Buy wireless greenhouse monitoring on evidence that observations reach the place where you use them, at an acceptable age, from the actual crop locations. A quoted radio range or a dashboard that updates near the gateway cannot establish coverage and data delivery across the production house.
The useful pre-purchase question is concrete: will this proposed sensor, gateway and receiving application provide the observations needed for this monitoring decision under representative normal conditions? Agree what “fresh enough” and “complete enough” mean before a demonstration.
Define the observation you need
Start with the use. Reviewing yesterday's climate trend and checking a current crop-zone condition may tolerate different delays or gaps. Ask the grower and controls specialist to state the maximum acceptable observation age, missing-data tolerance and evaluation period for each intended monitoring use. Do not borrow a universal percentage or reporting interval from another farm.
Define the vendor's terms as well. Sampling describes when the sensor takes a measurement. Reporting describes when information is transmitted or made available. A display refresh can redraw the same old value. Ask which timestamp identifies the measurement itself and which identifies arrival at the local receiver or dashboard.
NIST's 2018 Guide to Industrial Wireless Systems Deployments starts with measurable application requirements, including process behavior, information volume and delay or reliability needs. It calls for site assessment and warns against using a coverage map as the only performance indicator. Those industrial planning principles are useful here; the guide is not a greenhouse field trial.
Follow the data all the way to its user
Draw the proposed path in plain words: sensor, transmitter if separate, gateway or base station, local receiver, and any network or cloud service needed to reach the chosen application. Identify exact equipment and firmware. Record which part needs batteries, local power, a license or an external connection.
For example, Apogee's Aranet telemetry page distinguishes sensors, base stations, local integration and cloud access. That product architecture illustrates why receipt at a gateway and availability to a remote user are separate observations. It does not show how a particular site will perform.
Commercial greenhouse wireless sensing also includes products such as Heliospectra's DOXA light sensor. A greenhouse application on a product page establishes an intended use. The site's layout, crop state and receiving application still need their own delivery evidence. Optical accuracy is a separate question from transport reliability.
Two demonstrations that leave a gap
The convenient location. A supplier demonstrates a sensor close to the gateway, with an open screen and a clear aisle. Readings appear promptly. Your planned location is at crop height in a distant bay with a developed canopy and a normally closed screen. The demonstration supports only the conditions observed. It leaves that distant point unresolved, even if both locations fall inside the advertised range.
Ask for a supplier-supported observation trial covering the intended points and representative normal crop, screen and equipment states. If the proposed crop has not developed yet, record that limitation and agree what evidence will be needed later. Do not describe an empty-house check as a whole-season result.
The recovered history. In a hypothetical example, a measurement is taken at 10:00 but first appears in the receiving dashboard at 10:20. Its observation age on arrival is 20 minutes, assuming comparable clocks. A decision made at 10:05 could not use that record then. Later recovery may improve the historical dataset, but it does not turn late delivery into timely availability. These times are invented to explain the distinction, not acceptance limits.
Keep measurement, local-arrival and application-arrival timestamps separate. Record the timezone and clock basis before calculating delay. If the clocks cannot be compared, mark age as unresolved rather than producing a precise but misleading number.
Keep a record that can support acceptance
A short signal survey can help locate weak radio paths. Apogee's Radio Site Survey Ping-Pong Kit, for example, indicates radio signal strength between survey locations. A signal indicator is supporting evidence; it does not by itself report the age of a measurement at your dashboard or establish performance through future crop conditions.
| Record | What it tells you |
|---|---|
| Monitoring use and agreed limits | Whether the receiving information can serve the intended decision. |
| Location, height and normal operating state | Which crop and screen conditions the observation actually covers. |
| Sampling and reporting definitions | What records should exist and what their timestamps mean. |
| Sent and received record basis, if available | Whether completeness can be assessed on comparable records. |
| Measurement and arrival timestamps | Observation age at the intended receiving application. |
| Missing observations and stale intervals | Where the service failed to deliver the required information during the evaluation period. |
| Signal indicators and supplier logs | Additional evidence to help the specialist interpret a delivery problem. |
Agree whether counts represent measurements, messages or another record unit. Retries, duplicated messages or reports containing several observations can make unlike counts misleading. If sent records are unavailable, describe the missing-data assessment against the agreed reporting schedule and mark its limitations. Do not label it a measured packet-delivery rate.
Record the longest interval in which the application lacks an observation meeting its agreed age requirement. This differs from the largest spacing between points in a history export. Retain the evidence reference and note whether recovery occurred after the useful decision time.
Run a bounded monitoring trial, then decide
Use the monitoring-point card first, then the delivery observation log. Both are manual worksheets. They do not calculate percentages, synchronize clocks or approve a system. Complete one card per point and use, and identify the observation period in each log row. Summarize the result against the agreed limits on the card.
The trial should observe supplier-supported equipment in authorized, safe, normal operation. It should not involve deliberate interference, radio power or frequency changes, disconnection of live controls or bypassed interlocks. Safety-critical alarms and controls require their own approved design and validation.
Close the buying discussion with an acceptance or hold for the named path and conditions, plus any outstanding crop-stage check. Keep sensor placement and calibration as a separate measurement gate, and plan what remains available during a network outage separately. The next purchase specification can then describe demonstrated monitoring coverage, rather than repeating a range headline.