In 2026, the Badge Does the Reporting

Open any connected-worker pitch deck this year and it starts with the same slide: the market goes from $8.62 billion in 2025 to $20.18 billion by 2030, an 18.5% compound annual growth rate (ResearchAndMarkets, March 2026). Every vendor in this space has that number memorised.

The same report lists its challenges, and they’re a lot less glamorous. Battery life limitations. Device reliability in harsh environments. Those two lines describe about 80% of the field problems I hear about, and neither one gets solved by a better dashboard.

So let’s talk about the part of worker tracking that actually decides whether a site keeps its system running in year two: where the radio intelligence sits, and how many boxes you have to power, mount and maintain to get it.

What actually moved in 2026

Three things pushed this from “nice safety extra” to a line item that gets argued about.

Occupational safety bodies stopped treating wearables as optional. The ILO’s 2025 report on AI and digitalisation at work came out saying digital tools, sensors and wearables are now central to effective occupational health and safety — not supporting extras you add when the budget allows (ILO). That framing changes procurement conversations. A safety manager quoting an ILO position is harder to defer than one quoting a vendor brochure.

Manual monitoring is being retired. The safety trade press spent January writing about the shift away from people watching dashboards and calling workers to check they’re fine (OH&S, Five Safety Technology Shifts for Lone Workers in 2026). Automated check-ins, automatic escalation, no human in the loop for the first sixty seconds. That only works if the device knows where the person is, not just that a button was pressed.

The 2.4 GHz band got more crowded, not less. The Bluetooth SIG’s 2026 dashboard has roughly six billion Bluetooth devices shipping this year, heading to 8.1 billion by 2030 (Bluetooth SIG market dashboard 2026). Your badge is one of six billion things talking in the same band as the Wi-Fi, the forklift telemetry and everyone’s earbuds.

Put those together and the question buyers started asking in 2026 isn’t “does it locate people”. It’s “how many gateways do I have to hang from this ceiling, and who maintains them”.

Two ways to answer “where is Ana right now”

The default architecture is infrastructure-first. You blanket the ceiling with gateways, the badge advertises itself once a second, and each gateway reports the signal strength it heard. Your platform does the maths and turns three RSSI values into a zone.

It works. It’s also the architecture that makes people wince at the quote, because coverage has to reach every corner a person might stand in — stairwells, plant rooms, the back of the cold store. And every one of those gateways needs power, a network drop and a firmware policy.

There’s a second way, and it flips the direction of the traffic. Instead of the badge shouting at the building, the building stays quiet and cheap: small battery beacons fixed at zone boundaries, doorways, gangways, workshops. The badge — already worn, already charged, already there for the SOS button — scans those anchors as the worker moves. Then it packs what it heard into its own advertising packet: anchor addresses, signal strength, anchor battery state. A much sparser layer of gateways picks those digests up and forwards them to your server.

Your platform gets a sentence instead of a shout: badge 4471 heard anchor QM-12 at −71 dBm at 10:41:03, then anchor QM-13 at −68 dBm forty seconds later. Direction of travel, dwell time, last known zone — all derivable. You still own the maths. More on that boundary later.

The reason this matters commercially is simple. Anchors are battery devices with a five-to-eight-year rating. Gateways are powered infrastructure with a maintenance tail. Moving the reading work onto the badge changes the ratio between the two, and on a big site that ratio is the budget.

The bits nobody puts in the deck

I’d rather you hear the trade-offs from a hardware vendor than discover them in month three.

Relaying is still radio, not magic. A badge can only forward what its own transmitter can reach. If a worker spends eight hours in a basement with no gateway in range, the digest sits on the badge until they walk back up. That’s fine for mustering and zone history. It’s not fine if you promised second-by-second location in that basement — in which case put a gateway down there and stop pretending otherwise.

Airtime adds up in a way single-tag deployments hide. One badge digest per second is nothing. Two hundred badges doing it is a crowd on three advertising channels. The fix is boring and effective: longer intervals in low-risk zones, trigger-based forwarding around events, and filter rules at the gateway so it only reports what your software actually uses. Anyone who tells you congestion isn’t a design constraint hasn’t run 500 tags at 100 ms.

Scanning is cheap, but the badge is small. Receive windows cost less energy than transmitting, which is why the reverse architecture is viable at all. Still, a badge that scans and transmits runs on a coin cell or a small rechargeable, not a lantern battery. A 400 mAh-class cell with a 16-month rating between charges is a realistic shape for this class of device. Multiply that by 200 badges and you have a charging rota, and a rota needs an owner.

Wear compliance beats spec sheets. A badge on a lanyard gets left in a locker. Wristbands survive quay work; keychain formats survive technicians; ID-card formats survive anyone who already wears a card. Pick the format per job function, not per purchase order — the best radio in the world reports nothing from inside a locker.

Anchors need their own plan. They’re battery devices in ceilings and doorframes, which means they inherit every battery question you’d ask about a tag, including the EU serviceability direction that’s landing in 2027. Cheap to buy, still a fleet to track.

What KKM actually builds for this

KKM makes hardware and firmware, and the badge-with-a-receiver idea isn’t a concept deck here — it’s a product family. The C3S ID-card badge, W51S wristband, F1S keychain and K9PS asset tag all work as relay badges: they scan nearby beacon data — MAC address, battery level, RSSI — carry it in their advertising packet and hand it to an IoT gateway. Several of them carry an SOS button, which lets a distress event go out with the zone context attached instead of as an anonymous alarm.

The anchors are the ordinary beacon line. The K6 sensor beacon is a useful one here because it earns its keep twice: it broadcasts iBeacon and Eddystone frames, but it also reports temperature and humidity, so a canteen or a plant room anchor doubles as a comfort and conditions reading. Rated up to eight years on a 2600 mAh ER14505 at one-second advertising and 0 dBm, IP63, −40 °C to +70 °C.

The gateway layer doesn’t need to be dense, but it does need to be reliable. The KG01 outdoor gateway is the yard and dock box: IP54, PoE 802.3af or 5 V DC, 300-plus beacons per second, MQTT or HTTPS with TLS 1.2 into your broker or local server, local buffering when the link drops, optional LTE and GPS, and auto-rollback if a remote config change cuts it off from your cloud. The KG02 covers ceilings indoors — around 300 m with its amplifier, 200-plus beacons per second, hardware watchdog, PoE or 5 V.

And for the parts of a site where a cabled gateway is a fantasy — rail sidings, pipeline walk-downs, a mine’s outer road — there’s the LAC3 LoRaWAN badge. It’s a personnel badge on standard LoRaWAN: SOS button, BLE scanning for zone context, accelerometer, NFC for identity and access, IP67, 400 mAh rechargeable rated 16 months per charge, +20 dBm output, EU868/US915/AU915. It talks to any standard LoRaWAN network server, so it doesn’t need a KKM gateway or a KKM cloud anywhere in the chain.

A port terminal, nine gateways, 40 hectares

Composite of the pattern integrators describe, names changed. A regional port operator on the European coast: three berths, about 40 hectares of yard and sheds, 210 people on shift across stevedoring, maintenance and gate operations. Their old kit was 3G pendants, SOS only — no location, and a mobile contract per device. When the operator asked “who is in the reefer yard right now”, the honest answer was “we’d have to radio and ask”.

What went in:

Around 60 K6 anchors on gangways, shed doorframes, workshop entrances, the canteen and the quay walkways.
C3S ID-card badges for roughly 190 staff, and W51S wristbands for the 20-odd quay crew who kept breaking card holders.
Six KG02 gateways indoors across the sheds and workshops, plus three KG01 units on light poles covering the yard approaches.
LAC3 badges for the 20 people who work the rail sidings and the outer road, where no cabled gateway was going.

Nine gateways for 40 hectares, because the anchors did the reading. The blanket-coverage quote they’d had before ran to more than forty.

Two things went sideways, both worth stealing. First, week one showed about 12% of badges never appearing in the platform during a shift. Not a radio problem — lanyards in lockers. The fix was a format change for two job functions and a rule in the supervisor briefing, and it took a month to get under 3%. Second, the reefer yard turned out to be a mess of steel boxes that changed shape every day, so the digests arrived with two to five minutes of latency depending on where the badge was standing. The operations team looked at it, said zone-level with a few minutes of lag is what we actually need, and stopped asking for real-time there. They did add four anchors at blind corners nobody had drawn on the plan.

What they gained wasn’t a live map on a wall. It was a mustering report that takes minutes instead of a radio call to every gang, and an SOS that arrives with a zone attached.

Where the hardware stops

KKM doesn’t sell a platform, and this architecture is a good illustration of why that boundary matters. The badge reports what it heard. The gateway forwards it over MQTT or HTTPS with TLS. Everything after that — the zone map, the dwell-time rules, who gets paged when a badge goes quiet, the escalation policy, the evidence you hand a regulator — is your software, in your cloud or your server room.

That’s not a limitation dressed up as a feature. A port operator’s zone geometry and escalation rules are operational IP, and they shouldn’t be trapped in a hardware vendor’s portal. Swap the badge, keep the logic; swap the gateway, keep the logic.

Six questions before you spec a badge system

Does the badge forward what it hears, or only its own presence? Relay versus plain broadcast decides whether you’re buying anchors or a ceiling full of gateways.
What’s the forwarding policy, and what latency does it imply per zone? A digest on an event trigger and a digest every second are different products with different battery maths.
Can scan window and advertising interval be set separately, per site? This is your only real lever on airtime when badge counts climb.
How does the badge get charged, and who owns that rota? Do the multiplication before the purchase order: badges × charge cycles per year.
Does SOS jump the queue? A safety alarm that waits behind routine telemetry is a design flaw, not a tuning issue.
Do the anchors report their own battery in a frame your platform already ingests? If replacing anchor cells needs a walk-around survey, you’ve built a second maintenance project.

If you’re specifying worker tracking for a 2026 or 2027 rollout and the gateway count is still a spreadsheet cell someone guessed at, that’s a conversation worth having. KKM has been building beacons, gateways and LoRaWAN devices since 2008, and we’re happy to argue about architecture before you buy anything.

Blog