Skip to content
New
Latest News || 10 October 2026: Two recent RIS-assisted near-field localisation papers are now available in IEEE Xplore. || 7 October 2026: Quiz and Pizza Night — Titanic Crew placed fourth. || 3 October 2026: Serving as a Student Ambassador at Queen's University Belfast, welcoming prospective students and visitors and supporting EEECS tours and event activities. || 9 September 2026: Published at AP-S/URSI 2026: “Near Field-Aware UE Localization in RIS-Aided Wireless Networks Through ML-Regressor.” || 25 August 2026: Poster presentation at UCMMT 2026 in Birmingham — “Experimental Validation of Localisation in RIS-Assisted mmWave Networks.”
Knowledge Hubantenna and propagation2. Propagation: What the Channel Does

antenna and propagation learning note

2. Propagation: What the Channel Does

Read path loss, shadowing, reflection, delay spread, and Doppler as physical effects rather than disconnected channel-model parameters.

Direct, reflected, and blocked radio paths between base station and user; below, a power-delay profile shows paths arriving at distinct delays
Original figure: one physical scene produces several arrivals. The lower power-delay profile (PDP) records their relative power and excess delay, not a map of where reflectors sit.

Start With One Street-Corner Link

A receiver hears one direct wave and later echoes from buildings. Each arrival can have a different strength, delay, direction, and phase.

  • Channel: the combined effect of those paths on the transmitted signal.
  • Reflection: a path bounces off a surface; diffraction: it bends around an edge; shadowing: a large object weakens important paths.
  • Why it matters: a strong average signal can still fade when the paths cancel at a particular location or frequency.
References for this section3

Path Loss and Shadowing

Path loss is the average power reduction between transmitter and receiver. Shadowing is the slower extra change caused by large obstructions.

  • Physical view: distance reduces the available power; a wall or vehicle may remove the direct path.
  • Use: estimate coverage and the extra link margin needed for blocked positions.
  • Keep separate: antenna gain, feed loss, polarization loss, and propagation loss should not be counted twice.

For an ideal unobstructed far-field reference with isotropic antennas:

LFS,dB=20log⁡10 ⁣(4πdλ)L_{\mathrm{FS,dB}}=20\log_{10}\!\left(\frac{4\pi d}{\lambda}\right)
  • LFS,dBL_{\mathrm{FS,dB}}: free-space path loss in dB.
  • dd: transmitter-to-receiver distance in metres.
  • λ\lambda: wavelength in metres; λ=c/f\lambda=c/f for frequency ff and wave speed cc.
  • π\pi: the usual mathematical constant; the logarithm converts a power ratio into decibels.

Example: at 28 GHz and 100 m, λ≈0.0107\lambda\approx0.0107 m and free-space loss is about 101.4 dB. Doubling distance adds about 6 dB only in this ideal reference. Street geometry can change the measured slope.

References for this section3

Delay and Excess Delay

Delay is the travel time of one path. Excess delay compares an arrival with the first detected arrival, so the first has zero excess delay.

  • Direct path: often arrives first, but it may be blocked or too weak to detect.
  • Later reflection: travels farther and arrives later; it can still be stronger than the first detected path.
  • Use: estimate timing error and whether echoes overlap neighbouring symbols.
τℓ,excess=τℓ−τfirst\tau_{\ell,\mathrm{excess}}=\tau_\ell-\tau_{\mathrm{first}}
  • τℓ\tau_\ell: absolute arrival delay of path ℓ\ell.
  • τfirst\tau_{\mathrm{first}}: delay of the first detected path, using the same clock.
  • τℓ,excess\tau_{\ell,\mathrm{excess}}: extra delay of path ℓ\ell, usually reported in ns or μ\mus.

Example: arrivals at 330 ns and 430 ns have excess delays of 0 and 100 ns. Excess delay does not reveal where the reflector is without more geometry.

References for this section3

Power Delay Profile (PDP)

A PDP plots received path power against excess delay. Read its horizontal axis as when the echo arrives, and its vertical axis as how strong that echo is.

  • Tap: one resolved delay bin or modeled path; a measured tap can contain several unresolved rays.
  • Peak: a strong arrival, not necessarily the direct path.
  • Use: see whether significant echoes fit inside an OFDM cyclic prefix and assess frequency selectivity.

For an idealized discrete-path channel:

h(t,τ)=∑ℓ=1Laℓ(t) δ ⁣(τ−τℓ(t))h(t,\tau)=\sum_{\ell=1}^{L}a_\ell(t)\,\delta\!\left(\tau-\tau_\ell(t)\right)
  • h(t,τ)h(t,\tau): channel impulse response at observation time tt and path delay τ\tau.
  • LL and ℓ\ell: number of modeled paths and the path index.
  • aℓ(t)a_\ell(t): complex path amplitude; its magnitude gives strength and its angle gives phase.
  • τℓ(t)\tau_\ell(t): arrival delay of path ℓ\ell; δ\delta marks an idealized arrival at that delay.

At a fixed observation time, the idealized PDP can be written as:

P(τ)=∑ℓ=1LPℓ δ(τ−τℓ,excess),Pℓ=∣aℓ∣2P(\tau)=\sum_{\ell=1}^{L}P_\ell\,\delta(\tau-\tau_{\ell,\mathrm{excess}}),\qquad P_\ell=|a_\ell|^2
  • P(τ)P(\tau): power associated with each excess-delay location.
  • PℓP_\ell: linear power of path ℓ\ell; do not add dBm values directly.
  • aℓa_\ell, LL, ℓ\ell, δ\delta, and τℓ,excess\tau_{\ell,\mathrm{excess}}: path amplitude, count, index, idealized impulse, and excess delay defined above.
References for this section3

Mean Delay and RMS Delay Spread

Mean excess delay is the power-weighted centre of the PDP. RMS delay spread says how widely the significant arrivals sit around that centre.

τˉ=∑ℓ=1LPℓτℓ,excess∑ℓ=1LPℓ\bar\tau=\frac{\sum_{\ell=1}^{L}P_\ell\tau_{\ell,\mathrm{excess}}}{\sum_{\ell=1}^{L}P_\ell}
  • τˉ\bar\tau: mean excess delay, in seconds.
  • PℓP_\ell: linear power of path ℓ\ell; stronger paths carry more weight.
  • τℓ,excess\tau_{\ell,\mathrm{excess}}: excess delay of path ℓ\ell; LL is the number of paths.
στ=∑ℓ=1LPℓ(τℓ,excess−τˉ)2∑ℓ=1LPℓ\sigma_\tau=\sqrt{\frac{\sum_{\ell=1}^{L}P_\ell(\tau_{\ell,\mathrm{excess}}-\bar\tau)^2}{\sum_{\ell=1}^{L}P_\ell}}
  • στ\sigma_\tau: RMS delay spread, in seconds; larger values mean more time dispersion.

  • PℓP_\ell, τℓ,excess\tau_{\ell,\mathrm{excess}}, τˉ\bar\tau, and LL: path power, excess delay, mean delay, and path count from the preceding equation.

  • Example: two equally strong arrivals at 0 and 100 ns give mean delay 50 ns and RMS delay spread 50 ns.

  • Consequence: large spread can cause frequency-selective fading and inter-symbol interference.

  • Caution: coherence bandwidth is an approximate inverse-scale summary, not a universal fixed constant.

References for this section3

Doppler Shift and Doppler Spread

Doppler shift is a frequency change caused by motion along one path. Doppler spread describes the range of shifts across paths arriving from different directions.

fD,ℓ=vλcos⁡θℓf_{D,\ell}=\frac{v}{\lambda}\cos\theta_\ell
  • fD,ℓf_{D,\ell}: signed Doppler shift of path ℓ\ell in hertz; sign depends on the chosen direction convention.

  • vv: receiver speed in metres per second relative to the path geometry.

  • λ\lambda: wavelength in metres; θℓ\theta_\ell is the angle between movement and the path direction.

  • Physical view: motion toward a path compresses the received wave cycles; motion away stretches them.

  • Use: decide how often pilots or beam-tracking estimates need refreshing.

  • Do not mix up: delay spread describes variation across frequency; Doppler spread describes change over time.

  • Example: at 28 GHz and 1 m/s, maximum single-path Doppler magnitude is about 93 Hz.

  • Actual spread: depends on the set of arrival angles.

  • Caution: coherence time is approximate; it is not one universal inverse formula.

References for this section3

Worked Link Check

For the same illustrative 28 GHz, 100 m link, take 30 dBm EIRP, 0 dBi receive gain, 100 MHz bandwidth, and 7 dB receiver noise figure:

  1. Received power: 30−101.4=−71.430-101.4=-71.4 dBm before other losses.
  2. Noise estimate: calculate thermal noise plus receiver noise figure.
NdBm≈−174+10log⁡10(BHz)+NF=−87 dBmN_{\mathrm{dBm}}\approx-174+10\log_{10}(B_{\mathrm{Hz}})+NF=-87\ \mathrm{dBm}
  • NdBmN_{\mathrm{dBm}}: estimated receiver noise power in dBm.
  • −174-174 dBm/Hz: approximate room-temperature thermal-noise density.
  • BHzB_{\mathrm{Hz}}: measurement bandwidth in hertz; here 100 million hertz.
  • NFNF: receiver noise figure in dB; here 7 dB.
  1. Margin: noise-limited SNR is about 15.6 dB. A hypothetical 20 dB blockage lowers it to about −4.4-4.4 dB. This is a teaching calculation; interference and other losses are omitted.
ObservationPhysical interpretationDesign decision
High path lossLittle desired power reaches the receiverIncrease link margin, improve pointing, or change site geometry
Deep shadowingA large obstacle blocks important pathsConsider alternate sites, diversity, or a controlled reflected path
Large delay spreadMany important paths arrive at different timesChoose suitable cyclic prefix or equalization
Fast Doppler variationChannel changes before an old estimate stays usefulUpdate pilots or tracking more often
Narrow angular spreadEnergy arrives from a limited set of directionsDirectional beams may be especially useful
References for this section3

When Each Model Is Useful

  • Free-space model: a clean reference calculation.

  • Statistical path-loss model: broad coverage estimates.

  • Multipath model: OFDM, beamforming, localization, and mobility tests.

  • Near-field arrays/RIS: check whether a plane-wave model misses distance-dependent phase curvature.

  • Coverage survey: map median power and shadowing across positions.

  • OFDM test: use PDP and delay spread to choose guard interval and pilot density.

  • Beam tracking: use angular and Doppler variation to set update frequency.

  • Localization: a later strong reflection can bias time-of-arrival if mistaken for the direct path.

Report whether values come from simulation, a channel sounder, or field measurements. A model fitted to one street, antenna orientation, and frequency is not automatically valid for a different site.

References for this section3

Complete references and further reading