@tango156157, I sometimes get similar errors while compiling the sketches. Mostly because I am still using the MySensors 1.5 library for my legacy network, 2.1.0b for the RFM69 network (working RSSI report) and 2.1.1 for the LoRa network.
Mostly when dealing with MySensors 1.5 and/or older sensors libraries, I find the newer Arduino version to pop all sort of errors. But then I fire up one of the older Arduino IDE versions I have, 1.6.5, 1.6.9 and so on. Eventually one of them compiles the sketch without errors while the newer version don't work. This is far from being a solid solution for these errors, but it gets the job done.
You should try that too, especially if you're using older libraries for your sensors.
You can find them here: https://www.arduino.cc/en/Main/OldSoftwareReleases#previous

@mfalkvidd said in Transport overhead optimization:

Could you describe the advantages of using a rfm96 instead of rfm69, when not using LoRa mode?

Then I don't understand your statement.

OK, trying to make a sense of those tables now, but my main question remains: Is RFM95_BW31_25CR48SF512 a LoRa mode or not? Because this is what I'm currently using.

So what you're telling me is that I'm not using the RFM96 in LoRa mode? I now seem really dumb, but in all fairness, after a lot of digging, I only found these modem configrations available:

* | CONFIG                    | BW    | CR  | SF   | Comment
* |---------------------------|-------|-----|------|-----------------------------
* | RFM95_BW125CR45SF128      | 125   | 4/5 | 128  | Default, medium range
* | RFM95_BW500CR45SF128      | 500   | 4/5 | 128  | Fast, short range
* | RFM95_BW31_25CR48SF512    | 31.25 | 4/8 | 512  | Slow, long range
* | RFM95_BW125CR48SF4096     | 125   | 4/8 | 4096 | Slow, long range

I chose RFM95_BW31_25CR48SF512 after some extensive trial and error range tests.

If I'm not currently using LoRa, how can I actually switch to that?

I couldn't not notice the large overhead associated with the MySensors transport protocol. You see, as discussed before, the issue at hand is not using a coin cell for 3 years instead of 2, but it's about the RFM69 and RFM95/RFM96 modules which are capable of using really low bitrates or special modulations.
Eg. I had a 14250 battery that drained within a month with an RFM96 sending data every 5 minutes using RFM95_BW31_25CR48SF512 which was sending only one sensor value. This happened because unlike RF24 which can have a total awake cycle of 100-150ms, the RFM96 had an awake cycle of about 5000ms. It took it 5 seconds to send the single sensor value.
RFM69 is another case in which a lower bitrate -> longer range -> longer TX time. I went as low as 300bps only to see that a single packet is sent in about 1500ms. That's way too much for a battery powered node.

Some of us need long range sensors and we don't want a constant power messy setup, or the complex circuitry and increased cost associated with adding "UPS" like features to our boards.

My RFM96 node now sends a ping every 3h and the RFM96 nodes run in 9600bps so that I wouldn't have to replace those 14250 batteries monthly and actually use them for the intended life-cycle, that's more than 3 years.

The alternative is to give up MySensors altogether and use third party libraries, concatenate the messages as raw data then parse them on the remote gateway based on predefined rule set. But although this gets the job done, it's not a really good idea to mix various libraries and properly maintain the whole system in the long run.

Well, I'm glad that the posting got some traction and some very interesting ideas were shared. While the bits and pieces of information were all out there, your argument helped drawing some conclusions.

1. LoRa is NOT an option for HA. alexsh1 explained one scenario and I'm actually using LoRa (not LoRaWAN) in my building, but both cases make use of LoRa only because it's a new cool technology. @alexsh1 your friend could properly cover his home with 2G/3G signal either for free, by asking the provider to improve the coverage in that area, or by buying a GSM repeater. And I could ask a neighbor to share his WiFi connection so an ESP8266 module could do the job.
2. The protocol addresses very specific segments where a great urban coverage is required and for that to happen no node should exceed a radio power of over 100mW.
   There is another case where everything was traded off for the sake of range, the ham radio JT65 protocol. I was able to successfully transmit a signal from Romania to Brazil using 5W RF power and a 1m diameter magnetic loop antenna. It's great for long distance, narrow bandwidth (200Hz wide channel), low power but it sends data at a "whooping" speed of 13 characters per 50 sec, during which it draws about 1.8A from a 12V battery (21.6Wh).
3. My original idea of LoRa <-> RFM69/RF24 is not feasible unless, some serious downsampling is involved, as buffering the raw data before sending it out is really useless because LoRa doesn't have a serious overhead that would be addressed by concatenating larger chunks of data in a single packets.

So at the end of the day, it really leaves us with a couple of applicable scenarios when taking public LoRaWAN meshes into consideration. Smart meters and perhaps security devices which only have to send a daily keepalive ping and if ever needed, tripped sensor alerts, provided that they would be immune to jammers. I'm not taking into consideration the close range stations where a daily 30s air time would suffice, because this defeats the purpose of LongRange. Other than that I see no real use of LoRaWAN, but feel free to share your ideas, perhaps I'm missing something.

And after apparently trashing both LoRa and LoRaWAN I will only say that I can barely wait to get myself a LoRaWAN gateway and set it up in my area

@alexsh1 said in Any RFM95 user reports?:

@gohan in a highly populated area Lora is better from my perspective. However, as already mentioned the cost is high (GW - €300, node - €55+ etc). Plus one is expected to develop

And why is that? Dense population = likely more users = frequent gw congestions.

Indeed, I see LoRa as an option for either experimenting, or as a solid alternative where there's no other option. Because at the end of the day, those who'll require a reliable transport layer will not resort to the open bands but use 2G/3G modems or even radio modems or satellite uplinks. While the LoRaWAN users will have to cope with the drastic FAP limitations, or use point-to-point systems for setups that allows no alternative, like I do.

I see, but the end nodes shouldn't necessary route their data in an a synchronous manner. The gateway can buffer the data and commit the reading asynchronously over to the LoRaWAN network once every 5 minutes. Sensor data downsampling could be performed by the gateway in order to shorten the TX window to the utmost minimum. Downsampling and/or a heavy compression algorithm could help.
On the other hand, I do realize that all these feature could lead to a fairly complex solution that's going to be so complicated to use that's never gonna get any traction.

I'm planning to get myself one of those and set it up in my neighborhood, however at least for the time being I find it a bit useless since I don't know of any Home <-> LoRaWAN gateway that would effectively push the end nodes data over the LoRa network and up to a hub. Moreover, if that gateway would exist, then there should be some considerations regarding the way the LoRaWAN gateway handles packets, eg. authentication, message signing, encryption, ACL or rate limiting. This would enforce a private communication channel, free of any abuser that would hammer down the gateway by sending readings every 5 sec. from a set of couple of dozen sensors, congesting the gateway.
It may be that some manufactures such as Cisco have already considered these but I'm not aware of that, and testing their solutions is rather cost prohibitive.