Clock-synchronization feature based on the average time from the last N adverts

[liktor88]

In case of hidden and inaccessible devices on the roof I always add Arduino for restart device each 24hours to avoid complete system crash. But after reset, clock on RPT have wrong date (~2024) - and this is problem for adverts send from restarted RPT.

Clock sync function based on average time provided from last N adverts received. Function should be enable/disable in the settings with information how often synchronization should be executed and how many times.

Required:
enable/disable in settings
configurable how often synchronization runs (interval)
configurable how many samples to average (5 or 10)
configurable how many attempts to perform per scheduled syncimmediate re-sync-on-boot behavior if system clock is clearly wrong (after received N required adverts)

[IoTThinks]

I think we can reference to how NTP works.

Generally, it is similar to what you are proposing.

However, I still believe we should remove time in repeaters.
So we can avoid this problem totally.

[4np]

Average would be unreliable as time from other repeaters cannot be trusted either. Probably it's best to take the last n packets from different -low hopcount- sources that have a time, and determine the min/max variation. If the min / max exceeds a number of seconds, it can be deemed unreliable and ignored / discarded. Ideally you would only consider packets which devices contain an RTC, that's probably not advertised so it could be useful if that's sent along with the time.

There will always be some invalid time as the transmission duration is not taken into account and can be many seconds. But it could help to get rid of those dates that are a year off (or more).

[tonurics]

I was coming here to suggest something similar. But rather than the adjustment always happening: make it an optional setting for hardware with known clock issues.

We should also use the median value of the observed packets for the datetime; which would be less affected by any outliers than the mean (average).