FreeLists / opendtv / [opendtv] Re: ATSC Patent Suit Threaten Nets, Transition

["Allen Le Roy Limberg" <allimberg@xxxxxxxxxxxx>]

U. S. 5,243,627 (Betts & Zuranzki) refers to the earlier Betts, Martinez and
Bremer U. S. patent No. 4,677,625, which may have led the examiner(s) to
take short cuts in the searching for prior art.  Patent attorneys at Bell
and IBM often cited a voluminous number of prior patents in a so-called
prior art statement.  Many Examiners got used to riding these ponies instead
of conducting good searches on their own, and the U. S. Patent & Trademark
Office institutionalized the PAS, making it mandatory for references cited
in corresponding foreign applications.  I suspect Bell sold the Betts &
Zuranzki patent because it was a bad candidate to prevail in an infringement
suit with any company with good patent counsel.

The Betts applications did not cite U. S. patent No. 4,295,218 issued
October 13, 1981 to Robert M. Tanner; titled "Error-correcting coding
 system" and assigned to the Regents of the University of California.  There
is valid reason to doubt that U. S. patent No. 5,243,627 would have issued
to Betts and Zuranski had the Examiner been aware of U. S. patent No.
4,295,218.  Tanner teaches the breaking of long codes into subcodes, then
using distributed coding on a plural-phase basis.  Tanner also teaches
distributing the bits of symbols into different baud intervals in addition
to using plural phases.  In regard to claims 9 and 19 of '218 Tanner teaches
de-interleaving the plural-phase code into its individual codes for
statistical (e. g., Viterbi) decoding.

Since trellis coding was known prior to the earlier Betts and Martinez
patent application, there is valid reason to doubt that U. S. patent No.
4,677,626 would have issued to them had the Examiner been aware of U. S.
patent No. 4,295,218.  Tanner teaches distributed coding on a plural-phase
basis.  There is valid reason to doubt that U. S. patent No. 4,677,625 would
have issued to Betts, Martinez and Bremer had the Examiner been aware of U.
S. patent No. 4,295,218.

The Tanner patent is cited as prior art in ten U. S. patents prosecuted
before 1993, many involving parallel coding arrangements, so the Tanner
patent cannot be considered obscure or not generally known by those skilled
in the art of parallel coding arrangements.  The abstract and specification
of U. S. patent No. 5,157,671 are specific evidence that the Tanner patent
was familiar to an inventor at Space Systems/Loral, Inc. in Palo Alto, CA
and that he believed Tanner's algorithms were widely familiar to others in
the coding art.

One of the spectacular and most notorious events in the history of coding
was turbo coding.  The seminal paper  by C. Berrou, A. Glavieux, and P.
Thitimasjshima titled "Near Shannon Limit Error-Correcting Coding and
Decoding: Turbo-Codes" is memorialized in Proceedings of the IEEE
International Conference on Communications (May 1993, Geneva, Switzerland):
1064-70.  This was three months or so before Betts and Zuranzki filed their
application.  An underlying principle of turbo coding is that the bits of a
symbol contain redundant bits shifted from their original position by an
interleaver.  That is, Viterbi decoder performance in a data communication
system using 2N-dimensional channel symbols N>1 can be further enhanced by
an interleaving technique which uses a distributed trellis encoder in
combination with a signal point interleaver.

Claims 9 and 19 seem to me to read in terms on a turbo decoder.  That is,
the claims are anticipated by Berrou's U. S. patent No. 5,446,747, unless
Betts and Zuranzki are able to swear behind Berrou's 23 April 1991 French
priority date.

Note that Berrou or his U. S. examiner considered U. S. patent No. 4,677,626
of Betts and Martinez relevant to turbo decoding.  It is doubtful whether U.
S. 5,243,627 upon re-examination could withstand an obviousness type
double-patenting rejection based on U. S. patent No. 4,677,626 combined with
the Berrou et al paper or Berrou's patent.

Note that Berrou or his U. S. examiner also considered U. S. patent No.
4,677,625 of Betts, Martinez and Bremer relevant to turbo decoding.  Claims
9 and 19 of U. S. 5,243,627 upon re-examination could not withstand a
double-patenting rejection based on U. S. patent No. 4,677,625.   Claims 9
and 19 are directed to the same invention as claim 2 of U. S. patent No.
4,677,625, which recites "k trellis decoders, each trellis decoder having an
input from said decoder for receiving series of bits from said decoder and
generating n bits of plain text bits corresponding to n bits received at
least during two previous baud periods".  A bit from two baud periods back
is clearly not in an adjacent baud period.  (A basic principle of patent law
is the species anticipates the genus when claiming an invention.)

Al Limberg

 
 
----------------------------------------------------------------------
You can UNSUBSCRIBE from the OpenDTV list in two ways:

- Using the UNSUBSCRIBE command in your user configuration settings at 
FreeLists.org 

- By sending a message to: opendtv-request@xxxxxxxxxxxxx with the word 
unsubscribe in the subject line.