CN101305521B - Method of encoding and decoding using low density parity check code - Google Patents

Method of encoding and decoding using low density parity check code Download PDF

Info

Publication number
CN101305521B
CN101305521B CN2005800255866A CN200580025586A CN101305521B CN 101305521 B CN101305521 B CN 101305521B CN 2005800255866 A CN2005800255866 A CN 2005800255866A CN 200580025586 A CN200580025586 A CN 200580025586A CN 101305521 B CN101305521 B CN 101305521B
Authority
CN
China
Prior art keywords
row
submatrix
parity check
whole
weight
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Expired - Fee Related
Application number
CN2005800255866A
Other languages
Chinese (zh)
Other versions
CN101305521A (en
Inventor
吴旼锡
丁奎赫
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
LG Electronics Inc
Original Assignee
LG Electronics Inc
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Priority claimed from KR1020040060839A external-priority patent/KR101079087B1/en
Priority claimed from KR1020040062068A external-priority patent/KR20060013197A/en
Application filed by LG Electronics Inc filed Critical LG Electronics Inc
Priority claimed from PCT/KR2005/002421 external-priority patent/WO2006011744A2/en
Publication of CN101305521A publication Critical patent/CN101305521A/en
Application granted granted Critical
Publication of CN101305521B publication Critical patent/CN101305521B/en
Expired - Fee Related legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Images

Classifications

    • H—ELECTRICITY
    • H03—ELECTRONIC CIRCUITRY
    • H03M—CODING; DECODING; CODE CONVERSION IN GENERAL
    • H03M13/00—Coding, decoding or code conversion, for error detection or error correction; Coding theory basic assumptions; Coding bounds; Error probability evaluation methods; Channel models; Simulation or testing of codes
    • H03M13/03—Error detection or forward error correction by redundancy in data representation, i.e. code words containing more digits than the source words
    • H03M13/05—Error detection or forward error correction by redundancy in data representation, i.e. code words containing more digits than the source words using block codes, i.e. a predetermined number of check bits joined to a predetermined number of information bits
    • H03M13/11—Error detection or forward error correction by redundancy in data representation, i.e. code words containing more digits than the source words using block codes, i.e. a predetermined number of check bits joined to a predetermined number of information bits using multiple parity bits
    • H03M13/1102—Codes on graphs and decoding on graphs, e.g. low-density parity check [LDPC] codes
    • H03M13/1148—Structural properties of the code parity-check or generator matrix
    • H—ELECTRICITY
    • H03—ELECTRONIC CIRCUITRY
    • H03M—CODING; DECODING; CODE CONVERSION IN GENERAL
    • H03M13/00—Coding, decoding or code conversion, for error detection or error correction; Coding theory basic assumptions; Coding bounds; Error probability evaluation methods; Channel models; Simulation or testing of codes
    • H03M13/65—Purpose and implementation aspects
    • H03M13/6502—Reduction of hardware complexity or efficient processing
    • H03M13/6505—Memory efficient implementations

Landscapes

  • Physics & Mathematics (AREA)
  • Probability & Statistics with Applications (AREA)
  • Engineering & Computer Science (AREA)
  • Theoretical Computer Science (AREA)
  • Mathematical Physics (AREA)
  • Error Detection And Correction (AREA)

Abstract

A method of encoding and decoding using an LDPC code is disclosed, by which a memory for storing a parity check matrix necessary for the encoding or decoding using the LDPC code and calculation amount and complexity necessary for the encoding or decoding can be reduced. The present invention includes a step of encoding an input data using a parity check matrix H having a configuration of H=[Hd | Hp] (Hd is (n-k)xk dimensional, Hp is (n-k)x(n-k) dimensional, k is a bit number of the input data, and n is a bit number of a codeword), wherein if the Hd comprises a plurality of sub-matrices, each of the sub-matrices has predetermined regularity in a row or column weight.

Description

Use the method for low density parity check code Code And Decode
Technical field
The present invention relates to the method for Code And Decode, especially, relate to the method for using low-density checksum (being abbreviated as hereinafter LDPC) code to come Code And Decode.Although the present invention is suitable for wide range of application, it is suitable for having the performance of enhancing in particular.
Background technology
Usually, although coding is distorted signals, the loss of signal when transmitter side sends data via communication channel etc. caused error, transmitter side is the processing procedure that the receiver side executing data processes to recover initial data.And the processing procedure of decoding to be transmission data that receiver side will be encoded revert to initial data.
Recently, a lot of attentivenesss is placed in the coding method of using the LPDC code.This LDPC code is to have low-density linear block code, because the most elements of parity check matrix H is zero, it proposes in 1962 by adding glug (Gallager).Because technical difficulty was difficult to realize the LDPC code that it was very complicated at that time.But this LDPC code was reconsidered in nineteen ninety-five, so that its superior function is verified.Therefore, carry out many effort and removed to research and develop the LPDC code.(reference: [1] Robert G.Gallager, " Low-DensityParity-Check Codes " (low density parity check code), The MIT Press publishes, on September 15th, 1963.[2] D.J.C.Mackay, " Good error-correcting codes based onvery sparse matrics " (based on the unusual good error correcting code of sparse matrix), IEEE Trans.Inform.Theory (IEEE meeting, information, principle), IT-45, pp.399-431 (1999)).
The parity matrix of LDPC code is the binary matrix that comprises " 0 " and " 1 ".Because the number of " 1 " of the parity matrix of LDPC code is considerably less, in the situation that the large scale matrix, permission comes the parity matrix of decoding LDPC via repeat decoding.If matrix size is very large, similar turbo code, the parity matrix of LDPC code illustrate the performance close to the channel capacity limit of Shannon (Shannon).
The LDPC code can be explained by the parity check matrix H of (n-k) * n dimension.And, can obtain by equation 1 corresponding to the generator matrix G of parity check matrix H.
[equation 1]
H.G＝0
In the coding/decoding method that uses the LDPC code, transmitter side comes coded input data by the generator matrix G that equation 2 usefulness and parity check matrix H have the relation of equation 1.
[equation 2]
C=G.u, " c " is code word here, and " u " is Frame.
But, usually use now and utilize parity check matrix H, rather than utilize the input data-encoding scheme of generator matrix G.Therefore, described as explained above, in the coding/decoding method that utilizes the LDPC code, this parity check matrix H is most important factor.Because parity check matrix H has great yardstick, for the many computings of corresponding Code And Decode process need.And the realization of parity check matrix H is very complicated.In addition, this parity check matrix H needs very large memory space.
In the process (that is, increasing the number of " 1 ") that many weights is increased to the parity check matrix H of the binary matrix that comprises " 0 " and " 1 ", more variable is added to parity check equations.Therefore, in the Code And Decode method of using the LDPC code, better performance can be shown.
But, if more weight is added to parity check matrix H, can from whole parity matrix, produce 4-circle (cycle) or 6-circle.Therefore, use the performance of the Code And Decode method of LDPC code to be worsened.
Summary of the invention
Accordingly, the present invention proposes the method for a kind of use low-density checksum (LDPC) code Code And Decode, it has eliminated the problem that one or more restrictions owing to prior art and shortcoming cause basically.
An object of the present invention is to provide a kind of method of the LDPC of use code Code And Decode, can reduce be used to the memory that is stored as use LDPC code coding or the necessary parity matrix of decoding by it, and be coding or decode necessary amount of calculation and complexity.
Another object of the present invention provides a kind of method of the LDPC of use code Code And Decode, can be so that 4-circle or the 6-circle of parity check matrix H are reduced to the performance that minimum mode strengthens coding or decoding aspect by it.
Additional characteristics of the present invention and advantage will be set forth in description subsequently, and will be apparent to a certain extent from this description, perhaps can learn by putting into practice the present invention.Structure by specifically noting in the specification of book and claim thereof and appended accompanying drawing can realize and obtain purpose of the present invention and other advantage.
In order to realize the advantage of these and other, and according to purpose of the present invention, as implementing herein and describe widely, comprise step according to the method for the present invention a kind of use LDPC (low-density checksum) code coding: use to have H=[H d| H p] the parity check matrix H coded input data of structure, wherein H d(n-k) * k dimension, H pBe (n-k) * (n-k) dimension, k is the bit number of input data, and n is the bit number of code word, if H wherein dComprise a plurality of submatrixs, each of this submatrix is expert at or row weight aspect has predetermined regularity, and whole H wherein dAny random two row do not list at least two and have simultaneously 1.
Preferably, if H dComprising that m has the submatrix of (n-k)/m * k dimension, is under 1 the condition in the row weight of the whole row of specific submatrix, and the j of this specific submatrix of m submatrix is capable to have W jIndividual continuous 1.Further preferred, W jConsistent with the whole row of this specific submatrix.Further preferred, for the whole row W of this specific submatrix jIncrease brokenly or reduce.
Preferably, if H dComprise m the submatrix with (n-k)/m * k dimension, consist of this H dThe row or column weight of random submatrix be 1.
Further preferred, all make from whole H dRandom three row in two row that the choose situation that has 1 combination at identical point be equal to or less than the critical value (C that presets Max).
Preferably, random two row of whole parity check matrix H do not have 1 simultaneously at least two row.Further preferred, all situations that make two row that choose from random three row of parity check matrix H have 1 combination at identical point are equal to or less than the critical value (C that presets Max).
Further preferred, critical value (C Max) be the arbitrary value in the 10-100 scope.
Further preferred, H pIt is dual diagonal matrix.
For the further advantage that realizes these and other, and according to purpose of the present invention, aspect use parity check matrix H decoding input data, a kind of method of using the LDPC code to decode, the method comprising the steps of: use to have H=[H d| H p] the parity check matrix H decoding input data H wherein of structure d(n-k) * k dimension, H pBe (n-k) * (n-k) dimension, k is the bit number of input data, and n is the bit number of code word, if H wherein dComprise a plurality of submatrixs, each of this submatrix has predetermined regularity aspect the row or column weight, and whole H wherein dAny random two row do not list at least two and have simultaneously 1.
Should be understood that above general introduction and following detailed description are exemplary and illustrative, and intention provides to the further instruction of the present invention such as claim.
Description of drawings
Accompanying drawing is included to provide further to be understood the present invention, and is incorporated into and consists of the part of this specification, and it illustrates embodiments of the invention, and can work to explain the principle of the invention with this specification.
In the accompanying drawings:
Fig. 1 is the block diagram of explaining the communication system of a preferred embodiment of the invention;
Fig. 2 is for explaining H=[H d| H p] schematic diagram of relational expression;
Fig. 3 is the schematic diagram of dual diagonal matrix;
Fig. 4 is the H according to a preferred embodiment of the invention dThe schematic diagram of specific submatrix;
Fig. 5 is the H according to another preferred embodiment of the present invention dThe schematic diagram of specific submatrix;
Fig. 6 is in the situation that the H of m=4 (i) dSchematic diagram, to explain characteristics of the present invention;
Fig. 7 is that wherein two row of parity check matrix H side by side have " 1 " at two points for the schematic diagram of explaining another characteristics of the present invention;
Fig. 8 is that wherein all combinations make two row that choose from random three row of parity check matrix H have " 1 " at identical point respectively for the schematic diagram of explaining another characteristics of the present invention;
Fig. 9 is as the bipartite graph by the different expression of parity check matrix H;
Figure 10 is the flow chart that generates a process of parity check matrix H according to the present invention;
Figure 11 is the flow chart that generates another process of parity check matrix H according to the present invention; With
Figure 12 is the flow chart that generates another process of parity check matrix H according to the present invention.
Embodiment
To at length be introduced now the method for coming Code And Decode according to use LDPC (low-density checksum) code of the preferred embodiments of the present invention, example wherein is to illustrate in the accompanying drawings.
Fig. 1 is the block diagram of explaining the communication system of a preferred embodiment of the invention, and wherein technical characterstic of the present invention is applicable to for example wireless communication system.
With reference to figure 1, transmitter 10 and receiver 30 uses wireless channel 20 to communicate by letter mutually as medium.In transmitter 10, the k bit source data u that exports from data source 11 is converted into n bit codewords c by the LDPC coding LDPC coding module 13.Transmitted via antenna 17 by this code word c of modulation module 15 wireless-modulated, and then received by another antenna 31 of receiver 30.In receiver 30, initial data is recovered via the processing procedure opposite with transmitter 10.That is, source data u can be at last solution by demodulation module 33 be in harmonious proportion the decoding of LDPC decoder module 35 and obtain.
The data transmission/reception process of above explanation is to describe for explaining in the minimum zone that characteristics of the present invention need.Therefore, corresponding process need is apparent for the step of transfer of data/reception more for those skilled in the art.
In equation 1, the first parity check matrix H can be by H=[H d| H p] (H d(n-k) * k dimension, H p(n-k) * (n-k) dimension) expression.Fig. 2 is for explaining H=[H d| H p] schematic diagram of relational expression." k " is the length (bit base) that inputs to the source data of LDPC coding module 13, and " n " refers to the length (bit base) of the code word c of coding.
By equation 1 and H=[H d| H p] relational expression, can know, Therefore, this LDPC coding module 13 multiply by will input data u by equation 2
Figure GSB00000144245000062
Mode carry out coding.Therefore, equation 2 can be replaced by equation 4.Especially, k bit input source data S 1 * kBe encoded to n bit codewords x by equation 2 1 * kCode word x has x=[s p]=[s 0, s 1..., s K-1, p 0, p 1..., P M-1], (p here 0, p 1..., P M-1) be parity check bit, and (s 0, s 1..., s K-1) be system bits.
[equation 4]
c = [ I | ( H p - 1 H d ) t ] t · u
But it is very complicated using this encoding scheme of generator matrix G.In order to reduce above-mentioned complexity, preferred, the source data of input is used the parity check matrix H direct coding.That is, because x=[s p], if use the characteristic of H.x=0, then Hx=H[s p]=0.From this equation, can obtain parity check bit p finally to obtain code word x=[s p].
Preferably, (n-k) * (n-k) dual diagonal matrix of dimension is used as H pThis dual diagonal matrix refer to leading diagonal wherein and directly the diagonal below leading diagonal be " 1 ", and wherein the rest is zero matrix.And Fig. 3 illustrates the schematic diagram that helps to understand dual diagonal matrix.
With H dBe decomposed in the situation of a plurality of submatrixs, each of submatrix preferably is set to be expert at and has predetermined regularity (regularity) in weight and the row weight.That is, if each of submatrix is set to have regularity in the row and column weight, it can be saved for storage H dOperand or the complexity of the perhaps memory space of parity check matrix H, and minimizing in the process of coding or decoding.At the H that has predetermined regularity aspect the row and column weight dEmbodiment be explained as follows.
In first embodiment, with H dBeing decomposed in the situation of submatrix of m (n-k)/m * k dimension, is under the condition of " 1 " in each row weight of specific submatrix, H dThe j of this specific submatrix of m submatrix capablely have a nearly W j(j=1,2 ..., (n-k)/m) individual continuous " 1 ".W jCan be consistent with the whole row of this specific submatrix.And Wj can increase or reduce brokenly for the whole row of this specific submatrix.
Fig. 4 and Fig. 5 at length explain first embodiment and illustrate to be included in H dIn the schematic diagram of specific submatrix example.
Specific submatrix shown in Figure 4 is the matrix with 7 * 28 dimensions.But the matrix that in fact is used for the LDPC coding is the larger matrix of matrix than 7 * 28 dimensions.In Fig. 4, every row of specific submatrix has four continuous " 1 ", and remaining value of corresponding line is zero (that is, the weight of every row is 4).And the weight of every row of this specific submatrix is 1.Fig. 5 illustrates an example, and every row of specific submatrix has W jIndividual continuous " 1 ", and remaining value of corresponding line is zero.But, W jChange brokenly for every row.In this case, the weight of every row of this specific submatrix also is 1.
At the H that has predetermined regularity aspect the row and column weight dSecond embodiment in, this H dComprise the individual matrix H of r (1-r) with (n-k) * (n-k) dimension (i) d[here r=k/n and i=1,2 .., r/ (1-r)], random H (i) dComprise m * m submatrix, its each have (n-k)/m * (n-k)/m dimension, and consist of this H dRow or the row weight of random submatrix be 1.
This H dCan comprise at least one H according to code rate (r=k/n) (i) d[i=1 here, 2 ..., r/ (1-r)].Code rate r is source data length k and the ratio of coded data length n, and usually uses r=1/2,2/3,3/4,4/5 etc.H (i) dBe the matrix with (n-k) * (n-k) dimension, and have H d=[H (1) d| H (2) d| ... | H (r/ (1-r)) d] relation.
Each H (i) dIt is characterized in that, in the situation that divided by m * m submatrix (its each have (n-k)/m * (n-k)/m dimension), consist of H dRow or the row weight of random submatrix be 1." m " is a positive integer, and corresponding to this H dResolution factor (resolutionfactor).Preferably, " m " chooses from 4-12, so that top performance to be provided.
Fig. 6 is illustrated in H in the m=4 situation (i) dExample, H thereon (i) dComprise 16 submatrixs (1,1), (1,2) ..., (4,4).The row of each submatrix or row weight are that 1 the fact refers to, and " 1 " is present among the random row or row of each submatrix, and its residual value of this random row or row is zero respectively.
If parity check matrix H be configured to have (n-k)/m by the characteristics of expanding above explanation * (n-k)/a plurality of submatrixs of m dimension, it is possible that random one row of this submatrix or row weight are set to 1.That is, as the H of the element of parity check matrix H dCan be set to just have following regularity, that is, consist of H dRow or the row weight of submatrix be 1, perhaps parity check matrix H can entirely be set to have following regularity, that is, the weight or the row weight that consist of each submatrix of whole parity check matrix H are 1.
In the coding or coding/decoding method that use the LDPC code, H dPerhaps parity check matrix H preferably provides two following characteristics.That is, use the coding of LDPC code or coding/decoding method by according to consisting of H dPerhaps the channel status between the row of parity check matrix H (channelstatus) is carried out in the mode of repeated exchanged probabilistic information judgement.But in failing to provide the parity check matrix H of following two characteristics, the probabilistic information of every row is transmitted to another row once, then returns, and does not repeat fully.Therefore, superperformance is unexpected.
Preferably, H dPerhaps any two random row of parity check matrix H are set to do not have simultaneously " 1 " on the whole at least two points.This H dThe two any random row fact that is set at least two points, not have simultaneously on the whole " 1 " refer to, at more whole H dTwo random row in, only have a point, have on it " 1 " that exists, with overlapping with another point with " 1 " of existing on it.
Fig. 7 illustrates the situation of a demonstration, and two row of this parity check matrix H side by side have " 1 " on two points.
With reference to figure 7, two points that represented by the circle of two capable closures of i and be set to " 1 ", H by two points that the circle of two capable closures of j represents dPerhaps parity check matrix H should be avoided in the situation shown in Fig. 7, so that use coding or the coding/decoding method of LDPC code that superperformance can be provided.This H dPerhaps two points of two row of parity check matrix H situation of side by side having " 1 " is known as the 4-circle.Therefore, this H dPerhaps any two random row of parity check matrix H fact of not having " 1 " at least two points refers to, on the whole not via H dPerhaps parity check matrix H forms the 4-circle.
Fig. 8 illustrates the situation of a demonstration, and all combinations make two row that choose from random three row of parity check matrix H have " 1 " at identical point respectively.In other words, all combinations make two row that choose from i, j and k are capable, that is, i and j is capable, j and k are capable, perhaps k and i are capable, have " 1 " at identical point.In Fig. 8, if six points of closed circle connect, form a circle.And this circle is known as the 6-circle.Fig. 9 is the bipartite graph that the difference as parity check matrix H represents, it refers to 6 row, 9 row parity matrixs.The matrix that thick line represents among Fig. 9 partly consists of the 6-circle.Therefore, all combinations make two row that choose from random three row of parity check matrix H be equal to or less than critical value C in the situation that identical point has " 1 " Max, this part that refers to for whole parity check matrix H formation 6-circle is equal to or less than this critical value C Max
Preferably, all combinations make from this H dTwo row that perhaps choose in random three row of parity check matrix H are equal to or less than the critical value C that presets in the situation that identical point has " 1 " respectively MaxEven at H dPerhaps there is the 6-circle in the parity check matrix H, preferred, critical value C MaxBe defined in the scope of the performance degradation that can avoid using the parity check matrix H Code And Decode.Further preferably, by the property enhancement result who measures and relatively cause owing to the reduction that is present in the 6-circle in the parity check matrix H, and be to reduce 6-to enclose necessary calculated load, with this critical value C MaxDetermine in rational scope.As the result of simulation, the critical value C between 10-500 MaxScope in can obtain gratifying performance.And estimated performance is better in the scope of 10-100.But, this critical value C MaxBe not limited to above-described scope.
Figure 10 is according to the H that has predetermined regularity in the row and column weight dFirst embodiment produce the flow chart of the process of parity check matrix H.The method that is described below is only demonstrated, and the parity check matrix H with aforesaid characteristics can produce with diverse ways.
At first, for this H dThe specific submatrix (the first submatrix) with (n-k)/m * k dimension, it is under 1 the condition, to have W that j row of this specific submatrix is configured to row weight at all row of this specific submatrix j(j=1,2 ..., (n-k)/m) individual " 1 " continuously is (S11).
Subsequently, for this H dRemaining submatrix among different submatrix (the second submatrix) carry out column permutation so that any random two row side by side do not have " 1 " (S12) at least two points.And this step S12 is sequentially put on remaining submatrix, to configure whole submatrix to last submatrix (m submatrix) (S13).This submatrix is combined to produce this H d(S14).And, H dAnd H pCombined to produce H (S15).
Figure 11 does not have the 4-circle for explain producing, but the 6-circle be equal to or less than C MaxThe flow chart of parity check matrix H process.The method of explaining is below only demonstrated, and the parity check matrix H with aforesaid characteristics can produce with diverse ways.
With reference to Figure 11, for by determining that the position that has the element of weight in (n-k) * k parity check matrix H produces this parity check matrix H, " i " is the index of any row of parity check matrix H, and " j " is the index of any row of parity matrix X, and C wExpression is the current weight number of row j arbitrarily.
At first, weight begins to be added to does not have weight (C w=0) first row (j=1) (S21).In this case, this weight increase refers to corresponding to any element of any row of row and is set to " 1 ".
Weight is increased arbitrarily i the candidate row (S22) to first row provisionally.This interim weight increase refers to, and the weight that increases to corresponding row is not final, but can change by follow-up step.Subsequently, judge whether in whole parity check matrix H, to exist 4-circle (S23).If there is the 4-circle, then weight does not increase to i row, but increases to another row (S22), and then carries out follow-up step.If there is no the 4-circle then judges whether to exist 6-circle (S24) in whole parity check matrix H.
As the result who judges whether to exist the 6-circle in parity check matrix H, if there is no 6-encloses, and then weight finally is added to i row.If have the 6-circle, judge whether that then the 6-number of turns order of whole parity check matrix H exceeds the critical value C that presets Max(S25).If the 6-number of turns order of whole parity check matrix H is no more than the critical value C that presets Max, then finally to be added to i capable for weight.If the 6-number of turns order of whole parity check matrix H surpasses the critical value C that presets Max, then weight does not increase to i capablely, but increases to another row (S22), and then carries out next step.
In case weight finally is added to i capable (S26), the current weight number C of j row wIncrease progressively 1 (S27).Then judge whether the current weight number C of j row wEqual j the admissible weight limit number C of row Jmax(S28).If the current weight number C of j row wEqual j and be listed as admissible weight limit number C Jmax, the weight that increases to the j row is terminated.And, judge whether that j equals code word size (S29).If the current weight number C of j row wBe not equal to j and be listed as admissible weight limit number C Jmax, then turn back to step S22, provisionally weight is increased the different row to the j row, to continue to carry out corresponding follow-up step.
If j equals code word size, this weight increase of whole parity check matrix H is terminated.Therefore, increase the result according to corresponding weight, this parity check matrix H can be by final produce (S31).
If j is not equal to code word size, this refers to and still has the row that do not increase weight.Therefore, by j being added 1 (S30), increased to next column in above-described mode from step S22 weight.
Such as what mention in the specification formerly, whole parity check matrix H can produce in above-mentioned mode.Alternatively, has [H d| H p] in the parity check matrix H of structure, H dProduced according to the step of above explanation, and can use and have machine made H p
Figure 12 is according to the H that has predetermined regularity in the row and column weight dSecond embodiment (comparison diagram 6) produce the flow chart of parity check matrix H process.
Compare with embodiment shown in Figure 11, embodiment shown in Figure 12 is further provided with a condition,, consists of this H that is dRow or the row weight of each submatrix of (n-k)/m * (n-k)/m dimension should be 1.The method that is described below is only demonstrated, and the parity check matrix H with aforesaid characteristics can produce with diverse ways.
With reference to Figure 12, " i " is the index of any row of parity check matrix H, and " j " is the index of any row of parity matrix X, and C wExpression is the current weight number of row j arbitrarily.
At first, weight begins to be added to does not have weight (C w=0) first row (j=1) (S40).In this case, the weight increase refers to corresponding to any element of any row of row and is set to " 1 ".
Weight is increased arbitrarily i the candidate row (S41) to first row provisionally.This interim weight increase refers to, and the weight that increases to corresponding row is not final, but can change by follow-up step.
Subsequently, if parity check matrix H is configured to have (n-k)/m * (n-k)/a plurality of submatrixs of m dimension, whether then check in the row of the capable submatrix that belongs to of i of j row or row, to exist to have weight and be equal to or greater than 2 row or column (S42).If exist to have weight and be equal to, or greater than 2 row or row in the row of the capable submatrix that belongs to of i of j row or row, then weight does not increase to i capablely, but increase to different row (S41), and next step is performed.If in the row of the capable submatrix that belongs to of i of j row or row, do not exist to have weight and be equal to, or greater than 2 row or row, then judge whether in whole parity check matrix H, to exist 4-circle (S43).
If according to the judgement that whether in whole parity check matrix H, has the 4-circle, there is 4-circle (S43), then weight does not increase to i row, but increase is to another row (422), and then next step is performed.If there is no the 4-circle then judges whether to exist 6-circle (S44) in whole parity check matrix H.
As the result who judges whether to exist the 6-circle in parity check matrix H, if the 6-circle does not exist, then weight finally is added to i row.If have the 6-circle, judge whether that then the 6-number of turns order of whole parity check matrix H exceeds the critical value C that presets Max(S45).If the 6-number of turns order of whole parity check matrix H is no more than the critical value C that presets Max, then finally to be added to i capable for weight.If the 6-number of turns order of whole parity check matrix H surpasses the critical value C that presets Max, then weight does not increase to i capablely, but increases to another row (S42), and then carries out next step.
In case this weight finally is added to i capable (S46), the current weight number C of j row wIncrease progressively 1 (S47).Then judge whether the current weight number C of j row wEqual j and be listed as admissible weight limit number C Jmax(S48).If the current weight number C of j row wEqual j and be listed as admissible weight limit number C Jmax, the weight that increases to the j row is terminated.And, judge whether that j equals code word size (S49).
If the current weight number C of j row wBe not equal to j and be listed as admissible weight limit number C Jmax, it turns back to step S42, provisionally weight being increased the different row to j row, and continues to carry out the step of corresponding back.
If j equals code word size, this weight increase of whole parity check matrix H is terminated.Therefore, increase the result according to corresponding weight, this parity check matrix H can be by final produce (S51).
If j is not equal to code word size, this refers to and still has the row that do not increase weight.Therefore, by j being added 1 (S50), increased to next column in above-mentioned mode from step S22 weight.
Such as what mention in the specification formerly, whole parity check matrix H can produce in above-mentioned mode.Alternatively, has [H d| H p] in the parity check matrix H of structure, this H dProduced according to the step of above explanation, and can use and have machine made H pPreferably, (n-k) * (n-k) dual diagonal matrix of dimension is used as H p
In Fig. 1, receiver 30 receives and uses the equation 4 in the above described manner coded data of decoding.
[equation 4]
H·c＝0
That is, if " 0 " produces from coded data c be multiply by the parity check matrix H, this means not have transmission error.If " 0 " does not produce from coded data c be multiply by the parity check matrix H, this means to have transmission error.Therefore, source data can be correspondingly independently.
Should be understood that technical spirit of the present invention and scope can expand to CPU (controlled processing unit) readable medium recording program performing, such as CD-ROM, floppy disk, computer storage, mobile communication terminal memory etc.In this case, the H or the H that have these characteristics dData structure, and be used for to generate H or the H with these characteristics dProgram be recorded in this recording medium.
Industrial applicibility
Therefore, the Code And Decode method of use LDPC code of the present invention is applicable to the communication system such as mobile communication system, portable Internet system etc.
Although described with reference to its preferred embodiment and for example clear the present invention, for those skilled in the art, it is evident that, do not break away from the spirit and scope of the present invention, can carry out therein various modifications and variations.Therefore, this invention is intended to cover it and be included into modifications and variations of the present invention within appended claim and the equivalent scope thereof.

Claims (16)

1. one kind is used LDPC (low-density checksum) yard method of encoding, and the method comprising the steps of: use to have H=[H d| H p] the parity check matrix H coded input data of structure, wherein H d(n-k) * k dimension, H pBe (n-k) * (n-k) dimension, k is the bit number of input data, and n is the bit number of code word,
If H wherein dComprise a plurality of submatrixs, each submatrix is expert at or row weight aspect has predetermined regularity, and whole H wherein dAny random two row do not list at least two and have simultaneously 1.
2. if according to claim 1 method is H wherein dComprising that m has the submatrix of (n-k)/m * k dimension, is under 1 the condition in the row weight of the whole row of specific submatrix, and the j of this specific submatrix of this m submatrix is capable to have W jIndividual continuous 1.
3. according to claim 2 method is wherein for this W of whole row of this specific submatrix jConsistent.
4. according to claim 2 method is wherein for this W of whole row of specific submatrix jIncrease brokenly or reduce.
5. if according to claim 1 method is H wherein dComprise m the submatrix with (n-k)/m * k dimension, then consist of H dRow or the row weight of random submatrix be 1.
6. according to claim 1 method, wherein all make from whole H or whole H dRandom three row in two row that the choose situation that has 1 combination at identical point be equal to or less than the critical value (C that presets Max).
7. according to claim 6 method, wherein this critical value (C Max) be the arbitrary value in 10～100 scopes.
8. according to claim 1 method, wherein this H pBe that wherein leading diagonal is 1 with direct diagonal below leading diagonal, and wherein all the other are dual diagonal matrix of zero.
9. one kind is used LDPC (low-density checksum) yard method of decoding, and the method comprising the steps of: use to have H=[H d| H p] parity check matrix H decoding input data, the wherein H of structure d(n-k) * k dimension, H pBe (n-k) * (n-k) dimension, k is the bit number of input data, and n is the bit number of code word,
If H wherein dComprise a plurality of submatrixs, each of submatrix is expert at or row weight aspect has predetermined regularity, and whole H wherein dAny random two row do not list at least two and have simultaneously 1.
10. if according to claim 9 method is H wherein dComprising that m has the submatrix of (n-k)/m * k dimension, is under 1 the condition in the row weight of the whole row of specific submatrix, and the j of this specific submatrix of this m submatrix is capable to have W jIndividual continuous 1.
11. method according to claim 10 is wherein for this W of whole row of this specific submatrix jConsistent.
12. method according to claim 10 is wherein for the whole row W of specific submatrix jIncrease brokenly or reduce.
13. method according to claim 9, if H wherein dComprise m the submatrix with (n-k)/m * k dimension, then consist of this H dRow or the row weight of random submatrix be 1.
14. method according to claim 9, wherein all make from whole H or whole H dRandom three row in two row that the choose situation that has 1 combination at identical point be equal to or less than the critical value (C that presets Max).
15. method according to claim 14, wherein this critical value (C Max) be the arbitrary value in 10～100 scopes.
16. method according to claim 9, wherein this H pBe that wherein leading diagonal is 1 with direct diagonal below leading diagonal, and wherein all the other are dual diagonal matrix of zero.
CN2005800255866A 2004-07-27 2005-07-26 Method of encoding and decoding using low density parity check code Expired - Fee Related CN101305521B (en)

Applications Claiming Priority (7)

Application Number Priority Date Filing Date Title
KR1020040058647A KR101073907B1 (en) 2004-07-27 2004-07-27 Encoding method by using LDPC code and computer-readable medium for the same
KR10-2004-0058647 2004-07-27
KR1020040060839A KR101079087B1 (en) 2004-08-02 2004-08-02 Encoding method by using LDPC code and computer-readable medium for the same
KR10-2004-0060839 2004-08-02
KR1020040062068A KR20060013197A (en) 2004-08-06 2004-08-06 Encoding method by using ldpc code and computer-readable medium for the same
KR10-2004-0062068 2004-08-06
PCT/KR2005/002421 WO2006011744A2 (en) 2004-07-27 2005-07-26 Method of encoding and decoding using low density parity check code

Publications (2)

Publication Number Publication Date
CN101305521A CN101305521A (en) 2008-11-12
CN101305521B true CN101305521B (en) 2013-01-02

Family

ID=37120451

Family Applications (1)

Application Number Title Priority Date Filing Date
CN2005800255866A Expired - Fee Related CN101305521B (en) 2004-07-27 2005-07-26 Method of encoding and decoding using low density parity check code

Country Status (2)

Country Link
KR (1) KR101073907B1 (en)
CN (1) CN101305521B (en)

Families Citing this family (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
KR100929080B1 (en) * 2006-02-20 2009-11-30 삼성전자주식회사 Signal transceiving device and method in communication system
KR101455978B1 (en) 2007-03-27 2014-11-04 엘지전자 주식회사 Method for encoding data using a Low Density Parity Check code
KR101512078B1 (en) * 2013-02-07 2015-04-14 최수정 Method and device of encoding/decoding using specific density based sparse inverse code
KR101512081B1 (en) * 2013-02-07 2015-04-14 최수정 Method and device of encoding/decoding using sparse matrix based on specific density
CN107017965A (en) * 2017-01-17 2017-08-04 北京银河信通科技有限公司 A kind of permanent envelope code modulating method of high-order applied to Unmanned Aerial Vehicle Data Link
US10944429B1 (en) * 2020-01-02 2021-03-09 Silicon Motion, Inc. Data accessing method using data protection with aid of parity check matrix having partial sequential information, and associated apparatus

Family Cites Families (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US6895547B2 (en) 2001-07-11 2005-05-17 International Business Machines Corporation Method and apparatus for low density parity check encoding of data

Also Published As

Publication number Publication date
KR101073907B1 (en) 2011-10-17
CN101305521A (en) 2008-11-12
KR20060010068A (en) 2006-02-02

Similar Documents

Publication Publication Date Title
CN101128981B (en) Method of encoding and decoding using low density parity check matrix
CN101395804B (en) Method of encoding and decoding using ldpc code
CN100571044C (en) Code And Decode method with the self-adapting changeable code check of LDPC sign indicating number
CN100583650C (en) Method of encoding and decoding using LDPC code and apparatus thereof
CN101305521B (en) Method of encoding and decoding using low density parity check code
JP4672015B2 (en) Encoding and decoding method using low density parity check code
US20080294963A1 (en) Method and apparatus for designing low density parity check code with multiple code rates, and information storage medium thereof
US8443254B2 (en) Method of generating a parity check matrix for LDPC encoding and decoding
CN110352562B (en) Method for performing encoding based on parity check matrix of LDPC code in wireless communication system and terminal using the same
CN102301603A (en) Coding and decoding using ldpc quasi-cyclic codes
US20230421177A1 (en) Apparatus and method for channel encoding/decoding in communication or broadcasting system
KR101079087B1 (en) Encoding method by using LDPC code and computer-readable medium for the same
KR20060013197A (en) Encoding method by using ldpc code and computer-readable medium for the same
KR101435831B1 (en) Method of generating parity check matrix
WO2006031062A2 (en) Method of encoding and decoding using ldpc code and apparatus thereof
KR20090086744A (en) Method of encoding/decoding data using ldpc code and parity check matrix therefor

Legal Events

Date Code Title Description
C06 Publication
PB01 Publication
C10 Entry into substantive examination
SE01 Entry into force of request for substantive examination
C14 Grant of patent or utility model
GR01 Patent grant
CF01 Termination of patent right due to non-payment of annual fee

Granted publication date: 20130102

Termination date: 20200726

CF01 Termination of patent right due to non-payment of annual fee