· ..

---- ~--

Microwave and RF Wireless Systems

.. .. ~ ____.___-

Microwave and RF Wireless Systems

David M. Pozar

Department of Electrical and Computer Engineering University of Massachusetts at Amherst



New York - Chichester - Weinheim Brisbane - Toronto - Singapore




Thls ~OGk was set in Times Roman by reobBqQk~ and primed and bound by R.R. ,I\)nnnelie}' &; SOilS, The eover was pri rIled~y, Phoenix Color Carp.

Ihi.~, be I) k is pi" tn ted 011 aci d-lree [luper. @

Copyrigh; © 200 I John Wiley & Sons. toe. All rights reserved,

No part of this publication may be reproduced, stored in ~ retrieval system-or transmitted in any form or by lIll)' means, electronic. mechanical. photocopying recording, scanning or otherwis~,excep[,as permitted under Sections l07 or I ORo!" t~e 1'176 United Sl~les Copyright At;t. WiLhQu\ elther lite prior written pe-!'ITliSSIQ!1 of the Publisher or authorization through payment ot- the appropriate per-copy fee to the Copyright Clearance-Center, 222 Ro ewood Dlrive, Danvers, MA n 1923. (:9,78) 750;84DO. J;l~

(978-) 750-4470, Requests to-the Publisher for permission should be addressed to tile Permissions Departmen!. John W£ley &. Sons .. lnc., 605 Third Avenue, New )"(jr~. NY 10158,00 L2., ,(i I 2) ~50,,(jO,ll. fox (212) 8S0-fJOpS. B-MIllI; PBRNtREQ@WILEY.cOM. To order bookl\ please cnU1(800J-22'5-594S.

Library o/CQI!gre.ls Calalagillg ill Pubtloation Data POLm, David M.

Mi:crowave and RF wi relesss ysrems f D3;Vid M, Pozar,

p. em.

Includes bibliog,~I[)bic;ll referenC!ls. ISBN 0"471-32282-2 (oloUI: alk, paper)

L Wireless. communication Sysl~I'nS, ~L MIN wave eommunieation systems, 3. Radio I'I-eqUMCy, 4. Mobile communication stst(ll:n~. L Title,

TK51 Q.t2.P:S9 6~1.3~2.......uli21



• r ..... ~ • ..-, ....


Wireless system design is aile of the most exciting fields in electrical engineering today. In economic terms, wireless applications that include cellular and PCS telephony, wireless Iocal area networks (WLANs), global positioning satellite (GPS) service, direct broadcast televi sion service (DBS), local mu] ripoint distribution systems (LMDS) , and radio frequency identification systems (RFID) constitute a yearly market in excess of $1 OOB, and strong growth is predicted over the long term. Prom 11 technical perspective, wireless system design involves a close integration ofa variety of topics that include antennas and propagationeffeets, RF and microwave circuit design; noise and intermodulation effects, digital modulation methods, and digital signal processing.

The pllrpose of this text is to present a cohesive overview of the fundamental subjects reg uired for the design and analysis of the RF stages of modern wireless systems, i I1C ludi ng antennas, propagation, fading, noise, receiver design. modulation methods, and bit error rates. Material is also included on the design of key components used in wireless systems, such as filters, amplifiers, mixers, oscillators, and phase-locked loops. Major wireless applications, such as cellular and pes telephony, GPS,. DES, WLANs, and LMDS systems are described. and many designexamples are given in the context of these systems. Required fundamentals on transmission lines, S parameters, impedance matching, and random processes are also included.

A key premise of this book is that a coherent understanding of wireless system performance and design can only be obtained by treating the relevant technical topics in an integrated manner. A collection of individual COUJses in antennas, microwave engineering, and communications engineering is unlikely to provide- an understanding of the interplay between different stages and their effect on the overall performance of the system. Courses in antennas or microwave engineering, for example, generally will not discuss the effect of noise or Rayleigh fading on bite-nor rates in a digitalfadio. Similarly, a course in communications theory will probably not discuss component noise fi gure and i nterrnodulation requirements for different modulation schemes and data rates. \Vru.Le the emphasis of this book is on the RFand microwave stages of wireless systems, we have included a chapter on modulation methods because this allows us to provide a complete characterization of a wireless system from an input data stream through the transmi tter, the antennas and propagatl on channel, and the-receiver, resul ting in overall system performance measures in terms of bit error rate, data rate, OJ range.

There is enough material here for a full year course in RF and microwave design or wireless systems at the senior or first-year graduate student level. Prerequisites ideally would include junior-level electronics, electro magnetics, transmission lines, probabi I i ty, and


vi Preface

random variables, but Chapters 2-4 contain brief but reasonably complete reviews of these topics to the extent that they will be required later in the text. If students have a familiarity with transmission lines. S parameters, and RF eire uit design.a one-semester course caved ng Chapters 1,3,4,9, and 10 can be presented with a focus on wireless system analysis and design. Some teachers may prefer to cover the systems-oriented material in Chapters 1-4 and 9-10 first followed by selective coverage of component design in Chapters 5-8, Other combinations are possible, depending on the background of the students and the opinions 01 the instructor. Much of the material on microwave circuit design presented in Chapters 2, 5, 6, and 7 was drawn from my text Microwave Engineering, with additional topics that include ceramic bandpass filters, stability, power amplifiers, FET mixers. and nonlinear mixer analysis.

Computer codes relevant to some of the problems and examples in the text are available on the Wiley Web site at www.wiley.com/college/pozar, These can be used for computing the complementary error function calculating the noise figure and mtermodularion point of a cascade system, determining the stability parameters of 3 transistor amplifier, and other

applications.· 1


I would first like [0 thank the students who participated in our course on wireless systems at the University of Massachusetts, and who used notes and draft copies or this book for several years. Thanks also go to my colleagues in microwave engineering and [he Wireless Communications Center at the University of Massachusetts. I would especially like to acknowledge my late colleague and friend, Bob Mclntosh, who provided a guiding vision for many of us through the years, Several people in industry were helpful in providing photographs of wireless hardware: Lamberto Raffaelli and Earl Stewart of Arcom, Carl Marguerite and Peter Alfano of Sage Laboratories, Fred Dietrich of Globalstar, A. W. Love of Rockwell, Tuli Herscovici of Spike Technologies and Harry Syrigos of Alpha Industries. Juraj Bartolic deserves thanks for providing a simplified derivation of the J-L" parameter stability test ill Chapter 6, as does Dennis Goeckel fOT his advice on an early version of the manuscript, Finally, 1 would like to thank Bill Zobrist, Jennifer Welter, and Suzanne Ingrao for their invaluable help in completing this project.

David M. Pozar Amherst, MA


,. ..,.__ ..

of these iiliarity overing 'sis and ers 1-4 . Other Jillions hapters

topics nlinear

ailable puting oint of I other

.reless rf this nd the cially aiding riding , Carl \. W. ~lpha ne {1-early ~, and

)ozar IlA





1.1 WIreless Systems and Markets 2

Classification of Wireless Systems

Cellular Telephone Systems


• Personal Communioatlons Systems 5 • Satellite Systems for Wireless

Voice fwd Data 6 • Global Positioning Satellite System 7 •

Wireless Local Area Networks 9 • Other Wireless Systems 9

1.2 Design and Performance Issues 11

Choice of Operating Frequency J 2 •

• Circuit Switching versus Packet Switching

• Radiated Power and Safety 15 •

1.3 Introduction to Wireless System Components

Basic Radio System 17 • Antennas

Amplifiers 21 • Mixers 22

Baseband Processing 23

1.4 Cellular Telephone Systems and Standards 23

Cellula)" and the Publi Switched Telephone Network 23 • AMPS Cellular Telephone System 24 • Digital Personal Communications System Standards 26

Multiple Access and Duplcxing 13
13 • Propagation 14
Other Issues 16
19 • Filters 19 •
• Oscillators 22 • 2



2.1 Transmission Lines 29

Lumped Element Model for a Transmission Line 30 • Wave Propagation on a

Transmission Line 31 • Lossless Transmission Lines 32 •

Terminated Transmission Lines 3.3 • Special Cases of Terminated

Transmission Lines 36 • Generator and Load Mismatches 39

2.2 The Smith Chart 41

Derivation ofthe Smith Chan 42 • Basic Smith Chart Operations 44 •
Using Tho Admittance Smith Chart 4S
2.3 Microwave etwork Analysis 47
Impedance and Admittance Matrices 47 • The Scatteri ng Matrix SO •
The Transmission ABCD) Maoix 53 vii

viii Contents

2.4 Impedance Matching 55

The Quarter- Wave Transformer 56

Single-Stub Tuning 6.1

Matching U ing LvSecti ns





3.1 Rev iew of Random Processes 69

Probability and Random Variables 69 •

69 • The Probability Density Function Probability Density Functions 71 • Autocorrelation and Power Spectral Density

3.2 Thermal Noise 74

Noise Voltage and Power 74

3.3 Noise ill Linear Systems 77

Autocorrelation ami Power Spectral Density in Linear Systems 77

Gaussian White Noise through an Ide-ill Low-pass Filler 78 •

Noise through an Idea] Integrator 79 • Mixh1g of Noise

The Cumulative Distribution Function

70 • Some Important

Expected Values 71 •



Gaussian White

80 •

Narrowband Repre 'entation of Noise 81

3.4 Basic Threshold Detection 83

Probability of Error 84

3.~ Noi e Temperature and Noise Figure 87

Equivalent Noise Temperature 87 • Measurement of Noise Temperature 88

• Noise.Figure 89 • Noise Figure of a Lossy Line 90 • Noise

Figure of Cascaded Components 91

3.6 Noise Figure of Passive Networks 93

Noise Figure of 11 Passive '1Wo- P(lT( Net work 94 • Application to a Mismatched

Lossy Line 95 • Application to a Wilkinson Power Divider 96

3.7 Dynamic Range and Intermodulation Distortion 98

G,Un Compression 99 • Iruerm dulation Distortion 100 •

Third-Order lntercepr Point 101 • Dynamic Range J 02 •

Intercept Point of Cascaded Components 104 • Passive lntermodularion I 06






Antenna System Parameters


Fields and Power Radiated by an Antenna 112 • Far-Field Distance 114

• Radiation Intensity 114 • Radiation Pattern 115 •

Directivi ty I 16 • Radiation Efficiency 117 • Gain I 18

• Aperture Efficiency 118 • Effective Area I J 8 • Antenna

Polarization 1 19

The Friis Equation 120

The Friis Equation 120 •


Effective Isotropic Radiated Power

121 [23

• lrnpednnce Mismatch 122 • Polarization Mismatch

Equivalent Circuits for Transmit and Receive Antennas 124

Contents Ix


Antenna Noise Temperature


Background and Brightness Temperature

127 • OfT 129

4.4 Basic Practical Antennas 13"1


Antenna Noise Temperature

Electrically.Small Dipole Antenna 132 • [-I alf- Wave Dipole Antenna 134
• Monopole Antenna 135 • Sleeve Monopole Antenna 136 •
Electrically Small Loop Antenna 137
4.5 Propngation 138
Free-space Propagation 139 • Ground Reflections 140 • Path Loss
for Ground Reflections 142 • Realistic Path Loss 1.42 • Attenuation 143

4.6 Fading ]44

Rayleigh Failing 145




5.1 Filter Design by the Insertion Loss Method 152

Characterization by Power Loss Ratio 152 • Maximally Flat Low-Pass

Filter Prototype 154 • Equal-Ripple Low-Pass Filter Prototype 157 •

Linear Phase Low-Pass Filter Prototype 158

5.2 Filter Scaling and Transformation 158

I rrrpcdance Sea I' i ng 158 • Frequency scaling For low-pass filters 159 •

Low-pass to High-pass Transformatlon 162 • Bandpass and Bandstop

Transformation 164

5.3 Low-Pass and High-Pass Filters Using Transmission Line Stubs 168

Richard's Transformation 168 • Kuroda's lderuiries 169

5.4 Stepped-Impedance Low-Pass Filters J 73

Approximate Equivalent Circuits for Short Transmission Line Sections 174

5.5 Bandpass Filters Using Transmission Line Resonators 178

Impedance and Admittance Inverters 178 • Bandpass Filters Using

Quarter·Wave Coupled Quarter-Wave Resonators 179 • Bandpass Filters

Using Capacitively Coupled Quarter- Wave Resonators 183





6. I PET and Bipolar Transistor Models

Field Effect Transistors 190 •

6.2 Two-port Power Gains J 94

Den nirions of Two-Port Power Gai ns 194

• Further Discussion of TWO-P011 Power Gains

Bipolar Transistors


S peel al Cases 198





Stability Circles 200 • Tests for Unconditional Stability 202

6.4 Amplifier Design Using S Parameters 205

Design for Maximum Gain 205 • Maximum Stable Gain _07

• Constant Gain Circles and Design for Specified Gain 210

x Contents

6.5 Low-noise Amplifier Design

6.6 POwer Amplifiers 218

Characteristics of Power Amplifiers and Amplifier Classes 218 • Large-Signal Characterization of Transistors 219 • Design of Class A Power Amplifiers







7.1 Mixer Characteristics 225

Frequency Conversion

Loss 228- •

225 •

Noise Figure

Image Frequency 221 • Conversion

229 • Intermodulation Distort-ion 230


• Isolation Diode Mixers

230 230

Small-Signal Diode Characteristics

Large-Signal Model 233 •

PET Mixers 239

23'1 • Si ngle-Ended Mixer

Switching Model 237



Single-Ended FET Mixer 239

7.4 Other Mixer Circuits 243

Other PET Mixers


Balanced Mixers


Small-Signal Analysis of the Balanced Mixer


lruage Reject Mixer






8.1 Radio Frequency Oscillators 25]

General Analysis 251 • Oscillators Using a Common Emitter BJT 252

• Osciuatcrs Using a Common Gate FET 254 • Practical Considerations

255 • Crystal Oscillators 256 • Voltage-Controlled Oscillators 258

8.2 Microwave Oscillators 258

Negative Resistance Osci llators


Transi stor Osc i.1l ators


Dielectric Resonator Oscillators 264

8.3 Frequency Synthesis Methods 268

Direct Synthesis 268 • Digital Look-up Synthesis 269 •

Phase-Locked Loops 271 • Practical Synthesizer Circuits 272 •

Fractional-N Phase-Locked Loops 273

8.4 Phase-Locked Loop Analysis 273

Phase Detectors 274 • Transfer Function for the Voltage-Conuolled Oscillator

275 • Analysis of Linearized Phase-Locked Loop 275 • First-Order

Loop 277 • Second-Order Loop 278

8.5 Oscillator Phase Noise 280

Representation ofPhase Noise 281 • Leeson's Model for Oscillator Phase Noise

282 • Effect of Phase Noise on Receiver Performance 285

------- ,.-

-.... .. .. ~~

Contents xi




9.1 Analog Modulation 288

Single-Sideband Modulation 289 • DoubJe-Sideballd Suppressed-Carrier

Modulation 292 • Double-Sideband Large-Carrier Modulation 295

• Envelope Detection of Double-Sideband Modulation 296

• Frequency Modulation 298

9.2 Binary Digital Modulation 303

Binary Signals 304 • Amplitude Shift Keying 304 • Frequency

Shift Keying 306 • Phase Shift Keying 307 • Carrier

Synchronization 309

9.3 Error Probabilitie for Binary Modulation 309

PCM Signals and Detectors 3,10 • Synchronous ASK 311 •

Synchronous PSK 312 • Synchronous FSK 313 • Envelope

Detection of ASK 313 • Envelope Detection of FSK 316 • Btt

Rate and, B andw idth Efficiency 317 '. Compari son of ASK, FSK, and PSK

Systems 3 '18

9.4 Effect of Rayleigh Fading on Bit Err r Rates 320

Effect of Rayleigh Fading on Coherent PSK 321 • Effect of Rayleigh Fading

on Noncohcrent FSK 322 • Comparison of Faded and Nonfaded Error Rates 322


M-ary Digital Modulation Quadrature Phase Shift Keying

• M -ary Phase Shift Keying

• Channel Capacity 331

324 32.5 330

Probability of Error for QPSK Quadrature Amplitude Modulation

327 330




Receiver Architecture 335

Receiver Requirement


Tuned Radi 0 Frequency Receiver 337 •

Superheterodyne Receiver 338 •

Direct Conversion Receiver 337

Duplexing 333

Dynamic Range 340

Mi ni mum Detectable Signal 341

343 • Automatic Gain Control

lruermodul ation 346

10.3 Frequency Conversion and Filtering

SelectionofIf'Frequency 347 •




344 •

343 • Dynamic Range

Compression and Third-order

347 Filtering

Spurious-free Range


10,4 Examples of Practical Receivers 350

PM Broadcast Receiver 350 • Digital Cell u I ill Receiver 351 •

Mill i rueter Wave Poi nt-to-Poiru Radio Receiver 352 • Di rect -Conversiou GSM

Rec .. eiver 355


A Wireless System Frequency Bands 358

B Useful Mathematical Results 358

xii Contents


Fourier and Laplace Transforms The Complementary EITor Function

Chebyshev Polynomials 361

Decibels and Nepers 362




its 0 by U bye 60 n busi also



mnn SYSt4 and ehal worl

the 1 Up-I! less wire uses SySb data


J_ -.











Introduction to Wireless Systems

In 'the early 1980" a marketing fum hired by AT&T to survey the potential U.S. marker for its newly inaugurated cellular phone service arrived at an estimate of les than 900,000 users by the year 2000. Like many predictions of technological progress, this one turned out to be off by a wide margin-in 1998 the number of cellular subscribers in the United States was over 60 million (already an error of more than 6000 percent). It is now estimated that half of all business and personal communications will be wireless by the year 2010 [1]. Rapid growth is also occurring with other wireless systems, such as Direct Broadcast Satellite (DBS) television service, Wireless Local Area Networks (WLANs), paging systems, Global Positioning Satellite COPS) service, and Radio Frequency Identification (RFID) systems. It is estimated that the number of consumer wireless devices will exceed 300 million by the year 2000 L21-[3]. These systems promise to provide, for the first time in history, worldwide connectivity for voice, video, and data communications. The uccesses of wireless technology to date, and the technological challenges of future wireless systems, make this an exciting and rewarding field in which to work.

In this book we study the operation and design of wireless systems from the perspective of the radio freq uency (RF) or microwave su bs ystems, These include modu 1 ators and frequency up-conversion circuits in the wirele s transmitter, the transmit and receive antennas, the wireless propagation channel and the frequency down-conversion and demodulator circuits in the wireless receiver. Generally these subsystems are analog in nature, even if the wireless system use. digital modulation techniques. We will see that noise and other characteristics of these subrysterns set the ultimate limits on the performance of a wireless system, in terms of maximum data rate, operating range, power requirements. and error rates.


2 Chapter 1: Introduction to Wireless Systems



In this ectlon wegive a brief introduction to some of the major wireless systems in use today. These include wireless cellularand pes telephone systems. commercial satellite systems, wireless data networks, point-to-point radios, the global positioning system, and other wireless systems.

Classification of Wireless Systems

L1 the broadest sense, a wireless system allows the communication of information between two points without the use of a wired connection. This may be accomplished using sonic infrared, optical, or radio frequency energy. While early television remote controllers used ultrasonic signals, very low data ratesand poor immunity to interference make such systems a poor choice for modern applications. Infrared signals can provide moderate data rates, but the fact that infrared radiation is easily blocked by even small obstructions limits their use to short-range indoor application such as remote controllers and local. area data links. Similarly, optical signals propagating in an unobstructed environment can provide moderate to high data rates, but require a line-cf-sight path, and cannot be used where foliage, fog, or dust can block the signal. For these reasons. most modern wireless systems rely on RF Or microwave signals usuallyin the UHF (100 MHz) to millimeter wave 30GHz) frequency range. Because of spectrum crowding and the need for higher data rates, the trend is to use the higher frequencies in this range. so that the majority of wireless systems today operate at frequencies ranging from about SOO.MHz to a few gigahertz. RF and microwave signals offer wide bandwidths, and have the added advantage of being able to penetrate fog, dust, foliage, and even buildings and vehicles to some extent.

Historically, wireless communication using RF energy began with the theoretical work of Maxwell, followed by the experimental verification by Hertz of electromagnetic wave propagation, during the period from 1873 to L89!. Marconi built on this work to develop practical commercial radio communications systems in the early part of the 20th century, It is interesting to note that the term "wireless" dates back to this early period, and although replaced by the word "radio" for most of this century. wirele .I' is again the preferred description for most of today's cellular telephone data links, and satellite systems.

One way to categorize wireless systems is according to the nature and placement of the users, til a point-to-point radio system a single tran, mitrer communicates with a single receiver. Such systems generally u: e high-gain antennas in fixed positions to maximize received power and minimize interference with other radios that may be operating nearby in the same frequency range. Point-to-point radios are generally used for dedicated data communications by utility companies and for connection of cellular phone sites toa central switching office. Point-to-multipoint systems connect a central station to a large number of possible receivers. The most common examples are commercial AM and FM broadcast radio and broadca t television, where a central transmitter uses all antenna wi.th a broad beam to reach many listeners and viewers. Broadcast radio is similar in function to local mult ipoinr distribution systems (LMDS), which are presently bei llg depl oyed in urban areas to provide wireless television and Internet access to users within a small geographical area. Another example of a point-to-multipoint system is paging, where a central station can briefly communicate with many users over a large geographical region. Multipoinrto-multipoint systems allow simultaneous communication between individual users (who may not be in fixed locations). Such systems generally do not connect two users directly to each other, but instead rely on a gr.id of base stations to connect an individual user to a central switching office, Which then connects to the base station of the other user. Cellular

1.1 Wireless Systems and Markets (3

telephone systems and some types of wireless local area networks (WLANs) are examples of this type of application.

Another way to characterize wireless systems is in terms of the directionality of communication. In a simplex system, communication occurs only ill one direction from [he transmitter to the receiver. Examples of simplex systems include- broadcast radio and television. In a half-dupiex system, communication may occur in two directions, but not simultaneously. Early mobile radios and citizens band radio are examples of duplex systems, and generally rely on a "push-to-talk" function so thai a single channel can be u ed for both transmitting and receiving at different intervals. Some wireless data links also LIse half-duplex transmission. Full-duplex systems allow simultaneous two-way transmission and reception. Examples include cellular telephone and point-to-point radio systems. FulJduplex transmission clearly requires a duplexing technique to avoid interference between transmitted and received signals. This can be done by using separate frequency bands for transmit and receive ifrequency division dupiexing, PDD), or by allowing users to transmit and receive only in certain predefined time intervals (time division duplexing, TDD).

While most wireless systems are ground based, there is increasing interest in the development of satellite systems for voice, video. anddara communications. Satellite systems offer the possibility of communication with a large number of users over wide areas, perhaps including the entire planet. Satellites in a geosynchronous earth orbit (GEO) are po itioned approximately 36,000 krn above the Earth, and remain in a fixed position relative to the surface. Such satellites ale useful for point-to-point radio links between widely separated stati u , and are commonly used for television and data COlTllTIWliCaLi I1S thr ughout the world" At one time transcontinental telephone service relied heavily on such satellites, but undersea fiber optic cables have largely replaced satellites for transoceanic connections as being more economical, and avoiding the annoying delay caused by the very long round trip path between the satellite and the Earth. Another drawback of GEO satellites is that their high altitude greatly reduces the received signal strength, making it impractical for two-way communication with small transceivers. Low earth orbit (LEO) satellites orbit much closer to the Earth, typically in the range or 500 to 2000 km. TIle shorter path length allows communication between LEO satellites and handheld radios, but satellites in LEO orbits are visible from a given point on (he ground for only a short time, typically from a few minutes to perhaps 20 minutes. Effective coverage therefore requires a large number of satellites in different orbital planes.

Finally, wireless systems can. be grouped according to their operating frequency, The choice of operating frequency will be discussed in much more detail in a later section. but Table Ll lists the operating frequencies or some f the most common wireless systems

Cellular Telephone Systems

Cellular telephone systems were proposed in the 19708 in response to tbe problem of providing mobile radio service to a large number of users in urban areas. Early mobile radio systems could handle only a very limited number of users due LO inefficient use of the radio spectrum and interference between users. In 1976, for example the entire mobile phone system III New York City could support only 543 users [1]. The cellularradio concept introduced by Bell Laboratories solved this problem by dividing a geographical area into non-overlapping hexagonal cells, where each cell has its own transmitter and receiver (base L~lalion.) to communicate with the mobile users operating in that cell. Each cell site may allow as many as several hundred users to simultnneously communicate with other mobile users, or through the land-based telephone system.

The first cellular telephone system to offer commercial service was built by the Nippon Telephone and Telegraph company (NIT), and became operational in Japan in 1.97-9 [4J.

4 Chapter 1: Introduction to Wireless Systems

TABLE 1.1 Wireless System Frequencies

Wireless System

Operating Frequency

Advanced Mobile PhODe Service (AMPS)

T: 824--849 MHz R: 869-894 MHz

Global System Mobile (European GSM)

T: 880-9 15 'tY[Hz R; cn5~960 MHz

Personal Communications Services (PCS)

T: 1710-1785 MHz R: J805-1880 MHz

US Paging

93J-932 MHz

Global Positioning Satellite (GPS)

t.i: 1575,42MHz L2: 1227.60 MHz

Direct Broadcast Satellite CDBS)

11.7-12.5 GHz

Wireles: Local Area Networks (WLANs)

902-928 MFl,z 2.400-2.484 GHz 5.725-5.850 GHz

Local Multipoint Distribution Service (LMDS)

2.8 GHz

US Industrial, Medical, and Scientific bauds (ISM)

902-928 MHz 2.400-2.484 GHz 5.725,---5.850 GHz

'fIR = mobile Llnil transmit/receive frequency.

This was followed by the Nordic Mobile Telephone, (NJv1T) system in Europe, which began operation in 1981. The first cellular telephone system in the United Stares was the Advanced Mobile Phone System (AMPS), deployed by AT&T in J 983.. All of these systems use analog FM modulation and divide their allocated frequency hand, into everal hundred channels, each of which can support an individual telephone conversation, These early systems grew slowly at first, becaus e of the initial costs of developing an infrastructure of base rations and the initial expense of handsets, but by the 1990s growth became phenomenal.

In 1998 there were 64 million cellular phone subscribers and over 57,000 base stations in the United States, generating annual service revenues of $30 billion with a market penetration of about 35%. Worldwide there were about 2()Q million cellular ubscribers in 1997. While the approxirnately 700 million wired telephone lines far outnumber wireless telephone users.jhe growth rate of wireless is about 15 times that for wired lines.

In '199688% of all cellular telephones in the United States used the analog AMPS system, but newer digital standards have been growing in popularity and win soon replace the AMPS system. These systems are generally referred to as Second Generation Cellular, or Personal Communication Systems (peS). Third generation pes systems, which may inc] ude capabilities for email and Internet access, are in the planning rages.

1.1 Wireless Systems and Markets 5
TABLE 1.2 Majol' Worldwide Cellular and pes Telephone Systems
Year of Frequency Channel
Standard Country Introduction Type Band (MHz) Modulation Bandwidth
NTT Japan 1979 Cellular 86()-940 PM 25 kHz
NMT·450 Europe 1981 Cell ul ar 453--468' FM 25 kHz
AMPS United States 1983 Cellular 824--894 FM 30kHz.
E-TACS Em'ope 1985 Cellular 872-950 PM 25 kHz
C-450 Germany 1985 Cellular 450-466 PM 20 kHz
NMT-900 Europe 1986 Cellular 890-960 FM 12.5 kHz
J't .... CS Japan 1988 Cellular 860-92~ FM 25 kHz
GSM urope 1990 PCS 890-960 GMSK 200kHz
IS-54 United States 1991 PCS 824--894 DQPSK 30 kHz
AMPS United Stales 1992 Cellular 824-894 FM 10 kHz.
IS-95 United Slates 1993 PCS 824--894 QPSK L25MHz
"PDe Japan 1993 Cellular 810-1513 DQPSK 2S kHz
NTACS JapUJ1 1993 Cellular 843-922 PM 12.5 kHz Personal Communications Systems

Because of til rapidly growing consumer demand for wireless telephone service, a<; welt as advances in wireless technology, several econd. generation standard. have been proposed for improved service in lite United Stales, Europe, and Japan. These pes standards all employ digital modulation methods and provide better quality service and more efficient use of the radio spectrum than analog systems. Digital. systems also provide more security, preventing eavesdropping through the possible use of encryption.

PCS systems in the United States use either the IS-136 time division multiple access (IDMA) standard the 18-95 code division multiple access (COMA) standard, or the Eu ropean Global System Mo bile (GSM) system [11., [2]., [4 J. Many of the new PCS systems have been deployed using the same frequency bands as the AMPS system. This approach Lakes advantage of existing infra rrucrure, and facilitates the LIse of dual-mode handsets thai can operate on both the older AMPS system as well (lone of the newer digital pes systems. Additional spectrum has also been allocated by the Federal Communications Conunisaion (FCC) around 1.8 GHz, aod some of the newer PCS systems use this frequency band.

Outside the United States, the Global System Mobile (GSM) TDMA system is the most widespread, being used in over 100 countries [1:1. The uniformity of a single wireless telephone standard through ut Europe and much of Asia allows travelers to nse a single handset throughout these regions. In contrast, the different pes sy terns in (he United States are incompatible. Table 1.2 lists the major cellular and PCS telephone systems lhal nave been deployed throughout the world 1'1], [4].

It is interesting to compare how the development of first and second generation cellular services has differed in the United States and Europe [1]. The first V.£. cellular' SLeOl, A1v.1PS, provided a single standard allowing every cellular user in the United tales and Canada to communicate within range of a base station. In the Europe of the early 1980s. however, individual countries developed their own analog cellular standards with different frequency bands and modulation methods, so that there were at least fOLlY incompatible systems in use (see Table 1.2). These situations were reversed for second generation digital system . The organization of European countries under the European Union in the 19808

led to the establishment of GSM as a si rigle digital PCS standard which is now used by over 100 countries in Europe and elsewhere. In the United States, however, government poli ·ie. relati ng to the allocation of radio spectrum, as well as the structure of the teJeC011l11l unicati ons industry and the competitive nature of R&D in the United Slates, has allowed thetechnologleal and economic trade-offs between COMA, TDMA. and GSM pes systems to be decided in the marketplace, Meanwhile, wireless telephone consumers in the United States are left to choose between an out-of-date analog system and a variety of incompatible digital systems.

6 Chapter 1: Introduction to Wireless Systems

Satellite Systems for Wireless Voice and Data

The key advantage of satellite systems is that a relatively small number of satellite. can provide coverage to wireless users at any location, including the oceans, deserts, and mountains-areas for which it would otherwise be difficult to provide service. In principle, as few as three geosynchronous satellites can provide complete global coverage, but (as we will see in Chapter 4) the very high altitude of the geosynchronous orbit makes it difficult to communicate with handheld terminals because of very lowsignal strength. Satellites in lower orbits can provide usable levels of signal power, but many 1110re satellites are then needed to provide global coverage.

There are a large number of commercial satellite systems eilher currently in LIse, or in the development stage, for wireless communications. The e systems generally operate at frequencies above I GHz because of available spectrum. the possibility of high data rates, and me fact that such frequencies easily pass through the atmosphere and ionosphere. GEO satellite systems. such as TNl'v1ARSAT and MSAT, provide voice and low-data rate communications to users with 12" to 18"/ antennas. These systems are often referred to as very small aperture terminals (VSATs), and in [997 were being deployed al the rate of about 1500 per month to business users [I]. Other satellite systems operate in medium or low-earth orbits to provide mobile telephone and data service to IJserS.01l a worldwide basis.

Iridium, financed by L\ consortium of companies headed by Motorola, was the first commercial satellite system to offer handheld wireless telephone service. It consisted of 66 LEO satellites In near-polar orbits, and connects mobile phone and paging subscribers to the public telephone system through a series of iatersatellite relay links and land-based gateway terminals. The Iridium system cost was approximately $3.4 B. and it began service in 199ft Globalstar, proposed by L ral and Qualcomm, is another LEO satellite sy [em intended for wireles: telephone, fax and paging. This system uses 48 satellites to provide global coverage, and became operational in 2000. One drawback of using satellites for telephone service is that weak signal levels requ ire a Iine-of-sight path from the mobile user to the satellite. This means that satellite telephones generally cannot be used in buildings, automobiles, or even in many wooded or urban areas (the topics of propagation, fading, and .link loss, which relate to this problem, will be studied in Chapter 4). This places satellite phone service at a definite performance disadvantage relative to land-based cellular and pes wireless phone service. But an even greater problem with satellite phone ervice is the expense of deploying and maintaining a large fleet of LEO satellites, making it very difficult LO compete economically with land-based cellular or PCS·ervice. TIle typical co. I of acellular or pes call is in the range of $0.10 to $0.20 per minute, whi Ie in ]999 the estimated cost of a call placed through the Iridi urn or Globalstar satellite was about $2.00 per minute. In addition, the cost of a cellular or pes handset to Dew subscribers is usually minimal (or zero), wbile the cost of a satellite handset is several thousand dollars. For these reasons, it is hard to see how satellite telephone service can compete with land-based cellularand pes systems in terms of either performance or cost, even though satellite systems offer (in principle) the convenience of a single phone that can be used anywhere in the world. Table l . summarizes some of the current c mrnercial voice-communication satelli e systems.

1.1 Wireless Systems and Markets 7

TABLE 1.3 Commercial Wireless Satellite Systems

Number Of Operational
System Organization Satellites Orbit Dale
INMARSAT-M lnmarsat 5 OEQ 1996
Iridium Motorola 66 LEO J998
Globalstar LoraJ. Qualcomm 48 LEO 2000
I o Global Hughes 10 MEO 2000
Odyssey TRW 12 MEO 2000 111 August 1999 both Iridium LLC and the ICO Global Communications companies declared bankruptcy. It remains to be seen whether Globalstar and the other large LBG systems wiJI be financially viable but the future of such satellite services does not look promising when land-based systems offer better performance at lower costs. A satellite from the Globalstar system is shown io Figure 1 .1.

Global Positioning Satellite System

The Global Positioning Satellite system (OPS) uses 24 satellites in medium earth orbits to provide accurate position information (latitude, longitude, and elevation) to users on land, in the air, or at sea. Originally developed as the NAVSTAR system by the military, at a CO$t of about $12B, GPS has quickly become one of the In t pervasive applications of wireless technology tor consumers and businesses throughout the world. Today, GPS receivers can be

FIGURE 1.1 An artist's conception of one ofthe sateilires used ill the Globalstar satellite telephone

--------- system, (Courtesy ofF. Dietrich, Glcbalsrar, San Diego, CAl

8 Chapter 1: Introduction to Wireless Systems

FIGURE 1.2 Photograph of a NAVSTAR global positioning system satellite, showing the solar


punels and. the Lvband helix transmitting antennas. (Courtesy of Satellite and Space

Division, Rockwell International. Seal Beach, CA.)

found on commercial and private airplanes. boats and ships, and ground vehicles. Advances in technology have led to substantial reductions in size and cost, so that small handheld GPS receivers can be used by hikers and sportsmen. With differential OIlS, accuracies on the order of I em can be achieved-sa capability that has revolutionized the surveying industry L5]. A photograph of a NAVSTAR GPS satellite is shown in Figure 1.2.

The GPS positioning system operates by using triangulation with a minimum of four satellites. GPS satellites are in orbits 20,200 km above the Earth. with orbital periods of .12 hours. Distances from the user's receiver to these sate] lites are fou nd by ti rning the propagation delay between the satellites and the, receiver. The positions of the satellites (ephemeris) are known to very high accuracy; in addition, each satellite contai ns an extremely.accurate dock to provide a unique set of liming gut es. A GPS receiver decodes this timing information and performs the necessary calculations in order to find the positi J) and velocity of the receiver, The GPS receiver must have a line-of-sight view to at least four satellites in the GPS constellation, although three satellites are adequate if altitude position is known (as in the case of ships at sea). Because of the low gain antennas required for operation, the received signal level from a G PS satellite is very low--typicalJy on the order of -130 dBm (for a receiver antenna gainof (l dB). This signal level is usually below the noise power at the receiver, but spread spectrum techniques are 'Used to improve the received signal-to-noise ratio,

GPS operates at two frequency bands: Ll , at 1575.42 MHz; and L2, at 1227.60 MHz, transmitting spread pectrum : ignals with binary phase hift keying modulation. The LI

1.1 Wireless Systems and Markets 9

frequency is used to transmit ephemeris data for each satellite, as well as timing codes, which are available to any commercial or public user. This mode of operation is referred to as the Course/Acquisition eelA) code. In Contrast, the L2 frequency is reserved for military use and uses an encrypted timing code referred to as tile Protected (P) code (there is also a P code signal transmitted at the L 1 frequency). The P code offers much higher accuracy than the CIA code, and it is likely that this capabili ty will soon be made available to all users,

The typical accuracy that can be achieved with an Ll GPS receiver is about 100ft.

Accuracy is limited by timingerrors ill the clocks on the satellites and the receiver, as well as some error in the assumed position 01' the OPS satellites. The most significant error is generally caused by atmospheric and ionospheric effects, which introduce small but variable delays in signal propagation from the satellite to the receiver. Much better accuracies can be obtained through the use of differential GPS. which uses a GPS receiver at a known location to provide error correction information to other nearby GPS receivers. ill this way, positioning accuracies to within 1 em can be obtained relative to the reference position. Recei vers [hat have access to the P code can use the encrypted timing data at the L 1 and L2 frequencies to correct for the atmospheric and ionospheric propagation delays and thereby yield very accurate position Information.

Wireless Local Area Networks

Wireless local area networks (WLANs) provide connections between computers over short distances. Typical indoor applications may be in hospitals, office buildings, and factories, where coverage distances are usually less than a few hundred feet. 'Outdoors, in the absence of obstruction' and with the use of high gain antennas, ranges up to a few miles can be obtained. Wireless networks are especially useful when itis impossible or prohibitively expensive to place wiring in OJ' between buildings. or when only temporary access is needed between compu ters .. Mo bi le computers users, OfCOllTSe, can only be connected to a computer network by a wireless link.

In pite of their attractiveness. market penetration of WLAN products ha been . low, probably due to a combination of factors that include relatively high costs, relatively slow data rates, and poor immunity to fading and interference. In 1996 the market for WLANs was about $200M, which is a negligible fraction of the several billion dollar cellular telephoneindustry, It is expected, however that market growth for WLANs. will soon increase substantially. A major Dew WLAN initiative is the Bluetooth standard, where very mall and inexpensive RF transceivers will be used to link a wide variety of digital systems over relatively short distances, Possible Bluetooth applications include wirelessly networking printers. scanners, cell phones, notebook and desktop computers, personal digital assistants (PDAs). and even household appli ances, C urren l' Bl uetooth systems operate in the JSM band at 2.4 GHz, and offer data rates up to 1 Mbps, Market projections for Bluetooth devices are in the range of several hundred million units per year.

Currently most commercial WLAN products in the United States operate in the Industrial, Scientific, and Medica 1 (ISM) frequency bands, and LIse either frequency- hopping or direct-sequence spread spectrum rechnlquesin accordance with lEEE Standard 802.11. Maximum bit rates range from 1 to 2 Mbps, which are much slower than the data rates that can be achieved with wired Ethernet lines. WLANs almost universally use Internet protocols (TCPfIP) for communication between computers. In Europe. the HIPERLAN standard provides for WLAN operation with data rates up to 20 Mbps.

Other Wireless Systems

Besides the wireless systems described above there are many other applications of wireless technology. Wireless local loop (WLL} is similar to a cellular telephone system,

10 Chapter 1: Introduction to Wireless Systems

but provides service Over a smaller operating area. Hospitals, college campuses, factories, and 0 lfice buildings tan ernpl oy a WLL as a priva te brC//1{;!1 exchange (PB X) to provide users with a single telephone number and a mobile handsel with which they can communicate from any point within the operating area. The cell sizes 11· r WL typically range from 50 to 100 ft and for this reason WLL is sometimes referred to as microcellular phone service. An interesting application of WLLs is to provide service for towns and villages in lesser developed countries that do not have wi:red telephone service, since installing a wireless local loop system i much more economical than installing hard-wired copper lines. For these reasons. the demand for WLL products is expected to grow rapidly in the next few years. with over 60 million WLL users predicted by the year 2000

The Direct Broadcast Satellite {DBS) system provides television service (rom two geosynchronous satellites directly to horne users with a relatively small 18" diameter antenna. Previous to this development satellite TV service required all unsightly dish antenna as large as 6 ft in diameter. As we will see in Chapters 4 and 9, this advancement was made possible through the use of digital modulation techniques, which reduce the necessary received signal levels as compared to previous systems. which used analog modulation. The DBS system uses quadrature phase shij1 keying .(QPSK) with digi tal multiplexing and error correction to deliver digital data at a rate of 40 Mbps, Two satellites, DBS-l and DBS-2, located at 101.2° and 100.8° longitude, each provide 16 channels Wilhl20 W f radiated power per channel. These satellites use opposite circular polarizations to minimize loss due to precipitation. and to avoid interference with each other (polarization duplexing), DBS-l transmits with left-hand circular pol ari zation (LHCP), wbiJeDBS-2 uses right-hand circular polarization (RHCP).

DBS competes directly with wired cable TV service, but within one year of its introduction in 1994, DBS sold over I million units to break all previous records and become the consumer electronics product with the fastest market growtb in hi tory. The initial cost of a DBS antenna and receiver was about $700. but after 2.5, mil Lion un i ts were sold the price had dropped to about halfthis value. This cost reduction was the result of market competition, as well as significant economies of scale associated with large volume production rates

hundreds of thousands per month).

Local Multlpoim Distrlbutlon Systems (LMDS) and Multipoin: Multichannel Distribution Systems (!vIMDS) provide broadband wireless connections between a fixed base station and a cellular region of users. These systems are poised for rapid market growth because of the. trong demand for the 'last mile connectivity, where wireles systems offer one of the few economical solutions to the problem of providing high data rate connections to small. businesses and homes for Internet access, telephone. television, and data communications, LMDS and MMDS systems typically operate in the 2.1-2.7 GHz band, the 3.4-3.7 GHz band, or the 28 GHz millimeter wave band, and may offer two-way full-duplex data rates ranging from 50 Mbps to over 110 Mbps for each channel. These systems are sometimes referred to as broadband fixed wireless, because they are intended for connections between fixed, as opposed to mobile, users. Figure l.3 shows it commercial MMDS subscriber system.

Point-to-poinr radios are used by businesses to provide dedicated data connections between two points. Electric utility companies use point-to-point radios for tile transmission of telemetry information for the generation, transmission. and distribution of electric power between power stations and substations. Point-to-point radios are also used to connect cellular base stations to the public switched telephone network, and are generally much cheaper than running high-bandwidth coaxial or fiber-optic lines below ground. Such radios usually operate in the 18, 24, or 38 GHz bands, and use a variety of digital modulation methods to provide data rates in excess ofl 0 Mbp . High gain antennas are typically used to minimize power requirements and avoid interference with other users.

1.2 Design and Performance Issues 11

FIGURE 1.3 Photograph of the subscriber antenna and outdoor unit of an MI\1DS system operating


at2-A-2..6 GHz. prr viding a data rate of 20 Mb/sec, (Courtesy ol'N, Herscovici, Spike

Technologies, Nashua NH.)

Radio frequency identification (RFID) systems are used for inventory cracking, shipping. toll collection, personal security access, and other functions. Most expres's delivery services, for example, use handheld terminals that scan bar codes on packages and relay information to a central station. As another example, available now in 'several cities, automatic toll collection (A.TC) LISCS a small transponder in an automobile that can be interrogated by ali RP system mounted at the entrance to a highway or bridge. The transponder provides the vehicle's account number, which is then debited, and a monthly bi.11 sent to the driver. RFlD iysterns are much more specialized than cellular or WLAN systems, and use a wide range of modulation methods, operating frequencies, and duplexing schemes. It is expected that the market for R lD systems will reach $1.5 B by the year 2000,



In this section we discuss general considerations related to the design and performance of wireless systems. These include the choke of RF frequency, duplexing and multiple access methods, and a brief mention of some of the problems associated with propagation through the wireless channel. We will also discuss the differences between cornrnunication

12 Chapter 1: Introduction to Wireless Systems

using a circuit-switched and a packet-switched system, and possible health hazards associated with radiated RP power.

Choice of Operating Frequency

One of the first decisions that must be made during the design of a wireless system is the operating freq uency. The c ho iceof a transmit or recei ve freq uency is never completely free. as only small sections of the RF spectrum are available for specific applications. As listed in Appendix A, large portions ofthe spectrum. are allocated to AM radio (550 kHz-I.G MHz), PM radio (88-106 MHz), broadcast TV (54-88 MHz and 174-806 MHz). and a multitude of radio chan nels for airport, police, {lie, CB, amateur, and other users. I n the United Sta res, the Federal Communications Commission (PCC) is responsible for assigning frequency spectrum tQ competing-users. A." lisrad previously in Table 1.1, frequency bands have been reserved for cellular and pes telephonesys terns, GPS, DB S, poi n r- to-poi n t rad i os, and other maj 01' wireless appl ication S. An irnportan 1 category is the Industrial, Scientific, and M edical (ISM) bands, which reserve three microwave frequency bands for a variety of uses not covered II nder other spectrum allocations. The ISM bands are used for WLAN s, microwave ovens, Rt<"lD systems, and medical treatments using microwave power. For this reason, systems operating in the ISM bands are limited to a maximum of 1 W of radiated .power.

Besides the availability of spectrum, other important factors influenced by the choice of operating frequencyinclude noise, antenna gain, bandwidth, and cost. Noise power, for example, increases sharply ,It frequencies below 100 MHz due to a variety of sources that include lightning, ionospheric ducting, and interference from engine ignitions. and other electrical equipmen t, At frequenci es above 10 GHz, however, noise power steadily increases due to thermal noise of the aunosphereand interstellar radiation. Noise sources and noise effects arediscussed ill furtherdetail in Chapters 3 and 4.

As we will see in Chapter 4, the gain of an an tenna increases with frequency, fOT a fixed antenna size, Thus the use of higher frequencies. is an advantage for point-to-pointwireless systems where high antenna gain is required, as the resulting antenna will be smaller and less obtrusive. Higher gain 'antennas also receive less noise power frorn the surrounding environment.

In Chapter 9 we will see that the maximum data rate of a communications channel is determined by the available bandwidth, when noise is present". Thus a wireless system capable of high data rates will require a correspondingly high RF bandwidth, and. this is easier to obtain at high frequencies than at low frequencies. For example. for a moduiation method having a spectral efficiency of I bit per second per Hertz;a J Mbps data rate requires ] MIIz of bandwidth. This bandwidth could be obtained with a frequency band from 100 to 10 1 MHz, or from 10.000 to 10.001 GHz-the lower frequency band requires 1% fractional bandwidth, while the higher frequency band requires only 0.01 % fractional bandwidth.

While most of the preceding considerations argue for the use of H high operating frequency, there are points working in favor of lower frequencies as well. One is that the efficiency of RF transistors decreases with frequency, which increases the prime power required to operate wireless transmitters and receivers. This is especially true at millimeter wave frequencies, where active device efficiencies can be as low as 30%. Tn addi ti OJ] , component cost generallyincreases with operating frequency, so 'it is much mare economical to build an RF subsystem 0.1 lrequencies below 1 GHz limn at higher frequencies.

Finally, electromagnetic propagation characteristics vary considerably with frequency.

Electromagnetic signa Is at freq uencies above a few gigahertz propagate largely in straight line paths. ttl us, requiring an unobstructed line-of-sight path between a wireless trans mi uer and receiver. At lower frequencies, however, signals can more easily pass through or around obstructions such as foliage, buildiugs.vand vehicles. Thus lower frequencies give better propagationcharacteristics for wireless applications such as cellular and pes telephone

'.2 Design and Performance Issues 13

systems, while higher frequencies may be perfectlyadequate for point-to-point radios and satellite systems. As a-rough estimate, operating range decreases by 5% LO 10% as frequency increases from 900 MHz to 2.4 GHz. and another LO% at 5 GHz.

Multiple Access and Duplexing

Because frequency spectrum is limited.vand it is usually desired to accommodate as many simultaneous users as possible, several methods have been proposed for increasing the capacityof wireless ell-a nnels, One such mu liiple access method is to divide the available frequency range into many narrow frequency bands. This is caIledji'equenc), division multiple access (FDMA). The j)"lviPS telephone system, for example, uses FD"MA., dividing the 25 MHz mobile receive (869-894 MHz) and transmit (824-849 MHz) bands each into 833 channels of 30 k.fu bandwidth. Another method is time division.multip!« access (TDMA), where voice or data is transmitted and received over a shared frequency band only during preassigned time intervals of very short duration, and interlea ved wi til voice or data segments from other users, TDMA U1US multiplies the num bel" of users that call be accommodated wi th a single channel, but requires critical rinring and range information coordinated from a central station. In practice, TOM Ai s often com bined wi tb frequency divi sion dup lexing to allow several users for each of several frequency bands. The third popular multiple access method is code division multiple access (CDMA..) .. CD1\1A ise spread spectrum technique, whereby the relatively narrowband signal from each user is spread out in frequency using a unique spreading code. Several hundred signal>; can then occupy the same frequency band, arid yet be individually recovered at the receiver with knowledge of the original spreading code.

As mentioned earlier, full-duplex wireless communication requires a duplexing method [0 prov ide trausmi t and recei ve channelsthat do not.interfere witheach other. Because of the high sen si ti vity of most wireless receivers, the isol ation between transmi trer and receiver is typically required [0 be on the order of 120 dB. As a practical matter, this much isolation cannot he obtained unless frequency division duplexing is used, witli separate frequencies for transmit and receive. A bandpass .filter at the input to the receiver can then be used to attenuate transmitter signals. Often i.t is convenient to use a single antenna for both transmit and receive, in which case a duplexing filter is used to pass receive frequencies from the an tenna to the receiver, and transmitfrequencies from the transmitter to tile antenna, while providing enough attenuation between the transmit and receive bands to. achi eve the necessary isolation. A serious drawback of duplexing filters ... however. is that tile}, generally have several dB of insertion loss. This I eads to the .1 ass of transmit power, and i ncreases the noise figure of the receiver. A commercial dnplexing filter is shown in Figure l.4.

In half-duplex wireless systems, as used in many TDMA telephones and wireless LANs, duplexing can be accomplished by using a ITans7nir!receive(T/R) switch. This allows. a single antenna to be rapidly switched between the transmitter and the receiver at the appropriate times. Electronic RF switches generally provide more [han enough isolation in their off state to protect (he receiver from high transmit signal levels,

CIrcuit Switching versus Packet Switching

Both hard-wired and wireless (cellular and PCS) telephone systems are based on centralized. networks that provide a direct physical circuit between the communicating P,U"lies forme duration of the call. Thisis referred to as a circuit-switching 11 C!.lWCIrk. 111e circuit-swi tched telephone network has proven to be extremely reliable for voice cornmunications, with a very high·quality ofservice (QoS). Circuit-switched communication sy stems are i neffici en l, however, when used for transm itting data that occurs in bursts. such as computer data, email. and telemetry data. because the physical circuit is not fully utilized. In these cases,pClcke.t-n",'itched networks are preferred.In a packet-switched network.

14 Chapter 1: Introduction to Wireless Systems

FIGURE 1.4 Photograph of a dual-band diplexer (top), a hybrid coupler (bottom left) and a


two-way power divider (bouom Tight), These components operate over the SOG--

2200 MHz frequency band, providing coverage of both AMPS and PCS bands in a single component. (Courtesy of Sage Laboratories, Natick, MA.)

interconnected routers are used to provide multiple paths between any two points in the network. Messages and data are divided into packets of fixed length that are independently routed through the network from the sender to the receiver. In thi way, messages and data can be multiplexed over various paths through the ne-twork, which provides efficient and robust communication links without tying lip channel capacity unnecessarily,

The Internet is the most prevalent packet-switched network, and is used extensively for data, email, and multimedia communication between computers. While it is possible to use pscketswitching for voice communication, the fact that packet switching does not guarantee even a minimum quality of service means that Internet telephone calls often suffer from annoying delays and broken conversations. Newer protocols and standards, however, should improve this situation by implementing packet switching with priority levels that can be used for time-critical connections, such as for voice and real-time video links, In time, we can expect the majority of voice, video, and data communications to take place over packet-switched networks.


Wireless communication is made possible by the fact that electromagnetic waves call propagate through pace without the need for connecting wires or other conductors. We wilJ see that in free space the power density of an electromagnetic wave radiated by an an tenn a decreases as 11 R2. This simple model, perhaps augmented with a factor to aCCOWll for atmospheric attenuation. is usually adequate for line-of-sight (LOS) radio Jinks, such as point-to-point radios and satellite communications links. In other cases, such as cellular radio in urban environments or mobile radios in moving vehicles, the phenomena of electromagnetic energy propagation is much more comphcared. Effects such as reflections from the ground, buildings, and vehicles, as well as shadowing from natural and man-made

1.2 Design and Performance Issues 15

obstructions, can cause rapid variations in theamplitude of the recei ved signal over relatively short distances or rime intervals. These effects are referred to es fading, and are primarily due to the presence of In re than one possible propagation path between the transmitter and receiver. Because different propagation paths generally have different phase (or time) delays, (he superposition of signals at the receiver will involve constructive and destructive interference, leading to sharp variations in amplitude as much as 20 dB.

The large variation in received signal strength caused by fading is one of the most formidable problems facing the designer of a wireless system. Fading leads to decreased range, lower data rates. and decreased reliability and quality of service. Many of the most sophisticated techniques used in wireless communications have been devel ped primarily ill an attemp to alleviate the degrading effects of fading. hese include .pread spectrum techniques, the lise of antenna diversity, sophisticated modulation methods. and error-correcting codes. In all cases, such techniques increase the cost and complexity of the wireless system.

Radiated Power and Safety

Safety ls a legitimate concern of users of wireless equipment, particularly in regard to possible hazards caused by radiated electromagnetic fields. The body absorbs RP and microwave energy and converts it to heal; as in the GHSe of a microwave oven, th.is healing occurs within the body, and may not be felt at low power levels. Such heating is DlOSt dangerous in the brain, eyes, genitals, and stomach organs. Excessive radiation can cause cataracts, Cancer, or sterility, For this reason it is important [0 define a safe radiation level standard, so that user of wireless equipment are not exposed [0 harmful power levels.

The most recent U.S. safety standard for human exposure to electromagnetic radiation is given by ANSI/IEEE Standard C9S.1-1992. ]J1 the RF-microwave frequency range of j 00 MHz to 300 Gl-lz, exposure Ii mi t.s are set all the power density (i n Watts/crn-) as a function of frequency, as shown in Figure 1.5. The recommended safe power density limit is as low as O.2mW/crn2 at the lower end of this frequency range, because fields penetrate the body more easily at low frequencies. At frequencies above 15 GHz the power density limit rises to 10m W Icm"2, since most of the power absorption at these freq uencies occurs near


- / /
- V
- /
- ~
~E 10.0
'" 1.0
0... 0_1 0.1


] o.

Frequency (OHz)



FIGURE 1.5 LEEE Standard C9S. J -1991 recommended power density limits for human exposure


to RF and microwave electromagnetic field ...

16 Chapter 1: Introduction to Wireless Systems

the skin surface. By comparison, the sun radiates a power density as high as 100 mW/cm2 on a dear day, but the effect of rhis radiation is much less severe than a corresponding level of microwave frequency radiation because the sun heats the outside of the body, with much of the generated heat being reabsorbed by the air, while microwave power beats imide the body. At frequencies below ]00 MHz, electric and magnetic fields interact with the body differently than at higher frequencies, and so separate limits are given for field components at these frequencies.

In addition to the above power density limits, the FCC sets limits on the total radiated power of some specific wireless equipment. Vehicle-mounted cellular phones (using an external antenna) are limited to a maximum radiated power of 3 W. For handheld cellular phones, the FCC has set an exclusionary power level of 0.76 W below which phones are exempt from the A£\lSl standard on radiated power density. Most cellular phones radiate a maxirnumof Gji W, and newer PCS phones radiate even Ie s power. Cellular and PCS base stations are limited to a total.effective radiated power (see Chapter 4) of 500 W, depending on antenna height and location but most urban base stations radiate a maximum of IO W. Wireless products using the ISM bands are limited to a maximum radiated power of 1 W.

Whi le other ceumries have differen t (someti rnesl ower) standards for RF and microwave exposurelimi ts, most 'experts feel that the above lirni ts represent safe levels with a reasonable safety margin, Some researchers, however, are concerned that health hazards may OCCLL[ due to nonthermal effects oflong-terrn exposure to even low levels of microwave radiation.



Other Issues

Although tile above technical issues a:re criti cal to the perform ance of a wire less sy stem in fact the overriding consideration for the success of a given commercial wireless system is most often its cost. The cost of a system should of course include manufacturing and production costs, but also the cost of the infrastructure th at is necessary to su pport, operate and maintain the system. This may involve components such as base stations, antenna towers, satellite replacement costs, insurance, fees for right-of-ways for buried cables, technology licensing fees, advertising, billing, and nonrecoverable engineering co ts CNRE).

Many wireless devices are portable and operate from battery p wer. Battery life is a critical consideration for consumers, so it is important to design for (he minimization of prime power requirements through proper component selection, as well as design technique to minimize power consumption. These may include shutting down parts of the system when their function is not required, and lowering transmit power when possible.

Finally, it should be realized that consumers will expect a wireless system lO offer performance that is comparable to the w ired system that it replaces. For example.consumers will not find the convenience of a cellular phone to be worthwhile if sound quality is significantly worse than with a wired phone, or if conversations are often interrupted.

In this section we describe the basic block diagrams for tile RF stages of wireless transmitters and receivers, and provide an introductory discussion of the main RF and microwave components that are used in these systems. In later chapters we will discuss the operation and design of each of these components in much more detail, so the pllIpose here is simply to provide an initial broad view of the overall wireless system. In this way the reader will be able to see the larger context in which these individual components are used in practical wireless systems. Figure 1.6 shows a table of commonly used symbols hat are used ill block diagrams forRF and microwave components; symbols for filters are shown in Figure 1.7.

'.3 Introduction to Wireless System Components 17

COI11POllCIH Svrnbol

Cornpoueut N~l1le




Oscillut r

900 power divider

Frequency rnultlplier

Frequency divider


FLGURE 1.6 Block diagram symbols for commonly used RF and microwave cornponents. (Filter


symbols me shown in Figure 1.7.)

Basic Radio System

The RP stages of most wireless systems have a high degree of commonality, even though there may be many variations in practice, The block diagrams of a typical wireless transmitter and receiver are shown in Figures l.Sa.b, respectively.

The input LO a wireless transmitter may be voice, video. data, or other information to be transmitted to one or more distant receivers. These data are usually referred to as

(a) Low-pass filler

(h) Bandpass filler

F.IC URE 1.7 Symbols for Illters: (a) low-pass, (b) bandpass, (c) high-pass,



18 Chapter 1: Introduction to Wireless Systems

IF filter


filler amplifier Demodulator D:.lIa




FIGU RE 1.8 Block diagram of a basic radio system: (3) radio tran miner, (b) radio receiver.


the baseband ignal. The basic function of the transmitter is to modulate, or encode, the baseband information onto a high frequency sine wave carrier signal that can be radiated by the transmit an renna. The reaso 11 for this is that signals at higher frequencies can be racliated mare effectively, and use the RF spectrum more efficiently, than the direct radiation of the baseband Signal. The transmitter of Figure 1..8a operates by first 1I sing the baseband data to modulate anintermediat ine wave signal. As discussed in more detail in Chapter 9, there are many possible modulation methods. both analog and digital, that function by varying e' ther the amplitude. frequency, or phase of a sine wave. The output of the modulator is referred to as the intermediate frequency (IF) signal, and usually ranges between 10 and JOO :MHz. The IF signal is then shifted up in frequency, or upconverted, to the desired RF transmit frequency using a mixer. The mixer operate by producing the sum and difference of the input IF signal frequency and the frequency of a separate Local oscillator (LO). A bandpass .filter (BPF) allows the sum frequency to pass, while rejecting the much lower difference frequency. If necessary, a power amplifier is used to increase the output power of the transmitter. Finally, the antenna converts the modulated carrier Signal from the transmitter to a propagating electromagnetic plane wave.

The receiver of Figure L8b recovers the transmitted baseband data by essentially reversing tile functions of the transmitter components. The antenna receives electromagnetic waves radiated from many sources over: a relatively broad frequency range. A input bandpass filter provides some selectivity by filtering out received signals at undesired frequencies, and passing Signals within rhe desired frequency band. The bandpass filter is followed by a Low-noise amplifier (LNA), whose function is to amplify the possibly very weak received signal. while minimizing the noise power that is added to the received signal. Placing a bandpass filter before the LNA reduces the possibility that the sensitive amplifier will be overloaded by interfering signals of high power. Next, a mixer is used. to downconvert the received RF signal to a lower frequency signal, again called [he intermediatefrequency (IF). When the LO is et to a frequency near that of the RF input, the output difference frequency from the mixer will be relatively low (typically less than 100 MHz), and can be easily filtered

1.3 Introduction to Wireless System Components 19

by the IF bandpass filter. A high gain IF amplifier raise the power level of the signal so that the baseband information can be recovered easily. Thi process is calJed demodulation, and today is usually performed with digital signal processing (DSP) circuits. As discussed in more detail in Chapter J 0, [his type of receiver is known a a superheterodyne receiver. because it llsesjreq.leFlc.y conversion, converting the relatively high RF carrier frequency to a lower lf frequency before final demodulation.


As seen from the preceding discussion, the function of an antenna is to convert an RF signal from a transmitter to a propagaungelectromagnetic wave or, conversely, convert a propagating wave to an RF signal in a receiver.fn a transceiver, where a transmitter and a receiver are co-located for full-duplex communications. the same antenna may be used for both transmit and receive.

The aspects of antennas that are important for wireless systems are discussed in detail in Chapter 4, along with the characteristics of the propagation channel between the transmit and receive antennas. Some of the more obvious characteristics of an antenna include operating frequency range, size, and pattern coverage. The radiation pattern. ofan antenna is a plot of the transmitted or received signal strength Versus position around the antenna. H can be shown that the radiation pattern of an antenna i. the same for transmitring and receiving. Wireless systems that provide broadcast-type service. such as television and AMlFM radio. require antennas with pattern coverage that is uniform in all directions. Such patterns are called omnidirectional, and can be obtained with wire dipole and monopole ("whip") antennas, among others. Others systems. such as p int-to-point radio and DBS receivers, require antennas that radiate (or receive) power preferentially in one direction. The measure of the directionality of an antenna pattern is provided by the directivity, or gain, of the antenna, an omnidirectional antenna has low gain; while a highly directive antenna has high gain.

An important characteristic of all antennas is that there are unavoidable relationships between the operating frequency. size. and gain of an antenna 161-[1'1. Because of the nature of the electromagnetic operation of an antenna. effective radiation of a-signal requires that the antenna have minimum physical dimensions on the order of the electrical wavelength ().. = elf) at the operating frequency. This means that antenna size decreases with an increase in frequency, so that antennas at low frequencies will be very large, while antennas at microwave frequencies and higher may be very small. In addition, it can be shown that the gain of an antenna is proportional to its cross-sectional area divided by )..2 .so that higb antenna gain requ ires an electrically large antenna. Thus a low-gain antenna used for GPS at 1.575 GHz may be <IS small as a few square inches. whi le.a high gain parabolic dish antenna used in a point-to-point radio in the same frequency band may be several meters in diameter.

More sophisticated antennas are able to change the direction of their main beam electronically. Such antennas are called phased arrays, and ill the past have generally been limited to lise in military ystems because oJ their high cost. Phased array antenna technology, however, can be very usefu 1 in commercial wireless systems because the antenna beam can be directed at a given user, while avoiding interference from other users, Such systems. are called adaptive arrays, or omeiimes. 'mar! antennas. and may lead to increased channel capacity for cellular and pes telephone systems i f cost reductions can be achieved.


Filters are two-port components that are used to selectively Pi:"I!)S or reject Signals 011 the basis of frequency, An ideal low-pass filler (LPF) will pas' all frequency components below

20 Chapter 1: Introduction to Wireless Systems

its cutofffrequ.ehcy, while rejecting higher frequency components. Similarly, a high-pass filter (HPF) will pass frequency components above its cutoff frequency while rejecting lower frequencies. A bandpass filter (BPF) passes frequency components within a narrow passband. while rejecting frequency components outside the pa sband, Figure 1.7 shows two sets of block diagram symbols that are commonly used torepresentlow-pas " bandpass, and high-pass filters.

Filters are key components in all wireless transmitters and receivers. As can be seen from the block diagrams of Figures 1.8a,b,filters are used to reject interfering signals outside the operating band of receivers and transmitters, to reject unwanted products from the outputs of mixers and ampl ifiers, ~U1d to set Lhe IF bandwidth of receivers. Important filter parameters include the cutoff fre q uency, in sertion loss; and the ou t-of-band anenuauon rate, measured in dB per decade of frequency. Filters with sharper cutoffresponses provide more rejection of out-of-band. signals. Insertion loss. measured in dB, is the amount of attenuation seen by signals through the passband of the filter. Another important consideration is size and integrability with other circuit components. Today much of the front-end circuitry of receivers and transmitters in the heavily used frequency range from 800 MHz to 2 GHz can be menolithically integrated 'into a few integrated circuit packages. At the present time, however, it is not possible to construct high-performance bandpass filters in integrated circuit form. The inherent losse associated with RF and microwave integrated circuit leads to filters having relatively high insertion losses and 1 w out -of-band attenuation rate . For this reason, most wireless systems today use individual "off-chip" filters that are located




FIGURE 1.9 RF block diagrams for a 900 MHz GSM cellular telephone receiver and transmitter.


Each subsystem is highly integrated with commercial RF integrated circuits, but

note that the required bandpass fillers are not part of the integrated circuit packages: (a) receiver block diagram, (b) trunsmitter block diagram.

I •

'.3 Introduction to. Wireless System Components 21

an tile circui! board, rather than fully integrated filters. This results in a larger and more costly assembly, but critical filtering performance is optimized. Figure 1.9 shows the block diagrams fora commercial GSM telephone handset with arelatively high level of integration, where the necessary bandpass filters are separate from the integrated circuit packages.

There are many technologies available for the implementation of RF and microwave "filters [8], primarily dependent OIl frequency. In the frequency range from 800MHz roabout 4 GHz,. most bandpass filters today are made with dielectric resonators, which offer small size and high Q (sharp cutoff), with reasonable insertion loss. At IF 'frequencies (below 100 MHz) bandpass filters using quartz crystals or surface acoustic wave (SAW) devices are very common. SAW filters have very sharp cutoffs, but suffer from the disadvantage of insertion losses that may be as high as 20 dB. At higher microwave and millimeter wave freq uencies, wa vegu i de resonators are often used for bandpass filters. Low-pass filters used in wireless systems usually have less stringent requirements than do bandpass filters, and thus are often made with simple LC networks, parallel coupled lines, or transmission line stubs [8J.


There are three main categories of amplifiers used in wireless systems: low-noise amp I{[ie rs (LNAs) , used in the input stage of a receiver; power amplifiers (PAs), used in tile output stage of a transmitter; and IF amplijiers, used in the IF stages of both receivers and transmiuers.Important specifications [OJ: amplifiers include the power gain (in dB), the noise figure, and the insercept points. The noise figure of an amplifier is a measure of how much noise is added (0 the amplified signal by the amplitier circuitry. This is 1110st critical in the front end of a recei ver, where the i nputsignal level is very small. and. it is desired to minimize the noi se added by the recei ver circuitry. In additi on, as we will see in Chapter 3, the noi se power in a receiver is affected more by the first few components 'than by later components. Thus it is lmperati ve that the first amplifier ina receiver have as low a noise figure as possible.

Because transistors are non li near devices, transistor arnplif ers exhibi t nonlinear.effects.

Two important phenomenon that occur in amplifiers because of these effects are saturation and harmonic distortion .. At low signal levels the output power of an amplifier is linearly proportional to the input pOW12f. But because tile output voltage of an amplifier canna I exceed the bias voltage level, output power gradually reaches a saturation point as input power increases, Saturation is usually only an issue with power amplifiers ..

A more prevalent problem is related to the fact that harmonics of input signals are generated at the output of an amplifieriand in the case of III ul tip le input signal frequencies some of these harmonics will lie within the passband of the amplifier, These harmonics lead to signal distortion (harmonic distortion), Generally the power level of distortion harmonics is very low but, as shown in Chapter 3., the power level of some of these distortion products increases as the ell be of U1e input sign al level. The implication of this effecti s that distortion power can be significant even for input power levels well below the saturarion point of an amplifier, In practice it is often desi red to keep distortion leve Is ,)S low as 50 to 80 dB below the output signal level.

Advances in semiconductor technology have led to the development of RF amplifiers using inexpensive silicon (Si) transistors at frequencies up to several GHz. Previously gallium arsenide (GaAs) transistors were required for frequencies at OI above 1 GHz, but GaAs processing is very expensive and incompatible with silicon-based integrated circuit fabrication. This limits U1C level of integration that Can be achieved in a wireless system, and therefore increases cosl. Another semiconductor technologythat is very promising is silicon germanium (Sine), which can be used at higher frequencies than silicon, but with lower cost than gallium arsenide.

The design of transistor amplifiers is discussed in Chapter 6.

22 Chapter 1: Introduction to Wireless Systems


A mixer is a three-port component that ideally forms the sum and difference frequencie: from two sinusoidal inputs. This allows the important function offrequenc)' conversion to be performed in superheterodyne transmitters and receivers. In the case of the transmitter shown in Figure I .Sa, the modulated baseband signal is upconverted in frequency by mixing with a high-frequency local oscillator signal. In a superheterodyne receiver the received signal is downconverted in frequency by mixing with a local oscillator to produce a difference frequency (the IF frequency), In both cases filters are required to select the desired frequency products, while rejecting undesired frequencies that are produced as a by-product of the mixing operation.

In principle, frequency conversion can be accomplished with either nonlinear devices (diodes, transistors), or time-varying elements (switches). Modem mixers generally use diodes or transistors and produce many frequencies. based on the harmonics of the input signals and their combinations in addition to the desired-sum and difference frequencies. A passive mixer (one that uses diodes) always produces an output signal (IF) of less power than the input (RF) signal, because or dissipative losses in the mixer as well as inherent losses in the frequency conversion process. This loss is characterized by the mixer conversion loss, Mixers that use active components (e.g., transistors) generally have lower conversion loss, and may even have conversion gain. As in ihe ca e of amplifiers, harmonic distortion and noise are also important considerations in mixer performance.

In some receivers a low-noise amplifier may not be required if the received signal level is strong enough. This cost-saving measure re lilt in the mixer being the first component in the front end ( ecoodif a bandpass filter is used ahead of the mixer), Which means thar the noise and loss characteristics of the mixer will dominate the performance of the receiver, Another potemial problem with this approach is that power from the local oscillator may leak backwards through the mixer and be radiated by the receiver antenna. Such radiation must be minimized in order to avoid interference with other users and other systems, so there is of len a specification on the LO-Io-RF isolation for mixers. Active mixers generally have very good isolation, because transistors arc usually unilateral to a good degree. In addition, as described in more detail in Chapter 7, certain mixer circuits can yield very high isolation. Overall however, mixer design usually involves irade-offs between noise performance. isolation and conversion loss.


Oscillators are required in wireless receivers and transrniuers to provide frequency conversion, <1J1d to provide sinusoidal sources for modulation, Typical transmitters and receivers may each use as many as 4-6 oscillators. at frequencies ranging from several kilohertz to many gigahertz. Often these sources must be tunable over a set frequency range, and must provide very accurate output frequencies often to within a few parts per million).

The simplest oscillatoruses a transistor with an Lnetwork to control the frequency of oscillation. Frequency can be tuned by adjusting the values f the LC network, perhaps electronically with a varactor diode. Such oscillators are simple and inexpensive. but suffer "rom the fact that the output frequency is very susceptible to variations .JlJ supply voltage, changing load impedances, and temperature variations. Better frequency control can be obtained by rising a quartz crystal in place of the LC resonator, A crystalcontrolled oscillator (XCO) can provide a very accurate output frequency, especially if the crystal is in !I temperature controlled environment. Crystal osci llators, however, cannot easily be tuned in f-requency. A solution to this problem is provided by the phaselocked loop (PLL) , which uses a feedback control circuit and an accurate reference source


1 A Cellular Telephone Systems and Standards 23

(usually a crystal controlled oscillator) to provide an output that is tunable with very high accuracy.

Phase-locked loops and other circuits U1Ht provide accurate and tunable frequency outputs are called frequency syntheslrers. Virtually all modern wireless systems rely on frequency synthesizer circuits for the key stages of frequency conversion. Important parameters that characterize frequency synthesizers are tuning range, frequency switching time, frequency resolution.cost, and power consumption. Another very important parameter is the noise associatedwith the output spectrum of the synthesizer, in particular the phase noise. Phase noise is a measure of the sharpness of the frequency domain spectrum of an oscillator, and is critical for many modernwirelesssystems. Phase noise and oscillator design will be studied inmore detail in Chapter g.

Baseband Processing

Om main concern in this book is with the RF stages of transmitters and receivers, in contrast to the processing of signals between the IF stages and the baseband input or output data. Nevertheless, it is worthwhile to say a few words here about what happens to the signalin a receiver after the IF stage, in order to have a more complete view of the overall wireless system.

After down conversion (0 an IF signa] (which may occur in two or more stages), the received signal must be demodulated .. The majority of wireless systems today utilize coherent digital modulation methods (discussed in Chapter 9), for which demodulation requires a local oscillator synchronized in both frequency and phase with the down-converted earlier signal, These processes, called carrier acquisition and carrier synchronization. have traditionally been very difficult to implement, but the advent of powerful digital signal processing (DSP) chips allows these functions to be-performed easily and inexpensively. Demodulated baseband data can then be obtained from the output of the DSP stage, perhaps even includingerror correction. If the baseband information is analog, as hi the case of cellular telephones, the received digital data will be converted back to analog form with a digilad-Io·analog converter (DAC). Similarly, the transmission of baseband analog information would first involve conversion to digital form using a sampler and an analog-to-digital converter CADC). Framing and multiplexing functions may also be performed on the digital baseband data.


With the preceding introduction to wireless systems and some of the key components used in receivers and transmitters, we can now look at cellular telephone systems in more detail, Cellular telephony is the most significant, and perhaps complex, application of wireless technology, and 111 any of the topics discussed in later chapters are of direct importance to cellular systems.

Cellular and the Public Switched Telephone Network

As discussed earlier, cellular telephony represents by far the. largest commercial application of wireless technology. While there are many different standards and systems in worldwide use, all rely ot1 the concept of cellular coverage regions for frequency reuse, and rely on circuit-switched public telephone networks to transfer calls between users.

Figure J.1O shows how a geographical area can be covered with hexagonal-shaped cellular regions. Cellular telephone users within each cell are serviced by a base station at

24 Chapter 1: Introduction to Wireless Systems


coverage lirea

the center of that cell To avoid co-channel interference, adjacent cells are assigned different sets of channel frequencies. Frequencies can be reused at two different cells when there is at least one intervening cell with different frequency assignments. Frequency reuse is one of the key advantages of cellular radio systems because it permi ts more efficient utilization of valuable radio spectrum. This method is generally used for FDMA, TDMA. and CDMA multiple access systems. In the case ofCDMA, further interference suppression is obtained through the use of spread spectrum technique. (For marketing reasons. service providers often distinguish between "cellular telephone service" and PCS, but PCS still employs a cellular radio systern.)

In operation, a cellular telephone user communicates with the closest base station, even though j L is I ikely that an adjacent base station may receive a weaker signal from the same user. If the user is mobile. a hancl-oJlfrom one base station to the next will OCCLII when the received signal power from the closer base. tation becomes greater than the received signal power at the original base station. Ideally, this switchover occurs quickly and reliably and is not noticed by the user.

All the base stations within a given geographical area are connected to a mobile telephone switching offt 'e (MTSO). which typically can handle several thousand simultaneous telephone calls. The MTSO provides connections to the public switched telephone network (PSTN), as shown in Pigure 1.11. The PSTN includes high-capacity fiber-optic lines between cities, as well as transoceanic line. between countries. Local telephone exchange offices provide connections to individual private and business users, generally with twisred c pper wire pair. All cellular phone call thus are routed through the PSTN. even when both parties are using cellular phones. In some newer PCS systems, however. callers may be connected through the base station if they are within the same cell site.

FIG U RE 1.10 Layout of hexagonal cell areas !U1d base stations for cc 1I ular rad io sy stems.


AMPS Cellular Telephone System

Like other first generation analog cellular systems. the AMPS (Advanced Mobile Phone Service) system was based OD technology of the 1970$. While most other first generation

1.4 Cellular Telephone Systems and Standards 25


~!a Cellular

Q'fj handset


base stati on

Mobile telephone switching office CM'TSO)

Business customers

Fiber or radio link

Privll.tc branch exchange

exchange office


Local customers

FIGURE 1.11 Pi ctorial diagram show; ng the connection of a c-ell ular telephone and base station to the public switched telephone network.

systems have been replaced with digital cellular, for reasons discussed in Section 1.1 the U.S. AMPS system is only slowly being supplanted by newer technology,

The mobile transceiver in the Al\1PS system uses a transmit frequency in tbe range of 824-849 MH z.and a recei ve frequencyi n the range of 869-894 MHz.. Both of these bands are divided into 832 channels that are 30 kHz wide. The maximum frequency deviation of the frequency modulated (Fl\tl) signa] is 25 kHz, allowinga 5 kHz guard band on each channel [2], 14]. Since separate frequencies are used for tl'ansmjt arrd receive, full duplex operationis provided (an example of frequency di vision duplexing). Multiple users within a cell are assigned different transmit and receive channels (frequency division multiplexing). A typical ceU in the AMPS system has an area of about 10 sq uare miles.

Figure 1..12 shows the block diagram of a typical. AMPS transceiver. The receiver is dual-conversion, meaning that there are two stages of frequency down-conversion, with two mixers and two local oscillators. The desired received channel is selected at the first IF stage, which uses a phase-locked loop to provide the proper local oscillator frequency, The transmitter is single-conversion, but agam a phase-locked loop is used to provide the necessary carrier frequency .. The transmir-receive duplexer operates as a channel separation filter, passing the higher frequency band to the receiver circuitry while blocking the lower frequency transmitted signal; the converse operation is performed between the transmitter and antenna. The duplexer is usually a ceramic resonator filter, Early mobile transceivers used di screte components, and were large and had short battery life. Newer transceivers have most of the circui try of Figure I. J Z integrated with one or two chips .except for the f I ters,

Communication between a mobile cellular phone and the base station involves four separate simplex (one-way) channels, listed as follows:

FYC-forward voice channel (base to mobile) Rye-reverse voice channel (mobile to base)

26 Chapter 1: lntrcductlon tg Wire!essSy.stems


R:1:l69- LNA

894MHZ~. '~~ .

~. ~

~ r:»:»

1st rr '1\8 MHz

2nd If: IF

455 kHz amp


. demoo


12-15 MHz' XeD

FIGU I'm l.I2 B lock diagram 0 f· an AMPS mobile transceiver.

FCC-forward control channel (base to mobile) RCC-reverse control channel (mobile to base)

The two voice channels are used for voice transmission and reception, while the two control channels are used for data messages that control call initiation, termination, and hand-offs. Each base station uses a dedicated FCC and RCC for all users within its cell. When an AMPS cellular telephone is first turned on, it scans the group of preassigned system FCC channels, and determines which FCC signal is strongest. This FCC is assumed to belong to the closest base station When the signal strength of the FCC drops below a certain level, the mobile unit again scans the FCes to pick out a new base station, and re-establishes contact over a new FCe.

When a call is placed to a mobile telephone, the·MTSO sends out the request to every base station in the system, which broadcasts the called phone number over the Fees. If the mobile phone is on, and within the coverage area of the system, it responds by sending irs mobile identification number (MIN) over the ReC. The MTSO then sets u_p the can by having the base station assign an 1I111.1sed FVC and Rye, and sends a ring signal to thll' mobile phone. When acall is made from a mobile phone, a call initiation request is sent to the base station over the RCC, along with the MIN and the number being called. The MTSO makes the connection through the PSTN, and assigns voice channels with the base statiou These procedures typically occur ill a few seconds, Pmther details of cellular telephone protocols can be found in reference [4J.

Digita.1 Personal Communications System Standards

Second generation cellular telephone systems use digital modulation methods, with [he primary advantage of allowing more users in a given frequency bandwidth. Additional features may include lower transmit powers, longer battery life, the use of error correcting

References 27

TABLE 1,4 International Digital PCS System Standards

PCS System IS-54/IS-1 36 lS-95 GSM
Transmit Frequency (RVe) 824-849 MHz 82.4-849 MEiz 890-915 M E-lz.
Receive Frequency (PVC) 869-894 MI",\z 869-894 MHz 935-960 MHz
Duplexing Method FDD FDD FDD
Mu:ltiple Access Method TDMA COMA TDMA
Channel Bandwidth 30 kHz l.15 MHz. 200 kHz
Channel Bit Rate 48.6 kbps 1,228.8 kbps 270.833 kbps
Users per Channel 3 64 8
User Bit Rate 8 kbps 1.2-9.6 kbps L3 kbps
Number of U sers 2,496 15,960 992 codes for improved Quality of Service, and the possible use of encryption for privacy .. As in the case of first generation analog systems, there area multitude of competing systems proposed and in use for digital pes. In the United States, digital pes systems have initially been deployed using the same frequency bands as the AMPS system, but newer pes services are using a frequency band near 1900 MHz that was recently allocated by the FCC, and auctioned to service providers in 1995. In Europe and some other countries, second generation cellular services use frequency bands at either 900 MHz or 1800 MHz. Japan currently has several PCS services operating at various bands from 800 MHz to 1500 .!YUh and at 1.900 :MHz.

Specifications for the three most commonly used digital PCS standards are listed In Table] .4. Interim standalds (IS) are communications standards that have been agreed upon by members of the Telecommunications Industry Association. The two competing PCS standards in the US are IS-54, a TDMA system, and 18-95, which uses CDMA, (18- 136 is an upgraded version of 18-54.) The predominant pes system in Europe and many other countries is GSM, which uses TDMA technology. While first generation cellular systems were intended only for Yoke communications, most second generation systems tan provide some basic data services, such as paging, fax, and Low-data rate access to computer networks. Such services will be further enhanced when third generation personal communication systems are implemented.


[I] Nati onal Research Council, The Evolntion of U ntethered COmmunications, National Academy Press, Washington, D,C.. 1997.

[2] L. E. Larson, RF and Microwave Circutt Design for Wireless Cemmunlcatlons, Artech House, Dedham, MA, 1996.

[3] R. Schneiderman, Wireless Personal Communications. lEEBPress, Piscataway, NJ, 1994.

[4] T. S. Rappaport, Wire.less Communtcattons: Principles and Practice, Prentice Han, Englewood

Cliff, NI. 1996,

[5] I Hem, GPS-A Gli ide to the Next Utility, Trimble N avigati on. Sunnyvale, CA, 1989.

(6) C, A. Balanis, Antenna Theory: Analysis and Design, 2nd edition, Wiley .. New York, 1997,

[71 W L. Srutzman aadG, A Thiele, Antenna Theoryand Design, indedition. Wiley, New York, 1998. [8] D. M. POZH. Microwave Engineering, 2nd edition, WHey, New York. J998.

28 Chapter 1: Introduction to Wireless Systems


1.1 How many U.S. PCS/cellular telephone subscribers were there in the previous year? What was the revenue generated by this market. How many cellular users were there worldwide? Use these data in conjunction with the figures presented in the beginning of this chapter to estimate the number of" U,S, subscribers for the current and the following years.

1.2 Gather data on (he market size for wireless local area networks in the United Stales during the last five years. Plot this information. and extrapolate the curve to estimate the market ('orWLANs for the next three years,

1,.3 Consider the frequency spectrum fT(m150:MHz to 2 GHz. Using data [rom Appendix A on frequency allocations. find the percentage of this frequency band that has been freely allocated to broadcast FM radio and television ill the United Slates. Compare this lO the percentage that is allocated to wireless systems such as celiularJPCS telephone, OPS, paging, and the ISM bands. Write a short essay discussing your opinion of this situation, Should policies regulating the frequency spectrum be changed to better serve the public?

1.4 Esumate the amount of energy required to operate a typical cellular [cJepilone for OTIC minute of talk time. U' [his phone is charged for one hour each day by using a solar panel with an area of 6" x 6", estimate the amount of talk time that can be obtained daily. (Obtain reasonable estimates for the power cousumption of a typical cellular phone, the efficiency of solar cells, and the aile rage power density.of sunlight, and list your sources for this Information. Show your work, and discuss your assumption, .)

1.5 Research consumer satisfaction will] satellite-based telephones, ami compare with the satisfaction of ceJluJarlPCS subscribers. Consider monthly cos .s, availabil ity of service, and quality of service. Can you draw any conclusions as t the long-term outcome of land-based versus satellite-based wireless telephone service'!











Transmission Lines and Microwave Networks

Transmission lines are essential components LO modern wireless systems, being used to connect antennas to transmitters and receivers, for impedance matching in mixers and amplifiers, and as resonant elements in oscillators and filters. It is likely that the reader is already familiar with the fundamental topics of transmission line wave propagation, reflection, transmission, impedance transformation, and the Smith chart, but we will discuss this material here for 'completeness, and for those who may need a review of these topics. We also require some familiarity with S parameters, impedance matching techniques, and basic microwave network analysis for use in later chapters, and so these topics are also presented. Our treatment of these topics will be more than adequate for the purposes of this book, but thereader who wants to delve ·deeper into these topics can refer to references [l ]-[2J. Finally, we note that om presentation of nansmissicn 1 i nes and networks can be accomplished from a circuit model perspective, without recourse to electromagnetic analysis. The reader should realize, however, that transmission line tlreory ultimately is based on the rigorous application of electromagnetics.



The key difference between standard circuit analysis and transmission line theory is electrical size. Circuit analysis assumes that the physical dimensions of a network are much smaller than the electrical wavelength, while transmission lines may be a considerable fraction of a wavelength, or marty wavelengths, in size. Thus a transmission line is <I. distributed-parameter network, where voltages and currents canvaryin magnitude and phase over the length of the line. We begin our treatment of transmission lines with a Jumped circu it model 'for an incremental length of line, and then study the transmission and reflection of electric waves on the line.


30 Chapter 2: Transmission lines and Microwave Networks

i(z, f) -

------ 6.<:------ ..






FIGURE 2,1 Voltage-and current den nitions and the eq ui valent circuit for an incrementa I length of


transmission line. ea) Voltage and current definitions, (b) Lumped-elemenreqtrivalent


Lumped El:ement Model for a Transmission Line

As shown in Figure 2.1 a, a transmission line is of ton schematically represented as a two-wire line. since transmission lines usually consist of two parallel conductors. A short segrnente z of transmission line can be modeled as a lumped-element circuit, as shown in Figure 2.1 b, where R, L. G. and C are per unit length quantities defined as follows:

R = series resistance per unit length, for both conductors, in g/m. L = series i nductance per un it I ength, for both can ductal'S, in H/uJ. G = shunt conductances per unit length, in S/m.

C = shunt capacitance per unit length, in F/m.

The series inductance L represent,'; the total self-inductance of the two conductors, and the shunt capacitance C is due to the dose proximity of the two conductors. 111e series resistance R represents the resistance due to the finite conductivity of the conductors, and the shunt conductance 1S due to dielectric loss in the material between the conductors. R and G, therefore, represent loss. A finite length of transmission line can be viewed as a cascade of sections of the form of Pi gure 2.1 b.

From the circuit of Figure 2.t b. Kirchoff's voltage law can be applied to give

... . !Jl(z., n

v(z, I) - R6.u(z, l) - Ltlz-- - v(z + 6.:;:, r) = 0,

. ilt

(2.1 a)

and Kirchoff's current law leads t.o

. . i.1v(z + 6.z. t) .

I (z, t) - G L\_;:v(z + ..6.z, r) - C 6.z - I (z + /).z. t) = O.


(2. J b)

Dividing (2.I,a) and (2.tb) by 6.z and taking the limit as 8z ---+ o gives the following

2.1 Transmission Lines 31

differential equations for the voltageand current on the line:

ov(zt)_ h'( ) L~i(Z,I)

--- - -.1\{ Z. t - ---,

GZ fit

at (z, t) = -Gv(z I) _ C Bu(z, /) .

Bz; , at



These equations are the time-domain form of the transmission line, or telegrapher, equations. For the sinusoidal steady-state condition, with cosine-based phasors, (2.la) and (2.2b) simplify to


-- = -(R + jwL)/(z) dz

dl(z) = -(G + jwC)V(z). dz



The solution of these differential equations leads to traveling voltage and current waves on the transmission Iine.

Wave Propagation on a Transmission Line

Equations (2.3a) and(2.3b) can be solved simultaneously to give a single wave equation for either V(z) OT T(z), by eliminating either I (z) or \f(Z):




)I =0: + jf3 = JeR + jwL)(G + jo')C),


i the complex propagation constant. The imaginary part fJ of the complex propagation constant is called the phase .anstani, while the real part, 0:, is the attenuation constant. Note that the propagation constant is generally a function of frequency.

Traveling wave solutions to (2.4) can be found as

V(z) = Vo+ e-n + Vo- eYZ, I (z) = foTe-1/l'. + fn- eVe,

(2.6a) (2.6b)

where the e-Vl term represents wave propagation in the +z direct jon, and the el':' term represents wave propagation in the -z direction. Applying (2.3a) to the voltage of (2.6a) gives the current on the line:

Comparison with (2.6b) shows that if a characteristic impedance, Zo, is defined as


R+jwL G + jWC'


then the voltage and current on the transmission line call be related as

32 Chapter 2: Transmission Lines and Microwave Networks

vet -Vo-

+ =ZI1=--.

[0 10-

Then (2.6b) can be rewritten in the following form:

v+ v-

1(z) = _jJ__e-rz - ~eY~.

Z(j 20

Converting the phasor voltage of (2.6a) to the lime domain gives


v(z, t) = I VoTI cos(o.)/, - f.fz + ¢+)["-< + I Va-I cos(wt + fh + ¢-)e"", (2.9)

where e± is the phase angleofthe complex voltage Vl. The wavelength. of the traveling waves is defined as the distance between two successive points of equal phase all the wave ara fixed instant of time, whichis seen to be

(2. to)

The phase veiocitv of the Wave is defined as (he speed at which a constant phase point travels down the line, and is given by

dz w

v" = dE = Ii = 'Aj,


since w = 2IT!,

Lossless Transmlss.ion Lines

The preceding results apply to general transmission lines, including loss effects, and it is seen that the propagation constant and characteristic impedance are complex, In many. practical cases, however, the 10:58 of the Ii ne is very small and can be neglected, resulting ill a simplification of the above results, Thus, setting R = G = 0 i:n(2.S) gives the propagation constant as

y = a + if! = jW../LC.


f3 = ill"; LC,


Cl! =U.


As expected for thelossless case, the attenuation constant a is zero. The characteristic impedance Of (2.7} reduces to



which is now a real number. The general solutions for voltage and current on a lossless transmission line can then be written as

V(z) = Vite-J/Jt + VQ-eJ/iZ,

V+ 11-

fez) = _(_) e-J/h __ {_) eJfJz.

ZIJ z,



2.1 Transmission Lines 33

l1{z), I{z)

____ z_o,_fJ V_' ~ h


FIGRE 2.2 A transmission line terminated in a load impedance ZL.

The wavelength on the line is

21[ 2n


- {3 - (lh.lLE·


and the phase velocity on the line i.

t» I

v,; = "i = ,JT;E'


Terminated Transmission Lines

Figllre2.2 shows a lossless transmission line terminated in an arbitrary load impedance ZL. This problem will illustrate wave transmission and reflection on transmission Jines, which are fundamental properties of transmission line circuits.

Assume that an incident wave of the form Vo+ e-i1kis g nerated from a source at z < O. We have seen that the ratio of voltage to current for such a traveling wave is 20, the characteristic impedance. But when the line is terminated in an arbitrary load ZL =F Zo, the ratio of voltage to current at the load must equal 2/.. Thus; a reflected wave must be 'generated at the load with the appropriate amplitude to satisfy this condition, The total voltage on the line can then be written as in (2.14a), as a sum of incident and reflected voltage waves. Similarly, the total current on the line consists of incident and reflected waves, as described by (2.14b).

The total voltage and current at the load are related by the load impedance, so l1L Z = () we must have

Solving for VIi- gives

21 - 20

11.- - - .11.+

() - ZL + 20 o·

The amplitude of the reflected VOltage wave normalized to the amplitude of the incident voltage wave is defined as the voltage reflection coefficienr, I":


A current reflection coefficient. giving the normalized amplitude of the reflected current wave, can also be defined. But because such a current reflection coefficient is just the negative of the voltage reflection co fficient [as seen from (2,14)], we wi 11 avoid confusion by us ing only the voltage reflection coefficient in this book.

34 Chapter 2: Transmission Lines and Microwave Networks

The total voltage and current on the line can then be written using the voltage reflection coefficient as

v(.~) = Vo+Le-JP< + f(11tJz1,

~+ .

l(i} = _o_[rJ{I, - f'eJI'JI.].

'. 20' .

From theseequations it is seen that the voltage and current OIl the line consist of a superposition of an incident and reflected wave; such waves are called standing wuws. Only when r = 0 is there no reflected wave. To obtain r = 0, the load impedance Zl. must be equal to the characteristic impedance 20 of the line, as seen from (2.17). Such a load i then aid to be matched LO the line, since there is no reflection of the incident wave.

Now consider the time-average power fI w along the line at the point z:



I J 1 \1.+12 .'

Pa"g = '2 ReI V(z)I"(z)) = '2 +ORel] - r~e-2Jl'lz + re2JI'l< - ifl2l.

where (2. 1.8) has been used. The mi ddle two. terms j n the brackets are of the form A - A * = 2) Im [.4.), and so are purely imaginary. This simplifies the result to

1 1 \1,+ 11

P = -_. _O-(1-lrI2)

~vg 2 20 '


which shows thai the average power fiow hi Constant ttl any point all the line, and that the total power delivered to the load is equal to the Incident power (JVo+ e (220), minus the reflected power (W(t 12 W 12/220). If J' = 0, maximum power is delivered to the load. while 110 power is delivered for 11"1 = 1 (all incident power is reflected from the load). The preceding discussion assumes that the generator is matched. so that there is no. re-reflection of the reflected wave from the generator at z < 0 (the case of a mismatched generator will be treated later).

When the load is mismatched, then, no! all of the available power from the generator is delivered to the I ad. This "loss' is called the return los» (R L), and is defined (in dB) as

RL = -20logjrjdB,


so that a matched load (I' = 0) has :l return loss of 00 dB (no reflected power), whereas a total reflection (If I = 1) has a return loss of 0 dB (all incident power is reflected).

If the load is matched to the line, r = 0 and the magnitude of the voltage on the line is I V (z) I = I Vo+ I, which is a constant, For this reason such a line is sometimes said to be "flat," When the load is mismatched, however, the presence of a reflected wave leads to standing waves where the magnitude of' the voltage on the line is no! a constant, Thus, (rom (2.18a),


where e = -z is the positive distance measured from the load at ? = 0 back toward the generator (see Figure 2.2), and e is the phase of the reflectioncoefficient (F = Wlefo). This result shows that the voltage magnitude oscillates with position z along [he line. T11e maximum value occurs when the phase term e.!w-2.tJe) = I, and is given by


Similarly, the I11ll1ll11UI11 value of voltage magnitude occurs when the phase term eJ(/)-2{1f.) = -I ,aad is given by

Vmln = I ¥tt 10 - Ifl),


2.1 Transmission Unes 35

As I I' I increases, the ratio of Vln"" to Vmln increases, so a measure of the mi smatch of a line, called the standing wave ratio (SWR), can be defined as

SWR = v.nal< = 1 + I r I .

Vmin 1 - WI

This quantity is also known as the voltage standing wave ratio (VSWR). From (2.23) it is seen that the SWR is a real number such that: '] S SWR < 00, where SWR = 1 implies a matched load.

From (2.21), it is seen that the distance between two succes ive voltage maxima (or minima) is e = 2IT /2{3 = IT 'A/'br = A/2, while the distance between a maximum and a minimum is e = Tr/2J3 = )",/4, where A is the wavelength on the transmission line.

The reflecti all coeff cient of (2.17) was deft ned as the ratio of the reflected to the incident voltage wave amplitudes at the load (£ = 0), but the reflection coefficient can be generalized to .LOY poinr f ~ 0 on the line as follows. From (2. 14a), with z = -e, the ratio of the reflected voltage to the incident voltage is


V- e-JiJe

j(£) = 0 = r(0)e,-2.ifJ1

11+ 'fJ/!. . '

Vo e1


where 1"(0) is tile reflection coefficient at. z = 0, as given by (2.17). This form is u eful when transforming the effect of a load mismatch down the Line.

We have seen that the real power flow on the line is a constant, while the voltage amplitude, at least for a mismatched line, is oscillatory with position along the line. The perceptive reader may therefore conclude that the impedance " een looking into UIe line must vary with position, and this is indeed the case. At a distance Z = -z from the load, the input impedance seen looking toward the Load is given by

z. _ lI(-£') _ l1ol'[eJ13e+fe-if:U]_z l+fe-2J1Jr.

In - 1(-1:.) - Za Vo+ eJ/il _ fe-j{Jf - 0 J - rr2j,Bt'

where (2.ISa,b) have been used for V(-) and 1('(.). A more usable form of this Jesuit may be obtained by using (2.17) for r in (2,25):


eZL. + Zo)eJ/I£ + (ZL. - Zo)a-jPf.

Zill = Zo '1Jf 'pe

(ZL + Zo)e1 - (ZL - Zo)rJ

_ Z ZL cos f3l+ .i Zo sin {3£

- 0 Zo cos j3t + j ZL sin tJ.e

Z,_ + j 20 tan (3£

= 20 .

20 + J ZL tan {J,e


This is an important result giving the impedance at the input of a length of transmission line with an arbitrary load impedance. We will refer to this result as the transmission line impedance equation. Some special cases of this result will be considered next.


A load impedance of 130 + j 90 n terminates a 50 !J transmission line that is 0.3 )" long. Find the reflection coefficient at the load, the reflection coefficient at the input La the 1i11e, the SWR n the line, the return loss, and the impedance een at the input t the line.


The reflection coefficient at the load can be found from (2.17):

36 Chapter 2: Transmission Lines and Microwave Networks

- = ZL - Zo _ (130 + j90) - 50 = 0598 ')' 8 I - .. Ll.

ZI_+ZO (130+.190)+50

Equation (2.24) can be used to transform th is reflection coeffi c i ent dow n the J i ne to the input. using the fact that (j,C = (2rr I },,)(O.3?) = I 08~:

f'(.e) = r(O)e-2jjll = (CL598 L21.8o)e~2I(I()H') = 0598 Ll65.gc

The SWR on the line is given by (2.23):

SWR = I + WI = 1 -I- O.59B = 3.98 l-Ifl 1-0.598

The returnloss can be calculated using (2.20)

RL = -20 Log I I"] = -20 log(0.598) = 4.47 dB

Finally. the inpin impedance seen at the input to the Ii ne is found from (2.26):

21. + .I 20 tan ae (130 + .190) + j50 tan(.lOgO)

Zlll = 20 . = 50---=----:-::-----"c:--=---_:__,-------:..

20 + j ZL tan fie 50 + j( 130 + j90) tan( 108")

= 12.75+.i5.80


Special Cases of Terminated Transmission Lines

A number of special cases of lossless terminated transmission Jines are useful in practice, so i.e is helpful to consider the properties of such cases here.

Consider tirsr the transmission line circuit shown in Figure 2-.3, where a line is terminated in it short circuit, ZL =0. From (2.17) it is seen that the reflection coefficient for a short circuit is C = -1; it then follows from (2.23) that the standing wave ratio is infinite. Prom (2.18) the voltage and current on the short-circuited line can be written as

v (2) = ~t[e- jf1~ - eJfl,] = -2) Vo+ sin f3z,

V-r 2V+

I(Z) = +[e-JfJ~ + eN!,] = _0_ cos (3z.

z, 20

which shows that V = 0 at the load (as expected, for a short circuit), while the current is a maximum there. From (2.26), or the ratio V (-l)jl( -.e), rhei nput i rnpedance can be found




z., = j 20 tan (3C,




PIGURE 2.3 A transmission line terminated in a short circuit.

2.1 Transmission Lines 37


l{z)Zo 2V~'



FIG RE 2.4 (a) Voltage. (b) current, and c) impedance (Rill = 0 or ) variation along a short-

---------- circuited transmission line.

which is seen to be purely imaginary for any length. E. and to take on all values between -I-joo and -j . For example, when e = 0 we have Z'D = 0, but for e = )./4 we have Zin = +joo (open circuit). Equation (2.28) also shows that the impedance is periodic in e, repeating for multiples of },_./2. The voltage, current, and input reactance for the short-circuited line are plotted in Figure 2.4,

N ext consider the open-circu i ted li ne shown in Pi gure 2.5, where Z L = co. D ividi I1g the numerator and denomimi.torof(2.17) by ZL and allowing ZL -+ 00 shows that the reflection coefficient for this Case is r = I, and thus the standing wave ratio is again infinite, From



FIGURE 2.5 A transmission line terminated in an open circuit.


38 Chapter 2: Transmission Lines and Microwave Networks


1(;;)2(1 -2jVO+ )-



Xi" Zo


FIGURE 2.6 (a) Voltage, (b) current, and (c) impedance Rill = 0 or ) variation along <In OPCll--------- circuited transmission line.

(2.18) the voltage and current 011 the open-circuited line are

V(Z) = Vol·l.e-h~l + e.ilizJ = 2Vo~ cos/3z,

Vo+ .,,_ ',,_ ~2j Vo+ .

I(Z)=-[e-J"--e.ll'<]=· SlllfjZ,

z, z,

which shows that now J = 0 at the load. as expecred for an open. circuit, while the voltage is a maximum. The input impedance can be found from (2.26). or from the ratio It' (<.)11 (z), as

Z;n = - j Zo cot {Jl,




which is also purely imaginary for any lenglh,E. The voltage; current, and input reactance of the open-circuited line are plotted in Figure 2,6.

Finally. consider terminated transmission lines with some special lengths. For example, if e = ),,/2, 2.26) SllOWS that


meaning that a half-wavelength section (or any multiple of ),,/2) of transmission Iine does nol alter r transform the load impedance. regardless of the characteristic impedance of the line.

2.1 Transmission Lines 39

V'~~~_I" __ /m _

Zoo t3 §
0 z -/

FIGURE 2.7 Transmission line circuit [or mismatched load and generator.


If the line is a quarter-wavelength long or, more generally, ~e = 1../4 + nA/2, for n = 1,2,3, ... , (2.26) shows that the input impedance i given by

Z2 Zin = _0. 21.


Such a line is called a quarter- ... vave transformer because it has the effect of transforming the Load impedance, in an inverse manner, dependi n g on the ch aracteristic impedance of the line. This provides a useful practical method of impedance matching, which we will study in more detail in a later, ection.

Generator and Load Mismatches

In the preceding example, of transmission line circuits it was assumed that the generator was matched to the transmission line, so that no reflections occurred at the generator. In general. however, both generator and load may present mismatched impedances to the transmission line, We will study this case here, and see that the condition for maximum power transfer from generator to load may. in some situations, require a standing wave on the line,

Figure 2.7 shows a transmission line circuit with arbitrary generator and load impedances, Z8 and Z£, which may be complex. The transmission line is assumed to be lossless, with length t and characteristic impedance Zo. This circuit is general enough to m del In st passive and active networks that occur in practice.

Because both the generator and load are mismatched, multiple reflections can occur on the line, since reflected waves from the load will re-reflect from the generator, and form an infinite sequence of reflections. Tn the steady state. the net result is a single wave traveling toward the load, and a single reflected wave traveling toward the generator. We can analyze the circuit of Figure 2.7 by first finding the input impedance looking into the terminated transmission line from the generator end. Thus, from (2.25) and (2.26);

_ 1 + fee-2.iW __ 20 Ze + j Zo tan fjf.,

Zi,., = Zo 2 pc

I-ffe-J• ZO+jZttan{3.e


where r f is the reflection coefficient of the load:

r Z, - Zo


Ze + Zo

The voltage on the line is given by (2.'l8a), and we can find Vol-, the amplitude of the incident wave, from the voltage at the generator end of the line, where z = -e:


so that

40 Chapter 2: Transmission Lines and Microwave Networks

v+ = V Zm 1

o R Zin + Zg (elf'11 + re-iPf)'

This expression can be rewritten. using (2.33). as

, 20 e-i{1(

11' = V, ..

(I ~ Zil + Z~ (I - r~r~e-2)tH)

where r,\' is the reflection coefficient seen looking into the generator:

r _ ZII- Zo g-

ZR+ Z(}

We can calculate the power delivered to the load as




Pe = ~Re!VinI~) = ~jVi,J';Re{-1 } = ~IV~121 Zin '11Re{_1 }.

2 2 Zin. 2 lin + z, z;

Now if 2in = Rill + .i X", and Zg = R~ + j X *' then (2.38) can be reduced to


J I I~ Rio

Pe = - Vg. ')~ (X X . 0 •

2 (Rill + RII • + in + g)-

We can now use these general results to consider several special cases of load impedance, for a fixed generator impedance.

Firs! assume the case in which the load is matched to the line, so that ll' = 20. Then re = O. and S'W'R= Ion the line. The input impedance is Zin = lu, and the power delivered to the load is, from (2.39).


p. _ !IV 12 Zn

I, -? ~ (Z + R )2 + X2

- rJ g !I


ext, consider the case in which the generator is matched LO the input impedance of a mismatched line. That is, the load impedance Zc and/or Ute transmission line parameters (Je and Zu are selected to make the input impedance 2111 = Z/i' so that the generator is matched to the load presented by tile terminated transmission line, Then the overall reflection coefficient, r. seen nt the input to the line is zero:

r = Zjl1 - Zg = O.

Zin +Zg

Note, however, that for this case in general r It # 0 and rl i= 0, and there I11Hy be a standing wave 011 the line. The power delivered to the load is

P, _ ~ 1 \I 12 Rg

C - 2 g 4( R~ + X~)"

Observe that even though the term mated line is matched to the generator. the power de livered lo the load may be less than the power delivered to the load from (2.40), where the line was marched to the load, but not to the generator. This leads to the question of what is the optimum load impedance, or equivalently, what is the optimum input impedance, to achieve maximum power transfer to the load for a given generator impedance.

I.f We assume the generator series impedance, Zg, is fixed, we mny vary the input impedance Zin until we achieve the maximum power delivered to the load. Knowing Zin, it is then easy to find the corresponding load impedance Ze via an impedance transformation

(2.41 )


2.2 The Smith Chart 41

along the line. To maximize Pe we differentiate with respect to the real and imaginary parts of Zio. Using 2.39) gives

2Rin(Rin + Rg) . = 0

[(Rin + R/!.? + (Xin + Xs)2J2 '





Solving (2A2a,b) simultaneously for Rin and Xingives RiM = Rg and Xin = -Xs' or

z, = Z;.


This condition is known as conjugate matching, and results in maximum power transfer to the load, for a fixed generator impedance. Under these conditions the power delivered to the load is

I I ~ I

Pe = 2, V~J 4R ' g


which is seen to be greater than or equal to the powers Of (2.40) or (2.41). Al so note that the reflection coefficients ["e, r g, and T may be nonzero, Physically, this means that in some case the multiple voltage reflections on a mismatched line may add in phase to deliver more power to the load than would be delivered If the line were matched (no reflections). If the generator impedance is real (Xg = 0), then the last two cases reduce to the same result. which is that maximum power is delivered to the load when the loaded line is matched to the generator (Rlil = Rg, with Xin = Xi: = 0).

Finally, note that neither matching for zero reflection eZe = Zo ), nor conjugate matching (,zin = Z;), necessarily yields a system with. the best efficiency. For example, if Zg = Ze = Zo then both load and generator are matched (tIC reflections), but. only half the power produced by the generator is delivered to the load (half is lost in Z/I)' for a transmission efficiency of 50%. This efficiency can only be improved by making Zg as small as possible.


The Smith chart, sh wn in Figure 2.8, is a graphical aid that is very useful for solving transrni 'sian line problems. The reader may ·eel the t, in this day of scientific calculators and powerful computer-aided design soItware(CAD), graphical solutions have no place in modern engineering practice. In fact. however, the Smith chart is more thanjust a graphical technique. Besides being an Integral partof much of the current CAD software and test equipment for microwave design, the Smith chart provides an extremely useful way of visualizing transmission line phenoruen on, and so is also important for pedagogical reasons. Microwave and RF engineers can develop intuirion about transmission line and impedance matching problems by learning co think in terms of the Smith chart.

42. Chapter 2: Transmission Lines and Microwave Networks

Derivation of the Smith Chart

FIGURE 2.8 The Smith chart.


At first glance the Smith chart may seem intimidating, but the key to its understanding is to realize that it is essentially a polar plot of the voltage reflection coefficient, r. Lei the reflection coefficient be expressed in magn itude and phase (polar) form as r = I r I ej{l. Then the magnitude [I" I is plo [ted as a radius (I I'] ::: 1) [rom the cen ter of the chart, and the angle e (- ·180° ::: e .:s 1.80°) is measured hom the right-hand side of the horizontal diameter. Any passively realizable reflection coefficient can then be plotted as a unique point On the Smith charlo

The real utility of the Smith chart, however, lies in the fact [hal it can be used to convert from reflection coefficients to normalized impedances. (or admittances), and vice versa, using the impedance (or admittance) circles primed on the chart, When dealing with impedances on a Smith chart, normalized quantities are generally used, which we win de- 110te by lowercase letters. The normalizati on constan t is II sually the characteristic impedance of the trans miss ion line. Thu s, z = Z I Zo represen ts the normalized version of the i mpedance Z.

If a lossless line of characteristic impedance Zo is terminuted with a load impedance ZL. the reflection coefficient at the load can be written from (2.17) as


2.2 The Smith Chart 43

where ZL = Z L1Zo is Ole normalized load impedanc . This .relation can be sol ved for ZL in terms of I' to give

I + Wlej@ GL = 1 _ Wle}f)·

This complex equation (which could also be derived from (2.25) with f = 0) can be reduced to two real eg uations by w ri ting r and ZL in terms of their real ami i lIlagi nary parts. Let r = J,- + jf" and 1:1. = rt. + .iXL. Then (2.46) can be written as


. (I + r,.) + .JJ,

rt. + JX[ = (I _ rr)- Jr;'

The real and imaginary parts of this equation can be found by multiplying the numerator and denominator by the complex conjugate of the denominator to give

1- J~ - r~ rt: = (I. - r,? + rf"



Rearranging (2.47) give


2 (' .1)2 (1)2

(T r - 1)' + I", - ~' = --:-' ,

, ,r.}. • )</.

which are seen to represent two families of circles in the J,., I~i plane. Resistance circles are defined by (2.48a), and reactance circles are defined by (2,48b). For example, the rt. = J circle has its center at I', = 0.5, fi = 0, and has a radius of O.S, and so passes through the center of the Smith chart. All of the resistance circles of (2.48a) have centers on the horizontal r; = 0 axis. and pass through the T = 1 painton the right-hand side of the chart. The centers of all of the reactance circles of (2.48b) lie on the vertical F, = I line (off tile chart), .and these circles also pass through the r = I point. The resistance and reactance circles are orthogonal.

Besides being useful for trnnsforrning between reflection coefficient and normalized impedance. the Smith chart can also be used to graphically solve the transmission line impedance equation of (2.25), In normalized form, this equation can be written as


1 + re-2j/l1.

Zil1 = 1 _ f'e-2jfJf'


where r is the reflection coefficient at [he load. and e is the (positive length of transmission Line. We see that (2,49) is of the same form as (2.46), differing only by the phase angles of the r terms. Thus, if we have plotted the reflection coefficient r = I r leI" at the load the normalized input impedance seen looking into a length E of transmission Line terminates with l.L can be found by rotating the point clockwise an amount 2/3£ (subtracting 2fJe from fJ) around the center of the chart. The radius remains constant, since I I' I does not change with position along the line.

To facilitate such rotations, the Smith chart has scales around its periphery calibrated in electrical wavelengths, both toward and away from the "generator" (the direction away from

44 Chapter 2: Transmission Lines and Microwave Networks

the load). These scales are relative, so only the difference in wavelengths between twe points on the Smith chan is meaningful. The scales cover a runge of 0 'LO 0.5 wavelengths, which reflects the fact that the Smith chart automatically includes the periodicity of transmission line phenomenon. Thus, a line of length A/2 (or any multiple) requires a rotation of2/Jf = 2;rr around the center of the chart to transform an impedance from the load end to the input, bringing the point. back [0 its original position,

Basic Smith Chart Operations

We can best illustrate the use of the Smith chart for basic transmission line problems through the use of an example.


A load impedance of' 130 + j90 Q terminates aSCI Q transmission line that is 0.3.1. long. Find the reflection coefficient at the load, the reflection coefficient at the input to the line. the SVVR 011 the line, the return loss. and the impedance seen at the input LO lhe line. (This is the same problem as. Example 2.1.)


We begin by calculating the normalizedload impedance:

130 + j90. .

o = 2.60 + ) 1.80

5 .

This point can be plotted on the Smith chart, as shown in Figure- 2.9. Using a compass and the voltage reflection coefficient magnitude scale thatis printed on most Smith charts. the reflection coefficient magnitude at the load can be read as WI = 0.60. This same C0l11paSS setting can then be applied to the standing wave ratio (SWR) scale to read SW'R = 3.98, and to the return loss scale (in dB) to read RL = 4.4 dB. The angle of the reflection coefficient call be found by drawing a radial line from the center of the chart: through the load impedance point, and reading the reflection coefficient angle on the outer scale of the chan as 21 .80. Note that these values Me in dose agreemen t w ith the results calc lila ted in ExampleTI.

Now draw a circle with center atthe center of the chart, and passing through tho: load j m ped ance poi nt. This circle .i s caned a constan t SWR circle. and it represents the locus of all possible values of reflection coefficient (and impedance) that the load can present along the line. Read the reference position of the load on the wavelengths-toward-generator (WTG) scale as a.220A. Moving along the line a distance of 0.31. toward the generator brings us to the position 0.2201. + 0.3).. = 0,52.011. on the WTG scale. Because the reflection coefficient repeatsevery 0.51., th is' is equivalentto 0.020A. Drawing a radial line at LIlli; position gives the normalized input impedance at the intersection ofthisline and the SWR circle of Gin = 0.255 + jO.l17. Then the input impedance at the input to the line is

z, = ZOZi" = 50(0.255 + jO.l17) = 12"7 + jS.8 n.

The reflection coefficienr at the input to the line still has a magnitude of WI = 0.60; the phase is read from the radial line at the input position and the phase scale as 165 .8~. These values, are in close agreement with the results calculated in Example 2 . .1. 0

2.2 The Smith Chart 45

FIGURE 2.9 The Smith chart for EK!lJ11pLe 2.2.


Us:ing The Admittance Smith Chart

Another powerful feature ofthe Smith chart is that it can be used with normalized admittances in the same w~iy it is USed with normalizedimpedances, and can be used to convert between impedance and admittance. The latter technique is based on the fact that, in normalized form, the input impedance ofa load ZL eonaected to a ),,/41ine is, from (2.32),

.". - t/~l

~H1 - , . .:::. ... ,


which has the effect of converting a normalized impedance to a normalized admittance.

Since a campi etc revo lution around the S rn ith chart corresponds to a line length of ,,/2, a A /4 [Tan sforrnation is equivalent to rotating the chan by 180Q: this isalso equivalent to imaging a given impedance (or admittance) point across the center of the chan to obtain the correspondi ng adm ittance (or im pedance) point,

The same circles can be used lor either impedance or admiuance, for both real and imaginary parts .. As labeled on the chart, however, a positive imaginary part corresponds to an inductive reactance, or a capacitive susceptance, while a negative imaginary part corresponds to a capacitive reactance, or an inductive susceptance.

In this way. a single Smith chart can be used for both impedance and admittance calculations during the solution of a given problem. At different stages of the solution, the

46 Chapter 2: Transmission Lines and Microwave Networks

chart may be used as either an impedance Smith chart or an admittance Smith chart. This L<: often required when solving impedance marching problems with stub tuners.

A load of ZL = 100 + )50 n terminates a 50 Q transmission line. What are the load admittance and the input admittance if the line is O. J SA long?



The normalized load impedance is ZL = (l00 + )50)/50 = 2 + j 1. We initially consider the Smith chart as an impedance chart, and plot the normalized load impedance point and draw the S\VR circle through this point. Next, we convert to admittance by rotating A/4 around the chart (or simply by drawing a straight line through ZL and the center of the chart to intersect the other side of the SWR circle), The chart is now considered M an admittance chan, and the normalized load admittance can be read as YL = 0.40 - JO.20 r (See Figure 2.10.)

Ttl transform the load admittance to the input end of the. line, first read the reference position of the load admittance on the wavelengths-toward-generator scale as 0.463)... Adding the O.15}, length f the llne brings LIS to a position of

FIG RE 2.10 The Smith chart "for Example 2.3.



2.3 Microwave Network Analysis 47

0.613;'. or O.H3)". The intersection of a radial line at this position with the SWR circle gives the normalized input admittance as Yin = 0.60+ }0.66. Then the actual input admittance is

0.60 + jO.66 . .

Ylll=Yil1/Z0= 50' = 0.0120+ JO.0132S.



In this section we show how the familial' concepts of low-frequency circuit analysis can be extended to characterize RF and microwave circuits and networks: We have seen earlier in this chapter that the distributed nature of a circuit becomes important at high frequencie . when physical dimensions become an appreciable [Taction of the electrical wavelength. In addition, it is helpful to be able to view voltages and currents in terms ofincident, reflected, and transmitted waves.

We begin by discussing the use of impedance and admittance matrices to describe the relationship between the total voltages and currents defined at the terminal ports of an arbitrary N-port microwave network, and show how these quantities can be decomposed toto the sum of incident and reflected waves. This leads to a discussion of the scattering matrix which gives an alternative characterization of an N-portnetworkin term of incident and reflected waves. The scattering matrix is central to modem RF and microwave circuit design and will be used extensively in later chapters on amplifier and oscillator design. Finally, we will describe the transmission, or ABeD, matrix.

Impedance and Admittance Matrices

Consider the arbitrary N -port microwave network shown in Figure 2.1 J, where incident and reflected voltages. v,;'" and Y,,:-. and incident and reflected currents. I: and r;;, are defined at each port. Also at each port is defined a terminal plane. In, to provide a phase reference point for the voltages and currents. From (2.14) we can write the total voltage and currents at the nth port as

V - \1+ + Y-

(/ - on II ~

(2.SJa) (2.51b)

I" = 1/ + l,~,

since the terminal plane corresponds to z = 0 in (2.14).

The impedance matrix [Z] of the microwave network then relates these voltages-and currents;


or in syrnbolie form as

rv'l = [Z][I].


48 Chapter 2: Transmission lines and Microwave Networks

FIGURE 2.11 Photograph f the HP872(lB vector network analyzer. This instrument can measure


two-port scattering parameters Lip to 20 GHz, with buill-in error correction, a syn-

thesized source, and a color display, (Courtesy or Hewlett-Packard ompany, Santa Rosa, CA.)

Similarly, we can define an admittanc matrix Ir'] as

or in symbolic form as

[lJ = [Ylrvl.


Of course, the [2] and [1 J matrices are the inverse of each other:


Note that both the [Z] and [Y] matrices relate the total port voltages and currents.

From (2.52), we can see that a given matrix element ZIj can be fund in terms of port voltages and currents as


Zlj= - .

If h =or(>rkt'.i


In words. (2.5S) states that the ijilJ. element of the impedance matrix can be found by

2.3 Microwave Network Analysis 49

driving port j with a current l j, open-circuiting all other ports (so that h = 0 for k =1= 0), and measuring lhe open-circuit voltage at port l, Thus. 2/1 is the input impedance seen looking into port i when all other ports are open-circuited, and Zij is the transfer impedance between ports i and j when all other ports me open-circuited. For this reason, r Z I is often called the open-circuit impedance matrix of the network.

Similarly, from (2.53), the ijth element of the admittance matrix can be 'found as


which states that Ytj call be determined by driving port j with a voltage VJ, short-circuiting all other ports (so that Vk = 0 for k =1= 0), and measuring the hort-circuit current <It port i. The [Y] matrix is often called the short-circuit admittance matrix of the network,

In general. each element of the [Z] Or [Y] matrix may be complex. For an N-port network, the impedance and admittance matrices are N x N in size, so there are 2N1. independent q uantities or degrees of freedom for an. arbitrary N -port network. IJl practice, however, many networks are either reciprocal or lossless, or both, U the network is reciprocal (not containing any nonreciprocal media or elements such as ferrites, plasmas, or acti ve devices), it can be shown that the impedance and admi trance matrices are symmetric, '0 that Zi.l = 2 Ji and Yjj = Yii [l ], If the network islossless, so that 00 power is dissipated in the network. then we will show that all the Zu and Yij elements are purely imaginary quantities. Ei ther of these spec i al cases serves to reduce the number of'inde penden I q uanri ti e " or degrees of freedom that an N-port network may have.

The pure imaginary property of impedance and admittance matrix elements for los Ies networks can be easily derived, If the network is Iossless, the net real power delivered to the network must be zero. Thus Rei P3Vg] = 0, where

Pn'fg = ~[VY[JJ* = ~U2J[I])I[l.l* = ~[I11[Z][/]*

1 * ...

= 2(1,Z1,/, +11212/2 + '2221/[ +.,.)


= 2' L L ,,,,Znlll/,;



In this expression we have used the result from matrix algebra that ([AJr BlY = r BJI[AJ', where [A l' is the transpose of [AJ.

Since the port currents 1/1 are independent in (2.S7), we can set all firS equal to zero except for L«. Then setting the real pan of (2.57) to zero gives


Rel2m",) = O.


Next, let all port currents be zero except for I", and I". Then (2.57) reduces to

since 2m" = 2"m for a reciprocal network. Bur (fir ll~ + Imi,,) is a purely rea! quantity which is, in general, nonzero. Thus we must have that


50 Chapter 2: Transmission Lines and Microwave Networks

Equations (2.58) and (2.59) together imply that Re{Zmn] = 0 for any 111., h. Thus the impedance matrix of a lossless network has purely imaginary elements. The reader can verify that the same conclusion applies to the admittance matrix as well.

The Scattering Matrix

Like the impedance or admittance matrix for an N -port network, the scattering matrix also provides a complete description of the network as seen at its N ports. While the impedance : nd admittance matrices relate the total voltages and currents at the ports, the scattering matrix relates the voltage waves incident on the ports to those reflected from the ports, The scattering matrix representation is especially useful at high frequencies where it is difficult to measure total voltages and currents, but easier to measure incident and reflected voltage. For some components and circuits, the scattering matrix elements can be calculated using network analysis technique .. Otberwi e, the scattering parameters can be measured directly with a vector network analyzer (see photo in Figure 2.1.1). Once the scattering parameters of the network are known, conversion to other matrix representations can be performed . .if needed.

Again consider the N -port network of Figure 2.12. where v,~ is the amplitude of the voltage wave incident at port ll, and V,; is the amplitude. of the voltage wave reflected from port n. The. catrering matrix, or [Sl matrix, is defined in relation to these incident and retlected voltage waves as

v+ 1

v:+ 2


v:- 2

V+ N




v:_'r 1--I-4lr

~ __ ./. ---+- rmJL--

I~ Vi. -I;;

FIGURE 2.12 An arbitrary N-port microwave network.

2.3 Microwave Network Analysis 51

A particular element of the r SJ matrix can be found as


III words, (2.61.) says that Si; is found by driving port j with an incident wave of voltage v: and measuri ng the reflected wave amplitude. Vi -, coming out of port i. The incident waves on all ports except the jth port are set to zero, which means thatall ports should be terminated with matched loads to avoid reflections from rhe connections (which would amount to incident waves), Thus, Sir is the reflection coefficient seen looking into port i when an other POIts are terminated in matched loads, and Sij is the transmission coefficient from port j to port i when all other ports are terminated in matched loads.

We can now show 110W the [S] matrix can be determined from the [Z] or [YJ matrix, and vice versa. First. we assume that the characteristic impedances of all ports are identical, a simplif-ying assumption that call be alleviated with generalized scattering parameters III Then. for further convenience, we can set Zo = 1. The total voltage and current at.the ntJl port can be written as in (2.51):


1 - I+ + 1- - V+ - fi-

n - n 11 - U Y,n'


Using the definition of [Z] from 2.52) with 2.62) gives

which can be rewritten as


where rUlis the {mil, or identity, matrix defined as

[U] =


o 0 000



Comparing (2.G3) to (2.60) Shows that

IS] = ([2] + [U])-l ([2J - [U J)


which gives tile scattering matrix in terms of the impedance matrix. Note that for the special ea .e of a one-pi rt network, (2.65) reduces [0

ZII - I SII =--~ZII + 1

in agreement with the result for the retlection coefficient seen looking into a load with a normalized input impedance of z II .

To find [Zlin terms of [S], rewrite (2.65) as [Z][S] + CUllS] = [Z] - [U], and solve for [Zl to give

I Z I = ([U 1 -I.S])-I ([U1 +, SI).




52 Chapter 2: Transmission Lines and Microwave Networks

FTGurill 2.1.3 A matched 3 dB attenuator wi III a .50 n characteristic impedance.


Recall that we have normalized the impedance to unity, so (2.66) must be multiplied by Zo to recover the actual impedance. Further properties of the [S] matrix, such as the symmetry of [S] for reciprocal networks, and the fact that [S] is unitary for IQ~sle-5S networks, are derived in reference IIJ.


From (2.61 Sil cal] be found as the reflection coefficient seen at port 1 when port 2 is terminated in a matched load (Z() = 50 Q);


. ')))) ij Find ihe S parameters of the matched 3 dB attenuator circuit shown in Figure 2. L3.

V-I Z(I) Z I

_ I _ (Ill _ in - ~()

511 - V+ +_ - r vt=o - 2.(1) _ Z .

I vl -0 III \1 20 "11 port "'-

where 2r~) is the input impedance seen at port 1 when port 2 is terminated W.iU1 a matched load. With reference to Figure 2:. J 3. this can be calculated as

(I) 8.56 + [141.8(8.56 + 50)] ~O

Z. = . =) Q

In 14 L.8 + 8.56 + 50) •

so SII = O. By ymrnetry of the circuit. we also have .$':\1 = O. The fact that SII = S22 = 0 means that the input port is matched when the output port is terminated In iii matched load, and vice versa. Such a network is referred to as marched. but. it is irnportant to realize that the ports may be mismatched if the other pons are not terminated in matched load ..

S21 can "be found by applying an incident wave at port I, vt, and measuring the outcorning wave at port 2. 112-, This is equivalent to the transmission coefficient from port 1 to port 2:

Prom the fact that Su = S22 = 0, we know that Vl- = 0 when port 2 is terminated in 20 = 50 Q, and that v/ = O. In this case we then have that vt = VI and v2- = V2. So by applying a v ltage VI al port 1 and using voltage division twice we can find the voltage across the 50 Q load resistor at port 2:

( 4 ! .44 ) ( 50 )

V2- = V2 = VI 41.44 + 8.56 50 + 8.56 . = 0.707VI•

where 4'.44 Q is the resistance resulting from the parallel combination of the 50 Q load and the 8.56 n and 141.8 Q resistors in series, Thus, SI_ = S21 = 0.707.

2.3 Microwave Network Analysis 53

'I h

P;" ~>----IV' I [~( r-----------I ~

PM 2


- ~ -
+ [AI B] + [ A2 B2] +
V Vz V~
I C] D, D~
- - C~ - (b)

l"lGURE 2.14 fa) A two-parr network; (b) a cascade connection of two-pori networks.

If the input power is 1 Vtl2 /220, then the output power is

Iv2-12 220

IS:n 121 vtl2 jv~ 12 220 = 420 '

which is one-half of the input power, as expected for a 3 dB attenuator. 0

The Transmission (ABeD) Matrix

The Z, y, and S parameter representations can be used to characterize a microwave network wi til an arbitrary number of ports, but in practice many microwave networks consi st of a cascade connection of two or more two-port networks. In this case it is convenient to define a 2 x 2 transmission, or ABCD, matrix, for each two-port network. We will then see that the ABeD matrix of the cascade connection of two or more two-I;l0rt networks can be easily found by multiplyi ng the ABCD matrices of the individual two-ports.

The ABeD matrix is defined for a two-port network in terms of the total voltages and currents as shown in Figure 2.1. 4a and the fa llowi ng re 1 auons between these quanti ties:

VI = AV2 +Bh it = CVz + Dh,

or i:n matrix form as

[VI] = [A B] [V2 ] .

11 C D lz


It is important to note from Figure 2.140 that a change in the sign convention of l z has been made from our previous definitions, which had h ill: the current flowing into port 2. The convention that h flows out of port 2: will be used when dealing with ABeD matrices so that in a cascade network 12 will be the same current that flows into the adjacent network, as shown in Figure 2.14b. Then the left-hand side.of (2.67) represents the voltage and curre-nt at port I of the network, while the right-hand side of' (2.67) represents the voltage-and current at port 2.

54 Chapter 2: Transmission Lines and Microwave Networks

In the cascade connection of two two-port networks shown in Figure 2.14b, we have that




Substituting (2.68b) into (2.68a) gives

which shows that the ABCD matrix of the cascade connection of the l wo network is equal to the product of the ABeD matrices representing the individual two-ports. Note that the order of multiplication of the matrices must be the same as the order in which the networks are arranged, since matrix. multiplication is 110l commutative.

The u efulness of the A BCD matrix representation is further enhanced by the fact that a library of ABCD matrices for elementary two-port networks can be compiled, and applied in building-block fashion to more complicated microwave networks thal consist of cascades of these simpler two-pons. Table 2.1 lists a number of useful two-port networks and their ABeD matrices,

The A BCD parameters can be derived in terms of the Z, Y or S parameters for a given network. To establish conversion from the impedance matrix, for example, we first change the sign convention for tile current at port 2 in the impedance matrix definition of 2.52) to be consistent with thatof LheABCD matrix:


VI = llZl1 - hZI2, V2 = II Z2J - hZ'12'

(2.70a) (2.70b)

Then from (2.67) we have that

8 - '!.!I

h. v~""u


(2.7 Ie)


]f the network is reciprocal, then Z:'_I = Z12, and (2.71) can be used to show that AD.BC = 1.

2.4 Impedance Matching 55

TABLE 2.1 The ABeD Parameters of Some Useful Two-Port Cirelli ts

c 0 0 ABCt) Parameters

o,.__,__-o..----, 0

:=je: : @5J @:

B=Z D == I

A == cos f31

C == jYo sin fiJI

B =jZo sin (31 D == cos [3/




The basic idea of impedance matchlng, or tuning, is illustrated in Figure 2.15, which shows an impedance match i ng network p I aced between a J oad i mpedance and a transmissien line. The matching network is ideally lossless, to avoid unnecessary loss of power, and is usually designed so that the impedance seen looking into the matching network is 20. the charac teri stic impedance of the reed Ii ne, Then reflecti ens are elimi nated on the transmission line to the I eft of the matching network, although there wi IJ be mu Iti pie reflections between the matching network and the load.

Impedance matching is important in wireless systems for several reasons:

Marching network

Lond z!.

FIGURE 2.15 A lossless network marching an arbitrary load impedance to a transmission line.

56 Chapter 2: Transmission Lines and Microwave Networks

• Maximum power is delivered to a load when itis matched to the feed line (assuming rhe generator is matched).

• Impedance matching sen: itivc receiver circuitry (antenna, low-noise amplifier. mixer) improves the signal-to-noi: e ratio of the system, and hence the maximum data rate.

• Impedance matching in transmitting system minimizes the required RF power. thus minimizing pri me power (max imizi ng battery 1 ife, red uci n g risk of rad i arion hazard).

The Quarter-Wave Transformer

As long as the load impedance, 2/_. has a positive real parr, a matching network can always be found. Many types of matching networks are available for practical use P]' but here we will limit ow: discussion to the design and performance of a few basic matching methods. These include the quarter-wave transformer, lumped element matching networks, and singlestub tuning. These techniques will be used in later chapters when We discuss the design oj amplifier and oscillators

As mentioned in Section 2. I, the quarter-wave transformer is a simple and useful circuit for matching a real load impedance to a transmission line, An additional feature of the quarter- wave. transformer is that it can be extended to multiseciion designs ill a methodical manner, to provide broader bandwidth [II. If only a narrow band impedance match is required. a single-section transformer may suffice. Although the quarter-wave transformer can only march a real loud impedance, a complex load impedance can alway! be transformed to a real impedance by [Ising an appropriate length of transmission line between the load and the transformer.

Here we will analyze the frequency performance of the quarter-wave transformer ali a function of load mismatch. The circuit 'is shown in Figure 2.16, where the characteristie impedance of the matching section is given by


Where Zl. is a real load impedance. AI' the design frequency, iil, the electrical length of the matching section is .1..0/4. but at other frequencies the electrical length is different, and ~ perfect match is no longer obtained, We will now derive an approxi mate expression for Ul( mismatch versus frequency.

The input. impedance seen looking into the matching section is

ZL + j211 2;0 = 21 Z + . 2 ' I } /,1

where I = [an /3e = tan e, and {3t = e = :n: /2 at the design center frequency, ./el' The reo flection coefficient seen at the input to the trans former is then


2111- 2(1 ZI(ZL - 20)+ j/(ZT - 20ZI.)

r = = --------':---;:.----;.-

ZIll + 20 21 (ZL + 20) + jf (Z~ + ZOZL) .


__ Z_-O_--(:_---Z--I----J Z, ,,,,,1,1

FIGURE 2.16 A quartet-wave matching transformer. e = 1./4 al the design frequency .fu.


2.4 Impedance Matching 57

Since, from (2.72), Z? = 202,-, (2.74) reduces CO

r = 2,- ~ 20

2,_ + 20 + j2tvZoZ,-

Then the reflection coefficient magnitude is


If I = I Zl. - 2Q I .

JeZl, + ZO)2 + 4(2 20Z" 1


1 =~==~~~~~==~~~==~

J 1 + 4Zo2L/(ZL - Zo? + 4ZoZl.t2/(ZL - ZoP



j 1 + [4Zo Zl./(Z 1. - Zo)2J sec'l 8

since 1 + (2 = 1 + tan2 e = sec2e.

Now if we assume that the frequency is near the design frequency, fo,. lien;C·~ ).,0/4 and e ;;;;: IT /2. Then sec2 e » I, and (2.76) simplifies to

12L - 201

WI ;;;;: ~.' [cose'], for e near n/2.

2", ZOZL


This result gives the approximate mismatch of the quarter-wave transformer near the design frequency, as shown in Figure 2.17.

If we set a maximum value, f""of the reflection coefficient magnitude that can be tolerated. then we can define the bandwidth of the matching transformer as

so = 2( ~ -em)'


since the response of (2.76) is symmetric about e = n/2, and I' = f", at e = B/II and at e = tt - em. Equating r m to the exact expression forrefiectioncoefficient magnitude in (2.76) allows us to solve [me",:

2.. _I + (2../. ZoZl. 8)2

? -. sec,

L;;, . Z/. - 20


e '" /31

FIGURE 2.17 Approximate behavior of the reflection coef6cient magnitude for a quar~er-W,1ve transformer operating near its design frequency.

I Plb


58 Chapter 2: Transmission Lines and Microwave Networks


FIGURE 2.18 Reflection coefficient magnliudc versus normalized frequency for II quarter-wave ---------- transformer with various load mismatches.


ron 2.JZQZ;_

co:se", =JI _ f;: IZ,- - 201'

If we assume TBM lines, then we have that

e = fU, = 2n/ ~ =rr.~ .

V/I 4/0 2.10

where u, is the phase velocity for the transmission line. Therefore the fractional bandwidth is, from (2.79),


tJ.f = 2Uo - f,,,) = 2 _ 2J,,, = 2 _ 4&m

fo fo f() 1T

_ 4 ... -1 [ rllT 2,y'Z()Z'L ]

-2--cos ~

If 'l-r2IZL -Zol



Fractional band width is usually ex pressed as II percentage, I 00 tJ.f! fu % . Note th at the bandwidth of the transformer increases as ZL becomes closer to 20 (a less mismatched load). Figure 2.18 shows a plot of the reflection coefficient magnitude versus normalized freq uenc y for various mismatched loads. N ate the trend of increased bandw i dth for smaller load mismatch.

Matching Using L-secttons

Another popular type of impedance matching network is the Lssection. which uses two reactive elements to match an arbitrary load impedance to a transmission line. This technique is used extensively in lower frequency circuit design, and bas the advantage over the quarter-wave transformer in that the load impedance need not be real.

There are two possible configurations for an L-Secl:ioD matching network, as shown in Figure 2.19. If the normalized load impedance, GL = ZL I Zo, is inside the 1 + jx circle on tbe Smith chart, then the circui t of Figure 2.19a should be used. If the norniulized load impedance is outside the J + j x circle. the circuit of Figure 2.19b should be used. The I +}. circle is the resistance circle on the impedance Smith chart for which r = I.

2.4 Impedance Matching 59



FIGURE 2.19 L-section matching networks. (a.) Network for ZL inside lite 1 + [x circle. ()J) Network for Z" outside the 1 + ix circle.

J n either of the configurati ons of Figure 2.19, the reactive elements may be either inductive or capacitive, depending on the load impedance. If the frequency is relatively low and/or the circuit size is electrically small, lumped-element inductors or capacitors Can be used. AI higher frequencies, however, it ill difficult to implement lumped element capacitors and inductors, so in this case tuning methods using transmission line stubs may be preferred.

While analytic solutions for the required values of series reactance jX and shunt susceptance jB are available [1], it is often convenien tin practice to use the Sill i th chart to fi nd these values for a given load impedance. Thi s procedure is best illustrated wi th an example.


Design an L-sectiOIl matching network to matcha series RC load having an impedance ZL = 200 - j 100 Q. to a 100 Q line, at a frequency of 500 MHz.


The normalized lcad impedance is z , = 2 - j 1, which is plotted OIl the Smitbchart of Figure 2.20a. Thi spain lis inside the 1 + j x circle, so we w ill use the matching ciroui t of Figure 2.1981. S i nee the first element from the load is a shu nt susceptance, it is helpful to convert toa load admittance )II. by drawing the SWR circle through the load impedance, and a straight 1 ine from the load th rough the center of the chart, as shown in Figure 2.20a. Now, after we add the shunt susceptance jfi and convert back to impedance, we wan I' to be OIl the 1 + j x circle, so that we can add a series reactancejX to match the load. Thi.s means that the shun! susceptance jji must move us from y t. to the 1 + j x circle on the admittance Smith chart. Thus, we construct the rotated 1 + j x circle as shown ill Figure 2.20a (center at 0.333). Then we see that adding a normalized.susceptance of jb = j03 will move us along a constant conductance circle to y = 0.4 + ,10.5 (this choice is the shortest distance from y [, to the sh if ted I + j x eire le). Converti ng back to impedance leaves us at z = 1 - j 1.2, indicating that the addi tion of a series reactance x = j J .2 will bri ng us to the center of the chart, to complete the solution.

The matchingci rcuit thus consists of a shun!' capaci tor and a series inductor, as shown in Figure 220b. At a frequency of f = 500 1\1Hz, the shunt capacitor has a value of

b .

C = -. -- = 0.92pF, 2;rrfZrj

and the series inductor has a value of


L = -- = 38.8 nI-1. 2;rrf

Rotated I + j;A' ,,1rcla 0;1 adnii u ~ncc churl

. / .

60 Chapter 2: Transmission Lines and Microwave Networks


2,.6IpF ~r-~~~


20= loon

Solution 2

FIG RE 2.20 Solution (0 Example 2.5. (a) The Smith chart for the L-seciJon matching networks, (bl ----------- The two possible L-s13clion matching circuits. (c) Reflection coefficient magnitudes versus frequency for the matching circuits or (b).

2.4 Impedance Matching 61


" ,


\ \ \

\ Solution

\ :2





\ \



Solution I

Irl 0.5


0.5 J(GH1.) (c)


FIGURE 2.20 (Co.lllill.l.led)

There is also a second solution for this tuning problem. If instead of adding a shunt susceptance of b = 0.3, We use <'I shunt susceptance of b = -0.7, wewill move to a poi nt on the lower half 0 f the rotated 1 + 1 x circle, to y = U.4 - j 0.5. Then converting to impedance and adding a series reactance of x = -1.2 leads LO a match as well. This matching circuit is also shown in Figure 2,20b, and is seen to have the positionsof the inductor and capacitor reversed from ihe first matching network. At a frequency of 500 MHz, the capacitor for this solution has a value of


c= =2.61pP.

, 2rrfx Zo

while (he inductor has a value of


L = -'-' = 46.1 nl-l, 2njb

Figure 2.20c shows the resu lting reflection coefficient ruagni tudes versus frequency for these two matching networks. ass liming that the load impedance of 2L = 200 - j IOOQ at 500 MHz consists of a 200 Q resistor and a 3.18 pF capacitor In series. There is not a substantial difference in bandwidth for these two solutions, bUL in other cases 111e difference may be more significant. 0

Single-Stub Tuning

Finally, we consider a matching technique that uses a single open-circuited or ShoJ1- circuited length of transmission line (a ,~tub), connected either in parallel or in series with (he transmission feed line at a certain distance from the load, as shown in Figure 2.21. Such a luning circuit is convenient from a microwave fabrication aspect, since lumped elements are not required, and the necessary transmission lines can easily be etched in planar circuit form. Single-stub tuning networksare used extensively in transistor amplifier and oscillator circuits, as discussed in Chapters 6 and S.

In si ngle-stub tuning the distance, d, from the load to the stub pas ition, and the value of the shunt susceptance (or series reactance) provided by the stub, are adjustable parameters. These two degrees of freedom can be used to match an arbitrary load impedance to any feed

----d--- ....

62 Chapter 2: Transmission Lines and Microwave Networks

Open or I sb rted 1 stub


----d--- .....





Open or shorted stull


FIGURE 2.21 Single-stub tuning circuits. (a) Shunt stub. (b) Series stub.


line (assuming the load impedance has a positive real part). For the shunt-stub case, the basic idea is to select d so that the admittance, Y, seen looking into the line at distance d from, U1e load is of the form Yo + j B, where Yo = 1/ Zoo Then the stub susceptance is chosen as - j B, resulting in a matched condition. For the series-s tub ca. e, the distance dis selected so that the impedance, Z. seen looking into the line at a distance d from the load is of the form 20 + j X. Then the stub reactance is chosen as - j X, resulting in a matched condition.

As discussed in Section 2.1, the proper length of an open- 01' short-circuited transmission line can provide any desired value of reactance or susceptance. Fora given susceptance' or reactance, the difference in lengths of an open- or short-circuited stub is /",/4. For transmission line media such as microstrip or stripline, open-circuited stubs are easier to fabricate since a short-circuiting via bole is not required.

Analytic solutions for both shunt- and series-stub tuning circuits can. be derived [1], but Smith chart solutions are usually accurate enough for practical work, and have the advantage of beinz quick and providing an intuitive view of the matching procedure. We will illustrate the method with an example for a shunt stub tuner,


.»))))~ For a load impedance ZL = 20 - jl5 n, design two single-stub shunt tuning networks to match this load to a 50 Q "line.


Begin by plotting L11e normalized load impedance, ZL = 0.4 - jO.3, as shown on the Smith chart of Figure 2.22. Since we are using a shunt stub, it is convenient

2.4 Impedance Matching 63

fIGURE 2.22 The Smith chart solution for Example 2.6.

to work with an admittance chart, and so we convert to a load admittance of )'L = 1.6+ j 1..2 by plotting a SWR circle through the load impedance and drawing a diameter. Note that the S WR circle intersects the 1 + j b circle at two points, denoted by )II and)'2 in Figure 2.22. Thus the distance d, fromthe load to the stub, is given by either of thesetwo intersections. Reading the WTG scale, we have

d, = 0.336 - 0.196 = 0.140);.,

ch. = (0.5 - 0,196) + 0.164 = 0.468)....

Of course, there are an infinite number of distances, d, 011 the SWR circle that intersect the J -I- .i b ci rete. U SII ally, however, it is desired to keep the matchi ng stub as close as possible to the load to improve the bandwidth of the match, and to minimize losses caused by a possibly large standing wave ratio onthe line between the stub and the load.

AI the two intersection points, the normalized admittances are

)II = 1 - jL061, )'2 =1 + JI ,,06J.

Thus. the first tuning solution requires a stub with a susceptance of j 1.061. The length of an open- ircuited : tub that gives this susceptance can be found on the Smith chart by starting at y = 0 (the open circuit) and moving along the outer edge of the chart (since g = 0) toward the generator to the j L061 point. The required stub length is thenz I = 0.130A. Similarly. the required open-circuit stub length for the second solution is £2 = O.370A. 0

64 Chapter 2: Transmission Llnes and Microwave Networks


[IJ D. M. Pozar, Microwave Engineering, 2nd edition, Wiley. New York. 1998,

[2] D. K. Cheng, Field and Wave Electromagnetics. 2nd edition, Addison-Wesley. Rending. MA. 1989.


2.1 A transmission line has the following per unit length parameters: L = 0.3 J,LH/m, C = 450 pF/m, /? = 5 Q/m, and G = 0.0.1 S/m. Calculate (he complex propagation constant and characteristic impedance u . this line at 880 MHz. Rccalcu late these quanti ties in the absence of I ass (R = 0 = 0).

2.2 A lossless iransrnission line of length Q.3A is terminated with a load impedance as shown below. Find the reflection coefficient at the load, the SWR on the line, the return loss, and the input impedance to the line.

[- 0.3'\ ------

Z'~ Z_O_=O_7_5_Q_. j Z,~W+ l,on

2 . .3 A lossless transmission line of characteristic impedance 20 is terminated with a load impedance of L50 n. If the SWR on the line is measured to be 1.6. find the two possible values for Zoo

2.4 A wireless transmitter is connected to <In antenna having an input impedance of 80 + j40 n through a 50 n coaxial cable, If the 50 n transmitter can denver 30 W when connected [0 a matched load, how much power is delivered (0 the antenna?

2.5 (a) Calculate the SWR and return loss £01' reflection coefficient magnitudes of 0.0 l, 0.1. 0,25. 0.5, and 0.75. (b) Calculate the SWR and reflection coefficient magnitudes for return losses of I dB,3 dB, 10 dB, 20 ua, and 30 dB..

2.6 The transmission line circuit shown below has Vj = 10 v nTIS, 28 = SO n. to = 50 n. 2" = 60- j40 n. and e = 0.6.:1.. Compute the power delivered to the load using three different techniques:

(a) find rand compute

( Vg)2 I "

Pt. = - -(I -Irn: 2 z,

(b) find 211\ and compute

(c) find VI,. and compute

Problems 65

Di$CUSS the rationale for each of these methods. Which of these methods elm be. used if the line is n losstess?

2.7 For a purely reactive load impedance of (he form ZL. = j X, show that the reflection coefficient magnitude If 115 always unity. Assume the characteristic impedance is real.

2.8 Consider the transmi ssion line circuit shown below. Compute the incident power, the refieered power, and the power transmltted into the infinite 75 n line. Show thai power conservation is satisfied.

10 V c--: :::--:l- __ Z_1_~-_7_5_n _

P,nc- -ptrlll\, Prof-

2.9 A load impedance of Z,_ = 60 + j30 n is to be marched to a 20 = 50 n line using a length e of lossless line of characteristic impedance Z,. ind values for the required real 2, and e.

2.10 Fur the circuit shown below, lind the power delivered to the load and the power dissipated in (he generator impedance for a load impedance of II = 30 + j40 Q. What value or load impedance wi II result LD maximum power delivered to the load'! What is this power?

2.ll Con ider the transmission line circuit below. se the Smith chart to find the SWR on the tine, the return loss, the reflection coefficient at the load, the load admittance. the input impedance to the line. the distance from the load to the first voltage rninlmurn, and the distance from the load to the firs! voltage maximum.

------- I", 0.8"'\ ----

z; ~ Z_D_=_S_O_Q ) Z, .70, j40 n

2.U Use rhe Smith churl to And the shortest lengths of a short-circuited 50 n transmission line stub to give the following input impedance:

(a) lin = 0 (b) 21n = 00

(c) Z", = j5() n (d) 21" = -j50 n

2.13 Repeal Problem 2.12 for an open-circuited length of 50 n line.



66 Chapter 2: Transmission Lines and Microwave Networks

2.1.4 Derive the [Z] <mel If I matrices for the two-port networks shown below,

'?em I

Port 2

2.15 A. particular two-port network is driven at both ports so that theport voltages and currents have the l'ollowing values:

VI = 5.0 L45" V2 = 3 .. 0 l-45°

11 = O.ll45" I: = O.2l90"

Determine the incident and reflected voltages at both ports, if the characteristic impedance is 50 Q.

2.16 A particular rwo-port network is driven at port j with a matched generator, and terminated at port 2 with a matchedload. Ifthe total voltage and current at port I are measured to be VI = 1.314 L 1.2.4" V and '1 = 15.4 L - 21.5" mA, and [he total voltage at port 2 is measured to be V2 = 0..8 i90" v. If the characterisric impedance is 50 n, find SII and ')'<1.

2.17 A three-port network has the scatteri ng rnatri x gi Yen below.

(a) Wllat is the return loss at each port, when all other ports are terminated in matched loads? (b) What is the insertion loss and phase between ports 2 and 3. when all ports are matched? (c) What is [he rerum loss at port J. when ports 2 and 3 are terminated in short circuits?

[0. j £900 [SJ = 0.4 c 180" 0.4 L ISO"

0.4 i_ 180· oj zo- 0.6 l45"

0.4 LiSO']. 0.6 L45~

0.2 LO°

2.18 Verify the ABeD parameters for the first three networks shown in Table 2.1.

2.19 Derive expressions giving the impedance matrix parameters in terms of the ABeD parameters. 2,20 Usc ABeD matrices to find tile voltage Vi. aCI;O$S the load resistor in the circuit shown below,

2.21 Design a quarter-wave transformer to match a 350 n load to ::1 'loon line. What is the percent band ... ~idth for this matching circuit, for an SWR :::: 27

2.22 In tile circuit shown below. a load impedance of ZL = 100 + j200 n is to be marched 10 a 50 Q feed line, using a I erigth , e, of lossless transmission lineof characteristic impedance. 21. Find ZI and ~. Determine, in general, what type of load impedances can be matched using such a circuit,


2, • 50 Q ., 2_1 ] Z,.' 100 +J200 Q

2,23 Design lWO loss less Lssection matching networks for each of the following normalized loud impedances

(a) 2'.1, = 0.5 - JO.8 (b) z: = 1.6 + jO.8

Problems 67

2.24 A load impedance of Z,. = 100 - j J 50 Q is to be matched to a 50 Q line using a single hunt-stub tuner. Find two solutions using open-circuited stubs.

2.25 Repeat Problem 2.24 using short-circuited stubs.

2.26 A load impedance of ZL = J 5 + j 50 n is to be matched to a JOO n line using a single series stub tuner, Find two solutions using open-circuited stubs.

2.27 Repeat Problem 2.26 using short-circuited stubs.













Noise and Distortion in Microwave Systems

The effect of noise L~ one of the most important considerations when evaluating the performance of wireless systems because noise ultimately determines the threshold for the minimum signal level that can be reliably detected by a receiver. Noise is a random process associated with a variety of sources, including thermal noise generated by RF components and devices, noise generated by the atmosphere and interstellar radiation. and man-made interference. Noise is omnipresent in RF and microwave systems, with noise power being introduced through the receiveantenna from the external environment, as well as generated internally by the receiver circuitry. In our later study of modulation methods, we will see that parameters such as signalto-noise ratio, biterror rates dynamic range, and the minimum detectable signal level are all directly dependent on noise effects.

Our objective in this chapter is to present a quantitative overview of noise and its characterization in RFand microwave systems .. Since noise is a random process, we begin with a review of random variables and associated techniques fat the mathematics] treatment of noise and its effects. Next we discuss the physical basis and a model for thermal noise sources, followed by an application to basic threshold detection of binary signals in the presence of noise. The noise power generated by passive and active RF components and devices can be characterized equivalently by either noise temperature or noise figure, and these parameters are discussed ill Section 3.4, foIJowed by the propagation and accumulation of Mise power through a cascade at two-port networks. A more detai led treatmen t of the noi se .0 gu re of general passive networks is given in Section 3.5. Finally, we consider the problem of dynamic range and signal distortion in general nonlinear systems. These effects are important for large signal levels in mixers and amplifiers. and can thus be viewed as complementary to the effect of noise, which is an issue for small signal levels.


3.1 Review of Random Processes 69


In this section we review some basic principles, definitions, and techniques of random processes that we will need in OUy study of noise and its effects in wireless communications systems, These include basic probability random variable, probability density functions, cumulative distribution functions, autocorrelation, power spectral density, and expected values, We assume the reader has had a beginning course in random variables. and so will uot require a full exposition of the subject References [1]-[3J should be useful for a more thorough discussion of the required concepts.

Probability and Random Variables

PrObability is the Likelihood of the occurrence of a particular event, and is written as P{eventj, The probability of an event 15 11 numerical value between zero and unity, where zero implies he event will never occur, and unity implie the event will always occur, Probability events may include the occurrence of an equality, such as P [x = 5), or events related to a range of values, such as P Ix ~ 5).

In contrast to the actual terminology, 11 random variable is neither random nor a variable, but is a function ilia! maps sample values 'from a random event or process into real numbers. Random variables may be used for both discrete and continuous processes. Examples of discrete processes include tossing coins and dice, counting pedestrians crossing a street, and the occurrence of errors in the transmission of data. Continuous random variables can be used for modeling smoothly varying real quantities such as temperature, noise voltage, and received signal amplitude or phase. We will be primarily concerned with continuous random variables.

Consider a continuous random variable X, representing a tandem process with real continuous sample values x, where -00 < x < 00. Since the random variable X may assume anyone 0 an uncountably infinite number of values, the probability that X 15 exactly equal to a specific value, Xa, must be zero. Thus, PIX = xoJ = 0, On the other band, the probability that X is less than a specific value of x may be greater than zero:

O:s P{X :s ':>':0) :s 1. 111 the limit as Xo ~ 00; PIX S xo) --4 1, as the event becomes a certainty,

We will not adopt any particular notation for random variables in this book, as it should be clear from the context which variables Are random and which are deterministic. In most cases the only random variables we will encounter will b associated with noise voltages, and typically denoted-as vn(t), or nCr),

The Cumulative Distribution Function

The cumulative distribution functior: (CDP), Fx(x), of the random variable X is defined as the probability that X is less than or equal to a particular value, x, Thus

Fx(x) = PIX ~ x].


It can be shown that the cumulative distribution function satisfies the following properties:

( I) Fx(x) :::: a (3.2a)
(2) Fx( ) = 1 (3,2b)
(3) Fx(- )=0 (3,2c)
(4) Fx(xl) ~ Fx(x'l) if XI :'5 X2 (3,2d) 70 Chapter 3: Noise and Distortion in Microwave Systems

The last property is a statement that the cumulative distribution function is a monotonic (nondecreasing) function The definition in (3.1) shows that the result of C3.2d) is equivalent to the statement that

The Probability Density Function

The probabilit)! densityfunction (PDF), J~ ex), of a random variable X is defined as the derivative of the CDF:

dFx(x) fx(x) = -a-' x


FxCx) = L\ /x(U} duo


Since the CDF is monotonically nondecreasing, /'x(x) :::: 0 fer all x, The PDF may contain delta functions. as in the case of discrete random variables, for which the CDF is a "stairtep" type of function.

By the fundamental theorem of calculus. (3.3) can be inverted to give the following useful result:


In addition, since F( -(0) = 0 from (3.2c), (3.4) reduces LO the following result that directly relates the CDF to the PDF:

Finally, because F(co) = 1 from (3.2b), (3.5) leads co the fact that the tcU1L area under a probabiliry density function is unity:

i: f,(x) d x = 1.


These results can be extended to cases of two random variables. Thus, the joint CnF associated with random variables X and Y is defined as


I"v(x, y) = --F'I'(x, y),

, ax(Jy'

Similar to the result of (3.4), the probability of x and y both ccurring in given ranges is found from


Fxy(x, y) = PIX .:S x and Y ::: y).


Then thejoinlPDF ls calculated as


The individual probability density functions for X and Yean be recovered from the joint PDF by integration over one of the variables:

f.~ (y) = f_oo fty(x. y) dx ,


Ir(x) = /, .(x, y) dy.



(3. lOb)

3.1 Review of Random Processes 71

For the special case where the random variables X and Y ate statistically independent, the joint PDF is the product of the PDFs of X and Y:

.f~y(X, .1') = .f,l,(x)l"(y).


Some Important Probability Density Functions

For reference, we List here some of 111e probability density functions that we will be using in this book. The mostbasic is the PDF of a uniform distribution, defined as a constant over a finite range of the independent variable:

. I

lrex) = -b. -Q

The constant I/eb-a) is chosen to properly normalize the PDF according to (3.6). Many random variables have gaussian statistics, with the general gaussian PDF given by



-00 < x < 00,


where m is the mean of the distribution, and a2 is the variance.

In our tudy offading and digital modulation we will encounter the Rayleigh PDF. given by

j~.(r) = ~e-r2/2<12 a

The reader can verify that these each satisfy the normalization condition of (~.6)


0::: r < 00.


Expected Values

Since random variables are nondeterministic, we cannot predict with certainty the value of a particular sample from a random event or process, but instead must rely on statistical averages such as the mean. variance, and standard deviation. We denote the expected value of the random variable X as X:, or E l X J. The expected value ·i s also sometimes called the mean, or average value, For discrete random variables the expected value is given as the sum of the N possible samples, X" weighted by the probabilities of the occurrence of hal sample:


X = E(K) = I'>iP{X =: xd.


This resu It directly generalizes to the ca .. e of continuous random variables:


i = E{X} = fOOl x/x(x) d x,



lt is easy to show that the process of taking the expected value of a random variable is a linear operation, and that the following two properties therefore apply (assume X and Y are random variables, with c.; a constant):

(I) E(cX} = cE[X}

(2) E{X + YJ = E[X} + E{Y).

(3. 14a) (3.14b)

We will also often be interested in finding the expected value of a function of a random variable, If we have a random variable x, and a function. Y = g(x) that maps values from x

72 Chapter 3: Noise and Distortion In Microwave Systems

to a new random variable y, then the expected value of y can be found for the discrete case as


5' = Ely) = E{g(x)} = Lg(x;}P{x = xd,



For the case of a continuous random variable this becomes

J' = E(yl = E{g(x)] = L: g(x)/Ai)dx.

The result of (3.15b) can be used to find higher-order statistical averages. such as the ntt:


moment of the random variable, X:


Xi' = E{x"] = x"fJx)dx.



The variance. ()l. of the random variable X is found by calculating the second moment of X after subtracting the mean of X:

0"2 = E{C.I: - .til = roo tx - 5;)z j,(x) d x


= E {x2 - Lxx + i2 j = .1'2 _ x2.

(3. J 7)

The root-mean-square (nn ) value of the distribution is 0". the square root of U1e variance. If a particular zero-mean random voltage is represented by the random variable .r. the power delivered to !l I ~ load by this voltage source will be equal to the variance of .r,

The expected value of a function of two random variables involves the joint PDF:

sex. y) = E'!g(x, y)l = r~ g(x, Y)iry(x, yJdxdy.



This result can be applied La the product of two random variables, x and )I, by letting the function g(x, y) = xy. For tile special case of independent random variables the joint PDF is the product of the individual PDFs by (3.11), so (3.18) reduces to

xy = El,xyj = 1 xjAx).dx 1 yJyfy)dy = E{x}E[yj.

-'00 -::)0


Autocorrelation and Power Spectral Density

An important characteristic of both deterministic and random signals is how rapidly their sample values vary with time. This characteristic call be quantified with the autocorrelation junction, defined for a complex deterministic signal, X(/), as the time average of the product of the conjugate of xV) and a time-shifted version, X(I + r ):

R(r)= f_. x*Cr)x(r-l-r)dr.


It can be shown that R(O) ::: RCr), and R(r) = R( -r). Also, R(O) is the normalized energy of the signal.

For stationary random processes, such ,\5 noise processes. the autocorrelation function is defined as

R(T) = E [z" (t)xCt -I- r j].


Because of the relation between the time variation of a signal and its frequency spectrum, we can also characterize the variation of random signals by examining the spectra of the

3.1 Review of Random Processes 73

autocorrelation function in the frequency domain. For stationary random processes, the power spectral del.1,silY (PSD), S (tv), is defined as the Fourier transform of the autocorrelation function:

Sew) = j' R(r)e-Jlor di . -00


The inverse transform can be used to find the autocorrelation from <1 known PSD:

I jC'O

R(T) = - . S(w)ej{IJt doi.

211: _00


For a noise voltage, the power spectral density represents the noise power density in the spectral (frequency) domain, assuming a ] Q load resistor. If vet) represents the noise voltage, the power delivered to a 1 Q load can be found as

PL = v2(t = E{u2(t») = R(O) = _1 100 Sv(W) dos = roo Su(21r.f) df W, (3.23)

211: -on 1-00

where S~(w) is the PSD of v(t). The last equality follows from a change of variable with to = 21rf. Writing this integral in terms of f (in Hz) is convenient because Sv(w) has dimension W /Hz, and therefore appears as a power density relative to frequency in Beltz.


01)))) -!) Consider a sinusoidal voltage source, Vo cos Wol, which is randomly sampled in time to form arandom process vet) = Vo cos 13 where e = WOf is a random variable representing the sample time. Assume e is uniformly distributed over the interval o ~ () < 2rr. since the cosine function is periodic with period 2Jt. Find the mean of the sample voltages the average power delivered to a I r210ad, the autocorrelation function of the.random process v(t), and the power spectral density.


The PDF for the random variable e is 1'0 (e) = 2~' for 0 <S f) < 211:. Then we can calculate the average voltage as

.1.1(1) = Elv(t)} = r 1J(t).fl1ce) dB, = _1 (21' cos e de = o.

10· 2n 10

The average power delivered to a I Q load ,is given by the variance of tJ(t):

-- [,2if V.212:Jr 1'2

Ps. = v2(t) = E{v2(t)] = v2(t)f~(e)de = _j!_ cos2()de = _0 W.

(I 21ro 2

The autocorrelation can be calculated using (3.21):

Rv(r) = B [V(t)V(l + r») = V~ E{cos wot cos wo(t + r)}

y:212rr V;2 £21'

= _Q_ cos e cos(e +wor)de = _j!_ [COSW()'l" + cos(2(;1 + wor)Jde

2JT 0 414 ,0


=! cosWor

Note that Ro(O) = Vo2j2. Which is the variance of v(r). The power spectral density is found using (3.22a):

74 Chapter 3: Noise and Distortion in Microwave Systems

This result shows that power is concentrated atw = wh and its image at -woo The total power can also be calculated by integrating the PSD over frequency, us in (3.23):

This result agrees with the earlier result obrainedas the variance using the roE a



Thermal noise, also known as Nyquist, or Johnson, noise, is caused by the random motion of charge carriers, and is the most prevalent type of noise encountered in RF and microwave systems. Thermal noise is generated in auy passive circuit element that contains, losasuch as resistors, lossy transmission lines, and otherlossy components. It can also be generated by atmospheric attenuation and interstellar background radiation, which similarly involve random motion of thermallyexcited charges. Other sources of noise include shot noise, due W the random motion of charge carriers in electron lubes and solid-state devices; flicker noise. also occurring in solid-sate devices and vacuum tubes: plasma noise,caused' by random Illations of charged particles in an ionized gas or sparking electrical contacts; and quantum noise, resulting from the quantized nature of charge carriers and photons. A I though these other types of 11 oise d i [fer from thermal noi se in terms of their origi n, their; characteristics are similar enough that they can generally be treated ill the same way as thermal noise.

Noise Voltage and Power

Figure 3.Ja shows a resistor of value R at temperature T degrees Kelvin (K). The electrons iu the resistor are in random motion, with il kinetic energy that is proportional to the temperature, T. These random motions produce small random voltage fluctuations across the terminals of tile resis LOt:. as ill LIS trated in Figure 3.1 b. The mean value of this voltage is zero; but its nonzero rms value in a narrow frequency bandwidth B is given by

VII = v'4k'l'J3R,



k: = 1.380 x 1O-23 J/K is Boltzmann's constant T is the temperature, in degrees Kelvin (K)

B is the bandwidth, in Hz

Ris the resistance. in n







FIGURE 3.1 (a) A resistor at temperature T produces the noise voltage u,,(I). (b) The random


noise vol [age _genemted by a resistor at temperature T.

3.2 Thermal Noise 15

FIGURE 3.2' (a) Tilt: Thevenin equivalent circuit for a noisy resistor. (b) Maximum power transfer of noise power [Tom a noisy resistor to a load over a bandwidth B,

The result in (3.24) is known as the Rayleigh-Jeans approximation, and is valid for frequencies up through the microwave band 14].

The noisy resistor can be modeled using a Thevenin equivalent circuit as an ideal (noiseless) resistor with a voltage generator to represent the noise voltage, as shown in Figure 3.2a, The available noise power is defined as the maximum power that can be delivered from the noise source to a load resistor. As shown in Figure 3.2b, maximum power transfer OCC\.lTS when the load is conjugately matched to the source. Then the available noise power can be calculated as

( Vn)2 1 v,?

P" = "2 Ii = 4R = kTB,


where v" is the rms n ise voltage fthe resistor. This is a fundamental result that is useful in a wide variety of problems involving 110i e. Note that the noise power decreases as the system bandwidth decrease. This implies that systems with smaller bandwidths collect less noise power. Also note that noise power decreases as temperature decreases, which implies that internally generated noise effects can be reduced by cooling a system to tow temperatures. Fi nally, note that the noise power of (3.25) depends on absolu te bandwidth, but TI()t on tile center frequency of the band. Since thermal noi se power is independent of frequency. it is referred to as white noise, because of Ole analogy with white light and its makeup of all other visible light frequencies. II has been found experimentally, and verified by quantum mechanics, that thermal noi e is independent of frequency for a < f < 1000 GHz.

The noise power of(3.25) can also be represented. in terms of [he power spectral density according to (3.23). Since the power given by (3.25) is independent of frequency the power spectral density must also be independent of frequency, and so we have that,

r. kT no

S,,(w) = - = - =~. (3.26)

28 2 2

This is known as the two-sidedpower spectral drmsil.y ofthermal noise, meaning that the frequency range from -B to B (Hz) is included in the integration of (3.23). This is the conventional definition as used in communication systems work. The notation defined in (3.26), where /1,u/2 = kT /2 is the two- ided power spectral density for white noise. will be used throughout this book. (Note that no is a constant with the subscript 'zero'. This should not be confused with the notation /'loW, which we will often use to denote a noise output signal. The subscript for this latter notation is 'ch', and will always be written as a function of time.)

Since the power spectral density of thermal noise is constant wi th frequency, its autocorrelation must be a delta function according to (3.22b):

1 10;:, nil· no

R(T:) = - _e-/WT dcl.l= -8(r).

2JT -c<l 2 2


By the central limit theorem, the probability density function of white noise is gaussian

76 Chapter .3: Noise and Distortion in Microwave Systems

with zero mean:


where (1"1. is the varianceof the gaussian noise. Thermal noise having a zero mean gaussian PDF is known as white gmJ.ssial1. noise. Since the variance of the sum of two independent random variables is the sum of the individual variances (see Problem 35), and the variance is equ i valent to power deli vered to a 1 Q load, the noi se powers generated by two i ndependem noi se sources add in a. common load . This is in cop trast to the case of deterministic sources, where voltages add,

Note that (3.27) is not completely consistent with (3.28),since(3.27) indicatesthai R(O), the variance of white noise; is infinite while (3.28) implies a finite variance. This problem arises because of the mathematical assumption that white noise has a constant power spectral density, and therefore infinite power. In fact, as we discussed earlier, thermal noise has a constant PSD only over a finite, but very wide, frequency band. We can resolve this issue if we understand our use of the concept of white noise to actually mean a band limited PSD having a finite frequency range, but broader than the system bandwidth with which we arc working.

p _ (~)2 _

"I - 2 RI + R'j_

P = (11;12)2 =

,,2 2 RI + lh

kTBRJ R.I + R2 kTBRz Rl + R2


1W0 noisy resistors, R I and R2,allemperatllre T, are shown in Figure 3.3. Calculate the available noi se power from these sou rees by considering the individual noise power from each resistor separately, Next, consider the resistors as-equivalent to a single resistor of value R I + R 2. and verify that the same <iva ilable noise power is obtained. Assume a bandwidth B for the system.

So ll.l 11:0 ri

The equivalent noise voltage from each resistor is found from (3.24):

\1,11 = j4kTBR1 V;,2 = j4kTBR2.

PDr maximum power transfer, the load resistance should be RJ + R2. Then the noise power delivered to the load from each noise source is

3.3 Noise In Linear Systems 77

So the total available noise power is

PI! = P", + P,,2 = kTB.

Considering the two resistors as a single resistor of value R, + 'Rz, with a load resistance of R[ + R2. gives an available noise power of

P" = kTB,

in agreement with the first result.



in a wireless radio receiver, both desired signals and undesired noise pass through various stages, such as RF amplifiers, filters, and mixers .. These functions generally alter the statistical properties of the noise, and soit is useful to-study these effects by considering the general case of transmission of noise through a linear system. We then consider some important sped al cases, such as fi I ters and integrators, and the non linear si tuation where noise undergoes frequency conversion bymixing,

Autocorrelation and Power Spectral Density in Linear Systems

In the case of deterrni nistic signal s, we call find the response of a linear time-invariant system to an input excitation in the time domain by using convolution with the impulse responseof the system, or ill the frequency domain by using the transfer function of the system. Similar results apply to wide-sense stationary random processes, in terms of either the autocorrelation function OJ the power spectral density.

Consider the linear time-invariant system shown in Figure 3.4, where the input random process, xCI), has an autocorrelation Rx("t) and power spectral density S",(0), and the output random process; y(l), has an autocorrelation Ry(T)a.nd powerspectral density Sy(w).lfthe impulse response of the system is h(t), we call calculate the output response as

)'(/) = L: h(u )x(t - fl) du, Similarly, a lime-shifted version of yet) is

Y(I +-r) = fro h(lJ)x(r + 7: -lJ)du.




So the.autocorrelation of y(.r) can be found as

j'OO (CO

Ry('!:) = E {)'(I)y(t + r)) = -00.1-00 h(i,l)h(l,,)E{x(t - u ).x(r + r - v)) du dv

= L: L: h(u)h(v)RAr + u - v)du du.



H.,frl. S,(IL»

_____ F_, 1_.G_U_R_E_3_.4 A linear system wi than impu lse response h (I) and transfer function H (aJ). The in put is a random processrtr), having autocorrelation l(r(T) and PSD SAev). The output random process is y(t), having autccorrelation R~(,) and PSD S .• (I)))_

-I~I' -I~I' -I~I'
Cl f GL
f f
Low-pass High-pass Bandpass 78 Chapter 3: Noise and Distortion in Microwave Systems

FIGURE 3.5 System symbols and frequency responses for low-pass. high-pass. and band-pass

--------- filters,

This result shows that the autocorrelation of the output isgiven by the double convol uti 011 of the autocorrelation of the input with the impulse response: R),(r) = h(r) €I he-r) ® Rx(T). We can derive the equivalent result in terms of power spectral density by taking the Fourier transform of both sides of (3.30), in view of (3.22a):

100 1 j''''' j'C<l

Ry(t)e-jwr de = . h(u)h(iJ) RAr + II. - v)e-jwt d c du dv,

--100 -DC! -C(J -00

Now perform a change of variable La 0: = r + u - u. so that do: = dt . Then we obtain L: Ry(r)e-jwr do = fop h(u)ejrull j_: h(v)e-}W" /-00 R_.(a)e-jo., do' du du, Since H(CtJ) is the Fourier tran s form of h(!.)

FI(w) = i: h(t)e-jw1dt.


the above simplifies to the following important result:

Gaussian White Noise through an Ideal Low-pass Filter


We will now demonstrate the utility of these results with several applications.

As we will see in Chapters S, 9, and 10, filters play an importantrole in wireless receivers and transmitter. The main function of a filter is to provide.fj.·equency selectii ity, by allowing a certain range of frequencies to pass, while blocking other frequencies. Figure 3.5 shows the symbols and associated idealized frequency responses for low-pass, high-pass, and bandpass filters. Here we examine the effect of an ideal low-pass filter on noise.

Figure 3.6 shows white nois passing through a low-pass filter. TIle filter has a transfer function, R(f). as shown. with a cutoff frequency of 6,/. Note that the transfer function is defined for both positive and negative frequency, since we will be using the two-sided power spectral density. OUT usual notation will be to use lowercase letters. such as 1'I,(r) and

"i(l). ""(I) r~
NI N" • f
-Aj' A/ FIGURE 3.6 White noise passing through an ideal low-pass filter, and the transfer function of the

--------- filter.

3.3 Noise In Linear Systems 79

nAt), for noise and signal voltages in the Lime domain, and capital letters, such as N, and Nf). for average powers of noise and. signals.

Since the input noise is white, the two-sided. power spectral density of' the input noise is constant, as given in (3.26):

. n.

S", (f) = 2° (allO.

Then from (3.32) the output power spectral density is given by



~ 2 .~ ?"f(.ltlfl<.6.j

S",,(f) - IH{f)1 S", (f) - ,- , .

o 1'OJ' If I > .6..f


The output noise power is then

No = (2.6..I')S"I'(/) = .6.fno.

We see that the output noi se power is proportional to the Ii Her bandwidth.


Gaussian White Noise through an Ideal Integrator

As we wi 1.1 see in Chapter 9, integrators are critical components-for the detection and demodulation of digital signals, Here we derive an expression for tile output noise power from an integratorwith white noise input; this result will be used Inter for the derivation of error probabilities for digital modulation in Chapter 9.

Figure 3.7 shows a noise signal, n.;{I), applied to the input of an ideal integrator. The output noise signal is no(r). The output of the integrator is the value of the integral, at time t = T, of the input signal. We need to find the average power of the output noise.

The tran sfer luncri on of the integration operation is


where T is the integration interval lime. Evaluating the magnitude squared of (3.36) gives

~ * (.I ~ e-J"'1'){.1 - ei'"7') 2 - 2 cos wT

IH(w)l~ = H(w)H (w) = ., = ------:,---

.w~ w2


since w = 2Jl"f.

If we assume white noise at the input, with Sn.1 ((1) = i1.u/2, then the output noise power can be calculated using (3.23) and (1.32) to give

Nn = [N r~[l I H(f)12 elf = nD;21OO (Sin ~fT) 2 d]

,-00 - - -00. n.f T

= noT }'oo (SiI1X)ZdX = noT

2rr· -00 x. 2


FIG RE 3.7 White noise passing through an ideal' intcgrntor.

80 Chapter 3: Noise and Distortion in Microwave Systems

FIGURE 3.8 White noise passing through a mixer with a ]0(;<11 oscillator signal, cCls(a!ol + e).

-------------- __ ----

The integra] was evaluated by using a change of variables, .. \"= tt f T; with dx = nTdj, and a standard integral listed ill Appendix B.

Mixing of Noise

One of the common functions of a receiver is to perform frequency conversion, by mixing a signal with a local oscillator to shift the original signal spectrum up or down ill frequency. When noise coexists with the signal, the noise spectrum will also be shifted in frequency. While we will study mixers in detail in Chapter 7, here we idealize the function of mixing by considering it as a process of multiplication of the input signal by a local o. cillator signal, as shown in Figure 3 . .8. We wish [0 find the average noise power of the output signal.

We assume that nCt) is a bandlimited white gaussian noise signal with valiance, or average power. a2 = E (112(t ). The local oscillator signal is given by cos(wo' + 8), where the phase, (), is a random variable uniformly distributed on the interval () < e ~ 2n, and Is independent of lI.(r). The output of the idealized mixer is

v(t) = nCt) cos (wol fI).


The average output power from the mixer can then be calculated as the variance of v(t):


This result shows that mixing reduces the average noise power by half. In this case, the factor ofone-half is due to the ensemble averaging of the cos2 wot term over the range oj random phase.

If we now consider a deterministic local oscillator signal of the form cos (()o£ (withou a random phase), the variance of the output signal becomes


The last result follows because the cos? wot term is unaffected by U1e expected value operator, since it is no k nger a random variable. In addition, v(l) is no longer stationary, and therefore does not have an autocorrelation function or power spectral density.) The variance of the output signal is now a function of time, and represents the instantaneou power of the output signal. To find tile time-average output power, we must take the time; average of the variance found in (3.41):

1 10 T tv £2:Jr1f)'rl (J2

No=- £[7i(l))dl=_o a2cos2wOldl=--.

T 0 2n .0 2


since T = J If = 211" Iwo. We see that the same average output power is obtained whethe the averaging is over the ensemble phase variation, or over time.

3,3 Noise In Linear Systems 81


Consider thecomplex mixing product xU) =: 1I(t)ej"~,,, formed by mixing noise voltage flU) with a complex exponential. If the autocorrelation and PSD of fI(t) are R!I{r) and S" (cv). find the autocorrelation and PSDof x(f).


Using the definition of autocorrelation for random processes given in (3.21) we have

R_.(.) = Blx*(t)x(1 + .)1 = E{n(l)e-jW[Jln(t + T)ej«Jo(I+t)j =: E {n(t)n(t + T)}ej',"",r = Ro (1:)ej~JDr

NOLe that x([) is stilJ a stationary process, since it has a proper autocorrelation function. From (3.22a) thepower spectral density is

S ( )_/00 R C')' -jml)I l t _/00 R (r) -J('"-wn)'d - S ( _ )

x Ci.J - -00 x r e (I. - ,-DO "r.e ,,1:" - "OJ Wt),

where the last result follows by replacing (u with w - Wo in S,,(w) = .1:..0 RII(r:}e-J,a'dr:. 0

Narrowband Representation of Noise

In many receiver circuits, signals and noise are passed through a bandpass filter. In this case, it becomes possible to represent the noise in a form that is more convenient for analysis, This is called the narrowband representation of noise, a result that will be very useful in Chapter 9 when analyzing the effect of noise 011 the demodulation of signals.

Figure 3.9 shows gaussian white noise passing through a bandpass filter whh 11 center frequency (1)0 and bandwidth fiw. If the two-sided power spectral density of the input noise is no/2. then the PSD of the output noise, n(r), is as shown in the figure. If fiw « wo, then n(t) can be represented as

11(1) = x(1) cos W{)/ + y(l)sinwQ/,


where x (I) and y(t} are random processes, bu t are slow Iy vary ing due to the n arrow bandwidth of the filter. To show that the above representation is valid, consider the circuit of Figure 3.10, which can be used to generate X(I) and y(t), Here the noise n(1) is divided and mixed separate] y wi th 2 cos 1iI~[ and 2 si n (Vo I. w h ich produces the follow ing results:

211.(/) cos' cvot = 2x(r)r::os2 Wet + 2y(1) sin Woi' cos WOI = X(I) + xU) cos 2w{lt + y(J) sin2wor 2n(t) sin Wut = 2x (t) cos w()t sin wot + 2y{t) sin1 wot = yV) ~ Y(/) cos 2W{)i + X(l) sin 2wol.

S,,({.<)) :'.JJ


FrGURE 3.9 White noise passing through a bandpass (iller, and the power spectral density of the


output noise.

82 Chapter 3: Noise and Distortion in Microwave Systems


H2C08 "'(II

T l.PF

r®-'I~I . ""1

49--1~1 ~ y(1)


(::) 2 sill Will

y(l)sil1 "'01 --- I /1(1)



FIGURE 3.10 Circuit to generate low-pass noise XCI) and yet) from 11 bandpass noise source n(l).

-------------- __ ----

After low-pass filtering with-a low-pass cutoff frequency of f~ = 8,1J)/4n. only the x(r) and y(t) terms will remain in the above results. Thus the output noises x(t and y(c are limited in bandwidth LO !::.w/2, and can be viewed as the bandpass noise at W\l shifted down to zero frequency. Since x(t) and y(l) represent the in-phase and quadrature components or 11(1), as indicated in the phasor diagram of Figure 3.10, (3.43) is also known as the quadrature representation of narrowband noise. We now find the statistics of xU) and .1'(1), and show that these gaussian random processes have zero mean, the same variance as n(l), and are uncorrelated. We will also find the power spectral densities of x(1) and yet).

Since E {I"I (t)} = 0, we have from (3.43) that

E {n(t.)} := 0 = E{x(1» cos WOI + E{y(t)} sjJJ(vo!.

Since cos WOI and sin WOI vary differently with time, we must have

E{x(t)} = E{y(t)} = O.


R,,(,) = E{n(/)I1U -I- r)1

= E (lx(t) cos wfll -I- Y(I) sill (VOl J[x (I + r ) em; (()o(t -I- r) -I- )'(1 + r ) sin (r)o(r + r)JJ = R,r(r) COSW!jl cos wo(t -I- r) + Rxy(r) cos wot Sill wo(J + r)

+ Ry ... (r)sinwol COStrJ()(i + r ) -/- RyC1:) sh1Wol sin wo(t + T)

Next, we evaluate the autocorrelation of n(l) using (3.43):

where R.,">,(r) = ElxU)y(t + r») and R~'.,(.) = E {y(J)x(f + r)) are the cross-correlation functions of x(c) and y(t). Using standard identities to expand the trigonometric produce gives the following:

Rile.) = ~R .r ("r)lcosw()1: -/-coswo(2t + r )]

+ ~R"y(T)[Sin wot -/- Sil1l.llo(2t -/- r)] -I- 4 Ryr(t)[ -sin WOT + sin No{21 -/- r)] + ~Ry("r)[coswor - cosw()(2t -I- r)].


Because n(1) is a stationary process, its autocorrelation must be a function only of r , and cannot ~'ary wi th t. Thus the coefficients of cos Wo (21 + r) and sin Wn (2t -I- r ) must vanish.


3.4 Basic Threshold Detectlon 83

FIGURE3.U Power specrraldensiry of x(t ) ,and y(l).

This gives

R,,(r.) = RyCr} Rxy(r) = -RyAr),

(3.46a) (3.46b)

fwd then (3.45) reduces to

R,I(r) = Rx{r) cos coQt + RxyC") sin 100'"' = RyC'!:) cos (!Jor - RyxCr) sin Wot. (3.47)

This res lilt shows that nCr), .r(O, and yO) all have the same variance, since R,,(O) = R~(O) = R>.(O),

We Call also find RII(r) directly by evaluating the inverse Fourier transform of S" (w), which is shown in Figure 3.9. Since R,i (r) is real. and symmetricabout co = 0, we have

1 100 " 1 1<Xl n !.W1)+iIWJ2

RIl(r) = -2 5,,(w)eJWT dw = ~ S,,(w) cOS,W't' dw = __2_ cos GIn dco

lC -00 _1t 0 2n '''Il-6.r42

no [ " (' 8W) ,(' /::"w) J' no, /::"wr

= -,-, sm lUa + -,- r - Sin WO - ~ l T = -- sm -- COSWOi,

2n: r . 2 , 2 tt r 2


Comparing (3.48) to (3.47) shows that

, no ' /::"ttJT

R,,(.) = R,,{!) = - sin --

- nr 2

Rxy(r) = R),t(r) = 0,

(3,4%) (3.4%)

The fact that Rx)'Cr) = 0 implies that x(t) and y(l) are statistically independent Finally, we can find the PSD of x(t) and Y(I) by taking the Fourier transform of R,,(r):

100 , ' {no for Iwi < /::"w/2

5,,(((.1) = 5y(w) = I?A7:)e-J'~·c.lT; = ,

~OO 0 for Icul > /:';.0)/2


where the required Fourier transform may be found in Appendix C. This power spectral densi ty is shown in Figure 3, I I., mid can be viewedas the bandl irni ted PSD of n (t) sh i fred up and down in frequency by the amount Wo, and low-pass filtered, Note (hal the peak value of the PSD of xU) and yet) is n(), twice that of the PSD of nCI),


We now have enough background in the topics of noise and systems to discuss an application to basic threshold detection, Threshold detection is relevant to most digital modulation schemes, and so we will see this topic again in more detail in Chapter 9,Here we evaluate the probability of error for a simple binary communications channel.


84 Chapter 3: Noise and Distortion in Microwave Systems

J'(t) + 11(1)


FIGURE 3.12 Input signal and noi: e \I{ ltage for <I basic threshold detection sy tern.


Consider a-communications system where binary signals are transmitted in the presence ofbandlimlted white gaussian noise. Thus, the received signal, 1'(1). can be written as the transmitted signal v ltage, S I), plus a noise voltage, n r):

r(t) = S(I)+ n{1)


where Il(r) has zero mean and variance (j2. A sketch of a possible received voltage is shown in Figure 3.12.

When a binary "1" is transmitted the signaling voltage will be s(r) = 1J(), and when a binary "0'" is trans mitted we will have s (/) = O. The receiver must be designed LO process the received voltage. and detect whether a ., I" or (I "0" has been transmitted. In the absence of noise we can simply sample the receive voltage and determine whether it is above Or below a threshold level. In this case, the logical choice for a threshold voltage would be lJo/2. so that if r(t) > 1J()/2 the receiver would detect a "1." and if r(i) < 110/2 the receiver would detect a "0." This detection process can be implemented uslng a simple sampler and comparator circuit. In practical receivers threshold detection could incorporate matched fillers or integrators to minimize the effect of noise, but here we consider only the sampling of the received signal ar its maximum Or minimum P9U1t.

Because the possible nois voltage amplitude ranges from -00 to 00. the received signal may someti roes be less than the threshold when a ·'1" has been sent, and may be greater than the threshold when a ·'0" has been sent. Either of these cases will result i11 a detection error. In fact. there arc four detection possibilities. as listed in Table 3. I.

Probability of Error

We can now find the probability of errario: threshold detection. We define P~ I) as !he. probability of an error ill detection when a binary "I" has been transmitted, and pJO) as th~ probability of an error when a binary "0" has been sent. Knowing these two probabilitia then defines the likelihood of all the outcomes in Table 3.1, SlJ1Ce the probability of a correct outcome is 1 - (probabili ty of an error),

Transmitted r(t);;.. un/2 Detection Correct
Binary Data S(l) ? Outcome Detection
0 0 no 0 yes!
0 0 yes error
Vo yes I. yes!
Vo no 0 error TABLE 3.1 Possible Outcomes of Threshold Detection

3.4 Basic Threshold Detection as

When a binary "1." is transmitted, a detection error will occur if the received signal and noise is less than the threshold level at the sampling time. For a threshold of vo/2. the probability of this ever:t is

(1) ,

P •. = P{r(t) = vo + n{t) < vo/2)

-1~1/2 , . ," _ [",,;2 e-(r-""J!/2iT' . ~ -00 .t,'(I)df~ J-oo.h7ra1 dr


where we have used (3.1.), (3.5), and the gaussian probability density function given in (3.12b). Since nCI) is gaussian with zero mean. the receive signal ret) is also gaussian, but with a mean value Of 110 when a binary "1" is being transmitted. The expression in (J.52) can be reduced to a standard form by using the change of variable x = (110. - r)I,.ff(Ji.

Then we have


where the lower limit is

110 xo= ---.


The integral occurring in (3.53) is rehired to the complementary errorfunction; written as

2 100 ,

elj"c(x) =, C , e~w duo

...."J! ,f


Details on properties of thecomplementary error function, including an algorithm for calculating efjc(x), can befound in Appendix 0., Using the definition of (3.55) allows (3.53) to be written as

, (1) 1 1 (uo)

P = ~eI1c(x.o) = -edc-,--,' ,

"2 2' 2~

which Is our final expression for PJ!). By a similar analysis we can find P?l; the probability of error when a binary "0" is sent. It is left as a probl em, to show that p;O) = pj I) , as might be e-xpected from the symmetry resulting from a threshold of vo/2. The result of (3.56) is dependent on the ratio lIo/a, which can be considered a signal-to-noise ratio (SNR), since Vo is the maximum signal voltage, and dis the rms value of the noise voltage, Since erfc (x) decreases mono toni c ally with x:, large SNR results in lower probability of error.

A graphical interpretation of' threshold detection is shown in Figure 3.13. The probability density functions are shown for the received signal and noise for the two cases of sending a binary "0;' or a" l." The former has a PDF centered at r = 0, while the latter has



r(l) '" n{r)


1'(1} '" l'u + 11.(1) ...---"-1"



3uo 12

FIGURE 3.13 Graphical interpretation of the probability of error far threshold detection.

86 Chapter 3: Noise and Distortion in Microwave Systems

a PDF centered at r = v(). The threshold of vo/2 is located midway between these values. The probabilities of error are the areas of the tails of the two PDFs either above or below the threshold value.


Calculate and plot the probability of error for threshold detection versus the signalto-noise ratio, vII/a, in dB. Use a logarithmic scale for the probability of error.


Since we are dealing with voltages, the signal-to-noise ratio in dB is calculated as

lIr) . Va

-(dB) = 20 log -.

a a

Theil (3.56) can be used [0 evaluate p:ll. The algorithm of Appendix D can be used to calculate values of the complementary error function. A sample calculation follows for vo/cr = 6 dB:

For vo/a = 6 dB we have a numerical value of

Uo = 10(,/20 = 2.0. a

Then the argument of the complementary error function is, from (3.54)

1)0 2.0

XCI = ~. = M = 0.707.

2 ...... 2a 2",2

Equation (3.56) gives

p2) =!e1fc(xo) = ~eljc(O.707) = ~(0.317) = 0.15').

The same method can be used for other values of vola, and the result is plotted in Figure 3.14. Note that for large values of SIN [he probability of error becomes Very small. Error probabilities in the range of 10-5 to 1.0-11 are often desired in ~~. 0

FIGURE 3.14 Probabi lily of error versus signal-to-noise ratio for threshold detection.


... 1£-03
~ LE-04
~ lE-06
g IE·07
I c-IO
0 5 I ,

III 15 20 log (~) dB




3.5 Noise Temperature and Noise Figure 87

/'[2]-'\' .

I -

I R ) N" . /I

\ /


-... .....

White noise. source


FIGURE 3.15 Equivalent noise temperature Man arbitrary white noise source.


Besides being received from theexternal environment by tile antenna, noise is also generated by passive and active components of a wireless receiver system. In this section we will study ways of characterizing the noise properties of components such as 3 Illp tifiers , mixers, couplers, and filters, and the transmission of noise through a multistage system.

Equivalent Noise Temperature

If an arbitrary noise SOUTce is white, so that its power spectral density is not a function of frequency (at least over the frequency range of interest), it can be modeled. as an equivalent thermal noise source" and characterized by an eq.d'valen.t noise temperature. This situation is illustratedin Figu re 3. 15, where an arbitrary while noise source of driving poi nt impedance R delivers noise power No to a load resistor R. This noise source can be replaced with a noisy resistor of val ue R, at temperature T, .• where T, is an equivalent temperature selected so that the same noise power is delivered to the load, Thus

T. _ No e - kB'


Wireless components and receiver systems can then be characterized in terms of their equi valent noise temperature, 1~, expressed in degrees Kelvin (K). N Ole that T, ::::. 0, and may be greater OT less than Iil = 290 K. In addi 660, note that the result in (3.57) implies some fixed bandwidth, B, which is generally the bandwidth of the component or system, As an example, consider a noisy amplifier having bandwidth B and power gain G .. Let. the ampli fier be matched to noiseless source and load resistors, as shown in Figure 3 .. 16a, If the source resistor of Figure 3,16a is at a (hypothetical) temperature of T, = 0 K, then the input noise power to the amplifier will be N, = 0, and the output noise power NQ will be due Only to the noise generated by the amplifier itself. We can obtain the same output noise power



:FIGURE 3.1.6 Equivalent noise temperature of a noisy ampllfier, (a) Noisy ampli fier, {b) Equi valent n oi se less am P lifier,