Re: Faster mobile broadband trials successful

"Kristoff Bonne" <skypro.be@xxxxxxxxxxxxxx> wrote in message
news:497397a9$0$2854$ba620e4c@xxxxxxxxxxxxxx

Upgrading from UMTS to HSPA bassicly means two things:
- move from dedicated 386 Kbps channels (one per user) to a shared
upstream/downstream channel of x Mbps (for all users in one cell).

Don't forget that UMTS is also limited in terms of capacity.
Quoting
from the article I originally gave a link to:
http://www.commsdesign.com/showArticle.jhtml?articleID=59100260
"Instead of limiting high-speed data access to fewer than five
users
in a cell, HSDPA can deliver 384-kbit/s data to many more, maybe
30,
users."

Note the "maybe" :-)

The author is compairing 384K dedicated and garanteed bandwidth with
a
system of a shared pool for all users.


But *how many* people in a cell could receive 384 kbps guaranteed
bandwidth on 3G?? Not many.


The "maybe 30" is based on an estimate of what you might gain by
applying statistical multiplexing to more users in one big pool
instead
of every user in his/her dedicated channel.


I think he'll know what he's talking about a lot better than you will
just based on one course at work.


The author doesn't say you can run 30 simultanous users, each
running a
384 Kbps stream in a typical HSPA-cell, and there's a very good
reason
he doesn't do that. :-)


If you think he's so wrong, explain why most Internet users woudl need
to have guaranteed 384 kbps of bandwidth anyway.


With W-CDMA, all the users' signals are transmitted on top of one
another, so if they all want to download bursty traffic at high
speed
then the interference from other users' signals will increase.

That's not how 3G is implemented.


I think you'll find that it is.


3G uses fixed CDMA-codes which are organised in different "levels"
based
on the symbolrate of the signal being carrier by it.


Thanks for teaching me the basics about CDMA. 3G was all the rage at
the time when I did my MSc, and I happened to specialise in all the
mobile comms subjects, so I think I just might know a thing or two
about how CDMA works, thanks.

CDMA systems use the PN codes as the multiple access mechanism, but
all signals definitely transmit on top of one another, and then the
receive multiplies by the same PN code at the receiver to recover the
transmitted data. If they didn't tell you that on your course, then it
must have been a very basic course considering that that's CDMA 101
stuff.


A code for a channel running at a SR "x" is equivalent then two
channels
using two CDMA-codes for SRs "x/2". That's why the CDMA-codes used
by 3G
are organised in the tree!

A HSPA shared downlink and shared uplink channel run at higher SRs
then
the 384 Kbps dedicated channels, so they take up the equivalent
number
of CDMA-codes.


And?


Migration from HSPA to HSPA+ includes this:
- Include the use MIMO-technology (using multiple antenna's to
increase
the signal-quality).
- Due to this increase of signal-quality, add still some other
encoding-scemes.

64-QAM instead of 16-QAM on HSPA in other words. Plus some of the
parameters are better on HSPA+, such as latency for ARQ and the
channel quality loop are shorter, IIRC.

That's true.

But as this does not really impact streaming-applications, I didn't
mention it.



- introduce an (optional) all-IP infrastructure.
So, now the questions:
- Can you upgrade a UMTS/R99 network to HSPA with only software?
Here the answer is simple: yes.

But it won't perform as well as if you upgraded the hardware as
well.

What do you mean by "upgraded hardware"?


Upgrades to the digital hardware that does the DSP processing and
other algorithm processing in the base station.


- But can you upgrade a HSPA network to HSPA+ in only software?

Well, the answer is a bit complex. In theory, it's simple: MIMO
needs
multiple antenna's (if possible, both at the transmitting-side and
the
receiving-side) so you need a hardware-intervention.

MIMO is only MIMO if there are multiple transmitters at both the
transmitting and receiving ends. M stands for multiple. So if
there's
a single antenna at either end you'd have a SIMO or MISO system.

True, but I left out this part not to make the text anymore
difficult
then necessairy.

For who's interested: the wikipedia has a interesting article on it:
http://en.wikipedia.org/wiki/MIMO
http://de.wikipedia.org/wiki/MIMO_(Nachrichtentechnik)



However,
1/ quite a lot of UMTS-transmission-towers do actually already
have
multiple antenna's.

Are they actually multiple antennas, or are they directional
antennas
implementing sectoring? Mobile phone systems have used sectoring
for
years, where they have three antennas on a base station with each
one
pointing in a direction that's 120 degrees different to the
adjacent
ones. Those aren't antenna arrays though, because they're only
implementing sectoring.

It depends.

In most cases the coverage-areas of the antennas partly an overlap.
It
actually depends on the local situation; especially the question if
the
the different sectors use the same frequency or not.

This all depends on local frequency- and planning-concideration.


Unless a base station uses antenna arrays rather than merely using
multiple antennas to provide sectorign, then the base station would
require new antennas to form the antenna array - you can't use
directional antennas as part fo an antenna array when they're pointing
in the wrong direction!


2/ Another aspect is the end-user equipement. If you have a
receiver
with multiple antenna's, you are also "partly" MIMO.

No, you're SIMO unless there are multiple transmitters at the base
station...

Hence the name "partly mimo" I used. :-)



It is interesting to actally make a sidenote about the speeds
given
for
the different 3G network.
1/ While people say "UMTS provides 384 Kbps, while HSDPA gives 3.4
Mbps,
that's 10 times the speed".

HSDPA actually goes up to 14.4 Mbps in theory, but that's
apparently
with no error correction, so the maximum would be 14.4 x 3/4, which
is
the maximum code rate.

That's one of the differences between "theory" and "practice".


Hardly. Shannon's channel capacity theorem is all about the fact that
the only way to receive error-free datastreams is if you use error
correction coding! So this is precisely what theory says.


Quite some operators limit the speeds of HSDPA for a number of
reasons,
ranging from tower-planning to backbone conciderations.

There aren't that much operators offering HSPA at full speed, and in
a
lot (if not all) cases, they just do it in parts of their network so
that the "marketing"-guys can say "we provide HSDPA which _can_
provide
speeds up to ...".


OK, that is true, but the 384 kbps figure for UMTS/R99 is PER
USER!
That's a dedicated channel PER USER.
http://www.commsdesign.com/showArticle.jhtml?articleID=59100260
"Instead of limiting high-speed data access to fewer than five
users
in a cell, HSDPA can deliver 384-kbit/s data to many more, maybe
30,
users."

See above.


HSDPA and HSUPA use SHARED channels, which are common for all
users
in a
cell. That 3.6 or 7.2 Mbps figures given are speeds of the ALL
terminals
in the same cell. This means you can BURST upto the speeds speeds,
if
you are only user of a cell, but in most cases that's not the
case.

Don't forget the effect of statistical multiplexing, because speeds
are bursty, but it's unlikely taht all users would be downloading
bursty data at the same time.

Correct, and that's exactly the problem with streaming over a
3G-network. That's is not "bursty" anymore and you brake the
statistical
multiplex which forms the basis of HSPA!


Bullshit. Stat muxing is not the basis of HSPA any more than it's the
basis of CDMA. Stat muxing is a fact of life in all telecoms networks.
Period.


It's completely wrong to think you HSPA multiplies the capacity of
the
cell by 10 or 20. That's simply not the case.

The article above reckons that it's by a factor of 6 or so.
HSPA+ will
make that even more.

See above!

3G network planning is a very specialised field with very
specialised
knowledge and -therefor- very expensive! The people who do work in
the
field are not going to write articles about it in the press. You can
be
sure about that!!!


Lots of people know what capacity 3G and HSPA can handle - thousands
of people in universities around the world have been simulating this
stuff for years.


I can tell you this: if you take all elements in account: practicle
planning-issues, frequency-issues, the mix of enduser equipement
with
different generations of technologies and different radio-quality
situations, ... the actual figures of are quite different then what
the
sales-brochures of the technology-companies tell you.


You seem to be making yourself out to be some kind of expert after
taking one short course at work.


... And LTE has about double the spectral efficiency
of HSPA+ apparently:
http://3g4g.blogspot.com/2008/01/comparison-hspa-vs-lte.html
plus there are other improvements compared to HSPA+.

True, but then, LTE doesn't exist yet and don't hold your breath
until
it get's there! :-)


It isn't that far away. The first global deployments are giong to
start within the next year or so.


Now, in this NG, the idea has been brought forward to use "mobile
internet" for internet-streaming.
The problem here is that this kind-of brakes the basic principle
of
HSDPA. A quasi-continues stream of (say) 100-or-so Kbps is still
half of
the 384 Kbps of a dedicated UMTS/R99 dedicated channel, but having
to
many continues streams on a HSDPA-cell will still kill the
network!

http://www.commsdesign.com/showArticle.jhtml?articleID=59100260
"Instead of limiting high-speed data access to fewer than five
users
in a cell, HSDPA can deliver 384-kbit/s data to many more, maybe
30,
users."

See above.

I propose you mail the author and ask him if he did actually mean
"30 streams at 384 Kbps/s simultanous, in a real life HSDPA-network,
under real life conditions", and if he has actually seen this
working in
real life.


You email him.


When the sales-documents talk of "3.6 Mbps" or "7.2 Mbps" or "13
Mbps",
that's only valid for BURSTABLE bitrate for a terminal with good
radio-conditions.
If you are at the edge of the cell, speeds will go much lower!

Internet radio streams use bit rates far below the Mbps level
though...

And the relevance to the difference between burstable but shared
bandwidth and garanteed dedicated traffic is?


The obvious relevance is that if a cell can deliver 384 kbps to 30
users, think how many 128 or 64 kbps streams can be handled! Isn't
that obvious??


So, as a cell usually carries multiple users, all under different
radio-conditions, things become much more difficult to evaluate.
Say you have a cell with two users: all close to the transmittor.
If
you
are on a 3.6 Mbps HSDPA-network, every user will be able to use
1.8
Mbps.
However, say that one of these two endusers in not in the center
of
the
cell, but at the edge. For her, the network will need to switch to
QPSK
instead of QAM16 or QAM64, just to make sure that the enduser will
be
able to decode this.
This means that that for that user, the maximum speed might be
something
like 400 Kbps.

Internet radio streams use bit rates far below the 400 kbps level
though...

OK, but the relevance of this with the fact that endusers under bad
radio-conditions negatively impact all users of a cell is what
exactly?


The relevance was that you had said that the max bandwidth at the edge
of the cell might be 400 kbps, and that that's higher than you need
for Internet radio.


As the total capacity of a 3G-cell is actually limited by the SR
(SYMBOLrate) of all the users, and not the nett bitrate, this in
fact
means that the total capacity of the cell will NOT be 3.6 Mbps (as
before), but 2.2 Mbps (1.8 Mbps for the "closeby" user and 400
Kbps
for
the "far away" user).

In short. What does this mean?
Eh?

If you upgrade a 3G network to HSPA+, the total capacity of a cell
will
increase due to the introduction of MIMO and the new
encoding-scemes,
you should not expect that -all of a sudden- the real-life
capacity
of
the network will double.

If a cell changed from using single-antenna HSPA to using 2x2 MIMO
HSPA+, I would expect the *capacity* to at least double - double
due
to the doubling of the antennas, plus a bit more due to the other
improvements for HSPA+ relative to HSPA.


But that's under the restriction that -all of a sudden- everybody
uses a
enduser equipement with two antenna's too.


Okay, I see that you said "all of a sudden", but you didn't exactly
make it clear that you meant that there would be lots of user
equipment with single antennas - I thought you were referrnig to when
MIMO is basically very widely supported.


But, of course, the reality is not like that: there are people using
3G
iphone's with just one antenna, UMTS/R99-only cellphones and PDAs,
USB-sticks with just one antenna, etc.




Don't get me wrong. I'm not saying that HSPA+ will not help at all,
but
there is a difference between what the marketing-boys and girls of
the
technology-companies tell you (based on a pure technical model) and
what
the people who really do 3G network-planning see in real life!!!


In real life, using 2x2 MIMO increases the capacity by a factor of 2.
Just because that won't be achieved overnight due to legacy hardware
still being used doesn't alter the fact that it does offer a factor of
2 increase in capacity. And one day we'll all be using MIMO-capable
handsets, and people change their mobile phones more frequently than
any other type of electronics kit.


That's what I wanted to point out. People should not get the wrong
impression about 3G.


I haven't got the wrong impression about 3G, thanks.



--
Steve - www.digitalradiotech.co.uk - Digital Radio News & Info

The adoption of DAB was the most incompetent technical
decision ever made in the history of UK broadcasting:
http://www.digitalradiotech.co.uk/dab/incompetent_adoption_of_dab.htm


.

Relevant Pages

  • Re: Faster mobile broadband trials successful
    ... upstream/downstream channel of x Mbps (for all users in one cell). ... A HSPA shared downlink and shared uplink channel run at higher SRs then ... Quite some operators limit the speeds of HSDPA for a number of reasons, ...
    (alt.radio.digital)
  • Re: Cellphone Reception/Transmission
    ... I guess a US cell tower and a BC cell tower were competing for my ... There are also various plans for do it thyself antennas. ... minimum power and excessive antenna gain is employed). ...
    (sci.electronics.design)
  • Re: Cellphone Reception/Transmission
    ... I guess a US cell tower and a BC cell tower were competing for my ... There are also various plans for do it thyself antennas. ... antennas and attaching them to your handset likely violates several ...
    (sci.electronics.design)
  • Re: Cellphone Reception/Transmission
    ... I guess a US cell tower and a BC cell tower were competing for my ... There are also various plans for do it thyself antennas. ... antennas and attaching them to your handset likely violates several ...
    (sci.electronics.design)
  • Re: Cellphone Reception/Transmission
    ... I guess a US cell tower and a BC cell tower were competing for my ... There are also various plans for do it thyself antennas. ... antennas and attaching them to your handset likely violates several ...
    (sci.electronics.design)