Simulcast mishmash
I find all this discussion about simulcast issues quite interesting. I used to work for a large company that was designing some of the first DAMPS cellular phones. We had a really brilliant individual who concentrated almost exclusively on what we called “the equalizer”. At the time, at first, I was a bit unclear what this beast was but I later got the gist of it (I think). My background is primarily RF hardware with some analog and digital hardware thrown in the mix, DSP algorithms are not my forte’, unfortunately. This “equalizer guy” was a DSP guru – all mathematics pretty much. And that was what the equalizer was – an algorithm with a very specific purpose. It was (as best my limited brain could comprehend) supposed to overcome the problems with decoding digitally modulated RF in heavy multipath situations. So, in a nutshell, (assuming I got this right – and I am sure I am oversimplifying) the equalizer was supposed to look at several simultaneously received identical but out of phase signals and sort it all out to get one nicely decoded data stream which was then passed on to the vocoder. It wasn’t the same as basic error correction in the sense that (beware, I am in my ignorance zone here) it didn’t just compensate for dropped or flipped bits; it actually looked at the data and tried to overcome specifically the problems associated with receiving multiple otherwise identical but with randomly or pseudo-randomly varying phase and amplitude components. From what I can remember the equalizer did this at the baseband level (not audio but at the IF after running through the I/Q demodulator and subsequent demodulation circuitry – the raw bits right after being extracted from the RF carrier which includes all of the parity, framing, etc. overhead) and I really can’t get my head around how that could be done (it’s bits, ok – how do you tell that they came from some out-of-phase mishmash at the RF level and how do you extract the “correct” pattern from same?) but, as I said, I am pretty much an idiot in this area.
Thinking back, now I wonder if maybe they were digitizing the downconverted IF, either before or after the I/Q demodulator, and running that through a dedicated DSP engine which ran algorithms on the digitized IF signal to detect and compensate for phasing issues and then converting the IF data back to (cleaned of multipath distortion) analog and then running it through the demodulation circuitry – I can sort of half see how that could be a possibility. But, for some reason, I am still thinking he did this using the raw bits that were finally extracted from the RF/IF carrier and, if so, it might as well be magic to me. And I do think that he had a dedicated DSP engine to handle all this either way.
My point here is that though cell phones didn’t deal so much with simulcast issues they did deal with some heavy multipath problems and the type of digital modulation we used (PI/4 DQPSK) was very sensitive to phasing and amplitude issues otherwise easily overlooked when using analog FM. And this “equalizer guy” was a really important dude for us! My thinking is that somebody similar or a team of same is on all the Motorola, M/A-Com, etc. design teams and probably puts a lot of effort into the “equalization” of the simulcast received signals. Unfortunately the consumer scanner manufacturers probably don’t have and can’t really afford to have the same so, well, hence the problem. Basic error correction is one thing but CQPSK is pretty sensitive to phasing issues and probably something pretty dedicated to that specific issue needs to be present in the final receiver design to overcome such issues which stem from both multipath and simulcast corrupted signals.
BTW – this guy had a pretty decent dry sense of humor and I recall one time after a rather long and arduous day when we were discussing digital vs. analog on RF when he said: “Well, if we had started with digital we’d be going to FM now!” I know he was joking buuuuut…
-Mike
