Wi-Fi for a factory or warehouse floor: coverage, not the Mbps on the box

A mill manager told us his Wi-Fi was fine in the office and useless on the floor, and asked which faster router to buy. He had already bought two, each advertising a bigger number than the last.

The number on the box was never the problem. Coverage was.

Why the big-number router fails on a floor

A consumer router is designed for a flat with people sitting still. A factory floor is neither.

Distance. Transmit power is limited by regulation, so an expensive router does not reach meaningfully further than a cheap one. Doubling the advertised Mbps changes nothing about how far the signal goes.

Steel. Racking, machinery, roller shutters, reinforced concrete, and in a textile setting, tightly packed rolls of fabric. Metal reflects, dense material absorbs. A signal that crosses 40 metres of open floor may not cross 15 metres of loaded racking. Floor plans lie: what matters is what is stacked in the aisles, not the dimensions of the building.

Uplink asymmetry. The access point shouts and a handheld scanner replies quietly. You can stand somewhere the device shows full bars, because it hears the access point, while the access point cannot hear the device. Coverage is decided by the weaker of the two.

Roaming. One router means everyone stays on one router. Several routers with their own network names means staff switching manually and devices clinging to a distant unit long after a nearer one appeared. This is the fault people describe as “it drops when I walk to the far end”.

Ceiling or wall

Ceiling-mounted, pointing down, is right for open floor area. The antennas in a typical indoor access point radiate outward and downward, so height puts the signal above racking rather than into the side of it.

Height has a limit, though. On a 10 metre roof, an access point in the trusses is far from the scanner in someone’s hand and has to shine down through whatever sits on the top shelves. Above roughly 6 to 8 metres you are usually better mounting lower, on columns or rack ends at 4 to 5 metres, pointing along the aisles.

Wall and column mounting suits aisles specifically. If your racking runs in long parallel rows, an access point at the end of an aisle beats one in the middle of the ceiling trying to get through six rows of steel. For high-bay storage it is often the only arrangement that works.

Never put an access point inside a metal cabinet, behind a steel beam, or above a metal-backed light fitting. We have seen all three.

2.4 GHz and 5 GHz do different jobs

2.4 GHz travels further and gets through obstacles better, but has only three non-overlapping channels, so it congests and picks up interference from Bluetooth and other equipment. 5 GHz carries far more data with many more channels, and is noticeably worse through solid obstacles.

You want both. Scanners, older devices and anything at the edge of coverage tend to sit on 2.4 GHz, and often that is correct: a barcode scanner needs a reliable few kilobits, not throughput. Laptops, tablets and QC stations belong on 5 GHz.

One avoidable mistake: turning transmit power to maximum everywhere. On 2.4 GHz that mainly makes your access points interfere with each other, and it encourages devices to hold on to distant ones instead of switching. Lower power with more access points beats higher power with fewer.

What Wi-Fi 6 is actually for

Wi-Fi 6 is worth having, for a reason unrelated to the headline speed. Its real gain is handling many devices at once: OFDMA lets one access point serve several small transmissions in the same slot instead of taking strict turns, and target wake time lets battery devices sleep properly between transmissions, which matters if your scanners must last a shift. On a floor with sixty scanners and tablets that efficiency is real.

What it does not do is extend range. The physics is unchanged. Anyone selling Wi-Fi 6 as the fix for a dead spot at the back of the warehouse is selling you the wrong thing; the fix for a dead spot is another access point and the cable to reach it. Old scanners also gain little, since the improvements need Wi-Fi 6 clients at both ends.

Past three or four access points, you need a controller

With one or two you can configure each by hand. Past that they need to cooperate rather than compete, and a controller, whether a hardware box, software or a cloud console depending on brand, is what makes that happen:

• One SSID everywhere, so staff connect once and devices move between access points without anyone choosing a network.
• Assisted roaming, nudging a device off a distant access point onto a nearer one. Standards like 802.11k, v and r speed this up where the client supports them. Without it, clients are famously sticky, and sticky clients are the usual cause of drops while walking.
• Coordinated channels and power, so neighbours stop shouting over each other.
• One place to see faults: which access point a scanner is on, its signal, when it dropped. Without that you are guessing.

Buy access points from one range, one brand, managed by one controller. Mixing brands loses you the roaming behaviour that was the reason to buy several.

Roaming is the real test

Do not stand under an access point and run a speed test. That proves the access point works, which was never in doubt.

Take the device your staff actually use, the scanner or the phone running your app, start something continuous, and walk the route they walk. Down the full aisle, behind the racking, into dispatch, out to the loading bay, past the machines while they run. Note where it pauses on handover and where scans fail. Those are your gaps: add an access point or move one, then walk it again.

Test with the racking loaded and the machinery on. A warehouse surveyed empty is a different radio environment from the same warehouse full.

Power them over the cable

Run Cat6 back to a PoE switch and power every access point over it. Putting a mains socket 5 metres up a column costs more and leaves you a box you cannot reboot from the rack.

Two constraints. The 100 metre Ethernet limit covers the whole run including patch leads, which on a large floor is a real limit; past it, put a small switch out in the building fed by fibre. And check the switch’s total PoE budget, not just its port count. A Wi-Fi 6 access point commonly draws somewhere between 12 and 25 W depending on model and load, and datasheets vary, so add up the actual figures rather than assuming.

Worked example: a 20,000 sq ft warehouse

There is no honest formula turning square feet into an access point count. It depends on racking layout, height, what is stored and where people work. Anyone quoting a fixed number from floor area alone has not thought about it.

For a 20,000 sq ft warehouse with, say, 6 metre racking in parallel aisles, a dispatch area and a small office, a sensible order looks roughly like this:

• 6 to 10 access points, and that range is wide on purpose. Open floor with light storage sits at the low end; dense high racking in long aisles sits at the high end, because each aisle effectively needs its own coverage.
• Aisle units on rack ends or columns at 4 to 5 metres pointing down the aisles, ceiling mounts over the open dispatch and packing areas.
• One PoE+ switch in the rack with headroom on its total budget, plus a small PoE switch out in the building on fibre if any run passes 100 metres.
• A controller, which at that count is not optional.
• One SSID for staff devices and scanners, a separate one for guests or office laptops if you want them kept apart.

Then walk it with a scanner before signing it off, and expect to move one or two access points after the first week. Everybody does.

Send us your floor layout, the rack height, what you store and which scanners your staff use, and we will put the access point and switch list together on WhatsApp.