INTRODUCTION TO DIGITAL TRANSMISSION

Transcription

1 CHAPTER 1 INTRODUCTION TO DIGITAL TRANSMISSION What is digital communication? How did it come about? Why did it come about? How can we engineer and anyalze systems? These questions are the motivation for this book, and the last question is its principal topic. Modern telecommunication is the confluence of three great trends over the last two centuries: First, the invention of electromagnetic signaling technology in the form of the telegraph, the telephone, and the radio; second, the development of mathematical theories that made these inventions practical and efficient; and finally microcircuitry, the "chip," which made these inventions small, fast, reliable, and very cheap. To these must be added an intangible that has no explanation and is perhaps simply an absolute: the need of people to communicate with each other. Time and again this urge has financed new communication service, whether it be telegraphing across London in the 1860s or cellular telephones in the 1990s. An axiom of investing once said that American Telephone and Telegraph was always a good investment: When times were good, all stocks go up; when times are bad, people call each other and complain. The AT&T company, for a hundred years a giant, has largely disappeared since 1985 and has been replaced by others. Why this happened is an interesting question but not one for this book. What is important to us now is that only a few fields transport and health care come to mind attract the investment and eager public support that communication does. This fact is the economic basis for digital communication. Although this is a book about engineering analysis, there is room in this first chapter for some history and for the fundamental question of why analog has given way to digital communication. Without these, no one would have been interested in the engineering. The plan of the book concludes the chapter. Digital Transmission Engineering, Second Edition. By John B. Anderson 1 Copyright 2005 the Institute of Electrical and Electronics Engineers, Inc.

2 2 INTRODUCTION TO DIGITAL TRANSMISSION 1.1 SOME HISTORY AND SOME THEMES Some major events in the long and fascinating history of electrical communication are listed in Table 1.1. A listing of dates can be misleading, because trends and inventions are most often born in confusion and in many places at once. For this reason, many dates are approximate, and it is especially difficult to define recent innovations. Some innovations, such as the writing of what we now call software algorithms, evolved only slowly over time. A particularly readable study of this history, its less tangible side, and its human consequences has been published by I. Lebow [1]. A briefer introduction appears in Ref. [4]. A reason to study history is to identify trends, which lead to the structure of the present day. We will look now at Table 1.1 and see what it says about trends in digital communication. Parts of the following will seem obvious, but we discuss them to avoid a greater danger, which is that we miss the subtleties among the commonplace. Communication as we know it today began in the nineteenth century. It is interesting that except for light fibers, the electromagnetic transmission technologies that we use today, namely radio and wireline telegraph and telephone, all appeared in the nineteenth century. So also did the scientific understanding of electricity and electromagnetism, without which the inventions would have made little sense. Radio found immediate use on board ships, but at first it was not at all the channelized medium that we use today. A major contribution to the Titanic disaster of 1912 was that all radio operators used the same spectrum space. Since the Titanic had the strongest transmitter and constantly used it, other nearby operators were overwhelmed and shut down their radio sets. Consequently, the early distress signals from the sinking ship went unheard by nearby ships. Here arises a theme that plays a major role in the book namely, the fact that radio transmission uses a channel and a carrier. It is often said that only by conversion to electromagnetic form can an audio signal be made strong enough for transmission. This may have been true in 1912, but kilowatt audio "transmitters," in the form of car stereos, are all too common today. The critical point is rather that electromagnetic transmission can be translated in spectrum, by shifting its carrier, to whatever frequency works best. For communication with submarines, the physics of salt water imply that the best frequencies lie near the audio range. But efficient antennas here are hundreds of kilometers long, and so any radio service subject to ordinary economics uses a much higher frequency. Transmissions that need to follow the Earth for a few hundred kilometers need medium-wave frequencies (these terms are defined in Table 1.2); for a few thousand kilometers they need shortwave frequencies, and space communication must take place in the micro- and millimeter-wave spectrum. All this derives from various physical laws, and translation by carrier is what makes radio economic. A second important reason for carrier transmission is that many users can share the same spectrum if each is translated by a different carrier; this was dramatized by the Titanic. Considerable technology had to be invented before the idea of narrowband channelized radio, which we take for granted today, became practical. New transmitters

3 Year TABLE 1.1 Important Milestones in History of Digital Communication Event ca Oersted shows electric currents create magnetic fields Henry discovers induction; Faraday and others show changing magnetic fields produce electric fields Various telegraphs demonstrated 1844 Morse commercial telegraph, Baltimore to Washington 1864 Maxwell publishes his theory of electromagnetism 1866 First permanent transatlantic telegraph Various telephone demonstrations by Bell and others 1878 First telephone exchange installed by Bell, Hamilton, Canada 1887 Experiments by Hertz verify Maxwell Marconi and others demonstrate radio over significant distances 1901 First transatlantic radio message by Marconi, United Kingdom to Canada Fleming announces diode tube; DeForest announces triode 1907 Fessenden transmits speech 320 km ca Armstrong devises superheterodyne receiver 1920 First modern radio broadcast by KDKA, Pittsburgh, PA ca Mechanical TV system demonstrations by Baird, London 1928 Gaussian thermal noise papers of Johnson and Nyquist 1929 Zworykin demonstrates electronic TV system ca Armstrong devises FM 1936 Commercial TV broadcasting by British Broadcasting Company, London ca First use of radar ca Matched filter devised, for radar Early computers constructed; proofs of sampling theorem appear; signal space theory applied to communication 1948 Transistor demonstrated by Brattain, Bardeen, and Shockley, United States; Shannon publishes his theory of information Beginnings of computer software; beginnings of microwave long-haul transmission 1953 First transatlantic telephone cable ca Matched filter applied to communication; first chips demonstrated ca Error-correcting codes begin rapid development ca Laser announced, United States, ca Communication satellites using active transponders; long-distance communication to space probes begins 1967 Forney proposes the trellis; Viterbi proposes his algorithm 1970 Low-loss optical fibers demonstrated Microprocessors appear; large-scale integrated circuits appear; speech and image digitization begins rapid development ca Bandwidth-efficient coded modulations begin to appear; digital telephone trunks first installed 1979 Images received from Jupiter ca Digital optical fiber telephone trunks begin to be installed Cellular mobile telephones become widespread in Europe ca Use of the Internet accelerates 1992 First digital mobile telephone system, GSM, begins in Europe Note: Most dates are approximate. 3

4 4 INTRODUCTION TO DIGITAL TRANSMISSION TABLE 1.2 Frequency Bands in Radio Spectrum Frequency Band Band Name Comments < 100 khz khz khz 3-30 MHz MHz MHz 1-10 GHz GHz > 200,000 GHz "Also called shortwave. Extra low frequency (ELF) Low frequency (LF) Medium wave (MW) High frequency (HF)" Very high frequency (VHF) Ultrahigh frequency (UHF) Microwave Millimeter wave Infrared Submarine communication Follows Earth surface AM broadcasting; follows Earth with loss Reflected by ionosphere TV and FM broadcasting Mobile radio Wideband links, Earth and space Space links Optical fiber links were developed that radiated a true amplitude-modulated (AM) signal. Receivers that could effectively reject all but one narrow frequency range came with Armstrong's superheterodyne concept in the 1920s. Another invention by Armstrong in the 1930s, frequency modulation (FM), was the first inkling of another major theme in electromagnetic communication, the bandwidth and energy? trade-off. Armstrong found that FM had a better signal-tonoise ratio at the same transmitter power than AM, apparently in proportion to its bandwidth expansion. Armstrong found it hard to convince his colleagues of this fact, but today we know that the same trade-off appears in many other places. Pulse-code-modulated (PCM) digitization of speech, for example, essentially freezes its signal-to-noise ratio, no matter how many retransmissions the signal passes through, but PCM also increases the transmission bandwidth. An even more radical idea was to appear in Shannon's 1948 article, "A mathematical theory of communication" [5]. Shannon showed that error could not only be traded for bandwidth but could in principle be reduced to zero by the use of coded signals. In a parallel development during the 1940s and 1950s, Kotelnikov, Shannon, Wiener, and others developed a new theory of optimal communication, a theory that worked best with symbolic, or "digital," transmission. This theory was mathematical, not electromagnetic, and its ideas were rooted in the twentieth century. These new digital ideas involved all sorts of complex processing and coding, and were it not for another evolving trend, they would have remained for the most part a curiosity. This trend was the idea of written processing algorithms on the one hand and ever cheaper hardware to run them with on the other. An algorithm is a step-bystep procedure to attain an end. Some algorithms like long division are ancient, but a major leap in the algorithm concept took place in the mid-twentieth century. With the ideas of von Neumann and the perfecting of computer languages and the stored program computer, algorithms took a mighty jump upward in complexity. What has evolved today is a technology in which various functions in a system can all be standard processor chips, each taking its function from the algorithm loaded into it.

5 1.1 SOME HISTORY AND SOME THEMES 5 It seems likely that the algorithm concept will continue to evolve beyond the sequential programs that dominate today, but the mid-century concept has been more than enough to implement the new theory of communication. Everyone knows the story of the large-scale integrated circuit, one of the major technology drivers of the last 40 years. But it is worth reiterating how revolutionary the chip really is. A single vacuum tube active circuit element in the 1950s cost about five late-1990s U.S. dollars. Transistor technology soon reduced this cost manifold and at the same time made the element more efficient and reliable. But photolithography and successive waves of miniaturization came, until the cost of this single device had dropped a millionfold and more. Further drops are still to come. It is interesting to imagine what would happen if another part of the economy, say the cost of energy or of cars, were to drop this much. The upheaval would be hard to imagine and hard to plan for. Just such a revolution is in progress in communication, because it is based on key commodities, processing and transmission, whose price has collapsed. As a small example, take the detector, a part of every receiver. With Fleming's 1904 diode, a great advance in its time, an AM detector was a vacuum tube that cost at least $100. By the 1950s, the detector was a vastly more reliable and cheaper solid-state diode. By the 1970s, this detector and considerably more sophisticated ones were a part of a larger integrated circuit chip that performed several functions. Today the detector and most of the radio can be a low-cost chip on which runs a stored program, whose identity as a "radio" can be changed at will. What has happened here is (a) a move away from communication as the study of physics and devices and (b) a move toward algorithms, information as symbols, and complex processing. Just as energy and bandwidth were found to trade off, now we see a three way trade-off among energy, bandwidth, and processing complexity. An increase in any one means the other two can be reduced, more or less, for the same performance. A PCM digitizer, for example, can be replaced by a more complex digitizer, which puts out fewer bits, which consume less transmission bandwidth. For another example, take Shannon's concept of coded communication. Originally a mathematical theory, it now has a practical meaning: Transmission error may be driven down by more complex coding, instead of more transmission energy. With modern coded modulation, coding complexity can even be exchanged for bandwidth. The cheapest of energy, bandwidth, and processing is now processor complexity. Digital communication is so full of concepts, processes, algorithms, and complexity because this is the route to lower cost. Although the processor revolution dominates our story, it should not be forgotten that many analog components have seen large cost declines. Perhaps the most significant of these is the optical fiber and its codevice, the laser. The much lower per-bit transmission cost of fibers is driving us to a two-choice world, in which fixed channels of any length are fibers and mobile channels are radio. Aside from fiber technology, radio-frequency (RF) components have been miniaturized and reduced in cost, a prime example being the analog RF chips in cellular telephones. Still, digital processor technology is steadily working its way toward the front of radios and eliminating more and more of the RF technology.

6 6 INTRODUCTION TO DIGITAL TRANSMISSION While the book takes its structure from these evolving trends, we should admit in closing that predicting the future and characterizing the present are hazardous, even pointless exercises. One needs to learn with caution and prepare for the unexpected. Historians of technology tell us that a major innovation is not felt completely for 50 years, and, if so, it is chastening to look at Table 1.1 and realize how much of it is not yet 50 years old. 1.2 WHY DIGITAL? Why indeed is communication more and more digitally based? When the stereo shop says its products are digital quality, is this an advertising slogan, or is digital really better? Let us look at the slogans and the reality and see why this revolution is really occurring. There are many solid reasons, and here they are in rough order of importance. 1. Cheap Hardware. First and foremost, digital hardware has become very cheap, as we have just stressed. This makes all the other advantages cheap to buy. 2. New Semices. We live in an age of , airline booking systems, computer modems, and electronic banking. Whereas voice and images are originally analog and may be transmitted either way, all these newer services are fundamentally digital and must be transmitted symbolically. Whether or not there is a new thirst for data by the human psyche is debatable, but it is clear that widely distributed enterprises and the "virtual workplace" and telephones that work "anytime, anywhere" are key to the way many of us choose to live. These new ways of life are more difficult in an analog world. 3. Control of Quality; Error Control. Here the digital story becomes more subtle. For high-quality music and television, digital format is not necessarily better per se. It is easily demonstrated that any reasonable frequency response, signal-to-noise ratio, and dynamic range may be achieved by analog recording and broadcasting. Where the digital format has a major advantage is in systems that are a chain of many tranmission links. These systems appear in many places, and two important ones are worth summarizing here. First, long-distance voice channels are usually chains of many repeaters. In former times, the links were microwave and nowadays they are more likely fiber-optic, but in either case a transcontinental channel can require links. We will see in the chapters that follow that with analog links the noises add, while with digital links the symbol error probabilities add. What this means is that a given signal-to-noise ratio over 100 analog links requires a ratio in each link that is 10 log , or 20 db, better; we will see that an error probability/? over 100 digital links needs only 1-2 db more energy in each link. The better telephone quality we enjoy over fiber lines is not so much a

7 1.3 CONTENTS OF THE BOOK 7 matter of fibers but rather digital transmission by whatever means when there are repeaters. A second, perhaps more subtle, example of a repeater chain is music recording. Now the chain is a microphone, amplifiers, mixing, more mixing, conversion to a medium, storage, conversion back, amplification, speakers. While this is not a sequence of similar links, there is still a sequence of troublesome points where distortion can enter. Again, digital signal handling has a major advantage. Once a quality is agreed upon and set in the initial analog-to-digital conversion, the digital chain can be designed to carry the music essentially error free to the speaker. 4. Compatibility and Flexibility. Once signals are digitized, it is possible at least in principle to think about transmitting them all by the same shared medium. Network functions such as switching and multiplexing are much easier. New network topologies and modes of multiple access become possible. Control and servicing information can be combined with revenue-producing signals. All sorts of features, from telephone voice mail to disc-player music search, become economic. 5. Cost of Transmission. In certain channels, the cost of digital transmission is less. We have already seen this in the channel with repeaters. Another case where digital transmission wins is a channel with low power, such as the deep-space channel. Since AM schemes cannot have better signal-to-noise ratios than the underlying channel, a bandwidth-expanding modulation such as FM is necessary at the least. This is the energy-bandwidth trade-off. Another alternative is digital transmission, combined with coding: This is trading energy for both complexity and bandwidth. Communication theory shows, in fact, that the most effective use of a weak channel is binary modulation combined with coding. Yet another case where digital modulation can win is a channel with a lot of nearby interferers, such as happens in cellular radio. 6. Message Security. With the growth of mobile telephony and electronic information banks, message security has attracted more attention. Older mobile systems are analog FM, which offers little security. Analog encryption is fundamentally difficult, but the encryption of symbols is just as fundamentally not difficult. 1.3 CONTENTS OF THE BOOK Figure 1.1 shows the overall plan of a digital communication link, and this is the plan of the book as well. To make the plan a little more concrete, the rough parameters of some common data sources are given in Table 1.3. These sources play a role in the examples of the succeeding chapters. Some additional orientation can be obtained from the spectral bands in Table 1.2. The core of a transmission link is of course its transmitter and receiver. It is convenient to break that subject down into baseband signaling, which has no carrier,

8 8 INTRODUCTION TO DIGITAL TRANSMISSION Data source Data sink! Source/channel encoding! : c-6 Source/channel decoding C-6 Modulator C-2, 3 Channel C-5 Demodulator C-2, 3, 7 Network synchronization & control C-4 Phase synchronization & symbol synchronization C-4 Figure 1.1 Plan of a communication link (and plan of the book). and bandpass (or RF) signaling, which works with a carrier. The baseband case is presented in Chapter 2, and here we encounter the ideas of pulse forming, modulation by linear superposition of pulses, modulation spectrum, and modulation probability of error. The last is derived via an elegant and general method called signal space theoiy, a major achievement of communication theory. Chapter 3 is the extension to earner transmission. The ideas of pulses, spectrum, and error probability extend easily to the RF case. Another thrust of the chapter is the parts that are unique to carrier transmission: nonlinear modulations like frequency-shift keying and distortions that occur in RF signaling. Signals must pass through a channel medium. Some of the many channel types are simple wires, coaxial cables, free space, the terrestrial surface and atmosphere, and optical fibers. Each has its own character and distortions, and these are collect- Signal Source TABLE 1.3 Packet Voice, simple Voice, complex coding Phone modem Phone trunk Video, simple Video, complex coding Parameters of Some Representative Digital Signal Sources Analog Bandwidth 4kHz 4kHz 3.4 khz 100 khz 4.5 MHz 4.5 MHz Bits b 64kb/s 4-6kb/s kb/s 1.5Mb/s 40 Mb/s < 1 Mb/s Comments Not real time; retransmission allowed PCM telephone Latest cellular speech coding Quadrature amplitude modulation (QAM) Lowest level line concentrator; 24 one-way lines PCM video Advanced coding; real time

9 1.3 CONTENTS OF THE BOOK 9 ed in Chapter 5. The chapter also provides the opportunity to study a particularly important and challenging medium, the fading mobile channel. Some further material appears in Chapter 7. Probably the most underrated area in transmission design is synchronization. Little time is spent in most books on this subject, yet synchronization of various types probably consumes half of transmission design effort. In reality, there are several layers to the synchronization of a system: An RF system needs receiver carrier synchronization, all systems need to identify the symbol boundaries, and most data transmissions have a frame structure that needs identifying. As well, the entire network needs to stay in synchronization. All of this is in Chapter 4. In addition, the chapter is introduces some of the physical side of networks. Chapters 2-5 are the core of the book. There are a great many allied topics to pursue, and space allows only two of these. Error-correction coding and an introduction to information theory form Chapter 6. Some advanced techniques for distorted and fading channels make up Chapter 7. Communication engineering goes much further, and indeed it is a constant challenge for the working engineer to keep up with developments. Here are some allied fields. Digital signal processing and software engineering play an important role in modern system design. The areas of speech and of image processing are fields in their own right. A rapidly evolving field is communication devices and electronics and RF engineering; unfortunately, it is hard to find courses and consequently engineers in these fields. For those with a theoretical bent, communication theory, detection theory, and information and coding theory are fascinating subjects. Further topics usually taught as separate subjects are antennas, cryptography, and communication networking. The prerequisites for this book are good courses in probability, transforms, linear system theory, and some introduction to analog communication. Stochastic process theory has been kept to a minimum in order to make the book more accessible, but this and other advanced disciplines are sometimes referred to in the text as a guide for the more advanced reader. Some useful notation for the rest of the book as well as some further insight into communication electronics are given in Fig This figure defines the parts of a standard transmitter and superheterodyne receiver. The last, devised by E. Armstrong before 1920, is the basis of almost all receivers and is one of the great inventions of electrical engineering. The receiver begins with a roughly tuned amplifier, the low noise amplifier (LNA), which is designed for moderate gain at the lowest possible noise contribution (for more, see Section 5.2); the LNA sets the noise level in the receiver. Next comes a mixer, whose purpose is to move all signals from their location in the RF spectrum to a single spectrum location, called the intermediate frequency (IF). The IF amplifier and bandpass filters illustrated in Fig. 1.2 achieve two purposes. They hugely amplify the signal to a level suitable for the detection circuit, and they strongly reject all neighboring signals, thus setting the narrowbandpass characteristic of the receiver (a property called selectivity). The genius in this arrangement is subtle to appreciate because it lies buried in the lore of circuit design: The goals of low noise, tunability, gain, and selectivity in circuit design all

10 10 INTRODUCTION TO DIGITAL TRANSMISSION Antenna -RF- Mixer -IF- Baseband Demodulator Decoder Data Wide BPF LNA Narrow High-gain BPF amplifier (D nt LO-Rx» Synchronizer Word ; synchronizer Receiver - Baseband- RF Mixer Data- Encoder Modulator Exciter PA Narrow BPF Transmitter LO-Tx Amplifier PA Power amplifier LNA Low-noise amplifier Oscillator LO Local oscillator IF Intermediate frequency RF Radio frequency Bandpass filter (BPF) Low-pass filter (BPF) Mixer Notation Figure 1.2 Simplified transmitter and superheterodyne receiver, showing notation used in practice and throughout this book. Not all parts are present in all systems. mildly contradict. Somehow, they need to be accomplished in separate, single-purpose circuits. Armstrong was one of the first to appreciate this fact about circuitry and to propose a solution. The top of Fig. 1.2 shows the superheterodyne circuit and the remaining detector, synchronizer, and decoder subsystems, and it also defines some common names and symbols in communication circuitry. These will be used on and off throughout

11 1.4 THE COMPUTER PROGRAMS 11 the book. The electronic design of tuned amplifiers and such blocks as mixers is explained in other books for example, Refs. [2] and [3]. The reminder of the figure shows a simplified transmitter design and some relevant acronyms and notation. Not all transmitters follow this chain: Many lack an exciter and of course many feed fibers or wires rather than antennas. 1.4 THE COMPUTER PROGRAMS A major aid, both for learning and for practice of engineering, is some sort of computing engine, especially one with a vector capability. Perhaps the most common such software tool is MATLAB, 1 but there are several others. For concreteness, MATLAB is used in this book. As the book progresses, short routines are given for the special calculations that arise in each subject. These are called Programs. The Programs and routines for basic operations such as convolution and the fast Fourier transform (FFT) also play a role in many of the homework problems. Software tools are related to the pervasive shift in communication engineering toward software-based signal processing and away from hardware circuitry. Filtering, for example, is increasingly time-domain convolution instead of analog hardware. The Programs and related homeworks help keep a focus on this shift. Rather than include the Programs on a disk at the back of the book, we have kept them short and simply written them out. This brevity is made possible by the vector nature of MATLAB. The fact is that disks-in-the-back seldom work the first time: The reader's platform is wrong, the tool version is wrong, the disk is incompatible, and so on. Murphy's Law is against us here. We believe it is easier for the reader to copy out short programs as need be, making the necessary modifications. Inexpensive student editions of MATLAB are widely available. Experience shows that significant time must be set aside to learn software tools. Not only are there the usual problems with the computer system, but many of us need practice with such underlying techniques as sampling, convolution, and numerical analysis. The problem is not limited to the readers of this book but affects all communication engineers. Time needs to be set aside for these techniques, in addition to that needed for studying digital communication. In closing, we offer some words of encouragement. Digital communication is a subtle subject, but it contains only a few methods and relatively few central lessons, which with a little care and patience can be recognized and mastered. This book has been designed to make that process easier. Here, as in engineering in general, cleverness is not so important as resourcefulness and dedication to the job. The book is not encyclopedic and it leaves out many smaller topics, so that the larger lessons can be stressed more often. Digital communication also has its engineering lore and its experience aspect, and we have included some of this along the way. The reward 'MATLAB is a trademark of The Math Works, Inc. Further information is available from The Math Works, Inc., 3 Apple Hill Drive, Natick, MA USA, orwww.mathworks.com. The simplest available versions of such tools are recommended for this book.

12 12 INTRODUCTION TO DIGITAL TRANSMISSION for studying this book is entry into a fascinating subject that is welcome and thriving everywhere in the world. REFERENCES 1. I. Lebow, Information Highways and Byways, IEEE Press, New York, K. K. Clarke and D. T. Hess, Communication Circuits: Analysis and Design, Addison- Wesley, Reading, MA, H. L. Krauss, C. W. Bostian, and F. H. Raab, Solid State Radio Engineering, Wiley, New York, J. B. Anderson and R. Johannesson, Understanding Information Transmission, IEEE Press, New York, C. E. Shannon, A mathematical theory of communication, Bell Syst. Tech. J., 27, ;