Problems with RX when connected to LoRaWAN gateway

[jtisaacs]

I have a RPi 3 and the Dragino LoRa / GPS Hat version 1.4. I am using modified version of the periodic application to connect to both TTN and Comcast MQCentral with either a Laird Sentrius RG191(https://assets.lairdtech.com/home/brandworld/files/Sentrius%20RG1xx%20User%20Guide%20v2_2.pdf) or a Comcast Area 8C (https://store.machineq.com/store/products/area-8c) gateway. In both cases I have no problem Joining with OTAA nor transmitting messages periodically as the application specifies.

Where I run into problems is when keeping the application alive for long periods of time. It seems as though I am not ever receiving a downlink message from the gateway. This results in my end-node eventually increasing the SF until the max and holding it there. Meanwhile the gateway is sending MAC commands requesting ADR to a lower SF. Also, eventually the AckRequested bit goes high since I haven't heard from the gateway.

I would like to be able to respond to the MAC commands from the gateway and be at the optimal DR/SF, but this doesn't seem to be working with this system as is. I noticed in your blog post that you said,"The examples work, however I would like to do some more analysis and fine tuning regarding the timing of the receive window, and I like to add more debug output to understand what is going on in the network in regards of the LoRaWAN protocol." I am just wondering if you have ideas of where I should start in debugging this issue. In the EV_TXCOMPLETE case I am checking LMIC.dataLen, but the value is always 0.

As a related question: How do people recognize that they are no longer connected to the gateway and that they should try joining again? I was considering putting a counter on the number of TX_COMPLETE with LMIC_dataLen 0 and then resetting after some number of transmissions.