Every warehouse Wi-Fi project starts with the same optimistic conversation. The building is enormous, it is mostly empty air, there are no walls to speak of, and surely radio ought to travel further here than in an office full of plasterboard. Then the racking goes in, the pallets arrive, and the network that surveyed beautifully in April stops holding a scanner session in September.
Nothing has broken. The building simply became a different RF environment, and the survey that produced the design never saw it.
Frequency, wavelength and why 5 GHz suffers more
Two things get worse as you go up in frequency, and both of them matter in a warehouse.
The first is free-space path loss, which rises with frequency. Moving from the middle of the 2.4 GHz band to the middle of the 5 GHz band costs you roughly 7 dB of received signal at the same distance, before anything gets in the way. That is a meaningful figure but it is not the interesting one, because it is predictable and you can design around it.
The second is what happens when something is in the way, and this is where 5 GHz genuinely struggles. Attenuation through most building materials increases with frequency, and the shorter wavelength also diffracts less effectively around obstacles. At 2.4 GHz a wavelength is around 12.5 cm; at 5 GHz it is around 6 cm; in the 6 GHz band it is under 5 cm. The smaller the wavelength relative to an obstruction, the more that obstruction behaves like a wall rather than something the signal bends around. A stack of pallets is a small obstacle to a 12.5 cm wave and a substantial one to a 6 cm wave.
So the honest summary is not "5 GHz has less range". In clear air the difference is modest. In a building full of stuff, the difference is large — and a warehouse is the definitive building full of stuff.
Loaded racking is a different building
Empty pallet racking is mostly air with a steel skeleton. Radio passes through the gaps between beams, reflects off the uprights, and fills the space reasonably well. Coverage plots look generous.
Loaded racking is not a skeleton. It is a wall — and depending on what is in it, a very good one:
- Metal. Steel banding, metal-clad pallets, wire decking, drums, tinned goods, appliances. Metal reflects rather than transmits. A rack bay loaded with metal is effectively opaque, and it also becomes a reflector that scatters signal in directions you did not plan for.
- Water. Water absorbs energy strongly at microwave frequencies, and a great deal of what warehouses hold is mostly water — chilled and ambient food, drinks, liquid chemicals, cosmetics, plants. Anything with a high water content is a strong absorber, and absorption is worse at 5 GHz than at 2.4.
- Density. Even benign cardboard and plastic attenuate; the question is how many metres of it a signal has to cross. A single bay of light goods is not a problem. Six bays deep, with each bay solid, is.
Why surveying an empty warehouse misleads you
This is the single most common way a warehouse Wi-Fi design goes wrong, and it is rarely anybody's bad faith. New buildings are surveyed before fit-out because that is when access is easy and the programme allows it. Extensions are surveyed while the new bay is still bare. Sites take on a new contract and fill racking that was half empty when the last survey was done.
In all three cases the measured data describes a building that will not exist by the time the network is in use. The access point count comes out low, the mounting positions get chosen for a space with clear sight lines, and the channel plan is built around a propagation pattern that the stock is about to destroy.
There are only a few honest ways around this:
- Survey it loaded. Where the building is already operating, survey it at representative stock levels and, ideally, more than once — peak and trough are different networks.
- Model the racking properly. In a predictive survey the racking has to be drawn as attenuating structure, not as open floor, with attenuation values chosen for what will actually be stored. A predictive model of an empty shed is a model of a shed.
- Design for the worst case and validate against it. Design to the loaded condition, then run a validation survey once stock is in, and correct what the model got wrong.
- Say what the assumption was. If a survey was done empty, that belongs in the report as a stated limitation, not buried.
Narrow aisles and the waveguide effect
A narrow aisle bounded on both sides by solid, largely metallic racking behaves rather like a waveguide. Signal that enters the aisle is repeatedly reflected off the rack faces and channelled along it, so it can propagate a surprising distance down the aisle while barely crossing into the next one.
That has two consequences, one helpful and one not.
The helpful one is that you can cover a long aisle from relatively few positions if you launch the signal into the aisle rather than across it. A directional antenna at an aisle end, or a downward-and-along pattern from a position over the aisle, does far more useful work than an omnidirectional radio sitting above the racking hoping to reach everything.
The unhelpful one is multipath. All that reflection means a device receives many copies of the same signal at slightly different delays. Modern OFDM radios handle multipath well and MIMO systems can exploit it, but in a highly reflective narrow aisle the delay spread and the constructive-and-destructive interference pattern can still produce deep, localised fades — spots a metre wide where the signal drops out and a step to either side fixes it. These are not visible in a design tool and they are one of the more compelling arguments for measuring in the building.
Very narrow aisle operations — VNA trucks, guided aisles, mobile shelving that closes up — are harder again, because the aisle geometry itself changes.
Mounting height, and the trade-off nobody mentions
High-bay warehouses invite high-bay mounting. It is convenient: the access points are out of the way of forklifts, containment can follow the roof steel, and one position appears to see everything.
The problem is that a scanner is at roughly 1.5 m in an aisle, and an access point at 12 m above the floor is a long way from it — and pointed the wrong way.
- Antenna pattern. A standard omnidirectional access point antenna radiates in a broad horizontal doughnut with a null directly beneath it. Mount it high and the strongest part of the pattern goes out sideways, over the top of the racking, to places nobody needs coverage — while the floor directly below sits in the weakest part of the pattern.
- Distance. The link budget has to cover the vertical drop as well as the horizontal run. Twelve metres of height is twelve metres of range you are not using horizontally.
- Co-channel reach. A high access point hears far more of the building than a low one, and is heard by far more of it. That enlarges the contention domain: more radios and more devices deferring to each other on the same channel, sharing the same finite airtime.
- Uplink asymmetry. The access point may be able to shout across the building. The handheld scanner, with a small antenna and a fraction of the transmit power, cannot shout back. Designing to what the access point can reach rather than to what the client can return is a reliable way to build a network with full signal bars and no throughput.
The usual answer in a racked warehouse is to mount lower than the roof and to use directional antennas — patch or sector antennas aimed down the aisles, or downward-facing units on drop poles below the racking top. Where high mounting is unavoidable, the antenna choice has to change with it. Access, safety and containment then become part of the design decision rather than an afterthought, which is one reason we do the design and the installation as one piece of work.
Why more access points at full power makes it worse
The instinctive fix for poor warehouse coverage is more radios, turned up. It is nearly always the wrong move, and it frequently makes a bad network worse.
Wi-Fi is a shared, contended medium. Devices and access points on the same channel take turns: before transmitting, a radio listens, and if it hears another transmission it waits. Every additional radio you can hear on your channel is another radio you have to wait for. Turning the power up widens the area over which everything hears everything else, which increases the number of participants sharing each slice of airtime.
So a warehouse with too many high-powered access points on overlapping channels can show excellent signal strength everywhere and still deliver poor throughput, long scan times and dropped sessions — because the airtime is spent deferring rather than transmitting. It also makes roaming worse: a device that can hear a strong signal from a distant access point has no reason to let go of it.
None of this can be decided from a floor area. It comes out of measurement — what the racking is actually doing to the signal, what the stock does to it at peak, and what the device fleet needs at the sharp end. That is what an on-site survey is for, and it is why our warehouse and logistics work starts there rather than with a product.
Let's find out what your Wi-Fi is actually doing.
Book a survey and you get measured data, a design you can build to, and a number you can budget against — not an opinion.