Do you see any packets in the gateway’s raw traffic page which hint that they might be from these nodes?
Yes, you can see in the image attached that the gateway seems to receive the raw packets from the TwasTe sensor.
The thing is that everytime I manually reboot the sensor, I can see the activation and uplink and downlink messages (in the application dashboard) and their counterparts in the gateway dashboard. But after these two uplink-dowlink messages the sensor seems to stop to transmit any data.
If you stop seeing any uplinks in the raw traffic page, that sounds like a bug in the sensor firmware.
It may only be doing joins and not suitable configured to send data.
Worth noting that repeatedly restarting a sensor is a bad idea - you’ll end up consuming the join nonce space. And unless you store the values you already use, it may take more and more tries each time to “walk” to a value which is legal by virtue of not having been previously used.
Also worth noting that one of our uplink messages seemed to be sending confirmed data.
That is probably a mistake, not only as it will rapidly burn through your allowance of downlinks by using them for confirmation.
But also because LoRaWAN’s confirmation protocol has what is arguably a design error - if the node transmits and the network confirms, but if for some reason (including the gateway failing to actually transmit it) the node doesn’t hear the confirmation, then the node will retry with the same frame count. But because the network has already seen that frame count, it will ignore the retry.
So if you miss the ack for a confirmed uplink that was actually received by the network, you are NEVER going to get an uplink for that packet with that frame count. The only way to get a result is to increment the frame count.
For most applications, it’s better not to use confirmed uplink and try to re-send increasingly stale data, but rather to send fresher data instead.
In the rare cases where you need absolute certainty that a message was received, do it at application level rather than LoRaWAN level and write your own responder which implements a strategy that first your need.
