CN104683072A - A Blind Identification Method of Parameters of Component Encoders with Punctured Turbo Codes - Google Patents

A Blind Identification Method of Parameters of Component Encoders with Punctured Turbo Codes Download PDF

Info

Publication number
CN104683072A
CN104683072A CN201510137808.9A CN201510137808A CN104683072A CN 104683072 A CN104683072 A CN 104683072A CN 201510137808 A CN201510137808 A CN 201510137808A CN 104683072 A CN104683072 A CN 104683072A
Authority
CN
China
Prior art keywords
matrix
code
sequence
check
punctured convolutional
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.)
Pending
Application number
CN201510137808.9A
Other languages
Chinese (zh)
Inventor
马丕明
李孟琪
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.)
Shandong University
Original Assignee
Shandong University
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 Shandong University filed Critical Shandong University
Priority to CN201510137808.9A priority Critical patent/CN104683072A/en
Publication of CN104683072A publication Critical patent/CN104683072A/en
Pending legal-status Critical Current

Links

Landscapes

  • Error Detection And Correction (AREA)

Abstract

The invention relates to a parameter blind identification method of a puncturing turbo code component coder, belonging to the technical field of the code identification in a communication system. The method comprises the steps of multiplexing a puncturing turbo code information bit and a verification bit to form a puncturing convolutional code, carrying out statistics and analysis for the code length and a code word starting point of the puncturing convolutional code by virtue of a linear matrix analysis method, identifying a verification matrix of the component coder by utilizing partial Walsh-Hadamard conversion, obtaining a puncturing mode and generating a matrix further by virtue of a Gaussian elimination method, carrying out the blind identification on the parameters of the puncturing turbo code component coder, and verifying the accuracy of the identified parameters. By adopting the method, the operation complexity of identifying the component coder parameter can be alleviated, the identification efficiency and the identification reliability can be improved, and the method is applicable to the fields of the intelligent communication, information processing and the like.

Description

一种删余turbo码分量编码器的参数盲识别方法A Blind Identification Method of Parameters of Component Encoders with Punctured Turbo Codes

技术领域technical field

本发明涉及数字通信系统中一种删余turbo码分量编码器的参数盲识别方法，属于通信系统中的码字识别技术领域。The invention relates to a parameter blind identification method of a punctured turbo code component encoder in a digital communication system, and belongs to the technical field of code word identification in a communication system.

背景技术Background technique

删余turbo码在现代通信中应用非常广泛，随着数字通信技术的发展，越来越多的领域都会产生对删余Turbo码盲识别技术的需求，删余turbo码盲识别技术也成为当今通信研究的前沿领域。The punctured turbo code is widely used in modern communication. With the development of digital communication technology, more and more fields will have the demand for the blind recognition technology of the punctured turbo code. The blind recognition technology of the punctured turbo code has also become the frontiers of research.

经典并行级联turbo码编码器主要由两个递归系统卷积码(RSC)分量编码器并行级联而成，分量编码器之间由交织器相连，一般情况下，各RSC编码器的结构相同，使用两个RSC分量编码器的turbo码的码率为1/3，每三个比特中有一个信息码元，两个校验码元。为提高码率，可对校验元进行周期性删余，而后与信息码元经过复用得到最终编码序列。The classic parallel concatenated turbo code encoder is mainly composed of two recursive systematic convolutional code (RSC) component encoders connected in parallel, and the component encoders are connected by an interleaver. In general, the structure of each RSC encoder is the same , the code rate of the turbo code using two RSC component encoders is 1/3, and there is one information symbol and two check symbols in every three bits. In order to improve the code rate, periodic puncturing can be performed on the check element, and then multiplexed with the information symbol to obtain the final coded sequence.

针对基于1/2源卷积码的(n-1)/n型删余卷积码的盲识别，传统的Walsh-Hadamard变换对内存空间的要求比较高，王磊、胡以华、王勇等在其2012年发表在《计算机工程与应用》16期“基于PWHT的删除卷积码识别方法”一文中，针对Walsh-Hadamard变换在删除卷积码校验矩阵识别中存在运算量和数据量过大的问题,对校验矩阵方程组进行了变形,提出部分Walsh-Hadamard变换(PWHT),但并没有给出应用于删余turbo码分量编码器参数盲识别的方法。为克服现有技术存在的缺陷和不足，本发明提出基于部分Walsh-Hadamard变换(PWHT)的删余turbo码分量编码器的参数盲识别方法，降低了分量编码器参数识别的运算复杂度，提高了识别效率。For the blind recognition of (n-1)/n-type punctured convolutional codes based on 1/2 source convolutional codes, the traditional Walsh-Hadamard transform requires relatively high memory space. Wang Lei, Hu Yihua, Wang Yong et al. Published in the 16th issue of "Computer Engineering and Application" "PWHT-based Deletion Convolutional Code Recognition Method", aiming at the Walsh-Hadamard transform in the detection matrix of the deletion convolutional code, there are too many calculations and data. The problem is that the parity-check matrix equations are deformed, and the partial Walsh-Hadamard transform (PWHT) is proposed, but no method for blind identification of parameters of component encoders for punctured turbo codes is given. In order to overcome the defects and deficiencies in the prior art, the present invention proposes a parameter blind identification method of a punctured turbo code component encoder based on a partial Walsh-Hadamard transform (PWHT), which reduces the computational complexity of component encoder parameter identification and improves recognition efficiency.

本发明采用计算机软件模拟的方式实现删余turbo码分量编码器的参数识别，实施过程中首先用软件模拟上述经典turbo码编码过程，将码字保存在文件中，识别时从文件中读取数据进行盲识别。The present invention adopts the mode of computer software simulation to realize the parameter identification of the punctured turbo code component encoder, first uses software to simulate the above-mentioned classical turbo code encoding process in the implementation process, saves the code word in the file, and reads the data from the file when identifying Perform blind recognition.

发明内容Contents of the invention

为了克服现有技术存在的缺陷和不足，本发明提出了运算量小、识别效率高的一种删余turbo码分量编码器的参数盲识别方法。In order to overcome the defects and deficiencies in the prior art, the present invention proposes a parameter blind identification method of a punctured turbo code component encoder with small calculation amount and high identification efficiency.

本发明技术方案如下：Technical scheme of the present invention is as follows:

一种删余turbo码分量编码器的参数盲识别方法，将删余turbo码信息位与第一位校验位复用构造删余卷积码，通过线性矩阵分析法统计分析删余卷积码的码长，码字起点，再利用部分Walsh—Hadamard变换识别分量编码器的校验矩阵，进一步由高斯消元法得到删余模式和生成矩阵，从而实现删余turbo码分量编码器的参数盲识别，并通过维特比译码得到置信度，以验证识别的准确性，该方法具体步骤如下：A parameter blind recognition method of a punctured turbo code component encoder, which multiplexes the punctured turbo code information bit and the first check bit to construct a punctured convolutional code, and statistically analyzes the punctured convolutional code through a linear matrix analysis method The code length and the starting point of the code word, and then use part of the Walsh-Hadamard transformation to identify the parity check matrix of the component encoder, and further obtain the puncturing mode and generator matrix by the Gaussian elimination method, so as to realize the parameter blindness of the component encoder of the punctured turbo code Identify, and obtain the degree of confidence through Viterbi decoding, to verify the accuracy of identification, the specific steps of the method are as follows:

1)从删余turbo码起点开始，选取其信息序列和分量编码器RSC1的校验位序列复用，构造删余卷积码序列，以1/3码率turbo码删余得到1/2码率的删余turbo码为例，若将原turbo码序列表示为x1a1b1x2a2b2x3a3b3x4a4b4……则删余后turbo码序列表示为x1a1x2b2x3a3x4b4……，构造删余卷积码为x1a1x2x3a3x4……，其中x1、x2、x3……表示不同时刻输出的信息位序列；a1、a2、a3……表示不同时刻输出的RSC1校验位序列；b1、b2、b3……表示不同时刻输出的RSC2校验位序列；1) Starting from the starting point of the punctured turbo code, its information sequence is selected and multiplexed with the check bit sequence of the component encoder RSC1 to construct a punctured convolutional code sequence, and a 1/2 code is obtained by puncturing the turbo code with a code rate of 1/3 Taking the punctured turbo code with high rate as an example, if the original turbo code sequence is expressed as x 1 a 1 b 1 x 2 a 2 b 2 x 3 a 3 b 3 x 4 a 4 b 4 ... then the punctured turbo code sequence Expressed as x 1 a 1 x 2 b 2 x 3 a 3 x 4 b 4 ..., the constructed punctured convolutional code is x 1 a 1 x 2 x 3 a 3 x 4 ..., where x 1 , x 2 , x 3 ......represents the information bit sequence output at different times; a 1 , a 2 , a 3 ......represents the RSC1 check bit sequence output at different times; b 1 , b 2 , b 3 ......represents the RSC2 output at different times check digit sequence;

2)将删余卷积码序列构成p×q矩阵，要求q＞N,p＞q，其中N表示编码约束度，取定列值q的范围，变化q得到不同矩阵，分别计算这些矩阵的秩，只留取矩阵秩不等于列数的矩阵，然后对这些矩阵进行初等变换单位化，记下单位化后左上角单位阵维数相等的矩阵列值，对留存列值取最大公约数，从而得到删余卷积码码长n；2) Construct the punctured convolutional code sequence into a p×q matrix, requiring q>N, p>q, where N represents the coding constraint degree, taking the range of the fixed column value q, changing q to obtain different matrices, and calculating the values of these matrices respectively Rank, only take the matrix whose rank is not equal to the number of columns, and then perform elementary transformation unitization on these matrices, record the column values of the matrix with the same dimension as the unit matrix in the upper left corner after normalization, and take the greatest common divisor for the retained column values, Thus, the code length n of the punctured convolutional code is obtained;

3)将删余卷积码序列重新构造矩阵，列数为步骤2)中留存的某列值N’，其中N’为码长n的倍数，取留存列值中的较小值，取矩阵行数大于列数，将码序列移位得到n-1个不同矩阵，连同码序列不移位矩阵共得到n个不同矩阵，n为码长，对这些矩阵进行初等变换单位化，分别记下单位化后矩阵的左上角单位阵维数，维数最小时的移位即删余卷积码起始点；3) Reconstruct the matrix with the punctured convolutional code sequence, the number of columns is a certain column value N' retained in step 2), where N' is a multiple of the code length n, take the smaller value of the retained column values, and take the matrix The number of rows is greater than the number of columns. The code sequence is shifted to obtain n-1 different matrices. Together with the code sequence without shifting the matrix, a total of n different matrices are obtained. The dimension of the unit matrix in the upper left corner of the matrix after unitization, and the shift when the dimension is the smallest is the starting point of the punctured convolutional code;

4)识别校验多项式矩阵，具体识别步骤如下：4) Identify the check polynomial matrix, the specific identification steps are as follows:

a)由构造的删余卷积码序列建立系数矩阵R(D)，将校验多项式矩阵表示为H(D)＝[h0(D),h1(D),...hn-1(D)]T，其中，hi(D)＝h0i+h1iD+h2iD2...hdiDd，i∈[0,n-1]，d＝max{deg hi(D)}，其中符号deg表示是求多项式的次数的函数，将R(D)×H(D)T＝0写成方程组的形式如下：a) The coefficient matrix R(D) is established from the constructed punctured convolutional code sequence, and the check polynomial matrix is expressed as H(D)=[h 0 (D),h 1 (D),...h n- 1 (D)] T , where h i (D)=h 0i +h 1i D+h 2i D 2 ...h di D d , i∈[0,n-1], d=max{deg h i (D)}, wherein the symbol deg represents the function of seeking the degree of polynomial, and the form of writing R(D)×H(D) T =0 as a system of equations is as follows:

rr dd 11 .. .. .. rr 0101 .. .. .. rr dndn .. .. .. rr 00 nno rr (( dd ++ 11 )) 11 .. .. .. rr 1111 .. .. .. rr (( dd ++ 11 )) nno .. .. .. rr 11 nno .. .. .. .. .. .. .. .. .. .. .. .. rr (( dd ++ NN )) 11 .. .. .. rr NN 11 .. .. .. rr (( dd ++ NN )) nno .. .. .. rr Nnn hh 0000 .. .. .. hh dd 00 .. .. .. hh 00 (( nno -- 11 )) .. .. .. hh dd (( nno -- 11 )) == 00 -- -- -- (( 11 ))

由于hn-1(D)中含常数项h0(n-1)＝1，将式(1)中的常数项部分移至方程右边，得下式：Since h n-1 (D) contains a constant term h 0(n-1) = 1, the constant term part in the formula (1) is moved to the right side of the equation, and the following formula is obtained:

rr dd 11 .. .. .. rr 0101 .. .. .. rr (( dd -- 11 )) nno .. .. .. rr 00 nno rr (( dd ++ 11 )) 11 .. .. .. rr 1111 .. .. .. rr dndn .. .. .. rr 11 nno .. .. .. .. .. .. .. .. .. .. .. .. rr (( dd ++ NN )) 11 .. .. .. rr NN 11 .. .. .. rr (( dd ++ NN -- 11 )) nno .. .. .. rr Nnn hh 0000 .. .. .. hh dd 00 .. .. .. hh 11 (( nno -- 11 )) .. .. .. hh dd (( nno -- 11 )) == -- rr dndn -- rr (( dd ++ 11 )) nno -- rr (( dd ++ NN )) nno -- -- -- (( 22 ))

b)将式(2)中的系数矩阵R'(D)分为两部分，矩阵长度均为N+1，宽度分别为R1和R2，其中R1与R2的和为(d+1)×n-1，(通常R2取值在16到24之间)，则第一、第二系数矩阵维数分别为(N+1)×R1、(N+1)×R2；b) Divide the coefficient matrix R'(D) in formula (2) into two parts, the matrix length is N+1, and the widths are R1 and R2 respectively, where the sum of R1 and R2 is (d+1)×n -1, (usually the value of R2 is between 16 and 24), then the dimensions of the first and second coefficient matrices are (N+1)×R1, (N+1)×R2 respectively;

c)设定一个长度为R1的二进制向量iTemp，大小从0至2R1进行循环，将每个固定的iTemp与第一系数矩阵的行向量模二加后取反，每行得到一个二进制数0或1，最终得到一个N+1维的二进制向量，表示为P，将P与式(2)等号右边的向量对应行的值相加；c) Set a binary vector iTemp with a length of R1, the size of which is cyclic from 0 to 2 R1 , add each fixed iTemp to the row vector of the first coefficient matrix modulo 2 and invert, and obtain a binary number 0 for each row or 1, and finally get an N+1-dimensional binary vector, expressed as P, and add P to the value of the corresponding row of the vector on the right side of the equal sign in formula (2);

d)再对第二系数矩阵进行状态统计，第二系数矩阵的行向量维数为1*R2，R2个任意‘0’和‘1’组合作为状态，共2R2-1个状态，每个行向量都是其中的一个状态，式(2)等号右边向量对应行的值作为该状态的输出，相同状态的输出值进行累加，不存在的状态输出值为0，得到一个(2R2-1)×1维状态向量；d) Perform state statistics on the second coefficient matrix, the row vector dimension of the second coefficient matrix is 1*R2, R2 any combination of '0' and '1' as the state, a total of 2 R2 -1 states, each The row vector is one of the states, the value of the corresponding row of the vector on the right side of the equation (2) is used as the output of the state, the output value of the same state is accumulated, and the output value of the non-existing state is 0, and a (2 R2 - 1) × 1-dimensional state vector;

e)对上述结果进行Walsh-Hadamard变换，找到大于置信度的解再转换为二进制向量，表示为Q，然后将此时的iTemp向量与二进制向量Q组合便得到我们要求的校验多项式H(D)；e) Perform Walsh-Hadamard transformation on the above results, find a solution greater than the confidence level and convert it into a binary vector, denoted as Q, and then combine the iTemp vector at this time with the binary vector Q to obtain the check polynomial H(D );

5)设定源卷积码的生成多项式矩阵阶数，遍历删余模式，构建线性方程组Gp(D)H(D)T＝0，其中GP(D)为生成矩阵，H(D)T为校验矩阵H(D)的转置，通过高斯消元法求解该方程组的唯一非零解，从而确定源卷积码的生成多项式和删余模式；5) Set the order of the generator polynomial matrix of the source convolutional code, traverse the puncturing mode, and construct a linear equation system Gp(D)H(D) T = 0, wherein G P (D) is the generator matrix, H(D) T is the transposition of the check matrix H(D), and the unique non-zero solution of the equation system is solved by the Gaussian elimination method, thereby determining the generator polynomial and the puncturing mode of the source convolutional code;

6)利用识别得到的分量编码器参数，将删余卷积码通过维特比译码得到置信度，验证识别参数的准确性。6) Using the identified component encoder parameters, the punctured convolutional code is decoded by Viterbi to obtain confidence, and the accuracy of the identified parameters is verified.

本发明较好的将删余turbo码分量编码器参数的识别转换为删余卷积码的识别，在线性矩阵分析法识别码长、起点的基础上，创造性的将改进的部分WH算法用于分量编码器校验矩阵的识别，并进一步识别得到删余模式和生成矩阵，最终实现删余turbo码分量编码器的参数盲识别，并通过维特比译码得到置信度，验证识别的准确性。本发明降低了分量编码器参数识别的运算复杂度，提高了识别效率和可靠性。The present invention preferably converts the recognition of the punctured turbo code component encoder parameters into the recognition of the punctured convolutional code, and creatively uses the improved partial WH algorithm for Identify the check matrix of the component encoder, and further identify the puncturing mode and generation matrix, and finally realize the parameter blind identification of the component encoder of the punctured turbo code, and obtain the confidence through Viterbi decoding to verify the accuracy of the identification. The invention reduces the computational complexity of component coder parameter identification, and improves identification efficiency and reliability.

具体实施方式detailed description

下面结合实施例对本发明作进一步说明，但不限于此。The present invention will be further described below in conjunction with the examples, but not limited thereto.

实施例：Example:

本发明实施例如下：一种删余turbo码分量编码器的参数盲识别方法，将删余turbo码信息位与第一位校验位复用构造删余卷积码，通过线性矩阵分析法统计分析删余卷积码的码长，码字起点，再利用部分Walsh—Hadamard变换识别分量编码器的校验矩阵，进一步由高斯消元法得到删余模式和生成矩阵，从而实现删余turbo码分量编码器的参数盲识别，并通过维特比译码得到置信度，以验证识别的准确性，该方法具体步骤如下：Embodiments of the present invention are as follows: a parameter blind identification method of a punctured turbo code component encoder, which multiplexes the punctured turbo code information bit and the first check bit to construct a punctured convolutional code, and makes statistics through the linear matrix analysis method Analyze the code length of the punctured convolutional code and the starting point of the code word, and then use part of the Walsh-Hadamard transformation to identify the parity check matrix of the component encoder, and further obtain the punctured mode and generation matrix by the Gaussian elimination method, thereby realizing the punctured turbo code The parameters of the component encoder are blindly identified, and the confidence is obtained through Viterbi decoding to verify the accuracy of the identification. The specific steps of the method are as follows:

1)从删余turbo码起点开始，选取其信息序列和分量编码器RSC1的校验位序列复用，构造删余卷积码序列，以1/3码率turbo码删余得到1/2码率的删余turbo码为例，若将原turbo码序列表示为x1a1b1x2a2b2x3a3b3x4a4b4……则删余后turbo码序列表示为x1a1x2b2x3a3x4b4……，构造删余卷积码为x1a1x2x3a3x4……，其中x1、x2、x3……表示不同时刻输出的信息位序列；a1、a2、a3……表示不同时刻输出的RSC1校验位序列；b1、b2、b3……表示不同时刻输出的RSC2校验位序列；1) Starting from the starting point of the punctured turbo code, its information sequence is selected and multiplexed with the check bit sequence of the component encoder RSC1 to construct a punctured convolutional code sequence, and a 1/2 code is obtained by puncturing the turbo code with a code rate of 1/3 Taking the punctured turbo code with high rate as an example, if the original turbo code sequence is expressed as x 1 a 1 b 1 x 2 a 2 b 2 x 3 a 3 b 3 x 4 a 4 b 4 ... then the punctured turbo code sequence Expressed as x 1 a 1 x 2 b 2 x 3 a 3 x 4 b 4 ..., the constructed punctured convolutional code is x 1 a 1 x 2 x 3 a 3 x 4 ..., where x 1 , x 2 , x 3 ......represents the information bit sequence output at different times; a 1 , a 2 , a 3 ......represents the RSC1 check bit sequence output at different times; b 1 , b 2 , b 3 ......represents the RSC2 output at different times check digit sequence;

2)将删余卷积码序列构成p×q矩阵，要求q＞N,p＞q，其中N表示编码约束度，取定列值q的范围，变化q得到不同矩阵，分别计算这些矩阵的秩，只留取矩阵秩不等于列数的矩阵，然后对这些矩阵进行初等变换单位化，记下单位化后左上角单位阵维数相等的矩阵列值，对留存列值取最大公约数，从而得到删余卷积码码长n；2) Construct the punctured convolutional code sequence into a p×q matrix, requiring q>N, p>q, where N represents the coding constraint degree, taking the range of the fixed column value q, changing q to obtain different matrices, and calculating the values of these matrices respectively Rank, only take the matrix whose rank is not equal to the number of columns, and then perform elementary transformation unitization on these matrices, record the column values of the matrix with the same dimension as the unit matrix in the upper left corner after normalization, and take the greatest common divisor for the retained column values, Thus, the code length n of the punctured convolutional code is obtained;

3)将删余卷积码序列重新构造矩阵，列数为步骤2)中留存的某列值N’，其中N’为码长n的倍数，取留存列值中的较小值，取矩阵行数大于列数，将码序列移位得到n-1个不同矩阵，连同码序列不移位矩阵共得到n个不同矩阵，n为码长，对这些矩阵进行初等变换单位化，分别记下单位化后矩阵的左上角单位阵维数，维数最小时的移位即删余卷积码起始点；3) Reconstruct the matrix with the punctured convolutional code sequence, the number of columns is a certain column value N' retained in step 2), where N' is a multiple of the code length n, take the smaller value of the retained column values, and take the matrix The number of rows is greater than the number of columns. The code sequence is shifted to obtain n-1 different matrices. Together with the code sequence without shifting the matrix, a total of n different matrices are obtained. The dimension of the unit matrix in the upper left corner of the matrix after unitization, and the shift when the dimension is the smallest is the starting point of the punctured convolutional code;

4)识别校验多项式矩阵，具体识别步骤如下：4) Identify the check polynomial matrix, the specific identification steps are as follows:

a)由构造的删余卷积码序列建立系数矩阵R(D)，将校验多项式矩阵表示为H(D)＝[h0(D),h1(D),...hn-1(D)]T，其中，hi(D)＝h0i+h1iD+h2iD2...hdiDd，i∈[0,n-1]，d＝max{deg hi(D)}，其中符号deg表示是求多项式的次数的函数，将R(D)×H(D)T＝0写成方程组的形式如下：a) The coefficient matrix R(D) is established from the constructed punctured convolutional code sequence, and the check polynomial matrix is expressed as H(D)=[h 0 (D),h 1 (D),...h n- 1 (D)] T , where h i (D)=h 0i +h 1i D+h 2i D 2 ...h di D d , i∈[0,n-1], d=max{deg h i (D)}, wherein the symbol deg represents the function of seeking the degree of polynomial, and the form of writing R(D)×H(D) T =0 as a system of equations is as follows:

rr dd 11 .. .. .. rr 0101 .. .. .. rr dndn .. .. .. rr 00 nno rr (( dd ++ 11 )) 11 .. .. .. rr 1111 .. .. .. rr (( dd ++ 11 )) nno .. .. .. rr 11 nno .. .. .. .. .. .. .. .. .. .. .. .. rr (( dd ++ NN )) 11 .. .. .. rr NN 11 .. .. .. rr (( dd ++ NN )) nno .. .. .. rr Nnn hh 0000 .. .. .. hh dd 00 .. .. .. hh 00 (( nno -- 11 )) .. .. .. hh dd (( nno -- 11 )) == 00 -- -- -- (( 11 ))

由于hn-1(D)中含常数项h0(n-1)＝1，将式(1)中的常数项部分移至方程右边，得下式：Since h n-1 (D) contains a constant term h 0(n-1) = 1, the constant term part in the formula (1) is moved to the right side of the equation, and the following formula is obtained:

rr dd 11 .. .. .. rr 0101 .. .. .. rr (( dd -- 11 )) nno .. .. .. rr 00 nno rr (( dd ++ 11 )) 11 .. .. .. rr 1111 .. .. .. rr dndn .. .. .. rr 11 nno .. .. .. .. .. .. .. .. .. .. .. .. rr (( dd ++ NN )) 11 .. .. .. rr NN 11 .. .. .. rr (( dd ++ NN -- 11 )) nno .. .. .. rr Nnn hh 0000 .. .. .. hh dd 00 .. .. .. hh 11 (( nno -- 11 )) .. .. .. hh dd (( nno -- 11 )) == -- rr dndn -- rr (( dd ++ 11 )) nno -- rr (( dd ++ NN )) nno -- -- -- (( 22 ))

b)将式(2)中的系数矩阵R'(D)分为两部分，矩阵长度均为N+1，宽度分别为R1和R2，其中R1与R2的和为(d+1)×n-1，(通常R2取值在16到24之间)，则第一、第二系数矩阵维数分别为(N+1)×R1、(N+1)×R2；b) Divide the coefficient matrix R'(D) in formula (2) into two parts, the matrix length is N+1, and the widths are R1 and R2 respectively, where the sum of R1 and R2 is (d+1)×n -1, (usually the value of R2 is between 16 and 24), then the dimensions of the first and second coefficient matrices are (N+1)×R1, (N+1)×R2 respectively;

c)设定一个长度为R1的二进制向量iTemp，大小从0至2R1进行循环，将每个固定的iTemp与第一系数矩阵的行向量模二加后取反，每行得到一个二进制数0或1，最终得到一个N+1维的二进制向量，表示为P，将P与式(2)等号右边的向量对应行的值相加；c) Set a binary vector iTemp with a length of R1, the size of which is cyclic from 0 to 2 R1 , add each fixed iTemp to the row vector of the first coefficient matrix modulo 2 and invert, and obtain a binary number 0 for each row or 1, and finally get an N+1-dimensional binary vector, expressed as P, and add P to the value of the corresponding row of the vector on the right side of the equal sign in formula (2);

d)再对第二系数矩阵进行状态统计，第二系数矩阵的行向量维数为1*R2，R2个任意‘0’和‘1’组合作为状态，共2R2-1个状态，每个行向量都是其中的一个状态，式(2)等号右边向量对应行的值作为该状态的输出，相同状态的输出值进行累加，不存在的状态输出值为0，得到一个(2R2-1)×1维状态向量；d) Perform state statistics on the second coefficient matrix, the row vector dimension of the second coefficient matrix is 1*R2, R2 any combination of '0' and '1' as the state, a total of 2 R2 -1 states, each The row vector is one of the states, the value of the corresponding row of the vector on the right side of the equation (2) is used as the output of the state, the output value of the same state is accumulated, and the output value of the non-existing state is 0, and a (2 R2 - 1) × 1-dimensional state vector;

e)对上述结果进行Walsh-Hadamard变换，找到大于置信度的解再转换为二进制向量，表示为Q，然后将此时的iTemp向量与二进制向量Q组合便得到我们要求的校验多项式H(D)；e) Perform Walsh-Hadamard transformation on the above results, find a solution greater than the confidence level and convert it into a binary vector, denoted as Q, and then combine the iTemp vector at this time with the binary vector Q to obtain the check polynomial H(D );

5)设定源卷积码的生成多项式矩阵阶数，遍历删余模式，构建线性方程组Gp(D)H(D)T＝0，其中GP(D)为生成矩阵，H(D)T为校验矩阵H(D)的转置，通过高斯消元法求解该方程组的唯一非零解，从而确定源卷积码的生成多项式和删余模式；5) Set the order of the generator polynomial matrix of the source convolutional code, traverse the puncturing mode, and construct a linear equation system Gp(D)H(D) T = 0, wherein G P (D) is the generator matrix, H(D) T is the transposition of the check matrix H(D), and the unique non-zero solution of the equation system is solved by the Gaussian elimination method, thereby determining the generator polynomial and the puncturing mode of the source convolutional code;

6)利用识别得到的分量编码器参数，将删余卷积码通过维特比译码得到置信度，验证识别参数的准确性。6) Using the identified component encoder parameters, the punctured convolutional code is obtained through Viterbi decoding to obtain confidence, and the accuracy of the identified parameters is verified.

Claims (1)

1. delete the parameter blind identification of remaining turbo code component coder for one kind, remaining turbo code information bit will be deleted and the first bit check bit multiplex constructs Punctured convolutional code, by the code length of linear matrix analytic approach statistical analysis Punctured convolutional code, code word starting point, recycling part Walsh-Hadamard conversion identifies the check matrix of component coder, obtained deleting complementary modul formula and generator matrix by Gaussian elimination method further, thus realize the parameter blind recognition deleting remaining turbo code component coder, and obtain confidence level by Viterbi decoding, to verify the accuracy identified, the method concrete steps are as follows:
1) from deleting remaining turbo code starting point, the check digit sequence choosing its information sequence and component coder RSC1 is multiplexing, structure Punctured convolutional code sequence, what obtain 1/2 code check more than deleting for 1/3 code check turbo code deletes remaining turbo code, if former turbo code sequence is expressed as x 1a 1b 1x 2a 2b 2x 3a 3b 3x 4a 4b 4then delete remaining rear turbo code sequence and be expressed as x 1a 1x 2b 2x 3a 3x 4b 4, structure Punctured convolutional code is x 1a 1x 2x 3a 3x 4, wherein x 1, x 2, x 3represent the signal bit sequence do not exported in the same time; a 1, a 2, a 3represent the RSC1 check digit sequence do not exported in the same time; b 1, b 2, b 3represent the RSC2 check digit sequence do not exported in the same time;
2) by Punctured convolutional code Sequence composition p × q matrix, require q > N, p > q, wherein N presentation code constraint degree, get the scope of determining train value q, change q obtains different matrix, calculate these ranks of matrix respectively, only leave and take the matrix that rank of matrix is not equal to columns, then elementary transformation is carried out to these matrixes unitization, write down the matrix train value that in unitization rear left, angular unit battle array dimension is equal, greatest common divisor is got to retention train value, thus obtains Punctured convolutional code code length n;
3) Punctured convolutional code sequence is re-constructed matrix, columns is step 2) middle certain train value N ' retained, wherein N ' the multiple that is code length n, get the smaller value retained in train value, get matrix line number and be greater than columns, code sequential shift is obtained n-1 different matrix, together with code sequence not shift matrix obtain n different matrix altogether, n is code length, elementary transformation is carried out to these matrixes unitization, write down the upper left corner unit matrix dimension of unitization rear matrix respectively, displacement when dimension is minimum and Punctured convolutional code starting point;
4) identify check polynomial matrix, concrete identification step is as follows:
A) set up coefficient matrix R (D) by the Punctured convolutional code sequence constructed, verification polynomial matrix is expressed as H (D)=[h 0(D), h 1(D) ... h n-1(D)] t, wherein, h i(D)=h 0i+ h 1id+h 2id 2... h did d, i ∈ [0, n-1], d=max{degh i(D) }, wherein symbol deg represents it is the function asking polynomial number of times, by R (D) × H (D) t=0 to be write as the form of equation group as follows:
r d 1 . . . r 01 . . . r dn . . . r 0 n r ( d + 1 ) 1 . . . r 11 . . . r ( d + 1 ) n . . . r 1 n . . . . . . . . . . . . r ( d + N ) 1 . . . r N 1 . . . r ( d + N ) n . . . r Nn h 00 . . . h d 0 . . . h 0 ( n - 1 ) . . . h d ( n - 1 ) - - - ( 1 )
Due to h n-1(D) containing constant term h in 0 (n-1)=1, the constant term part in formula (1) is moved on the right of equation, obtains following formula:
r d 1 . . . r 01 . . . r ( d - 1 ) n . . . r 0 n r ( d + 1 ) 1 . . . r 11 . . . r dn . . . r 1 n . . . . . . . . . . . . r ( d + N ) 1 . . . r N 1 . . . r ( d + N - 1 ) n . . . r Nn h 00 . . . h d 0 . . . h 1 ( n - 1 ) . . . h d ( n - 1 ) = - r dn - r ( d + 1 ) n - r ( d + N ) n - - - ( 2 )
B) by the coefficient matrix R'(D in formula (2)) be divided into two parts, matrix length is N+1, width is respectively R1 and R2, wherein R1 and R2's and be (d+1) × n-1, (usual R2 value is between 16 to 24), then first, second coefficient matrix dimension is respectively (N+1) × R1, (N+1) × R2;
C) set the binary vector iTemp that a length is R1, size is from 0 to 2 r1circulate, the row vector mould two of each fixing iTemp and the first coefficient matrix is added rear negate, and often row obtains a binary number 0 or 1, finally obtains the binary vector of a N+1 dimension, be expressed as P, P is added with the value of the vectorial corresponding row on the right of formula (2) equal sign;
D) again statistic is carried out to the second coefficient matrix, the row vector dimension of the second coefficient matrix be 1*R2, R2 arbitrarily ' 0 ' and ' 1 ' combination as state, totally 2 r2-1 state, each row vector is one of them state, and on the right of formula (2) equal sign, the value of vectorial corresponding row is as the output of this state, and the output valve of equal state adds up, and non-existent State-output value is 0, obtains one (2 r2-1) × 1 dimension state vector;
E) Walsh-Hadamard conversion is carried out to the above results, the solution being greater than confidence level is found to be converted to binary vector again, be expressed as Q, then iTemp vector now combined with binary vector Q the check polynomial H (D) just obtaining us and require;
5) set the generator polynomial matrix exponent number of source convolution code, traversal deletes complementary modul formula, builds system of linear equations Gp (D) H (D) t=0, wherein G p(D) be generator matrix, H (D) tfor the transposition of check matrix H (D), solved unique untrivialo solution of this equation group by Gaussian elimination method, thus determine the generator polynomial of source convolution code and delete complementary modul formula;
6) utilize the component coder parameter identifying and obtain, Punctured convolutional code is obtained confidence level by Viterbi decoding, the accuracy of checking identification parameter.
CN201510137808.9A 2015-03-26 2015-03-26 A Blind Identification Method of Parameters of Component Encoders with Punctured Turbo Codes Pending CN104683072A (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN201510137808.9A CN104683072A (en) 2015-03-26 2015-03-26 A Blind Identification Method of Parameters of Component Encoders with Punctured Turbo Codes

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN201510137808.9A CN104683072A (en) 2015-03-26 2015-03-26 A Blind Identification Method of Parameters of Component Encoders with Punctured Turbo Codes

Publications (1)

Publication Number Publication Date
CN104683072A true CN104683072A (en) 2015-06-03

Family

ID=53317716

Family Applications (1)

Application Number Title Priority Date Filing Date
CN201510137808.9A Pending CN104683072A (en) 2015-03-26 2015-03-26 A Blind Identification Method of Parameters of Component Encoders with Punctured Turbo Codes

Country Status (1)

Country Link
CN (1) CN104683072A (en)

Cited By (9)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN106411328A (en) * 2016-10-28 2017-02-15 华南理工大学 Soft-bit-based blind identification method for Turbo code interleaver
CN107370566A (en) * 2017-07-28 2017-11-21 西安电子科技大学 A kind of punctured Turbo codes blind-identification method under the conditions of error code
CN107566091A (en) * 2017-10-24 2018-01-09 重庆电子工程职业学院 A kind of blind recognition of convolutional code method for taking the dark information of Generalized Coupled into account
CN110535478A (en) * 2019-09-27 2019-12-03 电子科技大学 Dual input class Turbo code closed set recognition methods in a kind of DVB-RCS2 agreement
CN110690907A (en) * 2019-09-27 2020-01-14 电子科技大学 Known branch information turbo code deletion mode estimation method
CN111510164A (en) * 2020-05-14 2020-08-07 中国人民解放军海军航空大学 Turbo code component encoder identification method and system
CN115276886A (en) * 2021-04-29 2022-11-01 华为技术有限公司 A kind of code identification method and device
CN116566404A (en) * 2023-07-11 2023-08-08 北京谷数科技股份有限公司 Method and device for determining interleaving mapping relation of punctured Turbo codes
CN118074728A (en) * 2024-04-18 2024-05-24 北京邮电大学 A Turbo code puncturing pattern recognition method

Citations (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN102244520A (en) * 2010-05-11 2011-11-16 中国电子科技集团公司第三十六研究所 Blind recognition method of convolutional coding parameters
CN102244555A (en) * 2010-05-11 2011-11-16 中国电子科技集团公司第三十六研究所 Blind identification method for coding parameter of Turbo code
CN102244554A (en) * 2010-05-11 2011-11-16 中国电子科技集团公司第三十六研究所 Blind recognition method of punctured Turbo coding parameters
CN104184558A (en) * 2014-09-11 2014-12-03 山东大学 Fast blind recognition method for turbo code output block length
CN104270225A (en) * 2014-09-11 2015-01-07 山东大学 A Method for Blind Identification of Codeword Types in Error Control Coding

Patent Citations (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN102244520A (en) * 2010-05-11 2011-11-16 中国电子科技集团公司第三十六研究所 Blind recognition method of convolutional coding parameters
CN102244555A (en) * 2010-05-11 2011-11-16 中国电子科技集团公司第三十六研究所 Blind identification method for coding parameter of Turbo code
CN102244554A (en) * 2010-05-11 2011-11-16 中国电子科技集团公司第三十六研究所 Blind recognition method of punctured Turbo coding parameters
CN104184558A (en) * 2014-09-11 2014-12-03 山东大学 Fast blind recognition method for turbo code output block length
CN104270225A (en) * 2014-09-11 2015-01-07 山东大学 A Method for Blind Identification of Codeword Types in Error Control Coding

Non-Patent Citations (5)

* Cited by examiner, † Cited by third party
Title
刘健等: "基于Walsh-Hadamard变换的卷积码盲识别", 《电子与信息学报》 *
张永光: "一种Turbo码编码参数的盲识别方法", 《西安电子科技大学学报(自然科学版)》 *
张永光等: "信道编码及其识别分析", 《电子工业出版社》 *
李丹丹: "Turbo码的盲识别和基于LDPC码的压缩感知系统研究", 《中国优秀硕士学位论文全文数据库 信息科技辑》 *
陆佩忠: "删除卷积码的盲识别", 《中国科学 E辑 信息科学》 *

Cited By (12)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN106411328A (en) * 2016-10-28 2017-02-15 华南理工大学 Soft-bit-based blind identification method for Turbo code interleaver
CN107370566A (en) * 2017-07-28 2017-11-21 西安电子科技大学 A kind of punctured Turbo codes blind-identification method under the conditions of error code
CN107370566B (en) * 2017-07-28 2020-07-14 西安电子科技大学 A Blind Recognition Method of Punctured Turbo Codes under Error Condition
CN107566091A (en) * 2017-10-24 2018-01-09 重庆电子工程职业学院 A kind of blind recognition of convolutional code method for taking the dark information of Generalized Coupled into account
CN110535478A (en) * 2019-09-27 2019-12-03 电子科技大学 Dual input class Turbo code closed set recognition methods in a kind of DVB-RCS2 agreement
CN110690907A (en) * 2019-09-27 2020-01-14 电子科技大学 Known branch information turbo code deletion mode estimation method
CN110690907B (en) * 2019-09-27 2023-04-25 电子科技大学 A method for estimating the erasure mode of turbo codes with known branch information
CN111510164A (en) * 2020-05-14 2020-08-07 中国人民解放军海军航空大学 Turbo code component encoder identification method and system
CN115276886A (en) * 2021-04-29 2022-11-01 华为技术有限公司 A kind of code identification method and device
CN116566404A (en) * 2023-07-11 2023-08-08 北京谷数科技股份有限公司 Method and device for determining interleaving mapping relation of punctured Turbo codes
CN116566404B (en) * 2023-07-11 2023-09-19 北京谷数科技股份有限公司 Method and device for determining interleaving mapping relation of punctured Turbo codes
CN118074728A (en) * 2024-04-18 2024-05-24 北京邮电大学 A Turbo code puncturing pattern recognition method

Similar Documents

Publication Publication Date Title
CN104683072A (en) A Blind Identification Method of Parameters of Component Encoders with Punctured Turbo Codes
US10673462B2 (en) Coding method and coding device
CN106888025B (en) A kind of cascade Error-correcting Encoding and Decoding method and system based on polarization code
KR101438072B1 (en) Multiple programming of flash memory without erase
CN101803206B (en) ROM list-decoding of near codewords
CN101656541B (en) Coding method and device of RS codes
CN101867379B (en) A Decoding Method of Convolutional Codes Aided by Cyclic Redundancy Check
JP4185167B2 (en) Iterative decoding of product codes
US8103945B2 (en) Decoding method and decoding apparatus as well as program
CN101904101A (en) Encoder, Decoder, Encoding Method and Decoding Method
US12010352B2 (en) Coding method and related device
CN110098839B (en) Blind identification method for non-systematic convolutional code coding parameters under high error code
Barg et al. On the complexity of minimum distance decoding of long linear codes
CN104467875A (en) Blind recognition method for RS code and punctured convolutional code cascaded code parameters
CN104242957A (en) Decoding processing method and decoder
CN106708654A (en) Circuit structure for BCH error correcting code of NAND flash
KR20050063660A (en) An apparatus for encoding and decoding of low-density parity-check codes, and methods thereof
CN110690907B (en) A method for estimating the erasure mode of turbo codes with known branch information
CN103957016B (en) Turbo code encoder with low storage capacity and design method of Turbo code encoder
US7502983B2 (en) Binary error-correction coding method and decoding method for trellis with permuted and sequential symbols
Moradi Performance and computational analysis of polarization-adjusted convolutional (PAC) codes
CN111371465A (en) Bit interleaving method, system and medium for LDPC codeword
CN110708077B (en) LDPC code large number logic decoding method, device and decoder
US20050210363A1 (en) Error correction coding method using at least twice a same elementary code, coding method, corresponding coding and decoding devices
Grießer et al. A posteriory probability decoding of nonsystematically encoded block codes

Legal Events

Date Code Title Description
C06 Publication
PB01 Publication
C10 Entry into substantive examination
SE01 Entry into force of request for substantive examination
WD01 Invention patent application deemed withdrawn after publication
WD01 Invention patent application deemed withdrawn after publication

Application publication date: 20150603