KH6TY wrote:
>
>
> Alan,
>
> What happens, for example, if 100 DSSS stations are all on at the same
> time, on the same beginning and ending frequencies, because everyone
> assumes his presence at any one frequency is too short to be noticed?
>
> Will they interfere with each other, or will they collectively
> interfere with other users of the frequency, such as SSB stations?
All valid questions. You know the answer to most of them.
DSSS without CDMA, hold off, etc would neither work or be desired beyond
a certain "loading" (number of users).
> When you say "multiple" how many would that be with a spreading factor
> of 100?
Like you, I'd have to dig out the math, make some assumptions. There is
an answer, and it's greater than 1, and less than 100 for sure. :-)
Based on very rough math, and fuzzy assumptions, my initial calcs were
that it would take over 10 simultaneous DSSS to be detectable at psk
data rates with a spreading factor of 100.
More than that to be interference to a typical SSB signal. Remember,
just because a chip wanders into an SSB bandwidth slot does not mean it
will interfere with an SSB signal due to SSB filtering, response curves,
etc. That bit in the bottom 50 hz of an SSB slot will not be detected.
Likewise those in the "guard bands" between typical SSB signal spacing.
Likewise, since the energy is widely distributed there are no
significant sidebands that are much easier to detect/hear and become
interference.
But that was just a concept thrown out to make people realize that all
DSSS is not like ROS. Nor like the high data rate strong signal DSSS
seen on higher bands.
We need to separate the concept from the flawed implementation, that's
my point. I do believe in the future we will want to revisit DSSS with
CDMA as an alternative to the chaos of RTTY/WINMOR/P3/ALE/SSTV/whatever
we have now. Not to the exclusion of legacy weak signal modes. But as a
more efficient way to maximize throughput (users * data of any type) of
the very limited HF resource we have.
We'd have to do the math, but I'm pretty confident that for any chunk of
bandwidth (say, 20khz or greater) you could support more simultaneous
users at a given data rate with DSSS or similar wideband mode with CDMA
than the same chunk with SSB afsk modems. It's simply more efficient,
does not have the guard band issues, etc.
It will never happen in our lifetimes due to the hold that legacy modes
have. With some justification. But that does not mean we should paint
ourselves into a corner where it could never be discussed, much less
proposed.
>
> It seems to me that enough chips randomly spread over the band (by
> enough multiple stations) could also raise the general noise level,
> even if they were very weak. This was a concern of weak signal
operators.
This is true and valid for weak signal areas. It's not for strong signal
modes. Even including SSB, and you could do much in between FM channels
with minimal impact to FM qso's. There's nothing that states DSSS has to
be evenly spread across it's range, though it helps with processor gain.
You could have a sequence that only hit the guard bands between 10m FM
channels for example.
> For example, suppose it was decided to let multiple DSSS stations span
> the whole length of the 20m phone band so there was sufficient
> spreading. How many on the air at one time would it take to create
> noticeable QRM to SSB phone stations, or raise the noise background if
> they were on VHF?
There would have to be CDMA of some form. But the answer is still more
than one, less than many. You are still only using the net bandwidth
even when spread. IE: You are not truly using 50khz just because the
signal is spread across that range. Because you are not using it
exclusively. It's only when many, many users were active simultaneously
that it would reach interference levels. Likewise, the SSB signals would
surface as bit errors to the DSSS, so throughput would go down when it
was crowded with SSB signals.
> I ask this because I believe that the question arose several years ago
> regarding allowing hi-speed multimedia to operate over 20 kHz on 20m,
> which may be OK for one station, but what happens if there are 100
> stations doing the same thing?
High speed wide band is different than widely spread DSSS. It would
absolutely interfere with anything in that bandwidth, sounding like
white noise.
But similar questions pop up. Given 20 khz would typically handle 5-6
SSB signals with guardbands, could you beat the throughput with that one
20khz signal? Add CDMA, and would that channel carry more traffic than
the 5-6 SSB signals with P3? (Currently the ham legal throughput leader)
There are tradeoffs with multi-path, fading, etc. long/short symbol
lengths. None are perfect. But our current approach is not either. :-)
I'm not in favor of plopping hi-fi audio or multimedia wide band signals
in 20m SSB space. But do I think there should be options to experiment
(tightly controlled) with a CDMA approach on our wider/higher bands?
Sure. Even with great prop in a contest there is often large segments of
10m, and even 15m idle. What if we had something in place for the
upcoming sunspot cycle?
But as I saw the recently posted list of winmor freqs go by, and
mentally overlaid them with the similar P3 frequency list, really made
me think about optimizing aggregate throughput (users * data/time).
It's not clear to me why winmor/Pactor does not stack RMS's on channels
with geographic diversity to do some rudimentary CDMA. At any given
time, it's very unlikely that a single user can access all the RMS's,
due to prop/coverage. So there should be room for frequency reuse. We
keep ALE to 1 SSB width slot in each band for 99% of the activity. That
would not be practical for the RMS's due to volume, but I don't
understand why there is not more CDMA thinking utilized in our new
networks. Likewise, why do I have to call a specific RMS, could I not
just make a call for an RMS, and let the network determine which should
respond back?
> I studied communications theory and auto-correlation functions, etc.,
> 50 years ago in college, but unfortunately I don't remember much of it
> at all!
You and I both. I worked extensively with high speed data on RF back in
the 80's & 90's with the WA4DSY 56k modem. So some of the
theory/practical aspects are still fresh. And I've revisited the theory
a bit during the whole ROS fiasco.
While the WA4DSY modem was not DSSS, many of the other areas
(convolutional encoding, FEC, etc) still applied. As did CDMA. And same
logic to users * data for a given bandwidth applied. The same spectrum
used for 56k significantly beat multiple 9600 links using the same
bandwidth. Likewise, it was a direct IF modem similar to DSSS, not audio
based. While it was not a need, the core modem probably could have been
used to implement DSSS for a weak signal, lower data rate approach. Just
change the encoders.
OK, enough on DSSS. we can go back to PSK & RTTY. or JT* :-)
Have fun,
Alan
km4ba
