Installed fiber
Well, installing any kind of major fiber cable under water
will automatically mean terabits of capacity as it only
takes some 1/4 inch of fibers for that, the steel
enforcement to ensure physical strength takes
and costs more.
Maybe not reaching over the atlantic, but at least
making shorter fiber-hops around the finnish gulf.
I would not mix CAs energy problems with the way EU
and the scandinavian countries integrate the
St Petersburg region with EU. Historically St Pete
(not Florida) has been a part of the "nordic dimension"
for longer times than outside during this century of
major and cold wars.
The problem for EU, after having Estonia and the other
baltic nations included, is mostly drawing some
temporary limit between St Petersburg and the farthest
Mongolia and China until the economies are more comparable.
--
On modulation schemes, the modems used for 33.6kbps already
use very sophisticated, more than 1000 point QAM modulation
with lots of coding, although at very low bandwidths and
symbolrates, 3kHz or 3kBaud.
(33.6kbps transfers 10 bits/symbol at 3Khz = appr 1000-QAM
to simplify matters, 56kbps is another matter)
The text you quote is correct in that fixed connections
can use higher "densities", "transfer rates per Hz" like
64QAM, giving a 6 (2**6=64) fold increase in transfer
capacity compared to regular ones and zeroes.
However, as this needs a high signal-to-noise ratio as
well as tracking capabilty of changes in the channel,
overall linearity of the channel, etc, this won't work for
mobile transmission, but achievable for fixed.
The other limitation is pure CPU horsepower, easy to do
for some 3,000 QAM symbols per second (modems), not the same
for some 3,000,000 QAM symbols per second.
GSM allowing for "100% nonlinearity" in the GHz signal
path is one kind of optimum, CDMA, both Q and W,
demands more linearity, horsepower. Adding "higher
modulations" is just what has been done to regular modems,
radiolinks, etc as the "horsepower" of silicon
implementations have increased.
That is, the basic idea of GSM was to allow the handset
to work with transmitters only capable of turning
transistors barely on and off at the carrier frequency, but
was still beyond cheap, cost-effective implementations until
the middle 90s.
Now technology is ready for these "higher modulations",
especially in fixed, stationary implementations (where the
channel response doesn't change,etc)
The basic limitation is the width of the smallest possible
transistor, same as its maximum frequency of operation,
from 2-3um and X00Mhz to 0.1um and X0Ghz, since the
"great war of DRAM" mostly in the hands of the japanese
process technology.
(huge long term investments,high quality,low margins,etc)
When the process for making wafers exist, every wafer
costs "peanuts", only a question on how many chips
actually works from that wafer and how many are
ordered to pay for the investments (assuming the
chip is smaller than the maximum size of the random
error-density of the process, the factor of quality,
$1000 pentiums are far above that threshold and also
need a fan to cool them, speed needs power and battery)
Big thing if one gets 0-4 chips for $200 or 1-10,000
chips for the same money, and how many billions it
took to produce the first wafer (as well as how many
years it is profitable to keep that fab going)
That is, one of the most interesting "food-chains" in human
history, more critical to cooperation and not messing up
within that "chain", one part beeing "scales of production"
and ordering (allocating) parts in time and actually
paying cold cash for them when of delivered.
That is, it is one thing to claim this and that for
20-300 laboratory prototypes, and another to do the "real thing",
mastering the whole process, from wafer processes for
billions to consumers paying $50-100.
Ilmarinen, deep in holes and electrones.. |
