LoRaWAN FUOTA - Fragment size calculation

Hi. The fragment size you should use depends on your device: the size of the image, the number of fragments, and the fragment redundancy you will support. And what size payload you can carry in your messages (based on RF region and data rate). Do you know what's implemented in your device? For device firmwares based on LoRaMAC-node, the fragment settings are defined here: https://github.com/Lora-net/LoRaMac-node/blob/dcbcfb329b4a343ab007bc19ac43a8dc952b3354/src/apps/LoRaMac/common/LmHandler/packages/FragDecoder.h#L38-L57

And these settings are checked when AWS IoT Core for LoRaWAN tries to setup the fragmentation session: https://github.com/Lora-net/LoRaMac-node/blob/dcbcfb329b4a343ab007bc19ac43a8dc952b3354/src/apps/LoRaMac/common/LmHandler/packages/LmhpFragmentation.c#L355-L369

So if you suffer a "Not enough memory" error, it means AWS IoT Core is proposing a fragment size or number of fragments that exceeds what your device can handle. Hence you need to create the FUOTA task with a smaller fragment size, and/or change the settings on your device.

A bigger (maximum) fragment size will mean a quicker file transfer. But also more memory consumed because the redundancy setting will reserve extra memory for lost fragments/packets.