Digital and Analog Transmission Techniques

Introduction

Data communication is the transfer of information between devices through a transmission medium. Before information can travel through copper, fibre, radio, or another channel, it must be represented as a physical signal. A signal may be digital, with discrete voltage or light levels, or analog, with continuously varying amplitude, frequency, or phase.

Transmission technique selection is an engineering decision. It depends on bandwidth, synchronization, noise and interference, required data rate, implementation cost, compatibility with existing infrastructure, and the ability to regenerate or amplify the signal. Digital techniques are generally preferred for computer data because they support regeneration, error detection, encryption, and integration with digital networks. Analog techniques remain important in radio broadcasting, legacy voice systems, cable systems, and wireless carrier transmission.

This chapter examines line coding, block coding, scrambling, pulse-code modulation, sampling, digital-to-analog modulation, and analog modulation. The central distinction is that encoding changes the representation of information, whereas modulation adapts a signal to a transmission medium by varying a carrier or pulse characteristic.

Learning Objectives

After studying this chapter, you should be able to:

• distinguish digital and analog data and signals;
• explain major line-coding schemes and evaluate their bandwidth and synchronization properties;
• describe block coding and scrambling and explain why they are used together with line coding;
• apply the sampling theorem and calculate PCM bit rates;
• compare ASK, FSK, PSK, QAM, AM, FM, and PM;
• select a transmission technique according to bandwidth, noise resistance, synchronization, and efficiency;
• analyze a practical transmission design and justify its choices.

1. Signals, Data, and Transmission Impairments

Data is the information to be communicated. Data may be digital, such as a binary file, or analog, such as speech from a microphone. A signal is the electrical, optical, or electromagnetic representation of data. Digital data can be represented by digital or analog signals; analog data can also be represented by either type of signal.

A transmission medium limits signal quality. The principal impairments are:

• Attenuation: reduction in signal strength with distance. In decibels, attenuation or gain is commonly expressed as

where P₁ and P₂ are input and output powers in the same units. A negative value indicates loss. For voltage measurements across equal impedances, G_dB = 20 log₁₀(V₂/V₁).

• Distortion: alteration of the signal’s shape because different frequency components experience different attenuation or delay.
• Noise: unwanted energy added to the signal. Thermal noise, crosstalk, impulse noise, and electromagnetic interference are common forms.
• Limited bandwidth: the channel cannot transmit all frequencies equally well. A channel’s bandwidth is the difference between its upper and lower useful frequencies.

For a noiseless channel, the theoretical maximum bit rate is expressed by the Nyquist relationship:

Here R_max is the maximum bit rate in bit/s, B is bandwidth in hertz, and L is the number of discrete signal levels. The result assumes an ideal noiseless channel and ideal signaling. For a noisy channel, Shannon capacity is

where C is channel capacity in bit/s, B is bandwidth in hertz, and S/N is the received signal-to-noise power ratio, expressed as a linear ratio rather than decibels. Shannon capacity is a theoretical upper bound for reliable communication; practical systems operate below it.

2. Digital Signaling and Line Coding

Line coding converts a sequence of bits into a sequence of signal elements suitable for a baseband channel. A signal element may be a voltage level, voltage transition, light intensity, or other physical state. Line coding is not encryption: its purpose is transmission, not secrecy.

A good line code seeks a suitable balance among:

1. low or controlled bandwidth;
2. adequate synchronization information;
3. low direct-current (DC) component;
4. resistance to polarity inversion and noise;
5. simple implementation; and
6. a useful relationship between bits and signal transitions.

2.1 Unipolar and polar schemes

In unipolar NRZ, one voltage represents binary 1 and zero voltage represents binary 0. It is simple but has a substantial DC component and poor clock recovery when long runs of identical bits occur.

In polar NRZ, positive and negative voltages represent the two bit values. Two forms are common:

• NRZ-L: the voltage level represents the bit value.
• NRZ-I: a transition at the start of a bit interval represents one value, while no transition represents the other.

NRZ-I handles long runs of ones better than NRZ-L under its usual convention, but long runs of zeros still provide few transitions. NRZ signals are bandwidth-efficient for short distances, yet they are not ideal where reliable timing recovery is required.

2.2 Return-to-zero and Manchester coding

In RZ coding, the signal returns to a reference level during each bit interval. This creates more transitions and improves synchronization, but it requires more bandwidth.

Manchester coding has a transition in the middle of every bit interval. The direction of the transition identifies the bit. It is self-clocking and has no persistent DC level, but its minimum bandwidth is approximately twice that required by a comparable NRZ signal. Ethernet standards have historically used Manchester coding in some physical layers.

Differential Manchester also guarantees a mid-bit transition. The presence or absence of a transition at the beginning of the bit interval carries the data. Because information is represented by changes, differential Manchester is less sensitive to reversal of the two wires than level-based coding.

2.3 Bipolar coding

In AMI (alternate mark inversion), binary zero is represented by zero voltage, while successive binary ones alternate between positive and negative voltage. AMI reduces the DC component and permits some error detection because an unexpected pair of pulses with the same polarity is a bipolar violation. However, a long run of zeros produces no transitions.

Bipolar schemes such as pseudoternary coding and high-density bipolar variants address particular timing and zero-run problems. Their performance depends on the coding rules and the physical interface.

2.4 Comparison of line codes

The phrase “low bandwidth” must be interpreted carefully. A code with fewer transitions may have a narrower main spectral lobe, but it may also have poor clock recovery and baseline stability. A practical design evaluates the whole physical layer rather than bandwidth alone.

Figure: NRZ-L may remain at one level over several bit intervals, whereas Manchester guarantees a transition in the middle of every interval. The exact transition polarity depends on the adopted convention.

3. Block Coding and Scrambling

Line coding operates on symbols or bits, but it cannot by itself guarantee useful transition density. Two supporting techniques are block coding and scrambling.

3.1 Block coding

Block coding maps a block of m data bits into a block of n coded bits, where n > m. It introduces redundancy. The coding efficiency is

and the overhead fraction is

These expressions assume that all coded bits are transmitted and that additional framing overhead is ignored.

In 4B/5B, every group of four data bits is mapped to a five-bit codeword. The efficiency is 4/5 = 0.80, or 80 percent. The selected codewords avoid excessive runs of zero, thereby helping synchronization. The five-bit output is then line-coded, often using NRZI. Block coding can also reserve codewords for idle, start, stop, or error conditions.

8B/10B maps eight bits to ten bits, giving 80 percent efficiency while controlling run length and balancing ones and zeros. More recent high-speed systems use other codes, such as 64B/66B, which has efficiency 64/66 ≈ 96.97% and relies more heavily on scrambling and synchronization mechanisms.

3.2 Scrambling

Scrambling modifies a bit stream using a deterministic rule so that undesirable patterns, especially long runs of zeros or ones, are replaced with transition-rich patterns. Unlike encryption, a scrambler is normally reversible and is not intended to conceal content.

B8ZS replaces a run of eight zeros with a prescribed pattern containing bipolar violations. HDB3 limits runs of four zeros using substitution patterns. These methods are useful with bipolar physical signaling because intentional violations can be recognized by the receiver and decoded back into zeros.

A scrambler must be designed so that the receiver can reproduce the same sequence or detect the substitution unambiguously. Scrambling improves clock recovery and spectral balance, but it does not correct arbitrary errors and does not guarantee a transition for every possible input unless the protocol specifically provides that property.

4. Sampling and Pulse-Code Modulation

Pulse-code modulation (PCM) converts an analog signal into a digital bit stream through sampling, quantization, and encoding.

4.1 Sampling theorem

If an analog signal is band-limited to a highest frequency f_max hertz, ideal uniform sampling permits exact reconstruction when the sampling frequency satisfies

This is the Nyquist sampling condition. Here f_s is samples per second. The condition assumes a strictly band-limited signal, ideal sampling, and an appropriate reconstruction filter. In practical systems, f_s is usually chosen above 2f_max to allow for non-ideal filters.

If sampling is too slow, frequency components overlap after sampling. This phenomenon is aliasing: a high-frequency component appears as a false lower-frequency component. An anti-aliasing low-pass filter is therefore placed before the sampler.

4.2 Quantization and encoding

Sampling produces discrete-time values, but their amplitudes are initially continuous. Quantization rounds each sample to one of a finite number of amplitude levels. With n bits per sample, the number of quantization levels is

For a uniformly quantized signal with peak-to-peak range V_pp, the quantization step is approximately

where ∆ is in volts if V_pp is in volts. The maximum absolute quantization error for rounding to the nearest level is approximately ∆/2. Quantization introduces distortion that cannot be removed merely by increasing transmission power.

For a full-scale sinusoid and an ideal uniform quantizer, the approximate signal-to-quantization-noise ratio is

This special-case formula assumes an n-bit uniform PCM quantizer, a sinusoidal input using the full range, and quantization noise approximated as uniformly distributed. Speech codecs often use nonuniform quantization or companding, such as μ-law or A-law, to represent low-amplitude signals more effectively.

The PCM bit rate is

where R_b is bit/s, n is bits per sample, and f_s is samples/s. For k independent channels transmitted together,

before framing, error-control, and synchronization overhead.

Worked example: voice PCM

An engineering team digitizes a voice channel limited to 3.4 kHz using 8-bit PCM at 8,000 samples/s. The sampling rate satisfies the theoretical minimum because 8000 ≥ 2 × 3400. The bit rate is

If 24 such channels are multiplexed, the payload rate is 24 × 64 = 1536 kbit/s, before framing and signaling overhead. This explains the historical 64-kbit/s PCM voice channel and the 1.544-Mbit/s payload scale of a T1-oriented system, although actual line rates include additional structure.

5. Digital-to-Analog Modulation

Baseband digital signals cannot efficiently propagate through every medium. A modem therefore maps digital symbols onto an analog carrier. The carrier is characterized by amplitude, frequency, and phase. Digital-to-analog modulation changes one or more of these parameters.

5.1 ASK and OOK

Amplitude-shift keying (ASK) uses different carrier amplitudes for different symbols. On-off keying (OOK) is a binary ASK form in which one symbol is represented by carrier presence and the other by carrier absence. ASK is simple and can be effective in optical links, but amplitude noise directly affects the decision variable.

5.2 FSK

Frequency-shift keying (FSK) represents symbols by different carrier frequencies. Binary FSK uses two frequencies; M-ary FSK uses more. FSK has better amplitude-noise immunity than ASK because information is carried in frequency, but it generally consumes more bandwidth.

5.3 PSK

Phase-shift keying (PSK) changes carrier phase. BPSK uses two phases and carries one bit per symbol. QPSK uses four phases and carries two bits per symbol. With ideal symbol mapping,

where M is the number of distinct symbols. QPSK therefore carries log₂4 = 2 bits per symbol. Gray coding is commonly used so adjacent constellation points differ by one bit, reducing the average bit errors caused by a symbol decision near a boundary.

5.4 QAM

Quadrature amplitude modulation (QAM) varies both amplitude and phase. 16-QAM has 16 constellation points and carries four bits per symbol; 64-QAM carries six bits per symbol. Higher-order QAM increases spectral efficiency but places constellation points closer together, reducing tolerance to noise, distortion, and nonlinear amplification.

The approximate raw symbol rate for a serial stream is

where R_s is symbols/s and R_b is bit/s. This ideal relationship excludes coding, filtering, pulse-shaping, and guard-band overhead. In a bandwidth-limited channel, higher-order modulation can increase bit rate without proportionally increasing symbol rate, but it requires a higher signal-to-noise ratio.

6. Analog Modulation

Analog modulation maps a continuously varying message signal onto a carrier. It is still used in broadcast radio, some instrumentation, legacy systems, and specialized communication links.

6.1 Amplitude modulation

In conventional AM, the carrier amplitude varies with the instantaneous message amplitude while carrier frequency and phase remain nominally constant. For a sinusoidal message, the spectrum contains a carrier and two sidebands. If the message bandwidth is B_m, conventional double-sideband AM requires approximately

The carrier itself conveys no new information but consumes power. A large carrier simplifies envelope detection but reduces power efficiency.

6.2 Frequency and phase modulation

In FM, the instantaneous carrier frequency varies with the message signal. In PM, the instantaneous carrier phase varies. Both can provide strong resistance to amplitude noise because a limiter can remove many amplitude variations before demodulation.

For FM, Carson’s rule estimates occupied bandwidth as

where ∆f is peak frequency deviation and f_m,max is the highest message frequency. Carson’s rule is an engineering approximation, not an exact universal bandwidth boundary. Increasing deviation can improve noise performance but consumes bandwidth.

FM and PM require more complex demodulators than basic AM. They are therefore selected when noise performance and fidelity justify the additional bandwidth or circuitry.

7. Comparative Selection of Techniques

A useful decision process is:

1. Identify the data and medium. Digital baseband over copper may use a line code; a radio channel normally requires carrier modulation.
2. Determine the bandwidth constraint. Manchester and low-order FSK consume more bandwidth than many NRZ or high-order QAM choices.
3. Assess synchronization. Long runs and baseline wander require block coding, scrambling, transitions, or a separate clock.
4. Assess noise and linearity. Amplitude-based methods require a clean amplitude path; PSK and QAM require accurate phase and often linear amplification.
5. Evaluate spectral efficiency and power efficiency together. A high-order constellation saves bandwidth but needs better SNR.
6. Include implementation and compliance requirements. Spectrum masks, permitted power, electromagnetic compatibility, safety, and interoperability may eliminate theoretically attractive choices.

8. Applied Case: Digitizing and Transporting Remote Health-Centre Voice

Background

This illustrative case concerns a rural health centre that must transport two simultaneous voice channels to a regional hospital. The available link is a licensed microwave channel with limited bandwidth and variable interference. The voice source is analog, but the hospital network is digital.

Problem

The design must preserve intelligibility, keep the bit rate manageable, recover timing reliably, and operate under occasional amplitude fading. A simple analog AM link would be easy to understand but would be vulnerable to amplitude interference and would not integrate naturally with the hospital’s digital switching system.

Analysis

The voice band is limited to approximately 3.4 kHz. Sampling at 8 kHz satisfies the theoretical sampling condition. Eight bits per sample gives 64 kbit/s per channel, so two channels require 128 kbit/s of PCM payload before framing and error-control overhead.

The microwave modem cannot use an unstructured NRZ stream because long runs could impair clock recovery. A block code or scrambler is therefore required. QPSK can transmit two bits per symbol, so the ideal symbol rate for 128 kbit/s is

Actual occupied bandwidth is greater than a simple symbol-rate statement because pulse-shaping roll-off, filtering, guard bands, framing, and coding overhead must be included.

Approach and solution

The system uses an anti-aliasing filter, 8-kHz sampling, 8-bit PCM, framing, scrambling, and QPSK modulation. Differential or pilot-assisted carrier recovery is included because phase ambiguity is a practical concern. Forward error correction may be added if the microwave path experiences bursts of errors. The radio power amplifier and filtering are selected to meet the regulatory spectral mask.

Outcome

The proposed architecture preserves a digital interface, supports regeneration and monitoring, and is more resistant to amplitude fading than ASK. It uses less bandwidth than binary FSK at the same information rate, while avoiding the SNR demands of high-order QAM. The final engineering design would still require link-budget calculations, measured error performance, regulatory approval, and testing under worst-case weather conditions.

Learning points

The case demonstrates that no technique is universally best. PCM solves the analog-to-digital representation problem; scrambling solves transition and spectral-pattern problems; QPSK solves carrier transmission over a bandpass channel. The techniques operate at different layers of the overall communication chain and are complementary rather than competing alternatives.

9. Challenges, Limitations, and Contemporary Practice

Digital transmission is not automatically superior in every metric. ADCs, clock-recovery circuits, and digital signal processors consume power and introduce latency. Quantization creates distortion, and high-order modulation can fail rapidly when SNR declines. Analog systems may offer lower latency or simpler hardware in specialized applications.

Modern links commonly combine several techniques: pulse shaping, adaptive equalization, forward error correction, interleaving, scrambling, pilot symbols, and adaptive modulation. Wireless systems may shift between QPSK, 16-QAM, and higher-order QAM according to measured channel quality. Orthogonal frequency-division multiplexing divides a high-rate stream among many subcarriers, allowing equalization of frequency-selective channels. These developments do not remove the basic trade-off between bandwidth, power, noise, and complexity.

AI can assist with channel estimation, modulation classification, interference detection, and adaptive resource allocation. However, an AI-based receiver remains subject to physical-layer constraints. Training data bias, adversarial interference, explainability, safety, and conformance testing must be considered. AI should supplement, not replace, analytical link budgets and standards-based validation.

Professional practice also requires attention to lawful spectrum use, electromagnetic exposure requirements, accessibility of communication services, privacy of transmitted information, and secure management of network equipment. Scrambling must never be represented as encryption; confidential health or financial data require authenticated cryptographic protection.

Best-Practice Checklist

• Filter analog inputs before sampling.
• Choose a sampling rate above the theoretical minimum when real filters require transition bands.
• Use block coding or scrambling when the physical layer needs bounded runs and reliable clock recovery.
• Check both spectral occupancy and DC/baseline behavior of a line code.
• Select modulation order according to measured SNR, not desired throughput alone.
• Include coding, framing, pilots, roll-off, and guard bands in a practical bandwidth calculation.
• Test performance with attenuation, interference, oscillator error, and burst noise.
• Separate transmission coding from security mechanisms.
• Document assumptions, units, regulatory constraints, and fallback modes.

Chapter Summary

Digital and analog transmission techniques represent information in forms suited to particular channels. Line coding creates baseband signal elements and must balance bandwidth, synchronization, DC content, and implementation complexity. NRZ is economical but may lose timing during long runs; Manchester provides reliable transitions at the cost of bandwidth; bipolar schemes reduce DC content but may need zero-run control.

Block coding adds structured redundancy to control patterns and provide special symbols. Scrambling produces transition-rich, spectrally balanced sequences but is not encryption or general error correction. PCM converts analog information through filtering, sampling, quantization, and binary encoding. The sampling theorem requires a sampling frequency at least twice the highest signal frequency under ideal assumptions, while quantization determines amplitude precision and contributes noise.

Digital-to-analog modulation uses ASK, FSK, PSK, or QAM to place digital symbols on a carrier. Higher-order modulation improves bits per symbol and spectral efficiency but requires greater SNR and more accurate receiver operation. Analog AM, FM, and PM remain useful where continuous signals, broadcast conventions, or specialized equipment make them appropriate. Technique selection must consider bandwidth, synchronization, noise resistance, power, complexity, regulation, and interoperability as a combined engineering problem.

Key Terms

Aliasing: Misrepresentation of a high-frequency component as a lower-frequency component caused by insufficient sampling.

Amplitude modulation (AM): Analog modulation in which carrier amplitude follows the message signal.

ASK: Digital modulation that represents symbols using different carrier amplitudes.

Block coding: Mapping of m data bits to n coded bits to introduce controlled redundancy.

Carrier: A periodic signal whose amplitude, frequency, or phase is varied to convey information.

Companding: Nonlinear compression and expansion used to improve quantization performance, especially for speech.

FM: Analog modulation in which carrier frequency varies with the message signal.

Line coding: Conversion of bits into physical baseband signal elements.

Manchester coding: A self-clocking code with a transition in every bit interval.

Modulation: Adaptation of a signal to a channel by varying a carrier or pulse parameter.

NRZ: Non-return-to-zero line coding in which the signal does not necessarily return to a reference level during each bit interval.

PCM: Representation of an analog signal by sampled, quantized, and encoded values.

PSK: Digital modulation that conveys symbols through carrier phase changes.

QAM: Modulation that varies both carrier amplitude and phase.

Quantization: Mapping continuous sample amplitudes to a finite set of levels.

Sampling theorem: Condition requiring a sampling rate at least twice the highest frequency of an ideal band-limited signal.

Scrambling: Deterministic transformation of a bit stream to improve transitions and spectral properties.

Shannon capacity: Theoretical maximum reliable data rate of a noisy bandwidth-limited channel.

Synchronization: Recovery or maintenance of timing and carrier reference between transmitter and receiver.

Symbol: A signal state or waveform representing one or more bits.

Review Questions

Short Answer Questions

1. What is the principal purpose of line coding?
2. Why can long runs of identical bits cause a synchronization problem?
3. Distinguish block coding from scrambling.
4. State the ideal sampling condition for a band-limited signal.
5. Why does higher-order QAM require a higher SNR?

Long Answer Questions

1. Compare NRZ, Manchester, differential Manchester, and AMI with respect to bandwidth, synchronization, DC content, and limitations.
2. Explain the complete PCM process from analog input filtering to binary transmission.
3. Describe ASK, FSK, PSK, and QAM and discuss the trade-off between spectral efficiency and noise tolerance.
4. Compare AM, FM, and PM in terms of varying carrier parameters, bandwidth, noise behavior, and implementation.
5. Develop a technique-selection framework for a digital link operating over a noisy, bandwidth-limited channel.

Analytical / Application Questions

1. A sensor signal is band-limited to 12 kHz and is quantized using 10 bits per sample. Determine the minimum ideal sampling rate and the resulting raw PCM bit rate. State why a practical system might sample faster.
2. A 256-kbit/s stream is transmitted using 16-QAM. Assuming ideal symbol mapping and ignoring overhead, calculate the symbol rate and explain two practical factors that increase the required bandwidth.
3. A remote link experiences long runs of zeros and occasional amplitude fading. Recommend a combination of coding, scrambling, and modulation techniques, and justify the choice using synchronization, bandwidth, and noise considerations.

Multiple Choice Questions with Answers

Question 1

Which line code guarantees a transition in the middle of every bit interval?

A. Unipolar NRZ

B. Manchester

C. AMI

D. Polar NRZ-L

Correct Answer: B. Manchester

Question 2

What is the efficiency of a 4B/5B block code before other overhead is included?

A. 20%

B. 40%

C. 80%

D. 125%

Correct Answer: C. 80%

Question 3

An analog signal has a highest frequency of 15 kHz. What is the minimum ideal sampling frequency?

A. 7.5 kHz

B. 15 kHz

C. 20 kHz

D. 30 kHz

Correct Answer: D. 30 kHz

Question 4

An 8-bit PCM system samples at 8,000 samples/s for one channel. What is its raw bit rate?

A. 8 kbit/s

B. 16 kbit/s

C. 64 kbit/s

D. 128 kbit/s

Correct Answer: C. 64 kbit/s

Question 5

Which technique varies both carrier amplitude and phase?

A. FSK

B. QAM

C. OOK

D. BFSK

Correct Answer: B. QAM

Question 6

What is the main purpose of scrambling in a digital transmission system?

A. To provide confidentiality

B. To increase transition density and improve spectral properties

C. To convert analog data into samples

D. To eliminate all transmission errors

Correct Answer: B. To increase transition density and improve spectral properties

Question 7

How many bits per symbol does ideal 16-QAM convey?

A. 2

B. 4

C. 8

D. 16

Correct Answer: B. 4

Question 8

Which modulation is generally most directly affected by amplitude noise?

A. ASK

B. FSK

C. BPSK

D. PM

Correct Answer: A. ASK

Question 9

For an ideal noiseless channel with bandwidth B and L signal levels, which expression gives the Nyquist maximum bit rate?

A. R_max = B log₂L

B. R_max = 2B log₂L

C. R_max = B/(2 log₂L)

D. R_max = 2L log₂B

Correct Answer: B. R_max = 2B log₂L

Question 10

Why is an anti-aliasing filter placed before an analog-to-digital converter?

A. To increase the number of quantization levels

B. To remove high-frequency components that could alias

C. To provide carrier phase recovery

D. To encrypt the sample values

Correct Answer: B. To remove high-frequency components that could alias