If you have IKEA ALPSTUGA air quality monitors and devices on your Thread network keep dropping offline, especially battery sensors that only come back after a battery pull, this may be why.
The bug
Every Thread network regularly changes its security key, by default every 28 days (672 hours). Each device keeps a timer for this, and it should restart whenever the key changes.
ALPSTUGA firmware 1.0.26 reports an OpenThread version from October 2023. In that code, the first key change after a reboot doesn’t restart the timer. So a monitor can change the network key 28 days after its last reboot, even if the key changed minutes ago. OpenThread replaced this logic in March 2024, but the ALPSTUGA firmware build is dated October 2024. 1.0.26 is still the newest version in the Matter registry.
What it looks like
I power-cycled every device on my network except the border routers on 2026-10-06. Nothing changed for 28 days. Then, on 2026-11-03 (UTC):
| Key change | Started by | That monitor’s reboot, 10-06 | Its first frame on the new key, 11-03 |
|---|---|---|---|
| 90 → 91 | Monitor A | 10:44:55 | 10:44:46 |
| 91 → 92 | Monitor B | 11:22:58 | 11:22:58 |
| 91 → 92 | Monitor C | 11:23:48 | 11:23:49 |
The first change was the normal rotation. The other two came 38 minutes later, exactly 28 days after those monitors rebooted. This logic correctly predicted which key all 14 of my monitors ended up on.
Why it matters
Thread devices refuse a second key change too soon after the first, so a rapid change splits the network across two keys. My network is split that way right now. A device that falls two keys behind can no longer decrypt traffic, even though its radio still looks alive.
In September and October my key changed six times in 28 days, when it should have changed once. ALPSTUGAs started most of those changes, and smoke, CO, lock and presence devices from other brands got stranded. I no longer have the records to tie most of those changes to this bug, and nothing went offline this time, but this is the failure to watch for.
Signs to search for
- Thread battery devices go unavailable in Home Assistant while mains-powered devices stay fine
- A battery pull or power cycle brings them back, or they rejoin on their own minutes or hours later
Frame rx failed, error:Securityin the OpenThread Border Router logot-ctl keysequence counterjumps much sooner than every 28 days- Trouble starts about 28 days after you power-cycled your monitors
Other problems can cause the first two, so these are clues, not proof.
What needs to happen
IKEA needs to update OpenThread in the ALPSTUGA firmware, or backport the fix. I couldn’t find a bug tracker, so on 2026-11-03 I emailed the full details to the two addresses on the IKEA Home smart app’s Google Play listing. The support address rejected it (550 5.4.1 Access denied). The developer address didn’t bounce, but I haven’t had a reply, and I’ll update this thread if I do. If this is happening to you, please tell IKEA support too. The more reports they get, the better.
I’ve inferred the cause from the version the monitors report and from the timing. I haven’t seen IKEA’s source, and their build may include patches. One other product on my network, Eve Energy, reports a version with the same old logic, though I haven’t seen one do this.
I found this with threadwatch, my open-source Thread flight recorder. I’m happy to share more detail with anyone working on this.