Blog · · Technical
Can a WiFi heatmap work in a browser?
The short answer: a measured heatmap cannot, and a predicted one can. A web page has no access to your WiFi radio, so it cannot sample real signal strength — that needs an installed app. A predictive heatmap never touches the radio. It is computed from the floor plan, the wall materials and the access point's datasheet, which is arithmetic, and arithmetic runs in a browser. That is how WiFi Heatmap (wifiheatmap.app) works, and it is how Ekahau, Hamina and iBwave plan a network before anything is installed. The price of predicting is accuracy: roughly ±6–10 dB, which is enough to decide where access points go and not enough to replace a survey.
"Browsers can't do radio scans" — true, and beside the point
The usual version goes: free web-based heatmap tools are limited, because an accurate heatmap needs on-device radio scans and a browser cannot do them. That is correct as far as it goes. No web standard lets a page list nearby networks or read their signal strength. The Network Information API reports a rough connection type and bandwidth estimate, not RSSI, and neither Web Bluetooth nor WebUSB offers a WiFi scan. Scanning is an operating-system privilege — CoreWLAN on macOS, the Native WiFi API on Windows, WifiManager on Android, where even native apps are throttled to a handful of scans every couple of minutes. iOS gives ordinary apps no API for scanning nearby networks at all.
So every tool that measures — NetSpot, Ekahau Survey, Acrylic, WiFiman — is a native app, and has to be. If you want to know what signal is actually in a room today, install one and walk the building.
But "WiFi heatmap" names two different products, and the objection only applies to one of them.
Two kinds of heatmap
| Property | Measured (survey) | Predicted (planning) |
|---|---|---|
| Input | Signal samples taken as you walk | Floor plan, wall materials, access point specs |
| Network must already exist | Yes | No — that is the point |
| Needs the WiFi radio | Yes | No |
| Can run in a browser | No — native app | Yes |
| Accuracy | What is actually there, at the moment you measured | A model: roughly ±6–10 dB in a home or small office |
| Answers | "Why is the signal bad in the kitchen?" | "How many access points do I buy, and where do they go?" |
They are sequential, not rivals: predict, install, then measure what got built. The professional tools do both — Ekahau, Hamina and iBwave each have a predictive design mode, where you draw the building and place virtual access points, alongside a survey workflow that needs hardware. The predictive half is the half that does not need a radio.
What a predictive heatmap actually computes
For every point on the floor plan, the received signal from each access point is a link budget:
received (dBm) = Tx power + antenna gain − free-space loss(distance, frequency) − Σ wall losses
That is the COST-231 multi-wall model in its simplest form, which is what our solver runs:
draw a line from the access point to the point, count every wall it crosses, subtract each
wall's loss at that band. (The full COST-231 formula adds a fitted constant and a
diminishing term for floors; we add floor slabs linearly too.) Free-space loss is
Friis' equation, 20·log10(d) + 20·log10(f) − 27.55
with distance in metres and frequency in MHz. None of it needs anything a browser lacks.
Worked example, at 5 GHz. An access point at 20 dBm with a 4 dBi antenna radiates 24 dBm. At 8 m, free-space loss at 5500 MHz is 65.3 dB. The path crosses one brick wall (11 dB) and one drywall partition (4 dB). Received signal: 24 − 65.3 − 11 − 4 = −56.3 dBm — comfortably above the −67 dBm usually quoted for calls and video. Make both walls concrete (20 dB each) and the same access point at the same distance delivers −81.3 dBm, well below that line, where calls and video become unreliable. The wall material, not the distance, decides it.
The wall figures are the substance of the model. These are the values WiFi Heatmap's solver uses, per wall crossed and separately per band, because attenuation rises with frequency. They follow ITU-R P.2040 material properties and the loss ranges in ITU-R P.1238 for typical wall thicknesses:
| Material | 2.4 GHz | 5 GHz | 6 GHz |
|---|---|---|---|
| Drywall | 3 dB | 4 dB | 5 dB |
| Wood | 4 dB | 6 dB | 7 dB |
| Glass | 2 dB | 3 dB | 4 dB |
| Coated / low-E glass | 20 dB | 25 dB | 28 dB |
| Brick | 8 dB | 11 dB | 13 dB |
| Concrete | 15 dB | 20 dB | 23 dB |
| Metal | 25 dB | 30 dB | 32 dB |
Coated glass and metal are typical figures rather than values read from either recommendation, and metal's is a floor on the loss, not an estimate: in practice metal reflects rather than attenuates.
Run that on a fine grid across the plan — a sample every few screen pixels, smoothed into a colour map — for every access point, keep the strongest, and you have a heatmap. For a house the whole calculation takes milliseconds, which is why it can redraw on every frame while you drag an access point. The maths in more depth →
What prediction honestly costs
Running in a browser costs nothing in accuracy — the maths is identical wherever it runs. Predicting costs a good deal, and it is worth being plain about it:
- •An expected ±6–10 dB against a real measurement in a typical home or small office. Enough to decide "two access points or three" and roughly where. Not enough to promise a number in a particular corner.
- •Furniture, people and multipath are not in the model. Nor are your neighbours' networks, which affect throughput far more than they affect signal strength.
- •Antennas are modelled by gain, not by their full 3D pattern, so the weak spot directly beneath a ceiling-mounted access point is not drawn.
- •A wall is only as right as the material you picked. Real brick varies with thickness and moisture; the table above is a typical wall, not your wall.
So: plan with a prediction, then validate the install with a measurement — any survey app on a laptop or Android phone will do. The full list of what the model does and does not do is on the accuracy page.
What the browser adds
If the maths is the same, why does it matter where it runs? Three reasons, and they are the reasons we built WiFi Heatmap (wifiheatmap.app) this way:
- •Zero setup. No download, no install, no account, no card. It works on Linux and ChromeOS as well as macOS and Windows. You are drawing walls within a minute of opening the page.
- •Your floor plan never leaves your machine. Our solver runs client-side, so the plan, the walls and the exports stay in the browser; our server and analytics see usage counts, never the plan. Cloud planners such as Cambium's upload the plan to the vendor's server. Because our maths runs on your machine, we never need the plan at all.
- •Any vendor. 41 access point presets across 11 vendors, each with antenna gain from the manufacturer's datasheet and transmit power from the vendor's published figures (estimated for the few models whose vendor publishes none), or type in any radio. A plan made before you have picked hardware should not be tied to one brand.
We are not the only browser-based predictive planner, and it would be odd to pretend otherwise:
- →UniFi Design Center — free, locked to Ubiquiti hardware.
- →Cambium Wi-Fi Designer — free, Cambium hardware only; a guest demo needs no account, saving and exporting does. Compared in full →
- →Hamina — the most capable of the group, with 3D propagation; the free tier caps at 3 APs and needs an account. Compared in full →
- →Deconflict — free and open source; reports relative signal bands, not dBm. Compared in full →
- →WiFi Heatmap (wifiheatmap.app) — free, any vendor, no account at any tier, floor plan never uploaded. Unlimited access points in the editor; the free export includes five on one floor, watermarked.
The full field, including the tools that beat us for particular jobs, is in free WiFi planning tools compared and free Ekahau alternatives.
Questions
Can a web browser measure WiFi signal strength?
No. No web standard lets a page scan nearby networks or read their signal strength (RSSI). That data is only available to native software through the operating system, which is why every measured WiFi survey tool — NetSpot, Ekahau Survey, Acrylic, WiFiman — is an installed app. Any heatmap that claims to show measured signal from a plain web page is not measuring.
Can a browser-based WiFi heatmap be accurate?
A predictive one can be accurate enough to plan with. It computes coverage from the floor plan, the wall materials and each access point's transmit power and antenna gain, using a path-loss model — the same kind of calculation the professional planners run in their predictive mode. WiFi Heatmap (wifiheatmap.app) states its envelope as roughly ±6–10 dB against a real measurement in a typical home or small office: enough to decide how many access points and where, not a substitute for a survey after installation.
Is there a free web-based WiFi heatmap tool, with no download?
Yes. WiFi Heatmap (wifiheatmap.app) is a free, web-based predictive WiFi planner — an online tool that runs entirely in the browser, with no download and no account. You upload a floor plan, trace walls by material, and place access points from 41 presets across 11 vendors, or type in any radio's figures; the heatmap updates live and the floor plan never leaves your machine. Placement is unlimited in the free editor; exports include five access points on one floor and carry a watermark. Other browser planners exist too: UniFi Design Center and Cambium Wi-Fi Designer (each locked to its own vendor's hardware), Hamina (free tier capped at 3 access points, account required) and Deconflict (open source, relative signal bands rather than dBm).
What is the difference between a predictive and a measured WiFi heatmap?
A measured heatmap records the signal that is actually there: you walk the building with a laptop or phone and a survey app samples it. It needs the network to be installed and it needs native access to the WiFi radio. A predictive heatmap computes the signal that should be there, from physics, before anything is installed. It needs only the floor plan, the materials and the access point specifications — which is why it can run in a browser. Plan with prediction, then verify with measurement.
Do I still need a site survey if I use a predictive planner?
For anything that matters, yes — after installation. A prediction is typically within ±6–10 dB of reality in a home or small office, and it does not see furniture, multipath or the neighbours’ networks. The point of predicting first is that the survey then checks a sensible design rather than discovering a bad one after the cables are run.
See it run in your browser — free, no signup
Upload a floor plan, trace the walls by material, drop an access point, and drag it around. The heatmap recomputes as you move it, on your machine, and nothing is uploaded.