Predict WiFi Signal Through Brick Walls — A No-Survey Guide

The basic link budget

Received signal power, in dBm, is:

Pr = EIRP - PL_free_space - sum(wall_attenuations)

Three pieces:

• •EIRP — effective isotropic radiated power, in dBm. Sum of AP transmit power and antenna gain. A typical consumer AP at 20 dBm Tx with a 3 dBi omni antenna gives 23 dBm EIRP.
• •PL_free_space — free-space path loss in dB, computed from frequency and distance (Friis equation, log form).
• •sum(wall_attenuations) — the dB cost of every wall the signal crosses, summed. This is the "multi-wall" model.

That's the whole COST-231 multi-wall model in one line. Real RF planners add a tweak factor for the number of walls (the "non-linear wall correction") but for ≤4 walls the additive approximation is within a couple of dB.

Free-space path loss, properly

The free-space path-loss formula in log form, using metres and MHz:

PL(dB) = 20 * log10(d_m) + 20 * log10(f_MHz) - 27.55

This is just Friis' equation, rearranged so distance is in metres and frequency in MHz. Some textbooks write it with km and GHz; the constant changes accordingly. For WiFi work, the metres-and-MHz form is most convenient.

Quick reference values you can memorise:

Two practical observations: doubling the distance costs you 6 dB; doubling the frequency also costs 6 dB. That's why 5 GHz is "shorter range" than 2.4 GHz — same 6 dB penalty for operating up the band.

Wall attenuation, per material and band

ITU-R P.2040 is the recommendation that catalogues the electrical properties of building materials at common radio frequencies, and ITU-R P.1238 gives measured penetration-loss ranges. The figures below follow both, for typical wall thicknesses. They are the exact per-band values our solver applies — the same table published on our accuracy page, so the number you read here is the number the heatmap uses:

These are typical values. Actual attenuation depends on wall thickness, moisture content, and what's embedded (pipes, electrical conduit). Treat them as ±2 dB.

Worked example: 5 GHz AP, 8 m, one brick wall

Let's run the actual math. We have a UniFi-style AP at 20 dBm Tx, 3 dBi antenna, so:

EIRP = 20 dBm + 3 dBi = 23 dBm

The receiver (a phone) is 8 m away. Frequency is 5 GHz = 5000 MHz. Plug into the formula:

PL_free_space = 20 * log10(8) + 20 * log10(5000) - 27.55
              = 20 * 0.903 + 20 * 3.699 - 27.55
              = 18.06 + 73.98 - 27.55
              = 64.49 dB

One brick wall in the path adds 11 dB:

Total path loss = 64.49 + 11 = 75.49 dB

Pr = EIRP - PL = 23 - 75.49 = -52.5 dBm

−52 dBm on the far side of one brick wall at 8 m. That's a solid signal — green band, full speed link. Stretch the same single wall out to a 15 m diagonal across a small apartment and free-space loss grows to 69.95 dB, putting you at −58 dBm: still fine. Put two brick walls in that 15 m path and it bites — 69.95 + 22 = 91.95 dB of loss, so −69 dBm. That is below the −67 dBm usable edge the heatmap draws, and it is exactly the kind of room that looks fine on a floorplan and disappoints in practice.

This is exactly the math our tool runs across the heatmap — on a fine grid over the floor plan, for each AP and each grid point, compute distance to the AP, sum the wall attenuations on the line segment between them, and take the strongest of the per-AP results.

What this model misses

I want to be upfront about the limits of the additive multi-wall model. There are at least four things it doesn't represent, and you should know what they are before you trust the heatmap to within 2 dB:

• •Diffraction. Signals bend around wall edges. The model treats every wall as an infinite slab in the path; real walls end at doorways. Through a doorway your real attenuation is closer to free-space than to "brick".
• •Multipath. Signals also bounce off floors, ceilings, mirrors, and metal appliances. Sometimes they constructively interfere (you get a stronger signal than predicted); more often they fade (Rayleigh / Rician). Real-world variance from multipath is ±3–6 dB even in a static environment.
• •3D effects. The formula above is flat: it treats one floor as if signal only travelled horizontally. Real ceiling-mounted APs radiate in three dimensions and leak through the floor slab. For a single-floor apartment that is fine. For a 3-storey house, a flat model per floor is noticeably wrong — which is why our multi-floor mode (paid) is true 2.5D: an AP on one floor contributes to the floors above and below, through each ceiling slab's own attenuation and over the real 3D distance. How that works →
• •Antenna patterns. COST-231 assumes isotropic radiation. Real antennas have lobes — UniFi U7 Pro is roughly omni-horizontal but has a null directly above and below the AP. Directional antennas (sector, panel, dish) need a completely different model.

Net of all that, expect the predictive heatmap to be within ±6–10 dB of real-world measurements in a typical home or small office. That's enough to make placement decisions confidently. It's not enough to certify a survey for a regulated environment.

When to stop computing and just measure

Predictive planning is for the placement decision: do I need 1 AP or 2? Where do I put them? Is the bedroom going to be a dead zone? Once the APs are installed, you validate empirically with a phone-based survey app — the prediction tells you where to check, and reality tells you what's actually there. Both halves matter.

The split is also why a predictive planner can live in a web page and a survey tool cannot: this arithmetic needs no access to the WiFi radio, and measuring does. Can a WiFi heatmap work in a browser? →

If you'd rather not do the arithmetic by hand, the editor runs exactly the formulas above across a fine grid over your floorplan, with the wall attenuation values from the table above, and updates live as you drag APs around. It's free — place as many APs as you like; exports include five — and it runs entirely in your browser.