UBRARY
Michigan State
University

 

 

 

PLACE IN RETURN Box to remove this checkout from your record.
TO AVOID FINES return on or before date due.
MAY BE RECALLED with earlier due date if requested.

 

DATE DUE DATE DUE DATE DUE

 

3:92 19% 2005

O
1

 

Mm‘wmm.

 

MAR 0 5 2008

 

511167

 

 

 

 

 

 

 

 

 

 

 

6/01 cleIRC/DatoDuopss-p. 1 5

A SELF-STRUCTURING ANTENNA PROTOTYPE
By

Bradley Thomas Perry

A THESIS

Submitted to
Michigan State University
in partial fulfillment of the requirements
for the degree of

MASTER OF SCIENCE
Electrical and Computer Engineering

2002

ABSTRACT

A SELF-STRUCTURING ANTENNA PROTOTYPE
By

Bradley Thomas Perry

This thesis presents a class of antennas, called self-structuring antennas, from
an experimental perspective. The self-structuring antenna (SSA) is a concept that
involves creating a wideband antenna that can automatically adjust its electrical
shape in response to changes in its electromagnetic environment. This is done through
the use of simple on / off switches employing binary search algorithms to find optimal
configurations for the antenna.

Topics covered include a general overview of the self-structuring antenna to famil-
iarize the reader with the concepts. A brief history of the SSA and some practical
uses are also covered. Next, a prototype SSA is discussed. This includes the building
and testing of the antenna, including the application of a genetic algorithm (GA) to
the control of the prototype. Testing of this prototype for UHF and VHF television
reception follows, providing a proof of concept for both the SSA and GA.

Once the general framework of the self-structuring antenna has been laid down
through the above work, consideration is given to the occurrence of switch malfunc-
tions. Switch failures and their consequences are considered for single switches, as well

as groups of switches. Finally, conclusions drawn from this research are discussed.

For My Family

iii

ACKNOWLEDGMENTS

Throughout the course of this research, there have been many people who have
helped me to keep my sanity. Thanks to everyone who helped me out along the way.

First, I would like to thank my professors who helped me to find my way through
graduate work. Thanks to Dr. Edward Rothwell for being my major professor and
teacher, to Dr. Lee Kempel for helping me to decide to keep on the educational track,
and Dr. Dennis Nyquist for the instruction and example.

Thanks to friend and mentor, Dr Mike Havrilla, for taking time to teach me things
that I would not have learned otherwise. Your dedication to EM is nothing short of
admirable.

I would also like to thank friend and fellow graduate student Tony Lambert for
his help in C++ programming for the switch failure analysis presented in this thesis.
Without your help I wouldn’t have been able to complete this analysis in such an
efficient manner.

Thanks is also in order for Dr. Christopher Coleman for help with ETEX and for
the style file that made my life easier while writing this thesis.

Many thanks to Professor Percy Pierre for supporting my studies through the
graduate office fellowship. Additional thanks are in order for Dr. Lou Nagy and
Dr. Mark Krage of Delphi Research Labs for their support of SSA research at MSU.

Most of all I would like to thank my wife Beckie and the rest of my family for their
support through all of my studies. Your constant recognition and encouragement is

greatly appreciated.

iv

TABLE OF CONTENTS

LIST OF TABLES ................................. vii

LIST OF FIGURES ................................ viii

KEY TO SYMBOLS AND ABBREVIATIONS ................. x
CHAPTER 1

Introduction ..................................... 1
CHAPTER 2

Self-Structuring Antenna Concepts and Ideas .................. 2

2.1 Origination of the Self-Structuring Antenna .............. 2

2.2 Overview of the Self-Structuring Antenna System ........... 3

2.2.1 Self-Structuring Antenna Skeleton ................ 4

2.2.2 Control of the Self-Structuring Antenna ............ 4

2.3 Practical Applications of Self-Structuring Antennas .......... 5
CHAPTER 3

30 Switch Prototype Construction ......................... 8

3.1 Layout of the Self-Structuring Antenna ................. 8

3.1.1 Antenna Board .......................... 9

3.1.2 Control Board ........................... 9

3.1.3 Production of the SSA Prototype ................ 10

3.2 Functionality Testing of the Self-Structuring Antenna ......... 10

3.2.1 Control Board Testing ...................... 10

3.2.2 Antenna Skeleton Testing .................... 11
CHAPTER 4

Genetic Algorithm for the Self-Structuring Antenna ............... 20

4.1 Genetic Algorithm Concepts ....................... 20

4.1.1 General Concepts ......................... 20

4.1.1.1 Initialization ...................... 21

4.1.1.2 Preselection ....................... 21

4.1.1.3 Selection ........................ 22

4.1.1.4 Crossover and Mutation ................ 22

4.1.2 Specialization to the Self-Structuring Antenna ......... 23

4.2 Television Measurements ......................... 24

4.2.1 Testing Environment ....................... 25

4.2.2 UHF Test Results ......................... 26

4.2.3 VHF Test Results ......................... 27

4.3 Conclusions ................................ 28
CHAPTER 5
Switch Failure Analysis .............................. 39
5.1 Testing Environment and Procedure ................... 39
5.2 Single Switch Failures ........................... 40
5.2.1 Minimal Degradation under Single Switch Failures ....... 41
5.2.2 Intermediate Degradation under Single Switch Failure ..... 42
5.2.3 Large Degradation under Single Switch Failure ......... 42
5.3 Multiple Switch Failures ......................... 43
5.3.1 Minimal Degradation under Multiple Switch Failures ..... 44
5.3.2 Intermediate Degradation under Multiple Switch Failures . . . 45
5.3.3 Large Degradation under Multiple Switch Failures ....... 45
5.4 Conclusions ................................ 46
CHAPTER 6
Conclusions ..................................... 64
APPENDIX A
MT—BASIC Code: Genetic Algorithm for the Self-Structuring Antenna . . . . 67
APPENDIX B
BASIC Code: Network Analyzer Control ..................... 75
APPENDIX C
BASIC Code: SSA Control for Switch Failure Analysis ............. 91
APPENDIX D
CPLD Code: Control Algorithm for the Resetting of Latching Relays ..... 93
BIBLIOGRAPHY ................................. 96

vi

LIST OF TABLES

Table 5.1 Multiple Failures ........................... 44

vii

Figure 2.1
Figure 2.2
Figure 3.1
Figure 3.2
Figure 3.3
Figure 3.4
Figure 3.5
Figure 3.6
Figure 3.7
Figure 3.8
Figure 4.1
Figure 4.2
Figure 4.3
Figure 4.4
Figure 4.5
Figure 4.6
Figure 4.7
Figure 4.8
Figure 4.9
Figure 4.10
Figure 5.1
Figure 5.2
Figure 5.3

Figure 5.4
Figure 5.5
Figure 5.6
Figure 5.7
Figure 5.8
Figure 5.9

LIST OF FIGURES

Self-Structuring Antenna System (Reprinted with permission [1])
A Typical Self-Structuring Antenna Skeleton ............
Antenna Board Schematic ......................
Control Board Schematic .......................
Self-Structuring Antenna Skeleton ..................
Antenna Board Layout ........................
Control Board Layout ........................
Completed Control Board ......................
Completed 30—switch Self-Structuring Antenna ...........
Control Board Test Circuit Schematic ...............
Test Setup with horizontal placement of the SSA. .........
Test Setup with vertical placement of the SSA. ..........
Television Reception at a Fitness of 0.5. ..............
Television Reception at a Fitness of 0.8. ..............
Television Reception at a Fitness of 0.9. ..............
Television Reception at a Fitness of 1.0. ..............
UHF Test Results with Horizontal Placement of the SSA.

UHF Test Results with Vertical Placement of the SSA. ......
VHF Test Results with Horizontal Placement of the SSA.

VHF Test Results with Vertical Placement of the SSA. ......
Laboratory Test Setup .........................
Self Structuring Antenna Template ..................

Regions of the SSA Exhibiting Different Levels of Degradation un-
der Single Switch Failures .......................

Single Failure of Switch 1. ......................
Single Failure of Switch 2. ......................
Single Failure of Switch 3. ......................
Single Failure of Switch 4. ......................
Single Failure of Switch 5. ......................
Single Failure of Switch 6. ......................

viii

6

7
12
13
14
15
16
17
18
19
29
30
31
32
33
34
35
36
37
38
47
48

49
50
50
51
51
52
52

Figure 5.10
Figure 5.11
Figure 5.12
Figure 5.13
Figure 5.14
Figure 5.15
Figure 5.16
Figure 5.17
Figure 5.18
Figure 5.19
Figure 5.20
Figure 5.21
Figure 5.22
Figure 5.23
Figure 5.24
Figure 5.25
Figure 5.26
Figure 5.27
Figure 5.28
Figure 5.29
Figure 5.30
Figure 5.31

Single Failure of Switch 7. ...................... 53
Single Failure of Switch 8. ...................... 53
Single Failure of Switch 9. ...................... 54
Single Failure of Switch 10. ..................... 54
Single Failure of Switch 11. ..................... 55
Single Failure of Switch 12. ..................... 55
Single Failure of Switch 13. ..................... 56
Single Failure of Switch 14. ..................... 56
Single Failure of Switch 15. ..................... 57
Single Failure of Switch 16. ..................... 57
Single Failure of Switch 17. ..................... 58
Single Failure of Switch 18. ..................... 58
Single Failure of Switch 19. ..................... 59
Single Failure of Switch 20. ..................... 59
Single Failure of Switch 21. ..................... 60
Single Failure of Switch 22. ..................... 60
Single Failure of Switch 23. ..................... 61
Single Failure of Switch 24. ..................... 61
Multiple Failures Showing Minimal Degradation .......... 62
Multiple Failures Showing Intermediate Degradation ....... 62
Large Degradation due to One Switch under Multiple Failures . . 63
Large Degradation due to Two Switches under Multiple Failures . 63

KEY TO SYMBOLS AND ABBREVIATIONS

AGC: Automatic Gain Control

CPLD: Complex Programmable Logic Device

EA: Evolutionary Algorithm

GA: Genetic Algorithm

PCB: Printed Circuit Board

SSA: Self-Structuring Antenna

SWR: Standing Wave Ratio

UHF: Ultra High Frequency

VHF: Very High Frequency

CHAPTER 1

INTRODUCTION

This thesis introduces and investigates a new class of antennas called “Self-Structuring
Antennas” or SSAs. The SSA is a new concept developed by Dr Edward Rothwell
at Michigan State University. These antennas are designed to change their electrical
shapes in response to changes in their electromagnetic environment. This opens
the door for many uses of the SSA, especially in mobile applications, where the
electromagnetic environment is constantly shifting.

An overview of Self-Structuring Antennas is given in Chapter 2, including a de-
scription and basic operation of the SSA. Also included in this chapter is a brief
history of the SSA and some possible applications of the technology.

Chapter 3 covers the construction and operational testing of the first printed
circuit SSA built at Michigan State University.

The use of Genetic Algorithms, or GAs, for control of the SSA is covered in
Chapter 4. Also included is testing done using the prototype SSA. This testing served
as a proof of concept for the use of GAS in control of the SSA and has important
implications for future control through evolutionary algorithms.

With applications in mobile communications, where replacement of switches may
be difficult, it is important to understand how the SSA will respond when a switch,
or group of switches, malfunctions. For this reason, a study of the effects of switch
failures on the performance of the SSA was conducted. The results of this study are
the subject of Chapter 5.

Chapter 6 serves as a conclusion to this thesis. Contributions to this research are
discussed, along with future work needed to further the understanding of this new

technology.

CHAPTER 2

SELF-STRUCTURING ANTENNA CONCEPTS AND IDEAS

Traditional antennas face many unseen challenges when placed in their end applica-
tion. This is especially evident in automobile applications where rapid changes in
a car’s direction and the addition of aftermarket products can greatly affect perfor-
mance of the antenna. Primarily this is due to the fixed geometrical configurations
of traditional antennas. These antennas require costly measurement and testing in
order to perform well in a particular environment. The result is an antenna that
requires substantial engineering costs for a design that is very specific in application
and whose performance can degrade substantially due to changes in environment.
From these limitations, the self—structuring antenna has evolved.

The self—structuring antenna is a concept that involves creating a wideband an-
tenna that can automatically adjust its electrical shape in response to changes in
its electromagnetic environment. A brief history of the SSA is given in Section 2.1.
Concepts behind the Operation of the SSA are the subject of Section 2.2. Finally,

some practical applications of the SSA are given in Section 2.3.

2.1 Origination of the Self-Structuring Antenna

The self-structuring antenna was invented by Dr. Edward Rothwell at Michigan State
University in 1997. Several prototypes were constructed and testing showed that
the SSA had promising characteristics. Following these experiments, Dr. Rothwell
applied for a US. patent with the aid of Michigan State University. US. patent
number 6,175,723 was issued to Michigan State University in January of 2001. [1]
Research on the SSA began to accelerate in 1998-1999 when Dr. John Ross began
intensive numerical simulations. Dr. Christopher Coleman began research on the SSA

in 1999, culminating in a Ph.D. dissertation in 2002 [1]. Delphi Research Labs got

involved in SSA research in 2000 and began funding the research at Michigan State
in 2001. The first conference papers were presented at the 2000 IEEE AP-S / URSI
conference in Salt Lake City, Utah [2],[3]. A paper on this subject has also been
accepted for publication by the IEEE Antennas and Propagations magazine [4].

The author became involved in SSA research as an undergraduate through an
electrical engineering capstone project in the fall of 2000. Subsequently, he continued
research on the SSA as a graduate student of Dr. Rothwell. Several more conference
papers regarding the SSA were presented at the 2001 IEEE AP-S / URSI conference
in Boston, Massachusetts [5],[6], as well as at the 2002 IEEE AP-S / URSI conference
in San Antonio, Texas [7],]8].

Research on the self-structuring antenna continues at Michigan State University,

John Ross and Associates, and Delphi Research Labs.

2.2 Overview of the Self-Structuring Antenna System

The self-structuring antenna system consists of several components. These include
hardware and software used to control the states of the antenna, as well as the an-
tenna itself. Each component plays an important role in the overall performance of
the SSA. Figure 2.1 shows a general block diagram of the SSA system. The main
components of the SSA are shown, including the antenna skeleton, or template, a
receiver, and a microprocessor based controller. This diagram represents the mini—
mal components necessary for the SSA to work as intended. Occasionally, additional
control or measurement hardware is added when necessary.

The following sections discuss the components of the SSA in greater detail. Section
2.2.1 discusses the SSA template, focusing on the current versions of the SSA that have
been constructed using printed circuit boards (PCBs). Microcontrollers, computers,
and PCB control boards employing field programmable gate arrays (FPGAs) have

been used in controlling various versions of the SSA. The different control methods

for the SSA are the subject of Section 2.2.2.
2.2.1 Self-Structuring Antenna Skeleton

The self-structuring antenna skeleton, or template, is the primary structure for re-
ceiving or transmitting electromagnetic energy in the SSA system. The SSA skeleton
consists of radiating elements, switches, a feed network, and headers for connection
to the control hardware. In general, the radiating elements can be wires, patches,
or other radiating bodies interconnected by controllable switches [1]. The printed
circuit SSAs built during this research have used traces on the top side of a PCB
as their intended radiating elements. Figure 2.2 is the top of a typical SSA skeleton
showing the radiating elements, as well as other hardware. The antenna lines on the
SSA are interconnected using electromagnetic relays, which are shown as rectangles
in Figure 2.2. Control of the relays is accomplished through connections made to the
headers located on the antenna skeleton. The wiring between the relays and control
network are routed on the bottom side of the SSA template. Each SSA built in this
manner also required the use of driver hardware to supply the current necessary for
switching of the relays. In some cases the drivers were located on the SSA skeleton,
in others they were located on a separate control board. Further discussion is left for
Section 2.2.2. Finally, the feed networks for the printed circuit SSAs built to date
have consisted of standard 4:1 wideband television baluns. These have been used as
the feed network because the applications of the current SSAs are to mobile television
reception [5]. The end application for future generations of the SSA will need to be

considered in designing feed networks.
2.2.2 Control of the Self-Structuring Antenna

A microprocessor in the form of a microcontroller or computer is used to control the
self-structuring antenna. In each case a receiver is employed to measure the perfor-

mance of the SSA in a given configuration. The information attained by the receiver

can be in the form of signal strength, input impedance, or any other appropriate
measurable quantity [1]. This information is then used by the microprocessor to
make decisions on the future states of the SSA. The decision making process gener-
ally employs the use of a smart algorithm, such as a genetic algorithm as discussed
in Chapter 4.

Specific details on one control system used for the SSA can be found in Section

3.1.2.

2.3 Practical Applications of Self-Structuring Antennas

Current research activity on the self-structuring antenna is focused on an application
to mobile communications. Particularly, the research reported in this thesis focuses
on television reception in a changing environment, with application to automobiles.
Other practical applications to mobile communications include multitasking antennas
to do the jobs that currently require multiple antennas in automobiles, as well as
adaptable antennas for cellular phones and hand-held radios.

The applications to which the self-structuring antenna may be applied are numer-
ous, many of which probably haven’t even been conceived. This is the beauty of the
SSA — the engineering costs of design are low, while the opportunities for use are

high.

 

 

 

A
"

 

SSA / riz control lines

 

 

 

 

 

 

 

 

 

 

 

 

 

 

template 3
J
pmvlllwl [l “
line(s)
RECEIVER > MICRO-
PROCESSOR

 

 

 

Figure 2.1. Self-Structuring Antenna System (Reprinted with permission [1])

Switches Antenna Elements

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

\I \‘ \1 \\
\\ t l \‘
[:5 [i
la a
C33 [:3
El :fi ' fie
=3 [:33
., g E g
[Tm—l g 33:]
m if]
/ eat-:1 a] e: ] .:
\\ /
Microprocessor Relay Driver
Interface Hardware Antenna Feed

Figure 2.2. A Typical Self-Structuring Antenna Skeleton

 

CHAPTER 3

3O SWITCH PROTOTYPE CONSTRUCTION

Design and testing of the self—structuring antenna in printed circuit board (PCB)
form began with a 30 switch prototype. This board, along with subsequent ver-
sions of the antenna were created using ORCAD Capture and Layout software
(www.0rcad.com). They were then manufactured through Advanced Circuits Cor-
poration (www.4pcb.com). The creation of circuit schematics and board layouts for
the self-structuring antenna are covered in Section 3.1 along with production of the
SSA prototype. Section 3.2 covers the functional testing of the first printed circuit
board SSA.

3.1 Layout of the Self-Structuring Antenna

Layout and construction of the self-structuring antenna began with printed circuit
board layouts created through the use of ORCAD Capture and Layout software.
The circuit schematics for the antenna and control board were created using the
Capture software; these are included as Figure 3.1 and Figure 3.2, respectively. Once
the circuits were laid out using ORCAD Capture, netlists were created to tell the
computer how components should be connected. Next, footprints were created for the
various components to be placed on the circuit board. Footprints are a representation
of the connection of components to the circuit board and are used in connecting the
traces of the PCB. Finally, the components were placed in their positions within
the Layout software. Details regarding the layout and schematics of the antenna
and control boards for the 30 switch prototype are the subject of Sections 3.1.1 and
3.1.2. Section 3.1.3 explains the production process for the self-structuring antenna

prototype.

3. 1.1 Antenna Board

The antenna skeleton laid out for this prototype includes 30 electromechanical relays,
connections for a television balun, and several headers, as shown in Figure 3.3. Driver
and control hardware were placed on a separate circuit board as described in the next
section. Placement of the relays on this version of the SSA included some relays to
section off large portions of the board and some to create antenna elements of various
shapes and sizes. It is important to note that when sections of the board are cut
off from the feed structure, they still can play a part in reception through coupling.
In other words, even though a portion of the board is cut off from the rest, it is
still present and still effects the performance of the antenna. Also coupling into the
antenna and effecting the performance are the control lines that are routed on the
backside of the PCB. Studies as to the effect of the presence of the control lines on
reception have not been considered; it is only noted that they will in fact play some
role. ”The completed antenna board layout for the 30 switch SSA is included as Figure
3.4.

3.1.2 Control Board

The 30 switch SSA prototype required the use of a control board in order to reset
the latching relays used in the design. Included on this board are two complex
programmable logic devices (CPLDs) used to invert the control signals and reset the
relays, as well as drivers needed to provide current for turning the relays on and off.
The code used to program the CPLDS has been included as Appendix D. As in the
case of the antenna board, the control board was laid out using ORCAD Capture
and imported into ORCAD Layout for routing of traces on the PCB. The completed

circuit board layout is shown in Figure 3.5.

3.1.3 Production of the SSA Prototype

Once the antenna and control board layouts were completed, the printed circuit
boards needed to be fabricated and components needed to be attached.

The control board was fabricated at Michigan State University through the ECE
Shop. This was done using the Quick Circuit T-Tech Prototyping Machine, which
etches the traces onto a double sided copper board. Once the etching of the board
was complete, components were attached through hand soldering. The completed
control board is shown in Figure 3.6.

The antenna board was fabricated through Advanced Circuits Corporation
(www.4pcb.com) because of their ability to remove excess c0pper from the PCB.
This was necessary for proper operation of the antenna. Again, once the board was
etched the components were attached through hand soldering. The completed an-

tenna board is shown in Figure 3.7.

3.2 Functionality Testing of the Self-Structuring Antenna

Once the SSA prototype was built, it was important to verify that it would work as
expected. This was done in various stages, first through testing of the control board
for proper functionality, then through testing of the entire system. Testing of the
control board is covered in Section 3.2.1. Testing of the entire system, with emphasis

on functionality of the antenna skeleton, is left for Section 3.2.2.
3.2.1 Control Board Testing

Testing of the control board included the construction of/a test circuit to verify
prOper functionality. The test circuit was built on a protoboard using LEDs and
25052 resistors, chosen to match the resistance of the relays used on the antenna
board. A schematic of this test circuit is included as Figure 3.8. Headers identical
to those placed on the antenna board were connected to the protoboard and LEDs

and resistors were wired in series to simulate the switching of the relays. Signals

10

were applied to the control lines to turn the LEDs on and off. This testing proved

successful and the control board worked as expected.
3.2.2 Antenna Skeleton Testing

Testing of the antenna skeleton was a matter of connecting the entire SSA system
together, as explained in Chapter 2. Once the system was connected together, signals
were sent to the antenna skeleton to turn on certain relays that were subsequently
checked for functionality using a digital multimeter. Once all of the relays had been
checked, the SSA prototype was ready to be tested in an application. This was done
using a genetic algorithm and tests were run to measure the SSA’s performance for
the reception of UHF and VHF television signals. The genetic algorithm and its

testing are the subjects of Chapter 4.

11

 

 

 

O (D

 

'F‘
“HI- — l I

 

 

 

 

 

 

Figure 3.1. Antenna Board Schematic

12

 

 

 

 

mum

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

r

 

Figure 3.2. Control Board Schematic

13

 

 

 

 

 

 

 

 

Antenna Elements

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Sw1tcl\ies / \
\ "m H E 13“ / \.
.- 4.2::
[:z] .—

__. l l—
—-'r:::r-— 2E3]:
.. .——-—— :42]

* m“— [:5
l’::i'.:iE I
. a f “‘77—" I-—
;Q :22]

[me __
“"ile- AJZEL

___-_1 A I

E” “‘ E:—

i:::1 1.. l—

1 as:

[2:] 2,...

_ '22' L.

in] 22 22:1 ,
____. gi-

2:2]— 2 IE

. i T ._..
1:] L-\

 

 

 

 

 

 

 

 

 

/“ ‘\

Antenna Feed Control Interface

Figure 3.3. Self-Structuring Antenna Skeleton

14

 

Figure 3.4. Antenna Board Layout

15

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Figure 3.5. Control Board Layout

16

 

Figure 3.6. Completed Control Board

17

 

Figure 3.7. Completed 30—switch Self-Structuring Antenna

 

 

 

 

 

 

    
  

3383$23589933939939883883 “B
a .- n. 9:22!9£§23R&fi&&flfi§888

 

 

 

 

 

 

 

 

Figure 3.8. Control Board Test Circuit Schematic

19

 

 

 

CHAPTER 4

GENETIC ALGORITHM FOR THE SSA

The self-structuring antenna prototypes built during this project consist of 24—30
electromagnetic relays used to control restructuring of the antenna skeleton. With 11
relays, the number of electrical configurations available to the antenna is 2", which
is between 1.67 x 107 and 1.07 x 109 combinations. Due to the large number of
configurations available to the self-structuring antenna, the use of a search algorithm
to handle the choice of configurations becomes essential. The use of simple on/off
switches in the design of the self-structuring antenna makes a binary search algorithm,
such as a genetic algorithm, well suited for this task. The genetic algorithm used to

control the SSA is included as Appendix A.

4.1 Genetic Algorithm Concepts

Evolutionary algorithms, such as the genetic algorithm, use principles modeled after
nature to find a best fit solution to a given problem. This section will discuss the
principles behind the genetic algorithm, as well as application and specialization to

the self-structuring antenna.
4.1.1 General Concepts

Genetic Algorithms, or GA’s, consist primarily of three components modeled after
the natural selection process: preselection, selection, and crossover or breeding. In
addition to these processes, the makeup of individuals comprising the p0pulation
needs to be specified. These individuals, commonly known as chromosomes, are built
from a string of ones and zeros. In the case of the self-structuring antenna, the bit-
string that makes up the chromosomes represents the states of the switches, ones

being switches in the ‘on’ position and zeros in the ‘off’ position. Each of the above

20

components play an important role in the functionality of the algorithm and will be

covered in more detail in the following subsections.
4.1.1.1 Initialization

The first step in the use of a GA is creation of an original population of chromosomes.
The preferable method of filling this population is through a random process. The
original population was filled in a random manner in the case of the SSA. Once this

original population has been created, the GA is ready to be initiated.
4.1.1.2 Preselection

The preselection process involves choosing which individuals in a given population will
be allowed to breed in creating the next generation. The genetic algorithm for the
SSA uses the voltage present at the input to the automatic gain control (AGC) of the
television to evaluate the fitness of a given configuration. Consequently, individuals in
the population with the highest levels of fitness contribute more to the next pOpulation
than those with lower fitness levels. In other words, the strongest individuals are
allowed to breed more than other individuals in order to produce better offspring.

The fitness function used in the GA for the SSA was defined as

0.4

4.1
(’U — 0246002 - 0.4 ( )

f(v) =

 

where 12,480; is the ideal voltage at the input to the AGC of the television based
on measurements made using a dipole antenna attached to the television. 2) is the
measured voltage at the input to the AGC using the self-structuring antenna in a
given configuration. The above fitness function was chosen to give a value of unity
for what was determined to be a perfect picture on the television, while a value of 0.9
represented a good picture. The value of the fitness function was then used to fill a
breeding pool based on the relative fitness of an individual compared to the average

fitness of the population. If the population fails to fill the breeding pool completely,

21

random selection is used to bring the number of individuals up to the size of the

pepulation. This completes the preselection process and the selection process begins.
4.1.1.3 Selection

The selection process involves randomly choosing chromosome pairs from the breeding
pool created in the preselection process to breed, or crossover, in order to create the
next generation of combinations. This process is accomplished through stochastic
remainder sampling without replacement, allowing all members of the breeding pool
to create offspring [9]. Throughout this process, individuals are randomly selected
from the breeding pool to form a mating pair. Once an individual has been chosen to
be one half of a mating pair, that individual is removed from the breeding pool and
the pool size decreases. The process is repeated until there are no members left in

the breeding pool, having all been paired off for the crossover, or breeding, process.
4.1.1.4 Crossover and Mutation

The breeding process consists of mating pairs chosen in the selection process under-
going a crossover operation. This operation consists of two parents being split at a
random location along the chromosome, and recombining with a probability of Pm”.
A common value chosen for PM” is in the range of 0.6-0.8; the chosen value for the
SSA’s GA was 0.7. This means that the split chromosomes will recombine with each
other 70% of the time, giving offspring chromosomes made up of some information
from one parent and some from the other. The rest of the time the chromosomes will
not crossover, giving offspring identical to the parents. Once this process is complete,
there is a new population, identical in size to the original, with offspring made up of
the parents of the previous generation.

After the breeding process, the new population is subjected to a mutation process
in which single bits can be randomly changed with a probability of Pmut. The common

practice is a choice of Pmut = f where K is the population size. This was the chosen

22

value of Pm“, for the SSA. Once this mutation process is complete, the new population
takes the place of the old and the algorithm is repeated, starting with the preselection

process.
4.1.2 Specialization to the Self-Structuring Antenna

Several modifications were made to the standard genetic algorithm in order to speed
convergence of the population to a global maximum. The first of these specializations

was the use of a selective linear scaling of the fitness given by [9]

~

F = aF + b (4.2)

where F is the fitness of a given configuration and FM = Favg. If the maximum
fitness, Fm”, was less than 1.5 times the average fitness, Fm, then no scaling was
done. This was the case of having no individuals dominating the population. When
a given configuration appeared to dominate the population, scaling was performed to
give Fm“ = cmuthavg, where cm,“ is the scaling factor desired for the best population

member. In order to fulfill the above relations, a and b were chosen as

(Cmult — 1)Favg
= 4.
a Fmax — Favg ( 3)

 

b = (1 — a)Fa,,g (4.4)

If Fmin = aFm;n + b < 0, then a and b were recomputed such that Fmin = O and

Fwy = Favg . Relations which give this behavior are

Favg
= —.———— 4.5
a Favg — Fmin ( )
b = —aFm,n (4.6)

23

Scaling the fitness of the population in this manner helped to avoid having the pop-
ulation driven to a local maximum, rather than a global one.

The second specialization of the GA to the self-structuring antenna was the ad-
dition of a limited amount of elitism, or survival of the fittest, which insured that
the best states were not lost due to crossover or mutation. This addition gave more
favorable results, since the best states from several generations were able to produce
offspring in subsequent generations. Another feature of the genetic algorithm that
is unique to the SSA is the use of a single bit-string to make up a full chromosome.
Generally, a genetic algorithm uses several bit-strings pieced together to make up a
chromosome, allowing only changes in a few of the variables during crossover. Having
only one variable to optimize, that being the configuration of the antenna skeleton,
the GA for the SSA is inherently simpler, providing for a faster convergence. The
modifications made to the standard GA allow for better performance in the intended

application.

4.2 Television Measurements

Measurements were made on the 30-switch self-structuring antenna prototype using
a genetic algorithm for control of switch states. The relays on the antenna tem-
plate were controlled using a Multitrax microcontroller board (Control Technology,
www.controltrax.com). This board, based on the Hitachi HD64180 microcontroller,
has 48 lines of digital I/ O, 8 channels of 10-bit A/ D conversion, 64kB of RAM and
32kB of EPROM memory, and a ROM-based multitasking BASIC run—time compiler
(MT-BASIC). The digital I/O lines could not supply enough current to drive the
relays directly, so they were used in conjunction with 74AC541N driver ICs. The
genetic algorithm was written using MT—Basic and has been included as Appendix
A. The purpose of measurements made using the 30—switch prototype with genetic

algorithm control described below were for proof of concept. This testing shows the

24

self-structuring antenna’s ability to adapt itself in order to find optimum reception in
a given environment through the implementation of a genetic algorithm.

Section 4.2.1 describes the environment under which testing took place. Sections
4.2.2 and 4.2.3 give results of testing done for channels in the UHF and VHF bands.
Specifically, channel 23 (524-530 MHz) was used for UHF testing, while channel 10
(192-198 MHz) was used for VHF testing.

4.2.1 Testing Environment

Testing performed on the 30—switch SSA prototype took place outside of the Engi-
neering Research Complex on the campus of Michigan State University. The antenna
was placed on top of the television used for testing. Figure 4.1 shows the test setup
for the SSA placed horizontally atop the television, while Figure 4.2 displays a ver-
tical orientation. In both cases, the television was placed on a wooden table along
with power supplies, a computer and monitor, a control board, and the microcon-
troller. The computer was used to collect data as tests were run, and to provide an
interface to the microcontroller. The use of latching relays made the control board
containing drivers and logic hardware necessary, since the relays needed to be released
after each combination and required large amounts of current for switching. For more
information on the control board, see Section 3.1.2.

A baseline for what cbnstitutes a good received signal was provided through the
use of a standard television dipole antenna, or “rabbit ears”, using the voltage present
at the automatic gain control of the television. This voltage was then used to set the
ideal voltage in the objective function within the genetic algorithm as mentioned in
4.1. Using this function we were able to quantify how clear the television picture
would be when placed in a given configuration and consequently choose the best
parent configurations for the next generation.

The testing performed here included looking at two local channels, one in the Ultra

High Frequency (UHF) band and one in the Very High Frequency band (VHF). For

25

UHF testing channel 23 (524—530 MHz) was used and for VHF, channel 10 (192-198
MHz) was used. In both cases, tests were run using both a vertical and horizontal
orientation for the SSA, as mentioned earlier. The results were then compared to
improvements seen with a typical genetic algorithm. For a typical GA the average
value of the fitness increases in a fairly smooth fashion, then saturates at a steady
value. Evolution in maximum fitness for a typical GA often proceeds in “punctuated
equilibria”, with sudden improvements interspersed between periods in which fitness
remains relatively unchanged [10]. To get an idea of what a good signal is, consider
Figure 4.3, giving a fitness of 0.5. This figure shows what is considered to be a
bad picture. The television has ‘snow’ on the screen and is generally an unpleasant
viewing experience. An improvement is seen in Figure 4.4, where the fitness is 0.8 and
the ‘snow’ has decreased. The best pictures are seen between 0.9 and 1.0, shown in
Figure 4.5 and Figure 4.6. These pictures are crisp and free of ‘snow’, which is what
one wants to see when watching television. Further detail on this testing is given in

the following sections.
4.2.2 UHF Test Results

Testing was performed to evaluate the effectiveness of the SSA and its genetic al-
gorithm control for the frequency band known as Ultra High Frequency, or UHF.
Experiments were performed for both a horizontal and vertical orientations of the
SSA.

Results for the horizontal orientation are given in Figure 4.7. Comparing these
results to the typical case described above, the genetic algorithm, as well as the SSA
perform as desired. Specifically, there is a fairly smooth increase in the average fit-
ness of the pOpulation to a fitness of about 0.7, while the maximum fitness increases
quickly to a steady value of about 0.9 and up. Note that there are variations in the
maximum fitness with generation, suggesting that the best configuration is changing

from generation to generation. This is a result of the fitness scaling explained above.

26

The results for UHF testing using a horizontal orientation were the most promising,
telling us several things. First, the orientation of the antenna skeleton is important.
This was to be expected, since most US. television stations broadcast using a hori-
zontal polarization, with the exception of some VHF stations who broadcast with a
circular polarization. Second, the SSA is of a good dimension to receive signals at
around 500 MHz. At this frequency, the dimensions of the SSA used in testing are
about 0.8/\ by 0.53A, where A is the wavelength of the received signal.

Vertical orientation test results are given in Figure 4.8. These results reinforce
the conclusion given above, that the orientation of the SSA is important. Looking
at Figure 4.8, one can see that the performance of the SSA is generally random in
average, maximum, and minimum fitness. The performance is also of a generally
worse quality than in the horizontal case, with average values around 0.5-0.6 and
maximum values of 0.7-0.9. The ineffectiveness of the GA here is due to the lack of

good combinations with the SSA oriented vertically.

4.2.3 VHF Test Results

Testing for the VHF band show that the SSA template used for these experiments
was too small to effectively receive the signals in this frequency range. The size of
the antenna at around 195 MHz is about 0.3A by 0.2A. The fitness levels seen in
VHF testing were generally lower, giving steady state values averaging about 0.5
for horizontal orientation, with around 0.75 fitness levels for maximum values, as
seen in Figure 4.9. Figure 4.10 shows that vertical orientation again gives decreased
performance, with average fitness levels around 0.4 and maximum fitness levels around
0.6. The lower fitness levels seen in the vertical case, as well as the slightly more
random features, lead one to believe that the levels seen in the VHF case are indeed
due to the size of the antenna and not necessarily due to a circular polarization of

the received signal.

27

4.3 Conclusions

The use of a genetic algorithm and the testing done here provides a proof of concept for
the self-structuring antenna. It shows us that the antenna is capable of delivering good
performance under certain conditions. Specifically, it shows that the current template
design is effective for UHF television reception under a horizontal orientation. This
testing gives some insight into the necessary orientation and size of the antenna. As
seen in the VHF testing, the antenna is currently undersized for the frequency band
in which it is expected to perform. These are all considerations that need to be taken
into account for future antenna templates in order to create an antenna that is both

diverse and effective.

28

 

Figure 4.1. Test Setup with horizontal placement of the SSA.

29

 

Figure 4.2. Test Setup with vertical placement of the SSA.

30

”’ . 2222.22 «1'» l. ' 5 'r .EMHUNMTW .‘313. l E .
— I m W _. "A.

'1‘.
was... pm”... . . .

 

Figure 4.3. Television Reception at a Fitness of 0.5.

31

..‘!,.,;l, , ' . . inl .\;.\'.\‘..
‘ -. ‘2‘ 1., .25 .

f 3
3.
. f.

i

 

Figure 4.4. Television Reception at a Fitness of 0.8.

32

U3 .' \j
\

yxll“ “fl"

[m -\\l
lr'flutt A .

» .‘ . _ «‘n’l -‘ .
DCQ’H" ’4“ ‘ W}
uh\ttoll‘\\‘t
“K‘ ' .

 

Figure 4.5. Television Reception at a Fitness of 0.9.

33

 

Figure 4.6. Television Reception at a Fitness of 1.0.

34

 

— Minimum Fitness
-0- Average Fitness
-0- Maximum Fitness

 

 

 

 

I

0.8

0.7

0.2

0.1

 

I

 

 

12 14 16 18 20

8 1O

 

Generation Number

Figure 4.7. UHF Test Results with Horizontal Placement of the SSA.

35

 

0.9

I

 

.0
—L
I

 

l

 

 

 

 

 

— Minimum Fitness
-0- Average Fitness
-0- Maximum Fitness

 

 

 

o l l 1 l
O 5 10 15 20

Generation Number

25

30

35

Figure 4.8. UHF Test Results with Vertical Placement of the SSA.

36

 

Fitness

 

 

— Minimum Fitness
-0- Average Fitness
-0- Maximum Fitness

 

 

 

 

1 r . . . . .
0.9 ~ .,
0.8 - ..
0.7 r“ a
0.6 r -
0.5 r .
0.4 - .
0.3 - -
0.2 - 3
W
0.1 - q
00 5 110 1L5 210 215 310 35

Generation Number

Figure 4.9. VHF Test Results with Horizontal Placement of the SSA.

37

 

0.9

0.8

0.2

0.1

 

 

 

— Minimum Fitness
-0- Average Fitness
-0- Maximum Fitness

 

 

 

Generation Number

F -
0 5 10 15 20 25 30

35

Figure 4.10. VHF Test Results with Vertical Placement of the SSA.

38

 

CHAPTER 5

V

SWITCH FAILURE ANALYSIS

Many times antennas are subject to harsh environmental conditions, where their
physical degradation, alteration, or misuse results in decreased electrical performance.
Self-Structuring Antennas should be designed to operate under these difficult condi-
tions and respond to changes both in their environment and in their own physical
structure without loss of effectiveness. Since these antennas are to be used in many
consumer applicati ns where replacement and repair may be difficult, it is important
to consider the effect of switch failure on the performance of the antenna.

The successful oper ion of the SSA is dependant on the large number of available

 

states in which the anten a performs acceptably. .It is important that the antenna
template is designed in such a way that the failure of a switch, or a group of switches,
causes only minimal degradation in the performance of the antenna. That is, the per-
formance of the antenna does not fall below an acceptable level under switch failure.
For this reason, a study of the effect of switch failure on the SSA was performed.
Section 5.1 gives an overview of test procedures and the environment in which
the switch failure analysis was conducted. Results of single switch failures and their
consequences are covered in Section 5.2. Multiple switch failures are the topic of

Section 5.3.

5.1 Testing Environment and Procedure

Experiments were performed to measure the standing wave ratio (SWR) of the Self-
Structuring Antenna using the HP-8510 Network Analyzer. Figure 5.1 shows the
environment under which switch failure was analyzed. It can be seen in Figure 5.1
that the antenna was placed horizontally atop a styrofoam pedestal. The antenna

switches were controlled using a FlashTrax microcontroller board which was placed on

39

a neighboring pedestal. This board, based on the Hitachi HD64180 microcontroller,
has 24 lines of digital I/O, 8 channels of 10—bit A/ D conversion, 64k of flash RAM
memory, and a ROM-based optimized BASIC interpreter. The antenna feed was
connected to the network analyzer through a standard 4:1 wideband television balun.
Measured standing wave ratios for each switch combination were logged on a nearby
computer for future analysis. The BASIC code used to control the network analyzer
is included in Appendix B. The BASIC code used by the microcontroller to set the
antenna states is given in Appendix C.

The procedure for testing the SSA involved sending 10000 random states to the
antenna and measuring the resulting standing wave ratio. The switch combinations
sent to the antenna were statistically random and were consistent throughout the
experiments. That is, the states were sent to the SSA from a file of 10000 random
numbers that remained the same for all tests. This procedure was carried out for
operating frequencies in the range of 50 to 450 MHz, taken in 50 MHz steps.

The results of this testing were analyzed using a C++ program called switchcpp.
This program was used to compare the SWR measurements for failures of individual
switches as well as groups of switches. A baseline was established for comparison
using all 10000 switch combinations when no switches failed. The resultant figures

are explained in Sections 5.2 and 5.3.

5.2 Single Switch Failures

In order to get a fundamental understanding of the consequences of a switch failure,
the malfunctioning of a single switch is considered. A switch failure occurs when
any switch on the antenna template is permanently placed in either an “on” or “off”

position. To quantify the effect of a failed switch, the following function was defined:

frac( f) — N“ (5.1)

_ NAS

40

where N05 is the number of good states defined as those switch combinations which
give SWR S 2.0, and N AS is the number of available states. Available states are
defined as switch combinations that the antenna is still able to be placed into. With
this function, switches which give minimal, intermediate, or large degradation in the
performance of the antenna are placed into groups. It should be noted that the
number of available states is roughly cut in half with each switch failure. Therefore,
a switch failure reduces the total number of good states and causes a degradation in
performance, although its effect may be minimal. Figure 5.2 shows the location of

switches on the antenna template.
5.2.1 Minimal Degradation under Single Switch Failures

Switches that exhibit minimal degradation in SSA performance under failure are
those that only cause the efficiency of the control algorithm to deteriorate. The
algorithm’s ability to find good states depends on the large number of available good
states. That is, the more total good states available to the algorithm, the faster it
can optimize the population. This means that when the total number of states is cut
in half the algorithm becomes less efficient, even if the number of good states is also
cut in half. Switch failures showing minimal degradation represent the case of the
number of good states being cut in half when the total number of states is also cut in
half. For this case, the frac(f) function remains unchanged from the fully functional
case over the entire frequency band (50 to 450 MHz). Switches that exhibit this
behavior are generally located away from the antenna feed, controlling larger sections
of the antenna template. These switches are numbers 1 (Figure 5.4), 3 (Figure 5.6), 5
(Figure 5.8), 7 (Figure 5.10), 10 (Figure 5.13), 11 (Figure 5.14), 15 (Figure 5.18), 19
(Figure 5.22), 21 (Figure 5.24), and 23 (Figure 5.26) in Figure 5.2. The figures listed
above show cases in which the frac(f) function follows the baseline case, regardless of
whether a switch fails in the “on” or “off” position. Figures showing this behavior

represent switch failures causing minimal degradation.

41

5.2.2 Intermediate Degradation under Single Switch Failure

Switches exhibiting intermediate levels of performance degradation involve a dete-
rioration in the efficiency of the control algorithm, as well as small changes in the
frac(f) function for discrete frequencies. Switch 8 (Figure 5.11) is an example of a
switch failure which causes intermediate levels of degradation. Figure 5.11 shows a
noticeable change in the frac(f) function at both 200 and 400 MHz. This behavior
also occurs in switches 9 (Figure 5.12), 12 (Figure 5.15), and 17 (Figure 5.20) at dif-
fering frequencies in each case. It is interesting to note that these switches lie in the
area of the board bounding minimal and large degradation areas, as shown in Figure
5.3. These switches show better behavior than those which control shorter elements,
but worse behavior than those controlling longer ones. This suggests a relationship
between the electrical length of the antenna elements and the performance of the SSA
under switch failure. N o conclusions have been drawn as to why certain switches in

this group show changes in performance at differing frequencies.
5.2.3 Large Degradation under Single Switch Failure

Switches which show large levels of change in performance lower the efficiency of the
algorithm, as well as changing the frac(f) function large amounts at various frequen-
cies. Figure 5.5 shows a very pronounced case of this type of failure. This figure
represents a failure of switch 2 of Figure 5.2. It can be seen in this figure that the
performance of the antenna is changed dramatically when switch 2 fails. At low fre-
quencies there is a vast degradation in performance if the switch fails in the “on”
position, while performance seems to improve when the switch fails “off”. The op-
posite is true at higher frequencies. It is the unpredictability of the state in which
a switch may fail, as well as the wideband application of the SSA, that causes this
type of switch failure to be unacceptable. Other switches which cause large perfor-

mance changes are 4 (Figure 5.7), 6 (Figure 5.9), 13 (Figure 5.16), 14 (Figure 5.17),

42

16 (Figure 5.19), 18 (Figure 5.21), 20 (Figure 5.23), 22 (Figure 5.25), and 24 (Figure
5.27). Note that all of these switches lie close to the antenna feed and control the
configuration of short antenna elements, again suggesting a link between electrical
length and performance.

It should be emphasized that failures showing large performance degradation are
unacceptable to the design of the SSA template, since they lower the total number
of good states in an unpredictable fashion. These failures need to be accounted for
through redundancy in the layout of the SSA template in order to avoid complete

failure of the antenna in its end application.

5.3 Multiple Switch Failures

In order to obtain a better understanding of the effects of switch failure, the possibility
of multiple failures was explored. Multiple failures are ones in which more than one
switch gets permanently placed in either the “on” or “off” position. This means that
there are four possible ways in which a given pair of switches can fail. This can be
seen in Figure 5.28, with both switches failing on, both failing off, switch 7 failing
on and 11 off, and finally switch 7 off and 11 on. Also included in this and the rest
of the Figures in this section is the control case of no switch failures. With four
ways in which a pair of switches can fail, and 24 total switches, 1044 possible failures
exist. These failures were grouped into the following categories, shown in Table 5.1:
minimal degradation , intermediate degradation, large degradation due to one switch,
and large degradation due to both switches. Figure 5.28 shows an example of switches
exhibiting minimal degradation in antenna performance. Figure 5.29 is an example
of intermediate levels of degradation, while Figure 5.30 and Figure 5.31 are examples
of large degradation due to one and both switches, respectively. These examples are
representative of each type of performance degradation seen during multiple switch

failure analysis.

43

Minimal Change
Intermediate Change
Large Change

'0
.2
'E
u.
.r:
o
2:
3
a)
u
c
o
.2

1 = Minimal change infraction of good states

2 = Intermediate change in fraction of good states
3 = Large change due to single switch failure only
4 =

 

Table 5.1. Multiple Failures

5.3.1 Minimal Degradation under Multiple Switch Failures

Minimal degradation is regarded as having no appreciable change in the frac(f) func-
tion over the entire frequency band (50—450 MHz), as defined in Section 5.2. With
multiple switches failing, combinations which showed this property were primarily
those in which both switches showed minimal degradation under single switch failure.
This held true for most combinations of switches behaving in this fashion, although

some combinations did result in intermediate amounts of degradation when one or

44

more of the switches were near the minimal-intermediate boundary shown in Figure

5.3.
5.3.2 Intermediate Degradation under Multiple Switch Failures

Intermediate degradation involves similar behavior to the above case, with some slight
deviations at discrete frequencies. Figure 5.29 shows this behavior for switch com-
binations involving switches 7 and 12. Note that individually, one of these switches
was in the minimal degradation group, while the other was in the intermediate group.
This tends to be the case for multiple switch failures resulting in intermediate lev-
els of degradation, although in some cases a large change was noted when switches
were near the boundaries of minimal-intermediate, or intermediate-large change under

single switch failures, as shown in Figure 5.3.
5.3.3 Large Degradation under Multiple Switch Failures

Switch combinations that involved large degradation in performance of the SSA were
split into two groups in order to get a better understanding of the interaction between
switches. Combinations that exhibited large change independent of one of the failing
switches made up one group, while those that varied greatly due to both switches
made up a second group. Figure 5.30 shows a switch combination that belongs to the
first group. We can see that a large degradation occurs, but it is independent of the
position of one of the switches, specifically switch 22. In contrast, Figure 5.31 shows
a combination that belongs to the second group. In this case, the degradation that
occurs is dependent on both of the switches failing. The last group represents the
worst possible case in terms of switch failures. The design of the SSA template needs
to be reconsidered in order to avoid failures of this type, since they are unpredictable
in their consequences.

Generally, switch combinations that produced large amounts of degradation in-

dependent of one switch involved both a switch showing minimal degradation, and

45

one showing large degradation under single failures. This is what one might expect
to happen. Multiple failures that caused the largest amounts of performance degra-
dation due to both switches were also predictable. These were switches which caused
large amounts of degradation under single failures. Figure 5.31 is an example of this

second case, involving switches 22(Figure 5.25) and 24(Figure 5.27).

5.4 Conclusions

The current SSA template has various shortcomings that can be readily seen through
this switch failure analysis. First, at low frequencies, the antenna does not perform
well. This is true when no switches fail, due to the relatively small size of the template
in comparison to a wavelength at low frequency. In addition, switches controlling short
antenna elements cause a considerable loss in performance of the antenna under both
single and multiple failure, especially at higher frequencies. This results from a lack
of redundancy for the shorter antenna elements, i.e., longer elements can be built
from shorter ones, but the inverse is not true. Realization of these shortcomings can
be used in future generations of template layouts in order to create a more robust

design that will withstand failure of one or more switches.

46

 

Figure 5.1. Laboratory Test Setup.

47

 

 

 

 

 

 

 

 

 

is . a
’ 22 it
E E ii
~22 a [3 Bit .
E if! .. I}:
rm E3
Wm 22

 

 

 

mt? :5

 

Figure 5.2. Self Structuring Antenna Template.

48

 

 

 

 

fl

1!

 

 

’ fl Minimal Change d ‘5

23 11

2:: 6:5

7 u t:

 

 

   
 

Large Change Q:
E5

Intermediate 2 E23
lino-«”33 Change
1 E I Q E E

Figure 5.3. Regions of the SSA Exhibiting Different Levels of Degradation under
Single Switch Failures.

 

 

 

 

 

 

 

 

49

0.5

0.45

0.5

0.45

0.35

0.3

E 0.25

0.2

0.05

 

 

 

1 1 l l I. A l

1 00 1 50 200 250 300 350 400
Frequency

Figure 5.4. Single Failure of Switch 1.

 

— No Failures
-9- SW1 on
--N— SW1 oft

 

 

450

 

 

l 1 l l l 1 l

 

— No Failures
-e- SW2 on
+ SW2 oft

 

 

 

50

1 00 1 50 200 250 300 350 400
Frequency

Figure 5.5. Single Failure of Switch 2.

50

450

 

 

0.5
0.45

0.4

0.5
0.45
0.4
0.35
0.3

3: 0.25

0.2

0.1

0.05

 

 

 

L l l l L l l

1 00 1 50 200 250 300 350 400
Frequency

Figure 5.6. Single Failure of Switch 3.

 

— No Failures
-e— SW3 on
+ SW3 off

 

 

450

 

 

1 l 1 l

 

— No Failures
—e— SW4 on
+ SW4 oft

 

 

 

1 00 1 50 200 250 300 350 400
Frequency

Figure 5.7. Single Failure of Switch 4.

51

450

 

 

 

0.5
0.45 r-

0.4 r-

0.3 *-

§ 0.25

0.2 l-

T

0.15 '-

 

1 1 1 1

 

—— No Failures
-e— SW5 on
-N— SW5 off

 

 

 

1 00 1 50 200 250 300 350 400
Frequency

Figure 5.8. Single Failure of Switch 5.

450

 

0.5

0.4 *-

0.35 ’-

0.3 ’-

0.2 *-

 

1 1 1 1 1 1 1

 

— No Failures
—6- SW6 on
-- SW6 on

 

"V

 

 

1 00 1 50 200 250 300 350 400
Frequency

Figure 5.9. Single Failure of Switch 6.

52

450

 

 

 

0.5
0.45
0.4
0.35
0.3

g 0.25

0.2

 

 

— No Failures
—e— SW7 on
+ SW7 off

 

 

J. 1 1 1 1 1

 

100

1 50 200 250 300 350 400 450
Frequency

Figure 5.10. Single Failure of Switch 7.

 

0.5

0.45

0.35

E 0.25

*

 

 

 

— No Failures
-e— SW8 on
-— SW8 off

 

 

l 1 1 1 1 1

1 50 200 250 300 350 400 450
Frequency

Figure 5.11. Single Failure of Switch 8.

53

 

 

 

 

 

 

 

 

 

 

 

 

0-5 V I 1' I’ 1 Ti f _ No Failures
-6- SW9 on
0 45 _ q + SW9 off
0.4 - -
0.35 - ..
0.3
§ 0.25
0.2
0.15
0.1
0.05
50 1 00 1 50 200 250 300 350 400 450
Frequency
Figure 5.12. Single Failure of Switch 9.
0.5 U I I V I I I NO Failutaa
-e- SW10 on
O 45 _ _ —~— SW10 off
0.4 _ -

 

1 I 1 1 1 1 1

250
Frequency

0
50 1 00 1 50 200 300 350 400

Figure 5.13. Single Failure of Switch 10.

54

 

 

450

 

 

 

005 I I I ' U
' ' — No Failures

—-e— SW11 on
—N— SW11 off

 

0.45 ~ .

 

 

0.4 - «

 

 

 

I 1 1 1 L 1

o L
50 1 00 1 50 200 250 300 350 400 450
Frequency

Figure 5.14. Single Failure of Switch 11.

 

 

 

 

0.5 I T I I I I I No Failufes
-9- SW12 on
-N- SW12 off
0.45 b 1
0.4 r- a

 

 

 

1 1 1

O ‘ 1 1 l
50 1 OO 1 50 200 250 300 350 400 450
Frequency

Figure 5.15. Single Failure of Switch 12.

55

 

0.5 I I I f T r I

— No Failures
-6— SW13 on

 

 

 

0.4 - -*

 

 

 

 

G ’ 1 1 1 L 1 1 1
50 1 00 1 50 200 250 300 350 400 450
Frequency

Figure 5.16. Single Failure of Switch 13.

 

 

 

 

0'5 ' ‘ ' ' ' ‘ ‘ — No Failures
-e— SW14 on
O 45 _ . -N- SW14 off
0.4 ~ ~

 

 

 

J 1 1 1 1 1 1

o" v
50 1 00 1 50 200 250 300 350 400 450
Frequency

Figure 5.17. Single Failure of Switch 14.

56

 

— No Failures
-e— SW15 on
+ SW15 off

 

0.45 r- ..

 

 

0.4 - -[

 

 

 

1 1 1 1 1 1 1

O
50 1 00 1 50 200 250 300 350 400 450
Frequency

Figure 5.18. Single Failure of Switch 15.

 

0.5 I I I I I I M

— No Failures
-e— SW16 on
+ SW16 oft

 

0.45 '- _.

 

 

 

 

 

o 1 1 1 L 1 1 1

50 1 00 1 50 200 250 300 350 400 450
Frequency

Figure 5.19. Single Failure of Switch 16.

57

0.5

0.45

0.4

0.5
0.45
0.4
0.35
0.3

E 0.25

0.2

 

 

150

1

200

1

250

Frequency

1

300

350

 

400

Figure 5.20. Single Failure of Switch 17.

 

—— No Failures
-e- SW17 on
-N— SW17 off

 

 

 

 

150

200

250

Frequency

1

300

350

 

400

Figure 5.21. Single Failure of Switch 18.

 

— No Failures
-e— SW1 6 on
-u— SW1 6 off

 

 

450

 

 

 

0.5 I I fl I I I 17

— No Failures
-e— SW19 on
0 45 _ q -— SW19 oft

 

 

 

0.4 f -*

 

 

 

O 1 1 1 1 1 1 1
50 1 00 1 50 200 250 300 350 400 450

Frequency

 

Figure 5.22. Single Failure of Switch 19.

 

 

 

 

0:5 r I I I I I I No Failures
-e— SW20 on
0.4 t- ‘

 

 

 

1 1 1 1 1 1 1

50 1 00 1 50 200 250 300 350 400 450
Frequency

Figure 5.23. Single Failure of Switch 20.

59

 

0.5
0.45 ~
0.4 ~
0.35 -
0.3 F

E 0.25 -

0.2 *-

 

 

1 1 1 l

1 00 1 50 200 250 300 350 400
Frequency

Figure 5.24. Single Failure of Switch 21.

 

— No Failures
-e- SW21 on
+ SW21 oft

 

 

450

 

0.5

0.45 '-

0.4 *-

irac
.0
to
0:
I

 

h

 

1

1 00 1 50 200 250 300 350 400
Frequency

Figure 5.25. Single Failure of Switch 22.

60

 

— No Failures
-e- SW22 on
—- SW22 oft

 

 

450

 

 

0.5
0.45
0.4
0.35
0.3

g 0.25

0.2

0.5
0.45
0.4
0.35
0.3

g 0.25

0.2

 

 

 

— No Failures
—e— SW23 on
+ SW23 oft

 

 

1 1 M 1_ 1 1

 

100

1 50 200 250 300 350 400 450
Frequency

Figure 5.26. Single Failure of Switch 23.

 

 

 

 

— No Failures
-e— SW24 on
+ SW24 oft

 

 

1 1 1 1 1 1

1 50 200 250 300 350 400 450
Frequency

Figure 5.27. Single Failure of Switch 24.

61

 

 

 

 

 

 

 

 

 

0'5 ' ' ' ' ' ' ' — No Failures
-e— SW7andSW1 1on
O 45 - q + SW70nSW1 1ott
' —i— SW7oflSW1 1on
+ SW7andSW1 1011
0.4 - ~
0.35 - .,
o 1 #1 1 1 1 1 1
50 1 00 1 50 200 250 300 350 400 450
Frequency

Figure 5.28. Multiple Failures Showing Minimal Degradation

 

 

 

 

 

 

 

0.5 I I T I I j I
— No Failures
—e— SW7andSW120n
O 45 ,_ _ -— SW70nSW12oft
' —4— SW‘IoftSW120n
+ SW73ndSW120fl
0.4 - -
O . 1 1 1 I 1 1 1
50 1 OO 1 50 200 250 300 350 400 450
Frequency

Figure 5.29. Multiple Failures Showing Intermediate Degradation

62

 

 

 

 

 

 

0.5 I I I I I I I .

— No Failures

-e— SW228ndSW230n
O 45 _ j + SW220nSW23011
‘ —i— SW2zottSW230n

+ SW22andSW230fl‘
0.4 - ..

 

 

 

1 1 1 1 1 1 1

o 6
50 1 OO 1 50 200 250 300 350 400 450
Frequency

Figure 5.30. Large Degradation due to One Switch under Multiple Failures

 

0'5 ' T ' I I T I — No Failures

-e- SW22andSW24on
045 F _ + SW220nSW24oft

' -+— SW2201‘tSW24on
+ SW22andSW24otf

 

 

 

 

 

 

1

50 1 00 1 50 200 250 300 350 400 450
Frequency

Figure 5.31. Large Degradation due to Two Switches under Multiple Failures

63

CHAPTER 6

CONCLUSIONS

This thesis has presented an experimental perspective of a new class of antennas
known as self-structuring antennas, or SSAs. General concepts regarding the SSA
system, design, and testing process were covered. Several conclusions can be drawn
from the research done in constructing this thesis.

First, a prototype SSA was constructed using a printed circuit board for the
skeleton of the system. This prototype was built and tested, providing some insight
into the operation of the SSA. Specifically, the conclusions that were drawn from this
prototype were the necessity of consideration of orientation and size in the design
of the SSA. These factors are to be used in the design of future generations of the
self-structuring antenna. Also, the use of a genetic algorithm as an efficient control
for the SSA was shown using this prototype. This has implications in the use of
smart algorithms for control of the SSA in future applications, giving reason to test
the effectiveness of other algorithms, such as ant colony optimization or simulated
anneafing.

Switch failure analysis performed during this research shone some light on several
shortcomings of the SSA. First, the relatively small size of the self-structuring antenna
prototypes figured directly into the results of testing at low frequencies. It was shown
that the antenna performed poorly at these frequencies with or without switch failure.
In addition, testing showed that switches controlling short antenna elements created
a substantial degradation in performance under switch failure, especially at high
frequencies. A conclusion drawn from this was that the antenna did not have the
necessary redundancy to counteract the failure of a switch controlling a short antenna

element. Perhaps more clearly, short elements can build longer ones, but the inverse

64

is not true.

The shortcomings in the SSA designs that were exposed during this research give
some insight into how future generations should be built. These allow the designer
to take into account the application of the self-structuring antenna in deciding the
necessary orientation and size of the template. One can also take into account the
need for redundancy in the short antenna elements on a given template. All of these
conclusions can be put together to create a good framework from which future SSA

templates can be constructed.

65

APPENDICES

66

IXPWWETIEHDCIA

MT-BASIC CODE: GENETIC ALGORITHM FOR THE
SELF-STRUCTURIN G ANTENNA

5 ’This program works as the genetic algorithm for control of the SSA

6 ’Authors: Brad Perry and Dr. Ed Rothwell, Michigan State University

10
11
12
13
14
15

20
21
22
23

30
31
32
33
34
35

50
51
52
53
54

STRING FULL$(32), F1$(30), F$, MSB$, LSB$, MSB2$, LSB2$
STRING BXS, Q$, MATE1$C30), MATE2$(30), XX$(30)

STRING 0LDPOP$C30,30), NEWPOP$(30,30), 0FF1$<30), 0FF2$(30)
STRING A$(30), A1$(30), A2$(30), B$<30)

STRING Bl$(30), B2$(30), C$(30), D$(30)

STRING MX$(30), TEMP$(30), MUT$, MAX$<30), BREEDPL$(30,60)

INTEGER LS2B, MS2B, MSlB, LS1B

INTEGER STRINT, 00, SUN, GEN, II

INTEGER ISEED, N, M, X, Y, ZX, ZZ, IRND, 11, I2

INTEGER PDPSIZ, I, J, K, IMUT, AA, BB, JJ, KK, PP, P1, P

REAL SEED, XRND, TFIT, TMAX

REAL RDG, VOLTS, CH(10), CC, V1

REAL FIT(30), OBJ, CMULT, DELTA, AF, BF, HH
REAL PCRDSS, PMUT, FRAC(30), RR

REAL SUMFIT, FMIN, FMAX, AVG, EE

REAL GODDFIT

POPSIZ = 30 ’BE SURE TO CHANGE SIZE OF STRING TO MATCH
PCROSS = 0.7: PMUT = 1./POPSIZ

CH(1) = 3.05 ’CHANNEL ARRAY TO BE USED WITH KEYPAD

GEN = 1

CMULT = 1.5: GODDFIT = 0.9

1020 ’This is the seed for a random number generator
1021 SEED = .45342

1022 DEF XRNDz’Randomly generates a number between 0 and 1

67

1023
1024
1025

1026
1027
1027
1028

1040
1041

1044
1045
1046
1047
1048
1049
1050
1051
1052

1200
1220
1221
1222
1224
1225
1226
1227

1230
1231
1232
1234
1235
1236

SEED = SEED * 997. : ISEED = SEED : SEED = SEED - ISEED
XRND = SEED

FNEND

DEF IRND(I1,I2):

’Randomly selects an integer between I1 and I2

IRND = I1 + XRND * (I2 - 11 +1)

FNEND

’This function converts a binary string

’to an equivalent 8-bit integer

DEF STRINT(BX$)

00 = 128: SUM = 0

FOR X= 1 T0 8

Q$ = MID$(BX$,X,1)

IF GS = "1" THEN SUM = SUM + 00
00 = 00/2

NEXT X

STRINT = SUM

FNEND

’The next two functions perform crossover of two chromosomes
DEF C$ (A$,B$)

IF N = 30 THEN C$ = A$:GOT0 1226

IF N = 0 THEN C$ = A$:GOT0 1226

A1$ = MID$(A$,1,N)

82$ = MID$CB$,N+1,M)

C$ = CONCAT$(A1$,B2$)

FNEND

DEF D$ (A$.B$)

IF N = 30 THEN D$ = B$zGOT0 1236
IF N = 0 THEN D$ = B$:GOT0 1236
A2$ = MID$(A$.N+1.M)

81$ 8 MID$(B$,1,N)
DS = CDNCAT$(81$,A2$)

68

1237

1240
1241
1242
1243

1263

1264
1265
1266
1267
1268
1269
1270
1275
1280
1285
1290

1400

1415
1416
1417

1500
1505
1515

2999
3000
3100
3101
3103
3110

FNEND

DEF OBJ(V1)

CC = CH(1)

OBJ = 0.4 / ((V1-CC)*(V1-CC)+0.4)
FNEND

’ ***************************************************

’Initializing the population

FOR I = 1 T0 PDPSIZ

NEWPOP$(I) = ""

FOR J = 1 T0 30

IF XRND < 0.5 THEN NEWPOP$(I) = CONCAT$(NEWPOP$(I),"0"): GOTD 1270
NEWPOP$(I) = CONCAT$<NEWPOP$(I),"1")
NEXT J

NEXT I

FOR K = 1 T0 30

0LDPOP$(K) = NEWPOP$(K)

NEXT K

INITLCD 2,2,4

’The following commands initialize U14 and U15 as output ports
OUT 179, 128
OUT 195, 128

START 1,5
START 2
GOTO 1515

'*****************************************************

’The following task reads and interprets the AGC Voltage level
TASK 1

’The following commands initialize the LCD display

DEVICE 6

RDG = ADC(4)

69

3120
3200

3999
4000
4011
4020
4030
4040
4041

4050
4060
4070
4080

4135
4136
4141
4150
4160
4170

4187
4188
4189
4190
4200
4210

4213
4214
4215
4216
4217
4220
4225

VOLTS = (RDG*5.)/1024.

EXIT
’*****************************************************
’Task 2 evaluates a 30 bit binary string & outputs it
TASK 2

FOR Y = 1 T0 PDPSIZ

F1$ = 0LDPOP$(Y)

’32 bit strng to be broken into 4 8-bit strings:

FULL$ = CONCAT$("00",F1$)

MSB$ = MID$<FULL$,1,8)
MSB2$ = MID$(FULL$,9,8)
LSB2$ = MID$(FULL$,17,8)
LSB$ = MID$(FULL$,25,8)

’Using a function defined above,

’converting the strings to integers
LSlB = STRINT(LSB$)

LS2B = STRINT<LSB2$>

MS2B = STRINT(MSB2$)

MSlB = STRINT(MSB$)

’The following commands reset and set
’relays based on above integers

OUT 177, PEEK(ADR(LSIB))
OUT 178, PEEK(ADR(LS2B))
OUT 193, PEEK(ADR(MS2B))
OUT 194, PEEK(ADR(MSlB))

’The following commands send information to the LCD display
DEVICE 6

OUTLCD 192

PRINT " ";

OUTLCD 192

PRINT "JSB ";

OUTLCD 196

70

4230 PRINT PEEK(ADR(LSIB));

4233 OUTLCD 199

4234 PRINT " J5C ";

4235 OUTLCD 204

4240 PRINT PEEK(ADR(LS2B));

4243 OUTLCD 207

4244 PRINT " J7B ";

4245 OUTLCD 212

4250 PRINT PEEK(ADR(MS2B));

4253 OUTLCD 215

4254 PRINT " J70 ";

4255 OUTLCD 220

4260 PRINT PEEKCADR(MSlB));

4270 OUTLCD 222

4271 PRINT " ";

4272 FIT(Y) = OBJCVOLTS)

4275 TMAX = 0

4276 TFIT = 0BJ(VOLTS)

4277 OUTLCD 128

4278 PRINT TFIT;

4279 IF TFIT > TMAX THEN TMAX = TFIT
4280 IF TFIT >= GODDFIT THEN GOTO 4276
4281 FIT(Y) = TMAX

4380 IF Y = PDPSIZ THEN GOSUB 5000
4381 IF Y = PDPSIZ THEN GOSUB 5300
4900 OUTLCD 154

4910 PRINT " ";Y;" ";

4999 NEXT Y

5000 ’Subroutine that performs preselection, selection and crossover

5002 ’******************************************************

5003 ’COMPUTE AND DISPLAY STATISTICS
5004 SUMFIT = FIT(1)

5005 FMIN = FIT(1)
5006 FMAX = FIT(1)
5007 MAX$ = 0LDPOP$(1)

71

5008
5009
5010
5011
5012
5013
5014
5015

5020
5021
5025
5030
5035
5040
5050
5060
5070
5080
5085
5090
5098

5100
5101
5102
5103
5104
5105
5106
5107
5108
5109

5110
5111
5112

FOR JJ = 2 TO 30

EE = FIT(JJ)

SUMFIT = SUMFIT + EE
IF EE > FMAX THEN FMAX
IF EE < FMIN THEN FMIN
NEXT JJ

AVG = SUMFIT/30.

START 3

EE: MAX$ = OLDPOP$(JJ)
EE

’FITNESS SCALING ALGORITHM

IF FMAX < 1.5 * AVG THEN GOTO 5100

DELTA = FMAX - AVG: IF DELTA <= 0.0 THEN DELTA
AF (CMULT - 1.0)* AVG / DELTA

BF (1.0 - AF)* AVG

HH AF * FMIN + BF

IF HH > 0.0 THEN GOTO 5080

DELTA = AVG - FMIN: IF DELTA <= 0.0 THEN DELTA
AF = AVG / DELTA: BF = -AF * FMIN

FOR II = 1 TO POPSIZ

FIT(II) = AF* FITCII) + BF

IF FIT(II) <= 0.0 THEN FITCII) = 0.00001

NEXT II

0.00001

0.00001

’BEGIN PRESELECTION

KK = 0

FOR P = 1 TO POPSIZ

IF FIT(P) >= (1.2*AVG) THEN FITCP) = 2. * FIT(P)
RR = FIT(P) / AVG : PP = RR

FRAC(P) = RR - PP

IF PP <= 0 THEN GOTO 5109

FOR P1 = 1 TO PP: KK = KK + 1

BREEDPL$<KK> = OLDPOP$(P): NEXT P1

NEXT P

P = 0

IF KK >= POPSIZ THEN GOTO 5190
P = P + 1: IF P > POPSIZ THEN P = 1

72

5113 IF FRACCP) < 0 THEN GOTO 5111

5114 IF XRND > FRAC(P) THEN GOTO 5111

5115 KK = KK + 1 : BREEDPL$CKK) = OLDPOP$(P)
5116 FRAC(P) FRAC(P) - 1.

5117 GOTO 5111

5150 ’END PRESELECTION

5180 ’BEGIN SELECTION AND CROSSOVER
5190 FOR AA = 1 TO (POPSIZ/2)

5191 N IRND(1,30)

5192 M 30 - N

5193 ZX = IRND(1,POPSIZ)

5194 MATE1$ = BREEDPL$(ZX)

5195 BREEDPL$(ZX) = BREEDPL$(POPSIZ)
5196 POPSIZ = POPSIZ - 1

5197 22 = IRND(1,POPSIZ)

5198 MATE2$ = BREEDPL$(ZZ)

5199 BREEDPL$(ZZ) = BREEDPL$(POPSIZ)
5200 POPSIZ = POPSIZ - 1

5205 IF XRND > PCROSS THEN OFF1$ = MATEIS: OFF2$ = MATE2$z GOTO 5250

5215 OFF1$ = C$(MATE1$,MATE2$)
5216 OFF2$ = D$(MATE1$,MATE2$)

5250 NEWPOP$<AA) = OFF1$

5260 NEWPOP$(AA+15) = OFF2$

5262 NEXT AA

5263 POPSIZ = 30

5264 ’END SELECTION AND CROSSOVER

5265 ’Update the population
5270 FOR K = 1 TO POPSIZ
5280 OLDPOP$(K) = NEWPOP$(K)
5290 NEXT K

73

5291
5295

5300
5301
5302
5310
5320
5330
5340
5350
5360
5370
5380
5382
5383
5384
5385
5390
5395
5398
5400
5499

5999
6000
6014
6015
6017
6018

GEN = GEN + 1
RETURN

’Subroutine that performs mutation of chromosomes

FOR BB = 1 TO POPSIZ

MX$ = OLDPOP$(BB)

IF XRND > PMUT THEN GOTO 5395

IMUT = IRND (1,30)

IF MID$(MX$,IMUT,1) = "0" THEN MUT$ = "1": GOTO 5350

MUT$ = "0"

IF IMUT 1 THEN MX$ = CONCAT$(MUT$,MID$(MX$,2,29)):GOTO 5390
IF IMUT 3O THEN MX$ = CONCAT$(MID$(MX$,1,29),MUT$):GOTO 5390
TEMP$ = CONCAT$(MID$(MX$,1,IMUT-1),MUT$)

MX$ = CONCAT$(TEMP$,MID$(MX$,IMUT+1,30-IMUT))

OLDPOP$(BB) = MX$

OUTLCD 138

PRINT " ";

OUTLCD 138

PRINT "Mutate: ";IMUT;

NEXT BB

Y = 0

RETURN

EXIT

’Task 3 outputs statistics computed in task 2
TASK 3

PRINT "GENERATION #: ";GEN; " FMIN: ";FMIN
PRINT " FMAX: ";FMAX;" AVG= ";AVG

CANCEL 3

EXIT

74

IXFH?EEVEHD(]B

BASIC CODE: NETWORK ANALYZER CONTROL

Professional BASIC code for controlling the HP 8510C network analyzer
via HPIB, and for communicating with the FlashTrax controller via
serial connection. Sets all 24 switches to random states as determined
by random number read from file listr.txt.

Program: SSA_ED2.BAS

DECLARE SUB RunTest (isc&, device&)

DECLARE SUB Calibrate (isc&, device&)
DECLARE SUB ErrorTrap(device&)

DECLARE SUB FreqList (isc&, device&)

DECLARE SUB GetInfo ()

DECLARE SUB greeting ()

DECLARE SUB hp85108etup (isc&, device&)
DECLARE SUB init (isc&, device&)

DECLARE SUB Ioouts(device&, cmd\$)

DECLARE SUB MeasImpedance (isc&, device&, real.Z!, im.Z!)
DECLARE SUB MeasSWR (isc&, device&, swr!)
DECLARE SUB ReadyToMeasure (isc&, device&)
DECLARE SUB SpecifyParameters (isc&, device&)
DECLARE SUB waithere (t!)

DECLARE SUB randm ()

COMMON SHARED isc&, device& COMMON SHARED F

COMMON SHARED nstatesZ

COMMON SHARED twait

COMMON SHARED rowZ, col%

COMMON SHARED mach, maer, maxv% ’for download data

REM SINCLUDE: ’bigsetup’
’ This program performs automated measurements of
’ self-structuring antenna

’ using randomly-generated states.

’ The measurements are performed using an HP 8510 Network Analyzer.

75

’ Version of 15 September 2001

’ HP 8510 measurement implementation written by Chris Coleman from
’ HP 8720 measurement routines

’ HP 8720 initialization and error checking subroutines written
’ by John E. Ross III (taken from nam3.bas)

’ Modification for use with FlashTrax controller (www.multitrax.com)
’ by Ed Rothwell

REM Num% = RegisterFonts%("TMSRB.FON")
REM NumX = LoadFont%("N1/N2/N3/N4/N5/N6")

OPEN "debug.txt" FOR OUTPUT AS #10

twait = .03

’ row% and col% are variables used by the DrawSSA and DrasSSAGrid
’ subroutines.

row% = 4

col% = 1

’ Define the address of the HPIB interface (isc&) and
’ the HP 8510 (device&) isc& = 7 device& = isc& * 100 + 16

’ These variables are used for downloading data via HPIB

maxc% = 10

maxr% = 20

maxv% = 4096

’ ----- Start here -------

’set up serial communications with controller

OPEN "com2:2400,n,8,1,cd0,dsO,cs0,op0,rs,rb2048" FOR OUTPUT AS #1

76

CALL greeting

CALL GetInfo

CALL hp85108etup(isc&, device&)
CALL RunTest(isc&, devicek)

’ Put the HP8510 into continual measurement mode when completed
’ cmd$ = "cont;"

’ CALL Ioouts(device&, cmd$): CALL ErrorTrap(device&)

CLOSE #1 CLOSE #10 CLS

END

SUB Calibrate (isc&, device&)

CLS

PRINT "Calibration Options for HP 8510 Network Analyzer:"

PRINT " 0: Recall calibration from register 5 (default)"
PRINT " 1: Fully manual calibration using front panel"
INPUT "Choice: "; caltype$

SELECT CASE caltype$

CASE "", "O"
PRINT "Recalling calibration coefficients from register 5"
cmd$ = "reca5;"
CALL Ioouts(device&, cmd$): CALL ErrorTrap(device&)
CASE "1"
PRINT
PRINT
PRINT
PRINT "Press LOCAL on HP 8510, calibrate $11 1-port,"
PRINT "and press any key when finished"
WHILE INKEY$ = ""

77

WEND
END SELECT
CLS
PRINT "Calibration complete."
INPUT "Save coefficients to register 5? (Y/N) (default=N): "
INPUT calcoef$
SELECT CASE calcoef$
CASE II II , Ilnll , "N"
CASE llyll’ "Y"
’ Save the result in register 5.

cmd$ = "save5;"

CALL Ioouts(device&, cmd$): CALL ErrorTrap(device&)

END SELECT
END SUB
SUB ErrorTrap (device&)

’ This subroutine checks for errors
’ Check for BASIC or HPIB errors

GOSUB errorhp
’ Check for 8510C errors by checking status of 8510C Error que
chkque:
’ Read the status byte of the 8510C analyzer using the serial poll.

CALL iospoll(device&, statZ): GOSUB errorhp

78

LOCATE 23, 5: PRINT "ERROR ... STATUS = "; statX;
SLEEP (1)

’ If bit three of the status byte is set,

’ read the error queue summary list.
IF (statZ MOD 16) > 7 THEN
’ Tell 8510C to send out oldest error message in queue.

cmd$ = "outperro;"
CALL Ioouts(device&, cmd$): GOSUB errorhp

’ Receive error messages.
length% = 50
errdata$ = SPACE$(length%)
CALL ioenters(device&, errdata$, lengch, actualZ): GOSUB errorhp
’ Extract the error number from the string read in.
errnum% = VAL(LEFT$(errdata$, 5))
’ Initialize the string counter to begin after the error number.
1% = 9
’ Initialize the error message string.
errid$ = ""
’ Extract the error message from the string one character at a time.
DO UNTIL MID$(errdata$, 1%, 1) = CHR$(34)

errid$ = errid$ + MID$(errdata$, i2, 1)
i'/.=i'/.+1

79

LOOP
’ Display error message on status line.

LOCATE 23, 5: PRINT "HP 85100 ERROR "; errnumZ; ": "; errid$;
’ Beep to warn operator.

BEEP
’ Pause and then recheck status byte.

SLEEP 1

GOTO chkque

END IF
errorhp: IF PCIB.ERR <> NOERR THEN ERROR PCIB.BASERR
END SUB

SUB FreqList (isc&, device&)

cmd$ = "editlist;"
CALL Ioouts(device&, cmd$): CALL ErrorTrap(device&)

cmd$ = "sadd;"
CALL Ioouts(device&, cmd$): CALL ErrorTrap(device&)

singlefreq! = F * 1!

cmd$ = "cwfreq" + LTRIM$(STR$(singlefreq!) + "MHz") + ";"
CALL Ioouts(device&, cmd$): CALL ErrorTrap(device&)

cmd$ = "sdon;"
CALL Ioouts(device&, cmdS): CALL ErrorTrap(device&)

cmd$ = "lisfreq"

80

CALL Ioouts(device&, cmd$): CALL ErrorTrap(device&)

cmd$ = "sseg1;"

CALL Ioouts(device&, cmd$): CALL ErrorTrap(device&)

END SUB

SUB GetInfo

CLS

INPUT "Enter frequency (MHz) to be investigated"; F
INPUT "Enter number of antenna states to check"; nstates%

END SUB

SUB greeting

CLS

PRINT "Welcome to the automated self-structuring antenna testing system"
PRINT
PRINT "Press any key to continue"

WHILE INKEYS = ""
WEND

END SUB

SUB hp85108etup (isc&, device&)
CLS
PRINT "Initializing HP 8510 Network Analyzer"
PRINT

PRINT
CALL init(isc&, device&)

81

PRINT "Got past Init. Press any key to continue"
WHILE INKEY$ = ""
WEND

CALL SpecifyParameters<isc&, device&)

PRINT "Got past SpecifyParameters. Press any key to continue"
WHILE INKEY$ = ""

WEND

CALL FreqList(isc&, device&)

PRINT "Got past Freqlist. Press any key to continue"
WHILE INKEY$ = ""

WEND

CALL Calibrate(isc&, device&)

PRINT "Got past Calibrate. Press any key to continue"
WHILE INKEY$ = ""

WEND

CALL ReadyToMeasure(isc&, device&)

PRINT "Got past Readytomeasure"

PRINT "HP 8510 now set up. Press any key to continue"
WHILE INKEY$ = ""

WEND

END SUB

SUB init (isc&, device&)

This subroutine initializes the HPIB and HP 8510C Network Analyzer.

Reset the HPIB interface

CALL ioreset(isc&)

Define a system time-out of 10 seconds

82

time! = 10
CALL iotimeout<isc&, time!): CALL ErrorTrap(device&)

’ Abort any HPIB transfers
CALL ioabort(isc&): CALL ErrorTrap(device&)
’ Clear the HPIB interface
CALL ioclear(isc&): CALL ErrorTrap(device&)
’ Disable End-Or-Identify mode for transferring data
CALL ioeoi(isc&, 0): CALL ErrorTrap(device&)
’ Preset the 85100
CALL Ioouts(device&, "pres;"): CALL ErrorTrap(device&)
’ Enable HPIB debug mode for 85100
CALL Ioouts(device&, "debuon;"): CALL ErrorTrap(device&)
END SUB
SUB Ioouts (device&, cmd$)
’ This subroutine outputs a string command to the HPIB device

length% = LEN(cmd$)
CALL iooutputs(device&, cmd$, lengch)

END SUB

SUB MeasImpedance (isc&, device&, real.Z!, im.Z!)

83

 

’ This is the important measurement

END SUB

cmd$ = "sing;"
CALL Ioouts(device&, cmd$): CALL ErrorTrap(device&)

DIM impedance(9)

10

act.elem% 0

DIM frqd1!(1 TO 2, 0 TO 1), xr(0), xi(0)
STATIC frqdl!

max.e1em%

t0! = TIMER

cmd$ = "outpmark;"

CALL Ioouts(device&, cmd$): CALL ErrorTrap(device&)
CALL ioenter(device&, reply.1!, reply.2!, stim!)
CALL ErrorTrap(device&)

t! = TIMER - t0!

real.Z! = reply.1!

im.Z! = reply.2!

PRINT

PRINT

PRINT "Impedance = "; real.Z!; " + j"; im.Z!; " Ohms"
PRINT "Transfer time = "; t!; " seconds"

PRINT "Stimulus = "; stim!

PRINT

PRINT "press any key to continue"

WHILE INKEY$ = ""

WEND

SUB MeasSWR (isc&, device&, swr!)

’ This is the important measurement

cmd$ = "sing"
CALL Ioouts(device&, cmd$): CALL ErrorTrap(device&)

84

cmd$ = "outpmark"

CALL Ioouts(device&, cmd$): CALL ErrorTrap(device&)
CALL ioenter(device&, reply!): CALL ErrorTrap(device&)
swr! = reply!

END SUB
SUB ReadyToMeasure (isc&, device&)
’ cmd$ = "lisfreq"

’ CALL Ioouts(device&, cmd$): CALL ErrorTrap(device&)
’ cmd$ = "sseg1;"

’ CALL Ioouts(device&, cmd$): CALL ErrorTrap(device&)

’ cmd$ = "reca5;"

’ CALL Ioouts(device&, cmd$): CALL ErrorTrap(device&)

’ cmd$ = "swet?;"

’ CALL Ioouts(device&, cmd$): CALL ErrorTrap(device&)

’ CALL ioenter(device&, reply!): CALL ErrorTrap(device&)
’ PRINT

’ PRINT "Sweep time after single seg= "; reply!
cmd$ = "markl " + LTRIM$(STR$(F * 1!) + "MHz") + ";”
CALL Ioouts(device&, cmd$): CALL ErrorTrap(device&)
’ cmd$ = "smic;"
’ CALL Ioouts(device&, cmd$): CALL ErrorTrap(device&)

’ CALL MeasImpedance(isc&, device&, real.Z!, im.Z!)

’ CALL MeasMagPhase(isc&, device&, mag!, phase!)
’ PRINT "press any key to continue"
’ WHILE INKEY$ = ""
’ WEND
cmd$ = "swr;"

CALL Ioouts(device&, cmd$): CALL ErrorTrap(device&)

’ CALL MeasSWR(isc&, device&, swr!)
’ PRINT "SWR = "; swr!

END SUB

85

SUB RunTest (isc&, device&)

CLS

’RANDOMIZE TIMER

OPEN "result.dat" FOR OUTPUT AS #3
OPEN "listr.txt" FOR INPUT AS #4

’listr contains a file of random bytes stored as integers

’listr has no repeated 24-bit strings

’ Write Header to files "result.dat" and "test.dat"

PRINT #3, "This is an output file for test of the SSA"

PRINT #3, "using a random setting of states."

PRINT #3, "Date test was performed: "; DATE$, "Time of test: "; TIME$
PRINT #3, "f = "; F; "MHz"

PRINT #3, "no. states="; nstates%

PRINT #3, "Using file listr.txt"

’Begin test BEEP PRINT :

PRINT

PRINT "Ready to perform test."

PRINT "Attach antenna to test port."
PRINT

PRINT "Press any key to continue"
WHILE INKEY$ = ""

WEND

BEEP

CLS

FOR i% = 1 TO nstatesX

86

’Generate random byte for bank 1
p1% = 0
kk% = 1
FOR j% = 1 TO 8
INPUT #4, iinX
IF iinX = 1 THEN
p1% = p1% + ka
PRINT #1, CHR$(1);
ELSE
PRINT #1, CHR$(0);
END IF
kk'/.=kk'/.*2
CALL waithere(.03)
NEXT jX
’Generate random byte for bank 2
p2% = O
ka = 1
FOR j% = 1 TO 8
INPUT #4, iinX
IF iinZ = 1 THEN
p2% = p22 + ka
PRINT #1, CHR$(1);
ELSE
PRINT #1, CHR$(0);
END IF
ka = kk% * 2
CALL waithere(.03)
NEXT j%
’Generate random byte for bank 3
p3% = O
kkx = 1
FOR j% = 1 TO 8
INPUT #4, iinX
IF iin% = 1 THEN
p3% = p3% + ka
PRINT #1, CHR$(1);
ELSE

87

PRINT #1, CHR$(0);
END IF
ka = ka * 2
CALL waithere(.03)
NEXT jX

’ Measure SWR

’ and calculate reflection (p) and transmission (3) coefficient
CALL MeasSWR(isc&, device&, swr!)
p! = (swr! - 1) / (swr! + 1)

s! = 1 - p!

refl! = p!

IF 8! < 0 THEN
s! = .01
refl = 1 - 8!

END IF

PRINT #3, ix, p1%, p2%, p3%, swr!

’write results to screen
LOCATE 1, 1
PRINT "Bank1="; p1%; "Bank2="; p22; " Bank3="; p3%

LOCATE 4, 64

PRINT "num ="; 1%

LOCATE 5, 64

LOCATE 2, 25

PRINT " "
LOCATE 2, 25

PRINT "8 ="; s!

LOCATE 3, 25

PRINT " "
LOCATE 1, 52

PRINT "swr="; swr!

NEXT 1%

88

 

LOCATE 6, 12

PRINT "Test Complete. Press any key to continue"
WHILE INKEY$ = ""

WEND

BEEP

CLOSE #3

END SUB
SUB SpecifyParameters (isc&, device&)

’ Define start frequency
startfreq! = F * 1!
cmd$ = “star" + LTRIM$(STR$(startfreq!) + "MHz") + ";"
CALL Ioouts(device&, cmd$): CALL ErrorTrap(device&)

’ Define stop frequency
stopfreq! = 1.5
cmd$ = "stop" + LTRIM$(STR$(stopfreq!) + "GHz") + ";"
CALL Ioouts(device&, cmd$): CALL ErrorTrap(device&)

’ Define the number of points
nop% = 3
cmd$ = "poin" + LTRIM$(STR$(nop%)) + ";"
CALL Ioouts(device&, cmd$): CALL ErrorTrap(device&)

’ Turn on Averaging
cmd$ = "averon128" + ";"
CALL Ioouts(device&, cmd$): CALL ErrorTrap(device&)

” Define the number of averages

’ nav% = 4

’ cmd$ = "opc?;averfact" + LTRIM$(STR$(nav%)) + ";"

’ CALL Ioouts(device&, cmd$): CALL ErrorTrap(device&)

89

‘5.

 

 

END SUB

SUB waitforsrq (issc&, dev&)
STATIC SHARED PCIB.ERR, PCIB.BASERR, NOERR

squ = 1
checkstat: CALL iostatus(issc&, srqfl, statusZ)
IF PCIB.ERR <> NOERR THEN ERROR PCIB.BASERR
IF status% = 0 THEN GOTO checkstat
CALL iospoll(dev&, responseZ)
IF PCIB.ERR <> NOERR THEN ERROR PCIB.BASERR
IF (responseZ AND 68) <> 68 THEN GOTO checkstat
END SUB
SUB waithere (t)
’Pauses the program for t seconds ’ t can be less than 1

t1 = TIMER WHILE TIMER - t1 < t WEND

END SUB

90

1XPU?EHNI)IXL(3

BASIC CODE: SSA CONTROL FOR SWITCH FAILURE ANALYSIS

Program: swrprog2.txt

200 GOSUB 5000

210 GOSUB 5100

220 GOSUB 5200

300 LPRINT"SSA SWR Measurement Program"
400 GOSUB 5100

401 IF ABUFFER<8 THEN GOTO 401

402 p1=0:k=1:for i=1 to 8

403 p=aread:If p=1 then p1=p1+k

404 k=k*2:next i

405 IF ABUFFER<8 THEN GOTO 405

406 p2=0:k=1:for i=1 to 8

407 p=aread:If p=1 then p2=p2+k

408 k=k*2:next i

409 IF ABUFFER<8 THEN GOTO 409

410 p3=0:k=1:for i=1 to 8

411 p=aread:If p=1 then p3=p3+k

412 k=k*2:next i

420 ’Set i/o pins

422 OUT 192,P1:OUT 193,P2:OUT 194,P3
425 GOSUB 5050

426 LPRINT " Bl=";P1;" B2=";P2;" 83=";P3
440 GOTO 400

500 END

5000 ’Initialize LCD display

5010 OUT 224,56:OUT 224,56

5020 OUT 224,6:OUT 224,12

5030 RETURN 5050 ’Clear LCD display
5060 OUT 224,1

5070 RETURN

5100 ’Initialize auxilliary com port
5110 ACONFIG 2N81

91

 

 

5120 RETURN
5200 ’Initialize i/o pins and set all pins low

5210 OUT 195,128
5211 OUT 193,0:OUT 192,0:OUT 194,0 5220 RETURN

92

'x

 

 

JAFHPEHWEHD(13

CPLD CODE: CONTROL ALGORITHM FOR THE RESETTING OF
ILAEFCHJINHSIRJELADYS

Program for the resetting of latching relays used on the SSA control board. Written
in VHDL by Matt Freel as a part of an ECE capstone design.

entity rectrl is
port( a: in std_logic;
b: in std_logic;
s: in std_logic;
01: out std_logic;
02: out std_logic);

end rectrl;
architecture rectrl_behave of rectrl is

begin
ol<=(s and a)or(not s and b);
O2<=(not s and a)or s and b);

end rectrl_behave;

entity cont is
port( i: in STD_LOGIC_VECTOR (1 to 15);
01: out STD_LOGIC_VECTOR (1 to 15);
02: out STD_LOGIC VECTOR (1 to 15);
s: in STD_LOGIC );

end cont;

architechture cont_arch of cont is component rectrl is
port ( a: in std_logic;
b: in std_logic;
s: in std_logic;
oi: out std_logic;
02: out std_logic);

end component;

93

 

signal zero: std_logic;

begin

r1:
r2:
r3:
r4:
r5:
r6:
r7:
r8:
r9:

rectrl
rectrl
rectrl
rectrl
rectrl
rectrl
rectrl
rectrl

rectrl

port
port
port
port
port
port
port
port
port

r10:
r11:
r12:
r13:
r14:
r15:

end c

rectrl port
rectrl port
rectrl port
rectrl port
rectrl port
rectrl port
ont-arch;

map(i(1),
map(i(2),
map(i(3),
map(i(4).
map(i(5),
map(i(6),
map(i(7),
map(i(8),
map(i(9), zero,

zero,

0

zero,

zero,

zero,

zero,

zero,

U

zero,

zero,

mmmmmmmmm

b

map(i(10), zero,
map(i(11), zero,
map(i(12).
map(i(13).
map(i(14),
map(i(15).

zero,
zero,
zero,

zero,

01(1).
01(2).
01(3),
01(4),
01(5).
01(6).
01(7),
01(8),
01(9).

94

02(1));
02(2));
02(3));
02(4));
02(5));
02(6));
02(7));
02(8));
02(9));

01(10),
01(11).
01(12),
01(13),
01(14),
01(15),

o2(10));
o2(11));
o2(12)):
o2(13));
02(14));
02(15));

BIBLIOGRAPHY

95

 

BIBLIOGRAPHY

[1] CM. Coleman, Self-Structuring Antennas, Ph.D. Dissertation, Michigan State
University, East Lansing, MI, 2002.

[2] CM. Coleman, E.J. Rothwell, and J .E. Ross, “Self-structuring antennas,” IEEE
AP-S International Symposium and URSI Radio Science Meeting, Salt Lake City,
Utah, July 16—21, 2000.

[3] J .E. Ross, E.J. Rothwell, C.M. Coleman, and LL. Nagy, “Numerical simulation
of self-structuring antennas based on a genetic algorithm optimization scheme,”
IEEE AP-S International Symposium and URSI Radio Science Meeting, Salt
Lake City, Utah, July 16-21, 2000.

[4] CM. Coleman, E.J. Rothwell, J.E. Ross, and LL. Nagy, “Self-Structuring An-
tennas,” IEEE Antennas and Propagation Magazine,, to appear, June 2002.

[5] RT. Perry, C.M. Coleman, B.F. Basch, E.J. Rothwell, J.E. Ross, and LL. Nagy,
“Self-Structuring Antenna for Television Reception,” IEEE AP-S International

Symposium and URSI Radio Science Meeting, Boston, Massachusetts, July 8-13,
2001.

[6] CM. Coleman, E.J. Rothwell, J .E. Ross, and LL. Nagy, “Application of Two-
Level Evolutionary Algorithms to Self-Structuring Antennas,” IEEE AP-S Inter-
national Symposium and URSI Radio Science Meeting, Boston, Massachusetts,
July 8-13, 2001.

[7] C. Coleman, B. Perry, E. Rothwell, L. Kempel, J.E. Ross, and L. Nagy, “A
Study of Simple Self-Structuring Antenna Templates,” IEEE AP-S International
Symposium and URSI Radio Science Meeting, San Antonio, TX, June 16-21,
2002.

[8] B. Perry, C. Coleman, E. Rothwell, L. Kempel, J. Ross, and L. Nagy, ” Effect of
Switch Failure on the Performance of a Self-Structuring Antenna,” IEEE AP-S
International Symposium and URSI Radio Science Meeting, San Antonio, TX,
June 16-21, 2002.

[9] DE. Goldberg, Genetic Algorithms in Search, Optimization, and Machine Learn-
ing. Reading, Mass: Addison-Wesley, 1989.

[10] K. Mehrotra, C.K. Mohan, S. Ranka, Elements of Artificial Neural Networks.
Cambridge, Mass: The MIT Press, 1997.

96

General References

[11] W. Medenhall, Introduction to Probability and Statistics, Duxbury Press, North
Scituate, MA, 1975.

[12] J.H. Holland, Adaptation in Natural and Artificial Systems. Ann Arbor: The
University of Michigan Press, 1975.

[13] R. Haupt, “An Introduction to Genetic Algorithms for Electromagnetics,” IEEE
Tiansactions on Antennas and Propagation, vol. 17, pp. 7-15, April 1995.

[14] D.S. Weile and E. Michielssen, “Geneic Algorithm Optimization Applied to Elec-
tromagnetics: A Review,” IEEE Transactions on Antennas and Propagation, vol.
45, pp. 343-353, March 1997.

[15] J .M. Johnson and Y. Rahmat-Samii, “Genetic Algorithms in Engineering Elec-
tromagnetics,” IEEE Antennas and Propagation Magazine, vol. 39, pp. 7-25,
August 1997.

[16] Y. Rahmat-Samii and E. Michielssen, eds., Electromagnetic Optimization by
Genetic Algorithms. New York: John Wiley & Sons, 1999.

[17] E. Bonabeau, M. Dorigo, and G. Theraulaz, Swarm Intelligence: From Natural
to Artificial Systems. New York: Oxford University Press, 1999.

97

[l]]]]]][][]][W