Lab notebook · model notes
WSRM does not simulate radio waves. It adds up four things: how strong the access point is, how far the signal must travel, how many walls are in the way, and how cluttered the space is. Then it paints the sum.
That is a deliberately small model. It is easy to inspect, quick enough to redraw while you drag an access point, and wrong in ways you can reason about. This page lists each ingredient, where it comes from and where it will mislead you. It is an approximation, not a survey.
For every point on the plan, WSRM computes the signal from each access point and keeps the strongest:
Each term has its own entry below. The tool lets you change Ptx, the band (which sets PL at 1 m and scales wall losses), n (the clutter setting) and every wall’s loss.
dBm is power measured against one milliwatt on a logarithmic scale. 0 dBm is 1 mW. Every 10 dB is a factor of ten, and every 3 dB is roughly a factor of two. Because the scale is logarithmic, losses simply add: a wall that costs 3 dB takes the same bite out of a strong signal as out of a weak one.
The numbers you will see on the map are negative because the power arriving at a device is a tiny fraction of a milliwatt. −50 dBm is a hundred-thousandth of a milliwatt; −70 dBm is a ten-millionth. Closer to zero is stronger, so −55 beats −75.
The model starts from the free-space path loss at a distance of one metre, which depends only on frequency:
WSRM evaluates it at one centre frequency per band. Higher frequencies lose more in the first metre, which is why the same transmit power maps to a smaller area on 5 and 6 GHz even before walls are counted.
| Band | Centre used | PL at 1 m | Wall-loss scale |
|---|---|---|---|
| 2.4 GHz | 2437 MHz | 40.2 dB | × 1.00 |
| 5 GHz | 5500 MHz | 47.3 dB | × 1.25 |
| 6 GHz | 6500 MHz | 48.7 dB | × 1.40 |
The centre frequencies are single representative values, not statements about any particular channel. The wall-loss scale factors are illustrative round numbers that capture the general trend that walls cost more at higher frequencies.
Beyond one metre, the loss grows with the logarithm of distance, scaled by an exponent n:
In open space with nothing in the way, n is 2. Indoor measurements published over the years usually land somewhere between about 2 and 4, depending on the building and how it is furnished. Doubling the distance costs 3.01·n dB, so with n = 2.6 each doubling costs about 7.8 dB.
Because WSRM subtracts walls separately, the exponent only has to stand in for everything else: furniture, people, appliances and the general untidiness of rooms. The three presets (2.0, 2.6 and 3.2) are round numbers chosen so that the three clutter settings give clearly different maps. They are not fitted to any particular building.
WSRM draws a straight line from the access point to the point being evaluated. Every wall that line crosses subtracts that wall’s loss once. The angle of the crossing and the thickness of the wall are not considered; a diagonal wall costs the same as a square-on one.
| Material | 2.4 GHz | 5 GHz | 6 GHz |
|---|---|---|---|
| Glass / window | 2.0 dB | 2.5 dB | 2.8 dB |
| Drywall (stud) | 3.0 dB | 3.8 dB | 4.2 dB |
| Wood / door | 4.0 dB | 5.0 dB | 5.6 dB |
| Brick | 7.0 dB | 8.8 dB | 9.8 dB |
| Concrete | 12.0 dB | 15.0 dB | 16.8 dB |
| Metal / foil-backed | 20.0 dB | 25.0 dB | 28.0 dB |
Illustrative example values. These are round, plausible numbers meant to rank materials sensibly, not measurements of your walls. Real losses vary with thickness, moisture, wiring, insulation foil, rebar and whether a door is open. Edit any value in the mapper and the whole map follows.
Two details of the counting are worth knowing. Walls that share an end point count once where they meet, so a corner is not double-charged. Two walls drawn on top of each other each count in full, so draw one long wall rather than two overlapping ones.
Entry 06 · bench
A one-dimensional version of the same equation, so you can see its shape without a floor plan. Set a band, a power, a clutter exponent and a number of identical walls between you and the access point. The starting values are illustrative; change any of them.
| Distance | Signal | Quality band |
|---|
The solid, dashed and dotted horizontal lines mark −67, −75 and −85 dBm. The range figures solve the equation for the distance at which the curve meets each level, and say so when that falls under one metre or beyond a thousand.
With two or three access points WSRM evaluates each one separately, walls included, and keeps the strongest value at every point. That is how a device behaves when it is free to join whichever access point it likes best. The model does not add the signals together, does not model channel overlap or contention, and does not know that a device might cling to a distant access point instead of roaming.
The readout panel names the serving access point so you can see where the handover happens.
The ramp runs from indigo at weak signal through magenta and orange to yellow at strong signal. Between the anchor points the colour is interpolated, and anything below −85 dBm fades out to nothing by −95 dBm. Colour never carries the reading alone: the readout gives the number and a named band, and the contour lines mark two edges.
| Band name | Range | Rough meaning |
|---|---|---|
| Excellent | ≥ −50 dBm | Very close to the access point. |
| Strong | −50 to −60 | Comfortable for almost anything. |
| Good | −60 to −67 | A common design target for rooms that matter. |
| Fair | −67 to −75 | Fine for light use; sensitive to clutter. |
| Weak | −75 to −85 | Usable at low speed, easily upset. |
| Fringe | −85 to −95 | Barely there. Expect drops. |
| No signal | < −95 | Treated as nothing. |
The meanings are rules of thumb that tinkerers commonly use, not guarantees. What a given device can do at a given level depends on the device.
The coverage figures in the mapper count the share of sample points, spaced every 0.25 m across the plan, that reach each threshold. They include hallways, cupboards and anywhere else inside the rectangle.
These are the places the map will disagree with a real walk-around. None of them is small.
If you want the map to be a better guide to your own building, treat it like an instrument and calibrate it.
Even a calibrated map is a sketch, but it will be a sketch that agrees with the places you measured.
An illustrative case: 2.4 GHz band, 20 dBm transmit power, furnished rooms (n = 2.6), a point 10 m away with one drywall partition between it and the access point.
Put four more walls in the way and the same arithmetic gives about −61 dBm, the difference being the extra 12 dB of walls. Swap to 5 GHz and the first-metre term alone rises by about 7 dB. You can reproduce both in the bench plot above.