Skip to content
MatrixAILORAWAN

Technical article

Why LoRaWAN Meters Go Silent for 30 Minutes: Gateway Duty-Cycle Exhaustion in EU868

Regular gaps of 25–30 minutes across many meters at once, with healthy gateways and good signal, usually point to one cause: the gateway has used up its EU868 transmit budget.

By , Co-Founder, CTO & Head of R&D, MatrixAI · · 10 min read

The symptom

On one of our production networks, two separate fleets of electricity meters started showing the same pattern: many devices went silent at the same moment, stayed silent for roughly 26 to 30 minutes, then came back together. Gateways were online, signal strength and SNR were good, and nothing had changed on site.

When many devices fail together on a schedule rather than one by one, the problem is rarely the devices or the radio path. It is usually something shared — and on an EU868 network the most common shared resource that runs out is the gateway's transmit time.

The rule behind it: 1% duty cycle applies to gateways too

EU868 channels sit in sub-bands with regulatory duty-cycle limits. The default uplink channels (868.1, 868.3 and 868.5 MHz) are in a sub-band limited to 1%: any transmitter may be on air for at most 36 seconds per hour there. The RX2 downlink frequency, 869.525 MHz, sits in a sub-band that allows 10%.

Engineers usually think about duty cycle for devices. But a gateway is a transmitter as well, and in the default configuration it answers devices in the RX1 window on the same channel the device used — inside the 1% sub-band. Every downlink spends part of a budget that the whole fleet shares.

  • Acknowledgements for confirmed uplinks — one for every message from every device that asks for one
  • Join-accepts when devices join or rejoin
  • MAC commands from the network server, such as ADR adjustments and device-status requests
  • Application downlinks: configuration changes, time synchronisation, setpoints and relay commands

A long downlink at a slow data rate can take more than a second. At that cost, a gateway can only send a few dozen downlinks per hour in the 1% sub-band before its budget is gone.

How an exhausted budget turns into a 30-minute gap

Once the budget is used up, the gateway cannot transmit and the downlink is dropped. If meters send confirmed uplinks, they stop receiving acknowledgements. Depending on the firmware, they retransmit — consuming more airtime — and after several missed acknowledgements many devices assume the link is lost and start rejoining. Join-accepts are downlinks too, so they are also blocked.

The devices then sit in a join loop with growing back-off intervals until the gateway's budget recovers. Because the budget is shared, a large group of devices loses contact at the same moment and recovers at roughly the same time. That is exactly the synchronised, half-hour gap pattern we saw.

How to prove it

  • Line up the gap start times across devices. Gaps that begin within the same minute on many devices point to a shared cause.
  • Count downlinks per gateway per hour in the network server, and compare them with the gaps.
  • In ChirpStack, inspect the gateway's transmit acknowledgements: downlinks rejected for duty-cycle reasons are reported with a duty-cycle overflow status.
  • Check device profiles and firmware for confirmed uplinks and aggressive rejoin settings.
  • Look at the device frame log: a burst of join requests after the gap starts confirms the rejoin loop.

How to fix it

  • Use unconfirmed uplinks for periodic meter data. The next reading supersedes a lost one; acknowledging every reading wastes the gateway's budget.
  • Send downlinks in RX2 on 869.525 MHz where your network server and devices support it, so they use the 10% sub-band rather than the 1% one.
  • Reduce MAC and application downlinks: avoid unnecessary device-status requests, batch configuration changes and schedule time sync sparingly.
  • Add gateways. The network server sends each downlink through one gateway, so more gateways spread the transmit load and let devices use faster data rates.
  • Review device rejoin logic. A device should not rejoin after a handful of missed acknowledgements on a healthy network.

On our network, moving meter data to unconfirmed uplinks and cutting avoidable downlinks removed the gaps without changing any hardware.

Designing to avoid it

Duty cycle is a design input, not a tuning detail. Before roll-out, estimate the downlinks per gateway per hour for the whole fleet — acknowledgements, joins, ADR and application traffic — and check them against the budget at the slowest data rate you expect. If the numbers are tight, change the design before the devices are installed.

Deployments in the UAE commonly use the EU868 plan, and devices and network servers enforce its limits. Confirm local regulatory requirements during design.

Frequently asked questions

What is the duty cycle limit in EU868?

Most EU868 uplink channels are in sub-bands limited to 1%, which is 36 seconds of transmission per hour per transmitter. The 869.4–869.65 MHz sub-band, used for RX2 downlinks at 869.525 MHz, allows 10%.

Do LoRaWAN gateways have to respect duty cycle?

Yes. A gateway is a transmitter, so its downlinks are subject to the same sub-band limits. Network servers track the gateway's budget and drop downlinks that would exceed it.

Should LoRaWAN meters use confirmed uplinks?

Usually not for periodic readings. Confirmed uplinks generate one downlink per message and can exhaust the gateway's budget on large fleets. Use them sparingly, for critical events such as alarms.

Have an IoT project in mind?

Tell us about your site, assets and existing systems. A LoRaWAN specialist will review your requirements and propose a practical architecture — no obligation.