Most wireless network complaints come not from the device but from a lack of planning: access points hung in the wrong place, channels that interfere with each other, placement made without considering capacity, a handheld terminal dropping the connection when moving from one aisle to another. When these problems are tackled after installation, the ceiling is opened again and cable is pulled again. A site survey is the way to see them before installation.
1. Clarify the requirements
Before starting measurements, write down what the network will carry. For the same area, “guest internet” and “warehouse management with handheld terminals” require very different designs.
- Number of users and devices per area, peak hours
- Applications: voice/video calls, handheld terminals, IoT, guests
- Areas that need roaming
- Building structure: concrete, glass, metal racks, suspended ceilings, height
- Authentication model: WPA3-Enterprise, guest portal, device certificates
Design for the weakest client. At a point where a laptop connects easily, an old barcode scanner, an IoT sensor that only supports 2.4 GHz or a phone with a small antenna may struggle. A mistake you may encounter is measuring with the engineer’s laptop and approving the design based on it. Collect the model list of critical devices, the bands and standards they support (Wi-Fi 5, 6, 6E) and their roaming features at the very start. This list determines the channel plan, the security model and the minimum data rate setting.
Set the target values in writing too. For areas that will carry voice calls, many vendors’ design guides commonly use a target of about -67 dBm minimum signal level and around 25 dB signal-to-noise ratio; in areas carrying only data, the target can be more flexible. Which value applies to which area determines what counts as “passed” in the validation measurement.
2. Predictive design
A preliminary design is made in software on a scaled floor plan, with wall materials defined. This stage gives a good starting point for the number of APs and the budget; but since material assumptions can differ from the real site, it isn’t enough on its own.
The accuracy of a predictive model is limited by the scale of the floor plan and the wall definitions. An unscaled PDF, or a plan drawn years ago and changed since with partitions, makes the map the software produces meaningless. Scale the plan by measuring the distance between two known points on site. Define elements such as concrete shear walls, drywall, glass partitions, elevator shafts and fire doors separately. A model drawn with a single “interior wall” assumption shows either too many or too few APs on site.
3. On-site measurement
In critical areas, a temporary AP is set up (“AP on a stick”) to measure real signal loss. The existing wireless networks and channel use on site are also scanned; neighboring networks, old devices and interference sources at 2.4 GHz are noted. This measurement is especially important in warehouse, production and metal-heavy areas. If handheld terminals or mobile production equipment are used, plan the site test for industrial Wi-Fi and mobile terminal networks with the real devices and routes.
Install the temporary AP at the real mounting height and, if possible, with the planned model. An AP placed on a desk and an AP hung from the ceiling have different coverage shapes. In warehouses, do the measurement when the racks are full: attenuation changes noticeably between an empty metal rack and a pallet-loaded rack, and stock containing liquids and paper absorbs the signal further. A design approved in an empty warehouse comes back as dropouts at the ends of aisles when the stock arrives.
A passive scan shows neighboring networks and channel density; spectrum analysis reveals non-Wi-Fi interference sources (microwave ovens, wireless camera transmitters, old cordless phones, Bluetooth density). These sources are invisible in a scan done only with a Wi-Fi adapter. But they occupy the channel.
4. Capacity and channel plan
- Design with 5 GHz and 6 GHz first; use 2.4 GHz only to a limited extent for old devices.
- In dense areas, a 20 MHz or 40 MHz channel width gives healthier results than wide channels.
- Neighboring APs should be on non-overlapping channels, and transmit power should be limited to the coverage need.
- In areas that need roaming, 802.11k/v/r support and cell overlap should be planned.
In the 2.4 GHz band, 1, 6 and 11 are a common channel plan for a 20 MHz channel width; the choice is made according to national rules and interference on site. Keeping every AP’s radio on in this band produces dozens of cells sharing the same channel in a dense building. Turning off the 2.4 GHz radio on some APs often improves performance. If you will use DFS channels in the 5 GHz band, take into account that the AP will change channel when radar is detected and that clients may experience a brief interruption; at sites near airport or weather radar, this behavior shows up often in the logs.
The 6 GHz band (Wi-Fi 6E and later) works only with WPA3 or OWE; you can’t carry a WPA2 network name into this band. Plan the security model accordingly. The classic mistake with transmit power is running the AP at full power to increase coverage. The client hears the AP, but the AP can’t hear the client; the phone’s transmit power is lower than the AP’s. The result is a connection with full bars that doesn’t work. Keep AP power close to the client’s power. Disabling low data rates (e.g. 1, 2, 5.5 and 11 Mbps) prevents a distant client from sticking to an old AP and reduces beacon overhead.
5. AP placement and infrastructure
APs are positioned close to where users are, at ceiling level and away from metal obstacles. For each AP, the cable distance (100 m limit for copper), switch port and PoE budget are checked. Some Wi-Fi 6/6E APs need PoE+ (802.3at) or higher power for full performance; the switch should be chosen accordingly.
An AP hidden above a suspended ceiling, inside a metal trunking or behind a column looks tidy. Its coverage, however, is cut in half. In high-ceiling warehouse and production areas, it may be necessary to mount the AP not on the ceiling but lower, near the rack aisles, with a directional antenna. Some AP models running on insufficient PoE switch to a low-power mode by turning off the second radio or the USB port; it doesn’t look like a fault, but the designed capacity isn’t there on site. Check the PoE class on AP ports from the switch. Don’t forget uplink speed either: connecting an AP that supports 2.5 Gbps to a 1 Gbps port moves the bottleneck to the wired side.
6. Post-installation validation
After installation, a validation measurement is done in the same areas: signal level, signal-to-noise ratio, channel overlap and a roaming test. The results are reported; weak areas are fixed by AP position or power settings. This report also becomes the reference for future expansions.
Don’t stop validation at a passive map. Do active tests: open a voice call on a critical client and walk through roaming areas, run the real application on a handheld terminal while changing aisles, measure throughput in a meeting room during busy hours. Authentication times and roaming transitions should be reviewed in the wireless controller’s client logs; on networks using 802.1X, if the RADIUS response time is long, drops occur during roaming, and that isn’t an RF problem. An area that hasn’t been measured isn’t handed over.
Checklist
- Requirements and user density documented
- Predictive design + on-site measurement in critical areas done
- Channel and power plan prepared
- PoE budget and cable distances verified
- Security: WPA3-Enterprise or at least WPA2-Enterprise, guest isolation
- Post-installation validation report received
Frequently asked questions
Is a coverage map alone enough?
No. The signal reaching everywhere doesn’t mean many devices at once will get good performance. Capacity (user and application density) must be planned along with coverage.
Are more APs always better?
No. More APs than needed increase the number of devices sharing the same channel (co-channel interference) and can lower performance. The number of APs and transmit power should be planned together.
Is measurement needed after installation?
Yes. A post-installation validation measurement shows whether the design delivers the expected result on site, and allows problem areas to be fixed before going live.