Hi,
As we know that loraWAN have timing requirement. For example, if lora node rx1 delay set as 1 second, transform cost time between gateway and loraserver must less than 1 second. But internet always jitter, we can’t guarantee it. Did anyone have good idea about it?
Hope you can understand me. Thanks! 
OTAA has a RX1 delay of 5 seconds, so that should give you plenty of time. The for downlink data the RX1 receive-window delay is indeed 1 second by default. However, this value is configurable in the application and node settings using the RX1 delay setting.
By setting this to for example 5 seconds, you are able to increase the RX1 delay for downlink. Note that after changing this value, you need to perform a new OTAA.
Hi brocaar,
Yes OTAA has 5 seconds for RX1 delay, but how do I know how much delay should I set on lora node? Because different gateway may has different delay. Or I set all lora node as 5 seconds for RX1 delay? Seems it is enough…
That depends on your network latency. However as long as the RX1 delay > roundtrip from and to your gateway + a small margin for de-duplication etc, all should be fine. Please see the specification for all valid RX1 Delay values and more information: https://www.lora-alliance.org/lorawan-for-developers.
Have you think about distributed loraserver?
Every cloud node around world have it’s loraserver and lora-app-server, and one centre node only have lora-app-server. Centre node have all lora node nodeID, network session key, app session key… And those cloud node sync with centre node regularly.
Each gateway communicate nearest cloud node, and this cloud node will reply ACK of it’s loraserver and lora-app-server to avoid latency. I think this method will no need to caculate RX1 delay on each areal. And I know it is not easy…