Line-of-sight (LoS) is critical for mmWave 5G because signals at 26 to 28 GHz do not diffract around obstacles, penetrate buildings, or scatter effectively through foliage, unlike sub-6 GHz signals that survive significant non-line-of-sight (NLoS) conditions. A single building corner, tree canopy, or low wall between transmitter and receiver can cause 20 to 35 dB of additional loss, making the difference between gigabit throughput and no usable signal. Planning for LoS means treating 3D building geometry, vegetation, and terrain as the primary coverage determinants rather than just link budget parameters.
The fundamental reason mmWave signals require LoS is the combination of short wavelength and high atmospheric absorption:
The practical result: mmWave coverage exists only where there is a geometrically clear path, or a viable specular reflection path, between the small cell antenna and the receiver.
| Parameter | LoS at 28 GHz | NLoS at 28 GHz |
|---|---|---|
| Achievable throughput | Multi-Gbps (2 to 10+ Gbps) | 10 to 500 Mbps (via reflection) or effectively zero |
| Reliable coverage range | 150 to 400 m | 30 to 100 m via reflection |
| Link margin | Typically 10 to 20 dB margin | Near-zero or negative margin |
| Handover stability | Predictable boundaries | Highly environment-dependent |
| Planning model accuracy | High (geometry-deterministic) | Low (surface roughness and material variation) |
This binary nature means mmWave network planning is fundamentally about LoS corridor mapping, identifying which streets, open spaces, and building facades have clear sightlines to candidate antenna positions.
LoS computation requires a 3D model of all obstructions in the planning area. The minimum data requirements:
Two computational approaches are used:
Viewshed analysis: a geometric calculation that determines, for each ground point in the planning area, whether a clear sightline exists to the candidate antenna position. Viewshed analysis is fast and suitable for initial candidate screening. It does not account for Fresnel zone clearance, reflection paths, or atmospheric effects, but it correctly identifies LoS/NLoS zones.
Ray-tracing: a deterministic propagation model that computes all significant propagation paths, including direct LoS, specular reflections, and diffraction, from the 3D geometry. Ray-tracing is more accurate than viewshed analysis because it quantifies the received signal level on each path, including NLoS paths via reflection. It is also computationally intensive and requires LOD2 building models with material attributes.
For mmWave planning, ray-tracing is the recommended approach for final site selection and coverage verification. Viewshed analysis is appropriate for initial candidate screening across large numbers of candidate positions.
Even when there is geometric LoS between transmitter and receiver, the Fresnel zone, the ellipsoidal volume around the direct path within which reflections would cause destructive interference, must be clear. The first Fresnel zone radius at 28 GHz over a 200 m path is approximately 0.8 m. This means that a tree branch or parapet that does not block the geometric line of sight may still partially obstruct the Fresnel zone and cause measurable signal degradation.
At 28 GHz:
Partial Fresnel zone obstruction causes signal reduction proportional to the fraction obstructed. A planning dataset that captures building heights to ±1 to 2 m and includes per-tree canopy heights is necessary to correctly assess Fresnel zone clearance.
| Data Layer | Required Attribute | Accuracy Needed |
|---|---|---|
| Buildings | Footprint polygon and roof height | ±1 to 2 m height |
| Buildings | Roof geometry (LOD2) for parapet height | ±0.5 to 1 m for mmWave |
| DTM | Ground elevation | ±0.5 to 1 m |
| Vegetation | Individual tree canopy top height | ±1 to 2 m |
| Vegetation | Trunk base height | ±1 m |
| Walls/Fences | Height and location | ±0.5 m |
Missing or inaccurate data in any of these layers creates errors in the LoS determination that cannot be corrected through model calibration; they are geometric input errors, not parameterization errors.
Tree foliage state has a significant effect on mmWave propagation. In leaf-on conditions (spring through autumn in temperate climates), deciduous tree canopy adds 10 to 20 dB of attenuation compared to bare branches. This means:
LuxCarta's 3D vegetation data captures canopy extent with trunk/canopy height separation, enabling propagation models to correctly assign attenuation coefficients at each height layer, a necessary capability for seasonal coverage analysis.
LuxCarta's core product set directly addresses the LoS planning problem. Building extraction at 93%+ capture rate with individual building heights ensures that LoS computation is not invalidated by missing structures. LOD2 building models from LuxCarta's deep learning extraction technique (90%+ accuracy, 4x productivity per SPIE 2023) provide the roof geometry detail needed for Fresnel zone clearance analysis.
Vegetation data with trunk/canopy separation resolves the height-dependent attenuation problem that causes systematic LoS prediction errors along tree-lined streets. The IGARSS 2024 wall and fence extraction capability (80.31% precision, 86.32% recall) extends LoS accuracy to the sub-building obstacle level that matters most for FWA customer qualification and dense urban mmWave site selection.
Telefónica Deutschland's validation of LuxCarta data at 26 GHz for FWA planning confirms that data precision is sufficient to make premises-level LoS qualification decisions, the core commercial outcome that mmWave LoS planning exists to support.
LoS (line-of-sight) is a binary geometric condition: is there a direct unobstructed path between transmitter and receiver? Fresnel clearance goes further: even when geometric LoS exists, the ellipsoidal Fresnel zone around the path must be sufficiently clear of obstructions to avoid destructive interference. At 28 GHz the Fresnel zone is narrow (under 1 m radius for typical small cell link distances), so most geometric LoS paths also have adequate Fresnel clearance, but marginal cases near tree canopy or rooftop edges require explicit Fresnel analysis.
Outdoor mmWave cells provide negligible useful indoor coverage through standard building walls. However, mmWave is increasingly used for indoor applications with dedicated indoor small cells or distributed antenna systems (DAS). Indoor mmWave can deliver very high throughputs in venues like stadiums, airports, and convention centers where the cells are inside the building and LoS exists from ceiling or wall-mounted antennas to user devices.
Massive MIMO beamforming helps primarily in the LoS case by concentrating energy in the direction of the target device, improving range and throughput. In NLoS, beamforming can use multi-beam approaches to simultaneously illuminate a direct LoS path and reflection paths, potentially recovering NLoS links that a single-beam system cannot. However, the fundamental NLoS path loss at mmWave remains very high regardless of beamforming gain; beamforming does not overcome the physics of diffraction-free propagation.
Large vehicles (buses, trucks) are significant mmWave obstructions that cannot be represented in static geodata. Their effect is typically modeled as a stochastic blockage event in the link budget, using statistical models from 3GPP or ETSI that estimate the probability of vehicular blockage as a function of street type, traffic density, and antenna height. Static 3D building data does not need to represent vehicles, but the link budget must include a vehicular blockage margin for street-level small cells near roadways.
The 3GPP TR 38.901 model defines the following path loss exponents for 28 GHz Urban Micro (UMi) scenarios: LoS path loss exponent approximately 2.0 (close to free space); NLoS path loss exponent approximately 3.19 (significantly steeper rolloff). This difference means that at 100 m distance, the NLoS path loss is approximately 20 to 25 dB higher than the LoS path loss, confirming why LoS availability is the primary determinant of mmWave cell range and throughput.
LuxCarta provides AI-powered 3D geospatial data solutions for telecom, simulation, and smart city applications worldwide. Learn more at luxcarta.com or explore on-demand extraction at BrightEarth.