Digital Carrier Modulation

Introduction and Context

Digital carrier modulation is the process of representing digital information by controlled changes in the amplitude, frequency, or phase of a sinusoidal carrier. It is fundamental to wireless communication, satellite links, optical-fibre systems, microwave radio, cellular networks, Wi-Fi, and cable modems. Although a digital source produces symbols at discrete time instants, the transmission medium is often band-pass and therefore cannot efficiently transport an unmodulated baseband waveform. Modulation translates the information into a frequency range suitable for propagation, filtering, multiplexing, and radiation.

A digital communication link must balance several competing requirements: spectral efficiency, energy efficiency, implementation complexity, synchronization accuracy, robustness to fading and interference, and regulatory constraints. No modulation scheme is universally optimal. The appropriate choice depends on channel bandwidth, signal-to-noise ratio, mobility, power availability, hardware linearity, and the required data rate and error performance.

A useful abstraction is

where a_i is the complex symbol coefficient, p(t) is the pulse shape, f_c is the carrier frequency in hertz, M is the number of signal points, and Re{·} denotes the real part. Different selections of a_i produce amplitude-shift keying, phase-shift keying, quadrature-amplitude modulation, and related schemes.

Learning Objectives

After studying this chapter, the reader should be able to:

1. Explain the principles of digital carrier modulation and signal-space representation.
2. Distinguish ASK, FSK, PSK, QPSK, QAM, MSK, and related schemes.
3. Derive relationships among symbol rate, bit rate, bandwidth, and spectral efficiency.
4. Analyse coherent and non-coherent detection using Euclidean distance and signal-to-noise ratio.
5. Evaluate modulation schemes for power efficiency, bandwidth efficiency, fading resistance, and implementation complexity.
6. Apply the concepts to a practical communication-system design problem.
7. Interpret contemporary developments such as adaptive modulation and coded modulation.

1. Digital Symbols, Carriers, and Signal Space

A binary source produces bits, but a modulator usually processes groups of bits called symbols. If each symbol carries k = log₂ M bits, the bit rate and symbol rate are related by

where R_b is the bit rate in bits per second and R_s is the symbol rate, or baud rate, in symbols per second. This relation assumes equally likely symbols and exactly log₂ M information bits per symbol. For nonbinary source coding, framing, pilots, and error-control coding, the gross transmitted bit rate differs from the net information rate.

The carrier may be represented as

where A_c is carrier amplitude, f_c is carrier frequency, and φ_c is carrier phase. Digital modulation changes one or more of these parameters according to the transmitted symbol.

1.1 In-phase and quadrature representation

Using orthogonal basis functions, a passband signal can be written as

where I(t) and Q(t) are the in-phase and quadrature components. The complex envelope is

The two basis functions are orthogonal over an appropriate symbol interval. Consequently, two independent dimensions can be transmitted in the same frequency band. QPSK and QAM exploit these dimensions.

A constellation is a geometric representation of possible complex symbol values. The squared distance between two points is proportional to the energy required to distinguish them in additive white Gaussian noise (AWGN). Larger minimum distance generally improves error performance, whereas more points in a fixed region increase bit rate but reduce point spacing.

2. Amplitude-Shift Keying and On-Off Keying

In amplitude-shift keying (ASK), the carrier amplitude assumes one of several discrete values while frequency and phase remain nominally fixed. Binary ASK may be expressed as

for on-off keying (OOK), where T_b is the bit interval. More generally, binary amplitude levels can be positive and negative, although that form is often interpreted as antipodal signalling and is equivalent to binary PSK under a phase convention.

ASK has simple transmitters and receivers, but amplitude noise, fading, and nonlinear amplifier distortion directly affect the information. OOK is widely used in optical links, infrared remote controls, low-cost radio systems, and passive sensing. It is less attractive where the channel has severe amplitude fluctuations.

With coherent detection in AWGN, equiprobable antipodal binary signalling has bit error probability

where E_b is energy per information bit, N₀ is the one-sided noise power spectral density, and Q(x) is the tail probability of a standard normal random variable. This formula is not the general ASK result; it applies specifically to antipodal signalling with coherent detection in AWGN.

3. Frequency-Shift Keying

In frequency-shift keying (FSK), the carrier frequency is selected from a discrete set. Binary FSK can be written as

where f₀ and f₁ are the two frequencies. The amplitude is constant, which makes FSK suitable for efficient nonlinear power amplifiers.

The frequency separation determines orthogonality and bandwidth. For coherent orthogonal BFSK, the tone separation is commonly selected so that the two signals have zero correlation over the bit interval. A standard result is

for coherent orthogonal BFSK in AWGN. Non-coherent orthogonal BFSK has

under the usual equal-energy and ideal-detection assumptions. Non-coherent detection avoids explicit carrier phase recovery but incurs an energy-performance penalty.

Continuous-phase FSK (CPFSK) constrains phase continuity between symbols. Minimum-shift keying (MSK) is a particularly important CPFSK scheme with modulation index 0.5. It has constant envelope and compact spectral behaviour. Gaussian minimum-shift keying (GMSK) filters the data before MSK modulation with a Gaussian filter, reducing sidelobes at the expense of controlled intersymbol interference. GMSK was used extensively in GSM systems.

4. Phase-Shift Keying

In phase-shift keying (PSK), the carrier phase represents the symbol. M-PSK signals may be written as

where E_s is energy per symbol and T_s is symbol duration. The symbols lie on a circle of constant amplitude.

4.1 Binary PSK

BPSK uses two phases separated by 180°. Its coherent AWGN bit error probability is

BPSK has excellent power efficiency and a simple two-point constellation. Its disadvantages include the need for accurate carrier phase reference and sensitivity to phase ambiguity. A receiver may lock with a 180° ambiguity; differential encoding or a unique-word strategy can resolve it.

4.2 Differential PSK

In differential PSK, information is encoded in the phase difference between consecutive symbols rather than in their absolute phases. Differential detection is simpler because it does not require an absolute carrier phase reference. However, noise affects two consecutive observations, so differential detection generally performs worse than ideal coherent detection in AWGN.

4.3 Quadrature PSK

QPSK uses four phases, commonly separated by 90°, and carries two bits per symbol. One representation is

where I_k,Q_k independently take values proportional to ±1. Gray coding is normally used so adjacent constellation points differ by one bit. In AWGN with coherent detection and Gray coding, the bit error probability is approximately

the same as coherent BPSK when the average energy per bit is equal. QPSK doubles the bit rate of BPSK at the same symbol rate and has the same ideal power efficiency, but its phase transitions can produce amplitude fluctuations after pulse shaping. Offset QPSK delays one branch by half a symbol, limiting phase changes to 90° and reducing envelope variation.

5. Quadrature-Amplitude Modulation

Quadrature-amplitude modulation (QAM) varies both in-phase and quadrature amplitudes. A rectangular square-QAM constellation has M points, often with M = 2²ⁿ. For square M-QAM with Gray coding and coherent detection in AWGN, a widely used approximate bit-error expression is

This approximation is accurate at moderate and high signal-to-noise ratios for square QAM with equally likely symbols. It should not be used without modification for cross-QAM, probabilistic shaping, severe fading, or low-SNR regimes.

QAM is spectrally efficient because it places many points in two dimensions. However, increasing M reduces minimum Euclidean distance for a fixed average power. High-order QAM therefore requires high signal-to-noise ratio, accurate amplitude and phase estimation, linear amplifiers, and low phase noise. It is used in cable modems, DSL, Wi-Fi, LTE, 5G, microwave radio, and digital television.

6. Pulse Shaping and Bandwidth

Symbols must be shaped before transmission. A rectangular pulse has a narrow main lobe but large sidelobes and poor spectral containment. The raised-cosine family provides a controlled trade-off between bandwidth and time-domain transition speed. With roll-off factor α, where 0 ≤ α ≤ 1, the ideal baseband bandwidth is

for a one-sided low-pass bandwidth. The corresponding passband occupied bandwidth is approximately

assuming a real passband signal and ideal filtering. Root-raised-cosine filters are commonly divided equally between transmitter and receiver; their cascade produces the raised-cosine response.

The Nyquist zero-intersymbol-interference condition requires the overall pulse response to satisfy

and a nonzero value at n = 0, in the ideal timing case. Practical systems tolerate controlled interference and use adaptive equalization.

The uncoded spectral efficiency is

in bits per second per hertz, under the ideal raised-cosine assumptions. Coding, pilots, guard intervals, cyclic prefixes, framing, and spectral masks reduce net efficiency.

Worked Example 1: Symbol rate and bandwidth

A 16-QAM link transmits a gross bit rate of 20 Mb/s using a roll-off factor of 0.25. Since log₂ 16 = 4,

The approximate RF bandwidth is

This is an ideal occupied-bandwidth estimate. Practical filtering, guard bands, coding overhead, and spectral masks may change the assigned channel bandwidth.

7. Detection, Synchronization, and Error Performance

A coherent receiver estimates carrier frequency, phase, symbol timing, and channel amplitude before making decisions. Its matched filter maximizes the signal-to-noise ratio at the sampling instant in AWGN. The receiver then selects the constellation point nearest to the observed complex sample:

where r is the received sample and s_i is a candidate signal point. This minimum-distance rule is optimal for equally likely symbols in AWGN with equal covariance.

Non-coherent receivers do not require an absolute carrier phase reference. They are attractive in low-cost or rapidly varying systems but generally sacrifice error performance or bandwidth efficiency. Differential detection is one important non-coherent approach.

The signal-to-noise ratio per bit is

where γ_b is dimensionless. For coded systems, energy per coded bit and energy per information bit must be distinguished. If the code rate is R_c, measured in information bits per coded bit, then approximately

provided the energy definitions are consistent.

In a fading channel, the instantaneous SNR varies with time. Average AWGN formulas then cannot directly predict error rate; performance must be averaged over the fading distribution or evaluated using simulation or measured channel traces. Diversity, interleaving, channel estimation, equalization, and forward-error correction are often more important than selecting between nearby constellation types.

8. Comparative Design Trade-offs

The peak-to-average power ratio (PAPR) is especially significant in multicarrier systems. OFDM combines many subcarriers and may have high PAPR even when each subcarrier uses QAM. High PAPR requires amplifier back-off, lowering power efficiency. Single-carrier frequency-domain equalization and special waveforms can reduce this problem.

9. Adaptive Modulation and Coded Modulation

Modern systems rarely use a single fixed modulation order. Adaptive modulation selects BPSK, QPSK, 16-QAM, 64-QAM, 256-QAM, or higher orders according to channel quality. A scheduler may choose a modulation-and-coding scheme using channel-quality indicators, target block-error rate, latency requirements, and transmit-power limits.

Adaptive modulation improves throughput when the channel is favourable and preserves reliability when it is poor. Its effectiveness depends on channel-estimation accuracy, feedback delay, user mobility, and the design of switching thresholds. An outdated channel estimate can cause a high-order constellation to be selected during a deep fade.

Coded modulation combines modulation and error-control coding. In bit-interleaved coded modulation, coded bits are interleaved and mapped to constellation points. The mapping affects the reliability of bit positions; soft-demodulator outputs are therefore important to modern decoders. Trellis-coded modulation integrates coding and constellation expansion, achieving coding gain without a proportional increase in bandwidth.

Probabilistic constellation shaping changes the probability of transmitted points so that low-energy points occur more frequently. It can approach the theoretical capacity of the Gaussian channel while permitting flexible rate adaptation. Its practical design must account for distribution matching, finite block lengths, nonlinear hardware, and decoder complexity.

10. Practical Transmitter and Receiver Architecture

A typical digital carrier transmitter performs source processing, channel coding, interleaving, symbol mapping, pulse shaping, digital upconversion, digital-to-analogue conversion, frequency conversion, power amplification, and radiation or guided transmission. The receiver reverses these operations: filtering, downconversion, analogue-to-digital conversion, synchronization, matched filtering, channel estimation, equalization, demapping, deinterleaving, decoding, and source reconstruction.

Important impairments include carrier-frequency offset, phase noise, sampling-clock offset, IQ imbalance, DC offset, nonlinear amplification, multipath propagation, Doppler shift, and quantization noise. A constellation diagram is a practical diagnostic tool: radial spreading suggests amplitude noise, angular spreading suggests phase noise, rotation suggests frequency offset, and systematic elliptical distortion may indicate IQ imbalance.

Best practices include:

• specify whether rates are gross or net;
• state the roll-off factor and occupied-bandwidth definition;
• use Gray mapping unless another mapping is justified;
• reserve pilots or training sequences for synchronization and channel estimation;
• evaluate BER and block-error rate over the intended channel model;
• include amplifier back-off, filtering, coding, and implementation losses in link budgets;
• verify spectral masks and coexistence requirements through measurement;
• test worst-case frequency offset, phase noise, fading, and temperature variation.

11. Applied Case: Adaptive Modulation in a Microwave Backhaul Link

Background

Consider an illustrative point-to-point microwave backhaul link connecting a cellular base station to the core network. The channel bandwidth is 28 MHz, and the link must support variable traffic while maintaining a block-error rate below a specified operational target. Rain attenuation and multipath fading cause the received SNR to vary.

Problem

A fixed 256-QAM mode provides high capacity under clear-sky conditions but experiences unacceptable outages during rain. A fixed QPSK mode is reliable but wastes capacity when the channel is strong. The engineering problem is to design a robust adaptive strategy.

Analysis

Assume a raised-cosine roll-off factor of 0.2. The approximate symbol capacity of a 28 MHz channel is

Ignoring coding and framing overhead, the gross bit rates are approximately:

These values are planning estimates, not guaranteed throughputs. Coding overhead, pilots, guard intervals, and retransmissions reduce the net rates.

Approach and solution

The modem maintains several modulation-and-coding modes. It estimates post-equalization SNR, applies hysteresis to prevent rapid mode switching, and selects the highest mode whose measured error statistics satisfy the target. Strong interleaving and forward-error correction handle short fades, while automatic transmit-power control provides additional margin where regulations permit.

The system also uses a fallback mode. When channel quality deteriorates, it moves from 256-QAM to 64-QAM, 16-QAM, or QPSK before the error rate becomes unacceptable. The switching thresholds are determined experimentally using the actual radio, antenna, amplifier, and propagation environment rather than by AWGN theory alone.

Outcome

The adaptive link achieves high average throughput during clear conditions and maintains service continuity during moderate attenuation. Its performance depends on feedback latency: if the channel changes faster than the adaptation loop, the modem may select a mode that is too aggressive. Operational monitoring should therefore record mode occupancy, SNR margin, retransmissions, outage duration, and spectral compliance.

Learning points

The case illustrates that modulation order is a system-level decision. Higher-order QAM is not automatically better; it is beneficial only when the channel, amplifier, synchronization, coding, and regulatory conditions support it. A robust design combines modulation adaptation with coding, link-margin analysis, measurement, and fault management.

12. Limitations, Regulation, and Contemporary Directions

Digital modulation theory often begins with AWGN and ideal synchronization, whereas real channels contain interference, nonlinearities, multipath, mobility, and hardware imperfections. Closed-form BER expressions may therefore be useful benchmarks rather than final performance predictions. Simulation should use validated channel models, and field trials should confirm assumptions.

Regulatory authorities allocate spectrum and impose limits on occupied bandwidth, out-of-band emissions, adjacent-channel leakage, radiated power, and electromagnetic compatibility. Engineers must also consider lawful operation, coexistence with other users, exposure requirements, and secure handling of transmitted data. Modulation itself does not provide confidentiality; encryption and authentication must be implemented at appropriate protocol layers.

Current research includes highly shaped constellations, machine-learning-assisted demodulation, joint communication and sensing, reconfigurable intelligent surfaces, optical coherent modulation, terahertz links, and energy-efficient hardware. Machine learning may assist impairment compensation or adaptive selection, but it must be evaluated for generalization, computational cost, explainability, data dependence, and safety under distribution shifts. Classical signal-processing methods remain essential because they provide interpretable guarantees and predictable behaviour.

Chapter Summary

Digital carrier modulation maps discrete symbols onto controlled carrier amplitude, frequency, phase, or combinations of these parameters. ASK is simple but amplitude-sensitive; FSK provides constant-envelope operation at a bandwidth cost; PSK offers strong power efficiency; QPSK doubles the bits per symbol of BPSK without sacrificing ideal coherent BER; and QAM provides high spectral efficiency at the cost of higher SNR, linearity, and synchronization demands. Pulse shaping controls bandwidth and intersymbol interference, while the roll-off factor determines the ideal occupied bandwidth. Constellation geometry, especially minimum distance, links modulation design to error probability. Practical systems must address synchronization, fading, nonlinear amplification, coding, filtering, feedback, and regulatory constraints. Adaptive modulation and coded modulation allow communication links to respond intelligently to changing channel conditions.

Key Terms

Digital carrier modulation: Representation of digital symbols through controlled carrier parameters.

Symbol: One selected element of a modulation alphabet.

Symbol rate: Number of transmitted symbols per second.

Bit rate: Number of transmitted bits per second.

ASK: Modulation in which carrier amplitude represents the symbol.

OOK: Binary ASK in which one symbol is represented by carrier absence.

FSK: Modulation in which carrier frequency represents the symbol.

PSK: Modulation in which carrier phase represents the symbol.

BPSK: Two-phase phase-shift keying.

QPSK: Four-phase modulation carrying two bits per symbol.

QAM: Modulation using both in-phase and quadrature amplitudes.

Constellation: Geometric set of possible complex symbol points.

Coherent detection: Detection using a phase- and frequency-referenced carrier.

Non-coherent detection: Detection without an absolute carrier-phase reference.

Pulse shaping: Filtering used to control bandwidth and intersymbol interference.

Roll-off factor: Parameter describing excess bandwidth of a raised-cosine response.

Spectral efficiency: Information rate per unit bandwidth.

Minimum Euclidean distance: Smallest distance between constellation points.

Adaptive modulation: Dynamic selection of modulation according to channel conditions.

Coded modulation: Joint design of modulation and error-control coding.

Review Questions

Short Answer Questions

1. Define digital carrier modulation and explain why it is required in a band-pass communication system.
2. Distinguish bit rate from symbol rate.
3. Why does QAM generally require a more linear power amplifier than FSK?
4. State the purpose of a root-raised-cosine filter.
5. What is the role of Gray coding in a QAM or PSK constellation?

Long Answer Questions

1. Explain ASK, FSK, PSK, and QAM, comparing their bandwidth, power, and implementation characteristics.
2. Derive the relationship among modulation order, bit rate, symbol rate, and ideal raised-cosine bandwidth.
3. Discuss coherent and non-coherent detection, including synchronization requirements and performance trade-offs.
4. Explain constellation geometry and its relationship to AWGN error performance.
5. Describe the architecture and major impairments of a practical digital carrier transmitter and receiver.

Analytical / Application Questions

1. A 64-QAM system transmits 72 Mb/s with a roll-off factor of 0.25. Calculate its symbol rate and approximate RF bandwidth, ignoring coding and framing overhead.
2. A link can operate in QPSK, 16-QAM, or 64-QAM modes. Propose an adaptive-modulation policy for a slowly varying fading channel and identify the measurements needed to validate it.
3. A high-order QAM microwave link shows constellation rotation, radial spreading, and occasional clusters between ideal points. Diagnose likely impairments and recommend a measurement and mitigation plan.

Multiple Choice Questions with Answers

Question 1

For an M-ary modulation scheme carrying equally mapped uncoded symbols, how many bits are represented by each symbol?

A. log_2 M

B. M bits

C. 2M bits

D. M/2 bits

Correct Answer: A. log_2 M

Question 2

Which modulation scheme varies both the in-phase and quadrature amplitudes?

A. BFSK

B. BPSK

C. QAM

D. OOK

Correct Answer: C. QAM

Question 3

A 16-QAM system carries how many bits per symbol?

A. 2

B. 4

C. 8

D. 16

Correct Answer: B. 4

Question 4

For a raised-cosine passband signal, what is the ideal approximate RF bandwidth in terms of symbol rate R_s and roll-off factor α?

A. R_s/(1+α)

B. (1+α)R_s

C. αR_s/2

D. 2R_s/(1+α)

Correct Answer: B. (1+α)R_s

Question 5

Which scheme has a constant envelope in its ideal transmitted waveform and is therefore well suited to nonlinear power amplifiers?

A. High-order QAM

B. ASK

C. FSK

D. Rectangular 256-QAM

Correct Answer: C. FSK

Question 6

What is the principal purpose of Gray coding in a constellation?

A. To increase carrier frequency

B. To ensure adjacent points differ by one bit

C. To eliminate the need for filtering

D. To make all symbols have different energies

Correct Answer: B. To ensure adjacent points differ by one bit

Question 7

A 20 Mb/s 16-QAM link has a roll-off factor of 0.25. Ignoring overhead, what is its approximate RF bandwidth?

A. 4.00 MHz

B. 5.00 MHz

C. 6.25 MHz

D. 8.00 MHz

Correct Answer: C. 6.25 MHz

Question 8

Which receiver rule is optimal for equally likely signal points in AWGN under the standard Euclidean model?

A. Choose the point with the highest carrier frequency

B. Choose the nearest constellation point

C. Choose the point with maximum amplitude only

D. Choose the point transmitted in the previous symbol

Correct Answer: B. Choose the nearest constellation point

Question 9

Why does increasing QAM order usually require a higher SNR?

A. It reduces the carrier frequency to zero

B. It increases symbol duration automatically

C. It places more points closer together for a fixed average power

D. It removes the quadrature component

Correct Answer: C. It places more points closer together for a fixed average power

Question 10

What is the primary advantage of adaptive modulation?

A. It keeps the modulation fixed under all channel conditions

B. It selects modulation order according to channel quality

C. It eliminates the need for channel estimation

D. It guarantees zero error without coding

Correct Answer: B. It selects modulation order according to channel quality