Modern Channel Coding

Similar documents
Integration Continued. Integration by Parts Solving Definite Integrals: Area Under a Curve Improper Integrals

ME 522 PRINCIPLES OF ROBOTICS. FIRST MIDTERM EXAMINATION April 19, M. Kemal Özgören

Lecture 11 Waves in Periodic Potentials Today: Questions you should be able to address after today s lecture:

INTEGRALS. Chapter 7. d dx. 7.1 Overview Let d dx F (x) = f (x). Then, we write f ( x)

A general N-dimensional vector consists of N values. They can be arranged as a column or a row and can be real or complex.

A Sub-Optimal Log-Domain Decoding Algorithm for Non-Binary LDPC Codes

This Week. Computer Graphics. Introduction. Introduction. Graphics Maths by Example. Graphics Maths by Example

Computational Fluid Dynamics

Instructions for Section 1

SOLVED EXAMPLES. be the foci of an ellipse with eccentricity e. For any point P on the ellipse, prove that. tan

Winter 2016 COMP-250: Introduction to Computer Science. Lecture 23, April 5, 2016

Assignment 5-6 For Problems 1-4, find a general solution of each differential equation. *Solve for y in Problems 2 and 4.

Multi-Section Coupled Line Couplers

TMMI37, vt2, Lecture 8; Introductory 2-dimensional elastostatics; cont.

12. Traffic engineering

CIVL 8/ D Boundary Value Problems - Rectangular Elements 1/7

TOPIC 5: INTEGRATION

PH427/PH527: Periodic systems Spring Overview of the PH427 website (syllabus, assignments etc.) 2. Coupled oscillations.

3.4 Repeated Roots; Reduction of Order

COMP108 Algorithmic Foundations

Limits Indeterminate Forms and L Hospital s Rule

HIGHER ORDER DIFFERENTIAL EQUATIONS

Minimum Spanning Trees

Section 3: Antiderivatives of Formulas

Multiple-Choice Test Introduction to Partial Differential Equations COMPLETE SOLUTION SET

Text: WMM, Chapter 5. Sections , ,

Continuous Random Variables: Basics

Basic Polyhedral theory

FSA. CmSc 365 Theory of Computation. Finite State Automata and Regular Expressions (Chapter 2, Section 2.3) ALPHABET operations: U, concatenation, *

CONIC SECTIONS. MODULE-IV Co-ordinate Geometry OBJECTIVES. Conic Sections

CSC Design and Analysis of Algorithms. Example: Change-Making Problem

CSE 373: AVL trees. Warmup: Warmup. Interlude: Exploring the balance invariant. AVL Trees: Invariants. AVL tree invariants review

Addition of angular momentum

Minimum Spanning Trees

, between the vertical lines x a and x b. Given a demand curve, having price as a function of quantity, p f (x) at height k is the curve f ( x,

Theory of Spatial Problems

The Theory of Small Reflections

Functions of Two Random Variables

Why the Junction Tree Algorithm? The Junction Tree Algorithm. Clique Potential Representation. Overview. Chris Williams 1.

Lecture contents. Bloch theorem k-vector Brillouin zone Almost free-electron model Bands Effective mass Holes. NNSE 508 EM Lecture #9

The Trapezoidal Rule

SER/BER in a Fading Channel

cycle that does not cross any edges (including its own), then it has at least

Errata for Second Edition, First Printing

Electricity and Magnetism Electric Dipole Continuous Distribution of Charge

UNIT # 08 (PART - I)

CSE303 - Introduction to the Theory of Computing Sample Solutions for Exercises on Finite Automata

FINITE ELEMENT ANALYSIS OF

Polygons POLYGONS.

Errata for Second Edition, First Printing

INF5820/INF9820 LANGUAGE TECHNOLOGICAL APPLICATIONS. Jan Tore Lønning, Lecture 4, 14 Sep

DUET WITH DIAMONDS COLOR SHIFTING BRACELET By Leslie Rogalski

What do you know? Listen and find. Listen and circle. Listen and chant. Listen and say. Lesson 1. sheep. horse

Walk Like a Mathematician Learning Task:

Computer Vision. Fourier Analysis. Computer Science Tripos Part II. Dr Christopher Town. Fourier Analysis. Fourier Analysis

Page 1. Question 19.1b Electric Charge II Question 19.2a Conductors I. ConcepTest Clicker Questions Chapter 19. Physics, 4 th Edition James S.

LINEAR ALGEBRA APPLIED

Functions and Graphs 1. (a) (b) (c) (f) (e) (d) 2. (a) (b) (c) (d)

1. Twelve less than five times a number is thirty three. What is the number

CONTINUITY AND DIFFERENTIABILITY

Fundamentals of Continuum Mechanics. Seoul National University Graphics & Media Lab

Engineering 323 Beautiful HW #13 Page 1 of 6 Brown Problem 5-12

Image Enhancement in the Spatial Domain

Self-Adjointness and Its Relationship to Quantum Mechanics. Ronald I. Frank 2016

MATHEMATICS FOR MANAGEMENT BBMP1103

1 Introduction to Modulo 7 Arithmetic

Last time: introduced our first computational model the DFA.

Slide 1. Slide 2. Slide 3 DIGITAL SIGNAL PROCESSING CLASSIFICATION OF SIGNALS

Chapter 16. 1) is a particular point on the graph of the function. 1. y, where x y 1

ELEG 413 Lecture #6. Mark Mirotznik, Ph.D. Professor The University of Delaware

Chem 104A, Fall 2016, Midterm 1 Key

Addition of angular momentum

( ) Geometric Operations and Morphing. Geometric Transformation. Forward v.s. Inverse Mapping. I (x,y ) Image Processing - Lesson 4 IDC-CG 1

22.615, MHD Theory of Fusion Systems Prof. Freidberg Lecture 8: Effect of a Vertical Field on Tokamak Equilibrium

Problem Set 6 Solutions

Graphs. CSC 1300 Discrete Structures Villanova University. Villanova CSC Dr Papalaskari

Chapter 6 Notes, Larson/Hostetler 3e

Case Study VI Answers PHA 5127 Fall 2006

Density Evolution and Functional Threshold. for the Noisy Min-Sum Decoder

Strongly Connected Components

12/3/12. Outline. Part 10. Graphs. Circuits. Euler paths/circuits. Euler s bridge problem (Bridges of Konigsberg Problem)

glo beau bid point full man branch last ior s all for ap Sav tree tree God length per down ev the fect your er Cm7 a a our

Mathematics. Mathematics 3. hsn.uk.net. Higher HSN23000

5/9/13. Part 10. Graphs. Outline. Circuits. Introduction Terminology Implementing Graphs

Paths. Connectivity. Euler and Hamilton Paths. Planar graphs.

b. How many ternary words of length 23 with eight 0 s, nine 1 s and six 2 s?

CSE 373: More on graphs; DFS and BFS. Michael Lee Wednesday, Feb 14, 2018

Introduction - the economics of incomplete information

ES.182A Topic 30 Notes Jeremy Orloff

Lecture 21 : Graphene Bandstructure

a b c cat CAT A B C Aa Bb Cc cat cat Lesson 1 (Part 1) Verbal lesson: Capital Letters Make The Same Sound Lesson 1 (Part 1) continued...

MASTER CLASS PROGRAM UNIT 4 SPECIALIST MATHEMATICS SEMESTER TWO 2014 WEEK 11 WRITTEN EXAMINATION 1 SOLUTIONS

Floating Point Number System -(1.3)

ARITHMETIC OPERATIONS. The real numbers have the following properties: a b c ab ac

Floating Point Number System -(1.3)

2008 AP Calculus BC Multiple Choice Exam

Quantum Mechanics & Spectroscopy Prof. Jason Goodpaster. Problem Set #2 ANSWER KEY (5 questions, 10 points)

COMPUTER GENERATED HOLOGRAMS Optical Sciences 627 W.J. Dallas (Monday, April 04, 2005, 8:35 AM) PART I: CHAPTER TWO COMB MATH.

y cos x = cos xdx = sin x + c y = tan x + c sec x But, y = 1 when x = 0 giving c = 1. y = tan x + sec x (A1) (C4) OR y cos x = sin x + 1 [8]

Collisions between electrons and ions

Transcription:

Modrn Chnnl Coding Ingmr Lnd & Joss Sir Lctr 4: EXIT Chrts ACoRN Smmr School 27 Itrti Dcoding How dos th mtl informtion ol in n itrti dcoding lgorithm? W h lrnd tht it is possibl to optimiz LDPC cods so s to mimiz thir thrshold W will s tht w cn dsign cpcit-chiing, itrtil dcodbl fmilis of LDPC cods!! (i.., thrshold cpcit) Wht is th impliction in trms of mtl informtion?

Mtl Informtion Trjctor I(Xi;Y[it]) 2 3 4 5 6 7 8 9 Itrtions Mtl Informtion Trjctor Th L-ls clcltd in th tr r optiml in th sns of MAP-clcltor, i.., L(X i Y [it] ) is sfficint sttistic for Y [it] : I(X i ; L(X i Y [it] )) = I(X i ; Y [it] ) W cn lso drw th trjctor t hlf-itrtions (ftr ribl nods & ftr chck nods) Bt: th otpt mssgs of ribl nods nd chck nods r trinsic L-ls, whrs th mtl informtion trjctor w considr now is for -postriori L-ls

Mssg Pssing L ch L V A Vribl Nod L V E L C APP L C A Chck Nod L C E Trcking of Mssgs I () E This ssms tht th dcodr dpnds onl on mtl informtion! I (2) A I A () I E (2) Problm: How to compt th trnsfr fnctions f nd f 2?

Trcking of Mssgs Trcking of mssgs wold mn trcking of pdfs ( Dnsit Eoltion) Instd of trcking th pdfs w rdc th problm to trcking of mtl informtion btwn th mssgs nd th codword which r sclr qntitis I A,I E... rg smbolwis mtl informtion Etrinsic Chnnl Modl Enc Enc 2 comm. ch tr. ch pp A-priori mssgs r modld s indpndnt nois obsrtions of th ncodd sorc. Assmptions: - indpndnt obsrtions - modl for trinsic chnnl with qlit if th dcodr is optiml

Trnsfr Fnctions Enc Enc 2 comm. ch tr. ch pp Assming modl for th trinsic chnnl w cn r I A b ring th chnnl prmtr. At th otpt of th dcodr w cn msr/clclt I E I E = f(i A ) This is onl ct if th modl of th trinsic chnnl is corrct! EXIT Chrt of LDPC Cod.9.8.7 I A,chk.5.6 I E,r.4.3.2...2.3.4.5.6.7.8.9 I A,r I E,chk

Intrscting Crs.9.8.7 I A,chk.5.6 I E,r.4.3.2...2.3.4.5.6.7.8.9 I A,r I E,chk Etrinsic Informtion Trnsfr Chrts (Stphn tn Brink) (Stphn is th g on th right, not th clown on th lft) Photo b Joss Sir Stphn did his PhD t th U of Stttgrt, thn workd for Bll Lbs in th U.K., thn in Nw Jrs. H is crrntl with RlTk. H is rglr isitor of ftw. nd TU Win.

Trcking of Mssgs I () E This ssms tht th dcodr dpnds onl on mtl informtion! I (2) A I A () I E (2) Problm: How to compt th trnsfr fnctions f nd f 2? Trcking of Mssgs Trcking of mssgs wold mn trcking of pdfs. Instd of trcking th pdfs w rdc th problm to trcking of mtl informtion btwn th mssgs nd th codword which r sclr qntitis. I A,I E... rg smbolwis mtl informtion

Etrinsic Chnnl Modl Ch + + Dc + + + () pp () - + + - () π π + + (2) pp (2) + + - + Dc 2 + + - (2) π - Enc Enc 2 comm. ch tr. ch pp Etrinsic Chnnl Modl Enc Enc 2 comm. ch tr. ch pp A-priori mssgs r modld s indpndnt nois obsrtions of th ncodd sorc. Assmptions: - indpndnt obsrtions - modl for trinsic chnnl with qlit if th dcodr is optiml

Trnsfr Fnctions Enc Enc 2 comm. ch tr. ch pp Assming modl for th trinsic chnnl w cn r I A b ring th chnnl prmtr. At th otpt of th dcodr w cn msr/clclt I E I E = f(i A ) This is onl ct if th modl of th trinsic chnnl is corrct! Vribl Nods nd BEC rp d BEC q BEC p pp Etrinsic chnnl is modld s BEC (ct)..9.8.7.6.5 I E d.4 = 3, 4, 5.3.2...2.3.4.5.6.7.8.9 I A

Chck Nods nd BEC SPC d c BEC p pp SPC... singl prit chck.9.8.7 I A.6.5.4 d c = 6, 8,.3.2...2.3.4.5.6.7.8.9 I E EXIT Chrt of LDPC Cod.9.8.7 I A,chk I E,r.6.5.4.3.2...2.3.4.5.6.7.8.9 I A,r I E,chk

Othr Chnnls Modling th trinsic chnnl s BEC is ct if th commniction chnnl is BEC. For othr commniction chnnls, th ssmption of th trinsic chnnl is in gnrl n pproimtion. If th commniction chnnl is n AWGN chnnl, w modl th trinsic chnnl lso s n AWGN, bt this is onl n pproimtion! AWGN Chnnl ribl nods AWGN σ c rp d AWGN σ pp chck nods SPC d c AWGN σ pp

Conoltionl Cods Stphn tn Brink, Conrgnc Bhior of Itrtil Dcodd Prlll Conctntd Cods, IEEE Trns. Comm. Octobr 2 Sril / Prlll Conctntion Ch + + Dc + + + () pp () - + + - () π switchs opn sril conctntion π switchs closd prlll conctntion + + (2) pp (2) + + - + Dc 2 + + - (2) π - Sril conctntion: = pp - Prlll conctntion: = pp -

Informtion Combining BEC Wht is th ffct on mtl informtion whn w dd L-ls? SRC δ BEC LLR L δ 2 BEC2 LLR L 2 + I = I(X;L ) = - δ I 2 = I(X;L 2 ) = - δ 2 I(X;L L 2 ) = - δ δ 2 = (-I )(-I 2 ) Intrscting Crs.9.8.7 I A,chk I E,r.6.5.4.3.2...2.3.4.5.6.7.8.9 I A,r I E,chk

.. BER from EXIT Chrt (BEC) pp = + I(X;APP) = P b = (-I A )(-I E ).9.5...5...8.2..5.7.2.6.3.4.5.2..5.4.6.4.3.3.5.2..8.6.3..5.2.9.5.7..2.3.4.5.6.7.8.9.4 Informtion Combining AWGN σ SRC AWGN LLR L

Informtion Combining AWGN SRC σ AWGN LLR L σ 2 AWGN2 2 LLR L 2 + BER from EXIT Chrt (AWGN).9 E b /N =.db P b =.3 E b /N =.db P b =.7.8.7.6.5.4.3.5.6.7.8..2.3.4.5.9..6.7.8.2.9.3.4.5.5.6..2.7.8..9.5.6.8.7.2.6.7.8..9.5..2.9.3.2..6..2.3.4.5.6.7.8.9

Indpndnt Obsrtions Mssgs rcid from th trinsic chnnl r indpndnt obsrtions, which is onl flfilld if N Sttistics W s sttisticl qntitis, which r onl corrct if N P b thrshold E b /N

Smmr of Assmptions - Mssgs rcid from th trinsic chnnl r indpndnt obsrtions, which is onl flfilld if N - W s sttisticl qntitis, which r onl corrct if N - W modl trinsic mssgs with n trinsic chnnl. This cn onl b don ct for th BEC. Th Gssin ssmption is n pproimtion. Ar Proprt Enc comm. ch Enc 2 BEC p pp I E I A

Drition of Ar Proprt Drition of Ar Proprt 2

Drition of H(V Y) Vribl Nods rp d comm. ch BEC p pp

Chck Nods SPC d c BEC p pp Ar of LDPC Componnt Cods I E I A I A I E Ncssr condition for sccssfl dcoding:

Consqncs of Ar Proprt Srprising rslt: Th r proprt tlls s tht th dcodr cn onl conrg if th rt is smllr thn cpcit! Mor Consqncs... Sppos th condition for conrgnc is flfilld γ< Wht is th rslt of this inqlit?

Ar nd Rt Loss If γ w cn trnsmit t rts tht pproch cpcit. If γ < w r bondd from cpcit. γ mns tht - A = A c Frthrmor, th crs mst not intrsct. Th crs h to b mtchd. Cod dsign rdcs to cr fitting! Cr Fitting Cod Mitr W onl considrd rglr cods, whr r smbol hs th sm proprtis. Thrfor, rging or ll smbols is qilnt to th mtl informtion of n rbitrril smbol. If w prtition m into n grops j=...n ch with lngth l j, w cn writ I E s Th rslting EXIT fnction is th wightd rg of th EXIT fnctions of th grops.

Empl Vribl Mitr rp d BEC q BEC p pp 7% of th ribl nods h d =2 3% of th ribl nods h d =5 This is polnomil in p Not tht γ j = Empl Vribl Mitr.9 I E2.8.7 I E.6 I E.5.4.3.2...2.3.4.5.6.7.8.9 I A

Cr Fitting Lts fi th EXIT fnction of th chck nod dcodr. For cr fitting, w cn chng th following qntitis Thrfor, w cn writ th EXIT fnction of th ribl nod dcodr s th inrs EXIT fnction of th chck nod dcodr. Tlor Sris Epnsion Assming for mpl d c =5 w cn pnd I E s Tlor sris Trncting th Tlor sris nd normlizing th cofficints to rslts in Compr this with th trnsfr fnction of th mitr of ribl nods...

Cr Fitting.9.8.7.6.5.4.3.2...2.3.4.5.6.7.8.9 En mor Consqncs... Using th sm modl s for th ribl nd chck nod dcodr, it cn b shown tht th rs for sril conctntd cod with n otr cod R ot =k ot /n ot nd n innr cod R in =k in /n in r gin b Th sm ncssr condition -A ot < A in lds to If th innr cod hs rt <, i.. I(X;Y)/n in <C thn w cn not chi cpcit with sril conctntd cods!