Wireless signal propagation and terminal problems
Fundamental principles of how wireless signals travel, how the environment and devices change signal strength and quality, and the terminal-level problems engineers must handle when designing mobile systems.
Drafted with Aria, reviewed by the AiCanCode.org team · Updated . How we write and check our notes. Spotted an error? Use Give Feedback at the bottom of the page.
Why it matters
Two sentences: Wireless propagation determines how far and how reliably a radio link will work. Engineers use these ideas to make link budgets, choose frequencies, design antennas, estimate coverage, and solve terminal problems like low SNR, multipath fading and battery/antenna issues.
Key ideas
Start from first principles and simple models; build to practical terminal problems.
Electromagnetic wave: a coupled electric and magnetic field that travels through space. In the radio context we treat it as a transverse wave with frequency f (Hz) and wavelength λ = c/f (m), where c is the speed of light.
Power density: at distance r from an isotropic radiator that radiates power Pt (W) uniformly, the time-average power density S (W/m²) on a sphere of radius r is Pt divided by surface area 4πr² (far-field). This is the starting point for free-space models.
Antenna gain: Gain G (unitless or dBi) relates power radiated in a direction to power an isotropic antenna would radiate. Effective aperture Ae (m²) of a receiving antenna relates to gain by Ae = G·λ²/(4π). Define both before using them.
Friis transmission equation (exact under assumptions below): received power Pr at distance r in free space is Pt times transmitter gain Gt times receiver gain Gr times (λ/(4πr))². Assumptions: line-of-sight, far-field (r >> 2D²/λ where D is largest antenna dimension), matched polarization, no multipath, narrowband.
Free-space path loss (FSPL): ratio Pt/Pr (or in dB) due to geometry only. FSPL increases as (4πr/λ)² with distance and frequency.
Large-scale propagation (path loss + shadowing): real environments add excess loss from obstructions. Typical model: PL(d) (dB) = PL(d0) + 10·n·log10(d/d0) + Xσ where n is path-loss exponent (2 in free space, 2.7–3.5 suburban, 3–5 indoor), d0 is reference distance in far-field, and Xσ is a zero-mean Gaussian (dB) describing shadowing (slow fading) with standard deviation σ (dB).
Small-scale fading (multipath): when multiple delayed, phase-shifted copies of the signal combine at the receiver. If there is no dominant line-of-sight, the envelope of the received signal follows Rayleigh fading; with a dominant component it follows Rician fading (parameter K = power ratio dominant/other).
Delay spread and coherence bandwidth: multipath causes different propagation delays. Delay spread τrms (s) is the RMS of the power-delay profile; the approximate coherence bandwidth Bc ≈ 1/(5·τrms) (rough rule) — frequencies separated by less than Bc experience similar fading.
Doppler and coherence time: user motion shifts frequency by fd = v/λ·cos(θ) (max Doppler fd,max = v/λ). Coherence time Tc ≈ 0.423/fd,max (approx). If symbol duration << Tc channel is slowly varying; if >> Tc channel varies quickly causing time-selective fading.
Knife-edge diffraction: when a sharp obstacle blocks direct path, waves bend and a diffraction loss can be estimated from Fresnel–Kirchhoff theory; the simple knife-edge model gives loss as a function of Fresnel parameter v (dimensionless). For geometry with distances d1 and d2 from the edge to Tx/Rx, obstacle height above the LOS h (>0 when blocking), and wavelength λ:
v = h·sqrt( 2·(d1 + d2) / (λ·d1·d2) ). A widely used approximation for diffraction loss isL_dB ≈ 6.9 + 20·log10( sqrt((v - 0.1)² + 1) + v - 0.1 )(valid for v ≳ -0.7). Assumptions: single sharp edge, monochromatic.Reflection and scattering: reflection follows angle of incidence equals angle of reflection and is governed by Fresnel coefficients (depend on polarization and incidence angle); scattering from rough surfaces spreads energy.
Link budget and receiver problems: combine Pt, cable/connector losses, Gt, Gr, path loss, fading margin, and system noise. Noise power N = k·T·B (W) where k is Boltzmann constant, T is system temperature (K), and B is receiver noise bandwidth (Hz). Add receiver noise figure (NF, dB) to account for excess noise. Receiver sensitivity = minimum detectable signal for a target SNR. A practical formula: Sensitivity(dBm) ≈ N(dBm) + NF(dB) + required_SNR(dB).
Terminal problems: limited transmit power (battery), antenna size and orientation (gain/polarization mismatch), body effect (loss when handset held), near-far problem (in CDMA uplink many terminals: strong nearby signals can mask weak distant ones without power control), handover failures due to rapidly changing SNR, local oscillator drift, and thermal limits. Define each before using.
Diagrams to draw for exams (describe exactly):
- Free-space geometry: draw a transmitter at left and a spherical wavefront at radius r; label Pt, r, 4πr². Indicate receiving antenna with effective aperture Ae intercepting part of sphere.
- Multipath rays: draw transmitter, receiver, a reflecting plane (ground or building) and show direct ray and one reflected ray reaching receiver; label path lengths, phase difference Δφ and delays τ.
- Knife-edge diffraction: draw transmitter and receiver with a single rooftop edge between them; drop perpendiculars to the edge to compute Fresnel parameter v; label heights and distances.
Where more complex empirical models exist (Okumura, Hata, COST-231) refer to them by name for city/urban planning but do not re-derive their many empirical coefficients here.
Formulas
λ = c / f — λ is wavelength (m), c = 299792458 m/s, f is frequency (Hz). Applies always for electromagnetic wave in vacuum or air approximately.
S = P_t / (4·π·r²) — S is power density (W/m²), Pt is transmit power (W), r is distance (m). Assumes isotropic radiator and far-field.
A_e = G·λ² / (4·π) — A_e is effective aperture (m²), G is antenna gain (unitless). Assumes matched polarization and far-field. Received power relates to power density by P_r = S · A_e.
P_r = P_t · G_t · G_r · (λ / (4·π·r))² — Friis equation; Pr, Pt in W, gains unitless, r in m, λ in m. Assumes line-of-sight, far-field, matched polarization, no multipath.
FSPL(dB) = 20·log10(4·π·r / λ) = 20·log10(r) + 20·log10(f) + 20·log10(4·π/c) — free-space path loss in dB. r in m, f in Hz. Use when Friis applies. Equivalently, with r in km and f in MHz: FSPL(dB) = 32.44 + 20·log10(r_km) + 20·log10(f_MHz).
PL(dB) = PL(d0) + 10·n·log10(d/d0) + X_σ — log-distance path loss model. PL(d0) is measured/reference loss at d0, n is path-loss exponent, X_σ ~ N(0,σ²) dB.
N = k·T·B — thermal noise power in W; k = 1.380649e-23 J/K, T in K (standard 290 K used often), B in Hz.
SNR(dB) = P_r(dBm) - N(dBm) - NF(dB) — signal-to-noise ratio accounting for receiver noise figure NF. Convert units appropriately.
B_c ≈ 1 / (5·τ_rms) — coherence bandwidth approximation (Hz). τ_rms in seconds.
f_d = v / λ (max Doppler when cos(θ)=1) — Doppler shift (Hz) for speed v (m/s), λ in m.
Link budget in dB: P_r(dBm) = P_t(dBm) + G_t(dBi) + G_r(dBi) - FSPL(dB) - other_losses(dB) — use for practical calculations.
Worked examples
Example 1 — Friis received power (numerical) Given: Pt = 1 W (30 dBm), Gt = 2 (3 dBi), Gr = 2 (3 dBi), f = 2.4 GHz, r = 100 m. Compute Pr. Step 1: λ = c/f = 299792458 / 2.4e9 = 0.1249135 m. Step 2: Use Friis: Pr = Pt·Gt·Gr·(λ/(4πr))². Compute factor: λ/(4πr) = 0.1249135 / (4·π·100) = 0.1249135 / 1256.6371 = 9.942e-5. Square: (9.942e-5)² = 9.884e-9. Pr = 1 · 2 · 2 · 9.884e-9 = 3.954e-8 W. Convert to dBm: P(dBm) = 10·log10(3.954e-8) + 30 = -44.03 dBm. Answer: Pr = 3.95×10^-8 W (−44.0 dBm).
Example 2 — SNR and receiver sensitivity Given: Pt = 0.2 W (23.01 dBm), Gt = 0 dBi, Gr = 0 dBi, f = 900 MHz, r = 1 km, bandwidth B = 200 kHz, NF = 5 dB, T = 290 K. Find Pr, noise, SNR. Step 1: λ = 299792458 / 9e8 = 0.3331027 m. Step 2: FSPL(dB) = 20·log10(4·π·r/λ). Compute 4πr/λ = 4·π·1000 / 0.3331027 = 4·π·3000.9 = 4·π·3000.9 ≈ 37699.1. FSPL = 20·log10(37699.1) = 20·(4.576) = 91.52 dB (approx). Step 3: Link budget: Pr(dBm) = 23.01 + 0 + 0 - 91.52 = -68.51 dBm. Step 4: Noise N = kTB = 1.380649e-23·290·200000 = 8.0078e-16 W => N(dBm)=10·log10(8.0078e-16)+30 = -120.97 dBm. Include NF: effective noise = -120.97 + 5 = -115.97 dBm. SNR = Pr - effective noise = -68.51 - (-115.97) = 47.46 dB. Answer: Pr ≈ -68.5 dBm; SNR ≈ 47.5 dB.
(These examples assume line-of-sight, matched polarization, no additional losses.)
Common mistakes
- Forgetting antenna polarization mismatch or cable losses in link budget; these are often tens of dB if ignored.
- Mixing noise bandwidth and signal bandwidth (use the receiver noise equivalent bandwidth for kTB).
- Using simplistic coherence bandwidth Bc≈1/τrms without stating it is approximate; different rules (1/(5·τrms), 1/(50·τrms)) are used in texts.
For your exam
Question types you will see: definitions (e.g., define path loss, fading, Doppler), short notes (multipath, knife-edge diffraction), derivations (derive Friis from power density and effective aperture, derive FSPL in dB), numericals (compute Pr using Friis, compute SNR using kTB and NF), diagrams (draw multipath and knife-edge geometry), and link budget problems including margins. What full-marks answers include: clear definitions (bold terms if asked), list assumptions, step-by-step derivations with algebra, units at each step in numericals, a labelled diagram when asked, and mention practical factors (polarization, body loss, cable loss, fading margin) when discussing terminal performance.
Quick check
- Write Friis transmission equation and name three assumptions behind it.
- How does delay spread affect coherence bandwidth? Give the rough relation.
- A 1 W transmitter at 2.4 GHz, isotropic, what is the power density at 10 m? (Answer in W/m²) Answers: 1) P_r = P_t·G_t·G_r·(λ/(4πr))²; assumptions: line-of-sight/far-field, matched polarization, no multipath. 2) Coherence bandwidth B_c ≈ 1/(5·τ_rms); larger delay spread ⇒ smaller B_c. 3) S = 1/(4π·10²) = 1/(1256.64) = 7.96×10^-4 W/m².
Finished this topic? Mark it so your progress, study plan and readiness keep up.
Stuck on something here?