Introduction

A digital communication system is a system that carries information from one place to another by representing that information with discrete symbols, usually bits. A bit is a binary digit, either 0 or 1. A symbol is a chosen signal alternative used to carry one or more bits. For example, a simple transmitter may send one voltage pulse for bit 1 and another voltage pulse for bit 0. A more advanced transmitter may send one of sixteen possible signal points, so that each symbol carries four bits.

The purpose sounds simple: send the intended message and recover it correctly. The difficulty is that real communication links are never perfect. Signals become weaker with distance, noise is added by electronic devices and the environment, bandwidth is limited, clocks are not exactly aligned, radio waves reflect from buildings and terrain, and multiple users may compete for the same spectrum. Digital communication engineering is the discipline of designing systems that still work under those imperfect conditions.

This book is about that discipline.

It begins with signals and probability, then builds toward modulation, detection, coding, synchronization, equalization, OFDM, fading, MIMO, and full system evaluation. The goal is not only to recognize the names of techniques, but to understand why they exist and how they fit together.

The central problem

Every communication system has three essential parts:

1. a source, which produces information;
2. a channel, which carries a physical signal but may distort it or add noise;
3. a receiver, which tries to infer what the source sent.

A source might be a microphone, a camera, a file, a sensor, or a human typing a message. The channel might be a copper cable, an optical fiber, a radio link, or a storage medium such as a magnetic disk or flash memory. The receiver may be a phone, a satellite ground station, a Wi-Fi router, or a decoder inside a solid-state drive.

The important point is that the channel does not carry “information” directly. It carries a physical signal: a voltage, current, electromagnetic wave, optical intensity, or another measurable physical quantity. The receiver observes this physical signal and makes a decision about the message.

For example, suppose a transmitter wants to send a single bit:

• bit 0 is represented by a pulse of amplitude \(-1\);
• bit 1 is represented by a pulse of amplitude \(+1\).

If the receiver observes \(+0.95\), it will probably decide that bit 1 was sent. If it observes \(-1.10\), it will probably decide that bit 0 was sent. But if it observes \(+0.03\), the decision is uncertain. Noise or distortion may have moved the received value close to the decision boundary.

This small example already contains the heart of digital communication:

• represent messages by signals;
• transmit those signals through an imperfect channel;
• observe noisy data at the receiver;
• decide which message was most likely sent;
• measure how often the receiver is wrong.

The mathematical study of this problem was deeply shaped by Claude Shannon’s 1948 paper, which introduced a precise theory of information and channel capacity: the maximum reliable communication rate possible under a given channel model and set of assumptions (Shannon, 1948). Modern digital communication systems are practical engineering answers to the same basic question: how close can we come to reliable communication under real constraints?

Why digital communication is powerful

Digital communication is powerful because it separates information from the exact physical shape of the received waveform.

In an analog communication system, such as a simple amplitude-modulated voice link, the receiver tries to reproduce a continuously varying signal. If noise is added, the output waveform is also noisy. In a digital system, the receiver often needs only to decide among a finite set of possibilities. If the received signal is still close enough to the correct possibility, the receiver can recover the exact bit sequence.

Consider a text message sent over a wireless link. The letters are first represented as bits. Those bits are then protected by coding, mapped to signal symbols, transmitted as radio waves, received with noise and distortion, and processed by the receiver. If the receiver successfully estimates the bits, the text is recovered exactly. The received radio waveform does not need to look exactly like the transmitted waveform; it only needs to contain enough evidence for the receiver to make the right decisions.

This is why digital systems can support error detection, error correction, compression, encryption, packet switching, storage, and computer processing. Once information has been represented as bits, many mathematical and algorithmic tools become available.

But digital communication does not remove physics. It works by respecting physics carefully.

A radio transmitter still has limited power. A channel still has limited bandwidth. Thermal noise still exists. Antennas still have direction, gain, and propagation loss. A receiver still has imperfect oscillators and finite dynamic range. The engineering task is to design the digital layers so that they make efficient use of the physical link.

Bandwidth, power, and reliability

Three constraints appear throughout the book: bandwidth, power, and reliability.

Bandwidth describes how much frequency range a signal or channel occupies. Frequency is measured in hertz, where 1 hertz means one cycle per second. A channel that can pass frequencies from 0 to 1 MHz has a different capability from a channel that can pass frequencies from 0 to 20 MHz. Bandwidth matters because spectrum is scarce and regulated in wireless systems, and because cables and circuits cannot pass arbitrarily fast changes without distortion.

Power is the rate at which energy is transmitted or consumed. In communication systems, higher transmit power often makes the received signal easier to distinguish from noise, but power is limited by batteries, hardware safety, interference regulations, heat, and cost.

Reliability means the receiver recovers the intended information with acceptably low error probability. A common reliability measure is the bit error rate, often abbreviated BER. If a system has a BER of \(10^{-5}\), then on average about one bit out of every \(100{,}000\) received bits is wrong, under the tested conditions. In practice, systems may also measure packet error rate, frame error rate, outage probability, or application-level quality.

These three constraints interact. If we want to send more bits per second through the same bandwidth, we may need a higher signal-to-noise ratio, stronger coding, more complex receivers, or all three. If we lower transmit power, we may need to accept a lower data rate or use more redundancy. If we demand extremely low error rates, we may need additional delay for coding and retransmission.

This interaction is not merely practical; it is fundamental. Shannon showed that, for mathematical channel models, there are limits on reliable communication rate. For an ideal bandlimited additive white Gaussian noise channel, the famous capacity expression is often written as

\[ C = B \log_2(1+\mathrm{SNR}), \]

where \(C\) is capacity in bits per second, \(B\) is bandwidth in hertz, and \(\mathrm{SNR}\) is the received signal-to-noise ratio over that bandwidth, under the assumptions of the model (Shannon, 1948; Proakis and Salehi, 2008). This formula should not be treated as a complete design rule for every channel, but it gives a profound lesson: bandwidth and power can trade against each other, but neither can be ignored.

Noise and uncertainty

A communication receiver never has perfect knowledge. It observes a signal mixed with uncertainty.

One common model is additive white Gaussian noise, abbreviated AWGN. “Additive” means the noise is added to the signal. “Gaussian” means the noise samples have a normal distribution, the bell-shaped distribution widely used in probability theory. “White” means, in an idealized sense, that the noise has equal power per unit bandwidth over the frequency range of interest. AWGN is not the only noise model, but it is central because it is mathematically tractable and often approximates the combined effect of many independent noise sources in receiver electronics (Proakis and Salehi, 2008).

For example, if a transmitter sends amplitude \(+1\), the receiver might observe

\[ r = 1 + n, \]

where \(n\) is a random noise value. If \(n=-0.2\), then \(r=0.8\), and the bit may still be detected correctly. If \(n=-1.3\), then \(r=-0.3\), and the receiver may make an error. Communication theory uses probability to predict how often such errors occur.

This is why the early chapters of the book review random variables, probability distributions, expectation, variance, correlation, and random processes. These are not abstract decorations. They are the language needed to answer practical questions:

• How likely is a bit error?
• How much power is needed to achieve a target BER?
• How should a receiver decide between possible transmitted symbols?
• How much redundancy should an error-correcting code add?
• How does fading change link reliability?

Without probability, digital communication would be only a collection of circuits. With probability, it becomes a quantitative engineering science.

Modulation: turning bits into signals

A bit sequence cannot travel through space by itself. It must be converted into a signal suitable for the channel. This conversion is called modulation.

In a simple baseband system, bits may be represented directly by voltage pulses. In a passband radio system, bits are used to control a high-frequency carrier wave. A carrier is a sinusoidal waveform, such as

\[ \cos(2\pi f_c t), \]

where \(f_c\) is the carrier frequency. By changing the carrier’s amplitude, phase, or frequency, the transmitter embeds information into the waveform.

For example:

• Amplitude-shift keying changes amplitude.
• Phase-shift keying changes phase.
• Frequency-shift keying changes frequency.
• Quadrature amplitude modulation changes both amplitude and phase through in-phase and quadrature components.

A useful mental picture is a constellation, a diagram of possible transmitted signal points. In binary phase-shift keying, or BPSK, there are two points. In quadrature phase-shift keying, or QPSK, there are four points. In 16-QAM, there are sixteen points. The receiver observes a noisy version of one of these points and decides which point was most likely transmitted.

This geometric view is one of the most important ideas in the book. Once signals are represented as points or vectors, distance becomes meaningful. Larger distances between constellation points usually make symbols easier to distinguish in noise, but increasing those distances requires more energy or a larger signal range. This connects modulation design directly to power efficiency and error probability.

Coding: adding redundancy wisely

At first, redundancy may sound wasteful. If we want to send information efficiently, why add extra bits?

The answer is that carefully designed redundancy allows the receiver to detect or correct errors. This is the purpose of channel coding, also called forward error correction when the receiver corrects errors without asking the transmitter to resend the data.

A simple example is a repetition code. Suppose bit 1 is sent as 111 and bit 0 is sent as 000. If the receiver gets 101, it can use majority voting and decide that 1 was probably sent. This code is inefficient because it triples the number of transmitted bits, but it shows the basic idea: redundancy creates structure, and the receiver uses that structure to repair likely errors.

Real codes are much more powerful. Hamming codes, convolutional codes, turbo codes, low-density parity-check codes, and polar codes use mathematical structure to approach fundamental limits more efficiently than simple repetition. Shannon’s noisy-channel coding theorem showed that reliable communication is possible below channel capacity with sufficiently long and well-designed codes, while rates above capacity cannot be made arbitrarily reliable under the model assumptions (Shannon, 1948). Later chapters will explain how practical coding methods attempt to approach this promise.

Coding also illustrates a recurring engineering trade-off. Stronger codes can reduce error probability, but they may increase decoding complexity, latency, and overhead. A satellite link, a fiber-optic backbone, a low-power sensor, and a mobile phone may require different coding choices because their constraints are different.

Synchronization: agreeing on time and phase

Digital receivers must answer a basic question before they can detect symbols: when should the received signal be sampled?

If the transmitter sends one symbol every \(T\) seconds, the receiver must estimate the correct symbol timing. Sampling too early or too late can mix neighboring symbols and increase errors. This problem is called timing synchronization.

Passband systems also need carrier synchronization. The receiver must estimate the carrier frequency and phase well enough to interpret the modulation. If the receiver’s oscillator is slightly offset from the transmitter’s oscillator, the received constellation may rotate over time. If the phase is unknown, a symbol that should appear on the positive real axis may appear somewhere else.

For example, in QPSK the four ideal constellation points may be located at phases \(45^\circ\), \(135^\circ\), \(225^\circ\), and \(315^\circ\). If the receiver has a phase error of \(30^\circ\), every point appears rotated. Small rotations may be corrected by signal processing. Large or rapidly changing rotations can cause decision errors.

Synchronization is often invisible to users because successful systems hide it. But inside a receiver, timing recovery, frequency correction, phase tracking, and pilot processing are essential. A beautiful modulation scheme is useless if the receiver cannot align itself to the signal.

Channels are not always flat or stable

The simplest communication model says the received signal equals the transmitted signal plus noise. Real channels often do more.

A channel may attenuate different frequencies by different amounts. This is called frequency-selective distortion. It can spread one transmitted pulse into neighboring symbol intervals, causing intersymbol interference, or ISI. ISI means that the observation for one symbol is contaminated by energy from previous or future symbols.

A wireless channel may also vary with time. If a receiver moves, or if objects in the environment move, the received radio signal can fluctuate because multiple reflected paths combine constructively and destructively. This phenomenon is called fading. Wireless fading and diversity methods are central topics in modern communication theory, especially for mobile systems (Tse and Viswanath, 2005).

For example, a phone may receive a direct signal from a base station and several reflected copies from buildings. These copies travel different distances, so they arrive with different delays and phases. At one location, they may add strongly. A short distance away, they may partially cancel. This can make the received signal power change even when the transmitter power stays the same.

To handle such channels, communication systems use equalization, interleaving, diversity, OFDM, adaptive modulation, and multiple antennas. These techniques may seem advanced at first, but they all answer simple questions:

• How has the channel changed the signal?
• Can the receiver estimate that change?
• Can the transmitter or receiver compensate for it?
• Can the system use several independent paths so that not all of them fail at once?

The path through this book

The chapters are arranged so that each major idea has the tools it needs before it appears.

First, the book builds the system picture. You will learn what the source encoder, channel encoder, modulator, channel, detector, decoder, and destination do. This gives a map of the full communication chain.

Next, the book develops the mathematical language of signals and noise. Fourier analysis explains how signals occupy frequency. Linear systems explain filtering and channel response. Probability explains uncertainty, Gaussian noise, random processes, and error rates.

Then the book moves into information and source coding. Entropy measures uncertainty in a source, while source coding removes redundancy when compression is desired. This is different from channel coding, which deliberately adds controlled redundancy for reliability. The distinction is important: source coding reduces unnecessary bits; channel coding adds useful protection.

After that, the book studies sampling and quantization. These topics explain how analog signals such as voice and sensor waveforms become digital sequences. Pulse-code modulation is a key example: sample the waveform, quantize each sample, and represent the quantized values by bits.

The middle chapters focus on modulation and detection. You will learn baseband transmission, pulse shaping, signal-space geometry, matched filters, maximum-likelihood detection, and error probability. These chapters are the mathematical core of digital modulation.

The later chapters connect the theory to realistic links. Passband models explain I/Q modulation and complex baseband representation. Synchronization explains how receivers align with transmitters. Link budgets connect power, antenna gain, path loss, noise figure, and signal-to-noise ratio. Equalization handles channel distortion. Channel capacity describes fundamental limits.

Finally, the book studies coding, multiple access, OFDM, fading, diversity, MIMO, and simulation-based evaluation. These topics bring the earlier ideas into the style of modern communication systems.

The sequence matters. OFDM, for example, is much easier to understand after you know Fourier transforms, intersymbol interference, equalization, and complex baseband signals. MIMO is much clearer after you understand vectors, noise, detection, fading, and channel matrices. Error-correcting codes are more meaningful after you understand probability, distance, and reliability.

How to think like a communication engineer

As you read, try to develop three habits.

First, always ask what is random and what is known. Is the transmitted bit known? Not to the receiver. Is the channel known? Sometimes yes, sometimes only statistically. Is the noise sample known? No, but its distribution may be modeled. Clear thinking about knowledge and uncertainty prevents many mistakes.

Second, connect every equation to a physical or operational meaning. If you see energy per bit, ask how much energy is spent to send one information bit. If you see bandwidth, ask what frequency range the signal occupies. If you see probability of error, ask what decision rule produced it and under what channel model.

Third, look for trade-offs instead of perfect solutions. A practical system is rarely “best” in every way. It may be power-efficient but bandwidth-hungry, bandwidth-efficient but sensitive to noise, robust but complex, low-latency but less protected, or high-throughput but demanding on synchronization. Communication engineering is the art of choosing wisely under constraints.

A simple example is modulation order. QPSK sends 2 bits per symbol. 16-QAM sends 4 bits per symbol. If the symbol rate is the same, 16-QAM can carry twice as many bits per second. But its constellation points are closer together for a given average energy, so it generally needs a higher signal-to-noise ratio to achieve the same error rate. There is no free increase in data rate; the cost appears somewhere.

What this book will not assume

This book assumes basic undergraduate mathematics, including algebra, complex numbers, calculus, and introductory linear systems. It will use probability, Fourier analysis, and linear algebra, but it will introduce the communication meaning of those tools as they appear.

The book does not assume that you already know wireless standards or advanced coding theory. It also does not begin with a specific technology such as LTE, 5G, Wi-Fi, or satellite communication. Instead, it builds the principles that make such systems understandable.

When a result depends on an ideal model, the book will say so. For example, AWGN analysis is essential, but real receivers may also face phase noise, nonlinear amplifiers, quantization effects, interference, and fading. Ideal models are not useless; they are controlled starting points. A good engineer knows both what a model reveals and what it leaves out.

The first promise

By the end of this book, a block diagram of a digital communication system should no longer look like a chain of mysterious boxes. Each block should have a purpose:

• source coding reduces unnecessary representation;
• channel coding adds protection against errors;
• modulation maps bits to waveforms;
• pulse shaping controls bandwidth and intersymbol interference;
• the channel attenuates, delays, distorts, and adds noise;
• synchronization aligns time, frequency, and phase;
• equalization compensates channel distortion;
• detection chooses the most likely transmitted symbols;
• decoding uses redundancy to recover the message;
• performance evaluation tells whether the design meets its goals.

The second promise is more important: you will learn to reason quantitatively. Digital communication is not only about naming techniques. It is about predicting, designing, and testing systems that move information reliably through imperfect physical channels.

That is the journey ahead.

References

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

Shannon, C. E. (1948). “A Mathematical Theory of Communication.” Bell System Technical Journal, 27, 379–423, 623–656.

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