Wi-Fi for warehouses and logistics
A warehouse is the hardest RF environment most companies own, and the only one whose radio characteristics change every day as stock moves through it. Coverage that measures clean in an empty aisle can fail completely once the racking is full.
Why warehouses defeat Wi-Fi designs
The reason has very little to do with the equipment. It is that the building is not the same building twice.
Racking, stock and attenuation that moves
Steel racking reflects. Stock absorbs. A rack bay filled with palletised drinks, tinned goods, cosmetics or anything else with water or metal content is close to opaque at 5 GHz; the same bay empty is largely transparent. Survey on a quiet Sunday with half the racking cleared and you will have measured a building that does not exist on a Tuesday in November. This is why so many warehouse networks are commissioned successfully and then start generating tickets weeks later — nothing broke, the stock arrived.

Stock profile matters as much as fill level. Cardboard, plastics and air are comparatively kind. Liquids, foils, canned goods, dense mixed pallets and anything on metal cages are not, and a design that assumed one and got the other will be short on signal exactly where the picking happens.
Narrow aisle behaves nothing like wide aisle
A narrow aisle acts as a waveguide. Signal is channelled along its length and confined by racking on both sides, which is useful — one correctly chosen and correctly aimed access point can serve a long run — but it also means coverage does not bleed sideways to help the next aisle. Every aisle has to be considered on its own terms, and an access point that looks like it covers three aisles on a plan very often covers one.
Wide aisles, cross-dock areas and pick faces behave more conventionally, and bring the opposite problem: with nothing to contain the signal, access points mounted high and left at default power reach far further than they should, and start contending with each other for airtime instead of serving clients.
Height, downtilt and where the energy goes
Clear heights of ten to fifteen metres are routine, and the standard mistake is to mount omnidirectional access points on the roof steel pointing at everything and nothing. Most of the radiated energy ends up above the racking, where nobody is working, and the useful signal at pick-face height — roughly where a person is holding a scanner — is a fraction of what the plan predicted. Antenna choice and mounting height are not details in a warehouse; they are the design.
5 GHz costs you range
Higher frequencies carry more capacity and survive obstruction less well. In an office that trade-off is comfortable. Through several bays of full racking at 5 GHz, and more so at 6 GHz, it is not, which is why warehouse designs frequently need more access points than the same floor area in an office rather than fewer — the opposite of what a per-square-metre rule of thumb predicts.
The devices are less tolerant than the surveyor's laptop
Handheld scanners, vehicle-mounted terminals and voice-picking headsets have smaller antennas, lower transmit power and far less tolerance of a marginal link than a laptop. They also make their own roaming decisions, on thresholds set by the manufacturer, and many of them are notoriously reluctant to let go of an access point they can still just about hear. A network validated on a laptop can be comfortably and consistently wrong for the devices that actually do the work.
What we usually find
- A design done in an empty building. Either during fit-out or in a quiet period, with no allowance made for the stock profile the operation actually runs.
- One antenna type used everywhere. Omnidirectional access points across the whole shed, including in narrow aisle where a patch or directional antenna would do the job properly and stop the spill into neighbouring aisles.
- Everything at maximum transmit power. Usually applied as a fix for a coverage complaint. It extends the area over which every device contends, raises the noise floor and makes sticky-client behaviour worse rather than better.
- 2.4 GHz left wide open. Three usable channels in a building with dozens of radios, generating co-channel interference that nobody attributes to the Wi-Fi because the symptom is a scanner "being slow".
- Roaming left on defaults. Handhelds clinging to an access point four aisles back until the session drops, then re-associating and losing the transaction.
- No design at all for the edges. Loading bays, dispatch, yard, trailer parking, the mezzanine offices and the transition through shutter doors — the places where sessions break most often, and the ones least likely to appear on the original scope.
- Cabling done to the cheapest route. Access points positioned where a cable happened to reach rather than where the RF needed them.
Most of these are recoverable without replacing hardware. A good proportion of the warehouse networks we are called into have entirely capable equipment that was never designed or commissioned for the building it sits in — see troubleshooting and optimisation for how that gets diagnosed.
How we design a warehouse
Survey it as it operates
We survey with stock in, at representative fill, and we ask what the seasonal extremes look like. Where the operation swings hard between peak and trough, the design is worked to the harder case and the margin is stated rather than assumed. An on-site survey in a working warehouse also captures the interference sources a plan never shows — chargers, wireless bridges, neighbouring units, existing legacy kit nobody has switched off.
Antenna selection and mounting, aisle by aisle
Directional and patch antennas aimed down the aisle where the geometry calls for it; omnidirectional where the space is open. Mounting height chosen so the main lobe lands at pick-face and vehicle-cab height, not on the roof deck. Positions are checked for whether they can actually be fixed, cabled and reached again — which in a high-bay shed is a real constraint, not a formality.
Roaming designed for the worst device on site
We establish what the fleet actually is: scanner models, vehicle terminals, voice headsets, tablets, and anything embedded in the MHE. Signal thresholds, cell overlap along travel routes, minimum data rates and band steering are then set for the least capable of them. Voice picking is the strictest case — it needs continuous, low-latency coverage along every route a picker walks, with enough overlap that the handover completes before the old link degrades.
AGVs, AMRs and automation
Automated vehicles are less forgiving than people. They travel fixed routes at consistent speed, they hand over between access points at predictable points, and an interruption is not an inconvenience but a stopped vehicle in an aisle. Where AGVs or AMRs are in use or planned, we design the overlap along their routes specifically, confirm the handover behaviour their vendor requires, and validate it by walking those routes rather than sampling the floor. If automation is on your roadmap but not yet installed, say so at survey stage — retrofitting for it is considerably more expensive than allowing for it.
Cold stores, chill and freezer
Cold stores combine insulated panel construction that attenuates heavily, condensation and ice, and temperature limits that ordinary access points are not rated for. They usually need enclosures or externally mounted antennas with the radio kept outside the cold envelope, sealed penetrations, and cable specified for the environment. It is a small part of the floor area and a disproportionate part of the design work.
Mezzanines, offices and the transitions
Mezzanine floors change the vertical picture completely — an access point serving the floor below can be an interference source for the deck above. Offices inside the shed have office device density in a warehouse RF environment. And the transitions matter: through shutter doors, from warehouse into yard, from floor to mezzanine, in and out of chill. Sessions break at transitions far more often than in the middle of a well-covered area.
Loading bays, doors and yard
Loading bays are a moving RF environment — a shutter open to the yard and a shutter closed behind a trailer are different buildings. Yard coverage for trailer moves, gatehouse operations and vehicle-mounted terminals needs external-rated hardware, mounting on the building or on columns, and power and containment run to positions that were never intended to carry either. It is designed and priced separately from the indoor scheme because the constraints are genuinely different.
Delivering it without stopping the operation
Warehouses do not stop, and the ones that do stop are expensive to stop. Every warehouse installation we do is planned around that.
- Access at height. Scissor lifts, boom lifts or podium steps as the aisle geometry allows, with the plant, the operator tickets and the working-at-height method statements arranged in advance. Narrow aisle frequently means the racking has to be cleared or the aisle isolated to get plant in, which is a scheduling conversation to have early.
- High-level containment and cabling. Tray and basket at roof level, cable pulled to the designed positions, terminated, tested and certified. This is the part most often left unallocated between trades on a Wi-Fi project, and it is in scope from the start because we do it ourselves.
- Working around shifts. Nights, weekends, shutdown periods, or aisle-by-aisle phasing so picking continues in the rest of the building. We would rather plan the phasing with your operations manager than around them.
- MHE interaction. Pedestrian segregation, banksmen where required, and a plan for how our people and your trucks share aisles safely.
- RAMS and site paperwork as standard, with the accreditations most Wi-Fi companies do not carry — CHAS, Constructionline Gold, SafeContractor and TrustMark, held through The Specialist Electrical Group.
- Validated on completion. We re-survey against the design, including the aisles, the yard and the transitions, and hand you the evidence. See the coverage guarantee.
We have worked in warehouse and distribution environments of this kind — see the Magic Madhouse case study.
Common questions
Can you survey while the warehouse is full and running?
Yes, and we would rather. A survey in an empty building measures a building you do not operate. We work around picking, use banksmen and plant where the aisles require it, and phase the survey so the operation keeps running.
Our heatmap looks fine but the scanners still drop. Why?
Almost always one of three things: the heatmap was produced at laptop sensitivity rather than at the scanner's, the coverage was measured with the racking emptier than it normally is, or the problem is roaming rather than coverage — the device is holding onto a distant access point instead of moving to a nearer one. All three are measurable, and none of them are fixed by buying more access points.
Do we need to replace our access points?
Frequently not. A large share of the warehouse networks we are called into have capable hardware that was never designed or commissioned for the building — wrong antennas, wrong heights, wrong power, default roaming thresholds. Where the hardware genuinely cannot do the job, you will get the measurements that prove it rather than an assertion.
Can you cover the yard and trailer park as well?
Yes. Outdoor coverage is designed separately from the indoor scheme — external-rated hardware, mounting, weatherproofing, lightning protection where the position calls for it, and power and containment run to positions that were never built to carry them. It is quoted separately for the same reason.
We are bringing in AMRs next year. Does that change the design?
Significantly, and it is far cheaper to allow for it now than to retrofit. Automated vehicles need designed overlap along their exact travel routes and handover behaviour that meets the vendor's requirement, which is usually stricter than anything a human-operated fleet needs. Tell us at survey stage and we will design the routes in.
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.