[superkuh]

You're wrong. It's a mistake to treat terrain and foliage as the same thing. I have extensive experience with 3 completely different, narrowband and wideband, 902-928 MHz radio data systems as well as just playing around in the range with my software defined radios (hackrf, rtlsdr, etc).

902-928 has almost no advantage over 2.4 GHz when it comes to line of sight issues with terrain. In fact, it's more problematic due to the increased size of the fresnel zone. Sure, it does better through trees but a slight rise of ground is just as much of a problem for 915 and 2400. The freq here isn't helping much. It's the chirp and LORA modulation helping the link budget. But no line of sight is no line of sight.

[generatorguy]

It sounds like you agree 900 MHz will be much better than 2.4 GHz in this non line of sight application so long as the obstructions are trees and foliage and not terrain.

I would expect to be able to send a few bytes between two stations maybe 400m apart through non-line of sight forest with 900 MHz and both stations at the same elevation. Certainly the grandparent quoting miles would be for mountain top to mountain top.

[madengr]

This is all assuming they use antennas that are optimized for the application. If they use a “chip” antenna, performance will suck.

[generatorguy]

the video on their web page shows an external antenna, looks like a dipole.

[bigiain]

This stack overflow question has some nice pictures of the LoRa "chirps". They use rising/falling frequency modulation to allow operation at startlingly low signal to noise ratios.

(Also, the hardware they're using is fairly generic, and pretty much all the vendors selling them offer them in 915/920MHz and also 868MHz and 433MHz variants. I've seen claims of over 10km range with 433MHz LoRa gear without special antennas or clear line-of-sight...)

[bigiain]

Just noticed I'd left out the link to the SO question I mentioned (and it's too long ago to let me edit that post):

https://electronics.stackexchange.com/questions/278192/under...

[dfox]

LoRa has very peculiar modulation scheme that in the end is not CDMA-ish spread spectrum but an interesting way how to extend straight FSK into spread spectrum modulation with respectable processing gain and interference rejection.

[bigiain]

I recall reading that part of the design was so that not only was it good at rejecting interference, but that it also caused little interference to other users of the spectrum.

If someone else's ISM radio is using a specific frequency (perhaps with CDMA or TDMA), a "chirp" that's spears over about a hundred kHz and -20db or so down in the noise is unlikely to bother them, but is quite useable/reliable for the LORA gear to detect.

[jki275]

Not sure why this was downvoted, it's exactly correct.