Angle-of-Arrival is showing up in more and more RTLS specs because it hits a sweet spot: high accuracy, simpler tags, and you don’t have to carpet-bomb a site with Anclajes UWB. In our deployments, the real challenge isn’t “does AoA work?” It’s “can we get 10 cm where it matters, fast, without turning installation into a science project?”
Let’s make it practical. We’ll talk BLE direction finding basics, then the deployment rules that keep install time sane, and finally, how Lansitec’s AoA gateways fit into a real 0.1 m plan.
What 10 cm Accuracy Really Means in BLE AoA Positioning
Bluetooth Direction Finding (AoA/AoD) adds a Constant Tone Extension (CTE) to BLE packets, so a receiver with an antenna array can capture I/Q samples and estimate the signal’s angle from phase differences.
Here’s the part people skip: angle is not the same as position.
To turn angles into coordinates, you need a location engine to fuse measurements from multiple locators (puertas de enlace), plus good geometry and calibration. Bluetooth’s own technical overview calls out antenna switching, sampling slots, and HCI configuration as core parts of the system.
So yes, 10 cm is achievable in ideal conditions, but the “ideal” word matters. Nordic’s direction finding material is blunt about this: centimeter-level accuracy can happen, yet real-world RF introduces multipath and other constraints. (1)
BLE AoA Deployment Best Practices for 2026 Installations
If you want 0.1 m in specific zones (dock doors, tool cribs, medical equipment rooms), you need three things. Miss any one, and you’ll chase ghosts for days.
- Geometry you can trust: multiple puertas de enlace see the tag from meaningfully different angles.
- RF you can live with: reduce strong reflectors, or at least keep them consistent.
- Calibration you can repeat: short, structured, and done per zone.
Silicon Labs’ direction-finding docs emphasize that antenna switching patterns and configurations matter, and that custom arrays need the right setup to avoid poor angle estimates. (2)
Three Deployment Rules That Make or Break AoA Accuracy
- Mount overhead when you can. Ceiling mounting reduces body shadowing and gives cleaner geometry for tags moving on the floor.
- Design for overlap, not coverage. “I can hear it” is not “I can locate it.”
- Calibrate only where you need 10 cm. Don’t pay the 0.1 m tax across the whole site.
Choosing the Right BLE AoA Gateways for Indoor and Outdoor RTLS
Lansitec’s AoA gateway line is built around antenna arrays and practical mounting, with clear limits on install height and coverage radius.
BLE AoA Gateway Selection by Use Case and Accuracy
| Puerta | Where it fits | Positioning claim | Mounting limits | Power + enclosure |
|---|---|---|---|---|
| Puerta de enlace AoA para interiores AG1 | Indoor zones, straightforward PoE deployments | “sub-meter accuracy (0.1 m)” | height max 15 m, radius up to 2× height (cap 15 m) | PoE 802.3af or 12–30 V DC, IP66 |
| AG3 Positioning AoA Gateway | Higher-precision indoor RTLS, multi-backhaul sites | 0,1–1 m | height max 15 m, radius up to 2× height (cap 15 m) | DC 12–30 V, Wi-Fi 802.11ac, IP66 |
| Puerta de enlace AoA para exteriores AG4 | Yards, semi-outdoor, open-air event spaces | 0.1 m | height max 30 m, radius up to 2× height (cap 30 m) | PoE 802.3af or 12–30 V DC, IP66 |
A practical note: Lansitec explicitly ties coverage radius to mount height (up to 2× height, with a hard cap). That’s gold for fast planning, because you can do first-pass layout math without a full RF survey.
How to Achieve 10 cm AoA Accuracy Without Over-Installing Hardware
Optimal Mounting Height for BLE AoA Gateways
- Indoors (AG1/AG3): treat 6–10 m as a “sweet range” when the ceiling allows it. You get better sightlines, and your radius budget stays realistic because Lansitec caps effective radius at 15 m anyway.
- Outdoor (AG4): you can mount higher (up to 30 m), but don’t assume higher always means better accuracy. Higher often means a weaker signal and more reflections off vehicles and fencing.
Gateway Density Planning for Reliable AoA Positioning
For 0.1 m zones, plan locator placement so a tag typically has:
- 3+ strong puertas de enlace in view
- wide angle separation between them (avoid all puertas de enlace being on the same wall line)
- consistent line-of-sight along the main movement path
This matches how AoA systems work at the spec level: the receiver’s antenna switching and I/Q sampling during CTE drives angle estimates, and the location engine needs multiple good angles to triangulate well. (1)
BLE AoA Calibration Process That Actually Works
If you want 10 cm, you will calibrate. The trick is to calibrate like a production team, not like a research lab.
Silicon Labs’ direction finding notes make the practical point: antenna array configuration and switching patterns matter, and custom setups need correct configuration to avoid bad results. (2)
A fast calibration workflow we like:
- Step A: pick 6–12 “truth points” in the 10 cm zone (corners, center, two mid-edges).
- Step B: collect short samples at each point (same tag orientation, same height).
- Step C: validate with a simple walk path (a rectangle loop, 2–3 minutes).
- Step D: lock the config, then move on.
No heroics. No 200-point grids. If the zone fails validation, you fix geometry first (add one locator or shift one), then recalibrate.
Common BLE AoA Accuracy Problems and How to Fix Them
| What breaks 0.1 m | What you’ll see | Fast fix (least install pain) |
|---|---|---|
| Heavy multipath from metal racks | “Good one minute, awful the next” | Raise locators, add one more locator for angle diversity |
| Poor geometry (angles too similar) | Errors stretch along one axis | Move one gateway to a different side, don’t just add more on the same wall |
| Inconsistent tag orientation | Jitter when people carry tags | Use a consistent mounting position (helmet, badge plane, asset bracket) |
| Rushed calibration | Whole zone biased by 0.3–1 m | Recalibrate with fewer, cleaner truth points |
This aligns with the fundamentals: AoA relies on stable phase measurements during CTE and on predictable antenna-switching behavior. Garbage in gives garbage angles out. (1)
Real-World BLE AoA Layout Examples (Indoor vs Outdoor)
Indoor Warehouse AoA Layout for Forklifts and Pallets (AG3)
- Put AG3 locators overhead, target 10–12 m mounting height where possible (within the 15 m max).
- Use the “2× height” radius rule to draft spacing, then tighten density around choke points (dock doors, battery swap area).
- Calibrate only the choke points for 10 cm, run the rest at 0.3–1 m and save time.
AG3 specifically targets high-accuracy indoor positioning, with Bluetooth 5.1, antenna array, and a stated 0.1–1 m accuracy range.
Outdoor AoA Layout for Yards and Open Industrial Sites (AG4)
- Montar AG4 on building edges or poles, keep within the 30 m height limit.
- Expect more RF chaos (vehicles, fences, stacked containers). Plan extra overlap and don’t overpromise 10 cm everywhere.
- Use 10 cm only for “must-not-lose” zones (high-value storage rows, entry gates).
AG4 is built for this: IP66, PoE 802.3af, and an AoA positioning claim “within 0.1 m.”
When BLE AoA Is the Right Choice for 10 cm Accuracy in 2026
AoA in 2026 sits in a rare sweet spot: you can push toward 0.1 m accuracy without turning your site into a UWB construction zone. But AoA only behaves when you treat it like a system, not a spec sheet. You need clean geometry, sensible gateway height and density, and a calibration routine you can repeat without heroics.
Our rule of thumb stays simple. Use Lansitec AoA puertas de enlace (AG1/AG3 indoors, AG4 outdoors) where their height and radius constraints make planning predictable, then “buy” 10 cm only in the zones that deserve it. The rest of the site can run happily at sub-meter. Do that, and you’ll hit high accuracy where it matters, while keeping install time and project risk under control.
Preguntas frecuentes
About BLE AoA Accuracy and Deployment FAQs
Can AoA really deliver 0.1 m accuracy, or is that marketing?
AoA uses BLE Direction Finding with a Constant Tone Extension (CTE) so a locator can sample I/Q data and estimate an angle. Then your location engine combines angles from multiple locators into a position. That’s why accuracy depends on geometry and RF conditions, not just “AoA supported.” (1)
How many puertas de enlace do I need for a 10 cm zone?
For reliable 10 cm in a real site, plan for 3+ locators with meaningful angular separation so the engine can fuse multiple angle measurements. Bluetooth direction finding describes the core mechanism (CTE sampling and angle estimation), which inherently benefits from multiple independent perspectives for position solving. (1)
What’s the fastest way to avoid a calibration time sink?
Keep calibration repeatable and configuration-correct. Vendor guidance focuses heavily on antenna array configuration, switching patterns, and correct direction-finding setup because “sloppy setup” creates bad angles. Calibrate only the zones where you truly need 10 cm, validate quickly, then lock the config. (2)
What mounting height should I target with Lansitec AoA puertas de enlace?
Use Lansitec’s published constraints for quick planning: AG1/AG3 supports up to 15 m of mounting height, and the radius is up to 2× the height (capped at 15 m). AG4 supports up to 30 m height and radius up to 2× height (capped at 30 m). These limits directly shape deployment density and zone design.
Why does AoA look good when it’s quiet and worse when the site is busy?
AoA relies on clean phase information during the CTE. Dynamic environments introduce shadowing and multipath, which increases angle noise. That’s why busy warehouses and people-heavy areas can show more jitter unless you design overlap and geometry properly. (1)
When is AoA not the best fit?
If your environment forces heavy multipath and unstable propagation (lots of moving metal, tight aisles, constant obstruction) and you can’t add overlap or control mounting, AoA performance becomes harder to stabilize because the direction-finding angle estimate depends on the received signal quality during CTE sampling and correct antenna switching behavior. (1) (2)
References and further readings:





