Industrial LoRaWAN Site Surveys: Coverage and Acceptance

Make the survey answer an operational question

Start with the endpoint register and the requirement for each device group. A daily stock observation, a regularly updated process trend and a maintenance alert have different useful delivery windows. State the payload, reporting schedule and the point at which a record must be available to the user. Agree whether the test concerns radio reception, delivery to the network server or the decoded application record.

Download the editable site survey and acceptance worksheet. It includes position, operating condition, configuration, expected and received records, timing, acceptance criteria and retest fields. Each row starts as not tested. Populate the criteria with the customer before visiting the site, and add a row for each endpoint and condition that needs its own decision.

Survey the position that will actually be installed

Use a device and antenna arrangement representative of the proposed installation. Record the height, orientation, enclosure and nearby surfaces, together with the gateway and antenna position. Include photographs or drawing references that allow another technician to repeat the observation. If a temporary test arrangement differs from the final mounting, record the difference and schedule a verification at the installed position.

Choose conditions that could change the radio path: closed metal doors, normal machine operation, parked vehicles, stocked racks or a device behind a tank. Identify which conditions occurred during the test and which remain outstanding. A convenient corridor test can help diagnose a problem, but it should remain distinguishable from the result at the endpoint's intended working position.

Record the radio configuration with the signal observations

Capture the device model and firmware, frequency-plan configuration, channel or frequency information available in the logs, spreading factor, bandwidth and transmit-power setting. Retain the gateway identifier with each received copy. RSSI is recorded in dBm and SNR in dB. Their meaning depends on the receiver, modulation settings and conditions, so a single generic threshold is a poor substitute for an agreed application test.

The Things Network documents the distinction between received power and the signal-to-noise ratio. It also describes adaptive data rate as a mechanism that changes transmission parameters. Record whether ADR is enabled and whether the observation represents initial or settled behaviour. If the network changes a setting during the trial, retain that change with the test interval so two apparently similar runs can be compared correctly.

Define the denominator before reporting a delivery percentage

An expected-record count must have a defensible source. Use an endpoint log, a controlled test sequence or a verified reporting schedule with recorded start and end conditions. An event-driven device needs an event reference; assuming it sends at a fixed interval can create an incorrect denominator. Counter resets, device restarts and suppressed reports should be identified separately from unexplained losses.

For an application-delivery test, divide the unique expected record identifiers received by the number generated during the agreed window. Copies heard by multiple gateways count as reception diversity, not additional business records. If a payload contains several measurements, decide whether the acceptance unit is the transmission or the individual measurement and retain the relationship. Describe the metric as application delivery when that is the boundary observed; investigate radio loss using the appropriate device and gateway evidence.

Treat freshness as a separate acceptance result

A record can arrive successfully after it is useful for the intended workflow. Define the deadline and identify the clocks used to measure it. A measurement timestamp and a server arrival timestamp permit an age calculation only when their meanings and clock relationship are understood. Where the endpoint has no reliable timestamp, use an approved event reference or report the timing limitation in the test result.

Keep late records separate from records still missing at the end of the observation period. State how long recovery was observed and whether retained data preserved their original time. A report that resumes showing a current value does not demonstrate that every earlier sample was recovered. The test should make clear which historical records became available and which gaps remain in the operational trend.

Exercise the expected workload and downlink pattern

Test planning includes the proposed device population, payload sizes, report intervals and configuration traffic. A small survey can establish candidate locations, while a staged commissioning test checks the workload of the installed group. Record any difference between the test load and the final operating load. Examine synchronised reporting, repeated joins and retry behaviour where they are relevant to the design.

Downlink configuration and acknowledgement traffic also need an agreed pattern. Use the intended settings in the acceptance trial rather than enabling acknowledgements for every message simply to make the demonstration look reassuring. The network designer should review the resulting airtime and operating constraints for the selected configuration. Retain those choices so later changes in reporting frequency trigger an informed capacity review.

Test recovery across the entire path

Plan approved interruptions that distinguish endpoint power, gateway power, 4G backhaul and application processing. Record the affected component and the exact test window. Check fault indication, retained history and the order in which records return. A device's ability to retain measurements and a gateway's handling of traffic during a backhaul interruption are separate configuration questions.

Use an isolated or simulated test where an interruption would interfere with operations. The customer should know which recovery behaviour was demonstrated directly and which still requires an installation test. Assign a responsible person to every failed or untested requirement. This converts the survey into a commissioning record that the operational team can use after the project team leaves.

Close each location with a repeatable decision

Keep the original result when moving an antenna or adding a gateway. Repeat the relevant conditions and record the new configuration. A final coverage drawing should point to these accepted observations, allowing maintenance staff to understand why a particular location or mounting arrangement was chosen. Changes to buildings, stock layouts and reporting workloads can then be assessed against a known installation record.

The industrial LoRaWAN deployment service turns this worksheet into a survey, installation and handover scope. The network architecture guide explains server and gateway choices. Use the Industrial IoT service and IoT solutions overview to identify the wider instrument, reporting and support responsibilities around the network.

Frequently asked questions

Method references

The Things Network: RSSI and SNR and adaptive data rate support the radio terminology above. Reviewed 5 November 2026. The worksheet and acceptance process are engineering planning materials; the customer and project team set the actual test conditions and thresholds.