I had an interesting consult with a major RF chip manufacturer about what it would take to bring this sort of capability in house. They were surprised that I had some pretty exact numbers, and I pointed out to them it had already been done and this was the project that came out of it. Sadly they never told me whether or not they built one for themselves.
One of the things about software radio that is similar to my experiences in robotics is that when things get "real" they get weird. In robotics it is noise in the sensors, environmental effects, Etc. In radios it is the way things reflect and refract different radio signals, as well as interference both intentional and unintentional. I built a phase based modulation scheme and it worked perfectly using a ADALM-PLUTO with its Tx connected to its Rx, it needed tweaking when I had it go between two PLUTOs (clock skew, timing differences). And it still doesn't work going through the air between a PLUTO and a LimeSDR :-). The "real world" is tough!
Hi Chuck, I was the System / Proj manager on this. I had always wondered why folks didn't just simulate the whole thing in software; then I realized, although doing it with Ettus Research SDRs + moby networking + hella big Dell box to coordinate it all was necessary because --- the digital went out to the SDR, and out of the SDR came RF *at the frequencies of interest*. Although all of it was cabled RF (using cables I spec'd and acquired), it could have just as easily been put to antennas and run 'over the air'. (except of course that environment is harder to completely control).
The artificial interferers were able to be consistent across the entire test. The successful entries had no coordination, and still maintained spectacular bps throughput.
This work was done several years ago now, so I don't know if the winning algorithms are now deployed in an 'real' on-the-air system.
This was a very technically challenging project, pretty much at the limit at what could be done at that time. But it worked!
RF always does weird stuff ;-) There's a path, driving through the mountains up around Tyndrum, where you can hear a radio repeater some 50 miles away near Stirling despite there being 3000 foot high mountains in the way, and as clearly as if you were a mile from it.
It must just bounce off stuff just right, or something.
One of the things about software radio that is similar to my experiences in robotics is that when things get "real" they get weird. In robotics it is noise in the sensors, environmental effects, Etc. In radios it is the way things reflect and refract different radio signals, as well as interference both intentional and unintentional. I built a phase based modulation scheme and it worked perfectly using a ADALM-PLUTO with its Tx connected to its Rx, it needed tweaking when I had it go between two PLUTOs (clock skew, timing differences). And it still doesn't work going through the air between a PLUTO and a LimeSDR :-). The "real world" is tough!