Sir
What happens if you power up 10 modules at the same time to make JOIN ?
Will this delay each one during JOIN ?
For example, one will try, but the channel is busy, then will try late ?
And how long should be this delay ?
Thanks!
You should avoid starting a bunch up at the same time, and/or their firmware should have random delay before the first attempt and between attempts. Even something like counting to 10 before turning each next one on would help, though a minute or two would be better.
There are a few uplink channels to work with which are supposed to be randomly chosen so if the nodes are at comparable not to close distances to the gateway, collisions there will vary. But the uplink to downlink timing only has two opportunities, and a gateway can only transmit one downlink at a time. So even if TTN received several of the join requests at the same time, it could only respond to one or two of those with a given gateway. With additional gateways in range there might be more parallelism, but the 2nd downlink opportunity uses a shared channel, so in terms of literally simultaneous join requests, the maximum that could be responded to would be 1 plus the smaller of the number of uplink channels involved or gateways involved.
@cslorabox
Hi, all your works really make sense!
I will implement a RANDOM delay before try to make the JOIN.
The good is that i wont use DOWNLINK messages.
What happens if i send a UPLINK message and 4 gateways receive that ? Will the TTN receive 4 messages ?
Thank you very much!
That’s good in general, But keep in mind that:
-
A join accept is a form of downlink
-
TTN itself will send some network management messages especially towards the start of a session. For some of these, if the node does not indicate it received them, they will be re-sent until it does.
-
Most non-mobile applications should use adaptive data rate (ADR) which relies on the occasional automatic downlink at long intervals.
If the gateways are functioning properly and their internet connections do not have undue delay they will be merged by the de-duplication algorithm into one message with four distinct signal reports. If the network needs to reply, it can then pick the gateway with the strongest uplink signal (that still has downlink aritme to spend) to assign to send the reply.
If one of the gateways is seriously delayed, you might see that come through as its own distinct packet, but that packet would have an illegally already used frame count, and so be dropped.
I am using