On 20180409 06:17, Support@... wrote:
/RE: /The basic problem I have with that is presuming the 20 MHz means 50 ns when you get I and Q at 10 MHz. I alone and Q alone don't give you a meaningful time.
The Airspy hardware/firmware does not generate or use I/Q. The ADC in the hardware does not and cannot sample in quadrature. There is a single 12 bit ADC which is set to sample at 20MSPS/20Mhz/50nS real. It can be set to sample at other rates, but 20MSPS is the most that you can reliably push out over an USB2 interface. Youssef has now confirmed the decoder runs 20MSPS real, so I don't have a dog in the "is 20MSPS real to 10MSPS I/Q is possible" fight. However, if it's not possible, then how do you think SDRSharp works @ 10MSPS I/Q with an Airspy?
RE : Argue with Youssef. Are you absolutely certain that the multiple packets do not represent multiple transmissions?
Yes - I'm as certain as Schrodinger and Heisenberg will allow anybody to be about anything. Anyhow, Youssef has kindly supplied me some extra information which will allow me to investigate. If I find anything useful then I'll feed it back to him so he can decide whether to incorporate it in airspy-adsb.
After sleeping on it there is a potential for gaining a better guess for what the signal really is. Perform a bitwise vote. If two of three agree pick that value for the bit. If two of four agree (hence two of four disagree) discard the transmission. The next level would be comparison to historical transmissions. Consider the N decodes as your redundancy factor to use when deciding apparent truth. That is one of the oldest error correction schemes around.
I'd at least give the concept a try before crying too hard. Youssef's decoding technique offers you the chance to perform the voting. Maybe it will do some good.