Big iron, small gateway — the unequal pair every off-highway deployment runs on.

When someone in IoT says “we do equipment monitoring,” there’s a fair chance they mean a forklift in a warehouse with wifi. That’s a real problem. It’s also not the problem we get called about.

Off-highway equipment — the iron that works where the road ends — is its own category. It looks superficially like on-road or factory telemetry. It is not.

Here’s what’s actually different, and why it changes the build.

Definitions, briefly

“Off-highway” is the catch-all for equipment that doesn’t operate on public roads and isn’t bolted to a factory floor:

  • Construction: excavators, dozers, skid-steers, articulated dumps, cranes
  • Mining: haul trucks, drills, loaders, dragline shovels
  • Agriculture: tractors, combines, sprayers, irrigation pivots
  • Forestry: feller-bunchers, skidders, harvesters
  • Energy: gensets, compressors, frac pumps, the temporary infrastructure that pops up around a wellhead or a pipeline
  • Marine and ports: yard cranes, RTGs, terminal tractors

The common thread isn’t the work. It’s the environment. Off-highway equipment lives where infrastructure stops — no reliable power, no IT department, no LAN, often no cell signal, and a duty cycle the cab is built to survive but the bolt-on electronics aren’t.

Connectivity is a planning problem, not a checkbox

A forklift in a warehouse is on wifi. A skid-steer at a remote earthmoving site is six miles from the nearest tower with a hill in the way.

Off-highway connectivity choices we actually make, in rough order of how often:

  • Cellular at the gateway, BLE at the asset. The asset doesn’t need its own modem and SIM — it needs a beacon. The site needs one cellular gateway in a good spot, and everything in range of that gateway gets to the cloud through it.
  • Multi-carrier SIMs on the gateway. Carrier coverage is a regional patchwork. We’ve never deployed a serious off-highway gateway with a single-carrier SIM by choice.
  • Store-and-forward on dropouts. The data has to survive an hour with no signal and reconcile itself when the link comes back. Anything that loses data when the backhaul flaps is a problem we’ve already had.
  • Satellite as the fallback, not the primary. Iridium and Inmarsat work, but at a cost and payload-size constraint that makes “send everything, sort it later” infeasible. We compress and prioritize before bytes go over satellite.

Designing for one of these and discovering you needed another is one of the more expensive mistakes in this category. Site surveys aren’t optional — we wrote up what we measure before quoting a remote install.

Power doesn’t exist

Industrial equipment is full of power. None of it is for you.

A haul truck has a 24V system that the OEM has very specific opinions about. A genset has fuel and an alternator, but tapping the battery without a fused, transient-protected harness is how you find out what voltage spikes look like in person. An ag tractor in winter storage has no power at all.

Practical implications:

  • Gateways at the site, not on the asset. Solar plus battery for the gateway, sized for the worst month — see the sizing math. Asset tags are battery-powered BLE, sized for years of service.
  • Asset-side telemetry is opportunistic. A beacon transmits when it can, sleeps deeply when it can’t, and is durable enough to survive the way iron actually gets washed — which is to say, with a pressure washer aimed at the place we’d prefer it not be aimed.
  • OEM CAN-bus integration is a separate decision. When the asset has a J1939 bus and we’re tapping it, we’re now in the OEM’s electrical system, drawing OEM power, and dealing with OEM transient behavior. That’s a real-power install with real-power risk. It’s not “we’ll plug it in.”

If a vendor’s pitch involves a USB-powered device zip-tied to a roll cage, they have not done this before.

The duty cycle is brutal

A factory PLC sits at room temperature its whole life. An off-highway asset spends days at -25°C, weeks at 50°C, every hour in vibration that loosens fasteners and cracks solder joints, and the occasional 9g jolt from a track hitting a rock.

We don’t get to test for this in a lab. We test for it by recovering devices that have been in the field for a year and seeing what the failures look like. The pattern, after enough recoveries:

  • Connectors are the first thing to die, not the silicon.
  • Mounting hardware fatigues before the electronics do. A bracket cracked at the weld is a $4 part that costs a return trip.
  • Sealed enclosures aren’t, unless the gasket was actually compressed during install. We’ve cut open more enclosures full of red dust than we’d like to admit.
  • The temperature spec on the datasheet is the spec the silicon survives. It’s not the spec at which the LCD remains readable, the battery charges, or the cell modem holds a connection.

Hardened-for-off-highway is a real engineering effort. It’s also one of the easiest places in this industry to be told it’s been done when it hasn’t.

The OEM problem

Big iron does have telemetry. Caterpillar has VisionLink. John Deere has Operations Center. Komatsu has Komtrax. Volvo has CareTrack. They’re all real systems, and they’re all stuck behind the OEM’s customer portal, with the OEM’s definitions of CMH and utilization, and no interop with each other.

If a fleet is single-OEM, single-vendor, and brand-new, the OEM portal is fine. We don’t sell into that fleet.

The fleet that calls us looks like this: 14 Caterpillar machines, 9 Komatsu, 6 Volvo, 4 John Deere ag tractors someone bought used, a half-dozen rental skid-steers from three rental houses, and four gensets nobody remembers the brand of. Five OEM portals, three rental dashboards, four spreadsheets, and one ops lead trying to answer “how many hours did the fleet run last week.”

Off-highway is overwhelmingly a heterogeneous, mixed-OEM, partly-rented, partly-owned, partly-thirty-years-old problem. The OEM telemetry stacks are real assets when they exist. They aren’t an answer to “I have a fleet.”

The retrofit approach — one gateway per site, beacons on the iron, OEM CAN integration where it makes sense, the same definitions and the same dashboard across every machine regardless of color — is the answer to that fleet. We wrote about how to think about CMH across mixed iron in the warranty post and in the telemetry-vs-clipboard piece.

Operator interaction is different

A factory operator can use a tablet, scan a barcode, check a screen. An equipment operator is wearing gloves, in a cab full of vibration, and has either zero hands free or one hand on a joystick. Anything that requires operator input from the cab has to assume they will not do it.

Things we’ve shipped that asked the operator to do something, that didn’t survive contact with the cab:

  • Press a button to log a fault — they don’t, because they’re working.
  • Confirm an asset transfer on a phone — they don’t, because the phone is in a truck a quarter-mile away.
  • Enter a passcode after a swap — they don’t, because passcodes are not what working iron rewards.

What works: automatic. Tag, hub, range — the system figures out which machine is at which site without an operator’s help. Operator input is a fallback for ambiguity, not the primary mechanism.

What this means for the build

The shape of an off-highway IoT deployment, distilled:

  • Site-level gateway with multi-carrier cellular and solar power, sized for the worst month
  • BLE asset tags rated for the cab’s actual environment, with a hardware design that assumes the connector dies first
  • Store-and-forward at every layer, because the backhaul is going to drop
  • OEM telemetry consumed where it exists, ignored where it doesn’t, harmonized to a single set of definitions across the fleet
  • An app and a process built on the assumption that the operator cannot input anything from the cab

Most of the difficulty in off-highway IoT is design discipline applied before the first PO. The hardware exists. The cellular networks exist. The protocols exist. What separates a deployment that works from one that calls us in year two is whether somebody held the line on those constraints in the planning phase.