I

I LA-UR-89-1420

SULTRA-WIDEBAND RADAR
1
Research and Development
Considerations
~I

5 June 1989

I
I prepared for the
Naval Ocean Systems Center
San Diego, California

DTIC
ELEC TE
%67 JAN 10 19911 D
I by the

Mechanical and Electronic Engineering Division

ZaMMUION TATE1MY A

0• ILos Alamos Natonal Laboratory

Los Alamos,New Mexico 87545
al
I
I
GLOSSARY
I DoD Department of Defense
EM electromagnetic
EMP electromagnetic pulse
ESD Electronic Systems Directorate, USAF
FFT fast Fourier transform
FM frequency modulation
HPM high-power microwave
HDL Harry Diamond Laboratory
ICBM intercontinental ballistic missile
IEEE Institute of Electrical and Electronics Engineers
3 ISRD
LAMPF
Institutional Supporting Research and Development
Los Alamos Meson Physics Facility
LANL Los Alamos National Laboratory
LHS left-hand side
LO low observable
LPI low probability of intercept
MMW millimeter wave
NAVSEA Naval Sea Systems Command
NCTR noncooperative target recognition
NOSC Naval Ocean Systems Center
NRL Naval Research Laboratory
ODE ordinary differential equation Acession-For
ONT Office of Naval Technology NTIS GRAM
OPNAV Office of Naval Operations Unannounced [T
OTH over-the-horizon Justifxication
PCPS photoconductive power switch
PDE partial differential equation By _
PIN P-type--intrinsic-N-type Distribut on/
PRF pulse repetition frequency Availability Codas
RADC Rome Air Development Center Avail and/or
RCS radar cross-section Dist Special

SAR synthetic-aperture radar

TEM transverse electromagnetic
3 USN
UWB United States Navy
ultra-wideband
Statement "A"
per telecon Jan
Gallegos.
Los Alamos National Laboratory/OS-6.
I1VG
Los Alamos, NM 87545.
1/10/91
I ' •T • '/ ('4%N

Ap~rovod fmr publWh ze~am

DIa~rihution Lh~mbd
I

Executive Summary
This report provides an initial investigation into the technology of impulse radar.
3 The principal purpose is to provide a road map for exploratory development by
defining critical technology issues for a research investment strategy.
3 There are urgent requirements for improving radar capability for target identifi-
cation, or noncooperative target recognition (NCTR). Progress in NCTR has been
SI primarily in information processing, with few technology advances in the sensor
domain. Impulse radar provides new and revolutionary ideas, which can offer new
capabilities for active surveillance. These capabilities are discussed in this report
along with issues that must be resolved before system development is undertaken.

Background
Radar, in its many variations, is a mature and sophisticated technology. The
capabilities of current radar systems are, however, inadequate for many important
applications such as target identification, detection of "stealth" targets, and target
imaging.
Recent advances in high-speed high-power switching devices have provided the
technology base for innovative radar concepts that will discriminate targets by
radar signature. These impulse-radar concepts have lacked substantive investigation
3 heretofore.
The idea of transmitting an impulsive waveform of very high peak power with
3 a frequency ipectrum extending from almost dc to beyond 1 GHz introduces a
revolutionary concept for wideband radar design. There are several potential ad-
3 vantages and advanced capabilities of such a radar as compared with conventional
narrow-band radar.

I * High-resolution range measurement (1 nis = 15 cm)

e. Significant target-discrimination and target-identification capability

I
U
"* Detection of low-observable (LO) targets (low-RCS and stealth)

"*High-resolution target-imaging capability (single and multiple targets)

"*Improved penetration capability (earth, walls, foliage)

"*Ultra-short-range capability

"* Rejection of clutter and multipath interference through range gating

"• Low probability of intercept and low probability of exploitation

"• Reduced sensitivity to electronic-warfare countermeasures

Ultra-Wideband Technology
The technology of ultra-wideband (UWB) radar differs greatly from that of con-
ventional radar.

"* The relative bandwidth of impulse radar is large and requires analysis of the
launching, propagation, reflection scattering, and reception of transient elec-
tromagnetic (EM) waves rather than the much simpler steady-state sinusoids,
which suffice for narrow-band systems such as conventional radar.

"• New antenna designs must be undertaken for tiansient waves, which require
starting from basic principles (Maxwell's equations), as standard antenna the-
ory is based on sinusoidal excitation.

"• New devices, such as the laser-activated photoconductive power switch (PCPS),
require development for high-power pulse generation at high pulse-repetition
rates. Some of this development is already in progress.

"*New concepts are required in receiver and transmitter designs, such as the
channelized receiver combined with rise-time control of high-power pulse gen-
erators for controlling the radiation field.

"* Studies in materials science are required for UWB radomes and for dielectrics
used in the high-power sections of an impulse radar.

ii
". Emphasis is required on time-domain electromagnetics in measuring and com-
Iputing impulse response of targets, radar-absorbing materials, transmitting
and receiving antennas, radomes, and various media.

""New aignal-proceaseing algorithms must be developed for impulse radar.

. Potential applications of impulse radar require a variety of syatem. ana/y-

se and simulations to investigate optimum signal and antenna design, EM
response of targets, jamming and countermeasures, clutter rejection, false
alarms, EM compatibility, detectability, performance evaluations and com-
parisons.

Recommended Program of Impulse-Radar R&D

. A collection of R&D initiatives is recommended to examine the basic system

and technological issues requiring investigation prior to system development.
I Some of the more important tasks include (1) low-power short-range measure-
ments of transient EM response of various targets and materials, (2) transient
measurements of proposed antenna designs, (3) determination of the need for
revised PCPS designs to provide rise-time control, (4) study of the ability of
5 the channelized receiver to provide rapid selection of various matched filters
and to counter jamming, (5) study of radome materials for use over eytremely
I wide bands, and (6) EM analysis and computation for antenna dm.ign, tran-
sient scattering, and parametric system design.

* A parallel study of potential applications of impulse radar is needed to guide

the R&D program in technology and to select an application for early proto-
type development.

3 * With sufficient answers to the technological questions, initiate the develop-

ment of a prototype impulse radar for the application selected in the system
3 study.

I1
The broad range of critical areas that can be effectively addressed with this
research approach can be categorized as follows.

"*Overall system design, analysis and simulation

"* UWB-radar transmitting and receiving antennas

"* EM propagation of UWB waves

"* EM response of targets to impulse-radar signals

"* UWB-radar target identification

"* UWB-.radar imaging of targets

The research areas listed above are described in detail in the body of this report
and are essential to the development of advanced radar capabilities using impulsive
signals.

iv
* Contents
I
1 Introduction 1

2 Why Ultra-Wideband Radar? 7

i 2.1 Overview of Conventional Radar .................... 7
2.1.1 Characteristics and Applications ........................ 8
I2.1.2 Range Resolution .. .. .. .. .. .. .. .. .. .. .... ........ 8
2.1.3 Limitations of Conventional Radar .................... 9
2.2 Approaches to Ultra-Wideband Radar ...................... 9
2.2.1 Spread-Spectrum Radar ........................... 10
2.2.2 Impulse Radar ........ .......................... 11
S2.3 Potential of Ultra-Wideband Radar ........................ 12

3 Ultra-Wideband Radar 15
3.1 Theoretical Considerations ............................... 15
3.1.1 Time-Domain Electromagnetics ...................... 15
3.1.2 Emission, Propagation, and Penetration ............ .... 17
3 3.1.3 Electromagnetic Response of Targets .................. 19
3.1.4 Target Identification ............................. 19
3 3.1.5 Target Imaging ................................. 21
3.2 Ultra-Wideband Technology ............................. 22
3 3.2.1 Impulse-Radar Transmitters ........................ 22

I v
I
3.2.2 Antennas ............................. 25
3.2.3 Receivers ........ ............................. 26
3.2.4 Signal Processing ................................ 31
3.3 Potential Capabilities and Applications ...................... 34
3.3.1 Target Identification ............................. 34
3.3.2 Low-Observable Target Detection .................... 36
3.3.3 Target Imaging ................................. 37
3.3.4 Penetration Radar ....... ........................ 39
3.3.5 Short-Range Radarr ....... ....................... 40
3.3.6 Target/Decoy Discrimination ........................ 41
3.3.7 Intrusion Detection ............................... 42
3.3.8 Detection of Sea-Skimming Missiles ..... .............. 42
3.3.9 Detection of Cruise Missiles ......................... 43
3.3.10 Radar-Cross-Section Diagnostics ...................... 44
3.3.11 Synthetic-Aperture Radar .......................... 44
3.3.12 Low-Angle Radar ................................ 45
3.3.13 Low Probability of Intercept ...... .................. 46
3.3.14 Low Probability of Exploitation ...................... 46
3.3.15 Nonapplications ............ ..................... 46
3.4 System Design and Analysis ............................. 48
3.4.1 Optimum Signal Design ........................... 49
3.4.2 Jamming and Countermeasures ...................... 51
3.4.3 Clutter Rejection Analysis .......................... 51
3.4.4 Probability of Intercept and Exploitation ............... 54
3.4.5 Electromagnetic Compatibility ...................... 55
3.4.6 False Alarms, Detectability, and Missed Targets ........ .. 56
3.4.7 Overall Performance Analysis and Comparison ........... 59

vi
4 Critical Issues in Impulse Radar 61O
S4.1 Optimum Signal Design ......................... 61

4.2 High-Power "Impulse" Generators . .................. 63

1 4.3 Antenna Design .............................. 63
4.4 Transient Electromagntic Theory .......................... 64
1
4.5 Receiver Design ..................................... 6
4.6 Signal Processing mad Analysis ........................... 68

A Time-Domain Electromagnetics 69
3 A.I Maxwell's Equations ........... .... ..................... 70
A.2 Transient Electromagnetic Scattering ........................ 73

I A.3 Transient Excitation of Antennas .......................... 77

B Impulse-Iladar Signals 79
B.1 Preliminary Analysis ................................... 79
SB.2 An Alternate Analysis ................................. .80
B.3 Array Antennas for Impulse Radar ......................... 83
3 B.4 Time Variation of the Far-Zone Electric Field ............. 86
B.5 The Received Signal ........ ........................... 89
U B.6 The Matched-Filter Receiver ...... ...................... 90
B.7 Range Resolution of Impulse Radar ........................ 93
B.8 Detection with a Nonmatched Filter ........................ 94
B.9 Approximate Maximum Range ........................... 94
B.9.1 The Range Equation for Impulse Radar ............ 95
3 B.9.2 Maximum Range of Target Detection ................. 98

C Propagation of UWB Waves 101

I C.1 Evolution of EM Theory and Applications ................... 101
C.2 Propagation of Transient Signals .......................... 102
3C.3 The Harmuth Approach ................................ 103

3 vii
C.4 Transient Propagation in Lossy Media ..................... 104
C.5 Calculations ........ ................................ 106
C.6 Conclusion ........ ................................ 110

D Los Alamos National Laboratory 111

D.1 Capabilitim Relevant to Impulse Radar ................ 114
D.I.1 Projects and Facilities ...................... 114
D.1.2 Scientists and Engineers ........................... 116
D.2 Mechanical and Electronic Engineering ..................... 118
D.2.1 Electronics Technology Applications... . ............. 119
D.2.2 Short Biographies of the Authors ..................... 122

E Relevant Contacts 125

F References 127

viii
I
U

* List of Figures
I
3.1 Experimental arrangement for PCPS tests ................... 24
3 3.2 An energy detector for radar ....... ...................... 28
3.3 The cross-correlator detector ....... ...................... 30
3.4 The energy detector for impulse radar ...................... 58

B.1 Typical current pulse in one array element ................... 82

B.2 Time-derivative of current pulse in each array element ........ ... 82
B.3 Energy spectrum of radiated signal from each element of the array . 83
3 B.4 A linear array of N current elements ...... .................. 84
B.5 Directivity of ten-element array ........................... 85
I B.6 Far-zone electric field from a ten-element array ..... ............ 87
B.7 Zznergy spectrum from a ten-element array ................... 88
SB.8 Far-zone electric field from a ten-element array ............ .... 88
B.9 Energy spectrum from a ten-element array................... 89
B.10 Matched-filter output from a ten-element array for 4 = 00 ...... .... 91
B.11 Matched-filter output from a ten-element array for -= 50 ...... 92
1B.12 Matched-filter output from a ten-element array for 4, = 15 ..... ... 92
B.13 Output when two targets of equal size are present .... .......... 93
B.14 Single-pulse detection range with a matched-filter receiver ...... .100

C.1 Propagation of a transient wave in a loasy medium .......... ... 105

C.2 Propagation of a transient wave in a lossless medium ......... .. 107

D.1 Los Alarmos National Laboratory ...... .................... 113

I
D.2 Test of the RADC antenna ...... ....................... 121

|ix
I
I
I
I
Chapter 1

-| Introduction
I
Radar, in its multiplicity of realizations, is a mature and sophisticated technology.
The capabilities of current state-of-the-art radar systems are, however, inadequate
for many important advanced requirements, such as target identification, "stealth"
I target detection, target imaging, and material penetration. These inadequacies do
not necessarily result from l*n,'!h.-ions bi current hardware technology, but they
are largely due to limitations and difficulties inherent in the basic process of radar
itself. The inherent difficulties of the radar process, as a technique of remote sens-
ing, result primarily because the radar process is an inverse problem. Therefore,
extracting significant information about a target from the returned radar signal
is an exceedingly difficult and sensitive problem of a mathematical nature. Thz',t
is why the achievement of advanced radar capabilities will not result simply from
advancements in hardware technology but will also require advancements resulting
3 from theoretical research in fundamental areas such as EM theory, systei-as theory,
and signal analysis.
3 In order to make efficient progress in the development of any capability based
on a complex physical process, a fundamental and complete understanding of the
overall problem is necessary. In the case of advanced, remote-sensing capabilities
based on UWB radar, the processes involved are varied and highly complex. Al-
3 though much information can be obtained through experimental development, this

* 1
information can only allow an empirical, and therefore limited, approach to further
development. In order to gain a fundamental understanding of the complex pro-
cesses involved so that a knowledgeable and efficient approach to development can
be undertaken, it is necessary to formulate and solve the complex mathematical
problems that define these processes and to develop a theoretical understanding of
the problem. For this reason, the achievement of advanced radar capabilities based
on UWB approaches will require a strong theoretical effort.
A number of potentially significant approaches to the achievement of advanced
radar capabilities are based on the use of UWB signals, of which impul8s radar
is a specific example. Fundamentally, these approaches are based on the concept
of using signals having a temporal structure that is significantly different from
that of conventional radar signals. The temporal structure of a conventional radar
signal is often an amplitude-modulated sinusoidal wave. The modulation generally
takes the form of a broad envelope containing many cycles of the single-frequency
sinusoidal carrier. Therefore, the signal has a small relative bandwidth (narrow
power spectral density). UWB radar approaches are based on signals having a large
relative bandwidth (broad power spectral density). The concept of impulse radar
suggests the use of a signal consisting of very short pulses, i.e. "impulses" having
a large relative bandwidth. Many temporal structures for UWB radar signals are
possible, all of which remain to be examined carefully to determine their potential
benefits and capabilities. In this report, the broad term UWB radar is used to
include impulse radar as well as a number of other important concepts that are
not necessarily based on an impulse structure, such as "chirp" radar and other
spread-8pectrum approaches [Cooper and McGillem 1986].

2
Some of the potential advantages and capabilities that are attributed to UWB
radar as compared with conventional narrow-band radar are the following.

. High-resolution range measurement (1 ns =* 15 cm)

* Significant target-discrimination and target-identification capability

* Detection of low-observable (LO) targets (low-RCS and stealth)

• High-resolution target-imaging capability (single and multiple targets)

1 Improved penetration capability (earth, walls, foliage, rain)

. Ultra-short-range capability

* Rejection of clutter and multipath-interference through range gating

* Low probability of intercept (LPI) and low probability of exploitation

* Reduced sensitivity to electronic-warfare countermeasures

Of these items, target identification, LO target detection, and target imaging are
of particular importance; however, a detailed understanding of the best ways to
achieve these potential advantages and capabilities remains to be developed.
Many of the basic hardware requircrnentB for a UWB radar system are under-
stood in general and much of the required hardware technology is currently avail-
able. Therefore, many of the hardware aspects of UWB radar are of an engineering
specification and design nature. Other hardware aspects will require some engi-
neering research and development, but these steps cannot proceed effectively until
detailed design requirements are determined. The problem of immediate concern is
that fundamental questions regarding the basic process and potential capabilities
of UWB radar currently do not have clear and detailed answers. Many questions
must be answered before the system-design and hardware-development process can
be knowledgeably and effectively undertaken.
Of critical importance, therefore, is the formulation, solution, and analysis of
problems describing the fundamental physical processes, which are problems in
EM theory requiring a strong mathematical approach, and the development and

* 3
I
application of state-of-the-art system analysis and signal analysis. One goal of the
needed research is to provide for the knowledgeable design of a UWB-radar signal
structure for a particular application to obtain a high level of performance.I
The development of impulse radar is a multifaceted task requiring research in a
number of technical areas. Research to determine the fundamental characteristics
and advantages and the potential capabilities of impulse radar is essential at this
time. The research areas outlined in this report are designed to address a num-
ber of important and fundamental problems. This work is critical to the efficient

development of advanced radar capabilities based on UWB concepts.

The performance of theoretical research in UWB radar based on system mod-
eling and analysis, transient EM theory, and signal analysis will answer many im- -
portant questions. This work will result in a fundamental understanding that is
critical to the efficient development of advanced radar capabilities based on UWB [
concepts. Thks research approach entails complete and detailed "system modeling"
I
and is based on solving in detail the fundamental EM problems and on applying
state-of-the-art system analysis and signal analysis. In some cases, the basic ideas
under consideration can also be demonstrated with low-power proof-or-principle ex-

periments. This research can then be followed by engineering analysis and design.
The broad range of critical areas that can be effectively addressed with this

research approach can be categorized as follows:

* Overall system design, simulation, and analysis

- impulse signal design

- signal-encoding schemes (random, orthogonal, cyclic difference set, etc.)

- signal recognition and discrimination algorithms

- coherent matched-filter design and analysis
- noise and interference modeling and analysis
- signal intercept and detection analysis

4I
I
* UWB-radar transmitting/receiving antennas
i- UWB antennas for "impulive" signals
- arrays of UWB antennas
- antenna structures for launching "focus-wave" propagation modes

* EM propagation of UWB waves

- signal-propagation characteristics (attenuation, dispersion, skipping,
ducting, surface modes, etc.)
- focus-wave and "electromagnetic-missile" propagation modes
- propagation through materials, plasmas, rain, dust, etc.
- EM interactions with matter

i * EM response of targets to impulse-radar signals

- transient EM scattering theory
- transient EM response of LO targets
- screened and embedded targets, material penetration
- resonance, surface waves, "delayed" returns, etc.
- time-domain reflectometry and homomorphic system theory
- signal design for UWB-radar target identification

3
i UWB-radar target identification
- transient inverse EM scattering theory
3 - time-domain identification signatures
- spectral system identification
3 - parametric system identification
- singularity-expansion-method approaches
5 - resonance-mode and surface-wave-mode identification

. UWB-radar imaging of targets

- UWB synthetic-aperture approaches
- image reco.istruction as time-domain reflectometry
- image reconstruction by ellipsoidal backprojection

*5
These items represent a number of research areas that are essenti.il to the devel-
opment of advanced radar capabilities utilizing UWB signals. Many of these items
also represent areas of research that are currently of great interest in the disciplines
of electrical engineering and applied mathematics.
In conclusion, the advanced radar capabilities that are potentially achievable
using UWB approaches are currently uncertain and, therefore, aystem modeling,
mathematical analyis, and proof-of-principle ezperiments that elucidate potential
capabilities are of great importance. These topics and others will be addressed in
more detail in the remainder of this report.

6
Chapter 2

Why Ultra-Wideband Radar?

UWB radar is considered to be any radar system in which the relative bandwidth
of the signal is larger than about 10%. If the upper and lower useful frequencies of
the signal power spectral density are denoted by f. and f, respectively, the relative
bandwidth R is defined as (fu - fi)/(fu + fl). This definition is useful not only for
conventional radars but also for impulse radars where no carrier frequency appears.
To understand the theoretical basis for and the practical application of UWB radars
it is necessary to consider first the theoretical limitations on conventional sinusoidal-
carrier radars. These limitations suggest that certain aspects of radar performance
can be improved only by creating radar systems that use UWB signals.

2.1 Overview of Conventional Radar

Conventional radar is one in which a sinusoidal carrier is amplitude modulated by a
periodic sequence of short pulses. The sinusoidal carrier may or may not be coherent
from pulse to pulse. If the carrier is coherent from pulse to pulse, the radar is called
a pulse Doppler radar. Since the pulse Doppler radar is the most capable form of
conventional pulse radar, it is this form that is considered in this overview.

I 7
2.1.1 Characteristics and Applications

The conventional pulse Doppler radar measures targets in four coordinates: range,
range-rate, azimuth angle, and elevation angle [Golden 1987]. The capability of
conventional radar is frequently expressed in terms of its resolution and its accu-
racy. The resolution of a radar is a measure of its ability to separate responses from
targets differing from one another in one or more of the four coordinates. The re-
quired resolution in any coordinate depends on the application of the radar. Typical
applications of pulse Doppler radar include early warning, area search, surveillance,
target tracking, fire control, ground mapping, and target identification. Of these
applications, the last one requires the greatest resolution in range and angles. As
shown below, range resolution depends on signal bandwidth and this parameter is
under the control of the signal designer. In contrast, Doppler resolution depends
on the observation time, and angle resolution depends on antenna size. These pa-
rameters do not depend directly on the form of the transmitted signal.

2.1.2 Range Resolution

The range resolution of any radar depends entirely on the signal bandwidth and is
frequently defined as [Burdic 1968]

6 = 2-- (2.1)

where c - 3 x 108 m/s is the propagation velocity of light in free space. The
parameter P, is the effective bandwidth of the signal defined by
,6 ='R(0) (2.2)
=f R2 (r)dr(2)
where . (r) is the autocorrelation function of the signal pulse. The range-rate res-
olution depends primarily upon the time interval over which the signal is observed
and the angular resolutions depend on the radar antenna dimensions in the corre-
sponding directions and the wavelength of the carrier frequency.

8
The accuracy with which measurements can be made in each of the four coordi-
nates is also a function of the signal-to-noise ratio (SNR). For example, the range
accuracy may be expressed as [Burdic 1968]
OR = (2.3)

I where Eis the received signal energy, No is the one-sided noise spectral density at
the receiver input, and fo is the rms bandwidth defined by

S= (2) f)df
f f $ (M4 (2.4)

i in which S(f) is the energy-density spectrum of the signal. The range-rate accuracy
and the angular accuracy also depend on SNR in a similar way.

2.1.3 Limitations of Conventional Radar

The important point in the above equations is that both the range resolution and
3 the range accuracy depend inversely upon the signal bandwidth. Since these are
theoretical limits, no amount of clever signal design can replace this dependence on
bandwidth. If one wishes to resolve closely spaced targets in range, the only way to
do it is to increase the signal bandwidth. As seen below, both target identification
and target imaging require extremely good range resolution.
Another important characteristic of periodically pulsed signals is the ambiguity
function. This finction reveals target ranges and target velocities at which there
are ambiguous responses and indicates the existence of any coupling between range
and velocity measurements. Ambiguous responses impose serious constraints on
combinations of range and velocity that can be observed by the conventional radar.

2.2 Approaches to Ultra-Wideband Radar

There are two general approaches to increasing the signal bandwidth of a radar
system. The first approach is referred to as spread-apectrum and consists of designing

9
signal waveforms that have a large bandwidth without decreasing the duration of
each signal pulse. This requires some sort of modulation within the pulse. Such
signals have a large time-bandwidth product. The second approach is to reduce
the duration of each signal pulse. Signals of this type will have a time-bandwidth
product near unity.

2.2.1 Spread-Spectrum Radar

A number of different techniques can be employed to create spread-spectrum signals.

A brief discussion of each of these approaches is provided below.

1. Stepped FM

In this type of signal, the frequency of the carrier is hopped in a pseudorandom

fashion over a wide range of frequencies during each pulse. Although this
approach has advantages in clutter reduction and in jamming resistance, the
large range sidelobes make it unsuitable as a means for achieving good range
resolution.

2. Linear FM

The frequency of the carrier is swept linearly over a wide range of frequencies
during each pulse. Excellent range resolution can be achieved through this
approach, but a coupling between range and range-rate results in moving tar-
gets appearing at the wrong range unless suitable compensation is employed.
This approach is frequently referred to as chirp radar.

3. Nonlinear FM

Increasing the FM sweep rate near the edges of the pulse and decreasing it in
the center can be used to lowez the sidelobes of the signal spectrum. This is
advantageous in the design of the corresponding matched filter.

10
I
U
1 4. Pseudorandom Coding

The carrier is biphase modulated throughout the pulse by a pseudorandom

binary sequence. This approach avoids the range-Doppler coupling of the
FM signals, but may produce large range sidelobes unless the pseudorandom
sequence is very long.

1I 5. Hybrid Methods

It is also possible to combine two of the above methods in order to achieve

3 the advantages of each without (hopefully) the disadvantages of either. The
most common hybrid approach is to add pseudorandom coding to each hop
3 in a stepped FM signal.

6. Random Signal

Still another method achieving extremely large time-bandwidth product sig-

nals is to transmit bursts of wideband noise. Signal detection is accomplished
by cross-correlating the received signal with samples of the transmitted t,;;-
.nals. Unlike any of the other methods, range sidelobes can be made arbitrzriv.
small, and range and velocity ambiguities can be controlled independently.
I
2.2.2 Impulse Radar
The second approach to achieving a UWB signal is to shorten the pulse duration
because the bandwidth is rouglily proportional to the reciprocal of the pulse dura-

tion. This approach is the basis for impulde radar in which there is no carrier at
all. The antenna is excited by a sequence of very short baseband current pulses. In
order to achieve sufficient pulse energy to accomplish target detection at reasonable
ranges, it is necessary that the peak power be very large.
While the relative bandwidth of spread-spectrum radar systems tends to be
considerably less than unity, the relative bandwidth of impulse-radar systems is
almost always close to unity. The spectrum of such a signal can easily span two or
1 11

I
three decades in frequency. Because of this unique feature of impulse radar, it is
this approach to UWB radar that is discussed in detail in this report.

2.3 Potential of Ultra-Wideband Radar

The answer to the question, "Why UWB radar?", lies in the unique capabilities
that such a radar approach may possess. Some of the potential advantages and
advanced capabilities that have been attributed to UWB radar as compared with
conventional radar are listed on page 3. Some of these capabilities are presented in
more detail below.

1. Target Identification

The extremely good range resolution that can be achieved by UWB radar
make it well suited for advanced capabilities such as target identification. The
typical target consists of a number of scattering regions whose responses are
merged into a single return signal in a conventional radar system. In a UWB
radar these distinct scattering regions can be separated, thus aiding in the
identification of the target. Also, because of the broad spectral density, various
target-response modes may be excited that may be used in identification.

2. Detection of Low-Observable Targets

The use of UWB signals may aid in the detection of LO targets. Such LO
"stealth" targets may be coated with radar-absorbing materials that perform

well over only a relatively narrow band. UWB radar signals may be able to
exploit this limitation. Alternatively, targets may be designed to have a low
RCS by careful avoidance of fiat surfaces, sharp edges, and any resonances
in the usual radar bands. A UWB signal may uncover such resonances an)d
produce a much larger return than would a conventional radar.

12
3. Target Imaging

A related task in which the UWB radar might be useful is that of target
imaging, such as in a scanning mode or in synthetic-aperture radar (SAR). In
general, target imaging requires both good range resolution and good angular
resolution. The excellent range resolution is achieved by the use of UWB
signals. The angular resolution, which depends either on antenna size or
antenna motion, may be achievable directly for close targets or at a distance
by employing a synthetic aperture. Image reconstruction techniques based on
separated transmitters and receivers may allow high-resolution 3-D capability.

4. Penetration Radar

The use of impulse radar for ground penetration to detect tunnels, bunkers,
mines, etc., has been widely proposed and appears to have considerable merit.
A related application is the penetration of foliage to detect manmade objects
that would otherwise be hidden from view. This application is of particular
concern in locating strategically relocatable targets such as missiles, launchers,
tanks, trucks, etc., which can be effectively hidden from conventional radar
and infrared sensors by both foliage and camouflage.

5. Rejection of Background Clutter

The detection of targets in the presence of background clutter is always a

challenging task. Often the good range resolution provided by the UWB
radar can be exploited to separate the target from the clutter. In this regard,
impulse radar is probably better than the spread-spectrum radar, because
clutter can be eliminated by range gating rather than depending on the low
correlation sidelobes of the spread-spectrum signal.

I
| 13

I!
6. Low Probability of Intercept and Exploitation

The LPI capability of any radar is dependent upon the capability that one as-
sumes for the intercept receiver. This is also true for impulse radar. However,
an intercept receiver designed to detect the presence of conventional radar
signals may not be effective in detecting the presence of an impulse-radar
signal. Even if the intercept receiver employs energy detection (as it must
in almost all cases), it will only detect the radar signal if it is observing in
the right direction at the right time. Exploitation of a radar signal implies
that the signal can be used to advantage by the object the signal illuminates.
A common example of this is the use of radar homing by a missile in which
the missile-guidance system homes in on the radar transmissions. Although
there are numerous known countermeasures for this type of exploitation with
conventional radars, most involve some sort of multimode operation. A radar
system that could avoid such exploitation without requiring a change in the
mode of operation would be advantageous. Impulse radar may provide such
a system.

14
I
I
I,I

Chapter 3

* Ultra-Wideband Radar
I
3.1 Theoretical Considerations
In order to make efficient progress in the development of any capability based on a
complex physical process, a fundamental and complete understanding of the overall
problem is necessary. In the case of advanced, remote-sensing capabilities based on
UWB radar, the processes involved are varied and highly complex. Although much
information can be obtained through experimental development, this information
can only allow an empirical, and therefore limited, approach to further development.
In order to gain a fundamental understanding of the complex processes involved so
that a knowledgeable and efficient approach to development can be undertaken, it
is necessary to formulate and solve the complex mathematical problems that define
these processes and to develop a theoretical understanding of the problem. For this
3 reason, the achievement of advanced radar ca.pabi~ities based on UWB approaches
will require a strong theoretical effort.

3.1.1 Time-Domain Electromagnetics

I The solution of fundamental theoretical problems, predominantly through the dis-

cipline of applied mathematics, is of critical importance to the understanding and
development of UWB radar. For tue short-duration EM pulses to be used in impulse

*15
radar, the solution of problems in antenna design, EM propagation, EM scattering,
and the EM response of targets is of critical importance. For the achievement of
advanced radar capabilities such as target identification and target imaging, iuverse
problems for transient EM scattering need to be formulated and solved.
Although highly developed frequency-domain approaches can be applied to these
problems in transient EM theory, it is natural and more efficient to formulate these
problems in the time domain [Bennett 1978, Ross 1986]. For almost all problems
of importance to the development of UWB radar, the electrical parameters of the
materials can be assumed constant with respect to the field amplitudes. Therefore,
the mathematical formulation of the problem is linear and the equivalence, through
the Fourier transform, of the frequency domain and the time domain holds. When
factors such as computational efficiency are considered, however, formulating and
solving these problems in the time-domain is crucial.
For effective formulation and solution, the transient electromagnetic processes
in impulse radar require a time-domain approach rather than the frequency-domain
approach traditionally used in engineering electromagnetics. The formulation and
application of finite-difference, time-domain (FDTD) methods is an emerging nu-
merical approach in computational electromagnetics (CEM). The FDTD solution
of Maxwell's curl equations is analogous to the finite-difference solution of scalar-
wave-propagation and fluid-flow problems in that the numerical model is based
upon the direct solution of the governing partial differential equations. Another
approach to solving Maxwell's equations in the time domain involves a Galerkin
finite-element (GFE) formulation for irregular, nonorthogonal grids. Employing
quadrilateral and/or triangular elements, the GFE approach allows more accurate
modeling of nonrectangular structures. FDTD and GFE are nontraditional ap-
proaches in the field of computational electromagnetics, where frequency-domain
integral equation approaches such as the method of moments have dominated for
25 years.

16
I
I
Because of the complexity of the physical processes involved, time-domain elec-
I tromagnetic theory is of critical importance to and must be an integral part of
the development of UWB radar. For this reason, an overview of time-domain elec-

tromagnetic theory and a quintessential problem in transient EM scattering are

presented in Appendix A.

3.1.2 Emission, Propagation, and Penetration

The emission and propagation of EM waves having temporal variations that are
nonsinusoidal is a subject that is not completely understood. Recently, there has
been a considerable amount of discussion over the controversial ideas of Henning
I Harmuth concerning nonsinusoidal EM waves. See the IEEE Tranaactions on Elec-
tromagnetic Compatibility over the last 5 years. The ideas of Harmuth have some
I implications for impulse radar. To the extent that the authors have had time to
check his work on transient propagation in lossy media, the results appear to be

correct [Appendix C]. We cannot endorse all of his work on antennas, a subject
which requires much further study.
The design of antennas for the emission of EM impulses is important to perfor-
mance capabilities. A number of different approaches to UWB antenna structures
are described in Section 3.2.2. The effective development of high-performance anten-
nas will certainly require a considerable amount of EM modeling. Several computer
codes exist that can be applied to this development. Thin-wire time-domain codes
such as TWTD and its relatives [Landt 1974, Miller 1980] are immediately available
for use in modeling certain antenna structures. Other time-domain codes that have
been developed for particle-accelerator modeling, such as the MAFIA code group
[MAFIA 1988] may be useful for electromagnetic scattering problems.

Recently, propagation modes that have a far-field energy distribution that re-
mains "localized" have become of interest [Brittingham 1983, Wu 1985, Ziolkowski
i 1988]. These interesting propagation modes are referred to by a variety of names

17
such as "focus waves", "EM missiles", "directed-energy pulse trains", etc. The
capability of directed-energy transfer in a radar system is of great potential impor-
tance. It is not yet known whether any of these ideas will find application in radar,
but they certainly deserve further investigation.
The theory of the penetration properties of nonsinusoidal EM waves in lossy ma-
terials has been explored in some depth in the literature [Harmuth 1986]. However,
this exposition is still very abstract. One can draw some conclusions from informa-
tion available concerning the penetration capabilities of conventional radar. It is
well known [Ramo 1965] that the attenuation constant of isotropic lossy materials
usually increases with frequency. Hence, a radar operating at low frequencies will
generally have better penetration properties than one operating at high frequencies.
The normal atmosphere has little effect on the sub-ionospheric propagation of
UWB waves, which contain little or no energy above 10 GHz. Dense clouds of
dust and heavy rain may produce substantial scattering and attenuation (up to
10 dB/km for rain). Snow is much less of a problem. None of these effects are

significantly dispersive for ultra-short pulses.

Propagation of UWB waves through the ionosphere is another matter. Any
components of such waves below 30 MHz may be assumed to be lost in transit
through the ionosphere, which is no problem for UWB waves with spectra extend-
ing to 10 GHz or further. The problem is that ultra-short pulses may be greatly

distorted depending on electron density, which varies with sunspot activity and will
be difficult to take into account. These preliminary considerations suggest that con-
ventional long-pulse narrow-band radar is likely to prove superior to UWB radar
for applications requiring trans-ionospheric propagation. For example, detection of
ICBMs by UWB ridar will probably require locating the radar on a satellite.

18
I
I
3.1.3 Electromagnetic Response of Targets

Following the emission of the EM signal from the transmitting antenna and the
propagation of that signal to the target, the response of the target to the incident
EM signal must be considered. This process, which returns the signal to the radar
receiving system, is the basis for any information to be obtained about the target.
It is, therefore, one of the most crucial aspects of the entire radar process. In
particular, if an advanced capability such as target identification is desired, this
process must be understood in detail so that signals can be designed that result in
sufficient information being returned from the target to allow it to be identified.
The determination of the EM response of targets can be understood through
I the discipline of EM scattering theory. For the case of impulse radar, this theory
must provide a correct description of the EM scattering of "transient" signals from
target structures. It is therefore a problem in time-domain EM theory as described
in Section 3.1.1 and Appendix A.
Early work on the time-domain solution of scattering by perfect conductors was
performed by Bennett [Bennett 1968]. A milestone paper on the representation of
transient scattered fields is that of Marin [Marin 1973]. In this paper, he pointed out
that the EM scattering from a large class of conducting scatterers is a meromorphic
function of frequency. This mathematical result allowed the transient scattering for
this class of problems to be represented as a sum of complex exponentials. This is an
important consideration when target identification is being considered as described
in the following section.

3.1.4 Target Identification

An area of research that is perhaps the most important in terms of achieving ad-
vanced radar capabilities is that of target identification and target imaging. Target
identification and target imaging are very closely coupled with EM theory in the
development of advanced capabilities based on impulse radar. There are a number

* 1
of approaches to target identification that need to be examined as to their applica-
bility to impulse radar. One of the most important approaches, which is based on
parametric system identification, is discussed in this section.
Parametric identification of transient systems is a concept that has only recently
begun to appear in EM and applied-mathematics journals. The initial motivation
for research in this area resulted from interest in transient EM scattering theory
where the excitation is the electromagnetic pulse (EMP) produced by a nuclear
weapon. This research is now driven primarily by the interest in the development
of advanced radar capabilities such as target identification.
Parametric system identification is a subclaus of system identification in which
the information concerning the system is obtained in the form of a parametric
8ct. Such a set, for example, might be the poles and zeros describing the system
in the frequency domain. The theory of parametric system identification has been
formalized by Ljung in a survey report on system identification [Ljung 1981] and in a
joint paper with Glover [Ljung and Glover 1981] on a comparison of parametric with
nonparametric methods. In addition, Ljung has contributed to the understanding
of iUa in the parameter estimates through investigation of stochastic noise models
[Ljung 1978].
Signal representation consisting of fitting data with a complex exponential series
dates back to the work of Prony [Prony 1795] whose method was resurrected by Mc-
Donough [McDonough 1963] and a number of workers in EM in the mid 1970's [Mof-
fatt and Mains 1975, Van Blaricum and Mittra 1975, Miller 1981]. Subsequently,
Dudley [Dudley 1979] showed that Prony's method is a special case of pole-zero
parametric identification. The analysis and modeling of transient EM scattering in
the established framework of system identification, particularly parametric system
identification, utilizes signal-processing and system-modeling techniques based on
advanced methods in applied mathematics. This approach is particularly valid for
the modeling and analysis of UWB radar to determine the fundamental character-

20
istics and advanced capabilities it would have as compared to conventional radar
I techniques.

i 3.1.5 Target Imaging

The use of radar techniques to produce images of targets exhibits a dichotomy

that portrays two vastly different stages of development. On one hand there is the
production of 2-D images of extended areas, as is exemplified by the use of SAR
for mapping purposes. Radars of this type are well developed and can produce
high-resolution pictures by virtue of using signal bandwidth to achieve resolution in
one dimension and using radar motion to achieve resolution in a second dimension.
I As discussed in Section 3.3.11, impulse radar appears to be capable of achieving
this type of imaging as well as the more conventional sinusoidal-signal SAR.
The other aspect of radar imaging is that of producing the equivalent of 3-D
images (that is, producing accurate 2-D projections of 3-D objects) of relatively
small objects such as aircraft or missiles. This aspect of radar imaging is much
fl less well developed and there are no existing operational radars that are capable of
producing such images at long range. The reason for this failure is that excellent
resolution in three dimensions is required, and the only way to achieve the requisite
resolution in the third dimension with a monostatic radar is to use an antenna
system with a very large aperture in that dimension.
Techniques for achieving true 3-D imaging with a multistatic radar have been
proposed. These techniques are roughly the radar equivalent of the tomographic
techniques used in X-ray technology to achieve 3-D images. One such technique
is under investigation by Forrest Anderson. His technique employs the modified
Radon transform to accomplish ellipsoidal backprojection of points in a 3-D space.
I This approach utilizes perhaps as many as ten or twelve antennas arranged in a
circular configuration. The signal is transmitted from a single antenna in the center
I of the configuration and is received at all of the antennas in the circle. Although the

* 21

I
computational requirements are great, the technique appears to provide a feasible
approach to identifying the number and relative locations of aircraft in a formation.
Whether the same approach can also be used to identify the scattering regions on
a single aircraft in three dimensions is still open to question. It is clear, however,
that some form of UWB radar is essential to achieve the necessary range resolution.

Most of the 3-D monostatic radar techniques for imaging that have been demon-
strated have been accomplished at short ranges, on the order of 1 km or less, and
with an extremely narrow-beam antenna, which can only be obtained with practical
antenna sizes by using millimeter waves (MMW). Several examples of such MMW
radar-imaging systems are available in the classified literature. Much better cross-
range resolution can be achieved using an infrared radar system, but neither this
type of system nor the MMW system provides an all-weather capability.

3.2 Ultra-Wideband Technology

3.2.1 Impulse-Radar Transmitters

Impulse radars for short-range applications have been commercially available for
several years. Ground-penetration radars are used by civil engineers and geologists
for a variety of applications: locating pipes, reinforcement bars, voids, geological
interfaces, hazardous wastes, lake and glacial ice profiles, etc. [Ulriksen 19821.
An impulse rsdar for the detection of intruders has been developed by ANRO En-

gineering Consultants of Lexington, Massachusetts. Pulse widths of a few nanosec-

onds are used to achieve detection ranges of 50-80 m. A detailed description of this
system is available [Ross et al 1988].
Experimentation on impulse-radar transmitters using larger peak powers for
longer-range applications has been conducted for many years by Paul Van Etten
and his colleagues at Rome Air Development Center [Van Etten 1979]. Much of the
RADC work has used spark gaps and transmission lines charged to high voltages

22
I
to generate nanosecond and subnanosecond pulses. Sequences of short pulses were
generated in the RADC experiments by "frozen-wave" generators consisting of mul-
tiple spark gaps interconnected by transmission-line sections. Maximum ranges of
a few kilometers were obtained. Further progress has waited on a practical means
of obtaining pulses of higher peak power. However, the utility of impulse radar in
5 target identification has been amply demonstrated by the RADC experiments.
Current emphasis of R&D in impulse radar is stimulated by the requirements
for ranges of tens of kilometers in such applications as target identification. To
produce high values of electric field intensity on targets at long range, extremely
high rates of change of antenna current are required.
Solid-state devices capable of producing pulses with extremely short rise times
and high peak power were developed at the Los Alamos National Laboratory in
the early 1980. [Nunnally and Hammond 1983, Lee 1984]. Peak powers of many
megawatts with subnanosecond rise times can be realized with these PCPS devices.
The development of theoretical models of PCPS devices has also been completed
[Iverson 1988]. Early consideration was given to the possibility of using these PCPS
devices in radar applications to achieve improved range resolution [Cooper 1986].
The fall time of current pulses produced using PCPS devices is substantially
longer than the rise time. The interval between pulses must be still longer to allow
time for heat dissipation. The initial transient behavior of the current pulse is
quadratic in time, so a reasonable model of the current pulse is t 2 e-t/T. The rise
time of the time derivative of such waveforms is less than half that of the waveform
itself. If a 100-ps rise time in electric field intensity is required on the target, a
current rise time of about 200 ps will suffice. Such short rise times can be realized
with PCPS devices at high levels of peak power. Peak power levels of 100 MW
delivered to a 25-0" load were realized in PCPS experiments at Los Alamos for
relat'vely long (200 ns) pulses. Exactly what levels of peak power can be achieved
3 with rise times of 200 ps requires further experimentation.

* 23
I
.A typical PCPS device is a block of semiconductor not more than a few centime-
ters in length. The cros-section is a rectangle, a fraction of a square centimeter in
area. The experimental arrangement for some of the PCPS wts• at Los Alamos is
shown in Figure 3.1. The axis of the laser is in the direction of the thin dimension
of the PCPS device. The circuit consists of a high-voltage pulse source driving the
PCPS device and load in series. The resistance of the PCPS device is quite high
until it is illuminated by the laser, when it drops suddenly to a small fraction of
the load resistance. The timing of each pulse of load current is determined by the
laser trigger, which can be accurately controlled co that pulse sequences relatively
free of jitter can be produced conveniently.

••.,••VOLTAGE
TWO 18-rm RG-19
F PROBE
7if
_(
___._ING
CURRENT-VIEWING
RESISTOR
m

MARX , LENS SYSTEM

CIRCUIT
100-300 kV PCS----

"4nF LOAD ELECTRODES

_L RESISTOR

0-SWITCHED IJ
Nd: GLASS LASER

Figure 3.1: Experimental arrangement for PCPS tests

PCPS devices are efficient and car be scaled over wide ranges to meet require-
ments of a variety of impulse-radar applications. Thermal recovery occurs within
tens of ls for typical PCPS designs, so that pulse repetition frequencies (PRF) of

24
I
I
tens of kHz can be sustained. Rates of change of antenna current of 1013 A/s are
I readily achievable. Scientists and engineers with extensive experience in impulse-
radar R&D have indicated great interest in PCPS devices. Work is now under way
to develop production PCPS devices by at least one commercial firm.

1 3.2.2 Antennas

The transmitting antenna is the "transducer" that couples the output from the
high-power electrical-signal generator to the external environz-.ent of the target.
3 The characteristics of the emitted EM signal resulting from this complex coupling
process are critical to the capabilities and limitations of the entire remote-sensing
radar process itself. Because of the UWB nature of the signal involved, the antenna
problem is considerably different than for the usual narrow-band signals.
5 Conventional narrow-band antennas generally rely on sinusoidal resonance phe-
nomena for their operation. For UWB signals, this resonance approach is not ap-
I propriate. A number of different approaches to coupling UWB signals from the
electrical generator to the free-space environment are available, however.
One of these approaches is based on the idea of providing a broadband impedance
transformation from the impedance of the electrical generator (often 50 fi) to the
impedance of free space (377 0"). This is accomplished through a smoothly varying,
3 tapered-transmission-line structure that flares out and "launches" the EM wave
into free space. These strictures often take the form of an open-sided horn antenna
which is basically a tapered parallel-plate transmission line. The parallel-plate
structure is tapered in both width and plate spacing in a number of possible ways
such as linear or exponential tapers.
Another approach is to build an antenna structure that is comprised of a number
j of narrowband elements that completely cover the entire bandwidth of the signal.
These structures are often based on the log-periodic Yagi idea. A third approach
5 is based on the idea of an antenna structure that emits an EM signal produced by

*25
an excitation resembling a time-varying current sheet.
Time-domain electromagnetics is key to impulse-radar R&D as it includes the
analysis and measurement of the emission, propagation, reflection and reception of
transient electromagnetic waves. The study of transmitting and receiving antennas
for impulse radar has to start from basic principles, because the standard antenna
theory is limited to the special case of sinusoidal excitation. The engineer with
extensive experience in classical narrow-band radar systems, who is starting to
work on impulse radar, is advised to heed Mr. Dooley's aphorism.

It wras not the things we didn't know that hurt us.

It was all the things we knew for sure that turned out to be wrong.

The differences between the performance of an antenna or an antenna array

excited by extremely short transients and that produced by steady-state sinusoidal
excitation are many and large. Antenna patterns for transient excitations do not
have the deep nulls and sidelobes characteristic of sinusoidal excitations. A similar
statement can be made for the variation of target RCS with aspect angle. As a
signal-processing operator, the transmitting antenna may be regarded as a differen-
tiator and the receiving antenna as a scalar multiplier. The opposite interpretation
is also possible, because t-differentiation and scalar multiplication commute.

Several examples of the significant differences between transient and steady-

state excitation of antennas are shown in Appendix B, where an array antenna in

an impulse-radar application is considered.

3.2.3 Receivers

This section outlines several important aspects of the hardware design of receivers
for impulse radar. Section 3.2.4 discusses the actual processing requirements in
the receiver in more detail. The particular aspects considered are the isolation of
the receiver from the transmitter, the receiver input circuit, the signal detection

processor, and post-detection processing.

26
I A critical problem in the design of the receiver is isolating the receiver input from
the high-power pulse produced by the transmitter. Because of the large bandwidth
of the transmitted signal and its extremely large peak power, it is unlikely that the
isolators and circulators used in conventional radar for this purpose will perform
satisfactorily in impulse radar. This suggests that it may be necessary to use sepa-
rate transmitting and receiving antennas. However, even with separate antennas it
3 is necessary to provide additional protection for the receiver because the distance
between them is likely to be small and shielding has only a limited effectiveness.
5 Two isolation techniques appear possible. One is the use of signal cancellation at
the receiver input by providing a reversed signal to cancel the leakage signal. For a
i fixed radar installation this may be possible, although it might be very difficult with
a scanning antenna. A more direct approach is to provide fast-acting switches that
I short-circuit the receiver input when the transmitter pulse is on, using low-power
devices operated under the control of the transmitter. Two possibilities are PIN
diodes or PCPS devices, the latter being more convenient if the transmitter uses
PCPS devices to generate the antenna-current pulses.
When signal detectability is a primary concern, the major problem with the
3
ireceiver input circuit is that of obtaining adequate bandwidth without introducing
an excessive amount of noise. This implies that careful attention must be given to
the design of suitable low-noise amplifiers covering a frequency range of perhaps
10 MHz to 10 GHz. It is possible that cooled input circuits may be necessary
in some cases. When there is ample signal power, as would be the case in many
target-identification applications, the effect of receiver noise will be less severe.
3 Two other complications arise in connection with the receiver input circuits.
One of these pertains to the antenna noise that arises in addition to the noise in the
I input circuit itself. Antenna noise results, in part, from cosmic radiation and can be
quite large at low frequencies. For example, an antenna pointing toward the sky will
I have a noise temperature in excess of 3000 K at frequencies below 100 MHz. This

*27
large noise at low frequencies may establish a limit on the lowest frequency that
can be employed usefully by impulse radar. The second complication in the input
circuit arises if notch filters are necessary to reduce interference from narrow-band
signal sources. Not only is the design of the input circuit made more difficult, but
the additional losses decrease the signal energy and increase the noise.
The form of signal-detection processor depends strongly upon the application
for which impulse radar is intended. Various possibilities for detection processors
and the circumstances in which they might be used are discussed in more detail in
Section 3.2.4. The three basic types of signal-detection processors mentioned are the
energy detector, the matched filter, and the cross-correlator. The implementation
of such processors is discussed brietly here.

Figure 3.2: An energy detector for radar

A block diagram of the typical energy detector is shown in Figure 3.2. The
square-law device can be almost any nonlinear device if it is operated at a low
signal level. The integrator may be replaced by a low-pass filter with a bandwidth
equal to the reciprocal of the desired integration time. As indicated elsewhere, the
integration time should be at least as large as the round-trip propagation time over
the extent of the target. This makýes it possible to include all of the sigiaal energy
from the target in each detection operation, but it does destroy the range resolution
that would otherwise be available from the large signal bandwidth. This type of

28
detector is useful when target detection is the primary application of the radar.
3
aThe second form of signal-detection processor is the matched filter. The matched
filter is one whose impulse response is the time-reversed replica of the signal returned
SI from a single reflecting point. The exact implementation of such a filter is probably
not possible, but approximations can be constructed. One possibility for such an
approximation is the use of a channelized input. This type of receiver input circuit
utilizes parallel filters, each having a bandwidth that is only a fraction of the total
bandwidth, but centered at different frequencies so that the total parallel combina-
tion of filters spans the entire desired bandwidth. The gain of each parallel filter
can be adjusted to be proportional to the magnitude of the transfer function of the
I true matched filter in that portion of the spectrum. Not only does this approach
approximate the transfer function of the matched filter, but it also provides a use-
I ful means for excising portions of the spectrum to exclude narrow-band interfering
signals. A careful analysis of the requirements on such parallel filters has not been
I done, but it is likely that vcry careful control of the phase characteristics of the
filters is necessary to achieve an adequate approximation to the desired matched
i filter.
Although the matched filter provides the maximum output SNR, this maximum
3 pertains to each scattering region on the target and not to the total signal returned
from the target. This is the appropriate response for target identification, but does
I not utilize all of the returned signal energy for detecting the presence of the target.
Thus, this type of detector is most useful after the target has been detected and is
close enough for a more detailed examination.
The third type of signal detection processor is the cross-correlator in which a
replica of the returned signal from a single scattering region is generated in the
3 receiver, with a delay corresponding to the round-trip propagation time to the
target. The replica multiplies the received signal and the product is then filtered in
an appropriate low-pass filter. A block diagram of this type of detector is shown in

*29
Figure 3.3. The maximum output SNR of the cross-correlator is identical to that
of the ideal matched filter. Because the form of the received signal from a single
scattering region is known, the reference signal can be generated in the receiver
with good accuracy. From this standpoint, the cross-correlator is probably a better
implementation of the signal-detection processor than the matched filter, which can
be implemented only approximately. However, in order to observe returns from a
number of ranges at the same time, a large number of cross-correlators (possibly
several thousand) need to be implemented and this poses a substantial hardware-
implementation problem. For example, to observe all poosible scattering points
on a target having a range extent of 20 m with an impulse radar having a range
resolution of 2 cm requires 1000 cross-correlators. In addition, bulk delay to place
all of the cross-correlators on the target is also necessary.

INPUT LOW-PASS OUTPUT

SIGNAL FIvLTER 00

SDELAY'/

Reference
Signal

Figure 3.3: The cross-correlator detector

It may turn out that a hybrid signal-detection processor will provide the best
solution to the target-detection problem. One possibility for this is to use an energy

30
I
detector, with the appropriate integration time, at the output of the matched filter,
I or to sum the outputs of the cross-correlators and put this sum into an energy
detector. Whether this combined operation will provide a better probability of
detection than the energy detector alone is a matter for further study. In any
event, this apy-roach has the advantage of retaining the excellent range resolution
of the matcheoi filter or cross-correlator while utilizing all of the energy from the
3 target for detection purposes.
The form of post-detection processing required is also critically dependent upon
5 the application for which impulse radar is intended. Post-detection processing will
almost certainly have to be digital when target identification is the major concern.
3 This requires that the peak outputs from the matched filter or the cross-correlators,
and the times at which these peaks occur, must be digitized and stored so that
5 identification algorithms can be implemented. Although complete sampling of the
uutputs of the matched filter or the cross-correlators appears to be difficult to
3 achieve, there seems to be no practical difficulty in accomplishing this digitization
for the peak outputs only. When the objective is target detection only, and range
I resolution is not an issue, the appropriate output must be compared with a threshold
and quantized whenever the threshold is exceeded. Because of the relatively poor
range resolution that occurs in the case of target detection only, this operation
appears to be easy to accomplish.

3.2.4 Signal Processing

There are two major aspects of signal analysis that need to be considered in con-
Snection with the development of a viable impulse radar. The first of these is the
design of optimum antenna-current pulse shapes and optimum groups of pulses.
This aspect is discussed in more detail in Section 3.4.1. The second aspect is to
determine the modifications in signal shape that take place as a function of the
antenna transient respcnse, target transient response, and the receiver input circuit

31

I
characteristics. The target transient response is addressed in more detail in Section
4.4. Very little has be done so far in determining the antenna transient response or
in examining the transient performance of the receiver input circuit. An analytical
program must be initiated to explore these aspects of system performance in detail
before embarking on any effort to design an impulse radar. In addition, the anal-
yses aimed at determining the optimum pulse shape and the transient response of
typical targets must be carried further.
The ultimate purpose of the radar receiver is to process whatever signal is re-
turned from the target to determine as much information about the target as possi--
ble. This information may range from a simple decision concerning target presence
to a detailed picture of the various scattering regions on the target. Hence, many
different signal-processing methods must be analyzed in order to determine those
that are best for each application. The possible signal-processing methods include
the energy detector for target detection, the matched filter or the cross-correlator
for identifying individual scattering regions on the target, the use of parallel in-
put channels to approximate the matched filter, and the use of composite matched
filters to identify known clusters of scattering points to enhance detectability.
It may turn out that it is desirable to use a combination of processing methods in
the typical impulse-radar receiver in order to achieve the benefits of the extreanely
good range resolution without sacrificing detectability. One possibility is to use
an energy detector with sufficient integration time to include all returns from the
target and also use a matched filter to achieve the resolution necessary to separate
the individual scattering regions. A more effective, but more complex, solution is
to use a large number of cross-correlators, adjusted to respond to potential returns
in adjacent range cells, and put the sum of their outputs into an energy detector.
A third possibility is to use an energy detector with sufficient integration time at
the output of a single matched filter. All of these concepts should be analyzed
in sufficient detail to establish their relative merits for accomplishing both signal

32
I
detection and good range resolution.
Another important aspect of the processing requirements in the receiver de-
pends upon whether the application intended for impulse radar involves the nearly
simultaneous processing of returns from multiple targets. This situation arises, for
I example, in a track-while-scan application or in attempting to discriminate between
decoys and warheads during a missile attack. Although the processing requirements
I for each target are essentially the same as outlined above, the multiplicity of targets
greatly complicates the situation. This type of application undoubtedly calls for at
5 least a portion of the processing to be done digitally.
Although the initial desigXA of signal-processing methods will probably involve
Sanalog processing, as noted abovy, it is desirable to consider the feasibility of using
digital techniques in future processors. Because of the extremely short duration
I. of the signal pulses, this is a challenging problem. It is likely, however, that the
methods employed in wideband oscilloscopes can be adapted to accomplish the
U desired processing tasks.
It has often been stated that the impulse-radar signal contains very little Doppler
information, and this statement is frequently cited as one of the disadvantages of
impulse radar. This conclusion is undoubtedly true so far as extraction of informa-
tion from a single received pulse is concerned. However, in most applications, and
certainly in the case of target identification, more than one pulse is received. In
such cases the time interval over which the received signal is observed is such that it
3 is theoretically possible to extract the Doppler information that is actually present
in the received signal. This possibility is addressed in the following paragraphs.
I The usual concepts of extracting target range-rate information from the Doppler
shift of the received signal frequency are based on the assumption that a sinusoidal
I waveform is employed. In the case of impulse radar there is no such sinusoidal wave
so that the usual concept of a frequency shift is not applicable. More fundamentally,
I the effect of target motion is not that of a frequency shift, but the time contraction

33
contraction or expansion of the received signal. Interpreting such a change in the
time scale as a frequency shift is an approximation that is valid only for relatively
narrowband signals.
A preliminary time-domain analysis of the received impulse-radar signal from
a moving target reveals that the information regarding target range-rate is indeed
contained in the sequence of returned pulses. The present analysis has assumed
the use of cross-correlator detection, but it is likely that similar results can be
obtained from other detectors. The analysis has not yet been carried far enough to
indicate the actual implementation procedures that should be used to extract the
desired range-rate information, but it seems clear that the necessary information is
present in the received signal. As in conventional radar, better estimates of target
range-rate are achieved by processing the received signal over a longer time interval.

3.3 Potential Capabilities and Applications

A vital concern with the development of impulse radar is whether there are applica-
tions that cannot be achieved with conventional pulse radar, or cannot be achieved
as easily. Hence, the purpose of this section is to propose certain applications for
which impulse radar appears potentially superior. One cannot be certain about the
viability of any the following potential applications without a great deal of addi-
tional analysis and experimentation. All that can be done here is to present some
arguments that suggest that these applications have potential.

3.3.1 Target Identification

Target identification is perhaps the single most important advanced capability of
any radar system, and is of great interest to a number of organizations. Target
identification provides the most complete information regarding a target zhort of
producing an actual image of the target. Some of the theoretical conside.rations

34
I
I
concerning target identification and some proposed techniques are addressed in
Section 3.1.4, The purpose of this section is to indicate the extent to which impulse
radar may be able to fulfill these theoretical requirements.
Perhaps the most important requirement for any method of target identifica-
3 tion is the use of a radar that has adequate range resolution to be able to resolve
individual scattering regions on the target. Impulse radar meets this requirement
3 extremely well. Not only does this make possible the determination of target extent
in range, but if the target is rotating, the size in other dimensions can 'e deter-
5 mined as well. If a nonrotating target is observed over a long enough interval that
its aspect with respect to the radar changes appreciably, the concepts of differential
geometry can be used to determine something about the curvature of the target
surface. Range resolution of 1 cm or so, which is readily achievable with impulse
3 radar, makes this technique a sensitive indicator of certain aspects of target shape.
Another approach to target identification is that of representing targets as a sum
I of complex exponentials. If a catalog of such complex exponentials at various aspect
angles were available for each target of interest, then the use of Prony's method on
U the received signal would provide a systematic approach to target identification.
Two other aspects of target identification involve the possibility of detecting
I "resonancesin the target-return signal and the detection of return-signal modulation
3 by the target such as might be produced by jet-engine rotors. It appears quite likely
that impulse radar, with its extremely large bandwidth, can do a better job of
5 detecting target resonances than conventional radar. This statement is particularly
true if the resonances occur at frequencies below the usual radar bands. The set
3 of frequencies at which such resonances occur is probably unique to each type of
target. No analysis has yet been performed to indicate how such resonances show
Sup in the received signal from impulse radar, but experience suggests that a spectral
analysis of a sequence of returned pulses should reveal this type of information.
5 Jet-engine modulation is conventionally observed by Doppler techniques. It is

1 35
I
not yet clear that such modulation can be observed over a sequence of pulses from
impulse radar, but as noted in Section 3.2.4, a preliminary analysis suggests that
this may be possible.
Because of the importance of target Identification to the Department of Defense
(DoD), and because impulse radar appears to have some unique capability in this
respect, a more complete analysis of this potential capability is urger tIl needed.

3.3.2 Low-Observable Target Detection

Detection of Stealth Targets

One of the suggested applications of impulse radar is the possibility of more easily
detecting LO targets having a small RCS by virtue of the use of a radar-absorbing

material such as ferrites. These materials are alleged to function successfully over
only a relatively narrow frequency band [Harmuth 1983a, Harmuth 1985]. In fact,
Harmuth suggests that ferrite materials can reduce radar signal reflections by 20
dB over a relative bandwidth of only about 0.125. It appears, therefore, that an

impulse-radar signal having a relative bandwidth on the order of unity would have
a significant portion of its energy reflected by the ferrite material and its metallic
backing, and thus make this type of LO target more readily detectable.

On the other hand, it has also been suggested [Thornton 1986] that this problem
can be ameliorated by using ferrites in conjunction with other unspecified materials
and with an appropriate spatial distribution. There does not appear to be any
data in the open literature that supports this claim. In fact, the book by Knott
et al [Knott 1985] has an entire chapter devoted to radar-absorbing materials and
suggests that even with four layers of hexagonal ferrite, the RCS reduction can be
as great as 20 dB over a relative bandwidth of only about 0.17. One can conclude,
therefore, that a thorough investigation of the classified literature is necessary before
one can assert that the use of UWB signals does or does not make this type of LO
target more detectable.

36
Detection of Shape-Controlled Low-Observable Targets

Another approach to designing LO targets is to control the shape of the target by

eliminating flat surfaces, sharp edges, and external features that might be resonant
in any of the commonly used radar bands. However, if the radar signal covers a very
wide range of frequencies, as does the impulse-radar signal, this approach becomes
much more difficult to implement.
Consider, for example, an aircraft that must carry communications antennas.
Even if the antennas are conformal to the surface, they must be nearly resonant at
their operating frequencies to do an effective job of both transmitting and receiving.
Many communication frequencies are on the order of a few hundred MHz, well below
the usual radar bands. However, the spectral density of impulse radar is almost
directly proportional to frequency from zero frequency up to near the maximum
spectral density (see Appendix B for example). Thus, there is a significant amount
of signal energy throughout the communications bands, and these antennas could
easily reflect more energy than the rest of the aircraft. It seems likely that any type
of vehicle will have some features that are resonant at some frequency within the
bandwidth of t~he impulse-radar signal, even if they have been carefully designed to
avoid such resonances in the conventional radar bands.
Although the above conclusions appear plausible, much work is required to de-
termine the time-domain response of such objects rather than reliance on frequency-
domain concepts. Even so, on the basis of what is currently known about impulse
radar, the conclusion that LO targets are more easily detected by impulse radar is
eminently reasonable.

3.3.3 Target Imaging

The term "radar-target imaging" suggests that the radar is capable of producing
a 2-D outline or even a 3-D picture of the target. An outstanding example of this
type of image is the high-quality ground map that can be produced by SAR. Such

37
maps are possible because range resolution can be used to provide the resolution in
one dimension, while the synthetic.aperture processing provides the resolution in
the other dimension.
A long-time dream of radar practitioners is that of obtaining similar high-quality
images of small 3-D objects such as aircraft, missiles, tanks, etc. That such a dream
has yet to be fulfilled in any real sense results from the limitation to resolution in
two dimensions only, unless an extremely large antenna structure is used. In its I
normal monostatic mode of operation, impulse radar does not provide any solution
to this problem. However, as noted in Section 3.1.5, multistatic operation may I
provide a solution, and research related to this possibility is already under way.
It is reasonable to question whether impulse radar provides any advantage over
conventional radar in a multistatic mode of operation. Preliminary consideration
of this problem suggests at least two possible advantages that can be obtained from
the use of impulse radar. It is assumed here that the transmitted signal comes from
a single antenna and that the signal returned from the target is received at a number
of receiving sites. Since the range extent of each pulse is essentially the same as the
range resolution, a group of pulses separated in time only by an amount somewhat
greater than that equivalent to the target size can be transmitted. Furthermore,
successive pulses in the group can be modified to place the maximum energy at
different places in the frequency spectrum so that the possibility of exciting unique
target resonances is enhanced. This does not appear to be possible with conventional
spread-spectrum radar because each pulse will itself have a duration longer than I
the equivalent target extent, and will not have as large a bandwidth as the impulse-
radar signal. I
The second possible advantage that impulse radar may offer in the multistatic
mode of target imaging is more precise timing of the received pulses. This is im- I
portant in multistatic operation in order to correctly assign each return pulse in
each receiver to the proper scattering source. The more precise timing is a con-

38
sequence of the greater bandwidth of impulse radar as compared to conventional
spread-spectrum radar.

3.3.4 Penetration Radar

3 The problem with conventional radar operating at low frequencies is that of obtain-
ing sufficient bandwidth in the signal to achieve the desired range resolution. Im-
t pulse radar does not suffer from this problem, because the radiated signal can have
a relative bandwidth approaching unity. Indeed, a number of ground-penetration
radars operating essentially on impulse-radar principles are known to exist [Har-
muth 1983b].
The use of impulse radar for ground penetration to detect tunnels, bunkers,
mines, etc., has been widely proposed and appears to have considerable merit. A
-related application is the penetration of foliage to detect manmade objects that
3 I would otherwise be hidden from view. This application is of particular concern
in locating missiles, launchers, tanks, trucks, etc., which can be effectively hid-
3 den from conventional radar and infrared sensors by both foliage and camouflage.
Recent work by John McCorlle at Harry Diamond Laboratory on SAR in foliage
5 penetration indicates some advantages in using impulse-radar techniques.
The possibility of using radar to penetrate buildings to identify the nature of
3 contents may be valuable in intelligence applications. The optimum frequency for
accomplishing such building penetration is strongly dependent on the type of con-
5 struction used in the building. If the walls and roofs are pre-stressed concrete or
wood, then low frequencies are probably best. If the walls and ceiling are steel re-
I inforced concrete, somewhat higher frequencies are likely to be better. If the walls
are metal, then penetration is possible only through openings such as windows and
doors. Whether impulse radar is any better than conventional radar for building
penetration is a matter of conjecture at this point. However, it seems likely that
the very wide frequency range of impulse radar may be useful in providing initial

*39
information as to the optimum frequencies that a narrow-band radar should use for
this application.
Another potential application pertains to the ability of impulne radar to pene-
trate plasma regions caused by nuclear explosions, rocket exhaust, re-entry wakes,
etc. Such plasma regions typically become more transparent as the incident fre-
quency increases. Because the impulse-radar signal contains such a large range of
frequencies, the higher end of this spectrum may be able to provide more detailed
information about targetm engulfed in such plasma regions than would a conven-
tional radar. Furthermore, the fact that lower frequencies are also present in the
impulse-radar signal provides a greater ability to penetrate the atmosphere and
provide better detection of targets that are not in plasmas. Thus, the same impulse
radar may be able to perform a combination of tasks that would otherwise require
two conventional radars.

3.3.5 Short-Range Radars

A problem with conventional radar is that of observing targets at very short ranges.
Even if adequate power control is available, the signals returning from short ranges
often arrive back at the receiver while the transmitted signal pulse is still on. This
problem is particularly severe if spread-spectrum techniques are used to improve
the range resolution, and an excellent range resolution is often a requirement of
short-range radars. Impulse radar, on the other hand, uses a very short pulse and
has a minimum range (assuming adequate power control) that is esmentiall:, the
same as the range resolution.
To illustrate this problem, suppose that a spread-spectrum radar uses a pseudo-
random sequence that is 511 chips long with a chip duration of 1 ns. This provides
a range resolution of about 15 cm with range sidelobea that are down by about 13.5
dB. The pulse duration is 511 us, which provides a minimum range of about 76 m,
hardly adequate for many of the applications mentioned below. In contrast to this,

40
an impulse radar with a pulse duration of 200 ps has both a minimum range and a
range resolution of about 3 cm and no range sidelobes. This is indeed a dramatic
3 •improvement in performance that can be achieved by impulse radar.
Applications that require the use of a short-range radar combined with excel-
5i lent range resolution include spacecraft docking, missile fuzing, intrusion detection,
vehicle-collision avoidance, and station keeping in aircraft formations. In all of these
3I applications, impulse radar is a viable candidate.

5 3.3.6 Target/Decoy Discrimination

An important radar application in connection with a missile attack is that of dis-

I criminating between actual warheads and decoys. In general, decoys are designed
to have a ballistic coefficient and a conventional radar signature that closely ap-
proximates that of a warhead although the decoys are physically much smaller.
Conventional long-range radars used to track such objects usually do not have suf-
ficient range resolution to discriminate well between the decoys and warheads.
I •Impulse radar does have adequate range resolution to discriminate physical size
when the decoys are dispersed separately. It is not so clear that iuch discrimination
5 Iis equally possible when decoys are tethered together to create an object more
nearly the same physical size as the warhead. It seems likely that, even in this case,
significant differences in the returns will lead to possible discrimination.
A possible problem in this application is that of obtaining sufficient impulse
5 Ipower to achieve this objective at the ranges of interest. As noted elsewhere in
this report, the detectability of any target depends primarily on the energy of the
returned signal and not the peak power. When the pulse duration is short, as
it must be in impulse radar to obtain the desired range resolution, the necessary
I peak powers may be in excess of limitations imposed by dielectric breakdown at
the transmitter. On the other hand, much larger potential differences are required
I to damage insulation as pulse durations are reduced to the ultra-short intervals

Il 41

U2
proposed for impulse radars.

3.3.7 Intrusion Detection

The application considered here is that of being able to detect the intrusion of per-
sonnel into a secure area such as that surrounding a temporary military installation.
Although infrared and ultrasonic devices are commonly used for this application,
radar may have an advantage in all-weather capability. Radars used for this pur-
pose may be low-power short-range devices with a range resolution on the order of a
few feet and good Doppler resolution. An important consideration in the design of
such intrusion-detection systems is that of minimizing false alarms caused by dogs,
squirrels, or other small animals.
Although impulse radar is not yet considered to have a Doppler capability, it
may still be useful in short-range applications by indicating motion as a consequerne
of range changes from observation to observation. The excellent range resolution

should make this approach feasible, as well as provide a better opportunity to

distinguish between humans and animals.
Another mode of operation that might be feasible for this application is to use
two or more impulse radars in widely separated positions, each with a very broad
beamwidth and no angular scan. Target positions could be accurately determined
by triangulating the range-only measurements from each radar. This approach has
the advantage of providing constant surveillance of an entire area rather than the
periodic surveillance associated with a scanning radar. The disadvantage is the
possibility of "ghost" targets that result from incorrect association of returns.

3.3.8 Detection of Sea-Skimming Missiles

The sea-skimming missile tends to be relatively small and to operate only a few feet
above the surface of the ocean. As a result of their small RCS and the existence
of a large amount of sea clutter, such missiles are very hard to detect. There is

42
I
3
a possibility, not yet evaluated in any qaantitative sense, that the excellent range

resolution of impulse radar, combined with the absence of any range sidelobes such
as are typical of spread-spectrum radars, could provide better detectability for this
type of target. Because of the short reaction time necessary for successful detection
of sea skimmers, the fact that no Doppler processing is necessary may also be an

advantage.

3.3.9 Detection of Cruise Missiles

Il Another important defense problem is that of detecting cruise missiles. Some of

these weapons fly at low altitudes to be below the radar horizon for surface-based
radars and to be obscured by ground (or sea) clutter from most airborne radars.

Furthermore, even without the use of radar-absorbing materials, cruise missiles

have a considerably smaller RCS than a typical aircraft, and if it also has a stealth
* surface, it becomes extremely difficult to detect.

As noted in Section 3.3.2, impulse radar may have a significant advantage in

detecting LO ta ;ets because of its laige bandwidth. This advantage, if real, would
be equally useful in aiding the detection of cruise missiles. Furthermore, the absence
5 of range sidelobes and angle sidelobes (Appendix B) may provide an improved
signal-to-clutter rastio for an airborne platform. Consider an impulse radar having
3 a range resolution of 0.1 m and employing an antenna with a beamwidth of 1 degree.
If the RCS of the surface per unit area is .-25 dB -.t a range of 100 kni, the equivalent

Sclutter cross-section is about 0.55 m in each range cell for impulse radar. Assuming
that each scattering region on the cruise missile has an RCS on the order of 0.1 mi2 ,
3 the signal-to-clutter ratio in each range cell is about -7.4 dB on a single-pulse basis.
Approximately 100 pulses would have to be integrated to cbtain a probability of
* detection on the order of 0.9.

Consider next a corresponding conven'tional spread-spectrum radar with a range

resolution of 6 m (just enough to encompass the cruise missile), and with range

*I43

Il
sidelobes down by 30 dB (an optimistic value). In this situation the radar would
produce an equivalent clutter cross-section of about 100 m2 . If the conventional
RCS of the cruise missile is 1 m 2 , and the clutter RCS is 1 m2 , the resulting signal-
to-clutter ratio in each range cell is about -20 dB. In this case, about 2000 pulses
would have to be integrated to achieve the same probability of detection.

3.3.10 Radar-Cross-Section Diagnostics

The above discussion of the ability of impulse radar to detect LO targets suggests
another application of impulse radar in the development of such LO vehicles. The
extremely good range resolution of such a radar should make it possible to identify
the location of flare spots and other major scattering surfaces by observing accurate
scale models of the devices. An impulse radar with a bandwidth of 10 -Hz has a
range resolution of about 1.5 cm. Thus, even on the one-fifth scale models used
in such fuze-testing facilities as the Encounter-Simulation Laboratory of the Naval
Weapons Center, the individual scattering regions should be readily identified.

3.3.11 Synthetic-Aperture Radar

SAR relies on motion of either the radar or the target to create a signal that
appears to come from an antenna aperture much larger than the actual physical
aperture. With appropriate signal processing, this results in achieving a cross-
range resolution (i.e., parallel to the direction of motion) that can be one-half the
physica* aperture dimension. The conventional SAR employs a spread-spectrum
signal having a bandwidth on the order of tens of Mlz to achieve a down-range
resolution (i.e., perpendicular to the direction of motion) that is the same order of
magnitude as the cross-range resolution. The result is a radar that is capable of
providing maps of the ground or images of a target.
The first operation that must be performed by the typical SAR receiver is that
of despreading the spread-spectrum signal so that the Doppler information can

44
be extracted. For impulse radar this operation is unnecessary because the signal
spectrum has not been spread. On the other hand, each pulse of the impulse
radar carries essentially no Doppler information because of its short duration. To
compensate for this short duration is the fact that returns from different angles
arrive at the receiver at different times and may be extracted on the basis of time
rather than Doppler frequency. The exact nature of the required processing has
not been studied in any depth, but it appears likely that an appropriate method of
combining data from the multitude of range cells would again lead to a SAR-type
image. It also appears likely that the required processing may be less difficult than
that required for the conventional SAR, although this is yet to be established.
Another potential advantage of using impulse radar for SAR operation is that
the bandwidth necessary for a desired range resolution can be achieved at lower
frequencies. As noted above, this enhances foliage penetration and increases the
likelihood that the SAR can detect and identify targets that are normally hidden
3 by foliage.
Offsetting the possible advantages of impulse radar for SAR applications is the
need for extremely good clock stability. A rough calculation for a SAR system
having a range resolution of 0.3 m at a range of 100 km suggests that a short-term
I clock stability on the order of 4 x 10-11 may be necessary to achieve the desired
performance.

3.3.12 Low-Angle Radar

Another application suggested for a land-based impulse radar is that of detecting

3 and identifying targets that are only a few degrees above the horizon. Most con-
ventional radaxs have great difficulty with low elevation angles because of the large
amount of ground clutter surrounding the range to the target. The effects of such
clutter can be reduced by improving the range resolution of the radar, because the
I amount of clutter power that is in the range cell containing the target is reduced.

45

"-I • - • • • • ' N• •• r• ' • • . . ""l H" " - 1 a •r '"• • "+ •1 •"l • ' • •• :' a, .
The even better resolution of impulse radar should improve this situation further.
However, a more detailed consideration of this problem suggests that the law of
diminishing returns operates here. As the range resolution is made smaller than
the target, the target return gets smaller at about the same rate as the clutter
power. A much more detailed analysis is necessary before it can be concluded that
the impulse radar has a viable role in low-angle applications.

3.3.13 Low Probability of Intercept

The LPI capability of any radar is dependent upon the capability that one assumes
for the intercept receiver. This is also true for impulse radar. First, we must assume
that at reasonable ranges the R2 advantage that the intercept receiver enjoys with
respect to the radar applies in all cases of interest. Hence, even if the intercept
receiver employs energy detection (as it must in almost all cases), it will always
detect the radar signal if it is observing in the right direction at the right time. The
clue to any potential LPI advantage of the impulse radar must lie in its extremely
low duty factor, which may be on the order of 10- to 1 0 -g, and its extremely large
bandwidth, which may be as large as 10 GHz. The analytical basis for an LPI
advantage is discussed in more detail in Section 3.4.4.

3.3.14 Low Probability of Exploitation

Impulse radar may provide a means of avoiding exploitation without requiring mul-
timod& operation. The basis for this conclusion is discussed in Section 3.4.4.

3.3.15 Nonapplications

Several applications of more conventional radar do not appear to be feasible for im-
pulse radar. Although impulse radar may have an advantage in the detection of LO
targets, it does not appear to be particularly useful for early warning applications.

46
I
I
Conventional radars used for early warning use a very low pulse repetition rate

I with a great deal of radiated signal energy in each Pulse. This large signal energy is
essential to accomplish detection of aircraft or missile targets at ranges up to sev-
eral thousand miles. Obtaining an equal amount of energy in a short intense pulse
does not appear possible with present technology. However, as the technology of
high-power pulse generation advances, this conclusion may be changed, particularly
if generation of closely spaced high-power pulses becomes feasible.
A frequently expressed opinion is that the limit on the peak power that can
3 be generated in a single pulse is set by limits on electrical breakdcywn. However,
electrical breakdown of air or any other material requires the electrical held to be
present for a non-zero interval of time. It has been demonstrated that pulses that
are shorter than this minimum time ("sneak-through") will not cause the material
1 to break down. Thus, the limit imposed by electrical breakdown may be higher
than that expected from long-pulse measurements.
3 Another nonapplication for impulse radar appears to be over-the-horizon (OTH)
radar. Operation of such a radar requires the use of frequencies below 30 MHz
5 because reflection from the ionosphere is required. Although pulses that concentrate
the energy at frequencies below 30 MHz are certainly possible, such pulses will have

M a duration sufficiently long that sneak-through cannot occur. Thus, there will be
a limitation on peak power due to electrical breakdown and a corresponding limit
I
on the aw-'xble energy per pulse. It appears likely that the limit on pulse energy
is such t..4dL OTH operation is not feasible. Furthermore, even if sufficient energy

could be achieved, the range resolution is such that there would be smaller returns
g from targets of interest. This loss would be partially compensated by a reduction
in clutter energy, but it is unlikely to be a sufficient reduction because of the diffuse
nature of reflections from the ionosphere.
Although measurement of target velocity appears to be feasible with impulse

.5 radar, such measurements are not likely to be as accurate as measurements made

* 47

I
with a sinusoidal-signal radar with the same amount of returned signal energy. This
conclusion has not been verified by analysis as yet, so it may be incorrect. However,
if it is correct, then any application of radar in which velocity measurement is the
primary objective may be better carried out by conventional sinusoidal radar. An
application that comes to mind is that of police radar used to observe speeders.
However, to offset this possible disadvantage is the fact that conventional police-
radar warning devices would probably not detect the impulse-radar signal.
As noted earlier in this report, propagation through the ionosphere would result
in substantial distortion of impulse-radar signals. Applications requiring tramn-
ionospheric propagation will probably be better served by conventional radar.
It now appears that suitable angular directivity for impulse-radar antennas can
be achieved only by the use of arrays. Because arrays are likely to be more complex
and more costly than the parabolic dish antennas used so often for conventional
radar, it may uneconomical to employ impulse radars where their unique capabilities
are not really required, even if somewhat better operation can be achieved with
impulse radar. Similarly, any application requiring the use of s single antenna may I
not be feasible for impulse radar because of deficiencies in existing isolators and
circulators. This may be a temporary limitation, however, as the technology of
such devices becomes more advanced.
I
3.4 System Design and Analysis I
Because impulse radar is significantly different from conventional radar, many new
aspects should be analyzed before a viable system design can be achieved. This
section outlines the nature of some of these aspects and suggests the approaches
that seem most reasonable. In almost all cases the nature of the required analysis
strongly depends on the applicatin for which impulse radar is employed. However,
it is not our intent to provide quantitative results on any of these topics, but simply j

48
I

to outline some of the analyses that ought to be done.

3.4.1 Optimum Signal Design

Unlike conventional sinusoidal-signal radars, it is necessary to distinguish between

3I the signal waveform at the transmitting antenna and the form of the radiated EM
field. Harmuth [Harmuth 1983b] has shown that the far-zone electric field is very
3 Inearly proportional to the time derivative of the transmitting antenna current.
Consequently, the bandwidth of the energy-density spectrum of the radiated signal
Sdepends primarily on the rise time and fall time of the antenna current pulse and
is almost independent of the pulse duration.
I Also, Harmuth has shown that some constraints on the form of the antenna
current must exist in any physical system. In particular, if the antenna current is
U gto be zero prior to the pulse, then both the antenna current and the derivative of
the antenna current must be initially zero.
The actual shape of the antenna-current pulse is determined primarily by the
g mechanism used to generate this pulse and may be difficult to control. It is assumed
here, however, that some control can be exercised over the rise time and fall time
3 Iof the antenna current pulse, and that the timing of antenna-current pulses can be
precisely controlled.
3 Avoiding range ambiguities in the intended application requires that the repeti-
tion interval between pulses or pulse groups be greater than the round-trip propaga-
tion times for the expected ranges. Apart from this, however, there is considerable
flexibility in designing pulse groups that have special properties. For example, one
3 can design a pulse group with variable spacings that has a "single-coincidence"
property. That is, for any relative time shift between the transmitted signal and
3 the received signal, other than zero, at most one pair of pulses coincide. Such a
pulse group has a correlation function with the low' correlation sidelobes necessary
I for good range resolution but provides increased signal energy for better detectabil-

I49
ity. Such pulse groups may well be optimum for those applications of impulse radar,
such as identification of LO targets, in which both excellent range resolution and
detection at large ranges are important.
Additional flexibility in signal design exists if the antenna current rise and fall
times can be controlled. This type of control makes it possible to select the range of
frequencies in which the maximum energy density occults. Thus, if the application
involves penetration of a conducting medium, increasing the rise and fall times of
the current pulse will put more radiated signal energy at lower frequencies and
improve the penetration capabilities. This flexibility also has implications for EM
compatibility, which is discussed in Section 3.4Z5.
A third aspect of optimum signal design has to do with power control. The LPI
capability of a radar is enhanced if the radiated signal energy is no greater than that
needed for detection of the desired target at the desired range. This is particularly
true at short ranges where the range advantage of an intercept receiver is not as
great. Most methods of generating antenna-current pulses lend themselves readily
to controlling the magnitude of the current and, hence, the radiated signal energy.
A final aspect of optimum signal design pertains to controlling the shape of the
energy-density spectrum of the signal by controlling the rise and fall times of the
antenna current pulse precisely. It is desirable to produce a spectrum having a more
nearly constant energy-density over its useful bandwidth. This requires a particular
shape that can be determined analytically for the leading and trailing edges of the
pulse. Whether this type of control over the pulse shape is feasible is not yet known.
The possibility of altering the pulse shape from pulse to pulse suggests some
interesting applications. For example, a mode of operation analogous to frequency
hopping can be achieved by changing the pulse rise and fall times from pulse to
pulse. This permits the frequencies at which maximum energy density occurs to
be changed during operation and enhances the possibility of detecting targets that
may have large responses at specific frequencies.

50
I
I
3.4.2 Jamming and Countermeasures

It is relatively easy to jam conventional narrow-band radars because the jamming

3
Isignal energy can also be concentrated in a narrow band. Typical approaches to jam-
ming ronventional radars include tone jamming, swept-frequency jamming, barrage

3 (or noise) jamming, and pulse jamming. In addition to such brute-force techniques
there are also "spoofing" techniques, such as repeat jammers, which cause the radar
signal to provide incorrect information.
Jamming an impulse radar is likely to be much more difficult. This is because a
3 typical narrow-band jammer signal can be almost totally filtered out at the input
to the radar with very little degradation to the desired radar signal. Although
there are various ways to achieve this result, use of a channelized input to the radar
receiver is convenient for eliminating the jamming signal and provides a means of
approximating the matched-filter receiver by controlling the gains of the individual
channels. Any channel in which a jamming signal occurs can possibly be switched
off while the clear channels remain intact.
-•The use of a repeat jammer for impulse radar also appears to be fairly difficult.
The repeater must have a bandwidth comparable to that of the impulse radar and
3 Ibe able to provide a constant delay throughout this bandwidth. Although this is
not impossible to achieve, it is much more difficult than the present approach.
3 Although impulse radar appears to be fairly immune to conventional jamming
techniques, much more analysis is required to evaluate jamming vulnerability com-
* pletely.

S3.4.3 Clutter Rejection Analysis

Clutter is any radar return from objects other than the desired target. For a land-
based radar, clutter arises most often when the desired targets are only few degrees
above the horizon and results from trees, buildings, hills, ocean waves, etc. For
airborne radar looking for targets on the ground, clutter from the ground fills the

* 51
entire radar beam and usually represents a signal power considerably larger than
the power from the desired target. Conventional radar systems use several different
techniques to reduce the effects of clutter. In the case of moving targets, there are
ways to cancel returns from stationary objects on a pulse-to-pulse basis. Improving
the range resolution also reduces the clutter power in each range cell and enhances
the possibility of seeing the target.
The excellent range resolution of impulse radar should minimize the clutter
return in each range cell. Not only can this conclusion be obtained from theoretical
considerations, but it has also been demonstrated in actual radar tests. An example
known to one of the authors (GRC) of this report occurred with the Random-
Signal Radar developed at Purdue University in the early 1970's. Although this
radar is a spread-spectrum radar rather than an impulse radar, it has a range
resolution of about 15 cm and, unlike conventional spread-spectrum radars, has no
range sidelobes because successive pulses are statistically independent. Thus, in this
regard at least, it is very similar to impulse radar. Tests were performed with this
radar by the Navy at Panama City, Florida, in competition with more conventional
radars. The targets were hollow steel balls, about 30 cm in diameter, floating in
the Gulf at a range of about 1200 m. The Random-Signal Radar was able to detect
these floating objects because the 15-cm range rezolution reduced the sea clutter by
a sufficient amount. No conventional radar was successful in detecting the floating
balls because of excessive sea clutter. Thus, it is clear that improving the range
resolution does in fact substantially reduce the clutter signal, and impulse radar
should be better in this regard than conventional spread-spectrum radar because
impulses have no range sidelobes.
The use of improved range resolution to reduce clutter is not an unmixed bless-
ing, however. When the target of interest is substantially larger than the size of
the range cell, as it must be for target identification, the energy of the return from
each scattering region on the target is substantially less than the total energy re-

52
I
U
turned from the target. Hence, even with reduced clutter there may not be a net
I improvement in detectability. In general, it is difficult to predict the results without
a substantial amount of analysis and a precise knowledge of the target scattering
regions and aspect angle.
An important aspect of clutter analysis for impulse radar is to have good mod-
els for the return signals from typical clutter sources when they are illuminated
by short-pulse EM waves. On the basis of available information, it appears that
such clutter signals may be significantly different from typical clutter signals for
3" conventional radar. In a conventional radar, for example, the clutter return looks
much like noise and is frequently modeled as additional Gaussian noise in the sys-
Stem. Performance calculations are then made on the basis of the usual Gaussian
noise models. For impulse radar, the clutter returns may well look like impulsive
3 noise and have completely different statistics. Whether these different statistics are
favorable or unfavorable to the detection process is not yet known.
3 Another aspect of the clutter performance of impulse radar is its performance in
the presence of radar chaff. Typical chaff packages contain dipoles of different sizes
I so that there are many frequencies at which near-resonant returns occur. The RCS
of such chaff for sinusoidal waves may be on the order of 40 m 2 over a frequency
I
Urange from 3 GHz to 12 GHz [Barton 19881. However, the return signal will look
quite different to an impulse radar. A typical chaff dispersion will result in about
one dipole per 17 mn, which suggests that the average spacing may be on the order of
2.5 m. Because this separation is greater than the range resolution of impulse radar,
each chaff dipole produces a distinct return. This should simplify the problem of
3 distinguishing chaff from targets, particularly since the chaff return may fluctuate
more from pulse to pulse than the target return. In order to analyze this problem
3 in more detail, it is necessary to determine the time response of chaff dipoles of
different sizes to impulse interrogation.

I
* 53
3.4.4 Probability of Intercept and Exploitation

Impulse radar might be applied to situations in which interception of the radar

6ignal i~tto be avoided. The usual way to give a radar a low probability of intercept
is to employ a spread-spectrum signal with a large time-bandwidth product. If the
processing gain achievable from the large time-bandwidth product is greater than
the range advantage of the intercept receiver, then detection can be avoided. If the
processing gain in not large enough (mid usually it is not), then LPI performance
is achieved by avoidanice, that is, by not transmitting at the time and place being
e~xamined by the intercept receiver, if possible.
In the case of impulse radar, the time-bandwidth product of the signal is es-
sentially unity. Hence, there is no possibility of using processing gain to avoid
detection. On the other hanid, the bandwidth is so large that typical intercept
receivers can observe only a small fraction of the total transmitted energy. This
concept is explored quantitatively below.
Assume that the intercept receiver is using a ,:ompressive receiver that is capable
of scanning over a 50-MHz bandwidth in 1.00 As and that it has a frequency reso-
lution of 5 kHz. These numbers are somewhat representative of existing intercept
receiver capabilities and are adequate to detect conventional radar signals. Next,
assume that the effective bandwidth of impulse radar signal is 20 GHz. This implies
that the fraction of the total signal energy that falls within any one resolution cell
of the compressive receiver is 2.5 x 10-1 or -66 dB. Because the range advantage
enjoyed by the intercept receiver is likely to be on the order of 50 to 60 dB, there is
a 6- to r16-dB margin of protection for impulse radar. This is quite likely to result
in a fairly small probability of interception.
The second aspect that is favorable to impulse radar is its very small duty
factor. Most intercept receivers search a given solid angle in space in a specified
frame time so that time spent in each angular resolution cell may be on the order
of 1 ms. If impulse radar is not transmitting during the particular 1 ms that its

54
angular resolution cell is being interrogated, it will not be intercepted. Hence, the
3 probability of intercept is simply the probability that the very narrow impulse radar
pulse falls within a given 1-ms time slot occurring once every frame time. This is
I also likely to be a small probability, particularly if impulse-radar PRF is low.
Exploitation of a radar signal is the ability to use an enemy radar signal for
some purpose, missile homing on the radar transmitter, for example. Obviously,
exploitation requires the interception of the radar signal to have already taken place.

Homing receivers tend to have a bandwidth that is as wide as the conventional radar
* bands so that any radar operating within a conventional band will be detected.
Because of the extremely large bandwidth of an impulse-radar signal, the fraction
3 of the total signal energy within the p&-s band of the homing receiver will be small,
particularly if impulse radar has its region of maximum energy density outside of the
3 conventional radar bands. Thus, there is a reasonable probability that the homing

receiver will not receive sufficient signal energy to accomplish its mission.
I
3.4.5 Electromagnetic Compatibility

I Because any radar must operate in an environment in which communication equip-

ment and. other radars are also operating, the potential for interference between
systems is great. Unless they are operating in close proximity, conventional narrow-
3 Iband systems are unlikely to be a serious source of interference for impulse radar.
However, if these narrowband systems are operating in close proximity to impulse
radar and within its antenna beamwidth, the total interference energy delivered
into the impulse-radar receiver may be sufficient to seriously degrade performance.
3 It is unlikely that impulse radar will be a serious source of interference for such
narrow-band radar and communication systems, unless they are operated in close
3 proximity. The large peak pulse powers of impulse radar can easily saturate the
front ends of most conventional receivers and disable their operation for a period of
3 time that is many times larger than the duration of impulse-radar signal. Modern

I 55
I
communication systems tend to be digital and to utilize extensive error-correction
capability. The short pulses from impulse radar, even if they saturate the communi-
cation receiver front ends, are not likely to destroy more data than can be corrected
by forward burst-error correction.
Nevertheless, there may well be other types of equipment for which the pos-
sibiLlity of interference is serious. For example, interference in the sidelobes of
angle-tracking systems can cause large errors. Thus, radar fire-control systems,
missile-guidance systems, and track-while-scan systems are all potential victims of
impulse radar. Because such systems usually operate at frequencies higher than 8
GHz, these difficulties may be alleviated by appropriate spectrum control of impulse
radar.
The problem of EM compatibility of impulse radar and conventional systems is
indeed a serious one and deserves a great deal of analysis. There are undoubtedly
teci niques that can ameliorate the situation. For example, peak signal limiting at
the inputs to conventional receivers may reduce the amount of interference. Infre-
quent operation of impulse radar, coordinated with other systems, will also minimize
interference. In addition, there are some applications of impulse radar for which the
interference may be negligible or non-existent. For example, the ground-penetration
application may result in minimal interference if the sensor is close to the ground.

3.4.6 False Alarms, Detectability, and Missed Targets

The probabilities of detection and false alarm are important parameters for any
radar system. The analytical determination of these probabilities depends primarily
on the assumed fluctuation model of the target, the signal energy per pulse, the type
of detection employed, and the receiver noise model.
In conventional radar, target fluctuation results from sinusoidal wave returns
from various parts of the target adding vectorially in different ways as the aspect
angle changes slightly. This phenomenon is not likely to occur for impulse radar

56
I
because the range differences between reflecting points of most targets is likely to
3 be greater than the range resolution. Hence, each reflecting region will appear as
a distinct target. This result is fine when target identification is the issue, but it
3
! is a serious disadvantage with respect to target detectability because the energy
returned from each reflecting region is smaller than the total energy returned from
3 the target. Therefore, the question arises: is there any way of combining all of the
energy from a given target before the detection process?
One way of acquiring all of the energy associated with a given target is to employ
an energy detector such as the one in Figure 3.4. It can be shown [Torrieri 1981]
that the probability of false alarm for such a detector is given by

IPFA[= Q 2TB] (3.1)

3 where Q is the ordinary Q-function for the normal probability density function, VT is
the detection threshold, No is the one-sided noise spectral density, and the product
TB is greater than about 30. The value of T must be selected large enough so that
the integration is over all the returns from the target. It is on this basis that it is
3
Ireasonable to assume that TB is sufficiently large. However, this large integration
time destroys the range resolution of the radar. The corresponding probability of
SI detection for this situation is given by

IPD (2EI1No)12
V1 + -(2E/No)/2vTB]
Q[ Q1(PPA)
f (3.2)

I where E is the signal energy available in all of the pulses included in the integration
time T.
The alternative to energy detection is the use of a matched filter. Unless the
target structure is known a priori, the matched filter must be matched to a single
3 returned pulse. Although the matched filter provides the maximum signal-to-noise
ratio for each return pulse, the energy per pulse will be less than the total energy
from the target. Hence, this approach may not provide the greatest detectability.

I 57
IPT BPF • :2 T :UTU

SIGNAL OUTPUT

Figure 3.4: The energy detector for impulse radar

If the matched filter is used, however, the probability of false alarm is given by
[Diftanco 1968]
PFA = Q(VT) (3.3)

and the probability of detection for a single observation of a nonfluctuating target

is given by the following equation.

PD =Q (VT -- V2 ,E/RO) (3.4)I

When the issue is target identification rather than detectability, the matched
filter is the obvious choice because it provides the grcatest rangr resolution and
range accuracy.
A•m a means of comparing the probabilities of detection for these two type of

detectors, assume that we have a target with 10 reflecting regions spanning a range
of 100 range cells and that each reflecting region is returning a signal energy so that
2E/No = 20. If the probability of false alarm is fixed at 10', the probability of
detection for the energy detector becomes 0.999 while that for the matched filter
is only 0.39. This is a clear example of how the choice of detector depends on the
application.

58
I
U
3.4.7 Overall Performance Analysis and Comparison

Although no extensive comparisons of overall performance have yet been made

3 between impulse radar and conventional systems, such studies are of great im-
portance and urgency. The analysis presented in Appendix B does make some
3 comparisons of antenna beamwidth and range resolution and finds that these as-
pects are in reasonable agreement with the conventional wisdom. There are many
other aspects of radar performance that also ueed to be compared. These include
the maximum range capability, performance against stealth targets, LPI capability,
3 clutter-rejection capability, and jam resistance.
In addition to comparing impulse radar with conventional radar, it is also es-
U sential to compare various system designs for impulse radar itself. Of particular
concern here are various methods of generating the antenna-current pulses and vax-
ious methods for implementing the detection process. For both of these items there
is little previous experience to serve as a guide.

59
THIS
PAGE

MISSING
IN
OR1IGINAL
DOCUMENT
I

I
I Chapter 4

IU Critical Issues in Impulse Radar

3 This chapter deals with a few of the critical issues that must be addressed in any
serious investigation of the potential for impulse radar. These issues have been
discussed in some detail in previous sections; the intent here is simply to outline
aspects that require further consideration and to suggest reasonable approaches.
I The ordering of these critical issues is not in order of priority, but rather in order
of occurrence in the radar process.

* 4.1 Optimum Signal Design

Each potential application of impulse radar will likely require special consideration
as to the signal waveform that is optimum for that particular application. For
3 example, applications that require target detection at large ranges must empha-
size techniques that produce the greatest amount of radiated signal energy in each
pulse. For these cases, the precise form of the signal is of less importance because
the achievement of extremely good range resolution is not necessary. Therefore,
relatively slow rise and fall times on the antenna current pulses, but very large
maximum currents, are desirable in order to put more energy at the lower frequen-

c6ies.

61
On the other hand, if the application involves target identification, then very
good range resolution is necessary and target detection is of lesser importance.
In this case, obtaining the largest bandwidth of the radiated signal is of great
importance and effort must be directed toward achieving very fast rise and fall
times and very short duration of the antenna-current pulses. Because the EM
scattering process at the target is complex, other considerations, such as polarization
effects, must be considered in designing the pulse structure. Therefore, in order to
determine the optimum signal design for a particular application, complete and

accurate modeling of the various processes will be required.

In either of the above situations, it seenms likely that a radiated. energy spectrum
that is as nearly uniform as possible over the range of useful frequencies is desir-
able. Achieving such a uniform spectrum requires additional analysis to determine
appropriate shapes for the antenna-current pulses. This must be done, of course,
with due regard to achieving such shapes with practical hardware.
A second aspect of optimum signal design is that of achieving the greatest pos-
sible efficiency in terms of the ratio of the radiated energy per pulse to the stored
energy per pulse in the current-pulse generator. As noted previously, the duration
of the pulse is almost irrelevant with regard to the radiated energy because the
radiated energy depends on the rise and fall times of the pulse. However, both

the stored energy in the pulse generator and the shape of the resulting radiated
energy spectrum (particularly the peaks and valleys of the spectrum) do depend on
the pulse duration. Thus, the relationships among all of these aspects of the pulse
shape of the antenna current need to be analyzed in detail.
A third important issue has to do with the possibility and desiri.bility of radiat-
ing pulse groups with coded pulse spacings and/or coded amplitudcs to increase the
energy available without seriously degrading the range resolution. Several analyti-
cal techniques for designing such pulse groups are knowIL, but the extent to which

they can be implemented with existing hardware is unknown.

62
I A final issue is that of changing the pulse shape from pulse to pulse to achieve
3 both excellent detectability and excellent range resolution concurrently. A subissue
is that of power control on individual pulses to enhance the LPI properties of the
* radiated signal.

3 4.2 High-Power "Impulse" Generators

3: Emphasis on impulse-radar R&D for the immediate future will be on applications

requiring higher peak power than that used in short-range applications already
Sdeveloped such as ground penetration and intrusion detection. An ideal device
for the generation of ultra-short pulses of extremely high peak power is the PCPS
described in Section 3.2.1. The parameters of the PCPS can be scaled [Nunnally
and Hammond 19831 over many orders of magnitude to satisfy a wide variety of

* applications.
Redesign for production of PCPS devices is required. For example, the feasibility
3 of integrating the laser and the semiconductor switch into a single device must be
considered. Such a redesign for production by a commercial firm is already in

I progress.
Further modeling and experimentation is required to determine what modifica-

tion of the PCPS may be desired to facilitate the use of rise-time control in the
impulse-radar transmitter. Much of this design can be accomplished using existing
PCPS modeling codes [Iverson 1988]. Also, versions of the PCPS device scaled down
3 in peak power ought to be considered for use in existing short-range applications of
impulse radar.
I
4.3 Antenna Design

Design of antennas from transmission and reception of transient signals is one of the
3 most critical areas of impulse-radar R&D. As noted earlier, the starting point must

63
be at the level of basic principles. The standard antenna theory based on sin usoidal
excitation will be of little use in impulse-radar R&D.
Analytical studies of proposed antenna designs are difficult to the point of im-
practicality, even for the simplest configurations. Extensive use of numerical tech-
niques will be required. Existing computer codes, such as TWTD will be used and
extended as required. Other time-do,'w.:,i ý EM codes will need to be developed for
use in new types of antennas.
Traveling-wave antennas appear to be one class of antennas applicable to im-
pulse radar. Examples include the classical Beverage and rhombic antennas. More
pertinent examples are the impulse-radar antenna developed at RADC [Figure D.2],
and the TEM horn used at the University of Texas, Arlington. The dimensions of a
6raveling-wave antenna are likely to be much greater than the distance light travels

in the rise time of the antenna current; 10-100 ps =4. 3-30 mm. Thus, disper-
sion effects in antennas must be carefully considered in the design of impulse-radar
systems.

Materials science studies of radome materials for transmission of UWB signals

are also critically needed in impulse-radar R&D. Dispersion and refraction effects of
radomes on impulse-radar signals must be studied and taken into account in system
design.

4.4 Transient Electromagnetic Theory

One of the principal reasons for pursuing the development of impulse radar is that
certain highly desirable, "advanced" radar capabilities may be achievable. These
capabilities include, but are not limited to, target ideritification, target imaging,
"stealth" target detection, and various "penetration" capabilities. Regardless of
whether advanced or rudimentary capabilities are being considered, the fundamen-
tal processes governing impulse-radar operation and performance are transient elec-

64
I
tromagnetic processes. In particular, all applications of impulse radar require the
emission, propagation, scattering, and reception of short-duration, electromagnetic
pulses. This sequence of transient electromagnetic processes is the foundation on
* which impulse radar is based. The development of impulse radar will, therefore,

require a strong effort in the formulation and solution of transient electromagnetic

3 problems.
In recent years, there has been a substantial expansion of interest in the engineer-
ing application of linear electromagnetic wave theory, especially in areas involving
penetration and scattering interactions with electrically large complex structures.
3 This interest is driven primarily by the need to understand the characteristics of the
RCS of complex targets for conventional sinusoidal radar signals. Typical structures
of engineering interest have shapes, apertures, cavities, and material compositions
which produce near fields that cannot be resolved into finite sets of modes or rays.
3 Generally, electromagnetic wave interactions with such structures pose intractable
problems when addressed by existing analytical and asymptotic theories. Therefore,
U computational. approaches mudt be used to address these important problems.
Accurate numerical computation of full-vector electromagnetic wave interactions
with arbitrary structures is difficult but necessary for the development of impulse

radar. These computations are required not only for the analysis of RCS and

the scattering of transient electromagnetic wave, but also for the development of
antenna structures for the transmission and reception of impulse radar signals.
For effective formulation and solution, the transient electromagnetic processes in
3 impulse radar require a time-domain approach rather than the frequency-domain
approach traditionally used in engineering electromagnetics.
3 Appendix A provides an overview of time-domain EM theory and presents a
quintessential problem in transient EM scattering and transient excitation of an-
3 tennas.

65
4.5 Receiver Design

The design of a receiver for impulse radar signals is critically dependent on the
intended application for the radar. As noted in a previous section, the range res-
olution of impulse radar is likely to be sufficiently good so that individual returns
from separate scattering regions on a target of reasonable size are distinct. Since
each distinct return has less energy than the sum of the energies from all of the
scattering points, the target detectability suffers unless some method of combining
signal energies can be achieved. One approach to doing this, the energy detector,
is outlined in Section 3.4.6 in a discussion of the probabilities of false alarm and
detection. By making the integration time of the energy detector long enough to
include all of the returns from the target, one can substantially increase the prob-
ability of detection. The penalty for this improvement is that the effective range
resolution is now degraded to a value that is approximately the same as the size of
the target.
However, the energy detector does not use the available signal energy in the most
effective manner. To achieve the largest output signal-to-noise ratio, it is necessary
to employ a matched filter that is matched to the returned signal. This implies
that for a complex target, the matched filter must be matched to the sum of all

returns from all scattering regions on the target. If this were actually possible, the
probability of detection would be much greater than that possible with the energy
detector. Unfortunately, the only way to do this is to know the precise nature of the
target before it is detected - a very unlikely circumstance. However, even in this
idealized situation, the individual scattering regions are not resolved so the range
resolution is again degraded.

To retain the range resolution it is necessary that the receiver matched filter is
matched to the return signal from each individual scattering region on the target.
Since the return from each individual scattering point is likely to be a replica of
the transmitted signal, the construction of such a matched filter is possible, at least

66
I in theory. In practice, it is necessary to approximate impulse response of the ideal
matched filter. This point is discussed in more detail below.
An alternative to the matched filter is to cross-correlate the received signal with
an appropriately delayed replica of the radiated signal (not t. "eplica of the antenna-
Icurrent pulse). This method is exact and does not involve ay approximations. The

problem with this approach is that a different replica of the radiated signal and a
separate correlator are needed for each range cell. Although such an approach
3 is possible in theory and has some very attractive features, it is very hardware
intensive.
3 The above discussion reveals two conflicting requirements on the design of an
impulse radar receiver, depending on whether the primary goal is target detection
3 or range resolution. Fortunately, there appears to be a way of approximating both
objectives within the same receiver. This approach requires parallel channels in the
receiver input, each covering a portion of the total signal bandwidth and each with
individually controlled gains. The detectability of the energy detector is achieved

SI by setting all gains equal and summing the outputs of the channels into a square-law
device and integrator. The matched filter can be approximated by adjusting the gain
of each channel to be proportional to the average amplitude of the radiated signal
frequency spectrum within the channel spectrum. Thus, it is possible to visualize a
dual-mode method of operation in which the approximate energy detector is used
for initial target detection, and the approximate matched filter is used for target

identification. A critical issue in the design of such a parallel-channel receiver is that

p of achieving sufficient phase linearity in each channel to avoid seriously distorting
the signal.
"3
IAnother critical issue in receiver design is that of providing sufficient isolation
between the transmitter and the receiver. The usual isolation devices, such as
3
Ibalanced duplexers and ferrite circulators, are inherently narrow-band devices and
are unlikely to be useful in impulse radar. Probably the most effective approach is to

I 67
I
I
use separate antennas for transmission and reception and use spatial 'solatioli. Even
in this case, however, it may be necessary to employ shorting switches at the receiver
input to reduce direct signal feed-through sufficiently. The main disadvantage is an
increase in the minimum range of the radar because of nonzero switching times.

4.6 Signal Processing and Analysis

Processing and analysis of the received signal is another critical issue. Signal pro-
cessing and analysis is of particular importance when advanced radar capabilities
such as target identification and target imaging are considered. The major aspects
of this issue were outlined in some detail in Section 3.2.4, and axe summarized here.
The principal objectives of signal processing and analysis ar , to extract the
signal containing information about the target from the output signal of the receiver

and then to extract information about the target from that signal. Of particular
importance is the determination of optimum transmitted signal structures for a
particular application through the complete modeling of the impulse-radar process.
It is essential that approaches to signal processing and analysis be developed prior
to the final design of the impulse-radar system.

A. detailed study of various signal-processing methods should be undertaken.

This study should evaluate the performance of energy detectors, matched filters, and
cross-correlators for the receiver and, for the major applications of the impulse radar,
make recommendations as to the type of signal processing that is most suitable. Of
particular concern is the feasibility of hybrid-processing methods that might be
required for target detection and target identification.

A final critical signal-processing issue is that of performing the necessary pro-

cessing tasks when several targets are present. This is of particular concern in
track-while-scan systems and in target-discrimination systems.

68
I
I
I
1 Appendix A
I
Time-Domain Electromagnetics

I The complex physical processes involved in impulse radar reside primarily in the
domain of classical electromagnetics. The equations of classical electromagnetics,

I Maxwell's equations, are partial differential equations (PDEs) for the electric and
magnetic fields as a function of spatial position and time. However, the exposition
of EM theory often presents Maxwell's equations in the frequency domain. This is
accomplished through the application of the Fourier transform. and results in ordi-

nary differential equations (ODEs) for the electric and magnetic fields as a function
3 of spatial position. Under the Fourier transform, time, which is an independent
variable in Maxwell's equations, becomes frequency, which is a parameter in the

resulting ODEs and not an independent variable. Frequency then becomes an im-
portant parameter in the subsequent theoretical development and in the solution
of practical probleim in physics and engineering. This approach has proven useful
in classical communications and radar systems engineering which involves signals
characterized by narrow relative bandwidths.
Impulse radar involves EM excitations that are transient signals having a wide
relative bandwidth. Therefore, it is more natural and efficient to formulate the
mathematical equations governing the physical processes in terms of Maxwell's
equations in the time domain. Although frequency-domain ideas are still useful, the
problems to be addressed in impulse radar are better approached using Maxwell's

*69
equations in their original form in terms of spatial position, and time. Also, many
of the measurements required in impulse-radar research and development are best
made in the time domain. Study of literature on impulse radar and related theory
and technology shows that most of the recent progress has been based. on analyses
and measurements in the time domain. This Appendix presents an overview of
theory and problems in the important area of time-domain electromagnetics.

A.1 Maxwell's Equations

This section provides a basic introduction to time-dependent or transient EM theory,
beginning with Maxwell's equations, as it applies to problems of importance to UWB
radar. Maxwell's equations describe the evolution of EM field quantities and can
be written in MKS rationalized units as:
V XH = J+ -- (A.1)
at
VxE = aB (A.2)
V-D = p (A.3)
V-B = 0 (A.4)

where E and D are the electric field and displacement vectors, H and B are the
magnetic field and induction vectors, J is the conduction current density, and p
is the electric charge density. Equation (A.1) is Amperes's law to which Maxwell
added the displacement-current term !-D which is required to make the system of
equations consistent. Equation (A.2) is Faraday's law, equation (A.3) is Coulomb's
law, and equation (A.4) states the absence of magnetic monopoles. Equations (A.3)
and (A.4) represent auxiliary conditions on the principal Maxwell equations (A.1)
and (A.2).
For macroscopic media, the dynamical response of the aggregate atoms is sum-
marized in the constitutive relations.

70
ID = E (A.5)

J = o-E (A.6)

I B = H (A.7)

In general, the electric permittivity E, the electric conductivity a, and the magnetic
permeability it are tensors. However, for the purposes of this report on UWB radar,
I the media of interest will initially be considered as being isotropic, so E,a and /t
reduce to scalar parameters.
"I Maxwell's equations consist of a set of four coupled first-order PDEs relating

I "the various components of electric and magnetic fields. It is useful to introduce

two potentials, which allow these four first-order equations to be written as two
second-order equations. These potentials are the scalar electric potential ib and the
magnetic vector potential A. Since V B = 0, B can be defined in terms of a vector
I potential A.
B = Vx A
(A.8)

Faraday's law (A.2) can then be written as

I v [ E+-± j =o (A.9)
This means that the quantity with the vanishing curl in equation (A.9) can be
I written as the gradient of a scalar function, namely, the scalar electric potential V.
SE + atA -v (A.1o)

E= -VI- dA (A.11)

I The definition of B and E in terms of the potentials A and 0 according to equations

(A.8) and (A.11) satisfies identically the two homogeneous Maxwell equations. Tly-.-
I two inhomogeneous equations determine the dynamic behavior of A and 40. We can
write these two inhomogeneous equations in terms of A and ip

I 71

I
V 2A -- _V A + EIpy- j= -PJ (A.12)

o a.
-- A =-.-A (A.13)

The set of four Maxwell equations has been reduced to a set of two coupled
equations, (A.12) and (A.13), by introducing the potentials A and tP. These equa-
tions can be uncoupled by exploiting the fact that an arbitrary factor can be added
to the definition of the potentials. For the definition of B in terms of A, the choice
of A is arbitrary in that the gradient of some scalar function 5 can be added. Thus
the definition of B is left unchanged by the followving transformation.

A -- ) A + Vý (A.14)

In order that the definition of the electric field E remain unchanged, the simulta-
neous transformation
¢ ' '¢-- V)at o t~b(A. 15)

is required. The potentials A and 0 are now free to be chosen such that
v. A+cL =0 'Ao6

which is referred to as the Lorentz gauge. This particular gauge uncouples equatiorks
(A.12) and (A.13) which can then. be written as follows.

V2 A - t-•- -- -A, (A.17)

a 2 4,
...p (A.18)

This set o• two uncoupled second-order equatiorn, along with equalt.on (A."I,;
is equivalent to the four Maxwvell equations in every respect. These equ..tionw.. are
hyperbolic and exemplify the propagating nature of EM disturbances. The quantity
,E is related to the velocity of propagation U of EM radiation by the following
equation.

72
U (A.19)

The electric permittivity c and the magnetic permeability A are generally written in
terms of their free space values, Co and •oj times their values relative to free space,
C. and Ar.
= O (A.20)

The value of the free space permittivity is Eo = 8.845 x 10-1 4 F/cm and the value
of the free space permeability is AO = 47r x 10- H/cm. The velocity of propagation
can then be written as
U 1 (A.22)

where c = 1/cV/"• is the velocity of light in free space.

These equations are given without a reduction to "harmonic" form and, there-
fore, represent the governing equations for complete time-dependent problems of
EM theory.

A.2 Transient Electromagnetic Scattering

The application of EM scattering theory to the analysis of the scattering of UWB
radar signals by a target can be demonstrated in the context of a classical EM
scattering problem. The-problem to be considered is that of a transient EM field
generated by a current-sheet source and scattering from a dielectric slab having a
perfectly conducting backplane. The problem is presented here as a one-dimensional
formulation derived by assuming an infinitely extended planar structure. Although
this formulation may not appear to represent a physically relevant problem, it does
in fact contain the essential features of an entire class of physically important prob-
lems and represents a model that would provide a first step towards understanding
a number of critical features of UWB radar.

73
To begin the formulation of this specific problem, consider Maxwell's equations
in the time domain, as presented in section A.1, for the case of an excitation hav-
ing the form of an impulse-radar signal with a temporal structure given by f (t).
Spatially, the excitation source is assumed to consist of an infinitely extended,
y-directed, current sheet J confined to a region of zero thickness at a position
X = d > 0 in the y, z plane. The current-sheet is parallel to and located at a dis-
tance d from the near surface (x = 0) of a dielectric slab of thickness h that is also
infinitely extended in the y-z plane. The far surface of the dielectric slab (x = -h)
is terminated by a perfect conductor. The current-sheet, having the temporal struz-
ture f(t), produces a wideband radiating field that scatters from the dielectric-slab
target structure.
The theory of Green's functions is utilized in obtaining the full time-dependent
solution to this problem, so the form of the driving current to be considered initially
is
J(r,t) =-(x - d)b(t)g (A.23)

where r is the Cartesian spatial-position vector, the function 6 is the Dirac delta
function, and 0 is the unit vector in the y direction. The complete solution for
the case of UWB radar excitation can be found formally by a convolution of the
solution obtained for the delta-function excitation (the Green's function) with the
wideband-radar temporal function f(t).
Assume also that the material parameters as well as the problem geometry are
independent of y and z.
a- 0 a- 0 (A.24)

With this assumption, Maxwell's equations (A.1) and (A.2) reduce to the coupled
partial differential equations for the field components E. and H,,.

aEs = - (A.25)
ax at
7, (A.26)

74
All other field components equal to zero. In equations (A.25) and (A.26), the field
components E. and H, are functions of x and t only. Because it is assumed that
a, j, and c are constants throughout the region of interest, the equations (A.25) and
(A.26) are linear.
The time-domain problem is now transformed to a frequency-domain problem
by the application of the Fourier integral. The Fourier integral transformation of a
time-domain function g(x, t) into a frequency-domain function ^(x, w) is defined as
follows.
F [g(x,t)] = J g(x,t)e-' idt =-(XW) (A.27)
Applying the Fourier transform defined in equation (A.27) to equations (A.25)
and (A.26), and considering the resulting equations as defining a boundary-value
problem in the form of ordinary differential equations in the spatial variable x with
the frequency variable w as a parameter yields

dE (x;w) = -iwtsHC(x;W) (A.28)

d/-H,(x;dw) = b(x - d) + iwiEE(x;W) (A.29)

where E is the complex permittivity defined as follows.

= E+ - (A.30)
SW

Equations (A.28) and (A.29) can be combined to form a Helmholtz wave equa-
tion for the electric field

d 2 E~ A~
1d+z2 E iwj~b(x -d) (A.31)

where n is the complex wave number.

= (A.32)

The field fI, can be found directly from 4• using equation (A.6).
The solution of this problem requires the consideration of both regions, the
iritrrace at x = 0, and the boundary at x = -h. In the first region, 0 < x < d, the

75
medium is considered to be free space. Therefore, E and p. take on their free space
values EOand /IO, respectively, and ux 0. The second region is that of the dielectric
slab, -h < x < 0, and is characterized by E= C, a = a,, and A = p, = po. In the
free-space region, the complex wave number nCo is given by

/,o = IOO/2 = (A.33)

and in the dielectric-slab region, the complex wave number ic, is given by
/ \ 1~~1/2 - 1/
'Ca [lnES~j~ +
[r(1 -) 2 (A.34)

where cr is the relative permittivity (dielectric constant) of the dielectric slab.

The solution of this two-region problem requires techniques from the applied
mathematics of boundary-value problerm where the matching of solutions at the
interface x = 0 and at the boundary x -- h is required. The formal solution for
the electric field E, in the two regions is given as follows. For the free-space region,

bounded by the current sheet at x = d and the neat surface of the dielectric slab at
x -: 0, the solution is found to be
!•Ttjo ~sin KO:,,,r..
=" + E1 e-' (A.35)
1'CO
and for the dielectritc-slab region -h < x <0.,the ,olution is found to be
sin'.(x- + d)
. ... _.E l.. (A .36)
sin r.ad
where El is the electric field at the interface x = 0 given as follows.

LEI -- [irco + r., cot(K,h)] E, = iwI~oe-id (A.37)

"The final step in obtaining the complete solution E•(x, t) to this EMvl scattering
problem is to perform a convolution by tnultiplying the frequxency-domain solution
given above for the cafe of impulse excitation with the Fourier transform 1(w) of
the pasticular temporal impulse function I (t) o(f iaterest, and then to transform the
result back into the time domain using the inverse Fourier transforul,

7( w' , (,5,)) e"' 1A.38)

76
I
When this operation is performed for the free-space region by substitution of equa-
tion (A.35) into (A.38), the result is

E70 [tf t+ f x d+ f 0 w t
E (X..0 = 1 0
2 CJ C 7r oi i(k,/k0 ,) cot k~h j
I (A.39)

where rio is the intrinsic impedance of free space. The integral in equation (A.39)
can be solved by the calculus of residues. This equation represents a general solution

to transient EM scattering problem for arbitrary excitation.

A.3 Transient Excitation of Antennas

Time.-domain electromagnetics provides the basis for analysis and design of trans-
mitting and receiving antennas for imipulse radar. Computer modeling using nu-
merical solutions of EM problems will be the main tool for these design studies.
Low-power short-range measurements based on time-domain reflectometry will be
used to verify the performance predicted by the design studie.i. High-power mea-
surements may be deferred until later ,itages of the R&D program.
Beamwidths of individual antenna elements axe expected. to be broad. SIbstan-
tial antenna gain will, therefore, have to be provided through the use of antenina
3 arrays.
Preliminary studies of impulse-.radar anteý-ira &rrays [Appendix B1 indicate the
I need to study interaction of off-axis response with the matched filters used in the
receiver. This additional complexity may be offset by the ability to obtain additional
Sinformation iii the -radar returns, as compared with conventional radar.
TWTD codes may be used for the analysis and design of some classes of anten-

nas. However, soflXe of the new types of antennas required for impulse radar lust be
modeled using ot,hher approaches such as finite-difference or Galerkin finite-element
algorithlms. Traveling-wave antennas, in partlicular, will require car.ful modeling
Wand aialyiis using these new codes.

I 77
As noted in Section 3.2.2, there are substantial differences between the perfor-
mance of an antenna or an array excited by extremely short transients as compared
with that produced by steady-state sinusoidal excitation. One of the most important

of these differences is that, in general, the reciprocity theorem for antennas does not
hold for transient excitation. Robert Kelly has derived conditions for reciprocity in
the case of transient excitation and is preparing a paper on this subject.
Standard antenna theory is based on the assumption of sinusoidal excitation.
For example, proofs of reciprocity theorems in textbooks on EM theory depend on
this assumption. The UWB signal of an impulse radar cannot be approximated
by a sinusoid. Removal of the assumption of sinusoidal excitation requires us to
reconsider antenna reciprocity starting from basic EM principles.
Consider the widely known theorem that the far-zone pattern of gain vs. angle
for a. given antenna is the same for reception as for transmission of a sinusoidal
waveform. Does this theorem hold for an arbitrary transient signal? Study of this
question is complicated by the fact (Appendix B) that the waveform of the far-zone
radiation varies with angle. Thus, the pattern of an impulse-radar antenna cannot
be described as a 3calar function of angle as in the simple case of sinusoidal excita-
tion. However, this complication does not imply that transmit.receive reciprocity
fails for transient excitation.

Initial study of the problem indicates that transmit-receive reciprocity does hold
in several examples of array patterns. Further work is needed to determine condi-
tions 1hr transmit-receive reciprocity in general for antenna elements and arrays.

7,'1
I
I
i
-- Appendix B

Impulse-Radar Signals
I
5 This appendix presents a summary of work that has been done in defining reasonable
characteristics for the transmitted signal for an impulse radar and in determining
the return signal when an array antenna is used with a matched-filter receiver. The
first step is to find solutions for the current in a circuit that might conceivably
3 be used to generate short pulses in the transmitting antermna of an impulse radar.
Considered next is the far-zone electric field when an array of such antennas is
U used. Finally, the return signal from an idealized point target is determined and
the response of a matched filter to this return signal is evaluated.

B.i Preliminary Analysis

A paper by Harmuth [Harmuth 1983b] outlines some conditions on the antenna
current that are necessary in a physically realizable situation. Specifically, the

author points out that the far-zone radiated electric field intensity is proportional
to the time-derivative of the e&ntenna current and that there is a time-dependent
radiation resistance for the Hertzian dipole.
Harmuth also notes that, because the energy radiated must be equal to the en-
3 ergy delivered to the radiation resistance, there are some constraints on the antenna
current. In pa•rticular, if the current is required to be zero before zero time, then the

7
U 79
initial value of the current and its time-derivative must also be zero immediately
after zero time. These initial conditions ( .n be used in solving for a physically
realizable antenna current.
In order to carry out a preliminary analysis, it was convenient to assume the
existence of a model to represent a simple circuit that might reasonably be expected
to satisfy all of the initial conditions. Furthermore, it was assumed that this circuit
was excited by means of a voltage source and a awitch that is closed for some short
period of time. It is recognized that such a simple model cannot truly represent the
actual physical situation, which must necessarily involve distributed parameters.
The antenna current that resulted from the assumed combination of circuit
elements can then be determined from circuit theory, taking into account the time
variation of radiation resistance.
Since the far-zone electric field is proportional to the time-derivative of the cur-
rent, this quantity can then be computed. An item of particular interest is the
energy spectrum of the radiated EM field, which is obtained from the square of the
magnitude of the Fourier transform of the time-derivative of the antenna current.
Since the analytical expression for the Fourier transform of the derivative of the cur-
rent turns out to be quite complicated, and since computed values of the derivative
are available, the Fourier transform can be obtained by computing a Fast Fourier
Transform (FFT) of samples of the current derivative. The squared magnitude of
this transfoirm, normalized to its maximum value, can then be obtained.

B.2 An Alternate Analysis

The preliminary analysis outlined in the previous section is complicated by the need
to solve a nonlinear differential equation of second order and third degree. The re-
sulting solution for the antenna current is quite complicated and it is not feasible
to do much analytical work with that waveform. Further work is required in nu-

80
I
I merical computation based on the preliminary analysis to obtain results applicable
to studies on impulse radar.
In order to proceed now with an initial study of transmitted and received signals
in impulse radar, a more tractable waveform for analysis is required. The basic form
of the antenna current, expressed in terms of normalized time x = t/T, is selected
as follows.
i(z) = IoTxc-,T sin#/Tx,
x > 0
(B.1)

This current is displayed in Figure B.1 for aT = 2 and PiT = 2. The time-derivative
of this current, expressed as a function of x, is
di =o[(1 - aTx)e"•Tzsin fTx + (Txe-Tx cosfi7'x], x> 0 (B.2)
dt

and this waveform, which is proportional to the far-zone electric field, is displayed
in Figure B.2. The corresponding energy-density spectrum for the far-zone electric
field is shown in Figure B.3.
It is of interest to look at some numerical values that might represent a practical
situation. Suppose, for example, that the 10 to 90 percent rise time of the antenna
current pulse is 10 ps. From Figure B.1 it may be noted that this rise time occupies

about 0.47T. Hence, T = 25 ps. From Figure B.2 the maximum of the energy
spectrum occurs at about fT = 0.4, which corresponds to an actual frequency of 16

GHz. The corresponding half-energy points of the spectrum are at about 9.2 GHz

and 31.2 GHz, which yields a bandwidth of 22 GHz.

81
1.2 - T- I

1.0- aT PT 2.0

S0.8

06
0.6

3 0.4
>0 .
0

-0.2
0 0.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0
Relative Time (tIT)

Figure B.I: Typical current pulse in one array element

3.0
-aT = 2.0
OT S

0 "2.0
,> 1.5
'i 1.0
a 015

-10-

a -1.5

.9-
-2.0 .. .IItI , _ _
0 0.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0
Relative Time (t/T)
Figure B.2: Time-derivative of current pulse in each array element

82
U
UU 0' OT 2.
V

E -
* [12

* ~ -16.

4 S

I 28
-32 L.. a
i102 10-1 1 10
Frequency-Time Product (ft)

I Figure 13.3: Erzergy spectrum of radiated signal from each eleMueIL of the array

U
B.3 Array . for Impulse Radar

A single current " .±oes

Lo not provide signficant directivity irn transmitttig•g

3 antenna for an impulse radar. Ljwever, the use of array ,itennas does provide a
method for achieving reasonable directivities, although there are some significani.

I differences in the performance of such arrays with UWB signais. Perhaps the most
significant difference is that the waveform of the electric field intensity is a function

U of the angle away from the antenna axis. This dependence on angle is examined in
some detail here because it has considerable impact on the optimum processing of

I the received signaJ.

Consider a lhaear array of current elements spaced d meters apart as shown in
Figure BA. Assume that these are rertical elements so that the angle 0 is the

g azimuth angle measured from the normal to the line of the array. The analysis can

83
easily be extended to include both elevation and azimuth, but this is not done here
in the interest of brevity.

0 0 0 0 0 0 0 0 0
1 2 3 N

Figure B.4: A linear array of N current elements

When this array is used as a transmitting antenna, the far-zone electric field
intensity is simply the sum of the contributions from each current element. The
electric field due to the current in the jth element is E3 (t).

E1 (t) •= 47rer
Zo__ di(t dt
- r/c) (B.3)

Z0 = 377 f]
a = length of each array, element, in meters
r = distance from the array, in meters

c = 3 x 108 meters/second

The resulting electric field at any angle 0 is simply

E(t,O)=ZE
j=1 I
- (i- 1)-asin]
0
(B.4)

or, in terms of the normalized time x, we have E(x, 0).

E(,q) =- Ej [- (i - 1)4sid (B.5)

The directivity pattern of an antenna is usually defined in terms of the ratio of

the average power density radiated in a given direction to the maximum value of
the average power density. In the case of impulse radar, it is more appropriate to

84
U define directivity in terms of energy density rather than average power density. The
energy density in a given direction ¢ is defined as follows.

6 (€) f= [E(t,¢0)12 dt (B.6)

This function has been evaluated numerically at each valL.. of j and divided by the
3 value of E
(0). The half-energy beamwidth is the difference in O-va'ues at which the
energy density drops to one-half of its value at q0 = 0.
3 The directivity of the array depends upon both the spacing of the aitenna
elements and the length of the pulse. The directivity of a ten-element array for tt2 *

particular case of d/cT = 1.0 is illustrated in Figure B.5. The same current pulse
is assumed in each array element. It may be noted that the half-energy beamwidth
is about 10'.

.2-2 Normalized Element Spacing (d/cT) = 1.0

m"-4-• Half-Energy Beamwidth =10 degrees -

aT1OT= 2.0

C- 8
i 10-
-io
IL -12
I -

-18-
-20 0 5 10 15 20- 25 30 35 40
A
45 50 55 60
Angle Off Array Boresight (degrees)

Figure B.5: Directivity of ten-element array

U 85
I
A significant characteristic of the energy patterns displayed in these figures is
the absence of any sidelobe structure. A linear array with uniform excitation would
have a pronounced sidelobe structure with the largest sidelobe being down by about
13 dB. Although this array also has unifoim excitation, these sidelobes are entirely

missing.
It is of interest to compare the short-pulse directivity of the Lrray with the di-
rectivity that would be achieved with sinusoidal waves. For the previously discussed
value of T = 25 ps, a value of d/cT = 1 leads to a value of d = 0.0075 m. A ten-
element array therefore has a total length of 9 x .0075 = .0675 m. At a frequency of
16 GHz (i.e., the frequency at which energy spectrum is a maximum) the half-power
beamwidth for a un' form linear array is approximately 150.
A _.01875

e = - 075 = 0.2778 rad = 15.90

-. (B.7)
(N - 1)d .0675
Since the half-energy bearr'width is 100 for this case, it seems that the array has
somewhat greater directivity in the short-pulse case. However, if the comparison
is made at the upper half-energy point of the spectrum (31.2 GHz), the sinusoidal
wave half-power beamwidth would be 8.20; while at the lower half-energy point (9.2
GHz), the sinusoidal half-power beamwidth would be 27.7'.

B.4 Time Variation of the Far-Zone Electric Field

Another item of considerable interest in connection with array antennas for impulse
radar is the tims variation of the far-zone electric field intensity as a function of
azimuth angle and the corresponding energy spectra.
The time variation and the energy spectra for two different azimuth angles are
showr, in Figures B.6 through B.9. At an azimuth angle of 0* there is no broadening
of the pulse and. no change in the energy spectrum (except for magnitude) from that
for the single element. At an azimuth angle of 5V, the half-energy point for this case,
there is a noticeable broadening of the pulse and a significant change in the energy

86
I
spectrum. As the azimuth angle gets larger, the pulse becomes much broader and S
the energy spectrum reveals many more minima and maxima.
Three comments are relevant in connection with the electric field intensity at
angles away from the beam axis. First, the total signal energy does decrease with
I angle and, thus, a well-defined beam is possible. Secondly, the time variation of the
electric field intensity becomes broader with angle. Hence, a receiver matched to an
on-axis signal will not be matched to an off-axis signal. Finally, the energy spectrum
is no longer smooth, but reveals many minima and maxima. The locations of these
minima and maxima may be useful in deriving angular information.

0 T= 2.0
OT3
1 Azimuth = 5 degrees
-C Element Spacing = 1.0

IW
0
.1 ! 1N

I iRelative Time (t/T)

i Figure B.6: Far-zone electric field from a ten-element array

I IC
I
87
U
20 1 1 1 lo l I toEli

10-
E
• o

WA
20-
'2 0"1101
cc aT=A3T=2.0
-30 Azimuth Angle =5 degrees
Element Spacing (d/cT) =1.0

10210"1 1 10
Frequency-Time Product (ft)

Figure B.7: Energy spectrum from a ten-element array

5 i I i • i

. 4 a =T2.0
OT
Azimuth Angle = 15 degrees
S3 Normalized Element Spacing (d/cT) = 1.0
C- 2

U.

w- 1
-21--

S-34

0 1 2 3 4 5 6 7 8
Relative Time (tVT)
Figure B.8: Far-zone electric field from a ten-element array

88
20 - 1 1%fi ll . 1 1 9 1 1 1 if lla

IE -10

Ca
I -o -

IT =OT =Angle
2.0 =15 degrees
-30- Azimuth
Normalized Element Spacing = 1.0

I0- 0 2 10"1 1 10
Frequency-Time Product (ft)
Figure B.9: Energy spectrum from a ten-element array.

3 B.5 The Received Signal

3 It is assumed in this discussion that the received signal that is to be detected results

from the far-zone electric field impinging on a single perfectly conducting scattering
U point so that the reflected signal is just a replica of the incident signal. It should be
recognized that this is an idealized first step to considering a model that assumes
3 that any complex target is simply a collection of a number of such point scatterers.
Such a target model is an oversimplification of the real situation and does not reveal
3 what may be one of the greatest advantages of impulse radar; that is, the ability to
detect and/or discriminate targets whose reflectivity is a function of frequency.
I Before considering the problem of detecting the return signal, some consideration
should be given to the shape of the return signal as a function of azimuth angle.
I The previous section noted the change in the shape of the electric field with azimuth
i angle because of the spread in time delays from the various elements of the array

U 89
antenna. If the same array antenna is used for reception, there is an additional
spread that occurs because each component of the return signal arrives at the various
antenna elements at different times. This additional spread produces the same sort
of pulse broadening and spectral modification that was observed previously.

B.6 The Matched-Filter Receiver

It is well known that the probability of detecting deterministic signals in the presence
of white noise is maximized by the use of the matched filter. A matched filter is one
that maximizes the signal-to-noise ratio at a particular instant of time, say to. The
impulse response of such a filter is simply the time reflection of the signal about the
time to, with any portion in negative time removed to preserve causality.
For the type of signal considered here, the electric field intensity at the receiving
antenna has the same time variation as the far-zone electric field at the target when
the target is on the beam axis of the array antenna. It has been shown [Harmuth
1983c] that when the load impedance connected to the receiving antenna is resistive,
the voltage developed across this load has the same time variation as the incident
electric field intensity. Thus, the impulse response of the matched filter designed to
detect this signal must be the time reflection of the time function shown in Figure
B.2. Although it is not possible to realize this impulse response in actual hardware,
it is instructive to use it to establish some upper bounds on performance.
The output of the matched filter is obtained by convolving the input signal with
the impulse response of the filter. In the present case this is most easily done by
employing the FFT. Accordingly, a 1024-point FFT is used to obtain the desired
response. It is noted above that the time variation of the receiving antenna voltage
changes if the target is off the array axis. Accordingly, it is appropriate to see what
happens to the output of a filter matched to the on-axis signal when an off-axis
return is received. Figure B.10 displays the matched-filter response to an on-axis

90
+I
U signal from a ten-element array for which d/cT = 1 and for a signal in which
3 aT = /3T = 2. Figure B.11 shows the same situation when the target is 50 away
from the beam axis of the transmitting array. This is the one-way half-energy point
3 for this array antenna. It may be noted that the output signal amplitude has been
reduced by a factor of 0.64 and that the location of the peak response has been
3I shifted to tOT = 4.4.
Figure B.12 illustrates the matched-filter output when the target is 150 away
from the beam axis of the array. In this case the response now has two peaks and
the amplitude is only about one-tenth of what it was for the on-axis target. The
3 significance of two peaks is considered further when the subject of range resolution

I is discussed.

2500

2000- aT = 3T = 2.0
"Azimuth= 0 degrees
U1500- Element Spacing = 1.0
3 0
,1000-
I 500
g 0o-

-500-
I ~ ~~-10001jIl l
10 -2 3 4 5 6 7 8
I ~Relative Time (t/T)

Figure B.10: Matched-filter output from a ten-element array for 0*

I
| 91
2000 1 1

1500- aT = PT = 2.0
Azimuth = 5 degrees
CL Element Spacing =1.0

50O

i0

-5000

-1000
0 1 2 3 4 5 6 7 8
Relative Time (t/T)

Figure B.11: Matched-filter output from a ten-element array for 4= 51

250
200 aT = PT = 2.0
150- Azimuth = 15 degrees
S00 Element Spacing = 1.0
0-
50-
0
-So-
-11• --
-150 -
-200 -

-20 11 1 2 3 4 5 6 7 8 9 10 11 12
Relative Time (t/T)
Figure B.12: Matched-filter output from a ten-element array for 0 =15

92
I
U B.7 Range Resolution of Impulse Radar
One of the most important attributes that impulse radar is alleged to pcwsess is
an extremely good range resolution. hi order to examine thic claim quantitatively,
3
!the output of the matched filter was computed for the case of two closely spaced
targets on the beam axis of the array. If the range separation of the two targets is
I denoted by AR, then the normalized range separation, indicated on the following
figures simply as target spacing, is AR/cT. Figure B.13 displays the matched-filter
I output when the normalized target spacing is 1.0. For this target spacing, there is
a pronounced double peak as shown in Figure B.13.
The significance of this double peak is that a large single target off the axis may
be indistinguishable from two targets on the axis. This has serious consequences
for any algorithms intended to accomplish target identification.

3 1500 -i

aT =PT = 2.0
1000 Azimuth = 0 degrees
Element Spacing = 1.0
3 0 500-
Target Spacing =1.0

I .00
5 -
.1
00 1- I I I A
0 1 2 3 4 5 6 7 8 9 10 11 12
Relative Time (tiT)
3 Figure B.13: Output when two targets of equal size axe present

U 93
I
I'
It is of interest to compare the observed range resolution for impulse radar with
the classical range resolution that is related to signal bandwidth. If a normalized
target spacing of 1.0 is taken to be a measure of the range resolution of impulse
radar, and the value of T is taken to be 25 ps as before, then the physical target
spacing, AR, becomes 0.0075 m. If the bandwidth of impulse radar is taken to be
the half-energy bandwidth of 22 GHz, the classical result for range resolution yields
a value of 0.0068 m. These results are certainly comparable.

B.8 Detection with a Nonmatched Filter

Since the impulse response of the true matched filter is not physically realizable
(and even difficult to approximate), it is of interest to investigate the response of
a realizable filter to this type of signal. Accordingly, a very simple filter has been
considered and the parameters adjusted by trial and error to achieve the largest
output signal-to-noise ratio in the presence of white noise. It was shown that the
output signal-to-noise ratio of the matched-filter receiver is 7.4 dB better than the
output of the nonmatched filter. It is likely that there are other physically realizable
filters that would produce results much closer to that of the true matched filter than
the simple filter considered here.

B.9 Approximate Maximum Range

This section provides a discussion of the radar equation for impulse tadar and a
calculation of the maximum range for a specific type of signal. Although some
appraximations are involved, the results are indicative of the range performance of
impulse radar.

94
U
3 B.9.1 The Range Equation for Impulse Radar

Because of the very large bandwidth of the impulse-radar signal, it is necessary

to modify the conventional radar equation, which is essentially a result based on

3 sinusoidal waves. In particular, any use of frequency or wavelength must be elim-

inated from the radar equation because these concepts have no validity for the
3 non-sinusoidal waveforms employed in impulse radar. Fortunately, it is quite pos-
sible to do this in a manner that is both consistent with physical principles and is
uniquely dependent upon the time-domain representation of the transmitted signal.
Although there are many forms for the radar equation, the form that is most
3 convenient is one that relates the received signal energy to the transmitted signal
energy. ETGTARa(

ER - (47r) 2RiL (B.8)

where
ER = received signal energy in one pulse
ET = transmitted signal energy in one pulse
GT = transmitting antenna gain
AR = receiving antenna effective area in m2
a = target radar cross-section in m 2
R = range to the target in m
L = all losses due to the system and to the atmosphere
In the usual application of this form of the radar equation the antenna gain is
3 considered to be a function of the frequency. However, when an array antenna is
employed, the antenna gain can often be approximately expressed as a function only
3 of the number of elements in the array. Although an exact analysis of the gain of an
array is quite involved for non-sinusoidal waveforms, it has been carried far enough
3 to verify that a simple approximation is reasonable. In particular, if the elements
are assumed to be sufficiently separated and to have the same current, then the
antenna gain is simply the number of elements in the array. This assumption is
employed here. It is shown in B.3 that there are no sidelobes with equal excitation
I
95
I
In
of the elements, as there would be in the sinusoidal case, so there is no real reason
to consider unequal element currents. Tht .a, for an array with N elements, the gain
is well-approximated by N.
Although the analysis presented here utilizes the concept of an array, the end
result is expressed in terms of the effective area of the antenna rather than the
number of elements in the array. In this final form the result is equally applicable
to aperture antennas with a single feed point.
It is also assumed here the aperture of the array antenna is rectangular with
N& elements in elevation and NA elements in azimuth. Hence, the total number of
elements is N = NE . NA. If the spacing between elements is d in both azimuth and
elevation, the area of the aperture is simply

AT = dNE . dNA = d2 N (B.9)

It is assumed here that the transmitting and receiving antennas have the same
number of elements and the same element spacing, regardless of whether they are
the same structure. Clearly, only a trivial change is required to accommodate trans-
mitting and receiving antennas with different numbers of elements. The effective
area of the antenna is not necessarily the same as the physical area of the antenna,
but the difference can be accounted for in the loss factor L.
The antenna element spacing d depends upon the duration of the antenna current
pulse. It has been shown by computation of array antenna energy patterns that a
reasonable spacing of the array elements is

d = cT (B.10)

where
c = 3x106
T = current pulse rise time
A spacing smaller than this yields a broader antenna pattern and a spacing
greater than this reduces the number of elements in the array and, hence, the gain

96
SI of the antenna. Furthermore, this is about the smallest element spacing for which
the approximation for the gain is well justified. If this criterion is employed, the
number of elements in the array can be expressed in terms of the pulse rise time
"andthe physical area of the antenna as
I Ar
N = (cT) 2 (B.11)

Note that this result implies that the antenna gain, which is equal to N, is
inversely proportional to the square of the current pulse rise time. This is analogous
to the sinusoidal result in which the antenna gain is inversely proportional to the
square of the wavelength. Furthermore, this result is identical to one described by
Harmuth (1989) except for a constant that is close to unity.
The final parameter that must be specified is the transmitted pulse energy ET.
If losses in the antenna are included in the loss factor L, the energy radiated by
I each element of the antenna is given by [Harmuth 1983b]
El=ýO--S .o di d
( B .1 2)
2f
E 11 6 c"wh ( ad)
er e
Zo = 377 ohms
3a = equivalent Hertzian dipole length
i(t) = antenna current in one element, or the total current
in the case of an aperture with a single feed point
The total energy radiated is just ET = NE I for the array, or is E 1 for an aperture
3 antenna. This energy can be evaluated for any specific form of antenna current.
The specific form of antenna current that is used for purposes of evaluation is

i,(t) = it2e20/ t >0 (B.13)

3 in which T is the time at which the current pulse is a maximum. It is convenient

to express 10 in terms of the maximum current because the maximum current is
limited by the pulse generation equipment. Thus,

JO ( 'Imax (13.14)

I 97
Upon combining the above results, and carrying out the integration required by
(B.12), the energy radiated by each element of the array, or by the whole aperture
in the single feed case, becomes

HPmax (B.15)
T
where
2 4
Zoa C
2
1927rc
For a = 0.1m, H has a numerical value of 3.7916 x 10-Is.
When all of the above results are incorporated into the radar equation (B.8), it
becomes
- ETaTAR
(41rcT) 2 R4L (B.16)

Er = N-[I'
T

with N = 1 in the single feed aperture case.

B.9.2 Maximum Range of Target Detection

The range at which a target can be detected depends not only on the received signal
energy but also upon the receiver noise. When a matched filter receiver is employed,
the probability of detection for a given probability of false alarm depends only on
the ratio ER/No, where No is the one-sided spectral density of the receiver noise

and is given by
No = kToF

where
F = system noise figure
kTo = 4 x 10-21 W-3
Equation (B.16) can be solved for the largest range at which ER/No equals the
desired value. The result is

98
__ , ETATAn•c/
11miicTx (B.17)k~j~)-

This maximum ranuige has been computed ws a function of the total radiated
energy for three different values of pulse rise time, T, with the following parameters:

Effective Antenna Area, AR = AT = 1 m2

Target, CroWs-Section, a = 0.1 m2
System Noise Figure, F = 3 dB
System Losses, L - 4 dB
Detectability Factor, ER/No = 13 dB

The specified detectability factor yields a single observation probability of de-

tection of OM.) and a probability of false alarm of 1 x 10-. The resulting maximum
ranges are displayed in Figure B.14.
It is interesting to note in Figure B.14 that the shorter pulse rise times result in

larger maximum ranges for a given total energy radiated. The reason, of course, is
that a given size array will contain more antenna elements when T is smaller and,
hence, the antenna gain is larger. Furthermore, for a given total radiated energy,
the required energy per element will be smaller with shorter pulse rise times. It
should be emphasized again that the maximum range given by equation (B.17) and
by Figure B.14 is applicable for ei`her an array antenna or a single-feed aperture.
Although the preceding analysis, and Figure B.14, suggest that single-pulse
detection is being considered, all results apply equally well to a sequence of pulses
(coded or uncoded) if the matched filter is matched to the entire sequence. Hence,
the required energy per pulse is reduced from the value specified by Figure B.14 by
the number of pulses in the sequence.

99
Single-Pulse Detection Range

70-

E 60-T 10p

C
.40-
C
m 30-

~20

cc 10

0 10 20 30 40 50 60 70 80 90 1;00
Total Pulse Energy Uoulos)

Figure B.14: Single-pulse detection range with a matched-filter receiver

100
I

I Appendix C

Propagation of UWB Waves

The literature of physics and electrical engineering on EM propagation in looay

media concentrates almost entirely on sinusoidal waves. Notable exceptions are the
works by James R. Wait [Wait 1970] and by Henning Harmuth [Harmuth 1986]. The

purpose of this note is to review one of Harmuth's results, after a brief summary
of the history of relevant theory and applications. Study of this area is relevant to
applications requiring large relative bandwidths, such as UWB radar.

C.A Evolution of EM Theory and Applications

Applications of EM theory and technology to communications and radar have
evolved at a rapid rate. Feasible upper limits on carrier frequencies have increased
rapidly as the technology of radio engineering has evolved. Systems operating at

100 GHz and even higher frequencies are now common. Requirements on signal
bandwidth have also increased, but at a slower rate. Typical radio communication
and radar systems use relative bandwidth R, extending from f, to f., of only 1%
o r les s . R_ <
R __-_ <1
f. + fA -

Such narrow-band applications may be analyzed by assuming sinusoidal excita-

tion, as is standard in textbooks of electrical engineering and classical EM theory.

101
Advanced military communication and radar systems require steadily wider
bands to improve anti-jam performance and, in the case of radar, to increase resolu-
tion and accuracy. An example is the use of pseudo-noise sequences of "chips" with
duration of 1 ns or less. Another example is an impulse radar using a monocycle
or similar waveform with a rise time of 100 ps or less. Such UWB signals involve
bandwidths up to three decades and relative bandwidths greater than 10%.

C.2 Propagation of Transient Signals

Any signal requires a beginning. It also requires an end, unless only one instance
of the signal is to be transmitted through the channel. Thus, the analysis of signal
propagation must be based on the propagation of transient waveforms. This problem
is treated adequately in the classical EM literature for the case of propagation
through lossless media. However, the study of EM transients propagating in lossy
media has been neglected. A notable exception is the recent work of Harmuth,
where the importance of time-domain EM techniques in analysis and measurement
is shown. Other authors have recently written on the importance of time-domain
electromagnetics, particularly C. L. Bennett and Gerald Ross.
Why have analyses of EM transients propagating in lossy media appeared so late
in the development and applications of EM theory? Perhaps the following reasons
are part of the answer.

"* Transients in linear media may be analyzed in all cases by transform methods,
and this approach is often convenient.

"*The atmosphere of Earth may, in many applications to date, be considered to

be a losles medium.

"* Sinusoids may be used, with tolerable errors, as excitation for analyses of
narrow-band systems.

102
1 Accurate
.a analysis of a transient propagating in a lossy medium is difficult.

3 . * The need for accurate analysis of EM transients propagating in lossy media

emerged late in the evolution of EM theory and applications; a good example
3 is the recent interest in possible future applications of UWB radar.

I .3 The Harmuth Approach

Papers and correspondence by Harmuth in the IEEE Transactions on Electromag-
netic Compatibility beginning in 1986 have precipitated much discussion and con-
3 troversy published in the same journal over the past several years. Titling these
papers "Corrections to Maxwell's Equations" may have triggered some of the con-
I troversy. What Harmuth actually does is to use a certain mathematical technique
to obtain solutions of Maxwell's equations that have not been found before, but his
procedure cannot be regarded as a correction to Maxwell, at least not yet.
The mathematical approach used by Harmuth to solve for transient waves in
lossy media is to symmetrize Maxwell's equations by the introduction of terms
representing isolated magnetic poles. After solutions are obtained in closed analytic
form, the terms related to isolated magnetic poles are set to zero before final results
3I are stated. This approach is used by other authors of texts on EM theory.
A book entitled Propagationof Non-sinusoidal Electromagnetic Waves by Har-
muth was published by Academic Press in 1986. In Chapter 5, he reviews the
classical treatment of EM propagation and introduces his interesting revision of
3 these ideas following the result of one of his analyses. Later in this note, that
particular result is checked without using anything associated with isolated poles.
3 One of the most interesting ideas suggested by Harmuth in his Chapter 5 is that
the concept of "signal velocity" must relate to noise level at the receiver. This sug-
I gestion is certainly a surprising variation on the usual textbook treatment of phase
velocity, group velocity, and even the signal velocity defined by Sommerfeld [Som-

103

I
merfeld 1952]. On the other hand, communication system engineers have accepted
Shannon's ideas on channel capacity, the maximum possible rate of information flow
through a communication system, which clearly depend on noise level. If channel
capacity depends on noise, perhaps we ought not be surprised to learn that signal
velocity does also.

C.4 Transient Propagation in Lossy Media

Suppose that a plane TEM wave is launched at t = 0 in the plane y = 0 of an
isotropic EM medium characterized by permeability a, permittivity c, and nonzero
conductivity a. Harmuth considers this problem in Chapter 2 of his book and uses
the result in his general discussion of EM-wave propagation in Chapter 5. We will
consider the example described in his Section 2.1, where the boundary condition is
that the electric field intensity E(y, t) is a step function at all points in the plane
where y = 0.
E(0, t)= Eo o<t
10 t<0

The PDE to be solved is the following equation, which is readily derived from
Maxwell's equations for any plane TEM wave propagating in the y direction.
0 2E a2 E aE

Harmuth obtains the following solution to the given PDE and boundary condition.

E(y, t) = Eo -{1Iet -d/3Afsin

[jU(t) #y d# sin fly lI

U(t) = coshBt+(a/B)sinhBt V(t) = cosDt + (a/D)sinDt

Integrations are over 3 = 2-rK where K is the wave number; K = I/A - f/c.

104
* 1
0.9 yY
0.87
0.7-y=2y
0.6

U10.4
0.3
0.2
0.1

0 (a) Electric Field Intensity

I 8
7-
66 -y
5

0 t
0 (b) Magnetic Field Intensity

I ~Figure CA1: Propagation of a transient wave in a lossy medium

105
The constants are defined as follows.
c = WE a/VrA
=ao./2c A = (a/2)Z
B = =vr2caT Z =
Typical plots of E(y, t) are shown in Figure C.1 for y = 0, y,, 2y,, and 3y,, where
the steps occur, respectively, at t = 0, y,/c, 2y,/c, and 3y,/c. The asymptote
E(y, oo) is Eo for all y. Evaluation of 8 E(y, t)/dt at t = y/c shows that the slope
is vertical.
Harmuth's results are certainly nothing like anything we have seen before! Per-
haps other engineers have had a similar reaction, which might explain in part the
controversy mentioned earlier in this note. However, it is easy to show that the
solution above approaches the boundary condition as y goes toward zero. It is a
more difficult exercise, but still straightforward, to show that Harmuth's solution
satisfies Maxwell'. equations and reduces to the classical result when a = 0. These
details are carried through in the next section of this note.

C.5 Calculations

Consider the solution in Section C.4 for a = 0. We see that a = 0 = A.

E(y,t)/Eo = 1-- 2 f(csct) Bi dfl

= I--ir1 h oosinfl(y- ct) d -- KJ

1j f sinC(y + ct) dfl

Causality limits attention to the region where 0 < t. The following result is obtained
from a table of definite integrals.

fJ sin mx d0
J r/2 m>0
=
:- 0 mn=O

-7r/2 m <0
Consider any point y = yt > 0 and t > 0. We find the result graphed in
Figure C-2 by applying the tabulated integral to three cases: ct < yl, ct = yl, and

106
I
I
Yi < ct. The result is a step wave propagating in the positive y direction without
attenuation.
Considei' any point yj < 0 and t > 0 for the three cases: -- et < yl,ct = yl,
and y, < -ct. The result is a step wave propagating in the negative y direction
without attenuation.
SI We see that the classical results are obtained.
The required partial derivatives of Harmuth's solution are found on page 104.
I Careful inspection of the following calculations shows that one of the integrals
does not converge. This difficulty is associated with the step response and is avoided
by replacing sin/ly with' the imaginary part of an exponential and rotating the path
of integration slightly in the complex plane. This assistance of Ross Graves and
Patrick Hagen in developing the following calculations is gratefully acknowledged.

1.0
I
-50 0.5 * YY1
I
0.0
0 Y0/c t

Figure C.2: Propagation of a tranmient wave in a lossless medium

I
I
I
I
l 107

I.
I. a2 E(y~t) 2Ca A t)-snldf+f' tjfsip dj

1 aE(y, t) _2ot e A , t /~)i~

in
#g d
Eo at 7r ea* jj Uf)~ 1+AVt) d
2 CotA BUsj'nBd#

+f' a cotsDt - DsinDt InfydI3}

I a2E(y, t) _2a
2 t fA in flysin/3g 2
Eo 9t2 7r I o 0 IA '/ f
2o!eag {A BUt)infl d

+ aco~tDsinDt]!l-l
f00 1(2)
a f t) 'i fly d
2 fnA 3 I
p2()rn

+f~[-D'cos - aDsinDt] l1}

108
i
i Substitution of the partial derivatives of E(y, t) into the LHS of the PDE yields
the following summation after division by (2Eob/r)e2-.

iA 2 (oshBt + sinhBt) L#•d'j'd3

II f (coso t + isinDt) !ý-d/]

+.Ica 2 [fA (coh Bt + a sinhBt Sin

3 Y)d fl]
+ (cs Dt+ '2sin/

-2/Aca [fjA (B coah.Bt + asinhBt) nfldl

i :
+/~[( B_,., .. o, _,
l
(a(cos Dt+-Dsin DO i)/3d,3]

I +,. [ (Bosh B+ asinhBt) d,3

109
i
II
Coefficients of the various integrand terms are tabulated below.

Term Coefficient
cosh Bt /32 + jsca 2 -21scaB +jscB2 -Aaa +AsaB
2 -iAa +AoO'B
(a/B) sinh Bt #32 + Aca' -2AcaB +AeB
cos Dt 3' + Aea 2 -2Ascax -. &eD' -Aaa +AWaa
(a/D) sin Dt /32 + Afa 2 +2AcD 2 -AED 2 -- Aara -gwD 2 /a

2 + -. B2
-c-(a
2-)o + .61A (a - 2-a)
Aa+
= /+2 Acs (2a2 P3 C') - jAa(2ca) + 0

/32 -Ae - AcD

-
2 =82 -JAC [a 2 + (.82C2 _a2)] = 62 (1 _ IUEC2) = 0

.82 + A, (02 + DZ) -2 (2a)a - s(2ca) (o2CI - a 1) /a

- 2 + Ac (#2 C2) - 2Ac -_2pe (#2Co - a2)

= /3 2c 2
_EP/= 0

All of the integrand coefficients are zero and, thus, Harmuth's result satisfies
the Maxwell PDE. Note that nothing in these calculations has anything to do with
isolated magnetic poles. Thus, this remarkably new and different result is proven,
regardle-s of the method used in its derivation.

C.6 Conclusion
Further research is required on the use of these and similar results in practical
applications to take account of spherical wave fronts, more practical excitations,
and variations of A, a, and c. But it is clear that Harmuth has led us to think more
deeply than we had thought before.

110
I
I

I Appendix D

Los Alamos National Laboratory

I
The Los Alamos National Laboratory has the primary mission to apply scientific and
engineering capabilities to assure the national deterrent through nuclear weapons
technology. To be responsive to national needs and to succeed in this primary
mission, we:

strongly support basic research in selected disciplines to help maintain

an outstanding science and technology base;

conduct other applied programs, primarily in defense and energy, that

U complement the primary mission; and

encourage new ideas that fit the long-term vision of a multidisciplinary

laboratory solving problems of national importance.

U At present, slightly more than half of the activities at Los Alamos are dedicated
to developing nuclear weapons technology, as we have a broad research-to-retirement
responsibility for nuclear weapons. Such broad responsibility not only ensures the
viability of the stockpile but also fosters an atmosphere of scientific innovation and
creativity. The research atmosphere is particularly conducive to multidisciplinary
collaboration, which crosses the boundaries of traditional physics, chemistry, math-
ematics, materials science, and engineering.
I
IIi
In support of our primary mission, we have expanded our research programs
and our capabilities. These expanded capabilities allow us to contribute to national
strength in areas such as energy, conventional weapons, strategic defense, health and
environment, computational science, and economic competitiveness. Nonnuclear
defense programs constitute about a fourth of the Laboratory's efforts. They are an
important element in developing technologies for national security. The remaining
fourth of our work focuses on basic and applied research not directly related to
defense projects. Our experience has demonstrated many times that en outstanding
basic research program and a challenging set of nondefense research efforts are key
ingredients for excellence. In this mix of defense and other R&D projects, each area
benefits from the other.
We currently have around 7700 employees of whom half are degreed technical
staff. Half of the technical staff have doctorates with the other half divided ap-
proximately equally between B.S. and M.S. degrees. Our staff is 33% engineers,
29% physicists, 12% chemists, 9% mathematicians and computer scientists, with
the remainder from other disciplines. The budget of the Laboratory for FY 1989 is
$917M. The Laboratory occupies 43 square miles with 32 separate technical areas
organized into 19 technical and 11 administrative divisions.
The Laboratory is operated by the University of California for the U.S. De-
partment of Energy. It is located at an altitude of 7,400 feet in the mountains of
northern New Mexico in a setting of great natural beauty and a pollution-free envi-
ronment. A map of the central areas of the Laboratory and surroundings appears
in Figure D.1.

112
LOS ALAMOS COUNTY

LoIlao
IA-
N0

I' White Roc

Bandelier
Monument

1 Figure DJ: Los Alamos National Laboratory

113
D.1 Capabilities Relevant to Impulse Radar

D.1.1 Projects and Facilities

Several groups at the Laboratory are equipped with electronic instrumentation of
the types required for R&D in impulse radar, including spectrum analyzers, time-
domain reflectometers, counters, pulse generators, synthesizers, network analyz-
ers, sampling oscilloscopes, etc., spanning the frequency spectrum from dc to the
millimeter-wave band. These facilities include standard instruments available from
leading manufacturers and specially designed instruments to meet the needs of the
wide variety of projects undertaken at the Laboratory. These facilities include spe-
cialized test ranges located in canyon areas well secluded from inhabited areas of
the Laboratory and surroundings.
Current activities that are closely related to impulse radar, particularly with
regard to required facilities and personnel include the following projects.
"* HPM program for DoD
"*Microwave system developments for LAMPF and other accelerators
"* Wideband instrumentation for measurements of nuclear explosions
"*Fast-transient measurements of EM pulse effects
"* Doppler radar for measurement of wind shear

The project most closely related to impulse radar is the HPM program for DoD
directed by Robert Hoeberling of Accelerator Technology Division. Several high-
power pulse sources have been developed for the specialized needs of the HPM
program. Some of these sources are listed in Table D.1.

114
I
I Table D.1. HPM Pulse Sources at Los Alamos

I
Power Pulse
3 Frequency Density Duration
0.080 1.0 0.2-100,000
0.425 1.0 0.1- 1000
0.805 2.5 0.3- 1000
1.300 25.0 0.1- 100
2.340 1.0 0.1- 2000
8.200 1.0 0.015
17 & 30-40 500-1000 0.015
1-45 250 0.001
GHz MW/mrs I

I Some of the other major facilities of the HPM Project are listed below.

o Wideband EM Pulse Environment (WEMPE) Pulsers and Antennas

WEMPE IV WEMPE V
Peak Voltage 12 kV 100 kV
Rise Time 300 ps 200 ps
Pulse Duration 8 rs 8 ns
PRF 1-2000 Hz Single Pulse

* Diagnostic Instrumentation

3 - EMP sensors, 1-10 GHz

- Tektronix oscilloscopes
10 Model 1776
3 Model 2104
1 Model 7250
4 DCS cameras
- Photodigitizer and utilities

115
- HP network analyzers
1 Model 8753B, 0.30-6.0 GHz
1 Model 8408, 0.05-18.0 GHz

e Other facilities

- MPF-14 laboratory
- MPF-18 laboratory
- TA-49 field test area
- High-explosive test vessel
- Mobile anechoic chamber
- Mobile TEM cell
- 1.3-G0Hz lens and mirror system
- Mobile instrument van

D.1.2 Scientists and Engineers

Los Alamos staff members are listed in Table D.2 who have backgrounds related to
some of the more significant problems to be investigated in an impulse-radar R&D
program. In the early stages of preparation of this report, an extensive checklist was
prepared of all the problems that we could identify that appeared to be significant
for R&D in inpulse radar. As work on the document progressed, contacts were
made with management and key personnel in several divisions of the Laboratory.
Within a month we were pleased to see that, for every problem area that we had
identified, several personnel at Los Alamos were identified that were exceptionally
well qualified by education and substantial recent experience. Severil of these staff
members made significant contributions to this report, for which the authors are
most grateful.

116
U
I Table D.2 Staff Members and Skills Related to UWB Radar

George Cooper Radar signal processing, radar system

design and analysis, applied mathematics
Evan Iverson EM theory and experimentation, PCPS
modeling, signal analysis and processiug,
target identification, applied mathematics
3 Robert Kelly EM theory and practice, wideband electronic
instrumentation for nuclear tests
Robert Hoeberling High-power microwave system design,
I testing and management
Patrick Hagen Applied mathematics, singular
perturbation theory
Robert Roussel-Dupre Theory and measurements of EM propagation
Daniel Holden through plasmas, dust, rain, etc.
Ross Graves Theory and practice of radar system design,
applied mathematics
Paul Lewis Signal-processing algorithms and architectures
of special-purpose computers
Wesley Unruh Microwave physics and electronics, Doppler
radar for wind-shear detection
Richard Hughes Modern theoretical physics, classical EM
theory, applied mathematics
Harold Frost Materials science for wide-band radomes and
3 high-voltage dielectrics
Richard Cooper EM theory, development and use of
programming codes for EM problems
Gloria Bennett Heat-transfer problems of high-power PCPS
and other semiconductor devices
3 Forrest Anderson Application of modified Radon transform to
biomedical and radar imaging
Dan Ross Electronic system design and analysis,
algorithm development, principal investigator

The principal authors of this report are George Cooper, Evan Iverson and
Dan Ross.

I j 117

I
D.2 Mechanical and Electronic Engineering
Management of R&D in impulse radar at Los Alamos is the responsibility of the
Electronics Technology Applications Group (MEE-5) of the Mechanical and Elec-
tronic Engineering Division. To address the wide variety of problems of impulse
radar, resources of other groups at Los Alamos will be utilized as required, the
following resources in partic-ilar.
"• HPM Project in Group AT-9
Robert Hoeberling, Group Leader
"*Optical Sciences and Engineering, Group CLS-8
Albert Saxman, Group Leader
Bert Kortegaard, Engineering R&D Section
""Scientists and engineers listed in Table D.1

The Mechanical and Electronic Engineering Division was formed in November

1986 with personnel from the Electronics, Physics, Energy, Materials Science, and
International Technology Divisions. The Division currently has 120 staff members,
80 technicians, and 25 administrative personnel located in seven groups and the di-
vision office, headed by Daniel Metzger. There are both applied and basic research
groups in electronics and mechanics, an electromechanical group, a computer ap-
plications group, and a processing support group for strategic nuclear materials.
There are roughly equal numbers of B.S., M.S. and Ph.D degreed staff members in
the Division, 60% with degrees in engineering.
The mission of the Division is to:
"Do world-class science and engineering for the development and execution of
Laboratory programs."
In support of this mission we have adopted the following goals.
"*Be recognized as a world-class science and engineering organization.
"* Be timely and cost effective.
"* Emphasize excellence in employee hiring and career development.

118
I
I aCollaborate with industry, universities, and other government laboratories
3 . Establish new technological ventures.

* Establish major multigroup programs emphasizing intelligence, advanced de-

3
Ifense technologies, energy systems, and technological competitiveness.

* Enhance capabilities in signal processing, electromagnetics, intelligent sys-

I terns, computer applications, sensors, electrochemistry, solid-state physics,
structural and thermal mechanics.

* Develop a funding base that provides continuity of effort.

I The budget of the Division in FY 1989 is approximately $37M of which half

comes from non-DOE sources, primarily the DoD. About one-fourth of the Division
I funds come from the nuclear weapons program. A sizable amount, $2M, is in
Institutional Supporting Research and Development (ISRD), which the Division
uses to conduct basic research and to seed new endeavors.
We are developing and expanding our efforts in signal-processing hardware and
algorithm development, parallel processing and advanced graphic techniques, ap-
3 plication of intelligent systems and decision aids, sensor and detector development,
electrochemical processes and fuel cells, solid-state materials and theory, supercon-
3 ductivity theory and materials, structural analysis techniques, and thermal systems
such as heat pipes and acoustic engines. We provide support to the Laboratory
U in instrumentation, mechanical design, quality assurance, and facility control. We
plan to increase our activities in intelligence, defense, energy programs and ISRD
3 projects.

3 D.2.1 Electronics Technology Applications

Bruce Noel, Group Leader of MEE-5, Electronics Technology Applications, has

been designated Program Manager for the Impulse-Radar Project with Dan Ross,
electronic systems engineer in MEE-5, as Principal Investigator. Internal critical
! 119

I
reviews will be necessary at frequent intervals to monitor progress on projects of
such technical breadth and complexity as R&D in impulse radar. MEE-5 uses two
computer-based tools to manage its projects: (1) an accounting tool for control
at the group level, which also provides input to the Laboratory system, and (2) a
PERT-chart system for tracking tasks and schedules. Some of the current MEE_5
projects are listed below:

"* Attack Sensor Systems and Environment Technology (ASSET) provides a

means of detecting an attack on a satellite and transmitting information re-
garding the attack.

"* Thermographic Phosphors provide means of measuring temperatures of tur-

bine blades and other surfaces inaccessible by clasical techniques.

"* Beam Diagnostic Instrumentation provides components of the Beam Experi-

ment Aboard Rocket (BEAR) project.

"* Predictive Fuze provides improved accuracy of aimable and isotropic fragmen-
tation warheads against ICBM and aircraft targets using measurements from
a proposed millimeter-wave radar.

" lnegrated Tactical Decision Aid (ITDA) provides a pilot with information on
nuclear explosions to guide evasive maneuvers.

"* Smart Information System. employ advanced data base technology and arti-
ficial intelligence in a variety of DOE and DoD applications.

MEE-5 has extensive laboratory equipment including time-domain reflectome-

ter, picosecond pulser and wideband oscilloscopes that are useful for R&D in im-
pulse radar. Figure D.2 shows antenna components obtained from Paul Van Etten
at RADC Ln a preliminary test arrangement in the laboratory.

120
.. . '• ".: . ..
. .8. .. ...

••i!• " .-..........

I
I

.!Figure D.2:- Tes t of the RADC antenna

121
D.2.2 Short Biographies of the Authors

George R. Cooper received his BSEE, MSEE, and Ph.D. from Purdue Uni-
versity. He was on the faculty of the School of Electrical Engineering at Purdue
University from 1949 to 1985, where he taught courses in system theory, estimation,
communication theory, random processes, and information theory. He did research
in these areas and supervised more than 50 Ph.D. theses and more than 20 Masters
theses. During his last 15 years at Purdue he also served as Coordinator of the
Electrical Engineering Graduate Program, in addition to his teaching and research.
He is currently Professor Emeritus of Electrical Engineering at Purdue University.
Dr. Cooper has also served as a consultant to many industries and government
laboratories in the areas of communication system analysis, radar system analysis,
signal processing, and parameter estimation. Since retiring from Purdue in 1985,
he has been self-employed as an independent consultant. His most recent activities
have been in the areas of spread-spectrum communication, detection of LPI signals,
data fusion in multiple-sensor networks, and novel radar systems. He has been a
consultant since January 1986 to Los Alamos National Laboratory, Group MEE-5.
Dr. Cooper is the author or co-author of five textbooks in the areas of system the-
ory, communication theory and random processes, and is the author or co-author
of more than 75 published technical papers. He is a Fellow of the IEEE.

A. Evan Iverson received his B.S. in electrical engineering from the University
of New Mexico and his M.S. and Ph.D. in applied mathematics from the Uni-
versity of Arizona. He joined Los Alamos in 1977 to work at the Meson Physics
Facility. As a beam-line physicist and accelerator engineer there, he designed elec-
tronic instrumentation and particle detectors for the accelerator and for particle
physics experiments. Since joining MEE-11 in 1980, his research activities have
included high-speed electronics and electro-optics, mathematical research in non-
linear dynamical systems, and theoretical modeling of phenomena in semiconduc-

122
tor and superconducting devices, the PCPS devices in particular. He specializes
in mathematical modeling and analysis, mathematical physics, nonlineax ordinary
and partial differential equations, and theoretical nonlinear dynamics. His current
research activities include the theory of non-steady-state electron transport in semi-
conductors; mathematical analysis of time-dependent convective/diffusive transport
equations; problems in nonlinear EM theory; and the derivation and analysis of the-
I oretical models for superconducting electronic devices. Under his internally funded
research project in impulse radar, he is beginning research on transient EM scat-
I tering theory (analytic and computational), overall system modeling and analysis,
parametric system identification applied to impulse-radar target identification, and

the development of antenna structures for transient EM signals. Dr. Iverson is a

member of the Society for Industrial and Applied Mathematics, the Applied Com-
putational Electromagnetics Society, and the Santa Fe Institute.

Dan C. Ross received his BSEE and MSEE from Purdue and has over 40 years
of experience in engineering, research, management and teaching. He was selected
by IBM as their first employee in the IBM Resident Graduate Study Program
and received the doctorate with distinction from The Johns Hopkins University.
He was an engineering executive in IBM Federal Systems for 16 years, where he
3 directed systems engineering and research in air defense, air traffic control, naviga-
tion, surveillance and telecommunications. Dr. Ross taught electrical engineering
3 at Purdue, Johns Hopkins, United States Military Academy, United States Naval
Academy and San Jose State University. He is currently an adjunct professor at
New Mexico Highlands University, where he is a consultant to the President and to
the Faculty. He has been active in IEEE and other professional societies as com-
mittee member, session organizer, reviewer and author. He was a member of the
National Academy of Sciences study for NASA on Peaceful Uses of Space. He has
consulted on the technology and policy of satellite communications for the Pres-
idential Task Force on Telecommunications Policy, Office of Telecommunications

i 123
I
Policy in the Executive Office of the President, and for IBM, Corporation for Pub-
lic Broadcasting, Control Data Corporation, Electronic Data Systems Corporation,
and other major corporations, He has served Los Alamos National Laboratory for
seven years, where he is currently working on a millimeter-wave sensor for use in an
advanced type of fuze for guided missiles and on developing new projects for Group
MEE-5. Dr. Ross is active in community affairs in Los Alanmos County and is a
member of the University of California Retirement. System Board.

124
I
II

i Appendix E

i Relevant Contacts
I
Representatives of Los Alamos National Laboratory have visited several Department
of Defense (DoD) organizations to gather information on UWB radar technology
and potential applications; and several DoD representatives have visited Los Alamos
during the past six months. Contacts have also been made with universities and
industrial corporations active in radar development.

Non-DoD Contacts Topics

ANRO Corporation
David Barton Radar systems analysis
iBoeing
"I Aerospace
Terence W. Barrett EM theory applied to impulse radar
I Catholic University
Henning Harmuth EM theory applied to impulse radar
I Power Spectra, Inc.
Steve Davis Manufacture of PCPS devices based
on previous development at Los Alamos
University of Texas
W. C. Nunnally Impulse-radar experiment based on
Kim Reid PCPS devices developed at Los Alamos

I1
I
125

I
DoD Contacts Topics

Defense Intelligence Agency

W. E. Thompson EM theory applied to UWB
LT;C August Golden radar
Herb Dimmoch LANL paper for NOSC

Defense Research and Engineering

LTC Barry Crane DoD program plans for UWB
radar

U.S. Air Force

LTC James Taylor, ESD UWB technology and applications
Al Dahlgren, ESD

Paul Van Etten, RADC Review of RADC experiments

Michael C. Wicks, RADC on impulse-radar transmitters
and antennas

U.S. Army

John David, HDL Potential SAR applications of

John McCorkle, HDL impulse-radar technology

U.S. Navy
Vince Pusateri, NOSC Discussion of paper to be
James R. Buss, NOSC prepared for NOSC by LANL
George Byram, NOSC on UWB radar and relevant
Warren Stevens, NOSC LANL capabilities
James K. Hall, ONT

Merrill Skolnik, NRL Need for theoretical and

experimental work on UWB-
radar technology

Alan Petty, NRL Potential SAR applications of

Brooks Dodson, NRL impulse-radar technology
Frank Shoop, OPNAV USN program plans for impulse
radar

Randy Coggins, NAVSEA Impulse radar

126
II
I
* Appendix F

I References
I
D. K. Barton, Modern Radar System Analysis; Axrtech House, Norwood, MA, 1988

I- C. L. Bennett and W. L, Weeks, "A Technique for Computing Approximate Elec-

tromagnetic Impulse Response of Conducting Bodies", TR-EE68-11, NSF GK2367,
Purdue University, School of Electrical Engineering, June 1968
C. L. Bennett, and G. F. Ross, "Time-Dc-nain Electromagnetics and Its Applica-
tions", Proceedings of the IEEE, Volume 66, Number 3, p 299; March 1978

J. N. Brittingham, "Focus Waves in Homogeneous Maxwell's Equations: Transverse

Electric Mode", Journal of Applied Physics, Volume 54, Number 3, p 1179; March
1983
W. S. Burdic, Radar Signal Analysis; Prentice-Hall, Englewood Cliffs, 1968

George R. Cooper; "High-Resolution Radar Using PCPS Devices", E-Division Mem-

orandum, Los Alamos National Laboratory; 27 February 1986

George R. Cooper and Clare D. McGillem, Modern Communications and Spread

Spectrum, McGraw-Hill, New York, 1986

J. V. DiFranco and W. L. Rubin, Radar Detertion, Prentice-Hall, Englewood Cliffs,

1968
D. G. Dudley, "Parametric Modeling of Transient Electromagnetic Systems", Radio
Science, Volume 14, Number 3, p 387; 1979

August Golden Jr., Radar Electronic Warfare, American Institute of Aeronautics

3 and Astronautics, Washington, DC, 1987

I
127
I
U
H. F. Harmuth, "On the Effect of Absorbing Materials on Electromagnetic Waves
with Large Relative Bandwidth", IEEE Transactions on Electromagnetic Compat-
ibility, Volume EMC-25, Number 1, p 32; February 1983 (a)

H. F. Harmuth, "Antennas for Nonsinusoidal Waves. I. Radiators", IEEE 7?ansac-

tione on Electromagnetic C impatibility, Volume EMC-25, Number 1, p 13; February
1983 (b)
H. F. Harmuth, "Antennas for Nonsinusoidal Waves. IH.Sensors", IEEE Transac-
tions on Electromagnetic Compatibility , Volume EMC-25, Number 2, p 107; May
1983 (c)
H. F. Harmuth, "Antennas for Nonsinusoidal Waves. III. Arrays", IEEE Transac-
tions on Electromagnetic Compatibility, Volume EMC-25, Number 3, p 346; August
1983 (d)
H. F. Harmut'J, "Use of Ferrites for Absorption of Electromagnetic Waves", IEEE
Transactions on Electromagnetic Compatibility, Volume 27, Number 2, p 100; May
1985
H. F. Harmuth, Propagation of Nonsinusoidal Electromagnetic Waves, Academic
Press, Orlando, 1986
A. E. Iverson, "The Mathematical Modeling of Photoconductive Power Switches",
Transactions of the Society for Computer Simulation, Volume 5, Number 3, pp.
175-191, July 1988
E. F. Knott, J. F. Schaeffer, and M. T. Tuley, Radar Cross Section, Artech House,
Norwood, MA; 1985

J. A. Landt, E. K. Miller, and M. Van Blaricum, "WT-MBA/LLL1B: A Computer

Program for the Time-Domain Electromagnetic Response of Thin-Wire Structures",
UCRL-51585; Lawrence Livermore National Laboratory, May 6, 1974

Chi H. Lee, Editor, Picosecond Optoelectronic Devices, Academic Press, Orlando,

1984

128
I Ljung, L., "System Identification", Department of Electrical Engineering, S-581
83, Linkoping University, Linkoping, Sweden; 1981

L. Ljung and K. Glover, "Frequency-Domain versus Time-Domain Methods in Sys-

tern Identification", Automatica, Volume 17, p 71; 1981

L. Ljung, "Convergence Analysis of Parametric Identification Methods", IEEE

Transactions on Automatic Control, Volume 23, Number 5, p 770; 1978

"MAFIA User Guide", The MAFIA Collaboration, Deutches Elektronen-Synchrotron,

Hamburg, Federal Republic of Germany; May 1988.

L. Matin, "Natural-mode Representation of Transient Scattered Fields", IEEE

Transactions on Antennas and Propagation, Volume 21, p 809; 1973

R. N. McDonough, "Matched Exponents for the Representation of Signals", Doc-

I toral Dissertation, The Johns Hopkins University; 1963

E. K. Miller and J. A. Landt, "Direct Time-Domain Techniques for Transient Ra-

diation and Scattering from Wires", Proceedings of thc IEEE, Volume 68, Number

11, p. 1396, 1980

E. K. Miller, "Natural-Mode Methods in Frequency- and Time-Domain Analysis",

in Theoretical Methods for Dctermining the Interaction of Electromagnetic Waves
with Structures, Sijthoff and Noordhoff, Netherlands; 1981

D. L. Moffatt and R. K. Mains, "Detection and Discrimination of Radar Targets",

IEFS, Transactions on Antennas and Propagation, Volume 23, p 358; 1975

W. C. Nimnally and R. B. Hammond; "Photoconductive Power Switches", LA-

9759-MS; Los Alamos National Laboratory, Los Alamos, New Mexico; April 1983

I R. Prony, "Essai experimental et analytique sur les lois dela dilatabilite de fluides
elastiques et sur celles de la force expansive de la vaperu de l'alkool, a differentes
temperatures", Journal l'Ecole Polytechnique, Volume 1, Number 2, p 24; 1795

S. Ramo, J. R. Whinnery and T. Van Duzer, Fields and Waves in Communication

Electronics, John Wiley and Sons, New York; 1965

I
1
129

I!
G. F. Ross, "Early Developments and Motivations for Time-Domain Analysis and
Application", in Time-Domain Measurements in Electromagnetics, Edited by E. K.
Miller, Van Nostrand Reinhold, New York; 1986

Q. F. Ross, J. D. DeLorenzo and K. W. Robbin; "ANRO's Baseband Reflectometer

for Area Protection of Nuclear Facilities", DNA-TR-88-127; ANRO Engineering
Consultants Inc., 5 Militia Drive, Suite 104, Lexington, MA; 31 May 1988

Arnold Sommerfeld, Electrodynamics, Translated by Edwaxd G. Ramberg, Volume

3, Lectures on Theoretical Physics; Academic Press, New York; 1952
Thornton, B. S., "Comments on Use of Ferrites for Absorption of Electromagnetic
Waves", IEEE Transactions on Electromagnrtic Compatibility, Volume EMC-28,
Number 4, p 285; November 1986

Torrieri, D. J., Principles of Military Communication Systems, Artech House, Ded-

ham, MA; 1981

C. Peter Ulriksen; "Application of Impulse Radar to Civil Engineering", Doctoral

Dissertation; Lund University of Technology, Sweden; 1982

M. L. Van Blaricum and R. Mittra, "A Technique for Extracting the Poles and
Residues of a System Directly from its Transient Response", IEEE Transactions on
Antennas and Propagation,
Paul Van Etten; "Impulse Radars", Proceeding of International Workshop on Re-
mote Estimation of Sea Ice Thickness, University of St. John's, Newfoundland;
25-26 September 1979
J. R. Wait, Electromagnetic Waves in Stratified Media, 2nd Edition, Pergamon,
Oxford, 1970
T. T. Wu, "Electromagnetic Missiles", Journal of Applied Physics, Volume 57,
Number 7, p 2370; April 1985

R. W. Ziolkowski, "New Electromagnetic Directed-Energy Pulses", Microwave and

Particle Beam Sources and Propagation, SPIE Volume 873, p 312; 1988

130