Wi-Fi in the hall and between racks
A survey with the racks full, access points positioned along the aisles, antennas chosen and power set so a truck moves between points without dropping.
A conversation about warehouse IT almost always opens with "the scanners lose signal between the racks". It almost always ends with a proposal to add access points or turn up the power, which in a racked hall usually makes matters worse. Below is what is actually happening, how to measure it and what to do about terminals running a system that has had no patches for years.
Where the boundary with neighbouring pages runs: this page is about handling goods on site, meaning terminals, coverage between racks and the warehouse system. Selling, tills and stock across several channels are covered under trading companies, while shop-floor machines and splitting the network between office and production are under manufacturing. The Wi-Fi basics common to office and hall are under the company network and we do not repeat them here. A materials yard with a site office for the length of a contract is different again, because all of it has to be dismantled afterwards, and we cover that under construction.
The symptom is always the same: the truck turns into an aisle, the session freezes for a dozen seconds, the picker scans again. The cause almost never lies where we look for it.
The network can suggest to the scanner that there is a better access point nearby, but it cannot force it to move. Older terminals follow simple logic: they hold on to the current access point until the signal disappears altogether, and switch only once the connection is already unstable. Hence those dozen seconds in the middle of an aisle.
This is the crux, and the thing a coverage map does not show. An access point transmits with far more power than a handheld terminal, so the scanner sees a strong signal from the far end of the hall and concludes all is well. It does not work the other way: the access point cannot hear the terminal weak transmitter, packets are lost and the session breaks, even though the screen shows full bars.
Turning up the power enlarges the cells, so the scanner sees yet more distant access points and clings to them yet harder. On top of that, access points on the same channel hear one another above the racks and have to wait for each other to transmit. The answer is to shrink the reach of a single access point and aim it along the aisle, not to add more antennas.
The standards that help here are 802.11k, a list of neighbouring access points handed to the device, and 802.11v, a suggestion to move. There is also 802.11r, which speeds up the switch itself, but on older scanners it can stop them joining the network at all, so we enable it first on a separate network name and test it on a few units before touching the rest. None of these standards will work if the terminals do not support them, and we say so before a purchase rather than after.
Surveys usually get done when the hall is quiet, which means with the racks half empty. That is the worst possible moment and the commonest reason why a network signed off as sound starts losing scanners in November.
The metal frame of a rack reflects the waves and turns an aisle into a kind of corridor in which the signal runs along its length but does not pass sideways into the next row. Filling the racks changes it again, because drinks, chemicals and densely packed cartons absorb the signal far more than air does. A map made with the warehouse empty shows green where there will be a dead spot at peak.
A laptop held at head height will show results a terminal will never reach, because the terminal has a weaker antenna and less power. So we walk the pickers routes and drive the aisles the way a truck drives them, and we look not only at signal strength but at how often packets have to be repeated, because that is what shows when a connection is about to break.
A new row of racking or a change in how goods are slotted shifts those corridors and can create a dead spot where things were fine for years. It is the commonest explanation for reports along the lines of "but we changed nothing in the network". You changed nothing in the network; you changed something in the hall.
That is why we arrange the survey for a time when the warehouse is as full as possible, rather than whenever it happens to be quiet. The survey before a quote is free and carries no commitment, and what comes out of it is a drawing with positions for access points and a list of what has to be reached with cable. We have more than once followed such a survey by advising against an expansion and limiting ourselves to repositioning and retuning what was already mounted.
Warehouses today routinely run on handheld computers bought ten years ago or more. The hardware is solid and hard to wear out, but the system inside has stopped receiving patches, and that is what sets the replacement date.
Windows Embedded Handheld 6.5, the system on most older handhelds, went out of support on 14 January 2020. Older Windows CE went earlier still. It is not only about missing security patches: these devices often cannot do newer wireless encryption, so a single old terminal can block raising Wi-Fi security across the whole warehouse.
The commonest worry is that new devices will force a change of warehouse system. Usually they do not. If pickers work on text screens served from the server, the same screen can be displayed on Android without rewriting the software. We check this on one device before ordering a whole batch, because the question settles in half a day rather than in a discussion.
Terminals and their batteries are consumables, not fixed assets for a decade. A broken handheld takes one person out for a whole shift, so we keep a few units ready to hand over, configured and labelled. When buying we choose a warranty with repair at your site, so a fault does not wait for us to drive over.
What we do not do, and we say it up front: we do not write your warehouse system and we do not stand in for its vendor. We look after what that system runs on, meaning the server, the database, the network and the devices, and we deal with the software vendor when it has to be established whose side a problem sits on. On choosing the devices themselves we help as part of hardware supply, at purchase prices.
Six areas where work with a warehouse tends to start. Each has its own page with the detail.
A survey with the racks full, access points positioned along the aisles, antennas chosen and power set so a truck moves between points without dropping.
Access points under a single management console, so changing a channel or a power level does not mean climbing a ladder to each one in turn.
Devices chosen to suit your warehouse software, configured in batches, with spares for replacement and a warranty covering on-site repair.
The machine and database the WMS runs on: sized to the number of transactions, updated in service windows and watched for load.
Copies made so that stock levels can actually be restored from them, and verified by a test restore rather than by a green status in a console.
Round-the-clock watch over the server and network, plus an on-call phone line if the warehouse works evenings or nights.
Honestly about availability, because the old version of this page promised otherwise: we do not run a team working shifts and we will not pretend we do. Monitoring runs around the clock and is included in the retainer, on-call cover comes with the higher packages, and a warehouse that genuinely needs a technician on site at three in the morning will be told so plainly, and helped to find a supplier with round-the-clock staffing.
The order follows from the fact that in a warehouse the costliest thing is a break in picking, so we start with measurement rather than purchases.
We walk the pickers routes with the racks as full as possible and note where sessions break. While there we check what the terminals run on and how old that system is. This is the free survey.
We write down what the survey shows: where access points are missing, where repositioning and retuning the existing ones is enough, and where the devices themselves are the problem. We separate what has to be done now from what can wait.
We reposition and retune access points in windows agreed with the shift, and test the new configuration on a few terminals before it covers all of them. The old settings stay available to fall back on for the first few days.
Day-to-day requests, monitoring of the server and network, and a repeat survey after racking is moved or before the season. We put that last point in the contract, because otherwise nobody remembers it until the first failure.
The pages people reading this one move on to most often.
Pick your industry and see how comprehensive IT support and IT outsourcing for SMEs can help your company.
Usually not, and it often makes things worse, because the scanner starts seeing even more distant access points and clings to them even harder. We measure first, and only then say whether the problem is the number of access points, their placement, the power they are set to, or the terminals themselves. Sometimes repositioning and retuning what is already mounted is enough.
Not overnight, but a date has to be planned, because that system has had no patches since January 2020. It becomes urgent when the old devices block raising wireless security, because then one handheld holds the whole warehouse on weaker encryption. We normally spread the replacement over batches rather than doing it in a single weekend.
Usually not. If pickers work on text screens served from the server, those same screens can be shown on Android devices without rewriting the software. We check it on a single unit before ordering a whole batch, so the decision does not rest on assurances.
It depends on the software, and it is the first question we put to its vendor: can the terminal work without a connection and send the transactions later. If it can, we agree with the warehouse manager how long working that way is acceptable. If it cannot, then the line and the server are plainly the point at which picking stops, and we choose the backups and standby hardware accordingly.
Monitoring of the server and network runs around the clock and is included in the retainer, and on-call cover by phone comes with the Business and Enterprise packages, with a response within thirty minutes. What we do not run is a team on shifts, and we do not promise a technician in the hall at three in the morning. If that is exactly what you need, we will say so at the first conversation.
This one is about handling goods on site: terminals, coverage between racks and the warehouse software. That one is about selling: tills, stock across several channels at once and preparing for the season. Many companies need both, and then we simply combine the scope in one contract.
Both, and we usually start with the manufacturing one, because that is where the split between the office network and the machine network gets decided, and that decision overrides everything else. Coverage in the warehouse and the terminals come in as a separate topic, on the same terms as described here.
Walking a warehouse usually takes half a day, and you get the findings on one page the next business day. The survey before a quote is free and carries no commitment, including when the finding is that nothing is worth buying. Only after that do we talk about scope and price.
The first meeting takes 30 minutes and commits you to nothing. We will tell you straight whether we can help and what we would propose.
2019 - 2024 Interactive Workspace Ltd. - All rights reserved.