Re: [Discuss-gnuradio] LoRa implementation

[Alberto Trentadue]

Hello Marcus

Thanks for your reply. I'll look into your advice, sounds doable.

Indeed in the past days I was thinking how to make good use on my USRP
+WBX to setup a testbench for smart applications using The Things
Network (thethingsnetwork.org) with Arduino UNO.

I will post my progresses.

Cheers
Alberto


---------Original message---------------------

Date: Sat, 24 Oct 2015 13:43:40 +0200
From: Marcus Muller <address@hidden>

Hi Alberto,

I think that sounds like a fun project. The LoRaWan website is a little
vague till you find the specs [1], and even they do not cite a full PHY
document, but a quick patent search helped me find out that it's a Chirp
spread spectrum modulation, which led me to Semtech's Application Note
1200.22 .
Basically, take your symbol-to-be-transmitted, spread it by transforming
it to a series of up- or downchirps, for the LoRa modulation (there's
also a GFSK option for some bands as per the LoRaWan spec
So: First approach would probably be to take a good-SNR RX signal,
filter it to the channel BW; apply "quadrature demod"; a positive value
of phase difference would mean "up-chirp", a negative "down-chirp".
Slice as appropriate.
Altenratively: Apply correlation with chirp, detect whether same
"direction" or reversed (maybe something like: signal->FFT (lenght=chirp
duration*f_sample)->multiply_const (const = conj(dft(ideal chirp)),
vlen=fftlen) ->sum of vector->slicer))

Cheers,
Marcus