Drag the router across the floor plan. Coverage is recomputed from the actual wall geometry using a dominant-path model — signal takes the cheapest route it can find, through masonry or around it via doorways. This is a two-room starter plan; use Import floor plan to load your own.
Click a router to select it, then drag it anywhere — including outdoors. “Find best spot” searches every reachable position for the placement with the most floor area above the target level.
A stairwell is a hole in the slab — signal crosses between floors there at no cost, which is usually the only reason an upstairs router reaches downstairs at all. Draw one over the stairs; it applies to both floors at once. Drag the router onto the other panel to move it between storeys.
The clutter exponent absorbs furniture, people and floor/ceiling reflections. N = 20 is pure free space; 28–30 is a heavily furnished apartment.
Openings are the main way signal reaches other rooms. Drag their handles on the plan to move them along a wall.
| Room | Area | Median | Weakest | Coverage | Speed at median |
|---|
Every point on the floor gets the strongest signal it can receive by any route. The floor is rasterised at 6 cm and a least-loss path search (Dijkstra) runs outward from the router across that raster. Each step pays two costs: geometric spreading, and whatever material it passes through. Because the search is free to route around a wall instead of through it, signal bends through doorways and along the courtyard exactly as it does in practice — which is the failure mode of naive line-of-sight ray casting, and the idea behind the Dominant Path Model of Wölfle & Landstorfer.
Materials are specified as loss per wall crossing at this plan's 37 cm thickness, then converted to dB per metre internally. The values are whole-assembly figures — what you measure with a radio either side of a real wall — not bare-slab laboratory transmission, which understates a built wall by several dB. Masonry is where the two bands diverge hardest: NIST measured 203 mm concrete at 29 dB on 2.4 GHz and 48 dB on 5 GHz, while drywall and glass are near-identical on both.
This is a 2D single-floor median-field predictor. It has no ceiling or floor slab, no explicit specular reflections, no furniture, and no fast fading. Real measurements scatter around a well-calibrated indoor prediction by roughly 5–8 dB, so read the contours as bands, not lines. Treat a result within about 6 dB of the target as "probably fine, go measure it."
Geometry was traced from the supplied floor plan and reconciled against its dimension chains: 37 + 400 + 74 + 98 + 136 + 538 + 37 = 1320 cm across the top, 1066 cm down the side. All six rooms reproduce their stated areas to within 2%. Interior circulation is inferred — the doors and openings are editable, and moving them changes the answer a great deal.