Signal strength targets are useful, but a wireless network designed to a single number often fails users. Here is what a proper set of design criteria looks like.
Ask most people what a good Wi-Fi design looks like and you will hear a number: -67 dBm. It is a widely quoted target, originally popularised by vendor guidance for voice over Wi-Fi, and it is a sensible starting point. On its own, though, it is not a design.
What -67 dBm actually tells you
Received signal strength (RSSI) describes how much RF energy from an access point reaches a device. At -67 dBm, most modern clients can sustain good data rates, provided nothing else gets in the way. That last part matters. Signal strength says nothing about:
- Noise and interference. A -67 dBm signal in a noisy environment can perform far worse than a -72 dBm signal in a quiet one.
- Contention. If several access points share the same channel within earshot of each other, they take turns to transmit. More signal from more access points can make things slower.
- Capacity. One access point with excellent signal will still struggle with 120 students opening the same cloud application at 09:01.
- Roaming. Mobile devices need a second access point at a usable level before they leave the first. Coverage from one access point alone is not enough for voice or scanners on the move.
A better set of criteria
A requirements-led design typically defines, for each area of the building:
| Criterion | What it addresses |
|---|---|
| Primary signal strength | Can the weakest supported device hear an access point well enough? |
| Secondary signal strength | Is there a roaming candidate before the current connection degrades? |
| Signal-to-noise ratio | Is the signal comfortably above the noise floor? 25 dB or more is a common target for demanding applications. |
| Co-channel overlap | How many access points on the same channel can a device hear? |
| Capacity | How many devices, doing what, at the busiest time? |
The exact values depend on the devices and applications. A warehouse with older scanners, a hospital with voice handsets and a lecture theatre with hundreds of laptops will each have different criteria.
Design for the weakest device
Access points are usually more capable than the devices connecting to them. Phones and handheld scanners have smaller antennas and lower transmit power. If the access point can hear the client but the client can barely hear the access point, or the other way round, performance suffers. We start with the least capable device the network must support and design upwards.
Then measure it
Criteria only matter if they are checked. A predictive design should state its criteria clearly, and a post-installation validation survey should measure the installed network against them. That is the difference between a network that looks good on a plan and one that has been proven to work.