[Date Prev][Date Next][Thread Prev][Thread Next] - [Date Index][Thread Index][Author Index]

Re: Digital DX pileups

>The CDMA technology commonly used in cellular networks can only
>support about 30 simultaneous users on a single RF channel.  To do

Isn't 30 still better than 1? :-) (We actually got the number up to 61
one night in an early field test, but that was in an isolated sector
without any interference from neighboring sectors).

>that, each user's transmit power is rigidly controlled by the network,
>such that each of the 30 users has the same signal strength at the
>cell site.  The capacity (max. number of simultaneous users) goes way

Correct. This power control scheme was Qualcomm's major innovation,
and it is indeed the best way to build a CDMA cellular network when
you control the whole system design.

But it's not the only way. There are iterative decoding schemes for
CDMA where you first demodulate the strongest signal, then re-modulate
it and subtract it from the total received signal. Then you look for
the 2nd strongest signal, subtract that off and repeat until there's
nothing left.

We don't do this because it doesn't buy much when you already have
tight power control. But it does buy you a lot when the signals are
imbalanced. Just not as much as really good power control, that's all.

>down without such rigid power control.  It is theoretically possible
>to implement if everybody got new radios purpose-built for this
>digital mode, but I doubt that any kind of power control scheme will
>ever be used in DX pileups.  DX pileups can have hundreds or even
>thousands of users calling at the same time.

Who says we have to rely totally on the interference resistance of the
modulation method?  None of our existing modes can withstand
co-channel interference that even approaches the strength of the
desired signal, much less exceed it as spread spectrum can. Some of
our modes, such as FM on the shuttle, are about the worst choices
possible for pileup conditions.

We could combine spread spectrum with a traditional contention
resolution method, such as polling by address (callsign) groups, to
reduce the number of simultaneous transmitters to a manageable
number. With ordinary modulation, that number must be exactly 1, not
counting those too weak to matter. With spread spectrum, it can
be something else.

Nothing says everybody has to get through on the very first try; we
certainly don't do that now! The DX stations could still do what
they've always done, which is to work the strongest stations first and
get them out of the way so the weaker guys can have their chance.

What you really want is a modulation and contention-resolution scheme
that maximizes the probability of successfully decoding *at least one*
signal in the pileup, not necessarily all of them at once. As long as
that number isn't usually zero, you'll make headway and everybody will
eventually get their chance. (Again, consider FM on the shuttle as a
lesson on exactly how *not* to solve the problem.)

I think if you the math, you'll see that very few applications are
more ideally suited to the use of spread spectrum than a ham DX

>CDMA technology requires each user to be ASSIGNED a unique PN code by
>a network access channel.  The access channel is a "logical channel"
>(a particular PN code), not a separate RF channel.  An access channel
>is feasible to implement on amateur satellites, where the satellite
>administers the access channel.  I doubt it would ever be widely used
>on terrestrial ham bands because a rare DX station is unlikely to
>administer an access channel, and a remote access channel would not
>have the same propagation as the DX station.

Actually, no. That's just how IS-95 CDMA happens to do it, because
that's how AMPS did it, and it's a good fit to the master/slave model
of cellular telephony. In the random simplex mode better suited to ham
radio, everyone could use the same, standard PN code but with each
station picking a random phase within that code. Just make the code
period long enough to minimize the chance of two stations picking
exactly the same phase, but short enough to search in a reasonable
time. Seems easy enough.

>As a practical matter, future digital modes on ham radio must be
>accomplished using a "standard" computer (no specialized modem
>hardware) connected to a "standard" ham radio (no precise power
>control, frequency agility, or I/Q modulators).  It is reasonable to
>modify radios for the required IF bandwidth.

This is a very reasonable first step; with luck, we'll grow beyond it
as better hardware becomes available. By the way, "standard" SSB
radios are quite capable of power control.  The computer simply varies
the level of the audio signal it feeds into the transmitter microphone
jack. (I assume it's acceptable to ask the operator to turn off any
transmitter AGC).

I see some very interesting low speed data projects that can be
accomplished with nothing more than an unmodified SSB transceiver and
a computer with a sound card. After that, the main thing I'd want is
to open up the transceiver's bandwidth, so I'm glad you make that
acceptable.  But we're already discovering that even if you're only
interested in narrowband SSB and CW, the best way to build a
transceiver is with DSP -- and such transceivers, with the proper I/O
hooks, can be a natural for data applications as well.

>A few years ago it might have been acceptable to have a digital ham
>radio mode that requires an external modem and/or specialized radio
>equipment.  But technology and expectations have changed.  No future
>digital mode will achieve widespread use unless the mode can be
>implemented with standard computers and standards radios.

What constitutes a "standard" radio? Years ago, TNCs were external boxes
plugged into "standard" radios. Now you'll find the TNC an integral
part of many of today's "standard" radios.

I don't think the features of today's "standard" radios should stop us
from designing new features in external boxes that will eventually be
incorporated into tomorrow's "standard" radios. Otherwise we'll never
get anywhere.

>Furthermore, terrestrial digital modes should not require any type of
>controlling infrastructure.

On this I agree fully.


Via the amsat-bb mailing list at AMSAT.ORG courtesy of AMSAT-NA.
To unsubscribe, send "unsubscribe amsat-bb" to Majordomo@amsat.org