SI
SI
discoversearch

We've detected that you're using an ad content blocking browser plug-in or feature. Ads provide a critical source of revenue to the continued operation of Silicon Investor.  We ask that you disable ad blocking while on Silicon Investor in the best interests of our community.  If you are not using an ad blocker but are still receiving this message, make sure your browser's tracking protection is set to the 'standard' level.
Technology Stocks : Nokia (NOK) -- Ignore unavailable to you. Want to Upgrade?


To: EJhonsa who wrote (8506)12/10/2000 4:59:30 AM
From: 49thMIMOMander  Read Replies (2) | Respond to of 34857
 
The ping-pong secrete of GPRS and GSM.

One basic thing of GSM and also of GPRs is
that the handset doesn't send while recieving, greatly
boosting battery efficiency, cheaper to achieve
higher dynamics (difference between max and min)
of transmit and recieve signal, size of cell.

For example, a GSM handset never transmits
at the same time it receives, no expensive
"duplex filters" (except for those USA models
needing to be AMPS compatible) the network
is responsible for selecting different time slots
for the two functions, even maybe considering neighbouring,
competing opertor cells and optimally leaving one "silent"
slot inbetween.

...speak........listen.........speak........listen

instead of doing both at the same time

For example, having a competing operator in a
next-next-cell having it's slots "sliding by"
will disturb "my cell", cooperation helps!!
(compare to CDMA situation were 3-4 basestations
interfere or cooperate with the same channel code,
synchronization, nonideal correlation, one basestation
indoors or underground, not GPS synchronizable,etc)

Anyways, it is obvious that a GPRS handset cannot
receive all 8 timeslots while transmitting on one
or all 8 timeslots, except if theoretically very close
to a basestation.

send.send.listen.listen.listen-listen.send.send.etc

The 4-4 division is one, except that the more practical one
is to transmitting one or two while receiving 2-4 or 5-7
max.

The packet-thing is obviously to only use the slots
needed, shut up if nothing to say (save battery
and cell capacity)

Another basic thing is that GSM "wastes" at least half of
the "capacity" to increase the efficiency of the
power amplifier to gain battery time, GSM is by design
working in C or D-class, transmitting only 01010101
sequences, not a "HiFi" linear signal (A or B class,
to stick to HiFi terms)

This is the "secrete" of EDGE, by using a linear
power amplifier, seemingly wasting power and
battery efficiency, cost,etc, one can easily
double the raw data rate.

However, if the grid of basestations is dense enough
or one is close enough to a basestation this doesn't
add any cost!!

Additionally one can use "higher orders" of modulation,
as 8-16QAM to further increase the bitrate when
the threshold of adding linear PA-stages is done.

Note that:

CDMA demands linear PA-stages not to waste most of
the neighbouring bands as "guardbands" (GSM cellplanning
includes this "loss in capacity")

With GSM-like PA amplifiers in the "pure" CDMA handsets
would have "severe problems" coexisting with other systems.

Basestations always have to have expensive, linear power
amplifiers!!

That is, the important thing is the cost of the number
of paying customers and handsets one basestation can
support compared to the cost of those handsets!!
(sometime setting up the mast and the connecting links might
be the major cost)

This including the problems of neighbouring basestations,
maybe competing operators, and number of handsets needing
one or three basestations to function (soft handoff,
sectoring,etc)

Anyway, back to GPRS:

8 time slots, max 7 or more practical 4 used to recieve,
one or two to transmit, battery and power saving
ping-pong, half duplex,TDAM system (like the ping pong
WCDMA of China and HDR).

If one is in a dense basestation grid one can additionally,
with EDGE, double, tripple or more, the basic datarate,
switch on more battery wasting EDGE power stages in
linear mode, use "higher order" modulations.

If one additionally moves slowly or not at all, these
"higher order" modulations can, just like with CDMA,
be used more easily, reaching a higher efficiency,
transmission rate per Hz!!

Additionally one use the "raw data rate", no extra
coding, bit interleaving needed to cope with
"moving channels", neighbouring cells,etc.

And while continuosly sending,recieving, non-TDMA-CDMA
always need to fight non-optimal, non-theoretical
channelcorrelations,etc, power ratios,etc, EDGE-GSM-TDMA
has everything under control!!

Thats why the one "in the business" takes many CDMA and
WCDMA claims with a kind of cynicism, reminding to
consider the whole system as well as the business idea!!

The mythical single cell Q-CDMA, isolated from reality
in the middle of the desert beating any GSM cell in
crowded Manhattan, but using either linear power stages,
tripple size batteries or quadraple band guardbands
for 100 miles is a regular joke!!

If one adds smart, adaptive antennas, like flash lights
in the night, at least tripling the cost of basestations
and making a handset user wearing a funny looking antenna
hat, one can hype the business even more..

Luckil most operators figure out these things by
actually trying out systems..

Ilmarinen.

Need to reboot another set of WinBoot installations...



To: EJhonsa who wrote (8506)12/12/2000 7:03:21 AM
From: rf_hombre  Read Replies (1) | Respond to of 34857
 
Thanks for a very good overview of the infrastructure and network performance differences of GPRS versus WCDMA.

Your summary does however beg the question: Can GPRS operators afford to displace billions of minutes of voice airtime to relatively low revenue generating packet traffic?

One answer is that since GPRS IP traffic takes a back seat to voice, meaning that the network grants priority to voice traffic, this is not really an issue.
When a voice request for service is made, and all GSM timeslots are used, a few GPRS users will be released from the network in favor of that additional voice user. So IP is not hogging up capacity, it is using up excess Erlangs that would have gone unused anyway. But at the same time, operators cannot guarantee that the desired Quality of Service will be achieved consistently.

I believe this is one of the bone of contention that the HDR and CDMA2000 folks have over shared voice and IP spectrums in general and over GPRS and WCDMA in particular.

From the standpoint of the user then, the problem with GPRS is not so much the lack of high speed throughput but rather the fact that QoS cannot be guaranteed. So while business folk are waiting for their train or plane and attempt to access their email during the system busy hour, they will likely get denied service. Corporate customers are likely to demand a consistent level of service and be willing to pay a premium for it as well. But if 10 years of GSM is any indication, operators are very reluctant to offer segmented service levels for different users, as it would be undemocratic and could hurt their corporate image if it came to be known that this form of wireless segregation was going on. Ericsson, for example, has in its software the option of granting priority to certain “gold” customers, but to the best of my knowledge, this special feature (“Efficient Priority Handling”) is used very very sparingly. And certainly not in european markets.

So GPRS operators will either have to get used to the idea of offering subscribers “what they pay for” and thereby guaranteeing somehow QoS or risk losing business customers altogether.

The alternative is to use GPRS as a data platform to enter low bandwidth mass consumer data markets. Bundled with intelligent network features, such as location based information, GPRS need not require a higher throughput than 14.4kb/s to offer a cheap and commercially viable step up from Wap. With the current state of handset technology, this backdoor approach to GPRS may in fact prove more beneficial to all parties. GPRS would create a latent demand and awareness of data services in the consumer; that would benefit WCDMA and CDMA2000 alike.

rf_hombre



To: EJhonsa who wrote (8506)12/12/2000 3:04:26 PM
From: 49thMIMOMander  Respond to of 34857
 
< in the case of GPRS, compression technologies allow this number to be increased by 50%>

No compression, instead less channel coding than for
voice when the channel is good (pedestrian, external
interferences, etc)

That is, the "raw" data rate in the GSM air interface is
more than 20kbps (200kHz per 8 time slots)
The "rawest" CS-1 (?) coding does not code at all, just adds
a checksum to detect if the data is OK or not, while the
one with most coding uses up more than half to "help buddy
bits in troubles", giving appr 9kbps.

The "overheating and radiating" power amplifiers is
mostly a myth, as power is only used when actual data
is _transmitted_, that is, it only applies for someone
setting up a server for transmitting streaming video
using a handset.

And even in that case the handset must additionally
be on the border of a cell or in constant multipath
fading to transmit on max power, which additionally
easily can be limited by lowering the bitrate, use
less timeslots, also obvious from the battery
perspective.

Ilmarinen

P.S. No coding for CS-1 except that FSK is in itself a 1/2 coder.
When one moves to EDGE the raw datarates are doubled
without actual amplitude modulation, FSK -> 4QAM.
But when going to 8 then 16QAM one does another doubling
and adds actual amplitude modulation.

That is, FSK must always change the phase forwards or backwards, +90 or -90 degree, while 4QAM additionally
can do nothing, 0 degree, and make a 180 degree jump,
doubling the datarate (2 bits per symbol).

This without any substantial degradations in capability
to withstand noise and interference, they are becoming
more severe at 16,32QAM (4,5 bits per symbol)

However, when going from FSK to 4QAM some basic linearity
is needed from the power stages, but not in the receiver
path!! (note difference handset-basestation, basestations
anyway have linear powestages, handling many channels)

However, 16,32QAM adds more factors, like the interference
from the same channel in "far-away" cells, but as one
can be adaptively switched down to 4QAM its not a
"limitation", just a way to "use what can be used".
Not even trying to drive big trucks on small bumpy roads,
but instantly switch to a tractor for that purpose and let
the trucks roam on the highways.