Introduction

A digital communication system has one central purpose: to move information from one place to another reliably enough, fast enough, and efficiently enough for the intended use.

That sentence is simple, but it contains much of electrical engineering. The “information” may be a text message, a video stream, a sensor measurement, a banking transaction, a satellite command, or a sequence of samples from a microphone. The “place” may be another chip on the same circuit board, a phone across the city, a spacecraft near Mars, or a server reached through optical fiber. The word “reliably” matters because real channels are imperfect: wires attenuate signals, radios encounter interference and fading, receivers contain thermal noise, and oscillators are never perfectly synchronized. The words “fast” and “efficiently” matter because spectrum, energy, hardware, and time are limited resources.

This book studies how engineers design systems that meet those requirements.

A useful first picture is this:

A source produces a message. A transmitter converts that message into a physical signal. A channel carries the signal but also distorts it and adds uncertainty. A receiver observes the corrupted signal and tries to reconstruct the message. A destination uses the recovered information.

For example, when you send the word hello over a wireless link, the phone does not transmit the letters as tiny printed shapes. It represents the message using bits, such as

01101000 01100101 01101100 01101100 01101111

depending on the character encoding. Those bits are then protected by coding, mapped to waveforms, shifted to a radio carrier frequency, amplified, radiated by an antenna, affected by propagation, received by another antenna, filtered, sampled, detected, decoded, and finally interpreted again as text. The user sees a short word; the communication engineer sees a chain of careful transformations.

The aim of this book is to make that chain understandable.

Why digital communication needs theory

A bit is a binary digit: it can take one of two values, commonly written as 0 and 1. Digital communication systems represent information using bits or groups of bits. This does not mean the physical world becomes digital. A voltage, an optical field, or an electromagnetic wave is still a continuous physical signal. The word digital describes the information representation and many of the processing operations, not the absence of analog physics.

This distinction is important. A digital receiver often observes an analog waveform first. Only after filtering, sampling, synchronization, and detection does it decide which digital symbol was most likely transmitted. For example, in binary phase-shift keying, one waveform may represent bit 0 and another waveform with opposite phase may represent bit 1. Noise may push the received waveform closer to the wrong choice. The receiver’s task is not magical recovery; it is an informed decision under uncertainty.

That uncertainty is why probability appears so early in communication theory. Noise is not usually predictable sample by sample, but it can often be modeled statistically. A common model is additive white Gaussian noise: “additive” because it is added to the signal, “Gaussian” because its amplitudes follow the normal distribution, and “white” because its idealized power is spread uniformly over frequency. This model is not perfect for every channel, but it is fundamental because it captures thermal-noise-limited behavior and leads to clean mathematical results used throughout digital communications (Proakis and Salehi, 2008; Gallager, 2008).

Theory also matters because intuition alone can be misleading. Increasing transmit power can improve reliability, but it may cause battery drain, interference, amplifier distortion, or regulatory problems. Increasing bandwidth can allow faster transmission, but spectrum may be scarce or expensive. Adding error-correcting codes can reduce the error probability, but it introduces redundancy, latency, and implementation complexity. Communication engineering is therefore the art of choosing trade-offs deliberately.

The deepest theoretical result behind these trade-offs is Shannon’s channel coding theorem. In 1948, Claude Shannon introduced a mathematical theory of communication, including measures such as entropy and channel capacity, and showed that reliable communication is possible below a channel’s capacity under suitable coding assumptions (Shannon, 1948). This result did not provide a simple practical code by itself, but it changed the question from “Can reliable communication over noisy channels exist?” to “How close can our real systems get to the fundamental limit?”

Much of modern digital communication can be read as an answer to that question.

Signals are physical carriers of information

A signal is a quantity that varies in one or more independent variables and is used to represent information. In this book, the independent variable is usually time. A microphone voltage changing with time is a signal. A sequence of numbers stored in memory is also a signal. A radio-frequency waveform emitted by an antenna is a signal.

A signal may be continuous-time, meaning it is defined for every time value in an interval, or discrete-time, meaning it is defined only at separated time instants. For example, the acoustic pressure from speech is naturally modeled as a continuous-time signal. The samples produced by an analog-to-digital converter are a discrete-time signal.

Signals also have frequency content. A slowly varying temperature sensor signal occupies very low frequencies. A Wi-Fi waveform occupies a band of radio frequencies. The range of frequencies that a signal or system significantly uses is called its bandwidth. Bandwidth is not merely a mathematical convenience; it is one of the most valuable resources in communication systems. If a system sends 1 megabit per second through approximately 1 megahertz of bandwidth, its raw spectral efficiency is about

\[ \frac{1\,\text{Mbit/s}}{1\,\text{MHz}} = 1\,\text{bit/s/Hz}. \]

This number, bits per second per hertz, is a common way to compare how efficiently systems use spectrum. The exact useful rate depends on overhead, coding, pilots, guard intervals, retransmissions, and other practical details, but the idea is simple: we care not only how fast a system sends data, but also how much bandwidth it consumes.

The relationship between pulses, bandwidth, and reliable symbol transmission has a long history. Nyquist’s work on telegraph transmission analyzed conditions for transmitting pulses without intersymbol interference in idealized bandlimited systems, an idea that later became central to pulse shaping in digital communication (Nyquist, 1928). In this book, we will revisit that idea carefully when we study baseband transmission and raised-cosine filtering.

Communication is decision-making under uncertainty

A receiver usually cannot observe the transmitted message directly. It observes a corrupted version of the transmitted signal. Therefore, receiving is a form of decision-making.

Suppose a transmitter sends one of two possible signal values:

\[ +1 \quad \text{for bit 1}, \]

\[ -1 \quad \text{for bit 0}. \]

If the receiver observes \(0.8\), deciding “bit 1” is natural. If it observes \(-0.9\), deciding “bit 0” is natural. But if it observes \(0.03\), the decision is uncertain. A simple decision rule might be:

\[ \text{decide bit 1 if the observation is positive,} \]

\[ \text{decide bit 0 if the observation is negative.} \]

This is a first glimpse of detection theory. The receiver uses a rule that divides possible observations into decision regions. In later chapters, we will show how such rules are derived from probability models and why, for many important cases, the optimum receiver chooses the signal point closest to the received observation in a geometric signal space.

The word optimum must be used carefully. It always means optimum according to a criterion and a model. A receiver may be optimum for minimizing symbol error probability in additive white Gaussian noise, but not optimum if the channel has impulsive noise, severe synchronization error, or nonlinear distortion. Engineering requires both mathematical precision and awareness of modeling assumptions.

Reliability is measured, not guessed

A digital system is reliable when it makes sufficiently few errors for its application. The relevant measure depends on the system.

The bit error rate or bit error probability describes how often bits are received incorrectly. If a receiver makes 100 bit errors while receiving \(10^8\) bits, the measured bit error rate is

\[ \frac{100}{10^8} = 10^{-6}. \]

A file transfer may require extremely low final error probability, often achieved by combining error correction with error detection and retransmission. A real-time voice system may tolerate occasional bit errors but may not tolerate large delays. A control link for a safety-critical system may require both very high reliability and bounded latency.

This is why there is no single “best” communication system. A deep-space link, a Bluetooth earbud, a fiber-optic backbone, a 5G cellular link, and a low-cost soil-moisture sensor all solve different optimization problems. They may use shared principles—modulation, coding, synchronization, filtering, estimation—but they choose different operating points.

The main resources: power, bandwidth, time, and complexity

Four resources appear again and again.

The first is power. A stronger transmitted signal is usually easier to detect in noise, but power is limited by battery capacity, heat, cost, interference rules, and hardware constraints. A satellite transmitter and a wearable sensor face very different power budgets.

The second is bandwidth. More bandwidth can support higher data rates or simpler signaling, but spectrum is regulated and shared. In wired systems, bandwidth may be limited by the cable, connectors, amplifiers, or allowed electromagnetic emissions. In wireless systems, spectrum allocation is a central design constraint.

The third is time. Communication takes time because symbols have duration, codes process blocks or streams, packets wait in queues, and retransmissions may be needed. The total delay experienced by data is called latency. A video download can tolerate more latency than remote surgery or industrial control.

The fourth is complexity. A mathematically powerful receiver may be too expensive, too power-hungry, or too slow. A decoding algorithm that is excellent in a base station may be unsuitable for a tiny sensor. Practical systems balance performance against implementation.

These resources are connected. For example, error-correcting codes can reduce the required signal-to-noise ratio for a target error rate, which can save power. But coding also adds redundancy and processing. Higher-order modulation can send more bits per symbol and improve spectral efficiency, but it usually requires a higher signal-to-noise ratio for the same error probability. Orthogonal frequency-division multiplexing can simplify equalization in frequency-selective channels, but it introduces a high peak-to-average power ratio and sensitivity to frequency offset.

The chapters of this book are organized around these connections.

The learning path of this book

We begin with the system-level picture. Before studying equations in detail, Chapter 1 explains what each block of a digital communication system does: source encoding, channel encoding, modulation, propagation, detection, decoding, and delivery. This gives you a map.

Chapters 2 and 3 build the mathematical foundation. Signals and spectra explain how information is carried by waveforms and how channels shape those waveforms. Probability explains how to reason about noise, uncertainty, and random messages. These tools are not optional; they are the language of the subject.

Chapter 4 introduces information measures such as entropy and mutual information. Entropy measures average uncertainty in a source, while mutual information measures how much observing one variable tells us about another. These ideas originate in Shannon’s mathematical theory and are central to compression and capacity (Shannon, 1948).

Chapter 5 explains how analog signals become digital through sampling and quantization. This is where we connect continuous physical signals to sequences of numbers. Sampling chooses values at discrete times; quantization maps amplitudes to a finite set of levels. Both operations are powerful, and both have limits.

Chapters 6 through 12 focus on modulation and reception. You will learn pulse shaping, signal constellations, matched filtering, optimum detection, binary and \(M\)-ary modulation, complex baseband modeling, and synchronization. These chapters explain how bits become waveforms and how waveforms become decisions.

Chapters 13 through 15 bring in channel behavior. Real channels attenuate, delay, reflect, disperse, and distort. We study link budgets, noise figure, frequency selectivity, intersymbol interference, and equalization. This is where the clean signal-space picture meets physical propagation and imperfect hardware.

Chapter 16 studies channel capacity and fundamental limits. Capacity tells us the maximum reliable communication rate under a specified channel model. For the additive white Gaussian noise channel, Shannon’s capacity formula reveals the trade-off among bandwidth, power, and noise (Shannon, 1948; Gallager, 2008).

Chapters 17 through 19 study error control coding. Error detection helps us know when something went wrong. Error correction helps us recover the message without retransmission. We begin with block codes and convolutional codes, then survey modern forward error correction such as turbo codes, low-density parity-check codes, and polar codes. These methods are essential in contemporary communication standards, and modern coding theory has brought practical systems much closer to Shannon limits than early systems could reach (Richardson and Urbanke, 2008).

Chapters 20 through 23 study advanced communication methods used in shared and wireless channels: spread spectrum, multiple access, OFDM, fading, diversity, and MIMO. Wireless systems must handle multipath propagation, motion, interference, and many users sharing spectrum. Modern wireless theory treats these effects not only as obstacles but sometimes as resources, especially in diversity and multiple-antenna communication (Tse and Viswanath, 2005).

Chapter 24 integrates the full path through simulation and design practice. You will learn how to generate bit error rate curves, normalize energy per bit, interpret Monte Carlo results, and avoid common simulation mistakes. This final technical chapter is about disciplined engineering judgment.

How to think like a communication engineer

A communication engineer asks questions such as:

What information must be delivered? At what rate? Over what distance? Through what channel? With how much bandwidth? With how much power? With what error probability? With what latency? At what cost?

Then the engineer chooses representations and algorithms. Source coding removes unnecessary redundancy. Channel coding adds carefully designed redundancy for protection. Modulation maps bits to signals. Pulse shaping controls bandwidth. Synchronization aligns clocks and carriers. Equalization fights channel distortion. Detection decides what was sent. Decoding uses code structure to correct errors.

This may sound like many separate topics, but they are connected by a few recurring ideas:

A signal can be represented mathematically.

Noise and messages can be modeled probabilistically.

Distance between signals affects error probability.

Bandwidth and power are limited resources.

Redundancy can be wasteful in source coding but valuable in channel coding.

A receiver should use all available information, but only within practical complexity.

A system should be judged by measurable performance, not by appearance.

Keep these ideas in mind as you read. When a new technique appears, ask: What problem does it solve? What assumption does it rely on? What resource does it consume? What performance measure does it improve?

A small example to carry forward

Imagine a simple wireless sensor that sends one temperature reading per second to a nearby receiver.

The source is the measured temperature. If the sensor reports values such as 23.6°C or 23.7°C, the measurement must first be represented using bits. That is source representation and possibly source coding.

The radio link may corrupt some bits. To protect the message, the transmitter may add parity bits or a stronger error-correcting code. That is channel coding.

The coded bits must become a waveform. The sensor may use a simple binary modulation method, such as frequency-shift keying, where one frequency represents bit 0 and another frequency represents bit 1. That is modulation.

The signal travels through air. It weakens with distance, may reflect from walls, and is mixed with noise at the receiver. That is the channel.

The receiver filters the signal, estimates timing, decides which bits were sent, checks or corrects errors, and outputs the temperature estimate. That is reception, detection, and decoding.

Even this small sensor link involves nearly every major idea in the book. More advanced systems use richer versions of the same principles.

What success will look like

By the end of this book, you should be able to read a digital communication system diagram and understand the purpose of each block. You should be able to explain why noise causes errors, why bandwidth limits pulse shapes, why coding can improve reliability, why synchronization matters, and why wireless channels require special tools such as diversity, OFDM, and MIMO.

You should also be able to compute basic quantities: signal-to-noise ratio, energy per bit, spectral efficiency, symbol rate, simple error probabilities, link margins, and coding parameters. Just as importantly, you should know what those quantities mean physically and when a formula’s assumptions are valid.

Digital communication systems are among the most successful applications of mathematical engineering. They turn probability, linear systems, Fourier analysis, optimization, and physical insight into working technologies that connect people and machines across the world. The subject is demanding, but it is also unusually coherent: once the main ideas fit together, many systems that first looked unrelated begin to look like variations on a common architecture.

We now begin with that architecture.

References

Gallager, R. G. (2008). Principles of Digital Communication. Cambridge University Press.

Nyquist, H. (1928). Certain topics in telegraph transmission theory. Transactions of the American Institute of Electrical Engineers, 47(2), 617–644.

Proakis, J. G., & Salehi, M. (2008). Digital Communications (5th ed.). McGraw-Hill.

Richardson, T., & Urbanke, R. (2008). Modern Coding Theory. Cambridge University Press.

Shannon, C. E. (1948). A mathematical theory of communication. The Bell System Technical Journal, 27, 379–423, 623–656.

Tse, D., & Viswanath, P. (2005). Fundamentals of Wireless Communication. Cambridge University Press.