Data Rate of LoRa experiments



  • hello,

    To test the lora data rates with respect to distance, i am sending packets of some size say 64 bytes and use a timer to record over a minute how many packets I receive at the other side. I repeat this with other distances. Is this a good approach? Also is there any suggestions on what else I could do?



  • @snehasg96 said in Data Rate of LoRa experiments:

    @bmarkus Thank you! So if I understood you correctly, sending a packet of payload say less than 64 bytes should be good? Also do you mean i should include the SF to a value between 6-12 in my code?

    Appreciate your feedback.

    I say use payload which is within the allowed size and represent a typical size let say around 8-10 bytes. Use the lowest DR (highest SF) which is DR0/SF12 for EU868 and add other DR, DR1...DR5 to see how distance varies by DR. SF6 is not valid for SX127x device chipsets and for Gateways.

    Please note, LoRa packets can be travel for a long distance randomly, but do not expect that long distance you see for few packets will work in a production environment. Be conservative and base your decision on larger number of samples.



  • @bmarkus Thank you! So if I understood you correctly, sending a packet of payload say less than 64 bytes should be good? Also do you mean i should include the SF to a value between 6-12 in my code?

    Appreciate your feedback.



  • @marcozennaro Interesting tools and result, though if I'm not mistaken, you have exceeded the allowed EIRP by quite some margin (for sure when you used 23 dB gain antennas, and probably when you reverted to the 6 dB antenna, depending on the actual channel used), did you not? I didn't read through the whole paper so I may have missed something.

    Another set of very interesting tools for planning: http://www.ve2dbe.com/rmonline.html

    The UI is really not the most convenient (or beautiful!) in the world, but it has lots of tools, including planning a link between two sites taking account terrain, curvature, etc, coverage maps based on terrain, and more.



  • @marcozennaro Thats awesome. Good work!



  • You can easily check line of sight and freznel zone using our Telegram Bot described here:

    https://arxiv.org/abs/1610.09248

    We used it to achieve a LoRa 300km link (http://wireless.ictp.it/Papers/Mobicom.pdf).

    Enjoy!



  • @jcaron Sadly this wasnt me.

    Can confirm no line of sight and certainly wasnt > 200ft up a tower (Its a car mounted test setup. the local area is pretty damned flat and at basically sea level)

    and that was very weather dependent. the person in question has only managed to replicate it a few times. all in similar weather conditions.

    When I speak to the person in question next ill ask about the entry on: 13/3/2018 as something does seem a little weird there.

    We are also pretty sure this wouldnt be able to happen above land. and that having mostly nothing but empty sea between Node -> Gateway is a big factor.



  • @paul-thornton That's quite impressive, nearly defying the laws of physics...

    You definitely don't have line of sight between those places due to the curvature of the Earth, even with both antennas set on top of 200 m towers. Over 200 km at SF7, with no line of sight... Free space loss alone (with line of sight) is about 137 dB.

    You probably had very very good weather conditions with air layers giving you a very effective waveguide...

    And that's not the most intriguing one... on 13/3/2018 you had a 290 km link (free space loss: 140 dB) with an RSSI at -75 dBm. I think on that one ttnmapper got the data quite wrong (either that or you were pumping waaaaaaay more power than what is allowed).



  • As mentioned 64 bytes is large for a lora packet most are a few bytes. if there's a couple of gateways locally you may also find it skips your local gateway and get's picked up miles away.

    We've had range tests running for one of the things network groups I help with, and instead of hitting the local gateways Its picked up in another country some 200+km away something to consider if you get strange results.

    alt text



  • @snehasg96 In such test you must include lowest DR. Its definition (mean SF) is depending on region, but 64 bytes payload is usually more than allowed (e.g. max payload size is 51 bytes for DR0 in EU868, for US915 it is even less) and not needed for the test.


 

Pycom on Twitter