Path: utzoo!utgpu!news-server.csri.toronto.edu!cs.utexas.edu!uunet!motcid!wang
From: wang@motcid.UUCP (Jerry Wang)
Newsgroups: sci.electronics
Subject: Re: IEEE/Globecom: Qualcomm Spread Spectrum
Keywords: spread spectrum qualcomm viterbi
Message-ID: <4528@manta5.UUCP>
Date: 20 Dec 90 19:02:10 GMT
References: <1990Dec13.122314.6448@world.std.com> <4eabf5aa.1423f@godzilla.UUCP> <1990Dec19.183013.17271@jarvis.csri.toronto.edu>
Organization: Motorola Inc., Cellular Infrastructure Div., Arlington Heights, IL
Lines: 75

chik@eecg.toronto.edu (Raymond Chik) writes:

>In article <4eabf5aa.1423f@godzilla.UUCP> dalyb@godzilla.UUCP (Brian Daly) writes:
>	Could someone enlighten the rest with a little more detailed 
>description of	CDMA and FDMA ?

Here is my limited knowledge:

I. FDMA - Frequency Division Multiple Access

   This is how the current analog cellular system works.  The spectrum
   is divided into multiple 30 kHz (25 kHz in UK TACS system) channels.
   Each cellular subscriber is assigned to one such channel during
   a cellular phone call without interfering with each other.

II. CDMA - Code Division Multiple Access

   A. Spectrum Spreading

   CDMA is a digital multiple access communication system based on
   Direct Sequence Spread Spectrum (DS-SS).  The digital information
   to be transmitted (speech, data or control) is multiplied by a
   pseudo random sequence prior to modulation.  As a result the
   symbol rate entering to the channel is increased in proportion to the
   length of the sequence.  This is why it's called 'spread' spectrum.  The
   key advantage of spread spectrum is 'processing gain' over white noise, as
   evidenced by Shannon's channel capacity theory (Shannon said,
   you can get any BER performance you want given a S/N, as long
   as you have the 'bandwidth' to do it.  So the bandwidth increase
   contributes to 'processing gain directly).  Another advantage
   is 'anti-jamming' capability due to the spectrum spreading effect.

   The multiple access capability comes from the auto-corelation and 
   cross-corelation properties of the set of pseudo-random sequences
   used for spreading.

   B. Multiple Access

   Ideally each sequence resembles a noise pattern.
   Auto-correlation of noise produces a 'delta function'
   which has value of infinity at t=0 and zero for the rest of t.
   Cross-correlation of two independent noise patterns produces
   over all time.

   Assuming that continuos '1' is spreaded by two independent pseudo 
   sequences (let's call that sequence A and sequence B) and transmitted 
   to the channel by two transmitters, and a receiver uses sequence
   A to corelate the combined signals: transmitter with sequence A
   will produce a peak (the pseudo 'delta function') for every
   transmitted '1' at the receiver and transmitter with sequence B
   will produce no response at all at the receiver.  The receiver's
   capability to selectively pick up the desired transmission
   by using different sequence (Code) is the main principle for
   CDMA.

   C. CDMA issues - a few example

   Not all the codes (pseudo random sequences) chosen have ideal
   auto-corelation and cross-corelation properties.  Multiple
   transmissions over different codes decreases the jaming margin
   (processing gain plus input S/N, desired output S/N).  Strong (near)
   transmitter further eats up jamming margin for a weaker transmitter
   (far).

CDMA has a lot of advatntages over the existing FDMA and TDMA
(Time Division Multiple Access, being adapted in Europe)
for cellular application, but it also has it's own issues to
be resolved.  These are beyond the scope of this discussion.

Qualcomm is headed by people who write 'bibles' in the field of communication
and information theory, e.g. Viterbi, Jacobs, etc.  However I have not 
seen the Qualcomm proposal in the public domain to verify it's claims.
(And I don't believe 'odinary people' like myself can challenge them anyway).

Jerry Wang - Motorola