Concentratord in US915

[gregory81]

Brand new to LoRa and trying to get concentratord working in the US915. I don’t understand what the sub bands are for? Can I only configure the gateway to use a single sub band? How does the end device know what sub band a gateway is listening on?

I also don’t understand what multi_sf vs lora_std vs concentrator.fsk is for…

Thanks!
super noob…

[gateway]
  region="US915"

  antenna_gain=0

  lorawan_public=true

  model="waveshare_sx1302_lorawan_gateway_hat"

  model_flags=[]

  time_fallback_enabled=true

  # Gateway ID.
  gateway_id="0000000000000000"

  [gateway.concentrator]

    multi_sf_channels=[
      903900000,
      904100000,
      904300000,
      904500000,
      904700000,
      904900000,
      905100000,
      905300000,
    ]

    [gateway.concentrator.lora_std]
      frequency=904600000
      bandwidth=500000
      spreading_factor=8

    [gateway.concentrator.fsk]
      frequency=905300000
      bandwidth=125000
      datarate=50000

Raspberrypi with the Waveshare SX1303 915M LoRaWAN Gateway HAT running Ubuntu 22.04, concentratord v4 and chirpstack-gateway-bridge v4

[sp193]

This is related to the capabilities of a gateway. Contemporary gateways have a Semtech SX1301 chip, which has:

• 8x multi-SF channels, 125KHz bandwidth.
• 1 downlink path.
• 1x fixed-SF channel, configurable bandwidth.
• 1x FSK mode channel.

SF stands for Spreading Factor (SF). Supporting multiple SFs automatically, allows the channels to support most of the official datarates that you would normally use.

LoRaWAN uplink normally involves the the 8x channels, thus most gateways that have a single SX1301 are known as 8-channel gateways.
To have support for 16 or more channels, multiple SX1301 chips are required. The difference is the cost, which is why the higher-spec gateways appear less often.

The fixed-SF (“LoRa STD”) and FSK channels are intended for backhauling and meshing, and thus usually go unused.

US915 has a very wide frequency range. By the LoRaWAN specification, we should support 64 uplink channels and the device should also support all 64… but you cannot do that with an 8-channel gateway. So a strategy is to sub-divide the 64 channels into groups of 8 and pick one. These are the sub-bands as you know of.

So as you can tell, this is somewhat non-standard. The information on which 8 channels to use, must be communicated out-of-band, between the LoRa network operator and the sensor node owner/manufacturer.

[brocaar]

@sp193

The information on which 8 channels to use, must be communicated out-of-band, between the LoRa network operator and the sensor node owner/manufacturer.

An OTAA device should support 64 (+ 8) channels by default. During OTAA it will try different channels (ideally switching between sub-band each attempt). It is expected that it needs to re-try several times. As part of the join-accept message it will receive the channel-mask This way, devices do not have to know the sub-band in advance but will receive all config from the network on join.