Re: Ira's Poetry Break

Ira Brodsky (
Fri, 15 Nov 1996 15:23:57 -0600

Eugene Leitl wrote:

>On Thu, 14 Nov 1996, Ira Brodsky wrote:
>> [...]
>> However, each of the conversations is digitized (i.e., the voices are sent
>> as bit streams) and assigned a unique "code." The base station receiver
>> picks up all of the signals at once; a computer is able to distinguish and
>> separate the different conversations by "decoding" the composite bit
>> stream.
>Isn't this the same technology enabling a NAVSTAR GPS receiver to sift
>through several simultaneous sat signals? (in contrast to GLONASS, which is
>conventional). Where's the novelty of the CDMA approach, can you disclose
>a bit more details? Sounds really intriguing, but achieving the same
>signal ratio must involve a feedback loop, to dynamically adjust the
>sender power. What the highest possible talker load on a single
>frequency? Some 10, some 100? 'twould be interesting to learn. The stuff
>operates in GHz range, I presume? Individual channel bandwidth???

Yes, GPS uses a one-way version of CDMA. However, there are typically just
3 - 5 "mobile transmitters" (satellites), and much less multipath
interference. Cellular CDMA involves several novelties, including dynamic
power control, "soft" handoffs, "RAKE" receivers, and variable rate speech

Unlike conventional cellular systems (both analog and digital), CDMA reuses
all of the frequencies in all of the cells. A handoff in a conventional
system involves changing frequencies, with a brief loss of audio. In a
CDMA system, the phone can communicate with two or even three cells
simultaneously. As these networks mature, there should be fewer calls
dropped due to bungled handoffs, because CDMA is the only system that never
breaks the connection.

RAKE receivers listen for and recombine multipath signals by adding
variable delays to the components. Instead of interfering with each other,
the compoents now reinforce each other.

And in a conventional cellular network, once a call is in process it
consumes all of its allocated bandwidth, whether or not both parties are
talking at any given moment. With CDMA, variable speech encoding
"upshifts" the bit rate only during voice activity. In fairness, the
competing time division multiple access (TDMA) technology claims it will
add this capability later, but it is clearly *much* harder to accomplish in
a TDMA system, because standard time-slot assignment is quite rigid.

I believe the maximum CDMA "talker" load per 1.25 MHz carrier is 40. At
first, this seems low, as you can fit 41 30-kHz analog channels in the same
space. But in the classic cellular network, you can only reuse one-seventh
of the channels in each cell. (You can improve analog reuse, but the same
techniques apply to CDMA, so there is no net gain for analog.)

>> To the CDMA engineer, this means designing a radio network in which the
>> base stations must control the transmit power of every handset to within an
>> extremely fine tolerance -- no easy thing to do since signals received from
>> mobile handsets naturally vary more than ten thousand-fold due to fading
>> and multipath interference (i.e., signals taking different paths and
>> recombining at the receiver out-of-phase).

>Ah, so they have an autofeedback control loop allright...

Yes. The fraud/conspiracy theory says that once you load up the system,
this control loop will become impossible to control. However, they ignore
the fact the system is intentionally designed to accommodate only about 50%
of its theoretical capacity. It is not normally allowed to get close to
"meltdown" -- assuming such a thing exists.

>> However, if one can make power control work it delivers significant
>> advantages to the end-user. Since the handset is continually compensating
>> for fading by adjusting its transmit power -- thousands of times each
>> second -- the hissing and fluttering sound normally associated with
>> cellular telephone completely vanishes. CDMA phones sound very much like
>Uh, I thought the cellular stuff was digital? The only artefacts being voice
>compression alias?

Yes, both CDMA and TDMA get rid of the hissing and flutter, and both
introduce "quantization noise" as they use lower bit rate encoding. But
consumer tests have consistently shown that CDMA sounds more like landline
telephone. In part, this is because the CDMA camp has good
noise-cancelling technology, and in part because of CDMA's power control
and RAKE receivers. (Fading causes clicking sounds in TDMA systems.)

>> desktop phones; some people will refuse to believe you are calling them
>> from your car. And there is another beneficial side-effect of granular
>> power control: it greatly extends handset battery life.
>An vanilla cell phone in Europe has about 5 W peak power. What does CDMA
>average about to? Does this result in nondeterministic battery life time,
>depending on your movement dynamics and EM environment?

I don't know the peak power off hand, but I'm sure if you spend all of your
time near cell edges you will use more power. Still, it will be the
minimal power needed for that type of operation, and a significant savings
over the coarse power control of analog and TDMA systems.
>> [...]
>> Well, I now have one of these CDMA personal communication phones in hand
>> and it does work. Sound quality is excellent. Due to its enhanced battery
>> life, you can leave the phone on all day. (Actually, it can be left in
>The usual good cell stuff has a standby of max 50 hours, I thought.

That's not an apples to apples comparison. The phone I am talking about
runs at 1.9 gigahertz. The 800 Mhz version gets 72 hours of standby...

>> standby for two days, or you can talk for four hours, or some combination
>4 hours? It's not too sensational.

...and 5 hours of talk time. I believe you are comparing it, however, to
special phones with extra-large (or dual?) batteries. For example, I know
Ericsson has matched the five hours at 800 Mhz, but by using a battery
that's nearly as heavy as the entire CDMA phone *with* battery.

>> inbetween.) Initial prices are 10% - 15% less than cellular telephone.
>Ah. That's an argument. Hardware prices, what about provider prices?

Actually, that was the service provider price. The CDMA phone itself is
more expensive right now, because there is just one supplier near full
production; but phone prices are subsidized in the U.S. (I paid $199 for
the phone; Qualcomm's quantity one price is $530).
>> However, CDMA claims ten times the capacity of conventional cellular, so
>Wowie. Ten times 9.6 kBaud???

No, ten times the voice traffic capacity. Europe's GSM (TDMA) system is
about 2.6 times analog, while the U.S. TDMA technology is barely 2 times
analog, but uses an 8 kbps vocoder with inferior audio (to put it

CDMA claims 10 - 20 times analog at 8 kbps. Carriers are reporting 8 - 10
times analog at 13 kbps, which is 3 - 4 times the capacity of European TDMA

>> prices will probably go much lower as competition heats up. It is also
>> easy to provide features like caller ID over an end-to-end digital network,
>> so they are part of the standard package. Some of the conspiracy theorists
>> are now buying time for themselves by focusing on CDMA's capacity claims,
>> saying the technology will blow up in operators' faces *later* -- as they
>> bring more subscribers online.
>But CDMA must be just cellular, with a (big) plus. They sure modeled
>expected user load, this must be a bogus argument.

Yes. I should also say something about the CDMA versus Europe's GSM
competition. (WARNING: The following is considered heresy by the European
telecomm industry.):

In Europe, the next-generation cellular technology was developed by
committee and *mandated* by governments of the participating countries. In
other words, CDMA is currently illegal for use in European cellular
networks. The excuse for doing this was to ensure a single system that
worked everywhere in Europe. After all, Europe's analog systems are all
incompatible, as the reigning philosophy at that time was national

So they created GSM, which originally was Group Special Mobile (in French),
later changed to Global System for Mobile communications. Personally, I
like to call it Global Socialist Mobile, or Government Standard Mandated.
Regardless of what you call it, and of how it is excused, it represents
merely a shift from national protectionism to continental protectionism.

Please don't take any of this personally; I think Europe is a swell
vacation destination. :-) Plus, you guys invented Western Civilization,
for which I am sincerely grateful. (No kidding.)

GSM fans accuse the U.S. of "cowboy capitalism." I wish that were the
case. However, it is true the U.S. went from a government-mandated analog
cellular standard to industry-voluntary digital standards (plural). So,
when the FCC announced it would auction spectrum for a new personal
communications service, we had something like 70 competing standards at
first, and now we are down to 6. (The other 64 or so didn't attract
sufficient funding.)

I confess that while the European statists did a reasonably good job with
GSM, it is a matter of faith with me that even a somewhat distorted free
market environment is bound to come up with something better. Actually,
the GSM committee considered CDMA, and I am told Siemens actually argued in
favor of it. The committee took what they perceived as the low-risk path
-- leave CDMA as a potential "next" next-generation solution.

Meanwhile, in the U.S. we came up with or "own" TDMA system. Although
developed by industry voluntarily, it is clearly inferior to GSM -- despite
what my friend Craig McCaw claimed. Some TDMA supporters were clearly
angered that Qualcomm muddied the water by developing a second "standard."
But hey, nobody put a gun to anyone's head to buy into it.

Qualcomm gradually convinced every major cellular manufacturer to license
CDMA. Even Sweden's Ericsson has a license, although they used a back
alley route, and tried to make it look like an accident. (They bought 50%
of the British firm Orbitel, who then licensed CDMA for satellite
applications, and then Ericsson bought the other 50%. Of course, that had
nothing to do with CDMA!)

>> In this case, the "conspiracy theory" is really just the predictable
>> reaction of the radio communication industry's equivalent of The Flat Earth
>> Society. It is peddled mainly by old-guard engineers (plus a couple of
>> well-known Stanford professors) who simply can't accept the shift from a
>> separate-frequency to a same-frequency paradigm.
>> There is also much progress being made toward accessing the Net via
>> wireless phones. AT&T is now selling a product called "PocketNet Phone"
>> that retrieves intelligently-filtered text from Web pages. Nokia has
>> introduced a phone with a somewhat larger screen and built-in Web browser.
>Yeah, I've seen it. A flat box looking a plain cellphone, you open it up
>like an oyster and see a diminuitive qwerty and a graphical LCD. I think
>they're using ARM chips. Btw ARM, have a look at StrongARM (at DEC site),
>it will be part of the next Newton, and some other products.
>> If there is interest, perhaps I can post more on this later...
>Yeah, please do. Most extropians like tech things, methinks ;)

OK, I will do that. And thanks for your questions/comments.

Ira Brodsky
Datacomm Research Company
Wilmette, Illinois