Use of Frequency Sub Bands for China

Gateways
1

We have a LoRa demo that we try to use in China (with TTN network). It is made of a LoRa device and a LoRa gateway. Both have been registered on The Things Network.

On the page giving the frequency plan per region (Frequency Plans | The Things Stack for LoRaWAN), we have seen that it indicates to use CN_470_510_FSB_11 rather than CN_470_510_FSB_1

Here are the frequencies for FSB 11: lorawan-frequency-plans/CN_470_510_FSB_11.yml at master · TheThingsNetwork/lorawan-frequency-plans · GitHub

And here are the ones for FSB 1: lorawan-frequency-plans/CN_470_510_FSB_1.yml at master · TheThingsNetwork/lorawan-frequency-plans · GitHub

We currently have an issue which seems related to the choice of this Frequency Sub Band.
For the moment the LoRa device and the LoRa gateway don’t understand eachother.

We have enabled some debugging info and seen that the LoRa device communicates on the following frequencies:

28s813:Channel:82 Frequency = 486700000
…
53s674:Channel:86 Frequency = 487500000
…
653s681:Channel:87 Frequency = 487700000
…
1253s688:Channel:85 Frequency = 487300000
…
1853s695:Channel:86 Frequency = 487500000
(those frequencies are in FSB 11)
wheras the LoRa gateway accepts frequencies between 471.5 MHz and 472.9 MHz (those frequencies are partly in FSB 1).

I think that the LoRa gateway is configured for FSB 1 whereas the device is trying to communicate on FSB 11.

Can you please give us some advices? Is a LoRa gateway supposed to scan every Frequency Sub Bands (here FSB 1 + FSB 11) or is it normal that it scans only some frequencies?

Here we own the gateway so we can try to change its configuration but how would it work if a LoRa device configured for FSB 11 tries to contact a Gateway configured for FSB 1?

Thank you

2

Simply put it wouldnt! They need to use same channels AND Freq plan.

Typically only scan supported channels. Some territories have many channels available (as a collision/congestion mitigation tactic for general ISM Band use) - e.g. IIRC US has 64, hence need to tell device and GW which sub band to use otherwise they will be off talking/listening on different channels and statistically never the twain shall meet - well not often enough to get meaningful message throughput! Similar situation is AU, where it can get even more complicated as not only different channels in supported bands, but also 2 bands available - AU915 & AS923 - so GW and node have to be chosen, and configured, to target both desired band AND required sub-band of channels. Not uncommon for AU sites to be set up with dual GW’s to cover both bands (our friends at Meshed often have to do that). Most GW’s only support 8 channels; some can be configured with additional concentrator card(s) to handle 16, though often just doubling up capacity on same 8 channels, its rare to see one configured for say 32 or even 64 channels (though I have seen that on some research networks in the past…)