I have never said this <You mentioned before that power control is easier to do now than in the past due to the work folks have done with cdma.>
Instead I have referred to it as neanderthalian idea of keeping the wives and kids suitably silent in the
cave. That is, it is such an obvious thing. Another thing I have made jokes about is that "medium fast
power control", that it, it is one thing to set the optimum powerlevel within +/- 1dB (10%) or less in
a static enviroment, gicen enough times "to settle", but a totally different when the channel, multipath,
hgh speed handset moving, is changing, going kind of upside-down in every 0.1-0.2 seconds (or even less
if one must catch those bits in the middle of the fading)
I have also tried to inform the "GSM has no pwer control, CDMA is greatest" that GSM has had
power control from the very start, 12dB in 12 steps, now some 15 dB. (however, the goal is not to
adjust for those fast and deep fadings when moving, instead every timeslot containes a pre
defined "training" or "measurment", "calibration" signal.
Power control for CDMA and GSM is the same thing, althouhg
- extremely critical for CDMA
- less so for GSM (and the issue now is to do power control over many cells, to improve their total
capacity, how they share both channels and interference)
< I take it that channel allocation to EDGE/GPRS/GSM is dynamic and therefore moving more data in a cleaner environment enables more time for allocation of GSM voice channels and less capacity degradation due to data transmission via GPRS only.>
Correct, additionally improving voice reception for "handsets in the worst places"
<Is voice on an EDGE channel and smaller time slices part of the current standard or potential future upgrades? >
Well, i have to admit I do not have time to follow the details of every old,frozen,implemented,etc release.
One general rule is that many things are "optional" in the way that a certain implemtation can basically
"not use them". gets messy when there is a lot different variation and version handsets, as well for
handsets to communicate their capabilities to any GSM basestation in the world.
<And, don't these move EDGE into the realm of cdma and Qualcomm's IPR? to measure that fading>
In no way, GSM has been measuring the channel from the very start, as well as adjusting the
overall power level.
- 8 phase shifting
Not that difficult to start with, make a circle and place 8 eually spaced dots on it. One is allowed
to go from any dot to any other dot or stay where one is. Note, if going 180 degree you are
going to "dive through zero, origo". By rotating the "next" set of dots 22.5 degree you will never
go through zero, just "fly close by". That is, the RF-amplifiers always produce a "carrier", a
certain minumum amplitude.
Bandlimiting, filtering the signal makes it slightly more difficult, like doing this with heavy,slow pen
who cannot make instant accelerations nor breaks, need to make "smooth curves". This will
make the signl point "fly even closer by the origo".
However, to calculate the result of any non-perfect filtering and nonlinearity of the RF parts
is where it gets tricky, especially as GSM allowes a lot of "side-garbage" on the closest
neighboring channel which are not used in the same cell, only in neighboring cells farther away.
That is, the really tough requirements not to interfere with others come after the second-third
neighboring channel.
That is, this is where "dynamic and multi-lateral power-and-channel-selection-amongst-neighboring cells"
becomes important.
Ilmarinen
Btw, on that circle above, basic GSM is only allowed to 90 degree left or right, along the circle, that
"constant amplitude thing". That is, the signal point just goes "back and forth", "left and right",
not sit in the same place nor dive through origo to the other side.
That "Gaussian stuff" means it goes back and forth in a "smooth way", old trick to "filter"
the signal, to lower the side-lobes falling ontop of neighboring channels.
That is, the base for the 3x increase, doing one 0 or 1 bit per phase jumo, compared to 8 different ones,
giving a "load" of three bits, that 000,001,010...110,111 per phase change. |
