Table of Contents:
LoRa transmission is a broadcast protocol. There is no fix connection between gateway and end node.
For uplink steam: Each End node will broadcast the uplink message and all-around gateways will get this message and forward to LoRaWAN Server.
For Downlink Stream: Gateway will get instruction from LoRaWAN server and then broadcast this message to LoRaWAN End Node.
If we consider using LHT65N as an example to calculate the traffic, we can get a rough idea for how many messages are there:
A general LoRaWAN gateway such as LPS8N uses a multi-channel LoRaWAN Concentrator. which supports 10 rx channels and 1 tx channel.
When two sensors send uplink in different frequencies, for example, 868.1 nad 868.3. both packets can arrive at the gateway, and the gateway will forward to LoRaWAN Server. Normally, we can consider the gateway support 8 packets per second ( The packets are of different frequencies).
When we check the capacity for the LoRaWAN network, we need to check how the sensor work. We can't simply say how many nodes the gateway can support because nodes' uplink frequency is very important.
In a network with low transmission frequency end node, for example, several hundred LHT65N, transmit every 20 minutes. can be easily supported by a LoRaWAN gateway.
but for a high-frequency application, for example, a tracker sends a message every 3 second, you wil then see packet loss for dozens of units.
Accordiing to the tech spec, simply add Gateways, listening the same frequencies can't improve a lot the capacity.
Above is a simple calculation with rough idea. For more precise, please check this LoRaWAN Capacitor White Paper