LoRa Rx window timing debugging

[Serafi Nebot 0]

Objective

Reliably join a LoRaWAN network though OTAA & send/receive data.

Setup

The following setup, except the end device, works for a few hundred other devices which have been working for months. None of the already working end devices are Pycom products.

End device

FiPy with the following version:

(sysname='FiPy', nodename='FiPy', release='1.20.2.r6', version='v1.11-c5a0a97 on 2021-10-28', machine='FiPy with ESP32', lorawan='1.0.2', sigfox='1.0.1', pybytes='1.7.1')

Gateway

MikroTik LoRa gateway configured as a packet forwarder.

The gateway LoRa channels are the following:

Number	Frequency (MHz)	Bandwidth
0	868.1	125 kHz
1	868.3	125kHz
2	868.5	125kHz
3	867.1	125kHz
4	867.3	125kHz
5	867.5	125kHz
6	867.7	125kHz
7	867.9	125kHz
8	868.3	250kHz
9	868.8	125kHz

LoRa Server

Debian machine with the whole ChirpStack server stack:

• ChirpStack Gateway Bridge
• ChirpStack Network Server
• ChirpStack Application Server

Note: ChirpStack is configured with the default configuration files used in this guide.

End device code

import time
import pycom
import binascii
import socket

from network import LoRa

if __name__ == "__main__":
    pycom.heartbeat(False)
    pycom.rgbled(0x7f0000)

    lora = LoRa(mode=LoRa.LORAWAN, region=LoRa.EU868, device_class=LoRa.CLASS_A, public=1, adr=0, tx_retries=0)

    for i in range(0, 16):
        lora.remove_channel(i)

    lora.add_channel(index=0, frequency=868100000, dr_min=0, dr_max=5)
    lora.add_channel(index=1, frequency=868300000, dr_min=0, dr_max=5)
    lora.add_channel(index=2, frequency=868500000, dr_min=0, dr_max=5)
    lora.add_channel(index=3, frequency=867100000, dr_min=0, dr_max=5)
    lora.add_channel(index=4, frequency=867300000, dr_min=0, dr_max=5)
    lora.add_channel(index=5, frequency=867500000, dr_min=0, dr_max=5)
    lora.add_channel(index=6, frequency=867700000, dr_min=0, dr_max=5)
    lora.add_channel(index=7, frequency=867900000, dr_min=0, dr_max=5)
    lora.add_channel(index=8, frequency=868300000, dr_min=0, dr_max=5)
    lora.add_channel(index=9, frequency=868800000, dr_min=0, dr_max=5)

    dev_eui = binascii.unhexlify('0000000000000000')
    app_eui = binascii.unhexlify('0000000000000000')
    app_key = binascii.unhexlify('00000000000000000000000000000000')

    lora.join(activation=LoRa.OTAA, auth=(dev_eui, app_eui, app_key), timeout=0, dr=0)

    def rxcb(arg):
        print("RX callback received: {}".format(arg))

    lora.callback(LoRa.RX_PACKET_EVENT, rxcb)

    while not lora.has_joined():
        pycom.rgbled(0x7f7f00)
        time.sleep(5)
        print("joining...")

    print("joined")
    pycom.rgbled(0x007f00)

    # create a LoRa socket
    sock = socket.socket(socket.AF_LORA, socket.SOCK_RAW)

    # set the LoRaWAN data rate
    sock.setsockopt(socket.SOL_LORA, socket.SO_DR, 5)

    # make the socket blocking
    # (waits for the data to be sent and for the 2 receive windows to expire)
    sock.setblocking(True)

    # send some data
    sock.send(bytes([0x01, 0x02, 0x03]))

    # make the socket non-blocking
    # (because if there's no data received it will block forever...)
    sock.setblocking(False)

    # get any data received (if any...)
    data = sock.recv(64)
    print(data)

Problem

Observed behaviour

Some JoinRequest messages appear to get "lost" as they do not appear in the gateway traffic log. To my understanding the FiPy sends a new JoinRequest message every ~10 seconds.

It a significant amount of JoinAccept messages to successfully join the network and, once joined, very rarely receives a scheduled downlink message. All JoinRequest messages received by the gateway are successfully sent to & authenticated by the LoRa server, the JoinAccept response however, does not reach the end device.

Advancements

Some forum threads have mentioned that similar issues may be related to an incompatibility with the gateway's channels. Therefore, the FiPy's code has been modified to remove the default channel configuration and replace it with the gateway's. This has resulted in a significant improvement with the JoinRequest messages being lost.

Hypothesis

Since downlinks only appear to work every few hundred attempts there might be a problem with the Rx window timing.
I would expect this to be thouroughly tested in the LoRa firmware as the issue would be present on all Pycom devices.

Is there any way to debug this issue?

Any other hypothesis/solutions are welcome.
Thank you