AU2013203761B2 - Method and apparatus for channel encoding and decoding in a communication system using low-density parity-check codes - Google Patents

Abstract

Abstract A method for channel decoding using a Low-Density Parity-Check (LDPC) code, the method comprising: demodulating a signal transmitted from a transmitter; determining position information of punctured parity bits by estimating information about a predetermined 5 order of parity bit sets to be punctured and the number of parity bit sets; and decoding data using the positions of the punctured parity bits, wherein determining position information of punctured parity bits comprises: determining a number of parity bits to be punctured; determining a number of parity bit sets to be punctured based on the determined number of parity bits to be punctured; and acquiring a predetermined order of parity bit sets. 4246168 1 (GHMatters) P84756.AU.1 11/04/2013 [Figure 9] - - -innn nni.i o a a O 0 0 0 0 0 0 Oj-O -0 o a 0 0 0 O 0 0 0 0O O r- 0 0 O CD----D----D-C>---<D-C----------- D------------------- a O~4: r40 0 :0 0 0 0 CD CD 00 C: CBBBBC: D DC> -ri0lri -i0r 0 0 0 0j: c::) C.' 1 :O -: C-|C.:::..: C-:: C:)~~ ~ ~ +D ci.) C: ..I C) C:. ) oco-0 0 0 0 0 Or-' 0 0 0 0 0 0 0 0 0 0 00- ,-O O O 0 0 0 0 0 0 O r - 0 0 O Osr 0 0 * .- :. -n : Cr0 00O O ,- 0 0 0 0 0 0 0 Z3 O O C O - -- C O C O O 0 0 O O 0 0 0 0 0 0 0 0 OO 0 i. - . C)< > :) c : CI., I>nC> (Z)CD -- O s- ,- Oj O iOfO O &O O O1. O O O 0 -- 0 airO O O O 0 0 0 0 0 - -n O O O O O O O 0 >- OCD O O 0 0 0 0 0 0 0 OC> :C.-": C> (:" Z4 : > C) " * . *n --- i - -:" -.- ~C aZ: CD a> 0 0i) - O O - -- O O O -r -r--iO .. n O Oyns - O ,- 4 ,MOO .nO O OniO 0- 0 O Or O O 0 0 0 0 0 0r 0 0 0|r -:. r .. ro-::oOO11:+: C > C) CD CO O O D I,- Oz i :- O: C,- O l.- M I.A:. i.. CD ~ ~ ~ ~ ii 44 1.. .(Z. .z: C: = ).C)O i.M. -N.. P.M.. i+ 1 :r: CD C- CDO O O: CD C O D C> k' - O r I .". -s M IN I a .i::"'4 " . "I || ".E. q~t c:>'. c::;:: c> j .) l- T; rri:

Description

DESCRIPTION Invention Title METHOD AND APPARATUS FOR CHANNEL ENCODING AND DECODING IN A COMMUNICATION SYSTEM USING LOW-DENSITY PARITY-CHECK CODES 5 Related Application This application is a divisional application of Australian application no. 2009217933 the disclosure of which is incorporated herein by reference. Most of the disclosure of that application is also included herein, however, reference may be made to the specification of application no. 2009217933 as filed or accepted to gain further understanding of the 10 invention claimed herein. Technical Field The present invention relates to a communication system using Low-Density Parity Check (LDPC) codes. More particularly, the present invention relates to a channel encoding/decoding method and apparatus for generating LDPC codes with various 15 codeword lengths and code rates from an LDPC code given in high-order modulation. Background Art In wireless communication systems, link performance significantly decreases due to various noises in the channels, a fading phenomenon, and Inter-Symbol Interference (ISI). Therefore, in order to realize high-speed digital communication systems, which require high 20 data throughput and reliability, such as the next-generation mobile communication, digital broadcasting, and portable Internet, it is important to develop technology for overcoming the channel noises, fading, and ISI. Recently, an intensive study of error-correcting codes has been conducted as a method for increasing communication reliability by efficiently recovering distorted information. 25 An LDPC code, first introduced by Gallager in 1960s, has lost favor over time due to its implementation complexity which could not be resolved by the technology at that time. However, as the turbo code, which was discovered by Berrou, Glavieux, and Thitimajshima in 1993, exhibits performance levels approximating Shannon's channel limit, research has been conducted on iterative decoding and channel encoding based on a graph along with 30 analyses on performance and characteristics of the turbo code. With this as a momentum, the LDPC code was restudied in the late 1990s, which proved that the LDPC code exhibits performance levels approximating the Shannon's channel limit if the LDPC undergoes decoding by applying iterative decoding based on a sum-product algorithm on a Tanner graph (a special case of a factor graph) corresponding to the LDPC code. 35 The LDPC code is commonly represented using a graph representation technique, and many characteristics can be analyzed through the methods based on graph theory, algebra, and probability theory. Generally, a graph model of channel codes is useful for description of codes, and by mapping information on encoded bits to vertexes in the graph and mapping 4246168 1 (GHMatters) P84756.AU.1 11/04/2013 2 relations between the bits to edges in the graph, it is possible to consider the graph as a communication network in which the vertexes exchange predetermined messages through the edges, thus making it possible to derive a natural decoding algorithm. For example, a decoding algorithm derived from a trellis, which can be regarded as a kind of graph, may 5 include the well-known Viterbi algorithm and a Bahl, Cocke, Jelinek and Raviv (BCJR) algorithm. The LDPC code is generally defined as a parity-check matrix, and can be expressed using a bipartite graph, which is referred to as a Tanner graph. The bipartite graph is a graph where vertexes forming the graph are divided into two different types, and the LDPC code is 10 represented with the bipartite graph consisting of vertexes, some of which are called variable nodes and the other of which are called check nodes. The variable nodes are one-to-one mapped to the encoded bits. A graph representation method for the LDPC code will be described with reference to FIGs. 1 and 2. 15 FIG. 1 illustrates an example of a parity-check matrix H, of an LDPC code with 4 rows and 8 columns. Referring to FIG. 1, because the number of columns is 8, the parity-check matrix H 1 signifies an LDPC code that generates a length-8 codeword, and the columns are mapped to 8 encoded bits on a one to one basis. FIG. 2 illustrates a Tanner graph corresponding to the parity-check matrix H, in FIG. 1. 20 Referring to FIG. 2, the Tanner graph of the LDPC code includes 8 variable nodes x, (202), x 2 (204), x 3 (206), x 4 (208), x 5 (210), x 6 (212), x 7 (214), and x 8 (216), and 4 check nodes 218, 220, 222, and 224. An ith column and aih row in the parity-check matrix H, of the LDPC code are mapped to a variable node xi and aih check node, respectively. In addition, a value of 1, i.e. a non-zero value, at the point where an /h column and ajth row in the parity 25 check matrix H, of the LDPC code cross each other, indicates that there is an edge between the variable node xi and the jh check node on the Tanner graph as shown in FIG. 2. In the Tanner graph of the LDPC code, a degree of the variable node and the check node indicates the number of edges connected to each respective node, and the degree is equal to the number of non-zero entries in a column or row corresponding to the pertinent 30 node in the parity-check matrix of the LDPC code. For example, in FIG. 2, degrees of the variable nodes x, (202), x 2 (204), x 3 (206), x 4 (208), x 5 (210), x 6 (212), x 7 (214), and x 8 (216) are 4, 3, 3, 3, 2, 2, 2, and 2, respectively, and degrees of check nodes 218, 220, 222, and 224 are 6, 5, 5, and 5, respectively. In addition, the numbers of non-zero entries in the columns of the parity-check matrix H, of FIG. 1, which correspond to the variable nodes in 35 FIG. 2, coincide with their degrees 4, 3, 3, 3, 2, 2, 2, and 2, and the numbers of non-zero entries in the rows of the parity-check matrix H, of FIG. 1, which correspond to the check nodes in FIG. 2, coincide with their degrees 6, 5, 5, and 5. 4246168 1 (GHMatters) P84756.AU.1 11/04/2013 3 In order to express degree distribution for the nodes of the LDPC code, a ratio of the number of degree-i variable nodes to the total number of variable nodes is defined as f 1 , and a ratio of the number of degree-j check nodes to the total number of check nodes is defined as g 1 . For example, for the LDPC code corresponding to FIGs. I and 2, f2=4/8, fa=3/8, f4=1/8, 5 and fj=0 for i0 2, 3, 4; and g,=3/4, g&=1 14, and g;=0 for j$ 5, 6. When a length of the LDPC code, i.e. the number of columns, is defined as N, and the number of rows is defined as N/2, the density of non-zero entries in the entire parity-check matrix having the above degree distribution is computed as shown in Equation (1). 2f2N+ 3f 3 N + 4fN _ 5.25 N-NP/2 N . .. .(1) 10 In Equation (1), as N increases, the density of 1's in the parity-check matrix decreases. Generally, as for the LDPC code, because the codeword length N is inversely proportional to the density of non-zero entries, the LDPC code with a large N has a very low density of non zero entries. The term "low-density" in the name of the LDPC code originates from the above-mrentioned relationship. 15 Next, with reference to FIG. 3, a description will be made of characteristics of a parity check matrix of a structured LDPC code applicable to the present invention. FIG. 3 illustrates an LDPC code adopted as the standard technology in Digital Video Broadcasting-Satellite transmission 2' generation(DVB-S2), which is one of the European digital broadcasting standards. 20 In FIG. 3, N 1 and K, denote a codeword length and an information length (or a length of an information word) of an LDPC code, respectively, and (N 1

-K

1 ) provides a parity length. Further, integers M, and q are determined to satisfy q-(N-K 1 )/M. Preferably, K 1 /M, should also be an integer. The parity-check matrix in FIG. 3 will be referred to herein as a first parity-check matrix H 1 , for convenience only. 25 Referring to FIG. 3, a structure of a parity part, Le. Kid column through (Nri-)f column, in the parity-check matrix, has a dual diagonal shape. Therefore, as for degree distribution over columns corresponding to the parity part, all columns have a degree '2, except for the last column having a degree '1'. In the parity-check matrix, a structure of an information part, Le. &* column through 30 (K-1 )fh column, is made using the following rules. Rule 1: A total of K 1 /M, column groups are generated by grouping K, columns corresponding to the information word in the parity-check matrix into multiple groups each including M, columns. A method for forming columns belonging to each column group follows Rule 2 below. 424lii8 1 CHu.sisPB47SGA1.1 tt'14.2013 4 Rule 2: First, positions of '1's in each Oth column in /h column groups (where i=1,...,K 1

/M

1 ) are determined. When a degree of a 0 th column in each fh column group is (1) p(2) denoted by D;, if positions of rows with 1 are assumed to be , positions Rj (k = 1,2,..., Dj)of rows with 1 are defined as shown in Equation (2), in a fh row (where 5 j=1,2,...,M 1 -1) in an t' column group. R =k) R _)+ q mod(N - K 1 ), i10j-1) % k =1,2,..., D, i=1,n..., K, /M/1, j1..M . . (2) According to the above rules, it can be appreciated that degrees of columns belonging to an ith column group are all equal to D;. For a better understanding of a structure of a DVB 10 S2 LDPC code that stores information on the parity-check matrix according to the above rules, the following detailed example will be described. As a detailed example, for N 1 =30, K1=15, M 1 =5, and q=3, three sequences for information on the positions of rows with 1 for Othcolumns in 3 column groups can be expressed as follows. Herein, these sequences are called "weight-1 position sequences." R(1) = 0,5 Rf =1, R ' 2 1,0 3,0 R(1) =0, R (2 =11, R(3=13, R(1)=0, R (2 =10, R ()=14. 15 Regarding the weight-1 position sequence for 0 th columns in each column group, only the corresponding position sequences can be expressed as follows for each column group. For example: 01 2 20 011 13 0 10 14. In other words, the th weight-1 position sequence in the thline sequentially represents information on the positions of rows with 1 in the th column group. It is possible to generate an LDPC code having the same concept as that of a DVB-S2 25 LDPC code in FIG. 4, by forming a parity-check matrix using the information corresponding to the detailed example, and Rules 1 and 2. It is known that the DVB-S2 LDPC code designed in accordance with Rules 1 and 2 can be efficiently encoded using the structural shape. Respective steps in a process of 4246168_1 (GHMatters) P84756.AU.1 11/04/2013 5 performing LDPC encoding using the DVB-S2 based parity-check matrix will be described below by way of example. In the following description, as a detailed example, a DVB-S2 LDPC code with

N

1 =16200, K1=10800, M 1 =360, and q=15 undergoes an encoding process. For convenience, 0i 1 ii - K -1) 5 information bits having a length K 1 are represented as , and parity bits having a length (N 1

-K

1 ) are expressed as (Pon 11, - , AT, -Ki -1 Step 1: An LDPC encoder initializes parity bits as follows: PA =P1 = =PN 1 -K- = 0 Step 2: The LDPC encoder reads information on rows where 1 is located in a column 10 group from a Oth weight-1 position sequence out of the stored sequences indicating the parity-check matrix. 0 2084 1613 1548 1286 1460 3196 4297 2481 3369 3451 4620 2622 Rp) = 0, _R= 2048, R = 1613 R 4 =1548, R = 1286, "1' -, 6,,0 -10 R, -1460, R") = 3196, R% = 4297, R'9)= 2481, R"" = 3369, Rll =3451, R11 -4620, R"13 = 2622. 15 The LDPC encoder updates particular parity bits px in accordance with Equation (3), R(k) using the read information and the first information bit io. Herein, x is a value of 1,0 for k =12,...,13 P 0 = 9 o P284 = P2 0 64 0 , P1613 = 513 1P9 0 O P1548 1548 + 9 P 1286 1286 C p P 1 460

P

1460 + 0 , P3196 -P3196 ') P4 97 P4297 0+ ' 2481 P2481 ®+0, P3369 =P3369 9 o P34s1= P 345 1 c 'D, P4620 P462o 9 0, P2622 P2622 9 ......... (3) Pr p, @ i0 Pr<+-P,9i 20 In Equation (3), can also be expressed as , and represents binary addition. 4246168_1 (GHMatters) P84756.AU.1 11/04/2013 6 Step 3: The LDPC encoder first determines a value of Equation (4) for the next 359 information bits im (where m=l, 2, ..., 359) after io. {x+ (mimodM 1 ) x q}mod(NT 1 - K 1 ), 1 =360, m=n1,2,...,359 (4) 5 In Equation (4), xis a value of R. . It should be noted that Equation (4) is similar to Equation (2). Next, the LDPC encoder performs an operation similar to Equation (3) using the values found in Equation (4). In other words, the LDPC encoder updates parity bits P(r+(m modMj)xq)mod(Nr -K,) for im. For example, for m=l, i.e. for i1, the LDPC encoder 10 updates parity bits E(x+q)mod(NV-Kj) as defined in Equation (5). Pi 5 = PEis 1 n 20 = P 20 99 + i, P 1 62 8 = P 162 8 0+ i1, P1563 -P1563 .0 1 P1301 =P1301 1' P1475 = P1475 ( '1,

P

32 11

=P

32 11 1 312 4 3 12 .j 1

P

24 9 6 2496 '1

P

33 8 4

=P

3 38 4 tj1 P3466 =P 34 66 C1 p 4 635

=

4 635 1 1

P

26 3 7

P

2 637 1 4 (5) It is to be noted that q=15 in Equation (5). The LDPC encoder performs the above process for m=l, 2, ..., 359, in the same manner as shown above. 15 Step 4: As in Step 2, the LDPC encoder reads information of the 1 s weight-1 position -R (k =,2.,13) sequence 210 (k for a 361s information bit 360, and updates particular RMk parity bits px, where x is 2 . The LDPC encoder updates p m =361,'3 62 7... 19 Px+(m modM)xq jmod(NM-K 1 )' ' -2 "7 by similarly applying Equation (4) to the next 359 information bits 361, 362, ... , '719 after i360. 20 Step 5: The LDPC encoder repeats Steps 2, 3, and 4 for all groups each having 360 information bits. Step 6: The LDPC encoder finally determines parity bits using Equation (6). p,=p, TD p,_,, =1, 2,...,NAT - K -1 .......... (6) The parity bits p; of Equation (6) are parity bits that have completely undergone LDPC 25 encoding. 4246168_1 (GHMatters) P84756.AU. 1 11/04/2013 7 As described above, DVB-S2 carries out encoding through the process of Steps 1 to 6. In order to apply the LDPC code to the actual communication system, the LDPC code should be designed to be suitable for the data rate required in the communication system. Particularly, not only in an adaptive communication system employing Hybrid Automatic 5 Retransmission reQuest (HARQ) and Adaptive Modulation and Coding (AMC), but also in a communication system supporting various broadcast services, LDPC codes having various codeword lengths are needed to support various data rates according to the system requirements. However, as described above, the LDPC code used in the DVB-S2 system has only .0 two types of codeword lengths due to its limited use, and each type of the LDPC code needs an independent parity-check matrix. For these reasons, there is a long-felt need in the art for a method for supporting various codeword lengths to increase extendibility and flexibility of the system. Particularly, in the DVB-S2 system, transmission of data having several hundreds to thousands of bits is needed for transmission of signaling information. However, .5 because only 16200 and 64800 are available for lengths of the DVB-S2 LDPC code, there is a still a need for support of various codeword lengths. Yet, since storing independent parity check matrixes for respective codeword lengths of the LDPC code may reduce memory efficiency, there is a need for a scheme capable of efficiently supporting various codeword lengths from the given existing parity-check matrix, without designing a new parity-check !0 matrix. Summary of the Invention In accordance with one aspect of the present invention, there is provided a method for channel decoding using a low-density parity-check (LDPC) code, the method comprising: demodulating a signal transmitted from a transmitter; 25 determining a number of punctured parity bits; determining a number of punctured parity bit sets based on the determined number of the punctured parity bits; acquiring a predetermined order of punctured parity bit sets; determining position information of the punctured parity bits based on the 30 predetermined order; and decoding data using the position information of the punctured parity bits. In accordance with another aspect of the present invention, there is provided an apparatus for channel decoding using a low-density parity-check (LDPC) code, the apparatus comprising: 35 a demodulator for demodulating a signal transmitted from a transmitter; a puncturing pattern estimator for determining a number of punctured parity bits, determining a number of punctured parity bit sets based on the determined number of the punctured parity bits, acquiring a predetermined order of punctured parity bit sets, and 6138860_1 (GHMatters) P84756.AU.1 SARAHVV 22/01/2015 8 determining position information of the punctured parity bits based on the predetermined order; and a decoder for decoding data using the position information of the punctured parity bits. Other aspects, advantages, and salient features of the invention will become apparent 5 to those skilled in the art from the following detailed description, which, taken in conjunction with the annexed drawings, discloses embodiments of the invention. One or more embodiments of the present invention can generate a separate LDPC code with a different codeword length by optimizing encoding/decoding performance using information on a given parity-check matrix in the communication system that uses high-order .0 modulation and LDPC codes. Description of Drawings The above and other aspects, features and advantages of certain exemplary embodiments of the present invention will be more apparent from the following description taken in conjunction with the accompanying drawings, in which: .5 FIG. 1 illustrates an example of a parity-check matrix of a length-8 LDPC code; FIG. 2 illustrates a Tanner graph for a parity-check matrix of a length-8 LDPC code; FIG. 3 illustrates a DVB-S2 LDPC code; FIG. 4 illustrates an example of a parity-check matrix of a DVB-S2 LDPC code; (a) of FIG. 5 illustrates a signal constellation for QPSK modulation used in a digital !0 communication system; (b) of FIG. 5 illustrates a signal constellation for 16-QAM modulation used in a digital communication system; (c) of FIG. 5 illustrates a signal constellation for 64-QAM modulation used in a digital communication system; 25 FIG. 6 is a block diagram of a transceiver in a communication system using an LDPC code; FIG. 7 illustrates an example where random puncturing is applied to the LDPC code of FIG. 4; FIG. 8 illustrates another example where non-random puncturing is applied to the 30 LDPC code of FIG. 4; FIG. 9 illustrates another example where non-random puncturing is applied to the LDPC code of FIG. 4; 6138860_1 (GHMatters) P84756.AU.1 SARAHVV 22/01/2015 9 FIG. 10 illustrates another example of a parity-check matrix of a DVB-S2 LDPC code; FIG. 11 illustrates an example of a puncturing pattern determined by considering BPSK or QPSK transmission in the LDPC code of FIG. 10; FIG. 12 illustrates an example of a puncturing pattern determined by considering 16 5 QAM transmission in the LDPC code of FIG. 10; FIG. 13 illustrates an example of a puncturing pattern determined by considering 64 QAM transmission in the LDPC code of FIG. 10; FIG. 14 is a flowchart illustrating a method for generating an LDPC code with a different codeword length from a parity-check matrix of a stored LDPC code according to an 10 embodiment of the present invention; FIG. 15 is a flowchart illustrating an LDPC decoding method in a reception apparatus according to an embodiment of the present invention; FIG. 16 is a block diagram of a transmission apparatus using a punctured/shortened LDPC code according to an embodiment of the present invention; and 15 FIG. 17 is a block diagram of a reception apparatus using a punctured/shortened LDPC code according to an embodiment of the present invention. Throughout the drawings, the same drawing reference numerals will be understood to refer to the same elements, features and structures. Mode for Invention 20 The following description with reference to the accompanying drawings is provided to assist in a comprehensive understanding of embodiments of the invention as defined by the claims and their equivalents. It includes various specific details to assist in that understanding but these are to be regarded as merely exemplary. Accordingly, those of ordinary skill in the art will recognize that various changes and modifications of the 25 embodiments described herein can be made without departing from the scope and spirit of the invention. In addition, descriptions of well-known functions and constructions are omitted for clarity and conciseness. The terms and words used in the following description and claims are not limited to the bibliographical meanings, but, are merely used by the inventor to enable a clear and 30 consistent understanding of the invention. Accordingly, it should be apparent to those skilled in the art that the following description of embodiments of the present invention are provided for illustration purpose only and not for the purpose of limiting the invention as defined by the appended claims and their equivalents. It is to be understood that the singular forms "a," "an," and "the" include plural referents 35 unless the context clearly dictates otherwise. Thus, for example, reference to "a component surface" includes reference to one or more of such surfaces. Reliability difference in high-order modulation will first be described before a description of the present invention. Attention should be paid in designing an LDPC code 4246168 1 (GHMatters) P84756.AU.1 11/04/2013 10 because reliabilities of bits constituting high-order modulation symbols are different when high-order modulation is used in the communication system requiring LDPC codes with various codeword lengths, unlike in the communication system employing only Binary Phase Shift Keying (BPSK) or Quadrature Phase Shift Keying (QPSK). 5 In order to explain the reliability difference in high-order modulation, a description of signal constellations for Quadrature Amplitude Modulation (QAM) which is high-order modulation commonly used in communication systems will be provided. A QAM-modulated symbol consists of a real part and an imaginary part, and various modulation symbols may be generated by differentiating magnitudes and signs of their real parts and imaginary parts. 10 QAM will be described together with QPSK modulation to consider the details of QAM characteristics. (a) of FIG. 5 illustrates a signal constellation for general QPSK modulation. Referring to (a) of FIG. 5, yo determines a sign of a real part while y, determines a sign of an imaginary part. That is, a sign of the real part is plus (+) for yo=O, and minus (-) for yo=1. 15 Also, a sign of the imaginary part is plus (+) for y 1 =O, and minus (-) for y 1 =1. Since yo and y, are equal in error occurrence probability as they are sign indication bits that indicate respective signs of the real part and the imaginary part, reliabilities of (yo, y 1 ) bits corresponding to one modulation signal are equally important, in QPSK modulation. For yo,q and yi,q, the second subscript index q indicates qth outputs of bits constituting a modulation 20 signal. (b) of FIG. 5 illustrates a signal constellation for general 16-QAM modulation. Referring to (b) of FIG. 5, the meaning of (yo, yi, y 2 , y 3 ), corresponding to bits of one modulation signal, is as follows. Bits yo and y 2 determine a sign and a magnitude of the real part, respectively, while bits y, and y 3 determine a sign and a magnitude of the imaginary 25 part, respectively. In other words, yo and y, determine signs of the real part and imaginary part of the modulation signal, while y 2 and y 3 determine magnitudes of the real part and imaginary part of the modulation signal. Because distinguishing a sign of a modulated signal is easier than distinguishing a magnitude of the modulated signal, y 2 and y 3 are higher than yo and y, in error occurrence probability. Therefore, in terms of non-error occurrence 30 probabilities (i.e. reliabilities) of the bits, yO=y 1 >y 2 =y 3 . That is, bits (yo, y 1 , y 2 , y 3 ) constituting a QAM modulation signal, unlike those of a QPSK modulation signal, have different reliabilities. In 16-QAM modulation, among the 4 bits constituting a signal, 2 bits determine signs of the real part and imaginary part of the signal and the remaining bits only need to determine 35 magnitudes of the real part and imaginary part of the signal. Thus, orders of (yo, yi, y 2 , y 3 ) and a role of each bit are subject to change. (c) of FIG. 5 illustrates a signal constellation for general 64-QAM modulation. 4246168 1 (GHMatters) P84756.AU.1 11/04/2013 11 Of (yo, y 1 , Y2 Y3, Y3, 5 ), corresponding to the bits of one modulation signal, bits yo, Y2 and y4 determine a magnitude and a sign of the real part, and yi, y 3 and y 5 determine a magnitude and a sign of the imaginary part. Here, yo and y1 determine signs of the real part and the imaginary part, respectively, and a combination of y 2 and y 4 and a combination of y 3 5 and y 5 determine magnitudes of the real part and the imaginary part, respectively. Because distinguishing signs of a modulated signal are easier than distinguishing magnitudes of the modulated signal, reliabilities of yo and y 1 are higher than reliabilities of y 2 , y 3 , y 4 , and y 5 . The bits y 2 and y 3 are determined depending on whether a magnitude of the modulated symbol is greater or less than 4, and the bits y 4 and y 5 are determined according to whether the 10 magnitude of the modulated symbol is closer to 4 or 0 with 2 centered there between, or closer to 4 or 8 with 6 centered there between. Thus, a range in which the magnitude is determined by y 2 and y 3 is 4, while a range for y 4 and y 5 is 2. As a result, y 2 and y 3 are higher than y 4 and y 5 in reliability. In conclusion, yo=y 1 > y 2 =y 3 > y 4 =y 5 in terms of non-error occurrence probabilities (i.e. reliabilities) of the bits. 15 In 64-QAM modulation, of the 6 bits constituting a signal, 2 bits determine signs of the real part and imaginary part of the signal, and 4 bits only need to determine magnitudes of the real part and imaginary part of the signal. Accordingly, orders of (yo, y 1 , y 2 , y 3 , y 4 , y 5 ) and a role of each bit are subject to change. Also, in a signal constellation of 256-QAM or higher, the roles and reliabilities of bits constituting a modulation signal are different as described 20 above. A detailed description thereof will be omitted. Therefore, the present invention provides a method and apparatus for supporting LDPC codes with various codeword lengths suitable for high-order modulation, using a parity-check matrix of a structured LDPC code in a particular form. In addition, the present invention provides an apparatus for supporting various codeword lengths according to high 25 order modulation in a communication system using an LDPC code in a particular form, and a method for controlling the same. In particular, the present invention provides a method for generating an LDPC code using a parity-check matrix of a given LDPC code, the generated LDPC code being less than the given LDPC code, and an apparatus thereof. FIG. 6 is a block diagram of a transceiver in a communication system using an LDPC 30 code. Referring to FIG. 6, a message u is input to an LDPC encoder 611 in a transmitter 610 before being transmitted to a receiver 630. The LDPC encoder 611 encodes the input message u, and outputs the encoded signal c to a modulator 613. The modulator 613 modulates the encoded signal c, and transmits the modulated signal s to the receiver 630 35 over a wireless channel 620. A demodulator 631 in the receiver 630 demodulates the received signal r, and outputs the demodulated signal x to an LDPC decoder 633. The LDPC decoder 633 finds an estimate of the message based on the data received through the wireless channel 620 by decoding the demodulated signal x. 4246168_1 (GHMatters) P84756.AU.1 11/04/2013 12 The LDPC encoder 611 generates a parity-check matrix according to a codeword length required by a communication system, using a preset scheme. Particularly, in accordance with an embodiment of the present invention, the LDPC encoder 611 may support various codeword lengths using the LDPC code without the separate need for 5 additional stored information. In accordance with an embodiment of the present invention, a method of acquiring various codeword lengths from a given LDPC code uses shortening or puncturing. The term "puncturing" as used herein indicates a method that does not transmit a specified part of an LDPC codeword after generating the LDPC codeword from a given particular parity-check 10 matrix by performing LDPC encoding. Hence, a receiver determines that the non-transmitted bits were erased. For a better understanding of the puncturing, a parity-check matrix of the DVB-S2 LDPC code illustrated in FIG. 3 will be described in more detail below. Regarding the parity-check matrix of the DVB-S2 LDPC code illustrated in FIG. 3, its 15 total length is N 1 , K 1 columns in the leading part of the parity-check matrix correspond to information bits 1 , and columns in the remaining rear part correspond to length-(N 1

-K

1 ) parity bits Generally, the puncturing can be applied to both the information bits and the parity bits. Although the puncturing and the shortening commonly reduce codeword lengths, the 20 puncturing, unlike the shortening, described herein above, does not limit values of particular bits. The puncturing is a method for simply not transmitting particular information bits or particular parts of generated parity bits, so that a receiver can erase the corresponding bits. In other words, by simply not transmitting bits in Np predefined positions in a generated length-N 1 LDPC codeword, the puncturing can obtain the same effect as that obtained by 25 transmitting a length-(N 1 -Np) LDPC codeword. Because columns corresponding to the bits punctured in the parity-check matrix are all used intact in a decoding process, with the columns erased, the puncturing is distinct from the shortening. Further, because position information for the punctured bits can be shared or estimated in common by the transmitter and the receiver when the system is set up, the 30 receiver may merely erase the corresponding punctured bits, before decoding. In the puncturing technique, because a length of a codeword that the transmitter actually transmits is N 1 -Np and a length of an information word is constantly K 1 , the code rate becomes K 1

I(N

1 -Np), which is always greater than the first given code rate K 1

/N

1 . A description will now be made of a shortening technique and a puncturing technique 35 suitable for the DVB-S2 LDPC code. The DVB-S2 LDPC code, as described above, is an 4246168_1 (GHMatters) P84756.AU.1 11/04/2013 13 LDPC code having a particular structure. Therefore, compared with the normal LDPC code, the DVB-S2 LDPC code is able to undergo more efficient shortening and puncturing. With reference to FIG. 4, a detailed description will be made of characteristics of a DVB-S2 LDPC code to which parity puncturing is applied. It is to be noted that for the DVB 5 S2 LDPC code of FIG. 4, N 1 =30, K1=15, M 1 =5, and q=3, and weight-1 position sequences for 0 th columns in three column groups are as follows: 0 1 2 0 11 13 0 10 14 10 An ith weight-1 position sequence in an /h column sequentially represents information on the positions of rows with 1 in an /h column group. FIG. 7 illustrates an example where random puncturing is applied to the LDPC code of FIG. 4. Because the parity bits punctured in FIG. 7 are subjected to erasure processing at a decoder, the punctured parity bits, compared with the other non-erased bits, are not greater 15 in a performance improvement effect in an LDPC decoding process, so their reliabilities decrease. Consequently, the other bits directly connected to the punctured parity bits, which are low in reliability, also suffer from a decrease in the performance improvement effect in the decoding process. The decrease in the performance improvement effect increases as the number of edges directly connected to the bits punctured on the Tanner graph increases. 20 In FIG. 7, for example, a 0 th information bit corresponding to a 0 th column is directly connected to the punctured parity bit twice, a 3 rd information bit corresponding to a 3 rd column is directly connected to the punctured parity bit once, and an 8 th information bit corresponding to an 8 th column is directly connected to the punctured parity bit three times. In this case, the 3 rd 0 th, and 8 th information bits are superior in the performance improvement 25 effect in order in the decoding process. In other words, when the degrees of the variable nodes are equal to each other, the performance improvement effect decreases as the number of the connected punctured bits increases. It can be appreciated from FIG. 7 that the numbers of punctured parity bits, which are directly connected to respective information bits, are random due to the random puncturing 30 pattern. Therefore, there is a high probability that reliabilities of respective information bits will also be random. In other words, while some information bits may gain higher-than needed decoding performance, other information bits may suffer significant performance degradation. This random puncturing pattern may lead to considerable irregularity of the reliabilities of the information bits in the decoding process. 35 FIG. 8 illustrates a second example where non-random puncturing is applied to the LDPC code of FIG. 4. More specifically, a relatively non-random puncturing pattern in a particular form is applied in the example illustrated in FIG. 8. 4246168 1 (GHMatters) P84756.AU.1 11/04/2013 14 Referring to FIG. 8, even though the relatively non-random puncturing pattern is applied, the connections to the information bits may be significantly irregular according to the corresponding puncturing pattern. The non-random puncturing pattern of FIG. 8 may be more irregular compared with the random puncturing pattern of FIG. 7. 5 In case of the LDPC code with a parity-check matrix having a particular structure like the DVB-S2 LDPC code, the connections between the information bits and the parity bits punctured according to the puncturing pattern can be significantly changed. The embodiments of the present invention suggest a puncturing pattern that provides stable decoding performance by maximally suppressing the irregularity of the reliabilities of 10 the information bits in the decoding process using the structural characteristics of the DVB S2 LDPC code. FIG. 9 illustrates a third example where non-random puncturing is applied to the LDPC code of FIG. 4. In the example of FIG. 9, a puncturing pattern that maintains a constant interval of 3 between punctured parity bits from q=3, which is one of the constituent variables, 15 is applied to the parity-check matrix illustrated in FIG. 4. As can be seen in FIG. 9, each of the information bits is equally and twice connected to the punctured bit. The irregularity between the punctured bits and information bits is remarkably reduced when the interval between the punctured parity bits is set according to the value of q due to the structure of the DVB-S2 LDPC code. This is better described with reference to FIG. 3. 20 Referring to Rules 1 and 2 and FIG. 3, regarding each column group, positions of '1's in the first column in the corresponding column group determine positions of '1's in the remaining columns. Indexes of rows where 1 is located in the remaining columns differ exactly by a multiple of q from an index of a row where 1 is located in the first column, in regard to modulo (N 1

-K

1 ), where N, denotes a length of an LDPC codeword and K, indicates 25 a length of an information word. More specifically, indexes of rows where 1 is located in two consecutive columns in a particular column group differ from each other exactly by q, for modulo (N 1

-K

1 ). Another characteristic of the DVB-S2 LDPC code lies in a submatrix corresponding to the parity part in the parity-check matrix. Referring to FIG. 3, the parity part has a structure of 30 a lower triangular matrix in which 1 exists in all of diagonal parts, and in this structure, an /h parity bit corresponds to '1' located in an /h row. In light of the structural characteristic of the DVB-S2 LDPC code, assuming that particular parity bits are punctured, if parity puncturing is repeated exactly at intervals of q, the number of edges of information bits connected to the parity bits punctured in a particular 35 column group is most regular. For example, assuming that an ith parity bit is punctured for Osi<q and an (i+kq)th parity bit is repeatedly punctured for Osk<M, if a certain information bit is connected to the lh parity bit, this indicates that '1' exists in an h row for a column corresponding to the information bit. Therefore, it can be understood that '1' exists in the 4246168 1 (GHMatters) P84756.AU.1 11/04/2013 15 (i+kq)th row in a column corresponding to an information bit which is separated by k from the above information bit among the columns in a column group according to Rule 1 and Rule 2. As a result, the information bit is connected to the punctured (i+kq)h bit. For the DVB-S2 LDPC code, because degrees of variable nodes corresponding to the 5 entire information word are equal to each other in one column group, and one or less of '1' is distributed in one row, when q-periodic puncturing is applied, the information bits corresponding to one column group are connected to a same number of punctured bits. Therefore, the connections between the punctured bits and the information bits become regular, so that stabilized decoding can be expected in the decoding process. 10 As described so far, it can be appreciated that for the DVB-S2 LDPC code, the q periodic puncturing technique can contribute to performance improvement by making the best use of the structural characteristics of the DVB-S2 LDPC code. However, to date, with regard to the optimized puncturing pattern obtained by considering the q-periodic puncturing, only the results obtained by setting BPSK or QPSK are known. 15 In addition to the q-periodic puncturing method, there are known methods that optimize the performance according to the code rate or codeword length when applying shortening or puncturing to the general LDPC code. However, because the known existing methods of finding a shortening/puncturing pattern perform the optimization process considering only BPSK or QPSK, only one optimized shortening/puncturing pattern could exist for a given 20 LDPC code. However, the optimized puncturing/shortening pattern obtained when high-order modulation is used and a signal constellation/bit mapping (bit mapping on the signal constellation) scheme has been determined, may differ from that in BPSK or QPSK modulation. 25 In BPSK or QPSK modulation, because reliabilities of bits constituting a symbol are equal, reliabilities of codeword bits are also equal in the LDPC codeword after it underwent shortening or puncturing, so there is no need to consider a modulation scheme in the process of finding the shortening/puncturing pattern. However, as described above, in high order modulation such as 16-QAM, 64-QAM and 256-QAM, because reliabilities of bits 30 constituting a symbol are different, when the high-order modulation scheme and the signal constellation/bit mapping scheme are determined, reliabilities of codeword bits in the LDPC codeword after application of shortening or puncturing may be different from those before application of puncturing or shortening. FIG. 10 illustrates another example of a parity-check matrix of a DVB-S2 LDPC code. 35 In FIG. 10, N 1 =40, K1=10, M 1 =5, and q=6, and weight-1 position sequences for 0 th columns in two column groups of an information word are as follows: 05 102025 7 1526 4246168 1 (GHMatters) P84756.AU.1 11/04/2013 16 An ith weight-1 position sequence in an th column sequentially represents information on the positions of rows with 1 in an ith column group. Referring to FIG. 10, a degree of each column corresponding to the first column group is 5, and a degree of each column corresponding to the second column group is 3. Generally, 5 as to an LDPC code, its performance improvement effect is excellent in a decoding process as the degrees increase. Thus, it is commonly expected that performance given after decoding bits corresponding to the first column group would be superior compared to that of the second column group. With reference to FIG. 11, a brief description will now be made of a puncturing pattern 10 which is suitable when BPSK or QPSK modulation is applied to the LDPC code with a parity check matrix of FIG. 10. In FIG. 11, yo and y indicate each BPSK symbol, or indicate two bits constituting one QPSK symbol. Hence, yo and yj are equal in reliability on the signal constellation. Referring to FIG. 11, after a parity bit corresponding to a 5 th column in a submatrix 15 corresponding to an arbitrary parity bit is punctured, one parity bit is punctured four times at a period of q. Here, information bits corresponding to degree-5 columns are connected to punctured parity bits through 2 edges, and information bits corresponding to degree-3 columns are not connected to punctured parity bits on the Tanner graph. Generally, the bits connected to many punctured bits show a poor performance 20 improvement effect in the decoding process. However, in FIG. 11, the degree-5 columns still have 3 edges which are not connected to the punctured bits, so the performance may not be significantly decreased in the decoding process. In addition, since information bits in the degree-3 columns are not directly connected to the punctured parity bits, significant performance degradation may not occur similarly in the decoding process. 25 It is assumed herein that the degree-5 columns are greater than the degree-3 columns in terms of the performance improvement effect in the decoding process. However, this assumption is valid only for BPSK or QPSK, and cannot always be applied for general high order modulation. For example, as illustrated in FIG. 12, 16-QAM modulation can be applied to the LDPC 30 code with the parity-check matrix of FIG. 10. In FIG. 12, yo and yj indicate high-reliability bits that determine signs of a real part and an imaginary part in a 16-QAM symbol, respectively. That is, the reliability relationship between the bits is defined as yo = y 1 > y 2 = y 3 . Referring to FIG. 12, degree-5 columns are mapped to y 3 and degree-3 columns are mapped to yi. In other words, the degree-5 columns are mapped to a lower-reliability bit and 35 the degree-3 are mapped to a higher-reliability bit on the signal constellation. In this case, it cannot be concluded that the degree-5 columns would have a higher performance improvement effect in the decoding process. The reasons are as follows. In light of the characteristics of the 16-QAM modulation, since the degree-5 columns are 4246168 1 (GHMatters) P84756.AU.1 11/04/2013 17 mapped to lower-reliability information in a signal received from a channel, their reliability improvement is made very slowly in the decoding process. On the other hand, the degree-3 columns, though they have a low degree, are mapped to lower-reliability information, so the reliability improvement effect occurs very fast. 5 As described above, it cannot be guaranteed that the bits corresponding to the higher degree columns always have superior performance in the LDPC code, to which high-order modulation is applied. Referring back to FIG. 12, in an implementation, after a parity bit corresponding to a 4 th column in a submatrix corresponding to an arbitrary parity bit is punctured, one parity bit is 10 punctured four times at a period of q. Here, information bits in both of degree-5 columns and degree-3 columns are connected to punctured parity bits through 1 edge on the Tanner graph. In the puncturing pattern applied in FIG. 11, the punctured bits are connected only to the degree-5 columns because the performance improvement effect of the degree-5 15 columns is high in BPSK or QPSK modulation. However, in the puncturing pattern illustrated in FIG. 12, the punctured bits are uniformly distributed considering the difference between reliabilities corresponding to each column group based on the modulation scheme. In an analysis of this case, since the degree-5 columns are connected to only one punctured bit, there is a high probability that no more significant performance degradation will happen. Also, 20 though the degree-3 columns are connected to one punctured bit, they correspond to high reliability information from a received signal, so the probability is high that the significant performance degradation will not occur. Similarly, even when 64-QAM modulation is applied to the LDPC code with the parity check matrix of FIG. 10 as illustrated in fig. 14, the characteristics may be different from 25 those of BPSK, QPSK and 16-QAM. In FIG. 13, yo and yj indicate high-reliability bits that determine signs of a real part and an imaginary part in a 64-QAM symbol, respectively. That is, reliability relationship between the bits is defined as yo y 1 > y 2 = y 3 > y 4 = y 5 . FIG. 13 illustrates an example of a puncturing pattern obtained by considering parity bits corresponding to degree-2 columns. In FIG. 12A for 16-QAM, though 3 bits among the 30 punctured parity bits are connected to the lowest-reliability bits y 2 and y 3 , since the information bits corresponding to the degree-5 or degree-3 columns have good performance, no performance degradation may occur. However, in FIG. 13 where 64-QAM is applied, when too many parity bits are connected to y 4 and y 5 having very low reliabilities, performance degradation may occur. Hence, the reliabilities corresponding to the parity bits 35 should also be carefully considered. It should be noted in FIG. 12 and FIG. 13 that when the codeword length of the LDPC code decreases due to shortening or puncturing, orders of bits corresponding to the signal constellation are equal, but the bits decrease in a predetermined ratio. For example, in FIG. 4246168 1 (GHMatters) P84756.AU.1 11/04/2013 18 12 and FIG. 13, for an LDPC code, orders (y, Y1, Yo, Y2, Y1, Y3, y2, yo) and (y5, y1, Y3, Y4, Yo, Y2, Y3, , yi, y2, y4, yo) of bits corresponding to the signal constellations are maintained, but a ratio of bits corresponding to each signal constellation decreases in proportion to a length of the LDPC codeword. 5 As described in FIGs. 10, 11, 12 and 13, it can be easily expected that the same puncturing pattern could be changed according to the modulation scheme. That is, when a high-order modulation scheme and a signal constellation/bit mapping scheme are determined for a given LDPC code, the optimal puncturing pattern is subject to change according to connections between punctured bits and non-punctured other bits. Thus, 10 different puncturing patterns should be applied according to transmission modulation schemes, in order to minimize the performance degradation caused by puncturing. The general process for applying the above-described puncturing scheme can be summarized in 5 steps as follows. For convenience, it is assumed herein that N 1 indicates a length of an LDPC codeword, each column group includes M 1 columns, and Np parity bits are 15 subject to puncturing. The following puncturing process is illustrated in fig. 14 in brief. Puncturing Step 1 A transmission apparatus generates an existing shortened/non-shortened DVB-S2 LDPC codeword in step 1301. Puncturing Step 2 20 The transmission apparatus determines the number Np of parity bits for puncturing in IN A =P step 1303, and finds in step 1305, where Lv] is the maximum integer which is less than or equal to x. Puncturing Step 3 The transmission apparatus determines puncturing parity bits for 25 0sx<A and 0si,<q according to a transmission modulation scheme in step 1307. It is assumed that for 05x<q, values of ixwere previously determined using a density evolution method considering the transmission modulation scheme(herein, a relationship of Asq is given). Puncturing Step 4 30 The transmission apparatus applies puncturing to all parity bits for 0 x<A and 0sk<M 1 in step 1307. Here, the constant B is a preset non-zero integer. Puncturing Step 5 4246168_1 (GHMatters) P84756.AU.1 11/04/2013 19 The transmission apparatus additionally punctures the parity bits for Osk<Np-AM 1 in step 1307. Thereafter, the transmission apparatus transmits bits except for the punctured bits in step 1309. It can be understood from Puncturing Step 1 through Puncturing Step 5 that the 5 puncturing pattern can be defined accurately when the number Np of bits to be punctured, sequence information defining values of i, and a value of q are known. To describe detailed examples in which the above puncturing steps are performed according to modulation schemes, less than optimal (i.e. suboptimized) puncturing patterns for an DVB-S2 LDPC code with N 1 =16200, K1=7200, M 1 =360, and q=25 are shown in Table 10 1A and Table 1B. A process of selecting the suboptimal puncturing patterns will be described in detail below. In Table 1A and Table 1B, (Po, P1, P2,..., p8999) indicates all parity bits of the DVB-S2 LDPC code, which are enumerated in a row. 4246168_1 (GHMatters) P84756.AU.1 11/04/2013 20 Table 1A Major vari'ables of SvN=16200, Ki=7200, M1360, q=25 DVB-S2 LDPC code Grouping of arity -Dfine. a set s atisfying the following characteristics for bits 0 jkg=25: -=pk B mo d 25, 0 k<9000. The setF Pyis a set having 360 p arity bits as its entries. Range.f oN Puncturing Method O!g<9000 I N For an integer m = , punctureall of parity bits corresponding torn parity bit sets P*O), P 4 i), ... , and additionally puncture (N,-360m) parity hits from arnong the parity bits corresponding Pto Here, -rindic ates a permutation function that is a puncturing p attern, and the relationships are shown at the bottom of the table, 4240168 1|Gcausi.1P4755AU.1 1t[X14.2013 21 Table 1B T(E (1 (2) TE(3) -g(4) -g(5). 71) (7) -7(8) ir(9) -; -(10) (1i1) Ti(.2) x(13) Ti(14) ;r(15) 7(16) i-(17) ir(18) -x(19) 2(20) (21) (22) (23) .(24) - Relationship between permutation functions suboptimize don BPSKIQP SK 6 4 18 9 13 8 15 20 5 17 2 2 24 10 1 2 121 3 16 23 1 14 0 21 19 7 11 - Relationship between permutationfunctions suboptimized on 16QAM 6 4 13 9 18 8 15 20 5 1 - 2 24 22 14 23 3 19 10 1 12 0 16 21 7 11 - Re lationshi p between permutation functions subo ptimize d on 64Q AM 15 3 23 6 13 18 .1 9 17 8 5 20 16 22 4 24 14 2 21 12 7 10 19 11 0 - Referring to Table 1A and Table 12, it can be appreciated that when a length of parity bits to be punctured is determined, the puncturing process based on suboptimal puncturing S patterns is performed through a predetermined process regardless of the modulation scheme, but relationships between permutation functions indicating optimized puncturing patterns are all different according to modulation schemes. That is, when the puncturing method is applied without considering the modulation scheme, significant performance degradation may occur according to modulation schemes. 10 From the puncturing process, it can also be appreciated that AM, parity bits are punctured in Puncturing Step 3 and Puncturing Step 4 and N-AM 1 parity bits are punctured in Puncturing Step 5, so that a total of N. parity bits are punctured. The suboptimal puncturing patterns shown in Table 1A and Table 11 may not be unique according to conditions for finding the puncturing patterns. Because several 15 selections are possible in the process of selecting the puncturing patterns, which will be described below in detail, many puncturing patterns showing good performance may be available. Actually, puncturing patterns shown in Table 2A and Table 22 can also provide very excellent performance, like the puncturing patterns defined in Table 1A and Table 1B. 424Gii 1 iGuwsaimsPB47S&A.1 tt'14.2013 22 Table 2A Major variables of MVB-r vLDPC cod I,= 1 =6200, Ki=7200, M 1 =360, q=25 DVB-S2 ISDPC code, Grouping of p arity Define a setPg satisfying the following characteristics for bits Os <q=25: PF={pyI k-j mod 25, 0k<9000} The set' P is a setihaving 360 p arity bits as its entries Range.f , Puncturing Method C0 ,<9000 I N For an integer m= , puncture.all of parity bits corresponding torn parity bit sets P*O), P 4 i), ... , and additionally puncture (N,-360m) parity hits from among the parity bits corresponding to PX Here, -rindic ates a permutation function that is a puncturing p attern, and the relationships are shown at the bottom of the table, 424GI58 1 uw.siamlPB4755AU.1 ttf1i4.T13 23 Table 2B ( (2) TE(3) g(4) -g(5) 16 (7) 71(8) (9) -(10) g(11) i (2) x(13) Ti(14) ;r(15) 7(16) -g(17) ir(18) -x(19) ir(20) )(21) (22) (23) 2(24) - Relationship between permutation functions suboptimize don BP SKIQP SK 6 4 18 9 13 8 15 20 5 17 2 24 10 1 2 121 3 16 23 1 14 0 21 19 7 11 - Relationship between permutationfunctions suboptimized on 16QAM 6 4 13 9 18 8 15 20 5 17 - 2 22 24 7 12 1 16 23 14 0 21 10 19 11 3 - Re lationshi p between permutation functions subo ptimize d on 64Q AM 6 15. 13 10 3 17 21 8 5 19 2 23 16 24 7 18 1 12 20 0 4 14 9 11 22 - The method of mapping bits corresponding to the signal constellations used in 16 QAM and 64-QAM modulations of Table 2A and Table 2B is the results obtained by applying S the same bit mapping schemes as those illustrated in FIG. 12 and FIG. 13. The DVB-S2 LDPC codeword transmitted after undergoing the puncturing is restored to its original signal from a received signal at a reception apparatus through a decoding process in fig. 15. Fig. 15 is a flowchart illustrating a reception method in a reception apparatus according 10 to an embodiment of the present invention. Referring to fig. 15, a reception apparatus determines or estimates a puncturing/shortening pattern from a received signal in step 1401. Thereafter, the reception apparatus determines in step 1403 whether there are any punctured or shortened bits. In the absence of punctured or shortened bits, the reception apparatus performs decoding in step 15 1409. However, when there are punctured or shortened bits, the reception apparatus provides the puncturinglshortening pattern to an LDPC encoder 1560, which will be described in connection with fig. 16, in step 1405. In step 1407, the LDPC encoder 1560 determines that punctured bits are erased bits and determines that the probability that values of shortened bits will be zero (0) is 1. 20 Thereafter, the LDPC encoder 1560 performs decoding in step 1409. 42lii6B 1 iGauwsisP475&Au.1 tt'1"2013 24 In the puncturing process, q-periodic puncturing is applied to stabilize performance of the DVB-S2 LDPC code using structural characteristics of the DVB-S2 LDPC code. A most significant difference between the present invention and the prior art lies in considering reliability of the transmission modulation scheme when determining parity bits to 5 be punctured in Puncturing Step 3. An example of a sequence selection procedure is provided here below to determine the bits to be punctured in the DVB-S2 LDPC code in Puncturing Step 3. The following selection procedure is subject to change, when applied to other LDPC codes. Selection Process 1 10 Bits, which are connected, if possible, to a less number of information bits, are first determined. Selection Process 2 Parity bits to be punctured, which exhibit the most asymptotic performance, are determined from the parity bits determined in Selection Process 1, using a density evolution 15 analysis method that considers both a modulation scheme and degree distribution. Selection Process 3 Based on the parity bits to be punctured, determined in Selection Process 2, Selection Process 1 and Selection Process 2 are repeated on the bits excluding the already selected bits to be punctured. 20 Commonly, when the number Np of bits to be punctured varies greatly, the puncturing patterns optimized based on the value of Np may have no correlation with each other. In other words, in a system where the value of Npvaries greatly, all of the puncturing patterns optimized based on the value of Np should be stored separately for optimized performance. However, although the puncturing patterns obtained by applying the above selection 25 procedure are not guaranteed to be optimal for all cases, they would have relatively stable performance from one puncturing pattern having a regular rule, regardless of a change in the value of Np, thereby securing the relatively stable performance and easy storage of the puncturing patterns. For example, supposing that orders of the parity bits to be punctured are set as P 1 , 30 P 2 ,..., Pq, only a sequence indicating the orders of the parity bits may be stored to enable efficient puncturing for an arbitrary value of Np through Puncturing Step 1 to Puncturing Step 5. The puncturing technique may contribute to an increase in a code rate because it can change a length of an LDPC codeword, and reduce the codeword length without changing 35 an information length. Preferably, the puncturing and the shortening can be applied together to obtain the code rate and codeword length needed in the system. Assuming, as described above, that a codeword length and an information length of an LDPC code, which is intended to be finally obtained from a given LDPC code with a 4246168 1 (GHMatters) P84756.AU.1 11/04/2013 25 codeword length N 1 and an information length K1 through shortening and puncturing, are N 2 and K2, respectively, if a definition of N 1

-N

2 =N and KI-K 2 =K is given, the LDPC code with a codeword length N 2 and an information length K2 can be generated by shortening K bits and puncturing Np (=N-K) bits from a parity-check matrix of the given LDPC code. In case of the Ni - N, 5 generated LDPC code, for N>0 or K>0, since the code rate is , the puncturing and KT, - KA shortening lengths can be set considering N 2 and fig. 16 is a block diagram of a transmission apparatus using a punctured/shortened LDPC code according to an embodiment of the present invention. Referring to fig. 16, a transmission apparatus includes a controller 1510, a shortening 10 pattern applier 1520, an LDPC code parity-check matrix extractor 1540, an LDPC encoder 1560, and a puncturing pattern applier 1580. The LDPC code parity-check matrix extractor 1540 extracts an LDPC code parity check matrix that underwent shortening. The LDPC code parity-check matrix can be extracted using a memory, can be given in the transmission apparatus, or can be generated 15 by the transmission apparatus. The controller 1510 controls the shortening pattern applier 1520 to determine a shortening pattern according to an information length. The shortening pattern applier 1520 inserts bits with a value of 0 in the positions corresponding to shortened bits, or removes columns corresponding to the shortened bits from a parity-check matrix of a given LDPC 20 code. The shortening pattern can determined by extracting a shortening pattern stored in a memory, generating a shortening pattern using a sequence generator (not shown), or obtaining a shortening pattern using a density evolution analysis algorithm for a parity-check matrix and a given information length. The shortening pattern applier 1520 is optional when shortening is not needed for the 25 code. Further, the controller 1510 controls the puncturing pattern applier 1580 to determine and apply a puncturing pattern according to a modulation scheme and a length of puncturing bits. The puncturing pattern applier 1580 determines the number of parity bits for puncturing, divides the parity bits at predetermined intervals, determines the number of 30 puncturing bits, which are subjected to puncturing within the predetermined intervals, determines a modulation scheme, determines positions of puncturing parity bits corresponding to the determined number of the puncturing bits within the predetermined intervals according to the determined modulation scheme, and repeatedly performs puncturing on the puncturing parity bits corresponding to the determined positions at the 35 predetermined intervals. The predetermined intervals are determined by dividing a length of the parity bits by a length of one column group in a parity-check matrix. 4246168_1 (GHMatters) P84756.AU.1 11/04/2013 26 The remaining bits except for the punctured bits are transmitted to a receiver according to the modulation scheme through a transmission unit (not shown). The LDPC encoder 1560 performs encoding based on the LDPC code shortened by the controller 1510 and the shortening pattern applier 1520. 5 fig. 17 is a block diagram of a reception apparatus according to an embodiment of the present invention, in which a signal transmitted from a communication system using a punctured/shortened DVB-S2 LDPC code is received, and restored to user-desired data. Referring to fig. 17, a reception apparatus includes a controller 1610, a shortening/puncturing pattern determination/estimation unit 1620, a demodulator 1630, and 10 an LDPC decoder 1640. The demodulator 1630 receives and demodulates a shortened/punctured LDPC code, and provides the demodulated signal to the shortening/puncturing pattern determination/estimation unit 1620 and the LDPC decoder 1640. The shortening/puncturing pattern determination/estimation unit 1620, under the 15 control of the controller 1610, estimates or determines information about a puncturing/shortening pattern of the LDPC code from demodulated signal, and provides position information of the punctured/shortened bits to the LDPC decoder 1640. The shortening/puncturing pattern determination/estimation unit 1620 can determine or estimate the puncturing/shortening pattern by extracting a puncturing/shortening pattern stored in a 20 memory, generating a puncturing/shortening pattern using a previously implemented generation method, or obtaining a puncturing/shortening pattern using a density evolution analysis algorithm for a parity-check matrix and a given information length. The LDPC decoder 1640 performs erasure processing on the punctured bits and performs decoding thereon. 25 When the transmission apparatus applies both shortening and puncturing, the shortening/puncturing pattern determination/estimation unit 1620 in the reception apparatus may carry out pattern determination/estimation on the shortening first, perform pattern determination/estimation on the puncturing first, or make pattern determination/estimation on both the shortening and puncturing. The shortening/puncturing pattern 30 determination/estimation unit 1620 determines presence/absence of puncturing bits in the demodulated signal. When there are puncturing bits, the shortening/puncturing pattern determination/estimation unit 1620 determines positions of the punctured parity bits by estimating information about the puncturing pattern. The LDPC decoder 1640 decodes data using the determined positions of the 35 punctured parity bits, assuming that the probability that the punctured bits would be zero (0) and the probability that the punctured bits would be 1 are equal to 1/2. Because the probability that values of the shortened bits will be zero is 1 (i.e. 100%), the LDPC decoder 1640 determines whether or not it will allow the shortened bits to take part in its decoding 4246168 1 (GHMatters) P84756.AU.1 11/04/2013 27 operation depending on the value 1 of the probability that the shortened bits would be zero. When the LDPC decoder 1640 receives information on a length of the shortened DVB-S2 LDPC code by means of the shortening/puncturing pattern determination/estimation unit 1620, the LDPC decoder 1640 restores the user-desired data from the received signals. 5 As described in connection with fig. 16, the shortening is performed in the input stage of the LDPC encoder 1560 and the puncturing is performed at the output stage of the LDPC encoder 1560. However, in the reception apparatus illustrated in fig. 17, the LDPC decoder 1640 should have information on the puncturing and the shortening simultaneously, to make the decoding possible. 10 While the invention has been shown and described with reference to a certain exemplary embodiments thereof, it will be understood by those skilled in the art that various changes in form and details may be made therein without departing from the spirit and scope of the invention as defined by the appended claims and their equivalents. In the claims which follow and in the preceding description of the invention, except 15 where the context requires otherwise due to express language or necessary implication, the word "comprise" or variations such as "comprises" or "comprising" is used in an inclusive sense, i.e. to specify the presence of the stated features but not to preclude the presence or addition of further features in various embodiments of the invention. 4246168 1 (GHMatters) P84756.AU.1 11/04/2013

Claims (3)

1. A method for channel decoding using a low-density parity-check (LDPC) code, the method comprising: 5 demodulating a signal transmitted from a transmitter; determining a number of punctured parity bits; determining a number of punctured parity bit sets based on the determined number of the punctured parity bits; acquiring a predetermined order of punctured parity bit sets; .0 determining position information of the punctured parity bits based on the predetermined order; and decoding data using the position information of the punctured parity bits.

2. The method of claim 1, wherein the predetermined order of the punctured parity bit .5 sets is 6, 4, 13, 9, 18, 8, 15, 20, 5, 17, 2, 22, 24, 7, 12, 1, 16, 23, 14, 0, 21, 10, 19, 11, 3, when a codeword length is 16200 and an information length is 7200, a modulation scheme is

16-quadrature amplitude modulation (16-QAM). 3. The method of claim 1, wherein the predetermined order of the punctured parity bit !0 sets is 6, 15, 13, 10, 3, 17, 21, 8, 5, 19, 2, 23, 16, 24, 7, 18, 1, 12, 20, 0, 4, 14, 9, 11, 22, when a codeword length is 16200 and an information length is 7200, a modulation scheme is 64 quadrature amplitude modulation (64-QAM). 4. The method of claim 1, wherein determining the position information of the punctured 25 parity bits, comprises: additionally determining positions of remaining punctured parity bits when the number of the punctured bits is not a multiple of the number of parity bit sets. 5. The method of claims 1, wherein the number of the punctured parity bit sets is 30 determined by below equation: A=L'M where A denotes the number of punctured parity bit sets, NP denotes a number of punctured parity bits, M, denotes the number of bits in one parity bit set. 35 6. The method of claims 5, wherein the parity bit sets are formed by below equation; P={Pk j=k mod q, Ok<N-K 1 }, 6138860_1 (GHMatters) P84756.AU.1 SARAHVV 22/01/2015 29 where P denotes the jth parity bit set, N, denotes a length of an LDPC codeword, K, denotes a length of an information word, M, denotes the number of bits in one parity bit set, q is a value satisfying q = (N, - K,) / MI, where K,/M, is an integer, and 0 : j < q. 5 7. An apparatus for channel decoding using a low-density parity-check (LDPC) code, the apparatus comprising: a demodulator for demodulating a signal transmitted from a transmitter; a puncturing pattern estimator for determining a number of punctured parity bits, determining a number of punctured parity bit sets based on the determined number of the .0 punctured parity bits, acquiring a predetermined order of punctured parity bit sets, and determining position information of the punctured parity bits based on the predetermined order; and a decoder for decoding data using the position information of the punctured parity bits. .5 8. The apparatus of claim 7, wherein the predetermined order of the punctured parity bit sets is 6, 4, 13, 9, 18, 8, 15, 20, 5, 17, 2, 22, 24, 7, 12, 1, 16, 23, 14, 0, 21, 10, 19, 11, 3, when a codeword length is 16200 and an information length is 7200, a modulation scheme is 16-quadrature amplitude modulation (16-QAM). !0 9. The apparatus of claim 7, wherein the predetermined order of the punctured parity bit sets is 6, 15, 13, 10, 3, 17, 21, 8, 5, 19, 2, 23, 16, 24, 7, 18, 1, 12, 20, 0, 4, 14, 9, 11, 22, when a codeword length is 16200 and an information length is 7200, a modulation scheme is 64 quadrature amplitude modulation (64-QAM). 25 10. The apparatus of claim 7, wherein the puncturing pattern estimator is adapted to additionally determine positions of remaining punctured parity bits when the number of the punctured bits is not a multiple of the number of parity bit sets. 11. The apparatus of claims 7, wherein the number of the punctured parity bit sets is 30 determined by below equation: A=L'] MI where A denotes the number of the punctured parity bit sets, NP denotes a number of the punctured parity bits, M, denotes the number of bits in one parity bit set. 35 12. The apparatus of claims 11, wherein the parity bit sets are formed by below equation; P={Pk j=k mod q, Ok<N,-K,}, 6138860_1 (GHMatters) P84756.AU.1 SARAHVV 22/01/2015 30 where P denotes the jth parity bit set, N, denotes a length of an LDPC codeword, K, denotes a length of an information word, M, denotes the number of bits in one parity bit set, q is a value satisfying q = (N, - K,) / M, where K,/M, is an integer, and 0 : j < q. 5 13. A method for channel decoding, substantially as herein described with reference to the accompanying drawings. 14. An apparatus for channel decoding, substantially as herein described with reference to the accompanying drawings. -0 6138860_1 (GHMatters) P84756.AU.1 SARAHVV 22/01/2015