Antenna Theory & Design
1. At a glance
An antenna is the transducer between a guided wave (current on a feed line, coax, or microstrip) and a radiated electromagnetic wave in free space. By reciprocity the same structure works for transmission and reception with identical pattern and impedance — so antenna design is a single discipline regardless of TX/RX role. The performance-defining numbers are a short list: gain G (dBi), directivity D, radiation efficiency η_rad, half-power beamwidth (HPBW), impedance bandwidth (typically -10 dB return loss), polarization (linear, circular, elliptical), voltage standing-wave ratio (VSWR), sidelobe level (SLL), front-to-back ratio (F/B), and axial ratio (AR) for circular polarization.
Application breadth is enormous. A modern smartphone carries 8–15 antennas — LTE main + diversity, carrier-aggregation, Wi-Fi 2.4/5/6 GHz, Bluetooth, GPS L1/L5, NFC, UWB (Apple U1, NXP Trimension), and a 5G mmWave module behind the bezel. A Wi-Fi router packs 4–8 dual-band PIFAs or monopoles. A 5G base station runs massive MIMO with 32×32 to 64×64 element arrays. Satellite systems use parabolic dishes (DBS, VSAT), helical antennas (Iridium, Inmarsat), and patch arrays (Starlink user terminal). Radar ranges from slot arrays (marine X-band) to AESA (F-22, Aegis SPY-6). Automotive integrates 77 GHz fascia radar, V2X 5.9 GHz, GNSS, and a shark-fin housing for cellular/SDARS. IoT modules sit on PCB chip antennas a few millimetres long. This note unifies the physics, geometry, and 2026-era practice that all of those share.
Foundation lives at [[Engineering/electromagnetics-engineering]] (Maxwell, free-space wave equation, η₀ = 377 Ω). The system-level wrapper is [[Engineering/rf-design]] (matching networks, S-parameters, noise figure, link budget at component level). PCB-level integration is [[Engineering/pcb-design]].
2. Why it matters
The antenna dominates wireless system performance in a way no other component does. A 10 dB antenna gain shortfall costs an order of magnitude in range — and link-budget recovery elsewhere (higher TX power, better LNA, FEC coding) is bounded by regulatory EIRP limits (FCC Part 15.247: 30 dBm EIRP for 2.4 GHz ISM, 36 dBm for 5 GHz UNII-1, etc.) and physics (kTB noise floor at -174 dBm/Hz). A poorly-designed handset antenna routinely loses 5–10 dB to hand grip and head proximity — recoverable only by adding an antenna (diversity) or expensive aperture-tuning IC’s (Qorvo QM77032, Murata, ST). On a 24 GHz mmWave 5G link a 1 mm fabrication tolerance on a patch element shifts resonance by ~600 MHz, killing the band. Antenna design is therefore typically the longest single-discipline schedule item in any new wireless product — 8–16 weeks of EM simulation + chamber iteration is normal for a phone, 6–12 months for a base station.
The flip side: an extra 3 dB of antenna gain is free range — no power, no battery cost, no thermal penalty. That’s why every wireless product team eventually invests heavily in antenna engineering.
3. First principles
Hertzian dipole — the elemental radiator
The Hertzian (infinitesimal) dipole is a short current element of length dℓ ≪ λ carrying uniform current I. Solving Maxwell’s equations in spherical coordinates (Schelkunoff 1943, after Hertz 1888):
- Far-field E_θ = j·η₀·(I·dℓ / 2λr) · sin(θ) · e^(-jkr)
- Far-field H_φ = E_θ / η₀ (the two are 90° in space but in time-phase in the far field; ⟨S⟩ is real Poynting power flowing radially)
- Radiation resistance R_rad = 80·π²·(dℓ/λ)² Ω — a tiny number for dℓ ≪ λ (a 1 mm rod at 100 MHz has R_rad ≈ 9 µΩ, dominated by I²R loss in any real conductor — why electrically short antennas are notoriously inefficient)
- Directivity D = 1.5 (1.76 dBi)
- Pattern sin²(θ) doughnut — peak broadside, null along the wire axis
- Near-field (r ≪ λ/2π): E_r and E_θ scale as 1/r³ (electrostatic-like quasi-static term); reactive energy stored, not radiated
- Radiation zone boundary: r ≈ λ/(2π) ≈ 0.16·λ separates near-field reactive region from radiating region
This is the building block. All wire antennas integrate Hertzian elements along a sinusoidal or arbitrary current distribution; arrays of Hertzian elements with progressive phase produce arbitrary patterns by superposition.
Half-wave dipole
A thin wire of physical length ≈ 0.48·λ (the 5% shortening accounts for end-capacitance effects) driven at the centre. Current distribution is sinusoidal with a maximum at the feed:
- R_rad ≈ 73 Ω (real part), X ≈ +42.5 Ω (slightly inductive); resonant length tuning zeroes X
- Directivity D = 1.64, gain G = 2.15 dBi (assuming η_rad ≈ 1 for a good conductor)
- HPBW ≈ 78° in the E-plane, omnidirectional in azimuth (H-plane)
- Bandwidth ≈ 8–16% depending on wire diameter (fatter dipoles are broader)
The half-wave dipole is the reference antenna. “dBi” means dB over a hypothetical isotropic; “dBd” means dB over a half-wave dipole. Always: dBi = dBd + 2.15.
Antenna pattern
The radiation pattern F(θ,φ) describes the angular distribution of radiated power. The two principal-plane cuts are:
- E-plane — the plane containing the E-field vector (for a vertical dipole, the vertical plane)
- H-plane — the plane containing the H-field vector (for a vertical dipole, the horizontal plane)
A full 3D pattern is captured as F(θ,φ); engineers typically inspect the two principal cuts plus a contour plot.
Directivity and gain
Directivity D measures how concentrated the radiated power is in the peak direction relative to an isotropic radiator:
D_max = 4π / Ω_A
where Ω_A is the beam solid angle — the equivalent solid angle into which all radiated power would flow if the pattern were uniform at peak amplitude. For a “pencil-beam” with two half-power beamwidths θ_1 and θ_2 (in radians):
D ≈ 4π / (θ_1 · θ_2) (Kraus approximation)
Gain G adds radiation efficiency (ohmic + dielectric losses inside the antenna):
G = η_rad · D η_rad = P_rad / P_input
For a well-designed antenna η_rad > 0.7 (-1.5 dB); for electrically small or lossy antennas it can drop to 0.1 or below.
Aperture efficiency
For aperture antennas (horns, dishes, patch arrays), the geometric area A_phys and the effective area A_eff = G·λ²/(4π) differ:
η_ap = A_eff / A_phys = G·λ² / (4π·A_phys)
Typical aperture efficiencies: parabolic dish 0.55–0.70 (illumination taper + spillover), horn 0.50–0.81 (depending on aperture phase error), patch array 0.70–0.85 (uniform illumination is realisable).
Impedance, VSWR, bandwidth
Bandwidth is conventionally defined where |Γ| ≤ -10 dB (VSWR ≤ 2:1) at the feed. The relation:
VSWR = (1 + |Γ|) / (1 - |Γ|) Γ = (Z_A - Z₀) / (Z_A + Z₀)
For a 50 Ω system, -10 dB return loss is the universal default; -14 dB (VSWR 1.5) for higher-quality systems; -20 dB (VSWR 1.22) for low-noise receivers and metrology.
Polarization
The direction of the radiated E-field vector defines polarization:
- Linear — vertical, horizontal, or any fixed angle. Cellular base stations use ±45° slant pair for diversity.
- Circular — RHCP (right-hand) or LHCP (left-hand). Used for satellite links to avoid polarization mismatch from Faraday rotation through the ionosphere (GPS L1 is RHCP).
- Elliptical — the general case; characterised by axial ratio (AR) = ratio of major to minor axis of the polarization ellipse. AR = 1 (0 dB) is perfect circular; AR < 3 dB is “good CP” by IEEE convention.
Cross-polarization mismatch loss:
- Linear-to-linear orthogonal: theoretically -∞ dB, practically -20 to -30 dB
- Linear-to-CP: fixed -3 dB
- CP-to-CP opposite hand: -20 dB or more
Beamwidth and aperture
For an aperture antenna, HPBW scales inversely with aperture in wavelengths:
HPBW ≈ k · λ / D k ≈ 60° to 80° depending on illumination taper
Uniform illumination: k ≈ 50.8°; tapered (Taylor -25 dB SLL): k ≈ 65°. Wider aperture → narrower beam → higher gain (G ≈ 32400 / (HPBW_az · HPBW_el) for pencil beams in degrees).
Friis transmission equation
The bedrock link-budget formula (Friis 1944) for free-space line-of-sight:
P_r = P_t · G_t · G_r · (λ / (4π·d))²
In dB: P_r[dBm] = P_t[dBm] + G_t[dBi] + G_r[dBi] - FSPL[dB] FSPL[dB] = 20·log₁₀(4π·d/λ) = 20·log₁₀(d_km) + 20·log₁₀(f_MHz) + 32.44
Reciprocity
A passive linear antenna has the same radiation pattern, gain, polarization, and impedance for transmission and reception. This is a consequence of Lorentz reciprocity in linear media. It collapses the design problem to one regime; everything proved for TX applies to RX. Active antennas (with non-reciprocal elements like circulators or PIN-diode duplexers) break this only at the system level. Practical consequences:
- Measure pattern in TX mode (drive antenna, rotate a probe) or RX mode (illuminate with a source horn, rotate antenna under test) — same result.
- Effective aperture A_eff and gain G are tied by A_eff = G·λ²/(4π). A high-gain antenna captures more power on receive and concentrates more power on transmit — same coin, two sides.
- TDD massive-MIMO base stations exploit reciprocity directly: channel learned from UL pilots is reused for DL beamforming, eliminating expensive DL CSI feedback.
Near-field vs far-field
The radiating far-field begins at:
d_ff = 2·D² / λ (D = largest aperture dimension)
Inside d_ff the field is dominated by reactive (storage) components falling as 1/r² and 1/r³. Friis and Gain only apply outside d_ff. For a 0.1 m antenna at 2.4 GHz, d_ff ≈ 0.16 m — most measurements are far-field. For a 30 m dish at 1 GHz, d_ff ≈ 6 km — too far for any indoor chamber, so compact-range or near-field-to-far-field transformation is used.
4. Antenna families
Wire antennas
| Family | Length | Gain | BW | Polarization | Use |
|---|---|---|---|---|---|
| Hertzian dipole | dℓ ≪ λ | 1.76 dBi | — | linear | theoretical reference |
| Half-wave dipole | ≈ 0.48·λ | 2.15 dBi | 8–16% | linear | reference, ham HF, FM rcvr |
| Folded dipole | ≈ 0.48·λ | 2.15 dBi | 10–20% | linear, ~300 Ω | FM/TV ribbon |
| Quarter-wave monopole | ≈ 0.24·λ + GP | 5.2 dBi | 8–10% | vertical lin | whip, car AM/FM, ISM |
| Yagi-Uda (Yagi 1926) | 3–20 elements | 6–17 dBi | 2–10% | linear | TV rooftop, ham VHF/UHF |
| Log-periodic (Carrel 1961) | many elements | 6–10 dBi | decade+ | linear | EMC test, broadband |
| Small loop | < λ/10 | -3 to 1 dBi | < 1% | magnetic-mode | AM rcvr, RFID, NFC |
| Large loop | ≈ λ | 2–4 dBi | 10% | linear (E-mode) | HF ham, quad loop |
| Axial-mode helix (Kraus 1947) | 3–20 turns | 10–18 dBi | 50%+ | circular | satcom uplink/downlink |
| Normal-mode helix | < λ/4 | 0–2 dBi | narrow | linear | rubber-duck handhelds |
Aperture antennas
| Family | Aperture | Gain | BW | Use |
|---|---|---|---|---|
| Pyramidal horn | a × b | 10–25 dBi | 50% | standard-gain reference (SGH), feedhorn |
| Conical horn | π·a² | 10–22 dBi | 60% | satellite TVRO LNB |
| Corrugated horn (Clarricoats 1969) | π·a² | 15–30 dBi | octave+ | low-sidelobe satellite, radio telescopes |
| Parabolic dish | π·D²/4 | 25–60 dBi | 10–25% | DBS, VSAT, microwave PtP, radio astronomy |
| Cassegrain dual-reflector | π·D²/4 | 30–65 dBi | 15% | high-gain satellite, deep-space (DSN 70 m) |
| Gregorian dual-reflector | π·D²/4 | 30–65 dBi | 15% | Arecibo (until 2020), ALMA optics |
| Slot (Booker 1946) | λ/2 × narrow | 2–8 dBi | 5–15% | flush-mount aircraft, military |
| Waveguide slotted array | many slots | 20–40 dBi | 3–10% | marine X-band radar, VOR |
Microstrip / printed antennas
| Family | Footprint | Gain | BW | Use |
|---|---|---|---|---|
| Rectangular patch | ≈ λ/2 × λ/2 | 5–9 dBi | 1–5% | Wi-Fi, GPS, RFID, ISM modules |
| Circular patch | πr² | 5–8 dBi | 1–5% | CP variants, satellite |
| PIFA (Planar Inverted-F) | < λ/4 | 2–6 dBi | 5–10% | handset cellular/Wi-Fi |
| IFA / FICA | < λ/4 | 1–4 dBi | 3–8% | IoT, wearable |
| Meandered monopole | λ/8 area | 0–3 dBi | 3–10% | sub-GHz IoT, key fob |
| Slot on PCB | λ/2 slot | 3–6 dBi | 3–10% | LTE handset diversity, laptop |
| Chip antenna (ceramic SMD) | 3 × 1 × 1 mm | -2 to 2 dBi | narrow | BLE / Wi-Fi modules, Antenova, Taoglas |
| Dielectric resonator | ε_r > 10 cube | 5–8 dBi | 5–20% | high-Q mmWave |
| Vivaldi (tapered slot, Gibson 1979) | 2–4λ taper | 6–10 dBi | decade+ | UWB, EW receivers |
Array antennas
| Family | Elements | Gain | Steering | Use |
|---|---|---|---|---|
| Broadside linear | 4–32 | 8–18 dBi | fixed | base-station sector |
| End-fire (Hansen-Woodyard 1938) | 4–10 | 8–14 dBi | fixed along array | radio astronomy, EW |
| Planar broadside | 4×4 to 32×32 | 13–35 dBi | fixed | satcom user terminal |
| Phased array (mechanical) | 100s | 30–50 dBi | electronic | air-traffic, weather radar |
| AESA (Active ESA) | 1000+ T/R modules | 30–50 dBi | wide scan | F-22 APG-77, Aegis SPY-6 |
| Massive MIMO | 64–256 | 25–35 dBi | digital BF | 5G base station (Ericsson AIR, Nokia AirScale) |
| Hybrid analog-digital | 64–256 / 4–16 RF chains | 25–35 dBi | beamspace | 5G mmWave (28/39/60 GHz) |
| Sparse / thinned | irregular | aperture-fill | wide | RFID localization, MIMO radar |
| Reflectarray | flat panel + scattering elements | 25–40 dBi | electronic (varactor) | Ku-band satcom (Starlink) |
| Metasurface (intelligent reflecting surface, 2020s) | 100–10000 unit cells | passive | electronic | 6G research, NTN |
5. Practical math + worked examples
Example A — Half-wave dipole at 2.4 GHz
λ = c / f = 3 × 10⁸ / 2.4 × 10⁹ = 0.125 m Physical length L = 0.48·λ = 60 mm (5% shortening from λ/2 = 62.5 mm for wire diameter ≈ λ/1000) R_rad = 73 Ω; X_in ≈ 0 Ω at resonance; |Γ| at 50 Ω = (73-50)/(73+50) = 0.187 → VSWR = 1.46:1 Return loss = -14.6 dB — usable in a 50 Ω system without a matching network for moderate-Q applications Gain G = 2.15 dBi, η_rad ≈ 0.97 for copper wire at 2.4 GHz Effective aperture A_eff = G·λ²/(4π) = 1.64 · (0.125)² / (4π) = 2.04 × 10⁻³ m² = 20.4 cm²
For tighter match a quarter-wave transformer of Z₀ = √(50·73) = 60.4 Ω inserted between dipole and feed gives a perfect match at f₀.
Skin-depth sanity check: at 2.4 GHz copper has δ ≈ 1.34 µm. A standard 24 AWG (0.51 mm diameter) wire has surface area enough that AC resistance R_AC ≈ ρ / (π·d·δ·L) for L = 60 mm ≈ 0.4 Ω, giving η_rad = R_rad / (R_rad + R_loss) ≈ 73 / 73.4 = 0.995 → -0.02 dB, negligible. Conductor loss dominates only for electrically-small antennas.
Example B — Friis link budget, Wi-Fi 6 outdoor LOS
P_t = 20 dBm (100 mW, regulatory limit at 2.4 GHz EIRP after subtracting cable loss) G_t = G_r = 2.15 dBi (omnidirectional whips) f = 2.4 GHz, λ = 0.125 m, d = 100 m line-of-sight FSPL = 20·log₁₀(4π·100/0.125) = 20·log₁₀(10053) = 80.0 dB P_r = 20 + 2.15 + 2.15 - 80.0 = -55.7 dBm RX sensitivity at MCS-7 (54 Mbps OFDM, 802.11a/g) ≈ -65 dBm → link margin 9.3 dB
9 dB margin in clear LOS is marginal for real-world deployment (multipath fading +/- 10 dB Rayleigh, wall losses for indoor, foliage absorption outdoor). Closing the budget requires either directional antennas (G ≈ 8–14 dBi yagi or panel), lower data rate (MCS-0 at -82 dBm), or 5 GHz UNII-3 with higher TX power.
Example C — Rectangular patch antenna design
Target f₀ = 2.45 GHz on Rogers RO4350B (ε_r = 3.66, tan δ = 0.0037, h = 1.524 mm = 60 mil).
Width W (Balanis eq. 14-6, sets radiation efficiency): W = (c / 2·f₀) · √(2 / (ε_r + 1)) W = (3 × 10⁸ / 4.9 × 10⁹) · √(2 / 4.66) = 0.0612 m · 0.655 = 40.1 mm
Effective permittivity (W/h = 26.3 ≫ 1): ε_eff = (ε_r+1)/2 + (ε_r-1)/2 · (1 + 12·h/W)^(-1/2) = 2.33 + 1.33 · (1 + 12·0.038)^(-1/2) = 2.33 + 1.33·0.835 = 2.33 + 1.11 = 3.44
Length extension ΔL (fringing field): ΔL = 0.412·h · (ε_eff + 0.3)·(W/h + 0.264) / ((ε_eff - 0.258)·(W/h + 0.8)) ≈ 0.412 · 1.524 · 3.74 · 26.56 / (3.18 · 27.10) ≈ 0.74 mm
Physical length L: L = c / (2·f₀·√ε_eff) - 2·ΔL = 3 × 10⁸ / (4.9 × 10⁹ · 1.855) - 1.48 mm = 33.0 - 1.48 = 31.5 mm
Edge resistance ≈ 200–300 Ω; for a 50 Ω inset feed the inset depth y₀ from edge: R_in(y₀) = R_in(0) · cos²(π·y₀/L) For R_in(0) = 240 Ω, R_in(y₀) = 50 Ω → cos²(π·y₀/L) = 0.208 → y₀ = 0.347·L = 10.9 mm
Full design then transfers to HFSS / CST / FEKO for parametric tuning (inset gap, feed-line width, ground-plane size). Bandwidth ≈ 2% for h = 1.5 mm; widen by stacking patches (10%+ BW), using aperture coupling, or moving to thicker substrate (h = λ/30 gives ~5% BW).
Example D — Phased-array beam-steering
A linear array of N = 16 elements with element spacing d = λ/2 at 28 GHz mmWave (λ = 10.71 mm, d = 5.36 mm). Each element fed with progressive phase shift β to steer the beam to angle θ_0 from broadside:
β = -k·d·sin(θ_0) = -(2π/λ)·(λ/2)·sin(θ_0) = -π·sin(θ_0)
For θ_0 = 30° (steered off broadside): β = -π·0.5 = -π/2 = -90° per element Beamwidth at broadside (uniform amplitude, λ/2 spacing): HPBW ≈ 102°/N = 6.4° Beamwidth at scan angle θ_0: HPBW(θ_0) ≈ HPBW_0 / cos(θ_0) = 6.4°/0.866 = 7.4° (beam broadens off-broadside) Array gain ≈ N · G_element = 16 · 6 dBi = 18 dB (for ideal uniform illumination)
Grating-lobe condition: d < λ / (1 + |sin(θ_max)|). For 60° scan max, d < 0.535·λ — so λ/2 spacing supports ±90° scan (in principle), but real arrays use 0.45–0.5·λ to control element-pattern roll-off.
Example E — Parabolic dish at 12 GHz (Ku-band DBS receive)
Common European DBS 60 cm offset-fed dish at f = 12 GHz (λ = 25 mm): Aperture A = π·(0.30)² = 0.283 m² Maximum directivity D_max = 4π·A/λ² = 4π·0.283 / (0.025)² = 5680 → 37.5 dBi With aperture efficiency η_ap = 0.65: G = 0.65 · 5680 = 3690 → 35.7 dBi HPBW ≈ 70°·λ/D = 70·0.025/0.60 = 2.9° — explains why DBS pointing is critical (±0.5° accuracy needed) Far-field distance d_ff = 2·D²/λ = 2·(0.6)²/0.025 = 28.8 m — chamber far-field requires a large room G/T for a typical LNB (T_LNA = 50 K, T_sky = 30 K, total ≈ 100 K): G/T = 35.7 - 10·log(100) = 15.7 dB/K, sufficient for the ~10 dB·Hz·K required link margin at 27.5 Mbaud QPSK
6. Antenna parameters in depth
- Radiation efficiency η_rad — ratio P_rad / P_input. Captures ohmic loss in conductors, dielectric loss in substrate, and any mismatched matching network loss. Reported in dB as 10·log₁₀(η_rad).
- Total efficiency η_total = η_rad · (1 - |Γ|²). The number a phone test lab actually cares about; degrades with poor impedance match.
- VSWR vs return loss vs reflection coefficient — three names for one quantity. See
[[Engineering/rf-design]]. - Axial ratio (AR) for circular polarization — AR = 1 (0 dB) ideal; AR < 3 dB acceptable; AR > 6 dB approaches linear behaviour.
- Cross-pol isolation — pattern of orthogonal polarization vs co-pol pattern. Target > 20 dB suppression below co-pol peak for a “clean” linear or CP antenna.
- Front-to-back ratio (F/B) — co-pol gain at boresight / co-pol gain at 180°. > 20 dB for a “directional” antenna (Yagi, dish, patch with backed ground); ≈ 0 dB for an omnidirectional dipole.
- Sidelobe level (SLL) — uniformly-illuminated aperture has first SLL at -13.2 dB (sinc-function pattern); Taylor n̄ = 4 distribution achieves -25 to -30 dB; Chebyshev gives equiripple sidelobes at any chosen level (Dolph 1946). Always at the cost of broadened main beam.
- Beam squint — frequency-dependent change in main-beam pointing of phased arrays using phase shifters (not true time delay). For an array of length L, squint Δθ ≈ (Δf/f)·tan(θ_0). At 5% bandwidth and 30° scan, squint ≈ 1.7° — significant for wideband radars and forced to time-delay units (TDUs) for >10% BW.
- Group delay flatness — important for digital modulation; non-flat group delay causes ISI in wideband data links.
- G/T (gain over temperature) — figure of merit for satcom receivers, dB/K. G_RX[dBi] - 10·log₁₀(T_system[K]). Combines antenna gain and LNA noise temperature.
- Effective isotropic radiated power (EIRP) — P_t[dBm] + G_t[dBi] - L_feed[dB]. The number regulators care about; FCC Part 15 limits 2.4 GHz ISM EIRP to +30 dBm (point-to-multipoint) or +36 dBm (point-to-point with 6 dBi+ antennas + 1:3 power reduction rule).
- Effective isotropic sensitivity (EIS) — receiver-side equivalent; S[dBm] - G_r[dBi]. The number to compare for OTA receive performance.
- Beam efficiency — fraction of total power within the main beam (defined by an angular region). Important for radio astronomy and sensitive radiometry where sidelobe pickup matters.
- Phase center — apparent point of origin of spherical wavefronts; usually frequency-dependent. Critical for GNSS antennas (varying phase centre causes 1–10 mm position error) and array calibration.
- Q factor (Chu limit, 1948) — for an electrically small antenna of radius a enclosed in a sphere of radius a < λ/(2π): Q_min ≈ 1/(k·a)³ + 1/(k·a). Sets the fundamental BW-vs-size trade: smaller than λ/(2π) makes high-BW radiation thermodynamically impossible (Wheeler limit, McLean refinement 1996).
Polarization types — reference
| Polarization | Description | Antenna producing | Loss to mismatch |
|---|---|---|---|
| Linear vertical (V) | E vector along Z | Vertical dipole, monopole, vertical patch | -∞ dB to horizontal (theory); -20 to -30 dB practical |
| Linear horizontal (H) | E vector along X or Y | Horizontal dipole, log-periodic | -∞ dB to vertical (theory); -20 to -30 dB practical |
| Slant ±45° | E rotated 45° | Cross-slant base-station panel | -3 dB to vertical (orthogonal slant is uncorrelated) |
| RHCP | E rotates clockwise viewing source | Axial-mode helix (right-wound), quadrature-fed patch, crossed dipoles | -∞ dB to LHCP (theory); -20 dB practical; -3 dB fixed to linear |
| LHCP | E rotates counter-clockwise | Left-wound helix, opposite-quadrature patch | -∞ to RHCP; -3 dB fixed to linear |
| Elliptical | AR > 1 (intermediate state) | Patch with imperfect quadrature feed, real-world CP | -1 to -3 dB depending on AR |
GPS L1 broadcasts RHCP at 1.575 GHz; consumer GPS receivers use a patch above a ground plane fed in quadrature (proximity-coupled or microstrip-line-fed) to capture RHCP with AR < 3 dB across the L1 bandwidth. A linear monopole receives GPS at -3 dB penalty plus polarization variation from multipath.
Engineering judgement — choosing an antenna
A short decision tree for typical wireless products:
| Question | If yes | If no |
|---|---|---|
| BW < 5%, low cost, line-of-sight outdoor? | Yagi or panel patch array | Dipole or PIFA |
| Sub-GHz IoT (433/868/915 MHz), tiny PCB? | Meandered monopole, chip antenna (Ignion, Antenova) | Wire whip (compromised by size) |
| Handset cellular, multi-band, hand-tolerant? | PIFA + aperture tuner IC (Qorvo QM77032) | Fixed monoband patch |
| Wi-Fi router, 2.4 + 5 GHz, omnidirectional? | Dual-band dipole or PIFA, 2–4 element MIMO | Single chip antenna |
| GPS / GNSS handheld? | RHCP patch over copper GP, LNA at antenna | Linear monopole (loses 3 dB) |
| Satellite TVRO (Ku-band)? | Offset parabolic + LNB | Anything smaller fails link budget |
| 5G mmWave (28/39 GHz) handheld? | Stacked patch sub-array (4–16 elements) with phase shifters | Single patch (no scan) |
| Automotive 77 GHz radar? | MIMO array on Rogers RO3003 substrate | Standalone patch (insufficient angular resolution) |
| Bluetooth wearable, no GP? | Slot or loop (radiates without ground plane) | Patch (needs GP, won’t fit) |
| Wide instantaneous BW (>30%)? | Vivaldi tapered slot, log-periodic, biconical | Resonant types fail |
| Low-profile aircraft skin-mount? | Slot, microstrip patch, conformal array | Whip (drag, ice) |
| Underground or underwater? | HF loop or ferrite-loaded magnetic antenna | Electric-field types attenuate in conductive media |
7. Phased arrays — 2026 practice
Phased arrays now dominate the mmWave 5G, automotive radar, satcom user-terminal, and military RF landscapes. The defining ability: electronic beam steering at the speed of a control voltage update (nanoseconds), with multiple simultaneous beams in massive-MIMO systems.
- Beamforming weights: w_n = a_n · exp(j·β_n), where a_n is amplitude (taper) and β_n = -n·k·d·sin(θ_0) is the linear phase progression for steering to θ_0. Patterns combine as array-factor × element-factor.
- Adaptive arrays — algorithms (LMS, RLS, MVDR/Capon, MUSIC for direction-finding) tune weights to maximise SNR or null interferers. Used in EW, satcom, and beam-tracking 5G UE.
- Hybrid analog-digital architectures — common at mmWave 5G (28/39 GHz). N elements share M ≪ N RF chains via analog phase shifters; digital baseband does beam selection and MIMO. Compromise between cost (one RF chain per element is unaffordable at 64+ elements) and flexibility.
- Massive MIMO — TDD reciprocity lets the base station learn the channel from uplink pilots and beamform on the downlink. Standardised in 3GPP TS 38.211/214 (5G NR). Typical config: 32 or 64 cross-polarized elements (gNB AAU), 4–8 layer spatial multiplexing.
- Beam codebook / beam management in 5G NR — UE searches over SS-block beams, reports the strongest with L1-RSRP, base station refines with CSI-RS. Fundamental to mmWave coverage.
- AESA (Active Electronically Scanned Array) — every element has its own transmit/receive module (TRM) with a HPA, LNA, phase shifter, attenuator, T/R switch. The architecture of every modern fighter, naval, and ground-based radar (APG-83, SPY-6, EL/M-2084). GaN HEMT TRMs from Qorvo, Wolfspeed, MACOM dominate at X-band and up.
- Subarray architecture — for very large apertures (10⁴+ elements) the array is partitioned into subarrays each with its own phase shifter at the subarray level + digital beamforming across subarrays. Reduces complexity at the cost of scan angle.
- Calibration — element-to-element gain/phase matching must be measured and corrected (typically with built-in coupling networks); thermal drift, ageing, and partial failures need closed-loop correction. Mil-grade arrays calibrate continuously in service.
- Thermal management — mmWave T/R modules dissipate 1–3 W each in a few mm³. Power density is a hard limit; liquid cooling appears at 10 W/cm² and above (Patriot, SPY-6).
- Power-amplifier back-off — class-AB and Doherty PAs at mmWave run 6–10 dB output back-off to keep ACLR below 3GPP TS 38.104 limits (-30 dBc OFDM EVM). Forces PA sizing at peak ≈ 4× average; total array DC consumption is dominated by PA bias.
- EIRP regulation per element vs total — FCC, ETSI, and ITU-R limits apply to total EIRP, not per-element. A 64-element array at -10 dBm/element radiates with coherent gain to reach +25 dBm EIRP (40 dB gain) — well above any per-element limit. Mass-market mmWave UE (3GPP TS 38.101-2) is capped at +43 dBm EIRP.
- Beam-tracking algorithms — mmWave 5G UE moves through a beam codebook (typically 32–64 beams) using L1-RSRP measurement; mobility scenarios (vehicular at 3 km/h up to 500 km/h NTN) drive the codebook refresh rate. Outage occurs when beam-loss exceeds the refresh interval — a key reliability metric.
Polarization-diversity arrays
A typical 5G base-station active antenna unit (AAU) is 32T32R with 16 cross-polarized (±45° slant) element pairs. This gives:
- 3 dB polarization diversity gain (averages out fading uncorrelated between polarizations)
- 2-layer spatial multiplexing per UE without additional aperture
- Beamforming in azimuth + elevation independently per polarization
Vendors: Ericsson AIR 6449, Nokia AirScale Massive MIMO AAU, Huawei MetaAAU, Samsung CDU50, ZTE A9611S. Power amplifiers per element have evolved from GaAs PHEMT to GaN-on-SiC HEMT (Wolfspeed, Qorvo) for higher PAE and thermal density.
8. Edge cases / gotchas
- Hand detuning — gripping a handset shifts antenna resonance by 50–200 MHz and adds 3–10 dB efficiency loss. Mitigation: closed-loop aperture-tuner ICs (Qorvo QM77032, ST PTIC, Murata) that retune the antenna with the cellular RF transceiver based on VSWR sensing.
- Body absorption (SAR) — FCC limits handset SAR to 1.6 W/kg averaged over 1 g (US) or 2.0 W/kg over 10 g (Europe IEC 62209). Forces antenna placement away from head/torso, transmit-power back-off (“MTPL”), and adaptive beam-steering at mmWave.
- Multipath fading — indoor Rayleigh fading produces ±10 dB envelope swings. Mitigated by MIMO, antenna diversity (spatial, polarization, pattern), or frequency hopping.
- Antenna near metal — adjacent ground planes, chassis, batteries detune resonance and degrade efficiency. Keep-out zones of ≥ 5 mm at 2.4 GHz; ≥ 1 mm at 28 GHz. Vendor app-notes (Antenova, Taoglas) specify exact keep-out for each chip antenna.
- Polarization mismatch indoors — multipath partially randomises polarization, costing 1–3 dB on average. CP antennas tolerate this better at fixed -3 dB penalty.
- Surface waves in PCB substrate — high-ε_r thick substrates trap power in TM₀ surface waves that radiate from board edges instead of the patch. Limit ε_r·h < 0.3·λ to suppress.
- Mutual coupling between adjacent elements (S₂₁ between feeds) — distorts patterns and impedances. Quantified by full-wave EM simulation or measurement; mitigated by ≥ 0.5·λ element spacing or decoupling networks (Lange couplers, EBG structures).
- Connector loss + cable loss at GHz — RG-178 / RG-316 cables show 1–3 dB/m at 2.4 GHz, 5–10 dB/m at 28 GHz. SMA connectors 0.1–0.3 dB each. Cumulative: a 2 m test cable at 6 GHz might cost 6 dB before the antenna sees the signal.
- Ground-plane truncation — a patch on a finite ground plane radiates differently than the infinite-GP analytical model predicts. Front-to-back ratio drops as GP shrinks below ~1.5·λ across.
- EMC + intentional antenna coexistence — an antenna can also be a leakage path for digital harmonics. FCC Part 15B (unintentional radiated) and 15C (intentional, the antenna itself) coexist in the same product and are tested separately. Aggressive notch filtering protects intentional radiator bands from digital harmonics.
- Phased-array graceful degradation — losing 1 of 256 elements drops gain by only 0.034 dB but raises sidelobes by ~6 dB. Calibration must detect and weight-around failed elements.
- Frequency-dependent beam squint — phase-shifter arrays scan vs frequency; time-delay-unit (TDU) arrays do not but cost more silicon.
- Quadrature imbalance in CP feed networks — 1 dB amplitude imbalance + 5° phase imbalance produces AR ≈ 2 dB; tight tolerances on quadrature hybrids are essential for CP fidelity.
- Manufacturing tolerance on patches — at 28 GHz a 50 µm copper-etch tolerance shifts patch resonance by ~70 MHz; on a 1 GHz-wide n257/n258 band that is a ~7% drift. PCB fabricator must hold ±25 µm or design includes post-fab tuning (laser-trim or screw-tuned cavity).
- Substrate anisotropy — woven-glass laminate (FR-4, Megtron) has ε_r 2–5% different parallel vs perpendicular to weave; in mmWave patch arrays this produces per-element resonance variation. Use Rogers RO3003 (PTFE-glass, isotropic) or low-Dk spread substrates.
- Radome / radome loss — protective covers (polycarbonate, ASA, fiberglass-epoxy) add 0.3–1.5 dB attenuation and can introduce reflection resonances if thickness is ≠ multiple of λ/(2·√ε_r). Tuned radomes (sandwich, A-sandwich, C-sandwich) optimize broadband transparency.
- Lightning / ESD protection — outdoor antennas must include DC grounding (gas-discharge tube, λ/4 shorted stub) to drain static buildup and survive nearby strikes (IEC 62305).
- Wind loading on dishes — a 1.2 m offset DBS dish at 35 m/s sees ~250 N drag; mount and feed-arm stiffness must avoid pattern blurring at hold-tracking accuracy.
9. Measurement / characterization
- Anechoic chamber — RF-absorber-lined room; valid far-field measurement requires d > 2D²/λ. Typical chambers: 5 m × 3 m × 3 m for f > 1 GHz; 20 m+ for sub-GHz. Anechoic performance > 30 dB quiet-zone for pattern tests, > 40 dB for radar cross-section.
- Near-field range — sample E-field on a plane / cylinder / sphere close to the antenna, FFT-transform to far-field. Standard technique for large apertures or where chamber far-field is impossible. Vendors: MVG/Satimo (Stargate, StarLab), NSI-MI, ETS-Lindgren. CTIA-OTA accredited labs almost always use spherical near-field.
- Reverberation chamber — metal-walled room with mode stirrer; gives statistical (Rayleigh) measurement of total radiated power. Faster than anechoic and standardized in IEC 61000-4-21 (immunity) and for OTA testing of small wireless devices. Bluetooth and Wi-Fi pre-compliance frequently uses RC.
- CTIA OTA test for handsets — Total Radiated Power (TRP) and Total Isotropic Sensitivity (TIS) averaged over the full sphere, with the device under test held in a SAM (Specific Anthropomorphic Mannequin) head/hand phantom. Operator certification reference (Verizon, AT&T, T-Mobile minimum thresholds per band).
- VNA-based pattern — Keysight PNA-X or Rohde&Schwarz ZVA drives the antenna under test and a fixed probe; rotation positioner sweeps θ and φ; complex S₂₁ at each angle reconstructs pattern + polarization. Single-frequency or swept.
- 3D pattern — full spherical scan (e.g. 360° azimuth × 180° elevation, 1–5° steps); processed into 2D contour plots, gain integration for directivity, and AR analysis.
- NFC / RFID antenna tests — H-field amplitude at fixed distances (ISO 14443 NFC: 1.5 A/m at the operating volume), mutual-coupling measurement between reader and tag.
- Calibration standards — standard gain horn (SGH) with NIST-traceable gain at every band (e.g. Narda 640-series); 3-antenna gain extrapolation when no calibration reference exists; substitution method with calibrated antenna of known gain.
Frequency-band recommendations
| Band | Frequency | Typical antenna | Where used |
|---|---|---|---|
| MF / AM broadcast | 0.3–3 MHz | Loaded vertical or loop | AM radio, marine, beacons |
| HF | 3–30 MHz | Dipole, Yagi, log-periodic, tapped vertical | Amateur, military, OTHR |
| VHF | 30–300 MHz | Dipole, Yagi, ground-plane, J-pole | FM broadcast, ATC, VHF marine |
| UHF | 300 MHz–3 GHz | Patch, PIFA, Yagi, helix, panel | TV, cellular sub-3 GHz, GPS L1/L5, Wi-Fi 2.4 |
| L-band | 1–2 GHz | Patch, helix | GPS, GLONASS, Inmarsat, Iridium |
| S-band | 2–4 GHz | Patch, slot, microstrip array | Wi-Fi 2.4, Bluetooth, weather radar, S-DARS |
| C-band | 4–8 GHz | Patch, dish, waveguide horn | Satellite TVRO downlink, fixed point-to-point, Wi-Fi 6E |
| X-band | 8–12 GHz | Patch, slot array, dish | Military radar, satellite uplink, marine X-band |
| Ku-band | 12–18 GHz | Offset dish, flat panel array | DBS, VSAT, FSS satcom |
| K-band | 18–27 GHz | Patch array, gap waveguide | Automotive K-band radar (gone to 77 GHz), point-to-point links |
| Ka-band | 27–40 GHz | Patch array, reflectarray, phased array | 5G n257/n258 (28 GHz), HTS satellite, weather radar |
| V-band | 40–75 GHz | Patch sub-array, gap waveguide | 60 GHz WiGig, 57–71 GHz unlicensed, automotive 77 GHz |
| W-band | 75–110 GHz | Sub-array, dielectric lens | Automotive 77 GHz radar, mm-wave imaging |
| Sub-THz | 110–300 GHz | Photonic / lens-coupled | 6G research, atmospheric sensing |
10. Tools / software
Full-wave EM solvers
- Ansys HFSS — frequency-domain FEM, industry default for antennas. Handles all common geometries; strong adaptive meshing.
- Dassault CST Studio Suite — time-domain FIT/FDTD (Transient Solver), frequency FEM (T-Solver), MoM, asymptotic SBR for ultra-large problems. Most comprehensive single tool.
- Altair FEKO — Method of Moments primary, with FEM and FDTD; strong for electrically large radiating structures (vehicle-mounted antennas, ship integration).
- Remcom XFdtd — FDTD specialist; favoured for transient + biological (SAR analysis on full body models).
- openEMS — open-source FDTD (Octave/Python front-end); credible accuracy with no license cost.
- NEC2 / NEC4 — Method-of-Moments wire codes (Lawrence Livermore origin); NEC2 free, NEC4 export-controlled. Front-ends: 4nec2, xnec2c, EZNEC.
- MMANA-GAL, AC6LA Antenna Modelling tools — free MoM with GUI; common in amateur radio.
- SimNEC — modern MoM with FOSS leanings.
Quick design + matching
- AntennaMagus (Altair) — large library of parametrised antenna designs; exports directly into FEKO/HFSS.
- Keysight ADS with Momentum — planar 2.5D MoM for PCB-integrated antennas + matching circuits; full circuit + EM co-simulation.
- AWR Microwave Office — circuit + planar EM; smith-chart-centric workflow.
- scikit-rf (Python, open-source) — S-parameter manipulation, matching networks, calibration math.
Phased array
- Ansys EMA3D Charge / Ansys HFSS SBR+ — large arrays via hybrid methods.
- MATLAB Phased Array System Toolbox — beamforming algorithms, array geometry studies, system-level simulation.
- Synopsys Saber / Ansys Twin Builder — system-level TRM simulation with thermal coupling.
Chamber + measurement
- ETS-Lindgren — anechoic chambers, antennas, instruments. The dominant North American supplier.
- Microwave Vision Group (MVG / Satimo) — Stargate (large spherical NF), StarLab (small mid-frequency NF), SG (compact NF). Dominant in CTIA-OTA labs.
- Rohde & Schwarz CMW500 / CMX500 + chamber — integrated test systems for cellular OTA.
Connector + cable ecosystem
Antenna engineering inevitably touches a small zoo of RF connectors. Picking the wrong one is a multi-week debug:
- SMA / RP-SMA (3.5 mm) — universal RF lab and Wi-Fi router workhorse; usable to ~18 GHz; female centre pin vs reverse-polarity (Wi-Fi consumer mandated RP-SMA by FCC to discourage user antenna swaps).
- N-type (7 mm) — outdoor cellular and microwave; rated to 18 GHz with high power handling (200 W CW typical).
- TNC — threaded BNC, used in GPS receivers and military for vibration resistance.
- MCX / MMCX / U.FL / W.FL / IPEX — miniature snap-on / surface-mount used to bring out cellular and Wi-Fi antennas from modules; usable to 6 GHz (U.FL) up to 18 GHz (W.FL gen 3).
- 2.92 mm / K-connector / 2.4 mm / 1.85 mm / 1.0 mm — precision instrumentation connectors for mmWave (40 / 50 / 67 / 110 GHz respectively).
- Waveguide flanges (WR-15, WR-12, WR-10, WR-8, WR-5) — at V/W/D-band beyond ~50 GHz, coaxial loses prohibitively and rectangular waveguide takes over.
Cable choices: RG-58 (lab, low cost, 18 GHz limit), RG-316/178 (thin teflon for modules), LMR-240/400 (low-loss outdoor runs), Sucoflex/Gore PHASEFLEX (phase-stable test cables). Typical 2.4 GHz loss: 0.5 dB/m (LMR-400), 1.5 dB/m (RG-58), 3 dB/m (RG-178); at 28 GHz: 2 dB/m (LMR-Ultraflex), 8 dB/m (RG-178), 15 dB/m thinner cables.
Industrial reference designs
- Ignion (formerly Fractus Antennas) — Virtual Antenna IP for IoT modules; small SMD parts that route on PCB ground.
- Antenova — ceramic SMD antennas (Lucida, Reefa, etc.) for BLE / Wi-Fi / GPS.
- Taoglas — extensive catalogue from chip antennas to large rugged GNSS pucks.
- Pulse Electronics (Yageo) — broad antenna + magnetics portfolio.
- Murata — chip antennas, LTCC modules, integrated SAW + antenna duplexers.
- TDK / Epcos — chip antennas and matching networks.
- Linx Technologies — module-integrated antennas and matching kits.
- Wireless silicon vendors (TI CC-series, Nordic nRF, ST BlueNRG, NXP/QN9080) ship reference PCB antennas as Gerber files and BoMs.
- Specialist mmWave IP — Sivers Semiconductors (5G NR FR2 beamformers, BFM06010), Renesas / IDT (mmWave RFICs), Anokiwave (Ka/Ku-band beamformer + array reference designs), Movandi (repeaters), MixComm / Sivers Wireless (5G EVK), Tower Semiconductor SiGe BiCMOS foundry process for 60/77 GHz transceivers.
EM solver method comparison
| Method | Domain | Best for | Limitation | Tools |
|---|---|---|---|---|
| Finite Element Method (FEM) | frequency | resonant cavities, antennas, packages | meshing 3D volumes is memory-heavy | HFSS, COMSOL RF, Maxwell 3D |
| Finite-Difference Time-Domain (FDTD) | time | broadband, transient (UWB, ESD), bio-EM | staircased curved surfaces, fine mesh near small features | CST Transient, XFdtd, openEMS, MEEP |
| Method of Moments (MoM) | frequency | wire antennas, PEC scattering, electrically large planar | dense matrix scales as O(N²) memory, O(N³) solve | NEC, FEKO, ADS Momentum |
| Finite Integration Technique (FIT) | time / freq | hybrid, broadband + resonant | combines FDTD-like grid with frequency-domain solve | CST Studio |
| Multilevel Fast Multipole Method (MLFMM) | frequency | electrically very large MoM | requires careful parameter tuning | FEKO MLFMM, HFSS IE |
| Shooting and Bouncing Rays (SBR) | high-freq asymptotic | radar cross-section of vehicles/aircraft | fails for small features < λ/2 | HFSS SBR+, CST Asymptotic |
| Physical Optics (PO) | high-freq asymptotic | reflector dishes, large parabolic | no diffraction (extend with PTD/UTD) | Ticra GRASP, FEKO PO |
| Transmission Line Matrix (TLM) | time | EMC, broadband | similar trade-offs to FDTD | CST TLM |
11. Cross-references
[[Engineering/electromagnetics-engineering]]— Maxwell foundation, free-space wave impedance (377 Ω), polarization, Friis equation, far-field criterion[[Engineering/rf-design]]— companion Tier 2; matching networks, Smith chart, noise figure, link budget at the component level[[Engineering/pcb-design]]— PCB-level antenna integration, controlled impedance, ground-plane keep-out, EMC interactions[[Engineering/semiconductor-devices]]— GaN/GaAs HEMT for HPA in T/R modules, silicon BiCMOS for mmWave phase shifters[[Engineering/signal-processing-dsp]]— digital beamforming, FFT-based DOA, adaptive nulling algorithms[[Engineering/transformers-power-systems]]— coupled-coil model behind wireless power and NFC (near-field)- planned
[[Engineering/photonics]]— optical antennas, free-space optical links, integrated photonic phased arrays (OPA) - planned
[[Engineering/rf-design]]— modulation, MIMO theory, channel models that drive antenna requirements [[Languages/Tier3/network-protocol-dsls]]— protocol-layer wrapper (Wi-Fi 802.11ax/be, LTE, 5G NR, Bluetooth LE, Zigbee, LoRa)[[Languages/Tier3/scientific]]— Phased Array System Toolbox, Antenna Toolbox workflows- planned
[[Engineering/spacecraft-attitude-control]]— orbital geometry, link-budget margin, doppler tracking for LEO constellations - planned
[[Engineering/rf-design]]— pulse compression, MTI/STAP, AESA architectures, range/Doppler resolution from antenna beamwidth - planned
[[Engineering/pcb-design]]— radiated emissions test, EMC chambers (close cousin of anechoic), intentional-vs-unintentional radiator distinction [[Robotics/sensors-pose-motion]]— RFID localization, UWB ranging (Decawave/Qorvo DW3000), GNSS receivers
12. Citations
- Balanis, C. A. (2016). Antenna Theory: Analysis and Design (4th ed.). Wiley. The canonical text — patch design equations, array theory, measurement.
- Stutzman, W. L. & Thiele, G. A. (2012). Antenna Theory and Design (3rd ed.). Wiley. Practitioner’s companion to Balanis; stronger on measurement and modern arrays.
- Kraus, J. D. & Marhefka, R. J. (2002). Antennas (3rd ed.). McGraw-Hill. Classic; helical antenna originator.
- Volakis, J. L. (Ed.) (2018). Antenna Engineering Handbook (4th ed.). McGraw-Hill. Industry reference; chapters by specialists.
- Mailloux, R. J. (2017). Phased Array Antenna Handbook (3rd ed.). Artech House. The canonical phased-array reference.
- Hansen, R. C. (Ed.) (1985, reprint). Microwave Scanning Antennas (Vols. I–III). Peninsula. Foundational AESA / phased-array engineering.
- Pozar, D. M. (2011). Microwave Engineering (4th ed.). Wiley. Transmission lines and matching networks underlying every antenna feed.
- Cheng, D. K. (1989). Field and Wave Electromagnetics (2nd ed.). Addison-Wesley. Engineering EM with antenna chapter.
- Hertz, H. (1888). Über die Ausbreitung der elektrischen Kraft. Annalen der Physik. The first experimental EM-wave dipole.
- Yagi, H. (1928). “Beam Transmission of Ultra Short Waves.” Proc. IRE 16(6): 715–740 — Yagi-Uda directional array (with Uda’s 1926 reflector/director work).
- Hansen, W. W. & Woodyard, J. R. (1938). “A New Principle in Directional Antenna Design.” Proc. IRE 26(3): 333–345. End-fire condition for maximum directivity.
- Schelkunoff, S. A. (1943). Electromagnetic Waves. Van Nostrand. Hertzian-dipole derivation in modern form; array-factor theory.
- Friis, H. T. (1946). “A Note on a Simple Transmission Formula.” Proc. IRE 34(5): 254–256. The Friis equation.
- Booker, H. G. (1946). “Slot Aerials and Their Relation to Complementary Wire Aerials (Babinet’s Principle).” J. IEE 93(IIIA). Slot antennas.
- Dolph, C. L. (1946). “A Current Distribution for Broadside Arrays Which Optimizes the Relationship Between Beam Width and Side-Lobe Level.” Proc. IRE 34(6): 335–348. Chebyshev arrays.
- Kraus, J. D. (1947). Antennas. McGraw-Hill, first edition. Axial-mode helix.
- Carrel, R. L. (1961). “The Design of Log-Periodic Dipole Antennas.” IRE National Convention Record Part 1: 61–75. Log-periodic theory.
- Clarricoats, P. J. B. & Saha, P. K. (1969). “Theoretical Analysis of Cylindrical Hybrid Modes in a Corrugated Horn.” Electronics Letters 5(9): 187–189. Corrugated horn / hybrid HE₁₁ mode.
- Gibson, P. J. (1979). “The Vivaldi Aerial.” 9th European Microwave Conference Proceedings: 101–105. Tapered-slot UWB antenna.
- Marzetta, T. L. (2010). “Noncooperative Cellular Wireless with Unlimited Numbers of Base Station Antennas.” IEEE Trans. Wireless Comm. 9(11): 3590–3600. Massive-MIMO origin paper.
- IEEE Std 145-2013. IEEE Standard Definitions of Terms for Antennas. Authoritative terminology.
- IEEE Std 149-2021. IEEE Recommended Practice for Antenna Measurements. Measurement methodology.
- 3GPP TS 38.101-1/2 (Release 17, 2024). NR; User Equipment (UE) radio transmission and reception. 5G antenna requirements (FR1 sub-6 GHz, FR2 mmWave).
- 3GPP TS 38.211 (Release 17). NR; Physical channels and modulation. Reference signals used for beam management.
- FCC 47 CFR Part 15 — Subpart B (unintentional radiators), Subpart C (intentional radiators including Section 15.247 for 2.4/5 GHz ISM, 15.407 for U-NII). EIRP limits.
- ITU-R Recommendations BS, BT, F, M, P, S, V series — international RF allocation, propagation models, antenna patterns by service.
- HFSS / CST / FEKO / NEC / openEMS user documentation — solver-specific best-practice references.
- McLean, J. S. (1996). “A re-examination of the fundamental limits on the radiation Q of electrically small antennas.” IEEE Trans. Antennas Propag. 44(5): 672–676. Modern correction to the Chu-Wheeler limit.
- Chu, L. J. (1948). “Physical Limitations of Omni-Directional Antennas.” J. Applied Physics 19(12): 1163–1175. Foundational small-antenna Q theorem.
- Wheeler, H. A. (1947). “Fundamental Limitations of Small Antennas.” Proc. IRE 35(12): 1479–1484. Companion Wheeler limit.
- Taylor, T. T. (1955). “Design of Line-Source Antennas for Narrow Beamwidth and Low Side Lobes.” IRE Trans. Antennas Propag. 3(1): 16–28. Taylor distribution for low SLL.
- CTIA 01.81 Over-the-Air Test Plan for User Equipment. Latest revision (2024) for CTIA-certified handset OTA testing.
- IEC 62209-1528 and IEEE Std 1528-2013 Recommended Practice for Determining the Peak Spatial-Average SAR. Handset SAR test methodology.