Produce a coverage map from measurements you took

applied · 60 min · Objective 2.3

Task

Walk a real space taking signal measurements on a grid, then turn the readings into a coverage map that identifies the gaps. A site survey is the answer to several exam scenarios, and doing a crude one teaches more than reading about an expensive tool.

Steps

  1. Sketch a floor plan and overlay a grid, roughly 3 metres per square. Mark where the access point is and mark walls, especially anything with metal, concrete or plumbing in it.
  2. At the centre of each square, stand still for ten seconds and record the RSSI in dBm and the negotiated link rate. Standing still matters — the reading swings by several dB as you move.
  3. Record at least twelve points, including one behind the worst wall, one at the furthest corner, and one right next to the AP for your reference maximum.
  4. Colour the grid: better than −65 dBm good, −65 to −70 adequate, −70 to −80 marginal, below −80 unusable. You now have a heat map.
  5. Identify the worst square and diagnose it. Is the cause distance, an obstruction, or interference? Move the AP a metre and re-measure three points to test your theory. A survey that does not change a decision was not worth walking.

Verify

for i in 1 2 3; do sudo iw dev wlan0 link | grep -E "signal|bitrate"; sleep 2; done
python3 -c "
readings = [('by the AP',-38),('hallway',-52),('kitchen',-61),('back bedroom',-74),('garage',-86)]
for place, dbm in readings:
    band = ('good' if dbm > -65 else 'adequate' if dbm > -70 else 'marginal' if dbm > -80 else 'unusable')
    print(f'{place:<16}{dbm:>5} dBm  {band}')
print()
print('every 3 dB lost is half the power; 10 dB is one tenth')
"

Your own twelve readings, classified into those four bands, are the deliverable. The sample above shows the shape a real set takes — a steep fall through the first wall and a long slow decline after it.

Notes

The decibel scale is logarithmic and that is the part worth internalising: 3 dB is half the power, 10 dB is a tenth. So a wall costing 6 dB has taken three quarters of your signal, which is why one badly placed wall matters more than ten metres of open air.

Two findings this exercise reliably produces. Signal is not symmetrical around an AP, because obstructions are not. And the AP's own placement is usually the cheapest fix available — moving it a metre away from a metal filing cabinet often does more than adding a second AP.

Note also that measuring downlink RSSI at the client is only half the story. A client may hear the AP well and still be too weak to be heard back, because the AP transmits with far more power than a phone does. That asymmetry is why "full bars and nothing works" happens.