CN104052501A - Multi-system LDPC decoding method low in complexity - Google Patents

Multi-system LDPC decoding method low in complexity Download PDF

Info

Publication number
CN104052501A
CN104052501A CN201410295270.XA CN201410295270A CN104052501A CN 104052501 A CN104052501 A CN 104052501A CN 201410295270 A CN201410295270 A CN 201410295270A CN 104052501 A CN104052501 A CN 104052501A
Authority
CN
China
Prior art keywords
information
confidence
codeword
check
th
Prior art date
Application number
CN201410295270.XA
Other languages
Chinese (zh)
Other versions
CN104052501B (en
Inventor
黄勤
王祖林
张睦
Original Assignee
北京航空航天大学
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
Application filed by 北京航空航天大学 filed Critical 北京航空航天大学
Priority to CN201410295270.XA priority Critical patent/CN104052501B/en
Publication of CN104052501A publication Critical patent/CN104052501A/en
Application granted granted Critical
Publication of CN104052501B publication Critical patent/CN104052501B/en

Links

Abstract

The invention provides a multi-system LDPC decoding method low in complexity, and belongs to the technical field of communication. In the method, the representation mode that the binary system with multi-system symbols is used for representing the simplified codon information confidence coefficient is adopted, the confidence coefficient of side information is updated through calculation of check nodes during each time of iteration, using efficiency for the side information by weighted and strengthened variable nodes is introduced into calculation of the variable nodes, and codon information and extrinsic information are calculated by adopting the information updating mode of the binary system. According to the method, the information length of each symbol of codons is far smaller than the length 2r in an existing method, the side information only has one finite field symbol and the confidence coefficient of the finite field symbol, the storage complexity is very low, calculation of the variable nodes and the check nodes is mainly integer addition and integer comparison operation, only little finite field operation and multiply operation are adopted, calculation complexity is very low, and very good decoding performance can be achieved for codes with various lengths and weights.

Description

低复杂度的多进制LDPC码译码方法 A low complexity method of decoding binary LDPC codes

技术领域 FIELD

[0001] 本发明涉及一种低复杂度的多进制LDPC码译码方法,属于通信技术领域。 [0001] The present invention relates to a low-complexity binary LDPC code decoding method, belonging to the field of communications technologies.

背景技术 Background technique

[0002] 低密度奇偶校验(LDPC)码是一种重要的纠错码,它的奇偶校验矩阵具有稀疏特性,采用置信传播(BP)算法时,LDPC码的性能能够接近香农极限。 [0002] Low Density Parity Check (LDPC) codes is an important error correction code, which has a sparse parity check matrix characteristics, using belief propagation (BP) algorithm upon the performance of the LDPC code can approach the Shannon limit. 目前已经有多种通信系统,例如数字卫星电视(DVB-S2)、第4代移动通信系统(4G)、全球卫星导航系统(GPS)等, 在其数据传输中采用LDPC码进行差错控制。 There are already a variety of communication systems, such as digital satellite (DVB-S2), fourth generation mobile communication systems (4G), the Global Navigation Satellite System (GPS) and the like, error control using the LDPC code in which the data transmission. 多进制LDPC码能够获得比二进制LDPC码更优异的性能,但是并未像二进制LDPC码一样获得迅速的推广。 Binary LDPC codes can get more excellent than binary LDPC code performance, but did not like the binary LDPC codes as access to rapid promotion. 这主要是因为目前并没有一种能兼顾译码性能和计算、存储复杂度的多进制LDPC码译码算法。 This is mainly because there is none able to combine and binary LDPC code decoding algorithm and decoding performance computing, storage complexity.

[0003] 多进制LDPC码的译码基于码校验矩阵的二分图,或者叫做Tanner图,图1所示即为一个LDPC码的二分部,它由变量节点、校验节点和连接两种节点的边构成。 Coding [0003] based on ary LDPC code bipartite graph check matrix codes, or called Tanner graph, that is, a two division LDPC code shown in FIG. 1, which consists of the variable nodes, check nodes and connecting the two edge node configuration. 其译码过程主要分为4个部分:码字信息的存储、边信息的存储、校验节点的计算和变量节点的计算。 Its decoding process is divided into four parts: the codeword information storage, storage, and variable node calculation of the check node side information. 现有的多进制LDPC码译码方法主要有两种,一种基于置信传播(BP)算法,另一种基于大数逻辑译码(MLgD)算法。 Conventional binary LDPC code decoding methods are mainly two types of belief propagation (BP) algorithm, another majority logic decoding (MLgD) algorithm. 这两种方法具有类似的码字信息,假设多进制LDPC码校验阵是有限域GF(2〇下的矩阵,则每个码字符号的信息均为一个长度为f的置信度向量,表征该符号分别为有限域GF(2〇下所有f个元素的可能性度量,需要存储f个置信度信息。而基于MLgD算法的多进制LDPC码译码方法具有比基于BP算法的多进制LDPC码译码方法简单的多的边信息。基于BP算法的多进制LDPC码译码方法的边信息,基于BP算法的不同实现方式,可以等于码字的信息向量,其长度为每条边需要存储f个有限域元素的置信度; 也可以根据码字信息向量中不同元素的置信度大小,只取置信度较大的部分元素,假设其长度为nm(实际中nm >> 1),这样每条边需要存储nm个有限域元素的值和相应的nm个置信度的值。基于MLgD算法的多进制LDPC码译码方法的边信息只由码字信息中置信度最大的那个元素构成,每条边只 Both methods have similar codeword information, assuming binary LDPC code parity matrix is ​​a finite field GF (2〇 in matrix, each of the information symbols are a codeword length of the confidence vector f, symbol characterizing the finite field GF (2〇 f respectively likelihood metrics for all elements necessary to store the f confidence information while binary LDPC code decoding algorithm based MLgD than BP algorithm based on a multiple Input the method of LDPC code decoding system much simpler side information. binary LDPC codes side information decoding method BP algorithm based on BP algorithm different implementations, the information may be equal to the vector codeword, the length of each f edges need to store a finite field element confidence; codeword information can also be based in the confidence vector elements of different sizes, just take a larger part of the element of confidence, assuming the length of nm (in practice nm >> 1) , so that each side needs to be stored nm finite field element values ​​and corresponding confidence values ​​nm based on binary LDPC codes side information decoding method MLgD algorithm only by the maximum codeword information in confidence that element configuration, each edge only 要存储1个有限域元素和1个置信度的值。多进制LDPC码译码中,校验节点的计算和变量节点的计算基于码字信息和边信息进行。基于BP算法的多进制LDPC码译码方法中,校验节点对每两个输入的边信息的计算复杂度为q 2量级(或者》;;, m量级,对于边信息长度为nm的实现),变量节点对每两个输入的边信息的计算复杂度为q 量级(或者2nm,对于边信息长度为nm的实现)。因为基于MLgD算法的多进制LDPC码译码方法的边信息更为简单,因此其校验节点和变量节点对每两个输入的边信息的计算复杂度均为1,是基于BP算法的多进制LDPC码译码方法的Ι/q 2 (或者1/< )。 To store a finite field elements values ​​and a confidence level. Ary LDPC code decoding, the variable node calculation and a check node-based side information and the codeword information. BP algorithm based ary LDPC code decoding process, check node calculation complexity side information for each of two inputs of the order q 2 (or ";;, m the order side for the message length to achieve nm), the variable node computational complexity side information for each of two inputs of the order q (2nm Alternatively, side information for the length of the implement nm). because the side-information based on binary LDPC code decoding method MLgD more simple algorithm, so which check node and variable node calculation complexity of information for each side of the two inputs are 1, it is based Ι / q 2 (or 1 / <) binary LDPC code decoding method BP algorithm.

[0004] 虽然基于MLgD算法的多进制LDPC码译码方法的存储和计算复杂度均远低于基于BP算法的多进制LDPC码的译码方法,但是基于MLgD算法的多进制LDPC码译码方法的性能非常依赖于LDPC码校验矩阵的列重。 [0004] Although based storage and computational complexity ary LDPC Decoding method MLgD algorithm decoding method are much lower than binary LDPC codes based on BP algorithm, but the algorithm-based MLgD binary LDPC codes the method of decoding performance is very dependent on the column weight of parity check matrix of the LDPC code. 当码的列重较低(小于6)时,其性能损失非常严重, 并且在误比特率(BER)高达ΚΓ 6左右即可能出现错误平台。 When the code column weight low (less than 6), which is a very serious loss of performance, and as high as ΚΓ 6 bit error rate (BER) may occur i.e. error internet. 目前并没有一种多进制LDPC 码译码方法能普适地在对各种参数的LDPC码译码时,在译码性能和计算、存储复杂度之前取得很好的均衡。 There is no currently-ary LDPC Decoding method can be universal in decoding of LDPC codes upon various parameters, to obtain well-balanced and before decoding performance computing, storage complexity.

发明内容 SUMMARY

[0005] 本发明的目的是为了解决多元域LDPC码译码复杂度和性能无法兼顾的问题,提出一种低复杂度的多进制LDPC码译码方法。 [0005] The object of the present invention is to solve the problem of the complexity and performance of the LDPC code decoding polyhydric domain can not take into account the proposed multi-LDPC Decoding method of low complexity. 本发明所述的译码方法是一种基于二分图的迭代译码方法,本方法在迭代中采用多进制符号的二进制表示简化码字信息置信度的表示方式,在每次迭代通过校验节点的计算更新边信息的置信度,在变量节点的计算中引入加权加强变量节点对边信息的使用效率,并且采用二进制的信息更新方式计算码字信息和外信息。 The decoding method according to the present invention is an iterative decoding method based on a bipartite graph, the method using a binary-ary symbol representation represented in a simplified iteration codeword information confidence, by checking at each iteration computing node update information confidence side, introduced in the calculation of the variable node weighting reinforcing efficiency in the use of the variable node side information, and calculates a codeword information and extrinsic information binary information update.

[0006] LDPC码译码方法首先进行译码初始化,然后迭代进行如下步骤:硬判决、计算边信息、校验节点计算以及变量节点计算,直到译码成功或失败。 [0006] LDPC code decoding method for decoding first initialized, then iterate the following steps: hard decision, calculating the side information, the check node calculation and variable node calculation, until decoding succeeds or fails. 本发明提供的低复杂度的多进制LDPC码译码方法,对码字信息,均采用二进制形式表示码字符号,在译码过程中,进行了如下方面的改进: The present invention provides a low-complexity binary LDPC code decoding method, codeword information, are expressed as binary codeword symbols, the decoding process, improved in the following aspects:

[0007] 1)译码初始化时,设定加权因子,使用比特置信度将码字信息进行初始化; [0007] 1) the decoding initialization, setting the weighting factors, using the confidence bit codeword information is initialized;

[0008] 2)校验节点和变量节点连接边的外部校验和的置信度转换成比特置信度向量,参与变量节点计算; [0008] 2) check nodes and variable nodes connected to the outer side of the confidence check bit and into confidence vector, participating variable node calculation;

[0009] 3)每次迭代更新的码字信息和校验节点的外信息都为比特置信度向量。 [0009] 3) each outer iteration update information and the codeword information bits are check nodes confidence vector.

[0010] 本发明译码方法,进行变量节点计算时,校验节点输入变量节点的信息,采用基于汉明距离的置信度加权方式来获得,具体是:根据校验节点和变量节点连接边的外部校验和的二进制表示与码字符号硬判决的二进制表示之间的汉明距离的大小设定加权因子,使用加权因子对连接边的外部校验和的置信度进行加权。 [0010] The decoding method of the present invention, when the variable node calculation performed, the check node information input variable node, based weighting scheme Hamming distance confidence obtained, in particular: The check nodes and variable nodes connected to the edges external test set and binary weighting factor represents the magnitude of the Hamming distance between the binary representation of the hard decision codeword symbols using the weighting factors of edges connecting external confidence are weighted checksum.

[0011] 本发明译码方法,在边信息计算时,边信息包括码字外信息的多进制硬判决符号及该符号的置信度;其中,符号的置信度为码字的二进制位外信息中所有绝对值中的最小值。 [0011] The decoding method of the present invention, when calculating the side information, the side information comprising an outer confidence ary codeword hard decision symbol and said information symbol; wherein the symbol is the confidence bit code words to the information All the absolute value of the minimum value.

[0012] 本发明译码方法中,所有码字信息、码字外信息、外部校验和的置信度、以及校验节点输入给变量节点的信息均为整数。 [0012] The decoding method of the present invention, all codeword information, information to the code words, the confidence and the external test, and input check node to variable node information are integers.

[0013] 本发明的优点和积极效果在于: [0013] The advantages and positive effects of the present invention:

[0014] (1)本发明中码字每个符号的信息长度仅为r,远低于现有算法的码字符号信息长度f ;同时,其边信息仅有一个有限域符号及其置信度,具有很低的存储复杂度; [0014] (1) of the present invention, the information length of each symbol in the codeword is only r, much lower than the information symbol codeword length f of existing algorithms; the same time, side information which is only a finite domain symbols and a confidence , having low complexity memory;

[0015] (2)本发明的变量节点和校验节点的计算主要为整数加法和整数比较运算,只有少量的有限域运算和乘法运算,具有很低的计算复杂度; [0015] (2) calculation of the variable node and a check node of the present invention mainly integers integer addition and comparison operations, only a small number of finite field multiplication operations and having very low computational complexity;

[0016] (3)本发明对于多进制LDPC码的码重不敏感,对于各种码长和码重的码均可获得很好的译码性能。 [0016] (3) of the present invention is insensitive to binary LDPC codes of code weight, for a variety of code length and code weight code can get good decoding performance.

附图说明 BRIEF DESCRIPTION

[0017] 图1是一个5行10列的LDPC码校验矩阵的二分图表示; [0017] FIG. 1 is a 5 rows and 10 columns of the LDPC code bipartite graph representation of parity check matrix;

[0018] 图2是本发明的多进制LDPC码译码方法的流程图。 [0018] FIG 2 is a flowchart ary LDPC code decoding method according to the present invention.

具体实施方式 Detailed ways

[0019] 下面将结合附图和实施实例对本发明作进一步的详细说明。 [0019] The accompanying drawings and the following examples of embodiments of the present invention will be further described in detail.

[0020] 如图2所示,为多进制LDPC码译码方法的整体流程图,首先进行译码初始化,然后迭代进行:硬判决、计算边信息、校验节点计算以及变量节点计算,直到译码成功或失败。 [0020] As shown, for the entire flowchart ary LDPC code decoding method, decoding is first initialized 2 then iteratively: hard decision, calculating the side information, the check node calculation and variable node calculation, until decoding success or failure. 本发明的低复杂度的多进制LDPC码译码方法,对码字信息的码字符号均采用二进制形式表示,在译码过程中,进行了如下方面的改进: Low complexity of the present invention is binary LDPC code decoding method, codeword symbol codeword information are represented in binary form, in the decoding process, improved in the following aspects:

[0021] 1)使用比特置信度将码字信息进行初始化; [0021] 1) Use the confidence bit codeword information is initialized;

[0022] 2)校验节点和变量节点连接边的外部校验和的置信度转换成比特置信度向量,参与变量节点计算; [0022] 2) check nodes and variable nodes connected to the outer side of the confidence check bit and into confidence vector, participating variable node calculation;

[0023] 3)每次迭代更新的码字信息和校验节点的外信息都为比特置信度向量。 [0023] 3) each outer iteration update information and the codeword information bits are check nodes confidence vector.

[0024] 下面结合具体实例来说明本发明的多进制LDPC码译码方法。 [0024] Specific examples will be described below with reference to binary LDPC code decoding method of the present invention.

[0025] 假设一个多进制LDPC码由其有限域GF(2〇下大小为mXn的奇偶校验矩阵Η的零空间定义,则该多进制LDPC码的码字是一个长度为η的GF(2〇下的向量,该码字向量可以用一个长度为nr的二进制向量等效表示。用c= (Cl,c2,…,cn)表示多进制码字,用Cj(l彡j彡η)表示码字的第j个符号,用Cj = (Cp Cj,2,…,Cj,J表示第j个符号的二进制表示,Cj,t表示第j个符号的第t个二进制位,值为0或1。当通信系统使用BPSK调制方式进行传输时,对码字二进制表示的每一比特做如下映射:〇- +1V,1 - -IV。经过二进制输入加性高斯白噪声(BI-AWGN)信道后,系统接收到的码字信息为y = (yi,y2, . . .,yn),其中乃=(y^ y"2,...,yy)为第j个码字符号r个比特的对数似然比 [0025] Suppose a binary LDPC code by finite field GF (2〇 size under the null space defined in the mXn parity check matrix Η is, the M-ary codeword is an LDPC code of length η GF (vector 2〇 in the codeword vector can be a binary vector of length equivalent representation nr. by c = (Cl, c2, ..., cn) represents a multi-ternary code word with Cj (l j San San [eta]) denotes the j-th symbol codeword by Cj = (Cp Cj, 2, ..., Cj, J represents a binary j-th symbol represents, Cj of, t represents a t-th bit of the j th symbol, the value of is 0 or 1. when the communication system uses BPSK modulation for transmission, each bit of the binary representation of the code word mapping follows: 〇- + 1V, 1 - -IV via binary input additive white gaussian noise (BI-. AWGN) channel after the system information received codeword is y = (yi, y2,..., yn), where is the = (y ^ y "2, ..., yy) for the j-th codeword symbols r bit log likelihood ratio

1彡j彡n,1彡t彡r。 1 San San j n, 1 San San t r. 多进制LDPC码译码器使用y和Η进行译码。 Ary LDPC decoder using Η y and decoding. 译码过程按以下步骤进行: Coding process performed by the following steps:

[0026] 步骤一、进行译码初始化设置。 [0026] Step a, decoding initialization settings. 设定加权因子Θ & Θ i,…,θ ρ设定最大迭代次数Imax。 Setting the weighting factor Θ & Θ i, ..., θ ρ set the maximum number of iterations Imax. 当前迭代次数k设为1。 The current iteration number k is set to 1.

[0027] 使用比特置信度将码字信息进行初始化,是将y以量化间隔Λ和量化比特数ω 均匀量化为整数赋值给第1次迭代的码字信息。 [0027] Using the confidence bit codeword information is initialized, y is a quantization interval Λ ω quantizing bits uniformly quantized to integer to the first iteration of codeword information. 本发明实施例按如下量化方式将码字信息初始化为整数: Example embodiments of the present invention will in the following manner quantized codeword information is initialized to an integer:

[0028] [0028]

(1) (1)

[0029] 其中1彡j彡η,1彡t彡r,ω、Λ分别为量化比特数和量化间隔,Ν为整数, 且-(2^-1) < Ν < 2〜-1,yj,t表示系统接收到的二进制码字信息y的第j个码字符号的第t个二进制位信息, [0029] wherein [eta] 1 San San j, 1 San San t r, ω, Λ respectively quantization bit number and a quantization interval, v is an integer and - (2 ^ -1) <Ν <2~-1, yj, t t th represents the j-th information bit codeword symbols received binary system codeword information of y,

的上角标1表示当前迭代次数, The superscript 1 indicates that the current number of iterations,

表示第1次迭代中第j个码字符号第t个二进制位信息。 1 represents the j-th iteration of t codeword symbols binary bits of information.

[0030] 通过式⑴得到第1次迭代的第j个码字符号信息 [0030] to obtain a first iteration j-th information symbol codeword by formula ⑴

进而得到第1次迭代的码字信息 Further information of the first codeword obtained iteration

[0031] 步骤二、对第k次迭代的码字信息进行硬判决: [0031] Step II of the k-th codeword information iteration hard decisions:

[0032] [0032]

(2) (2)

[0033] 其中 [0033] in which

是第k次迭代的码字信息,如果k = 1,则 Is the k-th iteration codeword information, if k = 1, then

在步骤一中初始化得到,否则 In a step to get initialized, otherwise

在上一次迭代的步骤五中更新得到, Steps on the next iteration of the Fifth been updated,

表示对第k次迭代的第j个码字符号第t个二进制位信息 K represents the j-th iteration of t codeword symbols to binary bits of information

的硬判决结果。 The hard decision result. 假设 Hypothesis

是多进制符号zf的二进制表示,则zf是码字第j个码字符号的硬判决结果。 Ary symbol is the binary representation of zf, zf is the hard decision codeword j th codeword symbols results. 将多进制硬判决向量 The hard-decision vector-ary

用校验矩阵Η进行校验。 Be verified with the check matrix Η. 如果z(k)HT = 0,则译码成功;如果z(k)HT尹0且k > Imax,则译码失败; 否则,进行步骤三。 If z (k) HT = 0, the decoding is successful; if z (k) HT Yin 0 and k> Imax, the decoding fails; otherwise, it proceeds to step three. 图2中判断奇偶校验结果是否为0就是判断z(k)HT是否为0,判断迭代次数时候超过最大迭代次数就是判断k是否大于1_。 In FIG. 2 determines whether a parity check result is judged as 0 z (k) HT is zero, it is judged when the number of iterations exceeds a maximum number of iterations is determined whether k is larger than 1_.

[0034] 步骤三、对于所有1彡j彡n, ie Μ」,计算第j个变量节点到第i个校验节点的边信息,其中,M」表示校验矩阵Η的第j列所有非零元素行位置的集合,假设Η的第i行第j 列的元素为比,」,则Μ」={i : 1彡i彡m,比,」关0}。 [0034] Step three, for all j San San 1 n, ie Μ "calculates the j-th variable node side information to the i-th check node, wherein, M 'represents the j-th column of all non-check matrix Η zero element row set position, the element j-th column of the i-th row is assumed Η than, "then Μ" = {i: 1 San i San m, ratio, "off 0}.

[0035] 首先,计算第k次迭代的码字外信息。 [0035] First, calculation of the k-th iteration of the outer codeword information. 如果本次迭代为第一次迭代,即k= 1,则码字外信息等于码字信息。 If the current iteration is the first iteration, i.e., k = 1, then the codeword information is equal to an outer codeword information. 校验节点的第j个码字符号的二进制位外信息 Bit information of the outer codeword symbols j check nodes

等于第一次迭代的第j个码字符号的二进制位信息 Information bit equal to the first iteration of the j-th code word symbols

如式(3): The formula (3):

[0036] [0036]

(3) (3)

[0037] 第k(k > 2)次迭代时,码字外信息由本次迭代的码字信息 To [0037] of k (k> 2) iterations, the outer codeword from the codeword information of the current iteration

减去上一次迭代时校验节点输入给与该码字符号对应的变量节点的信息 Input information given check node the variable node corresponding to the codeword symbols while subtracting the last iteration

得到: get:

[0038] [0038]

(4) (4)

[0039] 其中,1彡t彡r, [0039] wherein t 1 San San r,

表示第k次迭代时,对于第i个校验节点,第j个码字符号第t个二进制位的外信息。 K represents the iteration when, for the i th check node, the j-th codeword symbols t-th bit of the external information.

为上一次迭代时第i个校验节点输入给与第j个码字符号对应的变量节点的第t个二进制位信息。 Is the previous iteration i-th check node j-th input to AND codeword symbols corresponding to the t-th variable node information bit. 第j个码字符号对应的变量节点也就是第j个变量节点。 J-th variable node corresponding to a codeword symbol is the j-th variable node. (3)式和(4)式中所求的 (3) and (4) as required in the formula

的数值范围限定在(_2^2+1,2^ 2-1)范围内,当所求结果超出该范围时,设定 Numerical ranges defined in (_2 ^ 2 + 2 ^ 2-1) within the range, when the result exceeds the required range, setting

的值为最接近的边界值。 It values ​​nearest boundary value.

[0040] 然后,根据码字外信息计算变量节点到校验节点的边信息。 [0040] Then, according to the code words information calculation variable node to check node side information. 本步骤中的边信息分为两个部分,第一部分为码字外信息的多进制硬判决符号 In this step, the side information is divided into two portions, a first portion of the hard decision symbol is a multi-value information to the code words

,而 ,and

的二进制表示的每一位都是码字外信息相应二进制位的硬判决: Each and every one code word corresponding bits of information outside of the binary representation of the hard decision:

[0041] [0041]

(5) (5)

[0042] 其中<1·,表示^上的二进制表示的第t位,1彡t彡r ;边信息的第二部分为码字的r个二进制位外信息绝对值的最小值,S卩 [0042] where <1., represents a ^ t bit binary representation, 1 San San t r; the minimum value of the second portion of the side information into code words of r bits of the absolute value of the extrinsic information, S Jie

,它表示符号的置信度。 It represents a symbol of confidence. 这里,角标j - i表示连接第j个变量节点和第i个校验节点的边。 Here, a subscript j - i denotes the j-th edge connected to the i-th variable node and check nodes.

[0043] 步骤四、校验节点计算。 [0043] Step 4 check node calculation. 首先,对于所有1 < i < m,计算第i个校验节点的校验和 First, for all 1 <i <m, the checksum is calculated in the i-th check node and

[0044] [0044]

(6) (6)

[0045] 队表示校验矩阵Η的第i行所有非零元素行位置的集合,队= {j: 1 < j < n,hy 关0}。 [0045] Force check matrix represents the i-th row Η all non-zero elements of the set of row position, team = {j: 1 <j <n, hy Off 0}.

[0046] 连接校验节点和变量节点的边包括两部分信息:第一部分为该边的外部校验和, 第二部分为该边外部校验和的置信度;其中,边外部校验和的置信度的确定方法是:首先, 获得与校验节点相连的所有边的置信度的最小值MIR和次小值MIN 2,与校验节点相连的置信度最小的边的外部校验和的置信度为MIN2,剩余与校验节点相连的边的外部校验和的置信度均为MI&。 [0046] The check node connected to variable nodes and edges including two pieces of information: a first outer edge portion for the checksum, for the second portion outer side confidence checksum; wherein the outer edge of the checksum the method of determining the confidence level are: first, a minimum confidence connected MIR confidence of all edges connected to the check nodes and the second smallest value MIN 2, the minimum external check nodes and check edges confidence degree MIN2, the confidence side connected to external test and check nodes are remaining MI &. 下面以与第i个校验节点相连的边的信息进行说明。 The following side information to the i-th check node is connected will be described.

[0047] 计算与第i个校验节点相连的所有边的置信度信息的最小值和次小值。 [0047] The minimum times and smaller confidence value is calculated for all side information connected to the i-th check node. 将边信息的置信度的最小值和次小值分别记为 The confidence level of the side information and the second smallest of the minimum values ​​are referred to as

with

并将与第i个校验节点相连的所有边中置信度最小的边的序号记为ji。 All sides of the confidence in the minimum number of edges connected to the note and the i-th check node is ji.

[0048] 然后,对于与校验节点i,1彡i彡m,相连的所有边 [0048] Then, with respect to the check nodes i, 1 i San San m, all the edges connected

je队= {j : 1 < j < n,hu尹0},计算每条边的边信息。 je team = {j: 1 <j <n, hu Yin 0}, calculating the side information for each edge. 本步骤的边信息同样分为两个部分,第一部分为每条边对应的外部校验和 Side information in this step is also divided into two parts, the first part of each edge corresponding to the external test and

[0049] [0049]

(7) (7)

[0050] 第二部分为该边外部校验和的置信度,由下式求得: [0050] for the second portion of the outer edge and confidence check, by the following equation:

[0051] [0051]

(8) (8)

[0052] 这里,角标 [0052] Here, the subscript

表示连接第i个校验节点和第j个变量节点的边。 It indicates the i-th edge connected to the check nodes and the variable nodes j.

[0053] 步骤五、变量节点计算。 [0053] Step five, the variable node calculation. 变量节点的计算分为两个部分。 Variable node calculation is divided into two portions.

[0054] 第一部分,计算校验节点输入变量节点的信息,采用基于汉明距离的置信度加权方式来获得。 [0054] The first portion, the check node calculation information input variable node, based confidence weighting scheme to obtain a Hamming distance. 根据校验节点和变量节点连接边的外部校验和的二进制表示与码字符号硬判决的二进制表示之间的汉明距离的大小设定加权因子,使用加权因子对连接边的外部校验和的置信度进行加权。 The side of the check nodes and variable nodes connected to external test set and binary weighting factor represents the magnitude of the Hamming distance between the binary representation of the hard decision codeword symbols, the weighting factors of edges connecting external test and confidence weights. 对于所有1 < j < η和ie Mp计算步骤四中每条边信息的外部校验和 For all external test 1 <j <η ie Mp calculation step four and every edge information and

与本次迭代步骤二中的码字信息硬判决结果z(k)的二进制表示的汉明距离,将该距离用^ Hamming distance and the binary codeword information in the present iteration step two hard-decision result z (k) is represented by the distance ^

表示;假设符号 It represents; symbol hypothesis

的二进制表示为 The binary representation

则第i个校验节点输入给第j个变量节点的第t个二进制位的信息为 The t-th bit of the i-th check node is input to the j-th variable node information

[0055] [0055]

(9) (9)

[0056] 其中,1 < t < r,该信息将在下次迭代的步骤三中用来计算码字外信息。 [0056] wherein, in the 1 <t <r, the step iteration in the information used to calculate the next three outer codeword information.

为加权因子,具体取值在步骤一中已设定,根据所获得的距离 A weighting factor, the specific value is set in step one, according to the obtained distance

来确定加权因子的取值。 Determining a value of the weighting factor. 另外,本步骤将所求的 Further, the present step will ask

的数值范围限定在 Numerical ranges defined in

范围内,如果所求结果大于 The range, if the result is greater than the required

则将 Will

的值取为 The value is set at

如果所求结果小于 If the result is less than the required

则将死ΐ.,的值取为 Will die ΐ., The value is taken as

[0057] 变量节点计算的第二部分为码字信息的更新,按照下式进行: [0057] The variable node calculation section into a second codeword information is updated according to the following formula:

[0058] [0058]

〇⑴ 〇⑴

[0059] 其中,1彡j彡η,1彡t彡r,更新的码字信息将用在下一次迭代的步骤二中进行硬判决和奇偶校验。 [0059] wherein [eta] 1 San San j, 1 San San t r, the codeword information updated with the next iteration of step two hard decision and parity. 由式(10)可知,第k+Ι次迭代时的第j个码字符号的二进制位信息, 由两部分求和得到,第一部分为第一次迭代时的第j个码字符号的二进制位信息,第二部分为对第k次迭代时各校验节点输入给第j个变量节点的二进制位信息求和得到。 By the formula (10) can be seen, the information bits of the k + Ι iterations when the j-th code word symbols, obtained by summing the two portions, the first portion of the first binary codeword symbols j-th iteration time position information, the second part is a check node for each iteration k the first input to the j-th variable node information bit summation.

[0060] 另外,本步骤所求的 [0060] Further, the present step is required

的数值范围限定在 Numerical ranges defined in

范围内,如果所求结果大于: The range, if the result is greater than the required:

,则将 , Then

的值取为 The value is set at

,如果所求结果小于 If the result is less than required

,则将 , Then

的值取为 The value is set at

[0061] 步骤六、进入下一次迭代,即,令k = k+Ι,重新执行步骤二。 [0061] Step 6 to enter the next iteration, i.e., make k = k + Ι, repeat Step II.

[0062] 当译码在步骤二中译码成功或失败时,译码结束。 [0062] In the decoding step, when decoding success or failure of the two decoding end.

Claims (8)

1. 一种低复杂度的多进制LDPC码译码方法,首先进行译码初始化,然后迭代进行如下步骤:硬判决、边信息计算、校验节点计算以及变量节点计算,直到译码成功或失败;其特征在于,所述的译码方法,码字符号采用二进制形式表示,并包括如下方面: 1) 译码初始化时,设定加权因子,使用比特置信度将码字信息进行初始化; 2) 校验节点和变量节点连接边的外部校验和的置信度转换成比特置信度向量,参与变量节点计算; 3) 每次迭代更新的码字信息和校验节点的外信息都为比特置信度向量。 A low complexity method for decoding binary LDPC codes, initialization is first decoded, then iterate the following steps: hard decision, calculating the side information, the check node calculation and variable node calculation, until a successful decode or failed; wherein, said decoding method, codeword symbols expressed in binary format, and include the following: 1) the decoding initialization, setting the weighting factors, using the confidence bit codeword information is initialized; 2 ) and the external test confidence check nodes and edges connecting the variable nodes converted into bit confidence vector, participating variable node calculation; 3) each outer iteration update information and the codeword information bits are check nodes confidence degree vector.
2. 根据权利要求1所述的多进制LDPC码译码方法,其特征在于,所述的译码方法,在变量节点计算时,校验节点输入变量节点的信息,采用基于汉明距离的置信度加权方式来获得,具体是:根据校验节点和变量节点连接边的外部校验和的二进制表示与码字符号硬判决的二进制表示之间的汉明距离的大小设定加权因子,使用加权因子对连接边的外部校验和的置信度进行加权。 The binary LDPC code decoding method according to claim 1, wherein said decoding method, when the variable node calculation, variable node check node input information, the Hamming distance based confidence weighting scheme to obtain, in particular: indicates the size of the Hamming distance between the binary representation of the codeword symbols of the hard decision weighting factor is set according to the check nodes and edges connecting the variable nodes and check external binary, using confidence weighting factor edges connecting external weighted checksum.
3. 根据权利要求1所述的多进制LDPC码译码方法,其特征在于,所述的译码方法,在边信息计算时,边信息包括码字外信息的多进制硬判决符号及该符号的置信度;其中,符号的置信度为码字的二进制位外信息中所有绝对值中的最小值。 The binary LDPC code decoding method according to claim 1, wherein said decoding method, when calculating the side information, the side information comprising hard decision symbol-ary codeword information and outer the confidence symbol; wherein the symbol is the confidence to the information bits in the code word minimum of all absolute values.
4. 根据权利要求1〜3任一所述的多进制LDPC码译码方法,其特征在于,所述的译码方法中,所有码字信息、码字外信息、外部校验和的置信度、以及校验节点输入给变量节点的信息均为整数。 4. binary LDPC code decoding method according to any one of claims 1 ~ 3, wherein, in the decoding method, all of the outer codeword information, codeword information, and the external test confidence degrees, and input check node to variable node information are integers.
5. 根据权利要求1所述的多进制LDPC码译码方法,其特征在于,所述的使用比特置信度将码字信息进行初始化,具体是:将系统接收到的码字的二进制信息y以量化间隔Λ和量化比特数ω均匀量化为整数赋值给第1次迭代的码字信息。 The binary LDPC code decoding method according to claim 1, wherein said step of using the confidence bit codeword information is initialized, in particular: the system of the received codeword y binary information to quantify the number of bits and the quantization interval Λ ω uniformly quantized codeword information assigned to the first iteration is an integer.
6. 根据权利要求1或2所述的多进制LDPC码译码方法,其特征在于,所述的校验节点的外信息,具体是:第1次迭代时,校验节点的第j个码字符号的二进制位外信息等于第一次迭代的第j个码字符号的二进制位信息;第k(k > 2)次迭代时,校验节点的第j个码字符号的二进制位外信息由第k次迭代的码字信息减去第k-Ι次迭代时校验节点输入第j个变量节点的信息得到。 The binary LDPC code decoding method according to claim 1, characterized in that the outer check nodes according to the information, in particular: When the first iteration, j-th check node bit information bits to the information symbols in a codeword is equal to the first iteration of the j-th code word symbols; outer k-th (k> 2) when the iterations, the j-th bit codeword symbols check nodes check node information of the j-th input variable nodes when the information obtained by the k-th codeword information iteration k-Ι subtracting iterations.
7. 根据权利要求1所述的多进制LDPC码译码方法,其特征在于,所述的校验节点计算时,连接校验节点和变量节点的边包括两部分信息:第一部分为该边的外部校验和,第二部分为该边外部校验和的置信度;其中,边外部校验和的置信度的确定方法是:首先,获得与校验节点相连的所有边中置信度的最小值MIR和次小值MIN 2,与校验节点相连的置信度最小的边的外部校验和的置信度为MIN2,剩余与校验节点相连的边的外部校验和的置信度均为MIR。 The binary LDPC code decoding method according to claim 1, wherein, when said check node calculation, variable node connected to check nodes and edges including two pieces of information: a first side portion for external checksum, second portion for the outer side confidence checksum; wherein the method for determining the confidence level and the outside edge of the check are: first, get all the check nodes connected to edges in confidence minimum confidence outer edge checksum MIR minimum confidence value and the second smallest MIN 2, the check nodes connected to MIN2, the confidence side connected to external test and check nodes are remaining MIR.
8. 根据权利要求6所述的多进制LDPC码译码方法,其特征在于,所述的码字信息,更新方法是:第k+Ι次迭代时的第j个码字符号的二进制位信息,由两部分求和得到,第一部分为第一次迭代时的第j个码字符号的二进制位信息,第二部分为对第k次迭代时各校验节点输入给第j个变量节点的二进制位信息求和得到。 8. A binary LDPC code decoding method according to claim 6, wherein said codeword information, updating method is: j-th bit codeword symbols during a first iteration k + Ι information obtained from the summation of two parts, the first part of the j-th information bit codeword symbol at the first iteration, the second portion is divided into a j-th variable node to a check node for each iteration k the first inputs to the the summation of bits of information.
CN201410295270.XA 2014-06-26 2014-06-26 Low complexity code decoding method ary ldpc CN104052501B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN201410295270.XA CN104052501B (en) 2014-06-26 2014-06-26 Low complexity code decoding method ary ldpc

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN201410295270.XA CN104052501B (en) 2014-06-26 2014-06-26 Low complexity code decoding method ary ldpc

Publications (2)

Publication Number Publication Date
CN104052501A true CN104052501A (en) 2014-09-17
CN104052501B CN104052501B (en) 2017-03-29

Family

ID=51504909

Family Applications (1)

Application Number Title Priority Date Filing Date
CN201410295270.XA CN104052501B (en) 2014-06-26 2014-06-26 Low complexity code decoding method ary ldpc

Country Status (1)

Country Link
CN (1) CN104052501B (en)

Cited By (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN105846833A (en) * 2016-04-13 2016-08-10 国家电网公司 Short ring distribution based weighted message passing decoding method
CN106936445A (en) * 2017-03-14 2017-07-07 西安电子科技大学 Low-complexity maximum likelihood approximate q-ary LDPC (Low-Density Parity-Check) code decoding method

Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20050229087A1 (en) * 2004-04-13 2005-10-13 Sunghwan Kim Decoding apparatus for low-density parity-check codes using sequential decoding, and method thereof
CN101557232A (en) * 2008-04-08 2009-10-14 威望科技(苏州)有限公司 Decoding method of low density parity check codes
CN101707488A (en) * 2009-02-03 2010-05-12 天津博微科技有限公司 Method for controlling stopping of iteration on basis of multi-system LDPC iterative decoding
CN103095311A (en) * 2013-01-17 2013-05-08 上海交通大学 Collaboration decoding method of multi-system low density parity check (LDPC) code

Patent Citations (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20050229087A1 (en) * 2004-04-13 2005-10-13 Sunghwan Kim Decoding apparatus for low-density parity-check codes using sequential decoding, and method thereof
US7590914B2 (en) * 2004-04-13 2009-09-15 Electronics And Telecommunications Research Institute Decoding apparatus for low-density parity-check codes using sequential decoding, and method thereof
CN101557232A (en) * 2008-04-08 2009-10-14 威望科技(苏州)有限公司 Decoding method of low density parity check codes
CN101707488A (en) * 2009-02-03 2010-05-12 天津博微科技有限公司 Method for controlling stopping of iteration on basis of multi-system LDPC iterative decoding
CN103095311A (en) * 2013-01-17 2013-05-08 上海交通大学 Collaboration decoding method of multi-system low density parity check (LDPC) code

Non-Patent Citations (1)

* Cited by examiner, † Cited by third party
Title
王学鹏: "多元LDPC编码调制系统的低复杂度译码算法研究", 《中国优秀硕士学位论文全文数据库信息科技辑》 *

Cited By (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN105846833A (en) * 2016-04-13 2016-08-10 国家电网公司 Short ring distribution based weighted message passing decoding method
CN105846833B (en) * 2016-04-13 2019-06-25 国家电网公司 A kind of weighted messages transmitting interpretation method based on becate distribution
CN106936445A (en) * 2017-03-14 2017-07-07 西安电子科技大学 Low-complexity maximum likelihood approximate q-ary LDPC (Low-Density Parity-Check) code decoding method
CN106936445B (en) * 2017-03-14 2019-06-21 西安电子科技大学 A kind of multielement LDPC code coding method of low complex degree near-maximum-likelihood

Also Published As

Publication number Publication date
CN104052501B (en) 2017-03-29

Similar Documents

Publication Publication Date Title
Zhao et al. On implementation of min-sum algorithm and its modifications for decoding low-density parity-check (LDPC) codes
US8291279B2 (en) Memory-efficient LDPC decoder and method
US8683303B2 (en) Operational parameter adaptable LDPC (low density parity check) decoder
US8806307B2 (en) Interruption criteria for block decoding
US20070022354A1 (en) Method for encoding low-density parity check code
CN100440736C (en) Method and system for routing in low density parity check (ldpc) decoders
CA2465332C (en) Soft input decoding for linear codes
US20050229087A1 (en) Decoding apparatus for low-density parity-check codes using sequential decoding, and method thereof
JP3893383B2 (en) Ldpc code inspection matrix generation method and the check matrix generating device
US8301984B1 (en) QC-LDPC decoder with list-syndrome decoding
CN101159436B (en) Decoding equipment and method
CN102017427B (en) Method and apparatus for channel encoding and decoding in a communication system using low-density parity-check codes
EP1717959A1 (en) Method and device for controlling the decoding of a LDPC encoded codeword, in particular for DVB-S2 LDPC encoded codewords
US7401283B2 (en) Amplifying magnitude metric of received signals during iterative decoding of LDPC (Low Density Parity Check) code and LDPC coded modulation
US8219878B1 (en) Post-processing decoder of LDPC codes for improved error floors
JP5302972B2 (en) Channel coding method and decoding method and their device in a system using a Low-Density Parity Check Codes
US20090319860A1 (en) Overcoming ldpc trapping sets by decoder reset
US8898537B2 (en) Method and system for decoding
CN101103533B (en) Encoding method
WO2005077108A2 (en) Improved performance of coding schemes
JP5508549B2 (en) Reduction of error floor in fec code is repeatedly decoded
FR2909499A1 (en) Low density parity check code word decoding method for communication apparatus, involves determining messages, from variable node to control node, relative to set of symbols such that minimal value is null
EP2091156A2 (en) Apparatus and method for channel encoding and decoding in a communication system using low-density parity-check codes
Vangala et al. A comparative study of polar code constructions for the AWGN channel
JP5138221B2 (en) How to min-sum decoding an error correcting code

Legal Events

Date Code Title Description
C06 Publication
C10 Entry into substantive examination
GR01