AgriTech IoT: Field Sensor Networks That Survive Real Farms

A sensor that works in a lab does not necessarily work in a wheat field. AgriTech is one of the most unforgiving IoT environments and one of the most rewarding when you ship it well. Here is the playbook that has held up across deployments in three continents.

The non-obvious constraints

Before sensor selection, before network design, before any code: the constraints that engineers from urban-product backgrounds underestimate.

• Power is scarce. No mains, no convenient battery swaps. Solar with a small battery, sized to survive a week of overcast. Anything else fails by year two.
• Connectivity is intermittent. Cellular coverage on agricultural land is uneven. Wi-Fi is a fantasy. LoRaWAN works if you control the gateway placement.
• The environment is hostile. Dust, water, hail, frost, livestock, insects, agricultural chemicals. IP67 is the minimum. UV exposure shortens enclosure life. Birds will land on your antenna.
• The user is busy. A farmer cannot reset your device, find a Wi-Fi password, or call support during planting season. Onboarding has to happen in under two minutes.
• Service is hard. A field that is fifty kilometers from the nearest road is a long drive. A device that requires a site visit twice a year is uneconomic.

A device that survives all five constraints is a different artifact than a smart-home sensor.

Power architecture

Solar plus battery, sized for the worst case:

• Solar panel: 5-20 W depending on duty cycle and latitude. Mount it where the user cannot accidentally cover it (above the device, angled to local sun path).
• Battery: lithium chemistry suitable for the temperature range. LiFePO4 is the preferred choice for outdoor agricultural use — wider operating temperature than typical Li-Ion, better cycle life.
• Charge controller: MPPT for solar, simple linear is acceptable for very low duty cycle.
• Sleep mode current: target microamps, not milliamps. The device is asleep 99% of the time.

Build a power model in a spreadsheet before the first board spin. “Uses about 100 mA average” is not a model. Hours per day in each mode, sleep current, peak current, charge current, panel output by season — those are a model. The spreadsheet predicts whether the device survives a worst-case week.

Network architecture

For most field deployments, LoRaWAN is the right answer:

• Range: kilometers per gateway, more if line of sight is good.
• Power: device transmissions are brief enough that battery budget works.
• Cost: gateway is one-time capex; no per-device data fees.
• Resilience: works without cellular coverage; gateway can be on the user’s premises.

Variations:

• Mobile fleets (livestock, equipment) need NB-IoT or LTE-M for coverage where LoRa gateways cannot reach.
• Very dense, very low-data deployments might suit Sigfox or unlicensed sub-GHz mesh.
• Gateway-less devices (a sensor that uplinks directly to satellite IoT) are increasingly viable for remote deployments where even LoRa gateway placement is impractical.

Design the gateway placement before placing sensors. Your survey tool is the radio module on the actual hardware, not a coverage map.

Mechanical design

The enclosure is half the engineering. What we always do:

• Sealed connectors rated to the environment. Cable glands fail. Screw-down rated connectors do not.
• Drainage, not just sealing. Even sealed enclosures get internal moisture from temperature cycling. Design a drainage path.
• Mounting that survives users. Farmers will not match your specified torque. Over-build the mount.
• A serviceable battery. Not “tools-required, send-it-back” — actual screwdriver-and-replace serviceable. Five-year batteries fail at year four.
• Visible status indicator. A single LED that the farmer can see from the truck. Not a screen, not an app — an LED that says “this thing is alive.”

Onboarding

The two-minute onboarding test is real. If a farmer cannot install your device, scan a QR code, and see it appear in the app within two minutes, your conversion rate suffers.

The flow that works:

1. Mount the device (one minute, one tool).
2. Scan the QR code on the device with the app.
3. App shows the device on a map, prompts the farmer to confirm the location, suggests the nearest field.
4. Done. The device is in the system.

What is conspicuously absent:

• No Wi-Fi pairing.
• No Bluetooth pairing flow.
• No “register the device on a website.”
• No firmware update during onboarding.

If your protocol forces these, redesign until it does not.

Operational telemetry

Beyond the agronomic data the customer cares about, the device should report on itself:

• Battery voltage and charge state.
• Solar panel current.
• Internal temperature (and external if you have a sensor).
• Last successful uplink timestamp.
• Reset reason, restart counter.
• Signal strength of the most recent uplink.

These are the data the support team uses to decide whether a non-reporting device is broken, dead, or merely behind a barn. Without them, every “device offline” alert is a truck roll.

The boring path that wins

A summary checklist for an AgriTech sensor program:

• LoRaWAN with a customer-premises gateway.
• Solar power with LiFePO4 battery, sized to survive a worst-case week.
• IP67+ enclosure, sealed connectors, drainage, serviceable battery.
• QR-code onboarding, no Wi-Fi or Bluetooth pairing.
• Operational self-telemetry on every uplink.
• Aggressive sleep, microamp quiescent current.
• A spreadsheet power model that predicts outcomes before the first PCB spin.

Skip any of these and you ship a product that demos well and fails in the field.

If you are building a field sensor network, we have shipped these in farms, vineyards, and ranches.