Agricultural sensing is not a harder measurement problem than industrial sensing. It is a harder deployment problem: no mains, marginal or absent cellular coverage, distances measured in kilometres, livestock that destroy anything reachable, and a system that must survive being ignored for an entire off-season and still work in spring.
Each one has to survive a season outdoors, unattended, on a battery.
| Signal | How | What it changes |
|---|---|---|
| Soil moisture at depth | Capacitive or tensiometric probes at several depths | Irrigation driven by root-zone moisture rather than by a calendar |
| Soil temperature and conductivity | Combined probe at the same points | Planting timing, and salinity or nutrient movement over a season |
| Weather and evapotranspiration | On-holding station rather than a regional forecast | Water demand calculated for this field, not for the nearest town |
| Irrigation control | Valve actuation with confirmed state | Closed loop, with the flow confirming the valve actually moved |
| Livestock location and activity | GNSS or LoRaWAN collars with movement sensing | Grazing distribution, calving alerts and theft or stray detection |
| Machinery utilisation | Telematics on tractors and implements | Hours, coverage and idle time across a holding or a contractor fleet |
The sensing is the solved part. These are what determine whether the deployment is still running in year three.
This is the defining constraint. LoRaWAN with an on-holding gateway, or satellite where a holding exceeds any terrestrial reach, and a reporting policy that treats every transmission as expensive. The connectivity architecture is settled before the sensing is chosen.
Capacitive moisture readings vary with soil type, compaction and salinity. A factory calibration gives a number that trends correctly and is absolutely wrong, which matters when an irrigation threshold is set against it. Per-site calibration is part of commissioning.
Anything reachable will be rubbed against, chewed, or stood on, and everything else faces frost, UV and sustained wet. Enclosure, mounting and cable strategy carry as much design weight as the electronics.
Equipment is ignored for months, then needed immediately. Devices must recover from deep discharge, rejoin the network without intervention and report their own health, because nobody is going to check them beforehand.
A grower does not want a moisture graph. They want to know whether to irrigate tonight and for how long. Turning telemetry into that recommendation, at the right time of day and on a phone in a field, is where the value is realised or lost.
Arable, livestock and controlled environment.
Root-zone moisture, evapotranspiration and valve control, with water applied to measured demand.
Location, activity and grazing distribution, with calving and stray alerting across large holdings.
Utilisation, coverage and fuel across tractors and implements, including out of coverage.
Climate, substrate and dosing control where the growing environment is engineered rather than observed.
Range and power. A holding needs kilometres of coverage from devices running years on a battery, and LoRaWAN provides that where cellular coverage is marginal and mains power is absent. Where a holding exceeds any terrestrial coverage, satellite carries the exception traffic.
Yes, per site. Capacitive readings vary with soil type, compaction and salinity, so a factory calibration trends correctly but is absolutely wrong — which matters when an irrigation threshold is set against it.
Devices are designed to recover from deep discharge, rejoin the network without intervention and report their own health. Assuming someone will check them before the season starts is the most common way these deployments fail quietly.
A decision. A grower does not want a moisture graph; they want to know whether to irrigate tonight and for how long. Producing that, at the right time and on a phone in a field, is where the value is realised.
Tell us the specification, the constraint and the deadline. Programmes that cross silicon, radio, embedded and AI are where Faststream is strongest.