Radio Propagation and Channel Models

Introduction and Context

Wireless communication replaces a guided transmission medium with an unguided electromagnetic channel. The radio signal travels through an environment containing buildings, terrain, vegetation, vehicles, atmospheric layers, and other transmitters. Consequently, received power is not determined only by transmitter power and distance. It also depends on antenna characteristics, frequency, polarization, reflection, diffraction, scattering, motion, and bandwidth.

A channel model is a mathematical representation of these effects. Channel models support coverage planning, link-budget calculations, antenna selection, modulation and coding design, diversity analysis, multiple-input multiple-output (MIMO) evaluation, and system simulation. No single model is universally correct: a model must match the propagation environment, frequency range, antenna configuration, mobility, and required accuracy.

Learning Objectives

After studying this chapter, a student should be able to:

1. Explain the mechanisms of reflection, diffraction, scattering, absorption, and penetration.
2. Calculate free-space and empirical path loss.
3. Distinguish large-scale shadowing from small-scale fading.
4. Explain multipath delay spread, coherence bandwidth, Doppler shift, and coherence time.
5. Construct and interpret a wireless link budget.
6. Describe antenna gain, effective aperture, polarization loss, and radiation patterns.
7. Select suitable deterministic, empirical, and statistical channel models.
8. Evaluate practical limitations and design trade-offs in a wireless link.

1. The Radio Channel as a Physical System

A transmitted signal reaches a receiver through one or more propagation paths. The received signal may be represented conceptually as

where s(t) is the transmitted waveform, h(t) is the channel impulse response, * denotes convolution, n(t) is additive noise, and r(t) is the received waveform. If the channel changes with time, the response is written h(t,τ), where τ is excess delay.

A useful decomposition separates three scales:

• Path loss: average power reduction with distance and frequency.
• Shadowing: slower variation caused by obstacles and terrain.
• Small-scale fading: rapid amplitude and phase variation caused mainly by multipath and motion.

These effects are statistically distinct but occur simultaneously in a real link.

Propagation mechanisms

Reflection occurs when a wave encounters a surface large relative to its wavelength, such as a building wall or the ground. The reflected wave has a changed phase and generally changed amplitude. Conductive materials and incidence angle strongly influence reflection.

Diffraction permits energy to reach a region hidden behind an obstruction, such as a hill or rooftop. It is especially important when a direct line of sight is blocked. Knife-edge diffraction is a useful first approximation, although real obstructions have finite thickness and irregular shapes.

Scattering occurs when the environment contains objects comparable to or smaller than the wavelength, including foliage, lamp posts, rough walls, and signs. Scattering produces many weak paths arriving with different phases.

Absorption and penetration loss convert part of the electromagnetic energy into heat or otherwise attenuate it. Walls, glass, foliage, rain, and the human body can introduce substantial frequency-dependent loss. Penetration loss is often modeled as an additional empirical term rather than derived from ideal free-space theory.

2. Wavelength, Frequency, and the Link Geometry

Frequency and wavelength are related by

where λ is wavelength in metres, c ≈ 3 × 10⁸ m/s is the speed of light in free space, and f is frequency in hertz. Higher frequency means shorter wavelength. For example, at 2.4 GHz, the wavelength is approximately 0.125 m.

The first Fresnel zone is the ellipsoidal region around the direct path in which unobstructed propagation is desirable. For a path with transmitter-to-obstacle distance d₁ and obstacle-to-receiver distance d₂, the radius of the nth Fresnel zone at the obstacle is approximately

Here n is a positive integer. The first zone, n = 1, is most important. Significant obstruction of it can cause diffraction loss even when transmitter and receiver appear visually aligned.

3. Path Loss Models

3.1 Free-space propagation

For an unobstructed line-of-sight link with isotropic antennas, the Friis transmission equation is

where P_t and P_r are transmitted and received powers in watts, G_t and G_r are dimensionless antenna gains in the link direction, d is separation in metres, and λ is wavelength. The equation assumes far-field operation, matched polarization, free space, line-of-sight propagation, and no obstruction or multipath.

The free-space path loss is the ratio of transmitted power to received power when antenna gains are excluded:

In decibels,

Using distance in kilometres and frequency in megahertz gives the practical form

The constant is approximately 32.44 for these units. Free-space loss increases by 6.02 dB whenever distance or frequency is doubled.

3.2 Log-distance model

In real environments, average received power usually decreases according to an environment-dependent path-loss exponent rather than exactly the free-space square law. The log-distance model is

where d₀ is a reference distance in the far field, PL(d₀) is measured or calculated path loss at that distance, and n is the path-loss exponent. Typical approximate values are 2 in free space, 1.6–2.0 in some corridors, 2.7–3.5 in urban areas, and 3–5 in obstructed indoor or difficult environments. These ranges are indicative, not universal.

3.3 Log-normal shadowing model

Shadowing is commonly added as a zero-mean Gaussian random variable in decibels:

where X_σ is normally distributed with mean zero and standard deviation σ dB. Thus, shadowing is log-normal when expressed as a linear power factor. The model describes large-scale variation over locations separated by distances larger than the local fading scale. Measurements are normally required to estimate n and σ.

3.4 Empirical and deterministic models

Empirical models, such as Okumura–Hata and COST-type models, use frequency, distance, antenna heights, and correction factors fitted to measurements. They are efficient for network planning but valid only within their specified frequency, distance, height, and environment ranges.

Deterministic models use geographical databases, ray tracing, reflection, diffraction, and scattering calculations. They can provide spatially detailed predictions but require accurate maps, material properties, antenna patterns, and substantial computation. A model that is more complex is not automatically more accurate if its environmental inputs are poor.

4. Multipath and Small-Scale Fading

A transmitted symbol may arrive through many paths. If the paths have delays τ_i and complex gains a_i, a time-varying multipath channel can be represented as

where δ(·) is the impulse function. The received components add as complex phasors. Constructive addition produces a signal peak; destructive addition produces a deep fade. A movement of a fraction of a wavelength can therefore cause a major power change, especially at higher frequencies.

Flat and frequency-selective fading

Let B_s denote signal bandwidth and B_c denote coherence bandwidth. If B_s ≪ B_c, all significant frequency components experience approximately the same fading, called flat fading. If B_s > B_c, different frequencies experience different gains and phases, called frequency-selective fading. Frequency-selective fading can cause intersymbol interference (ISI).

The power delay profile describes average received power as a function of excess delay. Important quantities include mean excess delay and root-mean-square (RMS) delay spread. If P(τ) is the delay-power profile, then

and

for discrete paths. The quantities P_k and τ_k are path power and excess delay. Approximate coherence-bandwidth rules depend on the correlation criterion; a commonly used engineering estimate is B_c ≈ 1/(5σ_τ), while stricter correlation requirements may use a smaller coefficient such as 1/50. These are approximations, not universal identities.

Rayleigh and Rician fading

When there is no dominant line-of-sight component and many scattered components have independent phases, the envelope is often modeled by a Rayleigh distribution:

Here r is envelope amplitude and σ_h is the scale parameter. The model assumes many contributions, approximately independent phases, and no dominant deterministic component.

When a dominant line-of-sight or specular component exists in addition to diffuse components, a Rician model is often used. Its K-factor is

where the powers are linear, not decibel, quantities. In decibels, K_dB = 10 log₁0 K. Large K indicates a strong dominant component; K → 0 approaches Rayleigh behavior. Nakagami-m is another flexible empirical envelope model used when measured fading does not fit Rayleigh or Rician behavior well.

5. Doppler Effects and Time Selectivity

Relative motion changes the path length and therefore the received phase. For a single wave arriving at angle θ relative to the direction of motion, the Doppler shift is

where v is relative speed in metres per second. The maximum Doppler frequency is

The received spectrum is spread around the carrier, producing time-selective fading. A rough engineering estimate of coherence time is

for a commonly used correlation convention. Other definitions produce different constants; coherence time should therefore be treated as an order-of-magnitude measure.

If a symbol duration is much shorter than coherence time, the channel may be treated as approximately constant over that symbol. If channel variation is rapid relative to signaling, channel estimation and tracking become more difficult. Doppler is important in vehicular networks, high-speed rail, satellite links, and links involving moving users or objects.

6. Antenna Characteristics

An antenna converts guided electrical energy to electromagnetic radiation and vice versa. Important characteristics include:

• Radiation pattern: angular distribution of radiated power.
• Beamwidth: angular width, commonly measured between half-power points.
• Gain: directivity including radiation efficiency.
• Polarization: orientation and time behavior of the electric-field vector.
• Efficiency: ratio of radiated power to accepted input power.
• Front-to-back ratio: forward radiation relative to backward radiation.

Antenna gain in a particular direction is related to directivity and efficiency by

where η is dimensionless efficiency and D is directivity. Gain is often expressed in dBi relative to an isotropic radiator or dBd relative to a half-wave dipole. The conversion is approximately 0 dBd = 2.15 dBi.

The effective aperture of a receiving antenna is

where A_e is in square metres and G_r is dimensionless gain. This relation explains why frequency, antenna gain, and captured power are linked. Polarization mismatch introduces a polarization loss factor between zero and one; ideal matched polarization has factor one, whereas orthogonal linear polarizations ideally have zero received power.

7. Link Budgets

A link budget accounts for gains and losses between transmitter and receiver. In decibel form,

where L_p is propagation loss and L_other includes feeder, connector, polarization, penetration, implementation, and other losses. All additive terms must use consistent logarithmic units.

The receiver sensitivity is the minimum required received power for a specified data rate, modulation, coding, and target error performance. A link margin is

A positive margin does not guarantee service because shadowing, fading, interference, blockage, and model uncertainty may reduce instantaneous performance.

Worked example: outdoor line-of-sight link

Suppose a transmitter radiates P_t = 20 dBm at f = 2.4 GHz, with transmitter and receiver gains of 8 dBi and 2 dBi. The separation is 1 km. Cable and connector losses total 2 dB, and receiver sensitivity is −82 dBm.

Using the practical free-space equation with f = 2400 MHz and d = 1 km,

L_FS = 32.44 + 20 log₁0(2400) + 20 log₁0(1) ≈ 100.04 dB.

Therefore,

P_r = 20 + 8 + 2 − 100.04 − 2 ≈ −72.04 dBm.

The estimated link margin is M = −72.04 − (−82) ≈ 9.96 dB. This is a nominal free-space result. A practical design should reserve additional margin for shadowing, rain where relevant, antenna misalignment, and multipath fading.

8. Statistical Channel Models

A statistical model describes channel behavior using probability distributions and correlation properties rather than a detailed map. Common model families include:

A tapped-delay-line model approximates the channel as several delayed taps. Each tap may have a specified average power, delay, Doppler spectrum, and correlation. Such models are useful in link-level simulation because they reproduce delay spread and time variation without simulating every physical object.

A channel model should be selected by asking: Is the objective coverage, outage probability, equalizer design, MIMO throughput, or physical interpretation? Is the link indoor, urban macrocell, rural, highway, satellite, or industrial? What carrier frequency, bandwidth, antenna arrangement, and mobility apply? What measurement data validate the model?

9. Applied Case: Planning a Campus 2.4 GHz Wireless Link

Background. A university wishes to connect an outdoor laboratory to a network building 350 m away. The path includes several trees and a partial rooftop obstruction. Directional antennas are available, but the link must remain reliable during moderate wind and seasonal foliage changes.

Problem. A simple free-space calculation may overestimate performance because it ignores vegetation, Fresnel-zone obstruction, and shadowing.

Analysis. Engineers first calculate the free-space baseline at 2.4 GHz and then inspect the first Fresnel zone. They obtain a path-loss estimate using a site-appropriate log-distance or empirical model, add measured or conservatively estimated vegetation and equipment losses, and evaluate antenna alignment and polarization. A short measurement campaign records received signal strength at several nearby locations and under different foliage conditions. The data are used to estimate a path-loss exponent and shadowing standard deviation.

Approach and solution. The design uses directional antennas with adequate beamwidth, weatherproof low-loss cables, matched polarization, and a radio supporting adaptive modulation and coding. The link budget includes a target fade margin rather than relying only on mean received power. The rooftop obstruction is reduced where practical, and the antennas are mounted to improve Fresnel clearance. If the required margin cannot be achieved, a relay or a lower-frequency alternative is considered.

Outcome. The final decision is based not on a single received-power prediction but on an outage estimate that includes large-scale variability and expected fading. Periodic monitoring is specified because vegetation and construction can change the channel.

Learning points. A free-space result is a baseline, not a guarantee. Site geometry, antenna installation, uncertainty, and seasonal conditions can be as important as nominal transmit power. Measurement-based calibration generally improves confidence more than adding unjustified model complexity.

10. Limitations, Regulation, and Best Practice

Propagation models are approximations. Their errors arise from inaccurate maps, unknown building materials, changing vegetation, imperfect antenna patterns, weather, interference, and insufficient measurement samples. Statistical distributions also conceal location-specific nulls and local maxima. Model parameters should therefore be reported with their valid ranges and uncertainty.

Wireless systems must comply with national spectrum regulations concerning frequency allocation, maximum effective isotropic radiated power, out-of-band emissions, channel bandwidth, and exposure limits. Engineers must also respect privacy and safety when collecting drive-test or indoor measurement data. Location traces should be minimized, protected, and used only for legitimate engineering purposes.

Best practice includes:

1. Define the operating frequency, bandwidth, antenna configuration, environment, and mobility.
2. Use free-space theory as a sanity check.
3. Select an empirical or statistical model within its validated range.
4. Separate mean path loss, shadowing, and fast fading in analysis.
5. Use dB consistently in link budgets and linear units in power-ratio probability formulas.
6. Validate predictions with measurements and report uncertainty.
7. Include fade margin, installation losses, interference, and future environmental change.
8. Reassess the model when frequency, antenna height, bandwidth, or deployment geometry changes.

Contemporary Developments

Modern wireless design increasingly combines ray tracing, geographic information systems, channel sounding, and machine-learning-assisted calibration. Machine learning can interpolate measured coverage or estimate model parameters, but it does not remove the need for physically meaningful data and validation. For massive MIMO and millimetre-wave systems, spatial consistency, blockage, beam squint, clustered scattering, and near-field effects may require models beyond simple Rayleigh or Rician assumptions. Digital twins and real-time sensing are being explored for adaptive beam management, while sub-6 GHz and higher-frequency systems continue to require environment-specific calibration.

Chapter Summary

Radio propagation is governed by distance, frequency, antenna properties, and interaction with the physical environment. Free-space propagation provides a fundamental baseline through the Friis equation and free-space path-loss formula. Real links require path-loss exponents, shadowing terms, penetration losses, and often empirical calibration. Multipath creates delay spread and constructive or destructive interference; its statistical effects are commonly represented by Rayleigh, Rician, Nakagami, and tapped-delay-line models. Motion produces Doppler shift and time-selective fading, while bandwidth relative to coherence bandwidth determines whether fading is flat or frequency selective. Antenna gain, effective aperture, polarization, and radiation pattern are essential parts of a link budget. Reliable design combines theory, appropriate statistical assumptions, measurement validation, uncertainty margins, and regulatory compliance.

Key Terms

Path loss: Average reduction in received power caused by propagation.

Free-space propagation: Ideal unobstructed line-of-sight propagation.

Shadowing: Slow spatial variation caused by obstacles and terrain.

Small-scale fading: Rapid signal variation caused mainly by multipath and motion.

Reflection: Redirection of a wave from a large surface.

Diffraction: Bending or spreading that enables energy behind an obstruction.

Scattering: Redistribution of energy by irregular or small objects.

Multipath: Arrival of signal energy through multiple propagation paths.

Delay spread: Temporal dispersion caused by different path delays.

Coherence bandwidth: Frequency range over which channel response is correlated.

Doppler shift: Frequency change caused by relative motion.

Coherence time: Approximate time over which the channel remains correlated.

Rayleigh fading: Fading model with many diffuse components and no dominant path.

Rician fading: Fading model containing a dominant component and diffuse components.

K-factor: Ratio of dominant-path power to scattered power.

Antenna gain: Directional radiated or received power relative to an isotropic antenna, including efficiency.

Effective aperture: Equivalent receiving area of an antenna.

Link margin: Received power above receiver sensitivity.

Tapped-delay-line model: Channel model composed of delayed, independently or jointly fading taps.

Fresnel zone: Region around a direct path whose obstruction can cause diffraction loss.

Review Questions

5 Short Answer Questions

1. Distinguish path loss, shadowing, and small-scale fading.
2. State the assumptions of the Friis transmission equation.
3. What is the physical meaning of coherence bandwidth?
4. Why does relative motion produce Doppler spread?
5. Define antenna gain and effective aperture.

5 Long Answer Questions

1. Explain reflection, diffraction, scattering, and penetration loss, giving a wireless example of each.
2. Derive the practical free-space path-loss expression from the Friis equation and explain its units.
3. Compare Rayleigh, Rician, and Nakagami-m fading models, including their appropriate applications.
4. Explain delay spread, intersymbol interference, coherence bandwidth, and the distinction between flat and frequency-selective fading.
5. Describe how a professional engineer should construct and validate a wireless link budget.

3 Analytical / Application Questions

1. A 5 GHz link uses 15 dBm transmit power, 10 dBi transmit gain, 10 dBi receive gain, and 3 dB equipment loss over 2 km. Calculate the free-space path loss and nominal received power, then identify practical effects that could reduce the margin.
2. A measured channel has three excess-delay paths with powers 1, 0.5, and 0.25 and delays 0, 1 μs, and 3 μs. Calculate the mean excess delay and RMS delay spread, and discuss whether a wideband signal may experience frequency-selective fading.
3. A mobile user travels at 30 m/s at 2.4 GHz. Calculate the maximum Doppler frequency and explain the implications for channel estimation and link adaptation.

Multiple Choice Questions with Answers

Question 1

Which phenomenon is primarily responsible for rapid amplitude and phase variations over distances comparable to a wavelength?

A. Small-scale multipath fading

B. Free-space spreading only

C. Antenna radiation efficiency

D. Spectrum allocation

Correct Answer: A. Small-scale multipath fading

Question 2

Under the stated practical units, which expression gives free-space path loss?

A. 32.44 + 10 log₁₀(f_MHz) + 10 log₁₀(d_km)

B. 32.44 + 20 log₁₀(f_MHz) + 20 log₁₀(d_km)

C. 20 log₁₀(f_MHz/d_km)

D. 4πf_MHzd_km

Correct Answer: B. 32.44 + 20 log₁₀(f_MHz) + 20 log₁₀(d_km)

Question 3

A channel with no dominant line-of-sight component and many diffuse components is commonly modeled using which distribution?

A. Rician

B. Uniform

C. Rayleigh

D. Deterministic free-space

Correct Answer: C. Rayleigh

Question 4

The Rician K-factor is the ratio of:

A. Carrier frequency to bandwidth

B. Dominant-path power to scattered power

C. Transmit power to receiver sensitivity

D. Distance to wavelength

Correct Answer: B. Dominant-path power to scattered power

Question 5

If a signal bandwidth is much smaller than the coherence bandwidth, the channel is approximately:

A. Flat fading

B. Frequency-selective fading

C. Non-propagating

D. Free of noise

Correct Answer: A. Flat fading

Question 6

For a receiver moving directly toward an arriving wave, the magnitude of Doppler shift is greatest when the angle θ in f_D = (v/λ) cos θ is:

A. 0°

B. 45°

C. 90°

D. 180° only

Correct Answer: A. 0°

Question 7

A 2.4 GHz wave has an approximate wavelength of:

A. 1.25 m

B. 0.125 m

C. 12.5 m

D. 0.0125 m

Correct Answer: B. 0.125 m

Question 8

Which quantity is added to received power in a link budget when expressed as a positive antenna gain in dBi?

A. Antenna gain

B. Propagation loss

C. Cable loss

D. Shadowing standard deviation

Correct Answer: A. Antenna gain

Question 9

The effective aperture of a receiving antenna is related to gain by:

A. A_e = 4πG_r/λ²

B. A_e = G_rλ²/(4π)

C. A_e = G_rλ/(4π)

D. A_e = λ²/(4πG_r)

Correct Answer: B. A_e = G_rλ²/(4π)

Question 10

Which model explicitly represents several delayed channel components with specified tap powers and delays?

A. Tapped-delay-line model

B. Free-space model

C. Single-ray model

D. Polarization model

Correct Answer: A. Tapped-delay-line model