For radio heads: Decoding the LoRa PHY

Heater.

Just thought anyone into radio. modulation schemes, encoding, etc might be interested in seeing how the new LoRa wireless standard does it.

None of it is documented so these guys just listened to it and reverse engineered it:

If nothing else it's a brilliant piece of reverse engineering. The modulation is not like anything I heard of before.

Anyone out there care speculate why it is done that way?

Phil Pilgrim (PhiPi)

Brilliant presentation. I was able to follow about 3/4 of it. I do think that by implementing a DPLL on the receiving end during the sync frame, one could eliminate the multiple FFTs and just use Goertzel filters tuned to the eight candidate demod frequencies. A Prop could do that.

-Phil

Heater.

I love these kind of reverse engineering stories. CCC has hundreds of them! As does DEFCON.

A young girl round her found herself listening to some interesting signal. She proceeded to analyse it. Turned out to be the wireless link to the displays on bus stops that show arrival times. Soon she had that decoded...

Anyway, I guess if you have some SDR hardware and a PC then hitting the problem with the sledge hammer of FFTs is an easy way to proceed. I can't imagine that's what the actual LoRa chips do, given they are supposed to be tiny and very low power.

Think I have to watch that again to try and understand it better. I vaguely get an idea of the spread spectrum "chirp" being a way to deal with reflections and multi-path hops. Then the "chirp", "chirp", "chirp", start sequence makes sense. Then I get lost ...

All of which reinforces my desire to get hold of some LoRa modules to play with.

pmrobert

Heater, the better marine depth finders / bottom mappers / fish finders use the ultrasonic CHIRP concept, obviously in much lower freqs than RF. I acquired one last year and it has much better resolution than any of my previous devices. See http://www.thehulltruth.com/marine-electronics-forum/506221-chirp-sounders-fish-finder-explained.html#b for a Q&D explanation that made the concept clear to this guy anyhow...

Mike R...

Addendum: If in fact the chirp referred to in the LoRa stuff is similar in any way!

Heater.

That's an interesting write up on the modern state of ultrasonic ranging. I'm sure the chirp concepts apply there as they do radar and LoRa.

After all a sonar is a radar in sound. A radar is a transmitter and a receiver. The problems of bandwidth/power/noise/range etc are common to all of them.

Looks like it works in the LoRa case. With it's chirp modulation it achieves a range in kilometers as opposed to the few 10s of meters of WIFI etc.

Seems chirp modulation was first patented in the 1950's. I wonder what the LoRA patent adds to that.

Heater.

So Phil, with all this talk of chirp, sonar, range, etc, when are we going to see the Propeller using chirps with ping sensors to increase their range and resolution?

Seairth

http://hackaday.com/2017/01/06/ces17-arduino-unveils-lora-modules-for-the-internet-of-things/

Heater.

I'm looking forward to a Parallax PropelLoRa board this year

tonyp12

A usb dongle if you want to spy on air signals
https://www.disk91.com/2017/technology/hardware/take-a-look-at-sigfox-radio-signals-with-airspy-usb-software-defined-radio/

I have started to test sigfox on a CC1120 dev board and it never misses a message even if tower is miles away.
U.S is still just starting up, FCC requires 902MHz to be 600bps and most modules are still made for euro 868mhz 100bps

Yes you can only send 12bytes, 140x a day. But serial, timestamp and rough location is added for free.

I guess location will be more precise with more towers to triangulate and software done by the French,
as all towers receive the message and they filter out duplicates they probably could detect nanosecond time difference too.

Phil Pilgrim (PhiPi)

heater wrote:

So Phil, with all this talk of chirp, sonar, range, etc, when are we going to see the Propeller using chirps with ping sensors to increase their range and resolution?

I'd like to try it, but finding broadband transducers could be a challenge.

-Phil

Heater.

I did think about that Phil. I have no idea how "not broadband" a typical transducer is. Perhaps something can be done with them.

Phil Pilgrim (PhiPi)

A typical cheap piezo transducer, like those use in the Ping))), will have a very high Q -- probably on the order of 100 or more. They ring mercilessly. That's the reason for the dead-time requirement between sending a pulse and listening for a response.

-Phil