“CDMA won’t work with 40x capacity" was the more exact issue.
One of the factors was how large proportion of the handsets
would be in softhandoff, using double capacity.
Other factors were accuracy of power control in
realistic fast multipath fading conditions,
non-zero walsh correlation when timing isn't
perfect alternatively channel delay varies,etc.
Personally I have wondered how to in practice track both
doppler and multipath variations on low user speeds,
when the changes occure during a "long" symbol which hasn't
yet been fully received and detected.
(one can of course buffer the received signal and
run it through a lot of possible scenaries, kind
of Viterbi survivor style,pick the best one, but then it isn't that practical)
And as always one has the problem of what happens
with all the tracking and adapting when things
are "bad", with error rates of 1-10%, or worse,
when the signal is totally destroyed for a short
time by some external high energy transient or similar.
(or driving by a full power AM/FM/AMPS/etc sender,
jamming receiver RF parts)
That is, how the whole system recovers from a more
or less total failure, robustness, especially when
including handsets of various designs,etc.
The GSM channels are fairly uncorrelated and independent,
robust overall function, the opposite of the CDMA channels
which are highly connected, one messes up and all others
will soon do the same, chainreactions, kind of.
On the future I see a possibilty of having almost
any kind of system running on a platform with
- very linear RF parts
- high accuracy AD and DA converters
- lots of DSP horse power
Even a wide band GSM version with thus improved
multipath fading capabilities, and still the
optimal battery saving constant amplitude
feature, when saving battery is the most
important thing.
OFDMA should be a piece of cake on the same platform,
if it is programmable.
Compare modem chips who do everything from ADSL to
V.90 to fax and even 300bps FSK from the 60s.
Two lessons from the modem world was
- how ITU standards quickly replaced proprietory
standards
- how the fight over the V.90 standard almost drove
even major modem brands in the negative during the
years everyone postponed buying a new modems,
investing in modem pools, in spite of marketing tricks
like "free future upgrades" for shipping and handling costs,etc.
Ilmarinen.
P.S. The Ethernet collision detection protocol is/was
also a classic example of proven unstable system,
must reset,restart the whole network through timeouts to
make it work. |
