Institute for Mathematics and its Applications (IMA) - 1999 IMA Summer...
`
`https://www.ima.umn.edu/csg/
`
`PROGRAMS/ACTIVITIES
`
`Annual Thematic Program »
`
`Postdoctoral Fellowships »
`
`Hot Topics and Special »
`Public Lectures »
`
`New Directions »
`
`PI Programs »
`Industrial Programs »
`
`Seminars »
`
`Be an Organizer »
`Annual »
`Hot Topics »
`PI Summer »
`PI Conference »
`
`Applying to Participate »
`
`1999 IMA Summer Program:
`Codes, Systems and Graphical Models
`yong
`
`Schedule
`from Talks
`
`Participants
`
`Bibliographic Items Related to Week 1
`
`Bibliographic Items Related to Week 2 Material
`
`Partially supported by the National Security Agency
`
`August 2-13, 1999
`Organizers:
`
`G. David Forney, Jr.
`Massachusetts Institute of Technology
`LUSE27@email.mot.com
`forney@lids.mit.edu
`
`Brian Marcus
`IBM Almaden Research Center
`marcus@almaden.ibm.com
`
`Joachim Rosenthal
`University of Notre Dame
`rosen@nd.edu
`
`Alexander Vardy
`University of California, San Diego
`vardy@ece.ucsd.edu
`Note: The registration for this summer workshop has been closed due to an overwhelming response.
`
`The invention of turbo codes and other capacity-approaching codes has led to an exciting cross-
`fertilization of ideas between researchers from different backgrounds.
`The aim of the workshop is to bring together mathematicians, computer scientists, and electrical
`engineers in the area of coding theory, systems theory and symbolic dynamics so that the techniques
`from one area can be applied to problems in the other area. The two weeks of the workshop will be
`subdivided into two main focus areas:
`Week 1:
`Codes on Graphs and Iterative Decoding
`Week 2:
`Connections Among Coding Theory, System Theory and Symbolic Dynamics
`Week 1
`CODES ON GRAPHS AND ITERATIVE DECODING
`Belief propagation in Bayesian networks has been extensively studied in artificial intelligence since the
`work of Pearl a decade ago, and turbo codes have recently become a subject of much research in coding
`theory. In the past year or two it has been recognized that the iterative decoding algorithm used for turbo
`codes and other capacity-approaching schemes are instances of belief propagation. This has led to an
`explosion of work devoted to understanding and exploiting this connection. A related problem is that of
`representing a given code by a graph, such as a Bayesian network. A central impetus of much of this work
`is to understand why iterative algorithms work so well empirically on graphs with cycles, where practically
`no theoretical results are known. Experts in the dynamics of algorithms have also begun to be drawn into
`this work. The major focus of week 1 of the IMA workshop will be to bring together researchers in these
`various fields to better understand these emerging connections. This will be a natural follow-on to a
`special session on this subject at the upcoming 1998 MTNS conference (Mathematical Theory of Networks
`and Systems, among the most mathematical of the systems theory conferences).
`Topics for week 1 include: Codes defined on graphs, iterative decoding algorithms, factor graphs, turbo
`codes, connections with Bayesian networks.
`
`1 of 6
`
`10/11/2014 2:28 PM
`
`Hughes, Exh. 1037, p. 1
`
`

`
`Institute for Mathematics and its Applications (IMA) - 1999 IMA Summer...
`
`https://www.ima.umn.edu/csg/
`
`Week 2
`CONNECTIONS AMONG CODING THEORY, SYSTEM THEORY AND SYMBOLIC
`DYNAMICS
`Coding Theory, System Theory and Symbolic Dynamics have much in common as evidenced by the
`following list of research topics that play a prominent role in each:
`
`1.
`Construction of various types of finite- and finite-dimensional state representations of sequence
`spaces.
`2.
`Investigation of fundamental structural properties of sequence spaces, such as observability and
`controllability.
`3.
`Construction of input/output systems, i.e. mappings (or encoders) between sequence spaces.
`4.
`Understanding the special role that algebraic structure (in particular, linearity and duality) plays in
`1,2 and 3.
`
`Yet these subjects have developed somewhat independently, and each has its own language and points of
`view. Until recently there has been very little communication among researchers in these subjects. A
`main purpose of week 2 of the IMA workshop is to further the communication among researchers and
`stimulate connections among these subjects. Week 2 will aim to continue a successful series of
`interdisciplinary meetings that has included an IEEE Information Theory Workshop on Coding, Systems
`and Symbolic Dynamics in 1993 (Mansfield, MA), a special invited session at the IEEE Conference on
`Decision and Control in 1995 (New Orleans), and two special sessions at the MTNS in 1998 (Padova).
`Topics for week 2 include: Behavioral system theory, input/output mappings between spaces of
`sequences, state space representations, group codes, trellis codes, multi-dimensional systems and codes.
`The organizers plan a number of invited tutorial lectures specifically for interspecialty communication.
`Leading workers in each field will also be invited to present surveys of current research, with less
`emphasis on solved problems than on open ones. Finally, there will be both invited and contributed papers
`presenting recent research results.
`We expect the attendees to represent electrical engineering, mathematics and computer science
`departments in both academia and industry. As coding theory is the glue that holds the two weeks
`together, we expect that it will mostly be a subset of the coding theory participants who will attend both
`weeks.
`
`WORKSHOP SCHEDULE
`Week 1: August 2-6, 1999 Monday Tuesday Wednesday Thursday Friday
`
`Week 2: August 9-13, 1999 Monday Tuesday Wednesday Thursday Friday
`
`All talks are in Lecture Hall EE/CS 3-180 unless otherwise noted.
`
`WEEK 1: CODES ON GRAPHS AND ITERATIVE DECODING
`August 2-6, 1999
`SCHEDULE for MONDAY, AUGUST 2
`HISTORY AND TUTORIALS Day
`G. David Forney, Jr. (chair)
`Registration and Coffee
`Reception Room EE/CS 3-176
`Willard Miller, Fred Dulles,
`and G. David Forney
`R. Michael Tanner
`University of California-Santa Cruz
`
`8:30 am
`
`9:10 am
`
`9:30 - 10:30 am
`
`Introduction and Welcome
`
`Error-Correcting Codes and
`Graph-based Algorithms:
`Origins, Successes, the
`Current Quests
`Reception Room EE/CS 3-176
`Markov Chains, Error Control,
`and the Advent of Turbo
`Coding
`
`10:30 am
`Coffee Break
`11:00 am - 12:00 pm Stephen B. Wicker
`Cornell University
`
`12:00 pm
`
`2:00-3:00 pm
`
`4:00 pm
`
`9:15 am
`
`Lunch
`Frank R. Kschischang
`University of Toronto
`
`IMA Tea
`
`Factor Graphs and the
`Sum-Product Algorithm
`IMA East, 400 Lind Hall
`A variety of appetizers and
`beverages will be served.
`SCHEDULE for TUESDAY, AUGUST 3
`LOW DENSITY PARITY CHECK CODES DAY
`R. Michael Tanner (chair)
`Coffee
`Reception Room EE/CS 3-176
`
`2 of 6
`
`10/11/2014 2:28 PM
`
`Hughes, Exh. 1037, p. 2
`
`

`
`Institute for Mathematics and its Applications (IMA) - 1999 IMA Summer...
`
`https://www.ima.umn.edu/csg/
`
`9:30-10:30 am
`
`David J.C. MacKay
`Cambridge University
`10:30 am
`Coffee Break
`11:00 am - 12:00 pm Robert J. McEliece
`California Institute of Technology
`
`Sparse Graph Codes
`
`Reception Room EE/CS 3-176
`Some Simple Codes that Are
`Good in Both Theory and
`Practice
`
`12:00 pm
`2:00 - 3:00 pm
`
`3:00 pm
`
`3:30 pm
`
`4:00 pm
`
`4:30 pm
`
`Analysis and Design of
`Iterative Decoding Systems
`
`Lunch
`Thomas J. Richardson
`(Lucent Bell Labs)
`Ruediger Urbanke
`(Lucent Bell Labs)
`Reception Room EE/CS 3-176
`Coffee Break
`Contributed Talks and Informal Discussions
`Amin Shokrollahi
`Capacity Achieving
`Low-density Erasure Codes
`Bell Labs
`Gilles Zemor
`Iterative Decoding of Cycle
`Codes of Graphs
`ENST, Paris
`Dakshi Agrawal
`On the Phase Trajectories of
`the Turbo Decoding Algorithm
`University of Illinois-Urbana Champaign
`SCHEDULE for WEDNESDAY, AUGUST 4
`INFERENCE DAY
`Brendan J. Frey (chair)
`
`9:15 am
`
`9:30 - 10:30 am
`
`Coffee
`Tommi Jaakkola
`Massachusetts Institute of Technology
`10:30 am
`Coffee Break
`11:00 am - 12:00 pm Radford M. Neal
`University of Toronto
`
`Reception Room EE/CS 3-176
`Variational Methods for
`Inference
`Reception Room EE/CS 3-176
`Sparse Matrix Methods and
`Probabilistic Inference
`Algorithms
`
`The Sum-Product Algorithm in
`Gaussian Networks with
`Cycles
`
`Heeralal Janwa and Oscar Moreno
`University of Puerto Rico
`
`12:00 pm
`2:00 - 3:00 pm
`
`3:00 am
`
`3:30 pm
`
`4:00 pm
`
`4:30 pm
`
`9:15 am
`
`Lunch
`Brendan J. Frey
`University of Waterloo
`Yair Weiss
`University of California at Berkeley
`Reception Room EE/CS 3-176
`Coffee Break
`Contributed Talks and Informal Discussions
`John B. Anderson
`Properties of the Tailbiting
`BCJR Decoder
`University of Lund
`Amir Banihashemi
`Tanner Graphs for Group
`Block Codes and Lattices:
`Carleton University
`Construction and Complexity
`New Constructions of
`Ramanujan Graphs and Good
`Expander Graphs from
`Codes, Exponential Sums and
`Sequences
`SCHEDULE for THURSDAY, AUGUST 5
`Robert J. McEliece (chair)
`Coffee
`Reception Room EE/CS 3-176
`Symbolic Boolean
`Manipulation with Ordered
`Binary Decision Diagrams
`Reception Room EE/CS 3-176
`Trellises, Decision Diagrams,
`and Factor Graphs
`
`9:30 - 10:30 am
`
`Randall E. Bryant
`Carnegie Mellon University
`
`10:30 am
`Coffee Break
`11:00 am - 12:00 pm John Lafferty
`Carnegie Mellon University
`Lunch
`
`12:00 pm
`
`2:00 - 3:00 pm
`
`James L. Massey
`ETH Zurich and Lund University
`
`3:00 am
`6:00 pm
`
`Coffee Break
`Workshop Dinner
`
`Linear Systems over Fields
`and Rings, Linear Complexity,
`and Fourier Transforms
`Reception Room EE/CS 3-176
`Bona Vietnamese
`Restaurant
`Located near the IMA and the
`Day's Inn at 802 Washington
`Avenue, the south side of
`Washington very near the
`
`3 of 6
`
`10/11/2014 2:28 PM
`
`Hughes, Exh. 1037, p. 3
`
`

`
`Institute for Mathematics and its Applications (IMA) - 1999 IMA Summer...
`
`https://www.ima.umn.edu/csg/
`
`8:45 am
`
`9:00 - 10:00 am
`
`G. David Forney, Jr.
`Massachusetts Institute of Technology
`
`intersection of Washington
`and Oak St.
`Phone: 612-331-5011
`SCHEDULE for FRIDAY, AUGUST 6
`CODING THEORY DAY Alexander Vardy (chair)
`Coffee
`Reception Room EE/CS 3-176
`Codes and Systems on
`Graphs: Generalized State
`Realizations
`Reception Room EE/CS 3-176
`Factor Graphs, Trellis
`Formations, and Generalized
`State Realizations
`Reception Room EE/CS 3-176
`Decoding and Equalization:
`Iterative Algorithms and
`Analog Networks
`
`10:00 am
`
`Coffee Break
`
`10:15 - 11:15 am
`
`Ralf Koetter
`University of Illinois at Urbana-Champaign
`
`11:15 am
`11:30 am
`
`Coffee Break
`Hans-Andrea Loeliger
`Endora Tech AG, Switzerland
`
`Week 1: August 2-6, 1999 Monday Tuesday Wednesday Thursday Friday
`Week 2: August 9-13, 1999 Monday Tuesday Wednesday Thursday Friday
`
`All talks are in Lecture Hall EE/CS 3-180 unless otherwise noted.
`
`Symbolic Dynamics and Automata
`
`WEEK 2: CONNECTIONS AMONG CODING THEORY, SYSTEM
`THEORY AND SYMBOLIC DYNAMICS
`August 9-13, 1999
`SCHEDULE for MONDAY, AUGUST 9
`Registration and Coffee
`Reception Room EE/CS 3-176
`Willard Miller, Fred Dulles,
`Joachim Rosenthal, and Brian Marcus Introduction and Welcome
`Automata and Systems
`Jorn Justesen (Chair)
`Roger W. Brockett
`Dynamical Systems and their
`Associated Automata
`Harvard University
`10:30 am
`Coffee Break
`Reception Room EE/CS 3-176
`11:00 am - 12:00 pm Dominique Perrin
`Université de Marne-la-Vallée
`Algebra and Geometry Applied to Systems
`Ethan Coven (Chair)
`Paul A. Fuhrmann
`A Polynomial Module Approach to
`Linear Systems Theory
`Ben Gurion University
`Clyde Martin
`Linear Systems as Vector Bundles
`on Spheres
`Texas Tech University
`Coffee Break
`Reception Room EE/CS 3-176
`M.S. Ravi
`An Algebraic Geometric Point of
`View to Linear Systems Theory
`Eastern Carolina University
`IMA East, 400 Lind Hall
`A variety of appetizers and
`beverages will be served.
`SCHEDULE for TUESDAY, AUGUST 10
`Coffee
`Reception Room EE/CS 3-176
`Convolutional Codes
`Karl Petersen (Chair)
`Woven Convolutional Codes:
`Encoder Properties and Error
`Exponents
`Construction of Convolutional
`Codes with Large Free Distance
`Reception Room EE/CS 3-176
`On Convolutional Codes over Rings
`
`8:30 am
`
`9:10 am
`
`9:30 am
`
`1:30 pm
`
`2:30 pm
`
`3:30 pm
`
`4:00 pm
`
`5:00 pm
`
`8:45 am
`
`9:00 am
`
`10:00 am
`
`11:00 am
`11:30 am
`
`IMA Tea
`
`Rolf Johannesson
`University of Lund
`
`Roxana Smarandache
`University of Notre Dame
`Coffee Break
`Fabio Fagnani
`Politecnico di Torino
`Joint talk with Sandro Zampieri
`Universita di Padova
`
`4 of 6
`
`10/11/2014 2:28 PM
`
`Hughes, Exh. 1037, p. 4
`
`

`
`Institute for Mathematics and its Applications (IMA) - 1999 IMA Summer...
`
`https://www.ima.umn.edu/csg/
`
`Contributed Talks
`Joachim Rosenthal (Chair)
`All talks will be 25 minutes long, including questions.
`
`Thomas Mittelholzer
`IBM Zurich Research Laboratory
`Sergio R. Lopez-Permouth
`Ohio University
`Coffee Break
`Danrun Huang
`St. Cloud State
`Dharmendra S. Modha
`IBM Almaden Research Center
`
`Duals over Artinian Rings and the
`MacWilliams Identity
`Finite Fields, Permutations and
`Trellis
`Reception Room EE/CS 3-176
`Period Three, Chaos, and the
`Golden Mean Shift
`Art of Constructing Low-complexity
`Encoders/Decoders for
`Constrained Block Codes
`
`On Encoding in DNA Words
`
`Natasha Jonoska
`University of South Florida
`SCHEDULE for WEDNESDAY, AUGUST 11
`Coffee
`Reception Room EE/CS 3-176
`Multidimensional Systems
`Jon Hall (Chair)
`
`Multi-dimensional Symbolic
`Dynamical Systems
`Capacity of Constrained Systems
`in One and Two Dimensions
`Reception Room EE/CS 3-176
`Multidimensional Convolutional
`Codes
`
`Klaus Schmidt
`University of Vienna
`Paul H. Siegel
`University of California-San Diego
`Coffee Break
`Paul A. Weiner
`Saint Mary's University of Minnesota
`Systems Theory
`Roy Adler (Chair)
`Jan C. Willems
`Systems, States and their
`Representations
`University of Groningen
`Reception Room EE/CS 3-176
`Coffee Break
`Sanjoy Mitter
`Path Space View of Probabilistic
`Systems
`MIT
`SCHEDULE for THURSDAY, AUGUST 12
`Coffee
`Reception Room EE/CS 3-176
`Symbolic Dynamics and Applications
`Uwe Helmke (Chair)
`M. Michael Boyle
`Applications of Symbolic Dynamics
`to the Structure Theory of
`University of Maryland
`Nonnegative Matrices
`
`Multiplicities of SFT Covers
`
`Codings of Markov Chains and
`Weighted Graphs
`
`Natasha Jonoska
`University of South Florida
`Selim Tuncel
`University of Washington
`Contributed Talks
`Brian Marcus (Chair)
`Marie-Pierre Béal
`A Finite State Version of the Kraft-
`McMillan Theorem
`Université de Marne-la-Vallée
`Olivier Carton
`Université de Marne-la-Vallée
`Coffee Break
`Christiane Frougny
`LIAFA
`
`2:00 pm
`
`2:30 pm
`
`3:00 pm
`3:30 pm
`
`4:00 pm
`
`4:30 pm
`
`8:45 am
`
`9:00 am
`
`10:00 am
`
`11:00 am
`
`11:30 am
`
`2:00 pm
`
`3:00 pm
`
`3:30 pm
`
`8:45 am
`
`9:00 am
`
`10:00 am
`
`11:30 am
`
`2:00 pm
`
`2:30 pm
`
`3:00 pm
`3:30 pm
`
`4:00-4:30 pm
`
`4:30 - 5:00 pm
`
`6:00 pm
`
`Michael E. O'Sullivan
`University College Cork
`Fernando Guzmán
`Binghamton University
`Workshop Dinner
`
`Asynchronous Sliding Block Maps
`
`Reception Room EE/CS 3-176
`Deterministic Synchronization of
`Bounded Delay 2-tape Finite
`Automata
`The Key Equation for One-point
`Codes
`Ambiguity in Codes
`
`Campus club
`Located on the 4th floor of Coffman
`Student Union and serves a
`wide-ranging buffet. Coffman Union
`is located on the opposite side of
`Washington Avenue from the IMA
`and slightly to the west.
`
`5 of 6
`
`10/11/2014 2:28 PM
`
`Hughes, Exh. 1037, p. 5
`
`

`
`Institute for Mathematics and its Applications (IMA) - 1999 IMA Summer...
`
`https://www.ima.umn.edu/csg/
`
`8:45 am
`
`9:00 am
`
`SCHEDULE for FRIDAY, AUGUST 13
`Coffee
`Reception Room EE/CS 3-176
`Decoding and Interpolation
`Zhe-Xian Wan (Chair)
`Margreet Kuijper
`Algorithms for Decoding and
`Interpolation
`University of Melbourne
`
`Patrick Fitzpatrick
`National University of Ireland, Cork
`
`10:00 am
`
`11:00 am
`
`11:30 am
`
`Realization and Interpolation via
`Gröbner Bases
`Reception Room EE/CS 3-176
`Coffee Break
`Brian M. Allen
`Linear Systems Decoding of
`Convolutional Codes
`University of Notre Dame
`Informal Contributed Talks
`Lind Hall 409 with an option to switch to Lecture Hall EE/CS 3-180 contingent on the participants size
`2:00 pm
`Karl Petersen
`Good Measures for Bad Codes
`between SFT's
`University of North Carolina
`Ethan Coven
`The Symbolic Dynamics of Tiling
`the Integers
`Wesleyan University
`Coffee Break
`IMA East Lind Hall room 400
`Kimberly Johnson
`Automata, and Pumping Lemmas
`for Beta-shifts
`University of North Carolina
`Paul Trow
`Mappings between Group Shifts
`University of Memphis
`
`2:30 pm
`
`3:00 pm
`3:15 pm
`
`3:45 pm
`
`Week 1: August 2-6, 1999 Monday Tuesday Wednesday Thursday Friday
`Week 2: August 9-13, 1999 Monday Tuesday Wednesday Thursday Friday
`
`Back to top of page
`
`© 2014 Regents of the University of Minnesota. All rights reserved.
`The University of Minnesota is an equal opportunity educator and employer
`Last modified on October 06, 2011
`
`Connect With Us:
`
`Search IMA
`
`Go
`
`Twin Cities Campus: Parking & Transportation Maps & Directions
`Directories Contact U of M Privacy
`
`6 of 6
`
`10/11/2014 2:28 PM
`
`Hughes, Exh. 1037, p. 6
`
`

`
`INSTITUTE FOR
`MAIHEMArJCS
`AND ITS
`APPLICArJONS
`
`Gallager Codes -Recent Results
`August 3, 1999
`
`A talk that was presented by David J.C. MacKay, an IMA Visitor from the Cambridge University
`
`Hughes, Exh. 1037, p. 7
`
`

`
`Callaser- codes
`-
`fecent re.~ults
`
`-- -·---
`
`···· .
`
`. . . . ··-·· · .. -·· -
`
`· ····
`
`.. ··· ·~· · ·
`
`((_c,dfcrd 1--i~.J
`. -..
`~·-Ar~'rl,·,!~·-·) t(t.·~) I
`.
`
`• How goocl
`fe,Swor c:;'cr.U~r
`c.ocle~ o.re.
`EnnMc.;.n~ c;oilo.~ codes.
`
`•
`
`• 0~ ~"'ons.
`
`• Ga'~ ~ ~ '-'•s~
`fote, ~-" \)\oc.\< ~~tat~.
`
`I
`
`Hughes, Exh. 1037, p. 8
`
`

`
`.
`
`Pt\tt' : r e.t·{'l;)r- ¥~e.·. Ci:>""'~"'·'!\(IA,-\-\(>~
`'!i +· .. ~:~ o"~r M i!I.Y ~h.r,.•"··.d:,
`
`.
`
`.
`
`Me. \~"\c:.."C.\ :
`-
`
`-
`
`.Swr~ s -
`
`t
`-
`
`_j
`
`t)~C.Ol)E.R. >.-,.
`I $
`l' -
`
`"(
`
`-
`
`c.MANN&L
`
`~ .
`~i$£. n ..
`
`tlt:\t\~
`f9,dul':m~:J
`
`~c;od f\t.LU~ : ~ rtd\4<£ tv.~.r
`'P.!~•\;ty ~ ·
`~6.0 "\.>..)~: RC!Ou.n~ ~o.n5..
`'(~ ~ i\ rtd~Ktcl
`
`Hughes, Exh. 1037, p. 9
`
`

`
`. Code ~~'~;;· ~--
`
`VH~~ Cooo
`
`--'"J
`
`. . ti
`
`Q)b6
`toA~~
`-p~ "-;-.. 0
`A$ R.~c
`
`••
`\
`c
`
`~te.
`
`~
`
`\?Y::
`
`GoaD~
`(M
`~->o
`C ~~ R>O
`
`c
`
`>
`R .
`
`c
`
`R
`
`Hughes, Exh. 1037, p. 10
`
`

`
`G;ct$ -
`
`(pod d•ron~ .
`
`i'
`t
`
`d
`• ~.
`-
`4S N .. oo
`k
`Ccx\e~ hOAJt \ltr~ c-c cts~
`~~ ~ ..
`~c.v
`N
`N
`+\.l (~f) = 1 - R
`
`u.~ hout aotJ ~
`~ ~ .... 0
`
`~~~d;~
`·{ d :: const.
`
`Hughes, Exh. 1037, p. 11
`
`

`
`s
`
`ENCODER
`
`t
`
`CH ANNEL
`f = 7.5%
`~lt01JNOA~
`"fZs~!!p,~_f
`
`r
`
`DECODF:R
`
`A s
`
`...... 1<3"
`
`The decoder· takes
`the majority vote of
`the three signals.
`
`.
`Good news: only 1.6% of decoded bits are in error
`.
`Bad news: rate of cotrun unication reduced to 1/3
`
`Hughes, Exh. 1037, p. 12
`
`

`
`(7,4-) code
`
`·' .
`
`I(= 4 Source bits
`(nccde..s
`lr\to N = 7 tr-o."smitted bil-s .
`
`- --
`
`P R R I\Y C.HEc.K. M ATRIX ·
`.
`
`t
`M =3. (01) ~; 1'\ts
`t
`
`r I 0 ' 0 0
`I
`H= 0
`-
`-
`l ' 0 0
`' 0
`
`l
`
`I
`
`l 0
`
`I 0
`I
`
`..._ \(;;: 4 - · ......__---
`~0~
`N\::3
`f"' i 1-y bits
`bits
`
`Hughes, Exh. 1037, p. 13
`
`

`
`( 7, L+) <.ode
`
`'" o.c. -\-\or
`
`s
`t
`s
`0000 0000 000 0100
`0001 0001 011 0101
`0010 0010 111 0110
`0011 0011 100
`0111
`
`s
`t
`0 100 110 1000
`0101 101
`1001
`0110 001 1010
`0111 010 1011
`
`t
`1000 101
`1001 110
`1010 010
`1011 001
`
`s
`t
`1100 1100 011
`1101 1101 000
`1110 1110 100
`1111 1111 111
`
`s
`
`ENCODER
`
`t
`
`CHANNEL
`f = 7.5%
`
`r
`
`DECODER
`
`-s
`
`parit.y bits
`
`I
`I
`
`4% of decoded bit.s arc in error
`
`rate of communication is 4/7
`
`The d~:coder picks
`the s with maximum
`likelihood.
`
`I
`J
`
`Hughes, Exh. 1037, p. 14
`
`

`
`Dec.odi~
`
`-·
`
`"'
`-
`
`mod 2.
`
`H r -
`
`- Ht
`-
`-
`-
`\. , ,
`
`- -
`-~o Hn
`-
`
`'th~ .. s'/rd.rT:>fY\e •
`~ #.e. ~~'!.d
`"$\~\(
`z = H r Moct 2
`
`=:::
`
`-
`
`H n -(cid:173)~
`
`-- z -
`
`.
`
`-
`
`-
`
`. .
`
`-
`
`-
`
`-
`
`-
`
`·-
`
`-
`
`.
`
`·,
`
`Hughes, Exh. 1037, p. 15
`
`

`
`G.\~ Cob~S
`c::.~\.Lf'<;~ll l'ibZ.
`
`.
`
`1 -
`1
`
`1
`
`~
`J
`
`.1
`
`1
`
`1
`1
`~ 1
`
`1 1.
`'
`•
`
`1
`
`1
`1
`1
`'
`1
`1 1
`1
`
`A
`
`'
`I.
`1
`1
`1 1
`l t11~
`1 11:
`~I ct~ eo~n ~ . ~t 3
`
`H
`
`==
`
`1
`
`1
`1
`
`1
`
`1
`
`•
`1
`
`1
`
`1
`
`• ~e s••s ~·~
`
`4;rM!o1'A<YI~ . \,\h
`
`.. . . .
`
`Hughes, Exh. 1037, p. 16
`
`

`
`.
`
`'
`
`'
`
`.
`
`'
`
`c;
`coda Wiff\ ~Ked t
`c;oob, OJ\" ~
`Clte
`"~ -\STANCE.
`
`!: 3
`
`¥-
`
`~\ t w.r.s, "~~ coNs
`~. "tR) C.ootJ~ w l,a.rt
`~tl] ~-· ~,w-..
`* + o.
`
`t.~~~ r~ of
`~ uilt.. a.J t.Jiko!Jf
`~~. AA.~.d(~ lq~<j
`
`Hughes, Exh. 1037, p. 17
`
`

`
`~ -to ~d~t ~- tJecod·.~ ~eM
`. .
`.
`. .
`.
`. ·
`(DI Fr! c.v.Ll")
`.
`.
`
`t .....
`
`•
`
`•
`
`9
`~~~
`
`·------
`
`~\te\inood rat"fOl
`'Jt/
`C!
`
`Hughes, Exh. 1037, p. 18
`
`

`
`'l'HE l::NCODER
`
`. We demonstrate a large code that en(cid:173)
`code::; K = 10000 source bits into
`N = 20000 transJlli tten bits.
`Each pl'l.rity bit depends ou about
`5000 source bits.
`The encoder is derived from a very
`spmse 10000 x 20000 matrix H with
`three l s per column.
`
`T H.A NSMITTED :
`
`p a ri ty b its
`
`'
`
`'
`
`•
`
`'
`
`, I
`
`'
`
`:••
`
`\
`
`•
`
`' ,
`
`'
`
`•
`
`,• .• •, · . ' , l', ' ' :
`
`.
`
`:"• .·.. .. . ·· ~ 7: > .. ~ :_.;·.:..· . : .· . ..
`' .· .. . ' :··'>'·
`. ·:
`.·.·.:. .. . .. •' .
`.· . ><· . .'::.: .. : :·. : . .-:·. ·->~:~: ' :: ... : ·:: .
`•', _:, ... ~ ~." .... ;· ... ; .. · .•. · .. :·. ~-.· :.·~.;·.:~; :.· '::.·: .. :
`~ .: . ' ··>:· .. :<: ~ ...... :':\
`.
`, . .
`. ..
`..
`, .
`.
`. · . .. r: :
`....
`· • '.·
`. ,
`•'
`. .
`.
`.
`..
`.
`<
`. .
`:-::··· · . .. ..
`.
`.·:
`.•
`··.· •·.•. · .
`..- ~ .. . ·':: .·.,.=··.··· .• ::~ ·-···
`. ···.':
`. ·· .. ·
`.; :< . '.~ ··. :··..
`.. ,
`· .. ·· . ..
`. ..
`.. ,.
`...
`.
`...
`. .. ' . .
`· .···· ·.
`.·.
`.
`..
`.
`· :· ,. _
`.
`. .
`.
`:
`. " .. -
`. '
`.. : . : .
`. .. ·. ..
`.
`. . . ..
`.. . .. · . ·: .
`, •• :. ~ ·
`. ..
`.
`· .. · ......
`· ::,-·,-:·· ••• ••
`·. . . .
`. .
`·:. . .
`.
`.
`, . . •'
`. : · · ·
`... · •... ·: .. ·
`.· ..
`,
`.
`.. .
`. ~·
`. ,•
`.·\ . ;_:
`.
`··:-
`·.~ .
`.: ..
`.... ·.· . .. ·· ~
`:' .·. ·_,_·.:·:: · ..... ·. · .·..
`.·
`. ·t. ....
`.
`.
`..... .
`. ·· ".: '\ .
`.
`. ... -· ... ·.·: .. · ..
`-
`..: .
`. . .
`. ...
`. ..
`. · ..
`· ..
`·.
`..
`. ·· ...... .... . :
`' ·: . . ,.•,.: :: ·~····
`. ·:''.' .:· .·· _:.:··:. :' . . .. . ·.: :.
`·. ·. /.:.-:· .... ·~.--;~ .. ···.· .... , .- ·, <· ;<. ·.:· .. ~.~·.:.~:~ .... ~:.·.·: .... : .. /_.~·.·'·· ·_._··.·.:::.:Y::·:::~ ~.::
`.. ··· .. :: ..
`·· . .. . ; .
`· .. .
`, . ..... . ;,: .
`.· ..
`<<· ... · : ... :-:~.:·· .. -.·.•.·.· ... :· ... ·.· .. ;.~
`. '. ·~ ·: ·. : .. . ' .... ·.··
`-=.: ·~:.· ..... , ..... ··.· · . . ,_:· ·': ·· ··.: .·~ ... ~.: : ..... _; .. ': . . ·' .::·: .. :· ·:: ..... _.:-::· ... ~: ... ·. :. " .... .
`· .......
`:-:.: . . : .. · !
`.
`... . . '
`.
`.. . :' . ... · -· .
`.
`. ..
`. . .
`.
`- .· .
`.. . .
`: ,
`'
`. \
`..
`·. ·.~ .. . : '\~ .... : ·.· .. · ·. ·.··, ·:: · .. .. ·.
`·-
`.
`.
`.
`...
`.
`·· ·:.: .'.: .- ~·· :·
`.. .... .,_·:.· . .. · ·:. ': ·.
`·. ;: "-.·: '•
`.· .. · .. ;. . -: :.:. · ..•.. :.:.-::: .. ·:.:· ... , --:·.:-.·~ •, ................. . . : -·.:· .,,.··· .. ... .
`' ' •
`• ,. ••I• ' .:· .. ... ...... ;
`.
`: · ·. · ..
`·. .. . . . . '.·;. \• . .
`•• ~.
`.. . ·-•• ·.·.·~ ••• ·,1·.,_.· .. ,· . .-.• ·.·.·:: ..• •• .... ·.:.·.···::
`..... ·,·.·.: ..... :. · .. . ' ~
`. .
`.....
`·. .
`.
`.
`.~
`· :-.. . ·:···:..:.··:= .. ··: .
`. :
`·:.
`•• • \ :. •
`:.-· : .>: _; -:: ;-
`... ·.:· ... :
`
`H
`
`.
`
`.
`
`-
`
`.
`
`'
`
`.
`•
`
`•
`
`•
`
`•
`
`,, •
`
`t' ···
`•
`
`. •
`•' '
`
`.•
`•
`
`•
`
`•'
`
`,:
`
`• , :'. .,•
`
`. •
`•' •
`
`.
`:. '•, . ,
`
`• •
`
`'
`
`:
`
`<
`
`•
`
`• • •
`
`. :
`
`• •
`
`•
`
`··: : : : .:
`
`Hughes, Exh. 1037, p. 19
`
`

`
`Jt.era tive dA~oding
`After the transmission is ~f'nL over a chrumel with noi;;c level f == 7.5%:
`
`RECEIVED:
`
`1
`
`J1
`
`12
`
`13
`
`D ~~CODED :
`
`This fi nal de<:oding is error free.
`To the ease of an um1sually noisy transmission, thf' decoding algorit.!tm
`foils to find a valid dc(;odi!lg. F'v; this code and a channel wi~h f :...: 7.5%,
`such failures !n:tp pen abom once in every 100,000 trausntissi(l!lS.
`
`Hughes, Exh. 1037, p. 20
`
`

`
`0.1
`
`0.01
`
`0.001
`
`....
`~
`8 Q)
`"0
`0
`~ 0.0001
`
`·(cid:173)---.0
`
`n1
`~ 1e-05
`c.
`
`le-06
`
`: +
`
`.+
`
`I
`I
`I
`
`+
`
`I
`
`I
`v
`
`Our low density
`.
`parity check cod(;
`
`J...-.--" Shannon limit
`
`0
`
`0.2
`
`0.6
`0.4
`Rate
`
`o.a R 1
`
`Hughes, Exh. 1037, p. 21
`
`

`
`breccde.r
`errch 4
`~s.c.
`---r ...... - .-,........-=. -.-..-.-.-.-. • ..---,.----~
`~' '·' ~ 1--
`f •.
`(log
`SCttfe)0.01000 ·: •
`•• • •
`...
`
`··-· ··~
`
`. . . .
`
`• • •
`
`'<
`
`··-
`
`••
`
`R
`
`' 'SCHIRM
`•
`Gallager -..
`Sha nnon -
`
`0.3
`
`0.4
`
`0 .5
`
`C HA P"T't:R 7 · -- ERRon coJuts:c"T'tr-;c
`
`( :<.>u E9 & R t AL c uAN N &:L s
`
`II:, -·~ ...
`
`o.1oooo
`
`0 .00100
`
`0 .00010
`
`I
`
`0 .00001
`
`0
`
`G V
`
`0 . 1
`
`I
`
`J
`RO
`0.2
`
`Hughes, Exh. 1037, p. 22
`
`

`
`~\r error-
`prob~bi \ i ty
`''de. _ro.cto standaro -for
`t
`~o..tQ..!(tte. C:OMMW'l iC.O..+ic:{)S"'
`- r----------...,t---------.o"
`o.or
`
`voyo.:J<-f"' ~
`(i,i)~I(S
`
`o.oo
`
`10-S"
`
`0 . I
`
`o.ol
`
`~ . oo r ·
`
`10-S'
`
`10 -~ 1----'--+--t----__._~r---'--------'----- 10-b
`o:r
`1
`2.
`0
`sgrol to noise mtio
`· C:Hr6)
`/n
`(p.\i\eo
`\~~ta.r
`(IS,~+ RS) Co.\\~(" <:.ode
`. dB
`C:,o..ti.O;:~er
`G:x:i~ / ('.t:(~)
`
`N::.~ ooo
`•
`
`Hughes, Exh. 1037, p. 23
`
`

`
`ERtt<>tU M~o& ~Y
`
`c.obes
`
`<;o.rl~gv
`
`.
`
`E.o..t" \tqro..~Otl ~t . ~ ~u.a.u \~
`0.~- it· H~"' ! ~ h~t~· .
`
`" )
`
`.2.)
`
`Ut't\~J~.c-J~A. f.frtJri .
`- ~"' a~
`
`=
`
`'"
`0-.
`~dt:b · ~.trots
`-
`attokr ~
`
`*
`~Ll EAAt\Q~ PrJf.
`
`1lf.T({."ff.~ ~QJ2~S ~
`
`~ do not hGAIQ
`
`~-
`
`Hughes, Exh. 1037, p. 24
`
`

`
`~ tiW'te.s .. •
`9~~ ~es
`
`.
`
`.
`
`LAW
`
`'t. :!F i !-ero..4 "'1.S
`re~~--.·d
`
`Hughes, Exh. 1037, p. 25
`
`

`
`1J ECOt::> I NCJ
`
`/-IAV£ A
`
`Bio error p r<>l>ability
`
`Word err<>r prol.>ability
`
`P rof. I.,
`
`N::\Cf20 "··· .
`R -
`I
`0.0001 .. ..,.
`- 3
`
`O.OCH
`
`Oe.9.A&f GF2 • ....-.
`lnCU'-Ifer G F=-4 • ··•···
`RA "''6J+·-
`
`R..OWl>
`s
`C.35
`G
`S
`
`Col u •n!l.S
`2
`44!>
`3
`389
`4
`72
`1
`fi
`11
`2
`13
`4
`19
`S9
`
`1e-07 ~~- r-~-.-~ - r--4'
`0 .8
`1.2
`1.4 M:O
`1.0
`2
`1
`
`(a)
`
`\ ··~
`Figure 1. (a,b) Perfor mAnce c:>f Gall><ger cudes with N = 1920, .H = 1/3 , M " function o[ E•/No . rn (b) WP. also
`I
`\
`~how t he p elrfornuutce o! A rcp~Ht-accumulate coUtt ~:it.h N =-: 3000. (t.:) '"~'bH" profile o f the lrrHgulo.r codcc
`over GF(4) .
`f
`~
`
`J,-8
`
`2'
`
`(c)
`
`P(1:.) 0.2
`
`0 .15
`0.1
`0.0:;
`0
`
`0 .1
`0.01
`1e-3
`1e-4
`1<>-5
`
`0 10 20 so 40 50
`- 3
`
`0 10 20
`
`0.2
`0.1S
`0 .1
`o.os
`0
`
`0. I
`0.01
`1C•3
`1e-4
`le-5
`
`0 10 20 30 ~0 50
`
`--
`
`0 .2
`0.15
`0 .1
`0 .05
`0
`
`0.1
`0.01
`1&-3
`1e--'\
`1e-s
`
`0 .2 -
`0.15
`-
`0 . 1 -
`o.os -
`0
`
`0
`
`0.1
`0.01
`1e~
`1e-4
`1e-s
`
`10 20 30 40 50 'r
`-c..-10
`
`(a)
`
`(b)
`
`10
`
`100
`
`10
`
`100
`
`10
`
`100
`
`10
`
`100
`
`z /u
`(E./No)/<IU
`1-'w
`Po\ver, p
`
`O.!JO
`0.841.1
`O.lSJ
`3
`
`0.95
`1 .~ I:;
`0 .014
`1
`
`LOU
`1.7 61
`1.7 x 1o-•
`6
`
`}.05
`2 .185
`!1.0 x l o-u
`10
`
`FigtJr• ~- Hi~tograms s howing the tr~QUCncy d is tributi-on o r d~oding t.bn~ for t11E:!- binary Gatl~cr r..odc from figuNII
`l: (a) linear p lot; (b) l og-log plo<. The graphs ~he-w the numher o! i ter o.tinns Lalcen tn rc=h a valid
`rlpcodin g; the vc:.Jue of PVI give-s: th<H frcq uc.:mC}" wi th whidl. no v11.lid decoding wa.'-,1 reached ~r 1000
`it.ertt.t.ivns. T he vower J) which gives :l good fit of the J,Jt J".\"er lAw d.isl.ribution P(•) oc .,.--p (fo-r farge -r)
`is a lsu shown.
`
`2
`
`Hughes, Exh. 1037, p. 26
`
`

`
`25000
`
`20000
`
`15000
`
`10000
`
`5000
`
`(a)
`
`' '
`
`G.f"
`-rs~
`
`"\\
`
`•
`~""' \
`\ \ t.. .
`+ \ \
`~,
`X\
`x~. +"\
`' 4. \ #
`
`10000 ~
`1000 ~ .
`+
`
`...
`
`100 ~
`+
`
`10 1-
`
`0.1 ..
`
`I •
`
`(b)
`
`10 20 30 40
`
`50
`
`it'etW'ionS
`
`•
`
`Hughes, Exh. 1037, p. 27
`
`

`
`···-· · ·-
`Vor~\ng. co\U~n" we.~t t
`
`-··
`
`....
`
`-·
`
`' '
`
`--
`
`·-·
`
`. ·--
`
`--
`
`. .....
`
`1
`
`0.01
`
`0.001
`
`0.0001
`
`1e-05
`
`N:::.B 16, A-: 1/2: vary I
`
`1e-06 L.-... _
`1.5
`
`-
`
`__. -
`2
`
`-L-- -~ - --L--
`2.5
`3
`3.5
`
`Hughes, Exh. 1037, p. 28
`
`

`
`.
`R=1/2 1=3 (33)
`
`total~-
`undetected ···· .. ., ..... .
`
`0.1
`
`0.01
`
`0.001
`
`0.0001
`
`1e -05
`
`1 e-Q6
`
`-·- --···l'- -
`1
`2
`
`3
`
`4
`
`5
`
`N=204
`
`6EbfNo
`
`bloc.k \en~#) N
`
`R=1/3 t=4
`
`total -----7<(cid:173)
`undetected .. · 'lK· ..
`
`Block
`.ttrrcrr
`prob.
`
`1
`
`0.1
`
`0 .01
`
`0.001
`
`0.0001
`
`1e-05
`
`1e-06 ..
`
`1e-07
`
`...
`
`2
`
`Hughes, Exh. 1037, p. 29
`
`

`
`Eb I N 0
`-fir fl•vvt. P(br .. eJ.. E"m~v-)
`~Jfrvtce :lol'l1 s~ t:VIttrm
`
`/; 41. if
`
`0~
`
`Wh~r block
`~(ll)1'"S
`W')deh~.c..t-e.d ef"'r'tliS .
`
`o
`
`o
`
`0
`
`Hughes, Exh. 1037, p. 30
`
`

`
`''~~\~ ~ 'ouws ~ T~ ~•., ·
`W~Ct.r ooes better meon ?.
`
`a~t to Sho.nnon hm&t ~/No T
`1or ~Ntn t0\'"ol Qtc or f'O.fe
`
`"W"t'~ :
`
`'bt:~'f£D ~ R:I\\E ,.. ~Nb~a.TEI> ER~ Uf£
`
`~\uk' undU'tc.t"tA ~ ro.tt
`~ 1'1\\t\\1'1\WY\ c~o.~
`~\~ . d~,5
`~('~
`Wdt ~lt"iblt 0. ' N,ct
`
`T
`
`~
`G
`Cs
`
`~
`
`Hughes, Exh. 1037, p. 31
`
`

`
`E.NCOD•Nct
`
`N . ~
`•
`.a w K ---~:>~4 M ~
`" I
`
`. . . ' •
`
`-- =
`H
`
`l
`
`·
`
`·
`
`•
`
`• I
`
`1
`1
`I
`I ' t
`I I J
`I
`'
`
`.
`
`(ac-cm~ ~~t~t;,oJ,t~n
`0(N3 )
`
`-- U'
`
`Ht:o
`••
`
`•
`.. Ht•o
`••
`
`Ji -e~-J
`~·,
`••
`••
`'•
`
`..
`
`I
`!
`•
`
`0(N2
`)
`'if Edens~
`-
`
`.
`
`- F'
`
`_, .
`•
`
`
`-·
`-
`
`s
`
`r M
`.£
`
`I
`
`Hughes, Exh. 1037, p. 32
`
`

`
`.
`
`.
`
`Sur &\At~ codes ore bod ,
`( M0'j ~hr c&ewas)
`
`Hughes, Exh. 1037, p. 33
`
`

`
`-
`
`-
`
`H --
`
`-
`
`-
`
`-
`
`- - .J( - --
`
`- N--
`
`-
`
`- - ·-
`
`- -- - -
`-
`- - -
`- ---- ·
`- -.
`--
`- -
`
`ll
`
`- -
`
`I
`I
`l\1
`I
`
`I
`
`.~t I
`I I
`
`-
`-·C·
`
`t.
`I
`I .
`
`o
`
`I t
`
`~ '
`.. •
`
`•
`
`"
`
`t
`
`:
`
`>
`
`i•
`
`.
`
`~~' p
`I ;.
`!• • • •,•}
`•, , ,
`' l : '
`
`o o
`, I
`, .
`
`'•
`'•
`
`---~·-
`
`>
`•
`
`'
`
`1 ' l ••• .,,.-·••••••-" > OJ - 1,.,,,,+••••- • ' 0 ,,J,,,,,,'(',·; ·:
`i
`0
`)o
`
`~","'''''''''''''
`
`• •
`
`•••
`
`. . .
`
`i' ,
`' '·
`i
`'i
`.:
`
`.•
`
`l
`!
`l
`!
`'
`
`l
`
`I '
`
`' I ..
`•• .. . ' . '
`
`"
`
`' .
`: '
`
`!."l'o
`..
`: ·,
`· : :
`.
`••
`•
`
`.. •
`
`: o
`
`I
`
`• • t •
`~:~t\ . :,i ·~ ..
`' "
`•• t· ~:·a:t ··.,.
`-·-··- -·-1.
`..·~.:·:·.~ :·~!·
`····t·
`.... I ·' ..
`:: .·· .. .
`..
`[ .
`·· ··! ..
`.
`• •
`. .,.~.:.: ....... --·· . .. ...... J ......... -··. t:'~ :, ;: :::··~.:.:.:.::::, ,, -·-····
`• • . '
`• •
`..
`. •'
`I
`t ·l····: .......................... ,.~.~i"''~' .. ~ ·"····-- ..
`.
`I
`•
`,,
`.. .. .
`. I' .
`. . .
`• •
`
`•
`
`I
`
`t
`
`0
`
`0
`
`I
`
`I
`
`I
`
`0
`
`0
`
`'
`
`:
`
`I
`
`•
`• .. . ·1
`.. r
`
`•' •
`F-AST- ENC.ODINC.,
`
`Hughes, Exh. 1037, p. 34
`
`

`
`...
`
`rc·nc·x'a''rttt . ..,.. __
`
`. -·-·
`
`=== · -
`
`•
`
`C.\~Mp ~~ ~,gt~ MO
`tcuc.k
`<:oC"'A.lo..\\ons cluri~ dt~j
`
`Hughes, Exh. 1037, p. 35
`
`

`
`0 .1
`
`0.01
`
`0.001
`
`>.
`-
`:=:::
`:0
`Cll
`.0
`e
`c..
`.... e
`....
`UJ
`'
`:=: co
`(ti 0 .0001
`0
`·;:: ·-
`0..
`E
`UJ
`
`Turbo,
`1e-OS Rate 1/4
`
`.
`!.
`+MacKo.y <(g
`
`/:x..ve
`
`c;F(~)
`GF(2)
`
`'
`·.
`
`GF(4)
`
`'.
`
`\
`
`•
`\
`
`'
`'
`
`l
`
`~~
`GF(16)
`<;F(It,) \
`\
`' \
`'
`\
`\
`\
`\
`'
`\
`'
`\
`· ~
`\
`\ Turbo,
`I \Rate 1/3
`\
`\
`l
`
`<::;F(S).
`GF(8)
`
`1e-06
`
`-0.2
`
`0
`
`0.2
`
`0.4
`SNR/dB
`
`0.6
`
`0.8
`
`1
`
`Hughes, Exh. 1037, p. 36
`
`

`
`.......................
`
`.
`
`........... __ :
`
`........... = ·=--=mr::
`
`- - ~
`
`t
`
`MG.ke. 9~h •rf!za\AI\o..;
`l..~l:J/ Ato\iw.~ ~~~--, &
`.tT ~ )
`( <S" ie.\ M.O.I'o
`l)cw~::J M~ MO-L~ )
`
`• • • •
`
`~ . .. .
`
`o. pri'lil~
`b\t
`l»f'~'~pctt(S
`\n MO.t\J CoAstm:."ts
`
`Hughes, Exh. 1037, p. 37
`
`

`
`0. 1 ~
`
`0.01
`
`0 .001
`
`lrreg
`0.0001 GF{8}
`
`1e-05
`
`,_
`0 ....
`..... w
`
`' -
`
`c.
`E w
`
`r
`
`Reg
`GF(16)
`
`Turbo
`~Q,+e \;'""
`
`-0.2-0.1 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9
`SNR/db
`
`Figure 1: Empirica1results for Gaussian Channel, Rate 1/4 Left-Right: Irregular LDPC,
`GF(8) block!ength 24000 bit~; J PL Turbo, blocklength 65536 bits; Regular LDPC,
`a
`GF (16), blocklength 24448 bit.s: ln-egular LDPC , GF(2), blocklength 649.9.0 bits; Reg-
`ular LDPC, GF(2), blocklength 4.QQOQ hit3. (Reproduced from [1].)
`
`-
`
`Hughes, Exh. 1037, p. 38
`
`

`
`Col. Weight
`fraction
`
`2
`29
`23
`15
`3
`7
`9
`11
`43
`19
`0.700 0.191 0.037 0.024 0.013 0.017 0.008 0.005 0.003 0.002
`
`Table 4 .1: Parameters of good irregular codr. for G.F(8), rate 0.25
`
`Ro.+e '14-
`
`0.1
`
`0 .01
`
`Turbo, 1/4
`
`0 .001
`
`0.0001
`
`GF(4)
`•1/2
`
`£ -·-
`
`n;
`.g
`0.
`E w
`
`<;o.l l~r ·ci::des :
`c\\ error.s ~
`<k\W"~ . erroc:s
`'"'nx, axles
`f'r\O.'I<.e. W'd~
`~rn:x-5
`
`1e-os
`
`Turbo
`• GF(8), 1/4
`1/2
`1 e-06 '-'---J...:,...--l"---'-~---"'----..1.---'
`-0.2
`0
`0.2 0.4 0.6 0.8
`1
`SNR/db
`
`Figure 4-2: Current state of the art in error correction. Left to r ight: H.nte 1/4 irregular LDPC
`over GF'(S) bl...,c:klength 24000 bits; Rate 1/4 JPL turbo code fRi blnck!ength 65536; Rate .J./ 2
`
`Hughes, Exh. 1037, p. 39
`
`

`
`2......
`
`______
`
`____..
`
`_:___
`
`3+...
`
`______
`
`31.1..
`
`____n_
`
`__.___._
`
`____..
`
`_______
`
`_.__.___...
`
`____..
`
`____..
`
`21.1.
`
`__
`
`._I
`
`_____._
`
`..._._.._.....
`
`___.______.__ .._.____ w..____.:________.__,_____._
`
`___.__...
`._.___.
`__.__~
`______
`2....
`
`
`
`
`
`
`HII4......“..
`Z.___
`___:.._l|_—
`
`___I___
`
`-1:-E-—--u-.—a——u-.---—
`
`I—_n...--._-
`
`I
`_--__.g.__
`u'
`—II-—r—-I—-I--u-—---uh-|—u—-—
`
`____.__
`
`___L___
`
`_______
`
`.IL__....—_._...._..._.|!
`
`
`.F.._._.__.._Ft_.I__—L_I_!_I.I
`
`_________
`
`:______
`
`_.__..._____.._
`
`._:.__.
`
`In:I
`
`._+_T._._
`
`..:n..__
`
`_._+__.
`
`.2...
`
`_______
`..E.
`
`._.__.._._..
`
`__.__.__
`
`._...____
`
`.._.___.1__|._I_
`
`.__.____._._
`
`.__.__~__
`
`________
`
`.___..-__
`
`____q_._
`
`__t_.__
`
`__...w:._J:.
`
`2......
`
`‘._.__.._._.
`
`___:_._
`
`EH59
`
`0.153
`
`U. M?
`
`E1142
`
`{H35
`
`1
`
`* dgon,
`
`Ehcakmilah
`
`.
`
`.
`
`"
`
`'
`
`map (Q=*ӎ)
`0125
`{H31
`ruwcé Lir'bar1L-1e.
`(N=.lO )
`
`'
`
`*“
`
`6
`
`Hughes, Exh. 1037, p. 40
`
`

`
`7-Z
`'\\
`\\
`
`7 --
`
`•
`
`bir e (rt\r
`r~bcJ)\ 1,; ~J 1 o