On Jun 14, 10:03*am, Richard Clark wrote:

In the same graph, Shannon reveals how, if you code your bits (I will
leave it to the student to discover the meaning of that), you could
achieve the same 1:1000000 advantage with the addition of less than 1
dB of power boost.

73's
Richard Clark, KB7QHC

Until fairly recently, hams didn't do much coding, for a variety of
reasons. Computational horsepower is probably a big reason. Coding's
easy, decoding not so easy, at least in a "parts readily available
from Radio Shack" sort of sense. Obviously, today, one could do all
sorts of coding on a laptop PC, particularly at low bit rates, but
you'd still need to have an unusual convergence of someone who knows
how to implement the coding algorithms who's also interested in
amateur microwave operating. It's not anything like a turnkey thing,
or even a "go get gnuradio" thing. Where you see coding in common ham
use, it's buried in an application (PSK31, JT65, and the like)

The other problem is the frequency control issue. If you want low
rates and ragged edge of Shannon, you need good frequency stability
and control (and to a lesser extent, good phase noise). Until
recently (with GPS disciplined oscillators and surplus Rb sources)
this was a real challenge. As Rich commented with respect to antenna
pointing, you also have to be right on for frequency, and that's hard,
especially in a field situation. Tuning to 10Hz accuracy at 10GHz
implies 1E-9 frequency accuracy, which is challenging. To a certain
extent, processing power in a PC helps (get close, do parallel
demodulation, find the signal), but just like for coding, it requires
finding a person (or small group) who can deal with building low phase
noise stable oscillators AND with developing software that is somewhat
complex, compared to the usual "whack it out in a weekend of coding"
stuff.

I suspect there ARE hams experimenting with this, but it's a long way
from critical mass wide acceptance. You need something that you can
write an article in QST, and offer $100 widgets to make that
happen. There's not much cheap surplus gear either, since commercial
equipment these days tends to be more specialized and isn't as
amenable to hackery.

There are also proprietary rights issues with some coding techniques
(e.g. Turbo) but I suspect that legal issues aren't what's holding
hams back. For things like LDPC, there are published software
implementations that are free to use. I haven't looked but I imagine
that various convolutional codes and decoders are also publicly
available, along with Viterbi soft-decision decoders.