did not result in any noticeable improvement. The ns2 simulation did not re ect any
noticeable improvement with the base station?s delay being improved to 2.5 ms or to 1 ms.
Of course, the ns2 simulation assumed packets are being generated every 0.005 units and that the
packets are 500 bytes in size. This is not the case for a combat sensor network. Packet generation
would be irregular and the message size would tend to be small. Also the nodes would not always
be continuously transmitting at the same time to the base station.
(We tried to create a network with more nodes, but ns2 has a  le size limit between 4,049
54
and 4,496 bytes. That is, whatever instructions that one might create with the tcl programming
language, must be less than 5 kilobytes. Otherwise, the simulation software cannot run. As a result
the simulation runs were limited to working with four feeding nodes. Again, the link between 0 and
5 is used for the sink or destination of the packets.)
4.2 OPNET Simulation
OPNET?s Modeler software[83] is a commercial product that has the ability to do more than ns2.
It covers the common protocols and vendor devices. The Modeler uses several editors:
1. the Project Editor
2. the Node Editor (Used to create node models that describe the internal  ow of data within a
network object.)
3. the Process Model Editor (Used to create process models that describe the behavioral logic of
a module in a node model.)
4. the Link Model Editor
5. the Demand Editor
6. the Path Editor
7. the Packet Format Editor
8. the Antenna Pattern Editor (for use with the wireless module)
9. the ICI Editor
10. the Modulation Curve Editor (for use with the wireless module)
11. the Probability Density Function Editor
12. the Probe Editor
13. the Simulation Sequence Editor
55
14. the Filter Editor
In addition to the foregoing editors, there is the Analysis Tool.
There is a set of tutorials to help the new user to master each of the foregoing plus other features
of OPNET?s Modeler software. The provided tutorials could take 20 or more hours in order to
complete and not every feature is covered.
OPNET?s Modeler software can explore networks that range in size from an o ce to a campus
to an enterprise to the world. The network could be con gured as a bus, full mesh, randomized
mesh, ring, star, tree, or as an unconnected net.
One can select actual vendor products. For example, one could select to use a stack of two
3Com R SuperStack R II 1100 and two SuperStack R II 3300 (See http://www.3com.com/products/
en_US/detail.jsp?tab=support&pathtype=support&sku=3C16982-US)1 chassis with four slots,
52 auto-sensing Ethernet ports, 48 Ethernet ports, and 3 Gigabit Ethernet ports. The OPNET?s
Modeler is able to handle more nodes; the  rst tutorial uses a network that has 30 nodes. Modeler
is able to collect statistics for individual nodes or for the entire network. One key statistic is server
load and another one is Ethernet Delay. The wireless simulations can collect environmentally unique
statistics.
4.2.1 Details on the Created Wireless Simulation
Instead of creating the entire HEAPINGS sensor network, we looked at the base station, one trans-
mitting sensor node, and a moving jammer station. We narrowed our focus to exploring the di erence
between an omni-directional antenna and a directional antenna. We ran two di erent scenarios; the
 rst one had the jammer moving behind the two stations (See Fig. 4.3 for route.) and the other
had the jammer going between the two stations (See Fig. 4.4 for route.). Within each scenario,
we used an omni-directional antenna and a directional antenna. The transmitting node only has an
omni-directional antenna whereas the base station has the option of using either antenna. It does
not matter what kind of antenna that the jammer is using, but we opted to use an omni-directional
antenna.
1Both products have been discontinued for the SuperStack R III series.
56
Figure 4.3: Jammer Trajectory Behind the Nodes
Figure 4.4: Jammer Trajectory Between the Nodes
Packets would be transmitted at 1024 bits per second while using 100 percent of the channel
bandwidth. Or to express in di erent terms, the packet generator generates 1024-bit packets that
arrived at the mean rate of 1.0 packets per second with a constant interarrival time. (These are the
default values for OPNET wireless simulations.) We looked at the interference impact on receiving
complete packets. If the signal-to-noise ratio (SNR) is too low, then the packets would be corrupted
and must be rejected. We collected statistics on the bit error rate (BER) and on the throughput in
packets per second. We used 0.0 errors per bit as the cut-o point. That is, only totally error-free
packets were accepted; we did not considered using any error recovery protocol. (OPNET has the
ability to limit collection of statistics to successful exchanges between two nodes. We did not use
this option, because it would not show the amount of lost packets.)
To keep things simple, we did not vary the transmitter power, the frequency band, the modulation
type, nor the distances. The jammer is given more power (20 versus 1 for the sensor node) and its
signal modulation is changed{in order to create more impact on the receiver. The resulting signal
57
Table 4.1: The Four Scenarios
Omni Directional Omni Directional
and and and and
Behind Behind Between Between
Packets Created 1420 1420 1420 1420
{From Node 710 710 710 710
{From Jammer 710 710 710 710
Packets Destroyed 1296 724 1418 700
Packets Received 122 694 1418 718
would be perceived as noise. To have a more pronounced graphic for the second scenario, we followed
the suggestion of the tutorial and changed the antenna gain from 20 dB to 200 dB. True, this is not
a reasonable situation since in the real world antennas do not have that huge amount of antenna
gain; this was done merely to illustrate a point.
4.2.2 Running the Wireless Simulation
Table 4.1 presents the four scenarios in a tabular form. The  gures for the packets is the total
amount generated during the simulation run.
The Packets Destroyed and the Packets Received do not equal the Packets Created  gure, because
the simulation ended just before the last packet from the jammer and from the node reached the
base station. Otherwise, these would have been processed and included in the total.
4.2.2.1 Packet Reception Statistics
With the jammer being beyond both the sensor node and the base station, the impact was reduced.
So 122 packets were successfully received with the omni-directional antenna. When the directional
antenna is used, the amount of received packets improved to 694 or almost a six fold increase.
With the jammer traveling between the sensor node and the base station, the amount of good
packets was zero. The directional antenna enabled almost half of the packets to be received.
It appears that the second e ort was better than the  rst e ort. That is not necessary true. The
two tracks were di erent. There was no attempt to draw the two tracks to prove the same exposure
time of the jammer to the base station. The Fig. 4.3 track was quickly created while as the Fig. 4.4
58
track was drawn to be equal distance for the north south leg and parallel to the directional antenna?s
main beam. The other point is that an antenna with very high gain tends to have a very narrow
footprint{that is, a good balance between no gain and very high gain would enable the base station
to monitor more of the environment.
4.2.2.2 Bit Error Rate and Packet Throughput Results
In the  rst scenario (See Fig. 4.5), the BER starts o low and begins to increase. This follows the
rough arc path2 of the jammer. And of course, the packet throughput started o high and fell to
nil.
Figure 4.5: Jammer Trajectory Between the Nodes
When the  rst scenario is run with the high gain directional antenna, the BER is high, because
the jammer is inside the antenna?s cone (See Fig. 4.6.). When the jammer exits the antenna?s cone,
then the BER drops and the packet throughput goes up. Most of the time is jammer-free.
In the second scenario (See Fig. 4.7), the omni-directional antenna reviewed showed that the
BER gradually increased as the distance between the jammer came close to the base station. The
BER reached a maximum of about 0.32 errors per bit when the jammer is at the closest distance.
2The drawing re ects half of the curve, because the simulation was set to run as long as it took for the second
scenario. The second scenario did not require as much time as the  rst.
59
Figure 4.6: Jammer Trajectory Between the Nodes
There was no point when the environment appeared to be \free" of the jammer?s transmissions. The
two humps re ect the two points where the closest distance is the same.
Figure 4.7: Jammer Trajectory Between the Nodes
When the second scenario is run with the high gain directional antenna, the BER is the same
as the background noise and it appears that the jammer does not exist. When the jammer aligns
with the transmitter and the base station (comes into the antenna?s cone, the result is a total loss
of packets. Fig. 4.8 shows a pronounced spike for the period of the alignment.
Figure 4.8: Jammer Trajectory Between the Nodes
60
4.3 Analysis
In [10], we used mathematical analysis to show that HEAPINGS is able to generate more useful
work with almost no overhead.
HEAPINGS takes such a di erent approach from other protocols that it is hard to design a
simulation that does a good job of showing o HEAPINGS features. So we took a small part of
the protocol and designed a simulation to show one feature or bene t of HEAPINGS. That is, we
had made the point that a HEAPINGS node could transmit straight to the base station and the
base station?s directional antenna would provide a narrow beam, greater reach, and insights to the
location of the various nodes. The ns2 simulations showed that HEAPINGS is able to handle a
large tra c volume with very little lost information. Again this comes from changing the operating
assumptions. HEAPINGS has the bene t of using less resources to provide more information.
Furthermove, we stated that the nodes do not need to relay to other nodes. In other sensor
networks, the nodes and the base station all use omni-directional antenna and relaying is used. So
if a jammer was passing through the sensor  eld, communications within the network would be
harmed, even destroyed. The OPNET simulation with omni-directional antennas supported this
belief.
On the other hand, the jammer would only be e ective if can come between a node and the base
station{being in the general area will not give it the bene t of harming the communications. The
OPNET simulation with directional antennas supported this belief.
So the OPNET?s Modeler simulations showed the antenna behaviors and provided some useful
insights. It is clear that a directional antenna with a reasonable amount of gain can cover an area
and be able to receive a large amount of packets.
Now, the envisioned real world sensor network would have nodes that do not generate tra c
continuously. So overloading the base station is not likely to be an issue. Also the other side may
be more interested in shutting down the network than jamming it.
61
Chapter 5
EXPERIMENTAL WORK
The natural development or the next step would involve conducting some real world experiments.
Due to unrepairable test equipment, limited  exiblity in the current test equipment and radios, and
limited funds, it was not possible to conduct any proof-of-concept tests. This would need to be done
by others or done at Auburn when funding is provided.
One way that others could con rm that the concepts are valid is to conduct some simple mea-
surements. The following is an extract from a draft journal submission.
5.1 Introduction to the Draft Journal Submission
There is a symbiotic relationship between the military and science. What the military perceives as
its need, science will rise to the occasion and produce the solution. The United States military began
about in the early 1980s to shut down or greatly reduce their HF radio nets in favor of using the
satellites (clear channel and dialing ease), cellular telephones (not tied down to a desk and dialing
ease), and now the Internet (with access to much information). As a result, the current generation of
computer scientists and computer engineers tend not to have the radio communications background
and understanding that the older generations have had. They tend not to be interested in radio
communications as a hobby nor to become amateur radio operators. They may not understand why
there are problems with CrownCom?s lofty goals of \using dynamic spectrum access".[84]
When hardware kits are ordered for various purposes, there does not seem to be much under-
standing about the radio section. Many wireless devices tend to operate in the 2 and 5 GHz sub
bands. But sensor kits operate in much lower radio bands. A few universities are working with
CubeSats and these operate in the 420 to 450 MHz sub band. For a number of the users and stu-
dents, there is no thought about why the devices operate in these various sub bands. They may not
62
understand that the lower frequencies propagate better than the higher frequencies. They may not
understand antennas.
For example, the Auburn University Student Space Program has a document[85] that reports on
the program progress. In the section that reported on how the students determined the range of the
radios, this appeared on page 70:
... take the antenna and transceiver about a mile from the ground station to see if we
could hear a signal to and from the antenna.
If communications is achieved, then it would be a minor miracle, because the distance between
the satellite and the ground station would be closer to 700 kilometers, not the mile of the foregoing
experiment. This documentation is very detailed, but there is no information about how to calculate
the answer. Knowing the answer could determine if the right antenna is being used or if the proper
transmit power setting is being used. Space makes for an expensive place to conduct trial and error
experiments.
5.2 Working Background
So what do today?s computer scientists and computer engineers have as a working background?
They and others know that laptop computers can use Wi-Fi locations to access the Internet.
They may or may not understand why the range is so short for some locations and even shorter
elsewhere. They may notice that their cellular telephones may require more recharging when they
are away from large population centers, but they may not understand why. They have  gured out
that coverage around a spot is like a circle, but they may not realize the coverage is spherical.
What should computer scientists and computer engineers understand about the foregoing? They
should understand that radio signals are traveling in an expanding front in all directions at about
the speed of light or roughly 300,000,000 meters per second.[86] And they should understand that
at this speed a radio wave could circle the globe in about 1=7 of a second.[86]
As a radio signal travels, it loses some of its energy. This comes more from the spreading or
expanding wave front than from anything else. The strength drops in an inverse fashion. For
63
Table 5.1: De nitions of Terms in the Distance Equation
Term De nition
HTx Height of the transmit antenna.
HRx Height of the receiver antenna.
example, if the wave strength or  eld strength1 at 1 mile is 100 millivolts per meter2, then at 2
miles it will degrade to 50 millivolts, at 3 miles it will degrade to 25 millivolts, and so on. Computer
scientists and computer engineers should understand why lower frequencies survive the \trip" better
than higher frequencies.
The proposed solution might be to use directional antennas. So how does a high gain antenna
prevents this from happening? Computer scientists and computer engineers may not understand
that directional antennas do not prevent this from happening. Instead it controls the amount of
spread by reshaping the spreading pattern so that more energy or radio signal is traveling in the
same direction.
5.3 What Should Be Understood?
Dealing with line-of-sight (LOS) frequencies (above 30 MHz), one learns that the curving earth
will result in the signal traveling o into space. But LOS communications goes a bet beyond the
expected distance to the horizon. So the question becomes: \How far is line-of-sight?" Equation
5.1 answers that question. Equation 5.1 assumes a smooth or  at earth and the receive antenna is
on the ground. If the receive antenna has some height, then 5.2 would be better to use. Table 5.1
provides the de nitions of the terms.
Distance(kilometers) = 4:124
p
HTx(meters) (5.1)
Distance(km) = 4:124
 p
HTx(m) +
p
HRx(m)
 
(5.2)
1This is the more common expression.
2The common way of expressing this.
64
The complete picture involves gain that is added to a signal and path losses that take from a
signal. There are seven factors that contribute to the signal gain:
 Receiver sensitivity
 Receive antenna gain
 Receive antenna height
 Transmitter power
 Transmit antenna gain
 Transmit antenna height
 Required signal-to-noise ratio
These will be brie y explained in the following paragraphs.
5.3.1 Receiver sensitivity
Receiver sensitivity is the di erence between hearing noise and hearing a weak signal. Table 5.2 has
some rough values.3 The last two columns of Table 5.2 re ects the rough receiver sensitivity for two
signal widths4 as provided by [86].5 That is, the narrower the signal width, the better the values.
5.3.2 Antenna Gain
Antenna gain can range from 0 dBi for a whip antenna to 3 or so dBi for a dipole to 5 dBi or higher
for a yagi antenna. If an antenna is embedded in a device, it could actually have a negative gain.
5.3.3 Antenna Height
The height of an antenna can add gain, but only when the antenna is at least 30 feet above the
ground. An antenna mounted 50 foot above the ground sees about 4 dB gain improvement for a
3Those marked with an asterisk are provided values from [86]. The other values are interpolated by adding 3 dB
and tripling the frequency.
4Communications people call this \bandwidths." To avoid confusion with the computer usage of this word, the
normal communications term of \bandwidth" will not be used.
5[86] does not have an entry for the last column of the last row.
65
Table 5.2: Working Receive Noise Figures
Example Value Comments 3 kHz 100 kHz
Frequencies in dB
50 MHz 3 Amateur* 167 dBW 150 dBW
144 MHz 5 Amateur* 164 dBW 149 dBW
150 MHz 5 Interpolating 164 dBW 149 dBW
222 MHz 5 Amateur* 164 dBW 149 dBW
432 MHz 8 Amateur* 161 dBW 146 dBW
450 MHz 8 Interpolating 161 dBW 146 dBW
915 MHz 9 ISM Interpolating 160 dBW 145 dBW
1296 MHz 10 Amateur* 159 dBW 144 dBW
1350 MHz 11 Interpolating 159 dBW 143 dBW
2450 MHz 12 ISM Interpolating 157 dBW 142 dBW
4050 MHz 14 Interpolating 155 dBW 139 dBW
5800 MHz 15 ISM Interpolating 154 dBW ??? dBW
50-mile leg and about 2 dB for a 100-mile leg. The best value is a 100-foot mounted antenna for a
50-mile leg where the added gain is 8 dB.
5.3.4 Transmitter Power
The more power pushing a signal, the longer it will last. To express the transmitter power in the
same units as antenna, then one would take the log of the power and multiple the result by 10. For
an example, a 500-watt station has 26.9897 dB above 1 watt.
26:9897 = 10log(500) (5.3)
5.3.5 The Result
One would add the values to obtain the station gain. Then the matter of path loss must be consid-
ered. The  rst 100 miles, the path loss  gures increase quickly. After the 100-mile point, the rate
of increase tapers o .
For example, a station on 144 MHz is attempting to communicate with a station that is 300
miles away. The path loss of 214 dB. So combining the station gain, the signal is improved. Is the
distant station able to work with the resulting signal? That is, if the network needs to have better
66
Table 5.3: De nitions of Terms in Friis Equation
Term De nition
Pr Power as perceived by the receiver.
Pt Transmitter power.
Gt Gain (dBi) as added by the transmitter antenna.
Gr Gain (dBi) as added by the receiver antenna.
 The radio frequency in wavelengths.
R The distance in the same measurement unit as wavelength.
dB values, then changes would be needed. Either reduce the distance or increase the antenna gain
or raise the antenna higher or narrow the signal?s bandwidth in order to gain improvements.
This is not the complete picture. A slight adjustment is needed for fading. Although fading
is more common on frequencies below 30 MHz, VHF does experience some of this. Another small
adjustment is needed to re ect line loss. In \quick-and-dirty" discussions, these two adjustments
may be ignored and not be re ected in the calculations. One such equation is the Friis equation.
5.3.6 Friis Equation or Formula
The (Harald T.) Friis Transmission Equation may be used to determine the gain of the link (the
\power" of the signal). There are di erent forms. The more simple form is the transmission equation
5.4.
Pr
Pt = GtGr(
 
4 R)
2 (5.4)
Equation 5.4 is some times written from the viewpoint of the receiver as in equation 5.5.
Pr = PtGtGr(  4 R)2 (5.5)
Table 5.3 provides the de nitions of the terms. Note that the antenna gains are expressed in
dBi instead of other ratios and that the wavelength and the distance must be in the same units of
measurement.
There are more complex forms of this equation. Some of these are attempts to capture the
67
Table 5.4: De nitions of Terms in Hata Equations
Term De nition
LU For Urban areas. Answer in dB.
f Frequency expressed in MHz.
hB Height in meters of the base station transmitter antenna.
CH One of these sub equations:
Small to medium city: 0:8 + (1:1logf  0:7)hM  1:56logf
Large city (150  f  200): 8:29(log(1:54hM))2  1:1
Large city (200 < f  1500): 3:2(lo(1:75HM))2  4:97
d The distance expressed in kilometers.
impacts of impedance mismatches, directional antennas not pointing straight at each other, and
di erent antenna polarizations. One form that is a slight modi cation of the basic Friis two-antennas-
in-free-space equation adjusts the exponent \2" to be di erent values between \3" and \5" for urban
settings. Although these adjustments re ect the urban environment, if at least one of antennas is at
least 30 feet above the ground, then the measured gain will improve. This improvement is known
as \station" gain.
5.3.7 Hata Equation or Formula
Another predication equation is the Hata suite. These various forms predicate the degradation of
the signal. The equations attempt to capture what was found by real experiments in Tokoyo during
the 1960s. The basic form of the Hata equation 5.6 follows:
LU = 69:55 + 26:16logf  13:82loghB  CH + [44:9  6:55loghB]logd (5.6)
Table 5.4 provides the de nitions of the terms.
The Hata model for suburban areas uses equation 5.6, but improves the loss by two adjustments:
LSU = LU  2(log f28)2  5:4 (5.7)
The Hata model for open areas uses equation 5.6, but improves the loss by three adjustments:
68
Work Sheet for Each Frequency
Frequency
Antenna Gain
Transmitter Power
Distance
Antenna Height
Signal Width
Friis Measured
Result of Calculated Values
changing Values
Frequency
Antenna Gain
Transmitter Power
Distance
Antenna Height
Signal Width
Figure 5.1: Sample Form
LO = LU   4:78(logf)2 + 18:33logf  40:97 (5.8)
With the complex math steps, it is easy to forget that small values are better (less signal loss).
Whereas, in the Friis family of equations large values are better (received signal strength or power).
The Hata equation is not perfect. There is work to correct errors or to improve it. See Medeissis
and Kajackas[87] as an example.
5.4 The Approach
Previous generations of computer scientists and computer engineers gained an understanding of the
foregoing from interactions with the physical world. Current and future generations tend to interact
with a virtual world that may parallel the real world. Textbook study helps, but not completely.
Our approach would help the new computer scientists and the new computer engineers to gain a feel
for line-of-sight radio behaviors. The students would use a worksheet similar to Figure 5.1. These
are the  elds of the Friis equation plus the two  elds that are not included.
Then the student notes a baseline value. Varies each parameter and notices the changes.
69
Modi ed Work Sheet
Frequency Mxxx.xxxx
Antenna Gain nn dBi
Transmitter Power nn watts
Distance
Antenna Height
Signal Width AFSK FSK
Friis Measured
Result of Calculated Values
changing Values
Distance
Antenna Height
Signal Width
Figure 5.2: Modi ed Sample Form
5.4.1 Auburn University As An Example
Not all locations have fully stocked electrical/electrons lab. Auburn University is no di erent from
the other universities. Yet these insights can be taught with the less than ideal equipment.
5.4.1.1 What is Available at Auburn University
At Auburn University, we only had a  xed frequency transmitter with a dipole antenna and with a
 xed output power. So a student could vary the distance, the antenna heights, and the signal width.
Figure 5.2 re ects the changes to the form.
Using AFSK or Audio Frequency-Shift Keying is useful, because it has a wider signal than FSK
or Frequency-Shift Keying. AFSK varies audio tones whereas FSK varies or moves between two
nearby frequencies. Changing the height of one end and changing the distance are options that are
possible at any college or university. After the student has obtained a good spread of distances
values and of antenna heights for each emission, then the results could be collected into a table like
Figure 5.3. The Friis equation would be executed for each row.
The student should be able to notice some patterns and some \surprises."
70
Distance Height AFSK FSK FRIIS
A X * *
A Y * *
A Z * *
B X * *
B Y * *
B Z * *
C X * *
C Y * *
C Z * *
Figure 5.3: Results
5.5 Conclusion
Once a student has done a number of measurements, he or she will begin to understand the nature
of radio behavior. Also he or she will understand that real world does not always align with the
results of an equation. This discovery will enable the student to take equation results with a \grain
of salt."
Perhaps seeing that a narrower signal width has better results may encourage better hardware
designs. For example, Single Side Band (SSB) is used instead of amplitude modulation (AM) on
the HF bands, because SSB uses half the signal width of AM. The bene t is less power is needed
and a clearer signal. Both federal and non-federal radio nets are converting from 16 kHz wide
FM signals to 11 kHz or less. The bene t is a closer spacing of nets and a doubling of available
radio channels. Television stations are beginning to convert from 6 MHz-plus wide signals to much
narrower digital signals. The bene t is better sound and video while freeing up more radio spectrum
for other services to use. Today too many wireless designs do not seem to consider the need for a
small spectrum footprint. Maybe this teaching approach could help the students to think in terms
of radio spectrum.
Future work would be to use this approach in a networking course to validate that this helped
the students to understand more about wireless networking.
71
Chapter 6
CONCLUSIONS: ANTICIPATED BENEFITS
6.1 General Comments
I anticipate that the work on the helio protocol suite could result in the ability to track  re ghters
traveling through a burning building. The success of this e ort would mean no wired environment,
no need to carry a beacon, no need to have \fake" GPS or direction  nding stations deployed around
the burning building. The  re ghters would not need to carry another device or radio. Instead all
of the workload would be on the base station computer that could be on or in the  re chief?s vehicle.
Heaven forbid that a repeat of 9-11 should take place. But if it does, then it should be easier to zero
in on the exact locations of the trapped  re ghters and other safety workers.
Likewise, the helio protocol suite could track a person moving through a building. From access
point to access point, the person would be tracked. With directional antennas, the deployment of
access points could be done better without the wasteful overlapping coverage.
Aircraft  ying a grid pattern could have one aircraft functioning as a command-and-control
element. Without the need for a heavy duty radar system, the locations of each involved \player"
could be determined. The resources could be managed better. Aircraft would not bump into each
other and no search area is missed.
6.2 Security Comments
I anticipate that the helio protocol suite could enhance the security of a sensor network on several
levels.
6.2.1 Physcial Security
Physical security could be enhanced, because the base station does not need to be near any of the
sensor nodes. The command element and the base station could be placed at some distance to the
72
rear of the battle eld.
6.2.2 Mission Security
Mission security could be enhanced, because the nodes could be spread out for better coverage of an
area. Each node transmitting back to the base station would provide a piece of the big picture. The
base station could construct the big picture in real time and with greater precision. To achieve the
same coverage with traditional sensors networks would require a greater number of nodes{and such
an approach would require the network to deal with routing congestion and the great likelihood of
a chance discovery.
Again, by having the nodes far a part, this would make it harder for someone to make a chance
discovery. This is an always a present problem for the traditional sensor network.
Another part of mission security is the matter of lifetime. A network that dies quickly is a
network that is not collecting and forwarding information. Anything that can enhance the lifetime
of the network should be used. HEAPINGS achieves a longer lifetime by reducing service tra c
and cutting out  ooding messages. Also by using directional antennas, the nodes do not need to use
high power to reach the base station.
6.2.3 Communications Security
Communications security is a balance between speed and costs. Taking advantage of the radio prop-
agation behaviors and directional antennas would result in a low cost approach to communications
security. That is, by selecting the right frequency in order to reach the nodes and not go any further,
this would result in the other side not being able to intercept the radio signal.
Adding a bit of overhead, then the HEAPINGS base station could validate a node. That is,
it has a list of valid nodes. After a roll call, it will know what nodes are present and where they
are located. If a new node tries to enter the network or a known node appears to be in a di erent
location from where it checked in, then the base station can tag or mark the node as questionable
and will watch the node closely.
73
Encryption security is a question. Could the nodes? transmissions be encoded? This is not listed
as one of the central features for HEAPINGS, because such e ort would require extra overhead and
a greater drain on the batteries. However, it is possible that the mission may be such that the
commander would accept the reduced lifetime in order to have this feature. Thus, all decisions have
trade-o s.
Future researchers might  nd an solution that would reduce the impact and thus give the com-
mander greater  exibility.
Another way to have enhanced communications security is by reducing the amount of message
tra c. HEAPINGS is able to greatly cut down the amount of message tra c by cutting out  ooding
and reducing service messages. Thus there are no transmission, the other side cannot guess where
the transmitter is located.
6.3 The Future
HEAPINGS is a concept. It is still immature, but it does suggest a new direction for research. The
Internet and the routing protocols are based on the cold war envirnoment where a nuclear bomb
could take out a large part of the network. These \cold war" approaches have been applied to ad
hoc networks and to sensor networks. That may not be the best approach for networks that are not
trying for coast-to-coast coverage.
I did not know about triered networks when I  rst began my ponderings, but I was pleased to
noticed that these researchers are beginning to break with the \cold war" mind set{for example, the
nodes are not robust and do not have numerous abilities.
In short , my approach should result in better safety, better solutions to the geo-location problem,
saved lives, and better collection of information.
74
BIBLIOGRAPHY
[1] Answers.com. Biography: Arthur schopenhauer, Year unknown; accessed 2008. URL http:
//www.answers.com/topic/arthur-schopenhauer?cat=entertainment.
[2] Martin J. Feuerstein, Michael A. Zhao, Yonghai Gu, and Scot D. Gordon. the future of smart
antennas: Evolution to 3G and IP networks. In Proc. 11th IEEE International Symposium on
Personal, Indoor and Mobile Radio Communications (PIMRC 2000), volume 1, pages 48{54,
London, United Kingdom, September 18{21, 2000.
[3] K. Rambabu and R. Rajagopal. Smart base station antenna. In Proc. 2000 IEEE Interna-
tional Conference on Personal Wireless Communications, pages 303{306, Hyderabad, India,
December 17{20, 2000.
[4] R. Hoppe, P. Wertz, G. Wol e, and F. M. Landstorfer. Wideband propagation modelling for
indoor environments and for radio transmission into buildings. In Proc. 11th IEEE Interna-
tional Symposium on Personal, Indoor and Mobile Radio Communications (PIMRC 2000),
volume 1, pages 282{286, London, United Kingdom, September 18{21, 2000.
[5] Li Xinrong Li, Kaveh Pahlavan, Matti Latva-Aho, and Mika Ylianttila. Indoor geo-location
using OFDM signals in HIPERLAN/2 wireless LANs. In Proc. 11th IEEE International Sym-
posium on Personal, Indoor and Mobile Radio Communications (PIMRC 2000), volume 2,
pages 1449{1453, London, United Kingdom, September 18{21, 2000.
[6] Kent Rosengren, Per-Simon, Kildal, Jan Carlsson, and Olof Lunden. A new method to measure
radiation e ciency of terminal antennas. In Proc. 2000 IEEE-APS Conference on Antennas
and Propagation for Wireless Communications, pages 5{8, Waltham, Massachusetts, Novem-
ber 6{8, 2000.
[7] Science and Technology Division. Statement of Requirements for Public Safety Wireless
Communications and Interoperability. United States Department of Homeland Security,
1.0 edition, March 10, 2004. URL http://www.safecomprogram.gov/NR/rdonlyres/
3FFFBFBA-DC53-440E-B2EF-ABD391F13075/0/SAFECOM_Statement_of_Requirements_v1.
pdf.
[8] Wayne Rash. Mining for answers: Would RFID have helped in the utah mine tragedy? eWeek,
page 60, September 10, 2007. URL http://www.eweek-digital.com/eweek/20070910_stnd/
?pg=61.
[9] CAP Operations. Supporting our communities in times of need, Year unknown; accessed 2008.
URL http://www.cap.gov/about/PROGRAMS/OPS.html.
[10] Fred L. Strickland, Yu Wang, and Alvin S. Lim. HEAPINGS: A secure counterintuitive
sensor network protocol. In Hamid R. Arabnia, Victor A. Clincy, and Laurence T. Wang,
editors, Proc. 2007 International Conference on Wireless Netwoks (ICWN ?07), pages 474{
480, Las Vegas, Nevada, June 25{28, 2007. World Congress in Computer Science, Computer
Engineering, and Applied Computing (WORLDCOMP ?07).
[11] Fred L. Strickland and Yu Wang. HEAPINGS: From concept to simulations. In Hamid R.
Arabnia and Victor A. Clincy, editors, Proc. 2008 International Conference on Wireless Net-
woks (ICWN ?08), pages 590{596, Las Vegas, Nevada, July 14{17, 2008. World Congress in
Computer Science, Computer Engineering, and Applied Computing (WORLDCOMP ?08).
[12] Satish Jamadagni and M. N. Umesh. A PUSH download architecture for software de ned
radios. In Proc. 2000 IEEE International Conference on Personal Wireless Communications,
75
pages 404{407, Hyderabad, India, December 17{20, 2000.
[13] R. Vishwakarma and K. S. Shanmugan. Performance analysis of transmit antenna diversity
in 3G WCDMA system. In Proc. 2000 IEEE International Conference on Personal Wireless
Communications, pages 1{4, Hyderabad, India, December 17{20, 2000.
[14] Ashvin Chheda. On the forward link capacity of a wide band DS-CDMA system with transmit
diversity. In Proc. 2000 IEEE International Conference on Personal Wireless Communications,
pages 417{421, Hyderabad, India, December 17{20, 2000.
[15] Morten Jeppesen, Soren Holdt Jensen, and Gert Frolund Pedersen. Prediction based multi-
antenna single-receiver mobile terminal. In Proc. 12th IEEE International Symposium on
Personal, Indoor and Mobile Radio Communications (PIMRC 2001), pages A120{A124, San
Diego, California, September 30, /October 3, 2001.
[16] Mostafa Hefnawi and Giles Y. Delisle. Impact of wideband CDMA signals on smart antenna
systems. In Proc. 2000 IEEE International Conference on Personal Wireless Communications,
pages 5{8, Hyderabad, India, December 17{20, 2000.
[17] Andreas Czylwik. Downlink beamforming for mobile radio systems with frequency division
duplex. In Proc. 11th IEEE International Symposium on Personal, Indoor and Mobile Radio
Communications (PIMRC 2000), volume 1, pages 72{76, London, United Kingdom, September
18{21, 2000.
[18] Kapil R. Dandekar, Hao Ling, and Guangham Xu. Smart antenna array calibration procedure
including amplitude and phase mismatch and mutual coupling e ects. In Proc. 2000 IEEE
International Conference on Personal Wireless Communications, pages 293{297, Hyderabad,
India, December 17{20, 2000.
[19] Khalid AlMidfa, Eustace Tameh, and Andrew Nix. Improved DOA estimation using po-
larisation diversity: Simulations using a wideband propagation model. In Proc. 11th IEEE
International Symposium on Personal, Indoor and Mobile Radio Communications (PIMRC
2000), volume 1, pages 539{543, London, United Kingdom, September 18{21, 2000.
[20] R. Tingley and K. Pahlavan. A statistical model of space-time radio propagation in indoor
environments. In Proc. 2000 IEEE-APS Conference on Antennas and Propagation for Wireless
Communications, pages 61{64, Waltham, Massachusetts, November 6{8, 2000.
[21] Marshall D. Brain. How radar works, Year unknown; accessed 2008. URL http://www.
howstuffworks.com/radar2.htm.
[22] V. A. Kryachko. The variant of equation of direction  nding characteristic of the amplitude
sum-di erence direction  nder. In Proc. IVth International Conference on Antenna Theory
and Techniques, volume I, pages 376{377, Sevastopol, Ukraine, September 9{12, 2003. IEEE.
[23] S. T. Bagdasaryan, V. A. Tarshin, and V. A.Vasilyev. Measurements of reception direction
of deterministic signal against spatially-correlated interferences with the use of the adaptive
antenna array. In Proc. IVth International Conference on Antenna Theory and Techniques,
volume I, pages 381{383, Sevastopol, Ukraine, September 9{12, 2003. IEEE.
[24] S. V. Kukobko, A. Z. Sazonov, and O. I. Sukharevsky. Calculation of nose dielectric radome ef-
fect on direction- nding characteristics of an antenna system. In Proc. IVth International Con-
ference on Antenna Theory and Techniques, volume II, pages 657{660, Sevastopol, Ukraine,
September 9{12, 2003. IEEE.
[25] A. Yu. Grinev, I. A. Chebakov, and A. I. Gigolo. Solution of the inverse problems of subsurface
radiolocation. In Proc. IVth International Conference on Antenna Theory and Techniques,
volume II, pages 523{526, Sevastopol, Ukraine, September 9{12, 2003. IEEE.
[26] PBS-TV. Hunt for the super twister, March 30, 2005. URL http://www.pbs.org/wgbh/
nova/transcript/3107_tornado.html. This was a broadcast. Transcript is at web address.
[27] Vadim L. Pazynin and Yuriy K. Sirenko. Study of pulse signal propagation in urban envi-
76
ronment by the  nite-di erence method. In Proc. IVth International Conference on Antenna
Theory and Techniques, volume II, pages 705{707, Sevastopol, Ukraine, September 9{12, 2003.
IEEE.
[28] D. Davis, B. Segal, C. W. Trueman, R. Calzadilla, and T. Pavlasck. Measurement of indoor
propagation at 850 MHz and 1.9 GHz in hospital corridors. In Proc. 2000 IEEE-APS Con-
ference on Antennas and Propagation for Wireless Communications, pages 77{70, Waltham,
Massachusetts, November 6{8, 2000.
[29] C. W. Trueman, D. Davis, and B. Segal. Ray optical simulation of indoor corridor propagation
at 850 and 1900 MHz. In Proc. 2000 IEEE-APS Conference on Antennas and Propagation for
Wireless Communications, pages 81{86, Waltham, Massachusetts, November 6{8, 2000.
[30] R. W. Matthews, W. G. Scott, and C. C. Han. Multibeam antennas for data communications
satellites. In Proc. 5th Symposium on Data Communications, volume 2, pages 20{29, Snowbird,
Utah, September 27{29, 1977. ACM.
[31] K. Pahlava. Practice and experience: Wireless intrao ce networks. ACM Transactions on
O ce Information Systems, 6(3):277{302, July 1988.
[32] Z. Rosberg and J. Zande. Toward a framework for power control in cellular systems. Wireless
Networks, 4(3):215{222, March 1998.
[33] P. van Rooyen, R. Kohno, and I. Oppermann. DS-CDMA performance with maximum ratio
combining and antenna arrays. Wireless Networks, 4(6):479{488, November 1998.
[34] Evan Gabber and Avishai Wool. How to prove where you are: Tracking the location of customer
equipment. In Proc. 5th ACM Conference on Computer and Communications Security, pages
142{149, San Francisco, California, November 3{5, 1998.
[35] N. B. Pronios. Performance considerations for slotted spread-spectrum random-access networks
with directional antennas. In Proc. IEEE GLOBECOM?89, volume 3, pages 1613{1617, Dallas,
Texas, November 27{30, 1989.
[36] J. Ward and Jr R. T. Compton. Improving the performance of a slotted ALOHA packet
radio network with an adaptive array. IEEE Transactions on Communications, 40(2):292{
300, February 1992.
[37] J. Ward and Jr R. T. Compton. High throughput slotted ALOHA packet radio network with
an adaptive array. IEEE Transactions on Communications, 41(3):460{470, March 1993.
[38] A. Sugihara, K. Enomoto, and I. Sasase. Throughput performance of a slotted nonpersistent
CSMA with an adaptive array. In Proc. 6th IEEE International Symposium on Personal,
Indoor and Mobile Radio Communications (PIMRC ?95), volume 2, pages 75{80, Toronto,
Canada, September 27{29, 1995.
[39] F. Shad, T. D. Todd, V. Kezys, and J. Litva. Indoor SDMA capacity using a smart antenna
basestation. In Proc. 6th International Conference on Universal Personal Communications
(ICUPC 97), pages 868{872, San Diego, California, October 12{16, 1997. IEEE.
[40] F. Shad, T. D. Todd, V. Kezys, and J. Litva. Dynamic slot allocation (DSA) in indoor
SDMA/TDMA using a smart antenna basestation. IEEE/ACM Transactions on Networking,
9(1):69{81, February 2001.
[41] R. Sass. Communications networks for the Force XXI digitized battle eld. Mobile Networks
and Applications, 4(3):139{155, October 1999.
[42] J. Razavilar, F. Rashid-Farrokhi, and K. J. R. Liu. Software radio architecture with smart
antennas: A tutorial on algorithms and complexity. IEEE Journal on Selected Areas in Com-
munications, 17(4):662{676, April 1999.
[43] Fernado C. Gomes, Panos Pardalos, Carlos. S. Oliverira, and Mauricio G. C. Resende. Reac-
tive GRASP with path relinking for channel assignment in mobile phone networks. In Proc.
5th International Workshop on Discrete Algorithms and Methods for Mobile Computing and
77
Communications, pages 60{67, Rome, Italy, July 21, 2001. ACM.
[44] K. Kobayashi and M. Nakagaw. Spatially divided channel scheme using sectored antennas
for CSMA/CA - ?Directional CSMA/CA?. In Proc. 11th IEEE International Symposium on
Personal, Indoor and Mobile Radio Communications (PIMRC 2000), volume 1, pages 227{231,
London, United Kingdom, September 18{21, 2000.
[45] S. Yi, Y. Pei, and S. Kalyanaraman. On the capacity improvement of ad hoc wireless net-
works using directional antennas. In Proc. ACM International Symposium on Mobile Ad Hoc
Networking and Computing (MOBIHOC ?03), pages 108{116, Annapolis, Maryland, June 1{3,
2003.
[46] J. Cheng, M. Hashiguchi, K. Ligusa, and T. Ohira. Electronically steerable parasitic array
radiator antenna for omni- and sector pattern forming applications to wireless ad hoc networks.
IEE Proceedings Microwaves, Antennas and Propagation, CL(4):203{208, August 2003. Has
become IET Microwaves, Antennas & Propagation.
[47] R. Ramanathan. On the performance of ad hoc networks with beamforming antennas. In Proc.
2nd International ACM Symposium on Mobile Ad Hoc Networking and Computing (MOBI-
HOC 2001), pages 95{105, Long Beach, California, October 4{5, 2001.
[48] T. Korakis, G. Jakllari, and L. Tassiulas. A MAC protocol for full exploitation of directional
antennas in ad-hoc wireless networks. In Proc. ACM International Symposium on Mobile Ad
Hoc Networking and Computing (MOBIHOC ?03), pages 98{107, Annapolis, Maryland, June
1{3, 2003.
[49] M. Takai, J. Martin, A. Ren, and R. Bogrodia. Directional virtual carrier sensing for directional
antennas in mobile ad hoc networks. In Proc. ACM International Symposium on Mobile Ad
Hoc Networking and Computing (MOBIHOC ?02), pages 183{193, Lausanne, Switzerland,
June 9{11, 2002.
[50] Y-B. Ko and N H. Vaidya. Location-aided routing (LAR) in mobile ad hoc networks. Wireless
Networks, 6(4):307{321, 2000.
[51] A. Nasipuri and K. Li. A directionality based location discovery scheme for wireless sensor net-
works. In Proc. Eighth Annual International Conference on Mobile Computing and Networking
(MOBICOM ?02), pages 105{111, Atlanta, Georgia, September 23{28, 2002. ACM.
[52] T-S. Yum and K-W. Hung. Design algorithms for multihop packet radio networks with multi-
ple directional antennas stations. IEEE Transactions on Communications, 40(11):1716{1724,
November 1992.
[53] Martin Horne er and Dieter Plassmann. Directed antennas in the mobile broadband system.
In Proc. 15th Annual Joint Conference of the IEEE Computer Societies (IEEE INFOCOM
?96), volume 2, pages 704{712, San Francisco, California, March 24{28, 1996.
[54] C. Sakr and T. D. Todd. Carrier-sense protocols for packet-switched smart antenna basesta-
tions. In Proc. IEEE International Conference of Network Protocols, pages 45{52, Atlanta,
Georgia, October 28{31, 1997.
[55] Young-Bae Ko, Vinaychandra Shankarkumar, and Nitin H. Vaidya. Medium access control
protocols using directional antennas in ad hoc networks. In Proc. IEEE INFOCOM 2000,
volume 1, pages 13{21, Tel Aviv, Israel, March 26{30, 2000.
[56] Romit Roy Choudhury, Xue Yang, Ram Ramanathan, and Nitin H. Vaidya. Using directional
antennas for medium access control in ad hoc networks. In Proc. 8th Annual International
Conference on Mobile Computing and Networking (MOBICOM ?02), pages 59{70, Atlanta,
Georgia, September 23{28, 2002. ACM.
[57] Nitin Vaidya, Year unknown; accessed 2008. URL http://www.crhc.uiuc.edu/wireless/
talks/purdue.ppt. Versions of this talk presented at University of Pennsylvania (July 2002)
and Purdue University (August 2002). Slides used for a presentation at the 2002 CCW work-
78
shop, Santa Fe (there are more slides here than included in the workshop proceedings).
[58] A. Nasipuri, J. You S. Ye, and R. E. Hironmoto. A MAC protocol for mobile ad hoc networks
using directional antennas. In Proc. IEEE WCNC 2000, volume 3, pages 1214{1219, Chicago,
Illinois, September 23{28, 2000.
[59] A. Nasipuri, J. Mandava, H. Manchala, and R. E. Hiromoto. On-demand routing using direc-
tional antennas in mobile ad hoc networks. In Proc. 9th International Conference on Com-
puter Communications and Networks, pages 535{541, Las Vegas, Nevada, October 16{18, 2000.
IEEE.
[60] S. Bandyopadhyay, K. Hasuike, S. Horisawa, and S. Tawara. An adaptive MAC protocol for
wireless ad hoc community network (WACNet) using electronically steerable passive array
radiator antenna. In Proc. GLOBECOM 01: IEEE Global Telecommunications Conference
(GLOBECOM 2001), volume V, pages 2896{2900, San Antonio, Texas, November 25{29, 2001.
[61] S. Bandyopadhyay, K. Hasuike, S. Horisawa, and S. Tawara. An adaptive MAC and directional
routing protocol for ad hoc wireless network using ESPAR antenna. In Proc. 2nd International
ACM Symposium on Mobile Ad Hoc Networking and Computing (MOBIHOC 2001), pages
243{246, Long Beach, California, October 4{5, 2001.
[62] A. Nasipuri, K. Li, and U. R. Sappidi. Power consumption and throughput in mobile ad hoc
networks using directional antennas. In Proc. 11th International Conference on Computer
Communications and Networks, pages 620{626, Miami, Florida, October 14{16, 2002. IEEE.
[63] Lichun Bao and J.J. Garcia-Luna-Aceves. Transmission scheduling in ad hoc networks with
directional antennas. In Proc. 8th Annual International Conference on Mobile Computing and
Networking (MOBICOM ?02), pages 48{58, Atlanta, Georgia, September 23{28, 2002. ACM.
[64] S. Roy, D. Saha, S. Bandyopadhyay, T. Ueda, and S. Tanaka. A network-aware MAC and
routing protocol for e ective load balancing in ad hoc wireless networks with directional anten-
nas. In Proc. ACM International Symposium on Mobile Ad Hoc Networking and Computing
(MOBIHOC ?03), pages 88{97, Annapolis, Maryland, June 1{3, 2003.
[65] Bruno Sinopoli, Courtney Sharp, Luca Schenato Shawn Scha ert, and Shankar Sastry. Dis-
tributed control applications within sensor networks, August 2003. URL http://robotics.
eecs.berkeley.edu/~sinopoli/proceedings.pdf.
[66] Chee-Yee Chong and Srikanta P. Kumar. Sensor networks: Evolution, opportunities, and
challenges. Proceedings of the IEEE, 91(8):1247{1256, August 2003. URL http://www.cs.
utah.edu/classes/cs6935/papers/sensNet1.pdf.
[67] Wendi Rabiner Heinzelman, Joanna Kulik, and Hari Balakrishnan. Adaptive protocols for in-
formation dissemination in wireless sensor networks. In Proc. 5th annual ACM/IEEE Interna-
tional Conference on Mobile Computing and Networking, pages 174{185, Seattle, Washington,
August 15{19, 1999.
[68] Chalermek Intanagonwiwat, Ramesh Govindan, and Deborah Estrin. Directed di usion: a
scalable and robust communication paradigm for sensor networks. In Proc. 6th Annual Interna-
tional Conference on Mobile Computing and Networking, pages 56{67, Boston, Massachusetts,
August 6{11, 2000. ACM.
[69] Alvin Lim. Distributed services for information dissemination in self-organizing sensor net-
works, 2001. URL http://www.eng.auburn.edu/users/lim/sensit/service.pdf.
[70] Carme  Alvarez, Josep Diaz, Jordi Petit, Jos e Rolim, and Maria Serna. E cient and reliable
high levelcommunications in randomly deployed wireless sensor networks. In Proc. Second
International Workshop on Mobility Management and Wireless Access Protocols (MobiWac
?04), pages 106{110, Philadelphia, Pennsylvania, October 1, 2004. ACM. URL http://www.
ing.unipi.it/~o783499/research/docs/MSWiM04.pdf.
[71] Matt Welsh and Geo Mainland. Programming sensor networks using abstract regions. the
79
 rst USENIX/ACM symposium on networked systems design and implementation (NSDI ?04),
March 2004. URL http://www.usenix.org/events/nsdi04/tech/full_papers/welsh/
welsh_html/nsdi.html.
[72] Michael Arnone. Cybersecurity bill gains steam. Federal Computer Week, XIX(12):12, April
2005.
[73] Department of Computer Science and Software Engineering. Graduate program handbook,
general regulations, 2.6 graduate course requirements, Year unknown; accessed 2008. URL
http://eng.auburn.edu/programs/csse/programs/grad/handbook/gen-reg.html.
[74] George Washington University. Studying infosec at gwu, Year unknown; accessed 2007. URL
http://www.seas.gwu.edu/~infosec/.
[75] Andrew. S. Tanenbaum. Computer Networks. Prentice Hall PTR, fourth edition, 2003.
[76] David B. Johnson. Scalable support for transparent mobile host internetworking. Wireless
Networks, 1:311{321, October 1999.
[77] G. Racherla, J. L. Ellis, D. S. Furuno, and S. C. Lin. Ultra-wideband systems for data communi-
cations. In Proc. 2002 IEEE International Conference on Personal Wireless Communications,
pages 129{133, New Delhi, India, December 15{17, 2002.
[78] Harris. An/PRC-152 type-1 multiband multimisson handheld radio, 2008. URL http://www.
rfcomm.harris.com/products/tactical-radio-communications/an-prc-152.pdf.
[79] Cengiz Alaettino^glu, Klaudia Dussa-Zierger, Ibrahim Matta, and A. Udaya Shankar. MaRS
(maryland routing simulator) - version 1.0 user?s manual. University of Maryland College Park
Technical Report, 91(80):1{36, June 1991.
[80] Paul A. Fishwick. Simulation Model Design and Execution: Building Digital Worlds. Prentice-
Hall, Inc, Englewood Cli s, New Jersey, 1995.
[81] The network simulator -ns-2, 1995. URL http://www.isi.edu/nsnam/ns/.
[82] Marc Greis. Tutorial for the network simulator ns, Year unknown; accessed 2008. URL
http://www.isi.edu/nsnam/ns/tutorial/.
[83] OPNET. OPNET modeler wireless suite, Year unknown; accessed 2008. URL http://www.
opent.com/solutions/network_rd/modeler_wireless.html.
[84] CrownCom 2008. The third international conference on cognitive radio oriented wireless net-
works and communications, 2008. URL http://www.crowncom.org/index.shtml.
[85] Auburn University Student Space Program. AubieSAT-1: Auburn?s  rst student-built satellite,
April 2007. URL http://space.auburn.edu/page_attachments/0000/0046/Spring_2008.
pdf.
[86] The American Radio Relay League. Radio wave propagation. In R. Dean Straw, editor, The
ARRL Antenna Book, chapter 23, pages 1{35. The American Radio Relay League, Newington
CT 06111, 1994.
[87] A. Medeisis and A. Kajackas. On the use of the universal Okumura-Hata propagation predic-
tion model in rural areas. In Proc. 51st Vehicular Technology Conference IEEE VTC 2000,
volume 3, pages 1815{1818, Tokyo, Japan, May 15{18, 2000.
[88] Scott Health and Safety. Pak-alert se questions and answers, Year unknown; accessed 2004.
URL http://www.scotthealthsafety.com/PDFs/QA_pkalert_6097_702.pdf.
[89] Mine Safety Appliances Company. ICMTM 2000 and ICMTM 2000 plus inte-
grated computer modules, Year unknown; accessed 2004. URL http://media.
msanet.com/NA/UnitedStatesofAmerica/SAR/SelfContainedBreathingApparatus/
SCBAPartsandAccessories/ICM2000andICM2000Plus/0119-27-ICM2000.pdf. Also ex-
plore the website for other documents such as 10018725.pdf.
[90] Jean MacDougall-Tattan. Fire ghters safer thanks to new radios, March 24, 2003. URL
http://www.eagletribune.com/news/stories/20030324/HA_002.htm.
80
[91] Communications-Applied Technology. Intrinsically safe wireless intercom system, Year un-
known; accessed 2004 and 2008. URL http://www.c-at.com/ispages/is.html.
[92] On line Press Release. New kenwood trunking mobiles feature two-way paging  eetsync(tm)
paging protocol extends capability of TK-980/981 mobiles, Year unknown; accessed 2004 and
2008. URL http://www.kenwood.net/indexKenwood.cfm?do=PRDetail&PRID=10.
[93] Sierra Wireless. MP 775 GPS rugged wireless modem for global EDGE, GPRS and
GSM networks, Year unknown; accessed 2004.. URL http://www.sierrawireless.com/
ProductsOrdering/mp775.asp.
[94] Peter H. Dana. Global positioning system overview, Year unknown; accessed 2004 and 2008.
URL http://www.colorado.edu/geography/gcraft/notes//gps/gps_f.html.
[95] Oak Ridge National Laboratory. Sensor-based tagging and tracking system: Next-
generation time, space, and position information system, Year unknown; accessed
2004. URL http://www.ornl.gov/sci/engineering_science_technology/sms/Hardy%
20Fact%20Sheets/Sensor-Based%20Tagging.pdf.
[96] National Executive Committee for Space-Based PNT. U. s. space based positioning, naviga-
tion, and timing policy, December 15, 2004. URL http://pnt.gov/policy/.
[97] Rosum Corporation, Year unknown; accessed 2008. URL http://www.rosum.com/rosum_
tv-gps_indoor_location_technology.html. Information is on several web pages. See other
pages.
[98] Skyhook Wireless, Year unknown; accessed 2004.. URL http://www.skyhookwireless.com.
[99] Carmen Nobel. Skyhook rolls out wi- system. eWeek, 22(25):21, June 20, 2005.
[100] James Careless. First responder communications: Alternatives to CDPD. Advanced Rescue
Technology, VII(4):71{78, August/September 2004.
[101] RFID Journal. Frequently asked questions, Year unknown; accessed 2004. URL http://www.
rfidjournal.com/article/articleview/207#Anchor-What-363.
[102] News Scan. Target shares r d plans. InformationWeek, page 14, August 16{23, 2004. There
was a version on the Internet, but it is no longer available.
[103] Renee Boucher Ferguson. R d mandates spur industry action. eWeek, page 17, July 26, 2004.
[104] Laurie Sullivan. IBM shares RFID lessons. InformationWeek, page 64, October 25, 2004.
[105] Bob Brewin. No silver bullets: Initial DoD RFID hype gets tempered by reality. Federal
Computer Week, XIX(13):39+, May 2005.
[106] Andy Ward, Paul Webster, and Peter Batty. Local positioning system { technology overview
and applications, September 2003. URL http://web.archive.org/web/20051024070310/
http:/ubisense.net/Product/files/Ubisense+LBS+overview+white+paper+September+
2003.pdf.
[107] SWS Security. Sws security?s tracknet networked radio direction  nder system, Year unknown;
accessed 2004. URL http://www.swssec.com/tracknet.html.
[108] United States Fire Administration. Fire ghter fatality retrospective study, April 2002. URL
http://www.usfa.fema.gov/downloads/pdf/publications/fa-220.pdf.
[109] NOAA. Emergency personal locator beacon system becomes operational nationwide, Year
unknown; accessed 2004.. URL http://www.noaanews.noaa.gov/stories/s1168.htm.
[110] NOAA. Homepage, Year unknown; accessed 2004.. URL http://www.noaanews.noaa.gov.
[111] NOAA. Personal locator beacons{help from above, Year unknown; accessed 2004.. URL
http://www.noaanews.noaa.gov/magazine/stories/mag96.htm.
[112] Princeton MN Automated Flight Service Station. Direction  nding (DF) service, Year un-
known; accessed 2005. URL http://www.pnmafss.faa.gov/pages/DF.asp.
[113] Federal Communications Commission. Enhanced 911, Year unknown; accessed 2008. URL
http://www.fcc.gov/pshs/services/911-services/enhanced911/Welcome.html. Ac-
81
cessed ?Enhanced 911? on http://www.fcc.gov/911/enhanced/ on 11 October 2004. It has
been replaced by ?Enhanced 911 - Wireless Services? on http://www.fcc.gov/pshs/services/911-
services/enhanced911/Welcome.html, which was accessed on 15 May 2008.
[114] Kathryn Manzi. Rescue 21 update. On Scene, pages 7{8, Summer 2004.
[115] Skip Liepman. C2 constellation. Military Information Technology, 8(6):10{13, 2004.
[116] Clarence A. Robinson Jr. Sensors bolster army prowess, Year unknown; accessed 2008. URL
http://www.afcea.org/signal/articles/anmviewer.asp?a=30&z=10.
[117] Robert S. Dudney. Where do UAVs go from here? Air Force Magazine, 88(7):2, July 2005.
URL http://www.afa.org/magazine/july2005/0705editorial.pdf.
[118] California Institute for Telecommunications and Information Technology. ECE 156/MAE
149/SIO 238 sensor networks, Year unknown; accessed 2005. URL http://www.
calit2.net/technology/features/8-7-03_SensornetClass.htm. Accessed 19 July 2005.
Clayton Okino was an instructor for this course. See Student paper by Cheng-yu
Sung, Chris Hiestand, and Shadi Ghandchi on Battle eld Sensors (1 August 2003.
http://www.calit2.net/technology/features/sensorNetClass/ece156-battle eld.pdf).
[119] David Culler, Jason Hill, Mike Horton, Kris Pister, Robert Szewczyk, and Alec Wood. Mica:
the commercialization of microsensor motes. Sensor Technology and Design, April 2002. URL
http://www.sensorsmag.com/articles/0402/40/main.shtml.
[120] Sarah Yang. Researchers create wireless sensor chip the size of glitter, June 2003. URL
http://www.berkeley.edu/news/media/releases/2003/06/04_sensor.shtml.
[121] G. Anastasi, A. Falchi, A. Passarella, M. Conti, and E. Gregori. Performance measurements
of motes sensor networks. In Proc. 7th ACM International Symposium on Modeling, Analysis
and Simulation of Wireless and Mobile Systems, pages 174{181, Venice, Italy, October 4{6,
2004.
82
APPENDICES
83
Appendix A: A LISTING OF WRITTEN PAPERS
Five papers were created as result of this work. These are:
1. HEAPINGS
 HEAPINGS: A Secure Counterintuitive Sensor Network Protocol
 HEAPINGS: From Concept to Simulations
2. Accepted by never published by the conference:
 HERBSUDAI: A Counterintuitive Resource Tracking Protocol
3. Paper submitted, but not accepted:
 HAP and HAPDA: Two Geolocation Approaches
4. Journal submission:
 Teaching Radio Propagation and Friis Equation by Using Real World Measurements
84
Appendix B: COMMERCIAL AND INFORMAL INFORMATION
.1 Introduction
This will explore what has been done in the past. Some things are still being done today. There
are some peeks into the future. Much of this information has come from non-peer reviewed sources.
Fire ghters are used as a typical example of a  rst respondor. What applies to them, could be
applied to other emergency service activities. Reconnaissance is another way of saying sensors.
.2 What has been done without radios?
.2.1 Fire ghters without radios
Fire departments have used a tag system. A  re ghter would give to the senior person on the scene
a tag that had some information written on it. This would show that certain individuals were on
the scene and in the  re, but did not show their exact location.
Later when air packs became standard  re ghting equipment, alarms were added. These would
sound when the person is not moving. The assumption is that a  re ghter would always be moving
and doing something. If the person is motionless, then the conclusion is that the person is in trouble.
One example is the Scott Health Safety Company?s Pak-Alert SE distress alarm, which consists
of a  ashing red light emitting diodes (better known as \LEDs") and twin audible sounders [88].
See Figure 1 (courtsey of Mine Safety Appliances Company). Another example is the Mine Safety
Appliances Company?s o ering. The product comes in two models: the ICM 2000 and the ICM
2000 Plus. Both models use two high-pitched tones followed by a buzz and a  ashing red light [89].
See Figure 2 (courtsey of Mine Safety Appliances Company).
Figure 1: Scott Health and Safety?s Pak-Alert SE
These products and others on the market go by various names such as \chirpers" or Personal
Alert Safety System or PASS. The problem is that none of these are designed to transmit the exact
location of a down  re ghter.
85
Figure 2: MSA?s ICM 2000 and ICM 2000 Plus
.2.2 Reconnaissance without radios
When a nation needed information about the activities of the other side, spies were used. These
might work inside a foreign government building. Or they might crawl on their stomachs to peer
over the hilltop in order to see what the foreign military is doing.
Both activities are risky. Some individuals have been captured or killed.
.3 What has been done with radios?
.3.1 Fire ghters with radios
Using simplex radios is an improvement over Personal Alert Safety System since two-way commu-
nications is possible. However, handheld units have limited power and limited range [90].
Using a duplex system or a repeater system overcame the limited power and the limited range
problems [90]. Of course, the burning building needs to be in the coverage area of a nearby  re ghter
repeater. Communications-Applied Technology?s (CAT) solution is to install a portable repeater (for
example, the Intrinsically Safe Wireless Intercom System) on a large  re truck [91]. However, no
position information is transmitted. Figure 3 (courtsey to Communications-Applied Technology)
shows the company?s Aircraft Wireless Intercom System repeater.
Figure 3: CAT?s Aircraft Wireless Intercom System Repeater
86
.3.2 Reconnaissance with radios
When a nation needed information about the activities of the other side, technicians were used. They
would plant electronic listening devices (\bugs") inside a foreign government building. Or agents
might crawl on their stomachs to peer over the hilltop in order to see what the foreign military is
doing and report back via radio.
Electronic devices and radio communications has improved the level of safety, but there is still
some risk and some individuals have been captured or killed.
There is another part of electronic reconnaissance that does not involve planting listening devices.
These are called sensors. This has become a major area of interest. This will be covered in greater
detail at the end of this appendix.
.4 What has been done with the Global Positioning System?
The Global Positioning System (GPS) has been considered a simple solution for determining a
location. There are various devices on the market that will provide location information. Some
models will provide elevation too.
This ability has been combined with various communications systems. Some cellular telephone
companies supply special services that use GPS.
Kenwood combined a radio with an automatic vehicle location (AVL) device, a GPS module, and
a messaging system to create a  eet management product. FleetSyncTM is a proprietary product
o ering that can enhance the dispatching of vehicles [92]. The Kenwood website does not elaborate
about the details of the position reporting feature.
Another example is the Sierra Wireless? MP 880 GPS or MP GP 881 (the MP 775 GPS is an
older version) [93]. These products provide a ruggedized modem for data communications. If the
network supports voice, then voice communication is available. The supporting functions are placed
inside a black box and this is mounted elsewhere in the vehicle such as in the trunk. The user and
the home station can read the GPS location information. The system operates in the sub bands
between 869 MHz and 1990 MHz. See Figure 4 (courtsey of Sierra Wireless).
These systems and others that use GPS su er from one serious weakness: If the devices? antennas
cannot see four GPS satellites, then time and position information will not be available. Even with
access to four GPS satellites, the accuracy is 100 meters horizontally and 156 meters vertically [94].
In layman?s terms, that means a person or an object could be anywhere within a 50-meter (164-foot)
radius of a spot and could be anywhere up or down 78 meters (256 feet). That is still a large search
area.
Oak Ridge National Laboratory and the US Army are working to develop the next generation
of devices whereby tracking is better and GPS information is used when available [95]. It goes by
the name of Next-Generation Time, Space, and Position Information System (TSPI). It is still too
early to know how this will be done and what will be the results.
In December 2004, President George W. Bush issued a Presidential Decision Directive that
mentioned the need to develop \terrestrial augmentations" for position reporting [96].
Even before President Bush?s directive, the Rosum Corporation was formed in 2000 in order to
address the short comings of GPS. Their solution uses television signals for determining positions
inside a building or beyond the view of GPS satellites [97]. The approach is that a Rosum device
detects as many signals as possible (analog television signals, digital television signals, and/or satel-
lite signals) and measures the time of arrivals. This data is sent to a Rosum location server. The
location server calculates the device?s location in two dimensions and passes this information back
87
Figure 4: Sierra Wireless MP 880W GPS and MP 881W GPS Black Box
to the device for display plus to any tracking application [97]. This can meet most of the time the
FCC?s E911 position requirements.
However, as wonderful as this technology may be, there are at least seven major drawbacks.
First, high power (1,000 kilowatts or more) television stations tend to be located within 50 miles (80
kilometers) of major population centers and solid coverage goes out to about 200 miles. That is, if
a person is in an area that has no television towers and no clear view of three GPs satellites, then
the Rosum system will not work. Second, each television signal needs to have a synchronization
code transmission. Third, there needs to be at least three television signals. In some markets, that
may not be the case. Fourth, monitor units must be deployed at  xed locations to analyze the
stability and timing of nearby television signals and provide the information to the location server.
Fifth, communication links must exist between the monitor units and the location server. Sixth,
the user must have a Rosum chip on their arm or embedded in a radio. Seventh, information is
expressed in terms of two dimensions. To obtain a 3D position requires the deployment of \pseudo
TV transmitters (PTTs)"around the target of interest [97].
With the Rosum?s TV-GPS Plus system, the incident commander can pinpoint a  re ghter?s
location to the \level of a room in the building." This is very good, but the website grosses over
a few things. First, someone must deploy the PTTs and that takes time. Second, Knowing where
to place the PTT is not address on the Rosum website. The previous illustration shows that three
PTT units are used, but in another example of an amusement park four PTT units are used. If the
spacing must be equal distance, then what happens if the distances are not equal? Third, the PTT
88
units broadcast on unused TV channels and this means that Rosum or the user must obtain a radio
license from the FCC [97]. If a new television station enters the market, then the PTT frequencies
might need to be changed and the new frequency cannot legally be used until a new FCC radio
license is obtained. Not mentioned on the Rosum?s website is the fact that the United States over
the next few years is realigning the television broadcast band. That means the PTT units might not
be legal to use after the \sunset" of the old broadcast bands. Finally, the FCC license a television
station to an area and if the PTT are licensed in a similar fashion, then would mean the PTT units
cannot be used outside of the home area of the  re unit.
Skyhook Wireless uses the large number of Wi-Fi access points as a positioning system [98]. It
works best in a dense urban area, but it is only accurate within 20 meters. The company?s website is
thin on details. In a general magazine, the reported noted that the system can determine a position
within 30 meters, plus or minus 10 meters for 25 of the most populated United States cities [99].
.4.1 Fire ghters with Global Positioning Systems
GPS devices are being combined with other systems in order to create products for  re ghters. For
example, the Aurora Colorado Fire Department uses General Packet Radio Service (GPRS) with
a GPS wireless modem and Motorola?s Advanced Vehicle Locating software in order to provide
 re ghting crews with changing maps that will guide them to the site and details about the burning
structure [100].
At least one  re department is using the previously mentioned Kenwood product. The Kenwood
website quotes an unidenti ed  re department spokesperson. The person discussed how this system
has improved communications. But there was no mention of the GPS positioning features [90].
Again  re ghters go inside buildings and that takes them out of view of any GPS satellites.
Hence, any GPS enabled devices would be useless.
.4.2 Reconnaissance with Global Positioning Systems
In the battle eld environment, the military has used pilotless devices to view the terrain. Some of
these use GPS as part of the navigation systems.
.5 What has been done with the Radio Frequency Identi cation?
Radio Frequency Identi cation (RFID) uses radio frequencies and a passive device or tag. A trans-
mitter sends out a signal that envelopes the tag and generates a  eld. The embedded microchip
circuits are energized and a return message is sent. This message may contain information such as
serial numbers, ownership, and shipping container data. Normally, the tag information does not
change. The RFID tags will only react to a certain frequency from one of the RFID radio bands
(125 kHz, 13.56 MHz, 850-900 MHz, or 2.45 GHz). The RFID bands have certain behaviors. Low
frequencies are better at penetrating non-metallic materials and are cheaper to manufacture, but
have low data transfer rates. Whereas UHF frequencies are better for distance and faster data trans-
fers, but have poor abilities for penetrating materials and must have a line-of-sight view. Another
drawback is the overall distance. At best, passive RFID tags have about a 20-foot range. If a battery
powered RFID model is used, then the range is increased anywhere from 100 to 300 feet [101].
This technology has been around since the 1970s [101]. It is only recently that the cost-bene t
equation has improved to the point that Target, Wal-Mart, the United States Department of Defense,
and others are requiring vendors to use RFID on shipments to their warehouses [102] [103]. The
tags are manufactured in di erent shapes such as buttons, \lip stick" shapes, credit card size, key
89
rings, and so on. See Figure 5 (courtesy of Texas Instruments) for an example of a button shaped
RFID tag.
Figure 5: RFID Tag (Texas Instruments 13.56 MHz encapsulated Transponder)
However, these are not without problems. IBM Global Services discovered that RFID tags are
not free from radio frequency interference (RFI). Forklifts, bug zappers, cellular telephone towers,
walkie-talkies, and other handheld devices can create RFI [104]. In a recent real world test, the DoD
discovered that RFID readers are able to read a pallet tag almost 100% of the time, but attempting
to read stacked cases dropped the success level to 80% [105].
.5.1 Fire ghters with the Radio Frequency Identi cation
Having a  re ghter wear a RFID tag is not a workable solution. The read-only tag information does
not change. In order to pass the information, there must be a transmitter nearby and on the correct
frequency. Some buildings were constructed with metallic materials and in such an environment the
building would function as a screen room{that is, signals cannot enter or leave. Other buildings have
thick walls. Even with a battery powered RFID tag, the best distant is about 300 feet. That distance
would be exceeded when the structure has several  oors and has a large base. Another drawback is
the storage size; it is about 2 kilobytes (kB). Tags can be read-write or read-only. Read-write tags
can only be updated when a radio signal is received. A  re ghter would not have a simple way of
providing revised data while  ghting a  re.
.5.2 Reconnaissance with the Radio Frequency Identi cation
RFID tags would not work for CAP since the aircraft  y patterns where the distances are greater
than 300 feet.
RFID tags would not work in a sensor environment since data requirements are greater than 2
kB. Also the base station must be able to reach all of the RFID tags with a strong signal. That
increases the chance of being detected by the other side. The  ip side problem is that the RFID
tags might not be able to generate a strong enough signal to reach the base station receive antenna.
Also the nature of RFID tends to eliminate the option of relaying messages.
.6 What has been done with Local Positioning Systems?
Placing sensors around an area such as a prison or a controlled area is possible. These are called
\Local Positioning Systems" (LPS) and these are independent of GPS[106]. Individuals wear a tag
and the system can track their movements. See Figure 6 (courtsey to Sidney Fels, Changsong Shen,
Baosheng Wang, and Steve Oldridge) for an example of the system that the Department of Electrical
and Computer Engineering at the University of British Columbia is developing. This is  ne for a
regular environment, but there are some limitations.
90
Figure 6: A Local Positioning System
One problem is that there are no standards. The result is that one company?s tags may not
be recognized by another company?s system. Having a unique product tends to encourage repeat
purchases and keeps a purchaser from trying to  nd a suitable replacement from another company.
But it works against having the ability to have a person tracked from one environment to another
environment. This market has not reached the point where the customers want to have a common
standard.
Another form of LPS is a large geographical area that uses radio direction  nding and tags. The
SWS Security?s TRACKnet Networked Radio Finder System is such an example [107]. This system
requires a large infrastructure: computers, special software, a local GPS receiver, dedicated data
lines between all the stations, and a local map. The \target" carries a simple beacon transmitter.
This system takes the bearings from the stations to the target and plots these on a map. The
intersection of the lines determines the location of the target. The information is in two dimensions.
Even if a city was completely covered, the system would only be able to narrow the location down
to a city block.
NexGen City (2002 start-up) attempted the same coverage by using an ad hoc networking ap-
proach. All the active laptops and various devices that are mounted on sides of buildings and on
street lights are used to achieve city-wide coverage [100]. Each unit functions as a router and as a
repeater. So far it appears that the only deployed, real world system is in Garland Texas.1
.6.1 Fire ghters with Local Positioning Systems
The problem of no common standard means that a  re ghter with a single LPS tag would not be
tracked from one system to another system. If a  re ghter wanted to be tracked, then that would
mean carrying tags for each known LPS in the  re district.
There are three problems with the LPS approach.
The  rst problem is that a  re ghter would have to take time to select the correct tag. And time
is not the  re ghter?s friend. If the  re was huge and several buildings with di erence systems were
burning, then tracking would end upon entry into the other system?s coverage.
The other problem is that  re ghters carry many pounds (kilograms) of equipment with the
result that overexertion has caused a number of deaths. Fire ghters have a good reason for not
1In an e ort to con rm information and obtain permissions to use images, it appears this company had some
problems such as law suits and unhappy public. The website does not exist anymore.
91
wanting to carry another device. In the United States Fire Administration?s report, Fire ghter
Fatality Retrospective Study, it was found that the leading cause of death on the job is heart attack
at 44 percent with trauma a distanced second at 27 percent[108]. Fire ghters may not be willing
to add another bit of weight even if it is a few ounces (grams), because of the increased stress of
carrying the extra weight.
Still another problem is that LPS requires external power and during a  re that may not be
present. The  re may interrupt the power  ow or the power company may turn it o .
Darrell McClanahan, Garland telecommunications manager, wants to be able to collect a  re-
 ghter?s life signs via NexGen City system. But this is not likely to happen. The company admits
that their equipment has a location error of plus or minus 10 meters. The  re ghters would have
to wear something that would collect the life signs data and transmit it to a nearby laptop com-
puter. This will run headlong into the problem that  re ghters are not near a laptop computer
while  ghting a  re. So the extra weight would be unwelcomed.
.6.2 Reconnaissance with Local Positioning Systems
It is unlikely that AFAUX/CAP aircraft or AFAUX/CAP ground teams or agents would be operating
inside a \wired" environment.
.7 What has been done with Locator Transmitters and Beacons?
For years, aircraft have carried emergency locator transmitters (ELT) that operated on 121.5 MHz
and on 243 MHz. See Figure 7 (courtesy of Artex Aircraft Supplies) for an example of a small
ELT. When an ELT is transmitting, certain satellites will receive the signal and relay to the nearest
rescue coordination center. In the United States, this is the Air Force Rescue Coordination Center
on Langley Air Force Base, Virginia. The center calls HQ AFAUX/CAP National Operations Center
(NOC) and in turn the NOC dispatches teams to the general search area. The teams use trackers
that work when within line-of-sight distances (roughly 50 miles (80 kilometers) or less) of the target.
On 1 July 2003, the Personal Locator Beacon (PLB) System became operational [109]. A person
would purchase a PLB and would register with the National Oceanic and Atmospheric Adminis-
tration (NOAA) [110]. When the PLB is turned on, participating satellites2 hear the signal and
relay it to the nearest rescue coordination center. See Figure 8 (courtesy U.S. National Oceanic and
Atmospheric Administration (NOAA)).
.7.1 Fire ghters with Locator Transmitters and Beacons
Could this technology have helped to locate down  re ghters in the rubble of 9-11? It is true that
after the grounding of all aircraft in the United States, the  rst aircraft permitted back into the air
were AFAUX/CAP aircraft3. Even if all the  re ghters had ELTs, AFAUX/CAP still could not  x
a precise position on down  re ghters, because ELTs transmit a simple signal and do not provide
location information.
PLBs have saved many outdoorsmen, but they cannot help  re ghters. At best, the standard
406 MHz PLB can locate a person within a two-mile (3-kilometer) radius and the top-of-line PLBs
2Some satellites carry receivers for both the ELT and PLB. As the ELT receivers fail, there is no plan to repair or
replace them. The day will come when all ELT tracking will be done by ground based stations and any chance air
borne platforms.
3This is the source of the  rst images of the World Trade Center site.
92
Figure 7: Artex ELT 200 operates on 121.5 and 243 MHz
have GPS receivers that can reduce the search radius to less than 400 feet (122 meters) [111]. Of
course, if the satellites cannot hear the signal, then this enhancement is useless.
.7.2 Reconnaissance with Locator Transmitters and Beacons
It is unlikely that agents would be using these devices for collecting information. They wish to avoid
detection. The purpose of ELTs and PLBs is the opposite.
.8 What has been done with Radar?
How are aircraft tracked? Air tra c controllers use radar with an identi cation friend or foe (better
known as IFF or IFF/SIF) module. The radar locates the aircraft and the IFF transmits a query.
The aircraft?s IFF unit replies with a small data message. The sector air tra c controller will see
displayed on the monitor the aircraft as a blip and a  oating box. See Figure 9 (courtsey to NASA).
Air tra c controllers follow the movement of the blip in order to follow the progress of the aircraft.
A dialogue transpire between the air tra c controller and the pilot. Instructions are passed and the
pilot reports passing certain landmarks or navigational aid system (NAVAIDS).
93
Figure 8: COSPAS-SARSAT System Overview
In addition to the foregoing, a pilot uses NAVAIDS such as a VHF Omnidirectional Range
(better known as a VOR) navigation system, Tactical Air Navigation System (better known as
TACANS), and navigation beacons in order to stay on course. If a pilot is disoriented and if
the air tra c controller cannot see the aircraft on the radar monitor, then the pilot can contact a
nearby Automated Flight Service Station (AFSS) Radio and request Orientation Service. The Flight
Service Controller would collect key pieces of information and uses the station?s direction  nding
(DF) equipment to determine the bearing of the aircraft from the DF site. A secondary means is to
use several devices (another DF site, a VOR, Automatic Direction Finder (better known as ADF)
to triangulate the aircraft?s position and provide a new heading to the pilot. This is not a priority
service and may be not be available if the AFSS workload is too heavy [112].
.8.1 Fire ghters with Radar
Radar requires large targets in order to obtain a position. Fire ghters are smaller than an aircraft
and would tend to be inside a burning structure.
94
Figure 9: View of numerous aircraft on a radar scope.
.8.2 Reconnaissance with Radar
It is unlikely that agents would be using these devices for collecting information. Radars use high
power transmitters and large receiving antennas.
.9 What has been done with Enhanced 911 Service?
What about the new Federal Communications Commission (FCC) requirement that cellular tele-
phone providers must provide location information with 911 calls4? The Phase I requirement is
to locate callers by cellular radio towers and Phase II requirement is to locate the caller within a
50-meter (164 feet) radius in the best case [113].
.9.1 Fire ghters with Enhanced 911 Service
The location information is not precise enough and  re ghters would not have time to use a cellular
phone to report their position. There is another drawback. Cellular phone service may not be
available. Land lines to the towers could be broken. All the circuits could be busy. Or as have
happen during nature disasters, the cellular phone system could be o line. Finally, cellular phone
service is non-existent or spotty in rural areas. If the  re was near the boundary of two or more
telephone area codes and if a cellular telephone tower from a di erent area code was the closest,
then any dialed telephone number would need to have the area code included.
.9.2 Reconnaissance with Enhanced 911 Service
It is unlikely that agents would be using cellular telephones for collecting information.
4Network-based technologies must be able to locate a cell phone unit within 100 meters 67% of the time and within
300 meters 95% of the time. Handset-based technologies must be able to locate a cell phone unit within 50 meters
67% of the time and within 150 meters 95% of the time.
95
.10 What has been done to support the uniformed services?
Rescue 21 (R21) is the United States Coast Guard?s (USCG) program for upgrading its approach
to saving lives. One of R21 core features is the ability to track platforms (ships and aircraft). In
order to do this, it uses  xed sites and each platform carries a Rescue 21 transmitting package [114].
Figure 10 (courtsey to the United States Coast Guard) is the logo for this program5
Figure 10: US Coast Guard Rescue 21 Advanced Communications System (Logo)
The United States military has had an abiding interest in direction  nding for years. It could
be used to locate an o ending enemy station. It could be used to help an aircraft to determine its
location over the ocean. And of course, the great interest in  nding and recovering downed pilots.
The military has various nets and systems that are devoted to communications, computing, and
control. One example is the Command and Control Constellation, which is built on the Constel-
lationNet network. This consists of sensors, platforms, and command centers [115]. This supports
the Air Force requirements for command and control, intelligence, surveillance, and reconnaissance.
It is not intended to support civilian needs and it is designed for working with aircraft, not with
individuals on the ground or in a building.
.10.1 Fire ghters with support from uniformed services systems
USCG  re ghting and rescues tend to be on watercraft and a number of these have small \real
estate." Location information is not as big as a concern as it is for land based  re ghters. Hence,
the issues and the approaches are di erent from land based  res.
.10.2 Reconnaissance with support from uniformed services systems
It is unlikely that agents could use the USCG systems or the military nets for collecting information.
Again, these systems are intended to work with aircraft, not with individuals on the ground or in a
building.
5This was capture early in this research. When I contacted the USGS for permission, they could not  nd this
image anymore. They have a blanket policy of granting permission for all images as long as credit is given.
96
.11 What has been done with the Sensors?
Sensors are popular, because these can be deployed without putting a person in danger. However,
as wonderful and as successful current battle eld sensors are, there are still some problems. Edward
T. Blair, the United States Army?s program executive o cer for intelligence, electronic warfare and
sensors (PEO IEW&S) (Fort Monmouth, NJ)[116] stated that one major weakness is that the sensors
cannot track much more than 10 targets at the same time. Taking a big picture view, Gen John
P. Jumper, the Air Force Chief of Sta , pointed out that every military activity wants it own UAV
system with the resulting problems in safety and radio frequency con icts[117]. That is, sensors
need to have a collision-free channel.
Some undergraduate students in a University of California San Diego course (ECE 156 Sensor
Networks taught by Clayton Okino of the Jet Propulsion Laboratory) determined that sensor density
is another problem area[118]. Chee-Yee Chong and Srikanta P. Kumar noted some more problems
such as network discovery, control and routing, collaborative signal and information processing,
tasking and querying, and security[66].
Here is a list of these and other problem areas:
 Sensors cannot track a large number of targets at the same time.
 Energy shortfall.
 Availability of air platforms such as the OAVs and UAVs.
 UAVs loiter time over an area is limited.
 Returning data to home station is a problem. (UAVs are used sometimes.)
 UAVs cannot carry heavy sensors.
 Frequency coordination is not being done.
 UAVs can be detected by the other side.
 Sensor density is a problem.
 Network discovery.
 Control and routing.
 Collaborative signal and information processing.
 Tasking and querying.
 Security.
If the Crossbow?s MICA product is a typical sensor device, then many MICA sensor nodes are
designed to operate on 916 MHz[119]. That will result in congestion on this frequency and poor
distance coverage.
Berkeley Motes are even smaller than Crossbow?s MICA product and these Motes operate in the
same radio band. For example, the 5-millimeter square \smart dust" chip (see Figure 11 (courtsey
of Jason Hill and David E. Culler at University of California, Berkeley)) operates on 902 MHz[120].
Anastais et al. examined the MICA2 and the MICA2DOT Berkeley Motes and noticed that rain and
fog will degrade the signal[121]. For Anastais et al. , the factors that impact transmission range are
97
transmitter power, receiver sensitivity, the antenna gain, and data transmission rate. They did not
consider that the actual frequency used has the greatest impact. It is a known fact among radio
engineers that weather has a great impact on short wave length (that is, microwave) emissions.
Figure 11: \Smart Dust" Chip
98