spreading factor in LoRa PHY (CSS)

Question

I wonder how the spreading factor (SF) is defined in LoRa PHY. Specifically, in other spread spectrum techniques (e.g DS, FH), we have:

W = R * SF

W is the (spread) bandwidth, and R is the symbol rate.

In LoRa, we define SF as SF=K (no. of bits per symbol) and we have:

W = df * M df = R M = log2(K)

where df is the frequency separation, and M is the no. of symbols. So, in LoRa, it could be better to define SF as SF=M (=2^K) instead of K.

Thanks in advance for the time you generously spare for me!

Answer 1

I asked it directly from the father of LoRa, Olivier Seller. He said:

"Indeed what we name SF is actually the base 2 logarithm of the spreading factor. This came pretty naturally when we designed LoRa, as SF12 was shorter to spell than SF4096. By the way, this is not exactly the number of bits per symbol. For SF11 we only transmit 9 bits, and for SF12 10 bits. A reduced modulation alphabet is used to make synchronization and tracking easier."