CN104270225A - Code word type blind recognition method of error control coding - Google Patents

Abstract

The invention discloses a code word type blind recognition method of error control coding and belongs to the technical field of code word recognition in a communication system. The method includes the steps that the code length and the code word starting point of a code stream sequence are recognized through linear transformation with a coding parameter blind recognition method of a (n, k and m) system convolutional code, a punctured convolutional code and a Turbo code in the background technology; the code length range and the number of columns and orders of a plurality of row simplest matrixes are analyzed, and the Turbo code is distinguished out firstly; the (n, k and m) system convolutional code and the punctured convolutional code are distinguished out by observing the subblock matrix diagonal line rule of the row simplest matrixes and repeated emergence of elements of the column with the number equal to the row number of a first full-zero row, and blind recognition of code word types is finished finally. The code word type blind recognition method of error control coding has the advantages of being simple and direct in algorithm, high in recognition efficiency and accuracy and the like, and is applicable to the field of intelligent communication, information processing and the like.

Description

A kind of code-word type blind-identification method of error control coding

Technical field

The present invention relates to the code-word type blind-identification method of a kind of error control coding in digital communication system, be specially (n, k, m) systematic convolutional code and the code-word type blind-identification method deleting convolution code, Turbo code, be applicable to the field such as intelligent communication, information processing, belong to the code word recognition technology field in communication system.

Background technology

In collaboration communication, the known code-word type of recipient and coding parameter ability correct decoding, and in non-cooperating communication, recipient to the code stream received without any priori, first code-word type blind recognition must be carried out, because identification code-word type, it is the prerequisite really realizing Channel coding parameters blind recognition.

Nineteen fifty-five Ai Lisi (Elias) proposes convolution code, and has a lot of method for the blind recognition of (n, k, m) systematic convolutional code coding parameter.Zeng Jing according to the matrix arrangement regulation of convolution code sequence, gives the method solving true form check matrix in the master thesis " exploitation of VSAT Internet's detection system " of its University of Electronic Science and Technology in 2005.The people such as Xue Guoqing propose a kind of blind recognition matrix simplifying method for systematic convolutional code in the literary composition of " systematic convolutional code blind recognition " [information security and communication security, 2009].

Code check be 1/2 encoder for convolution codes code check lower, the redundant data being doubled in former data can be produced.By deleting remaining method, the source convolution code that can be 1/2 by code check generates deleting code of (n-1)/N-shaped code check.For deletion convolution code, Lu Peizhong, Shen Li etc. are published in Chinese science in 2005 at it: E collects in information science 35 (2) and proposes a kind of method solving deletion blind recognition of convolutional code by solving fault-tolerant system of linear equations in " blind recognition of deletion convolution code " literary composition.

Turbo code is applied widely in modern communications, and its maximum feature is by introducing interleaver and deinterleaver, realizes randomness coding, and realizes long code by effective combination of some short codes, reach the performance close to Shannon theoretical limit.Chinese patent CN102244555A discloses a kind of by the analysis that interweaves in Turbo code is converted into " convolution code+intertexture " pattern on November 16th, 2011, finally complete the method for Turbo code blind recognition.

The above-mentioned blind-identification method mentioned, when being all for known code-word type, to the coding parameter blind recognition of certain code word, and for the situation of code-word type the unknown, the blind recognition of coding parameter is not just known where to begin yet, therefore first will carry out the blind recognition of code-word type, the correct identification of code-word type is the primary prerequisite of Channel coding parameters identification.At present, consult existing list of references, there is not yet the data about code-word type blind recognition aspect.

Summary of the invention

In order to overcome defect and the deficiency of prior art existence, it is low that the present invention proposes a kind of computational complexity, high efficiency (n, k, m) systematic convolutional code and deletion convolution code, the code-word type blind-identification method of Turbo code and a kind of code-word type blind-identification method of error control coding, when priori there is no to the code stream received to solve, how code word is carried out to the problem of correct classification.

The technical scheme that the inventive method adopts is as follows:

A kind of code-word type blind-identification method of error control coding, take computer as hardware platform, realize that data to be identified are read in, the Presentation Function of identifying and recognition result, wherein data to be identified are mainly read in calculator memory by computer by identifying, then identified by this blind-identification method, on the computer screen, the step of this recognition methods is as follows in the result display identified:

(1) first define the Binary Zero of code word to be identified, 1 Bit data is code stream sequence, read in code stream sequence by computer and the length choosing recognition sequence is not less than (130 × 10) × 130, the length value of this recognition sequence is recognition matrix size maximum in a series of recognition matrixes set up in step 2;

(2) utilize recognition sequence to set up the recognition matrix of p × q size, wherein the size of matrix columns q gets 130 from 3 successively, and matrix line number p=10q;

(3) elementary transformation is carried out to recognition matrix and obtain the simplest matrix of row, calculate row the simplest rank of matrix r and upper left corner unit matrix dimension α thereof, the concrete grammar of elementary transformation is: to the from left to right column based processing of the data in recognition matrix, if the element on diagonal is 1, then the row at diagonal entry place is set to standard row, by same column, other have the row of the nonzero element first prime modulus two replaced with in this row and standard row to add the result of rear gained, if the element on diagonal is zero, then find the row at the nonzero element place below this row diagonal entry as standard row, by same column, other have the row of the nonzero element first prime modulus two replaced with in this row and standard row to add the result of rear gained, if the element of diagonal lower position is 0 entirely, then no longer carry out abbreviation,

(4) if row the simplest rank of matrix r and its columns q is unequal, then these four numerical value of difference q-r of columns q, the order r of current matrix, upper left corner unit matrix dimension α and columns and order are recorded, and with [q _i, r _i, α _i, q _i-r _i] form represents, subscript i represents i-th group of data, if the simplest rank of matrix r of row is equal with its columns q, goes to step 5;

(5) columns q is added 1, if q≤130, then return step 2, otherwise go to step 6;

(6) data [q of statistic record _i, r _i, α _i, q _i-r _i] group number scale is ReservedNum, if ReservedNum equals 0, then judge that code stream sequence is Turbo code, code-word type blind recognition terminates, if ReservedNum is not equal to 0, goes to step 7;

(7) the ReservedNum group data [q added up in comparison step 6 _i, r _i, α _i, q _i-r _i] in all upper left corner unit matrix dimension α _isize, obtain α _iminimum value be designated as α _min;

(8) these ReservedNum group data [q is observed _i, r _i, α _i, q _i-r _i] in all upper left corner unit matrix dimension α _isize, if there is a certain α _i> α _min, then this group [q is left out _i, r _i, α _i, q _i-r _i], ReservedNum subtracts 1 simultaneously, constantly repeats this operation, until remaining all data [q _i, r _i, α _i, q _i-r _i] in α _iall meet α _i=α _min, then the subscript of remaining data is renumberd by order from small to large;

(9) after step 8 data screening, more existing ReservedNum group data [q _i, r _i, α _i, q _i-r _i] in the simplest rank of matrix r of all row _isize, if there is a certain r _i>=r _i+1, then this group [q is left out _i, r _i, α _i, q _i-r _i], ReservedNum subtracts 1 simultaneously, and the subscript of remaining data is renumberd by order from small to large, constantly repeats this operation, until remaining all data [q _i, r _i, α _i, q _i-r _i] in r _iall meet r _i< r _i+1;

(10) after step 9 data screening, more existing ReservedNum group data [q _i, r _i, α _i, q _i-r _i] in the simplest matrix columns of all row and order difference q _i-r _isize, if there is a certain q _i-r _i>=q _i+1-r _i+1, then data [q is left out _i+1, r _i+1, α _i+1, q _i+1-r _i+1], ReservedNum subtracts 1 simultaneously, and the subscript of remaining data is renumberd by order from small to large, constantly repeats this operation, until remaining all data [q _i, r _i, α _i, q _i-r _i] in q _i-r _iall meet q _i-r _i< q _i+1-r _i+1;

(11) to remaining ReservedNum group data [q after previous step screening _i, r _i, α _i, q _i-r _i] in q _iask greatest common divisor, be code length ng;

(12) if code length ng > 8, then judge that code stream sequence is Turbo code, code-word type blind recognition terminates, otherwise goes to step 13;

(13) by code stream sequence by being less than 13 times of code lengths and setting up a recognition matrix p × q for the columns of code length multiple, and p > q, carry out elementary transformation to recognition matrix and obtain the simplest matrix of row, the concrete grammar of wherein elementary transformation is with in step 3;

(14) the simplest matrix diagonals line element of row is observed, see along the matrix lower right corner to angular direction, upper left, occur that element 0 and 1 is adjacent when first time on diagonal, and during the train value of the 0 column situation less than the train value of 1 column, calculate with code length ng with the train value q ' of this 0 element column, code word starting point can be obtained namely

(15) by the sequence after shiftg position in code stream sequence by being greater than 13 times of code lengths and setting up n abundant p for the columns of code length multiple _j× q _jthe recognition matrix of (1≤j≤n) size, and p _j> q _jif this n matrix column value is q ₁, q ₂, Lq _n, after the elementary transformation through describing in step 3, all abbreviation is the simplest matrix of row, and its order is respectively r ₁, r ₂, Lr _n;

(16) when n is tending towards infinite, if meet one of following two conditions, then judge that code stream sequence is Turbo code, code-word type blind recognition terminates, if following two conditions all do not meet, go to step 17, these two conditions are specially:

A) there is r _j=q _j(1≤j≤n);

b) $\underset{n\&RightArrow;\&infin;}{\lim }\frac{r_2-r_1}{q_2-q_1}=\frac{r_3-r_2}{q_3-q_2}=L=\frac{r_n-r_{n-1}}{q_n-q_{n-1}}$ Be false;

(17) series arrangement after shiftg position in code stream sequence is become the recognition matrix p × q being columns with 30 times of code lengths, and p > q, carry out elementary transformation to recognition matrix and obtain the simplest matrix of row, the concrete grammar of wherein elementary transformation is with in step 3;

(18) see along the matrix upper left corner to direction, the lower right corner, observe the matrix sub block in units of code length in the simplest matrix of row, if the number of 1 is not equal to code length and subtracts 1 in sub-block matrix diagonals line element, then judge that code stream sequence is as (n, k, m) systematic convolutional code, code-word type blind recognition terminates, otherwise goes to step 19;

(19) the full zero row row value of note row first, the simplest matrix is q ₀if, train value q ₀q from top to bottom in column ₀individual element is at q ₀+ d × ng (d ∈ N ^*) arrange and repeat, then judge that code stream sequence is as (n, k, m) systematic convolutional code, otherwise judge that code stream sequence is as deleting convolution code, code-word type blind recognition terminates.

Beneficial effect of the present invention is as follows:

(1) matrix-analysis method has fully been applied in code-word type blind recognition, the characteristic of the simplest matrix of row of code stream sequence after various coding has comprehensively been analyzed, determines the validity feature for type identification;

(2), in identifying, adopt by the stepping mode of rough sort to disaggregated classification, class condition is refinement step by step, finally completes the blind recognition of code-word type;

(3), in identifying, have employed less recognition matrix number and less recognition matrix dimension as far as possible, reduce computational complexity and improve recognition speed.

 

Embodiment

Below in conjunction with embodiment, the invention will be further described, but be not limited thereto.

Embodiment:

The embodiment of the present invention is as follows, a kind of code-word type blind-identification method of error control coding, take computer as hardware platform, realize that data to be identified are read in, the Presentation Function of identifying and recognition result, wherein data to be identified are mainly read in calculator memory by computer by identifying, then identified by this blind-identification method, on the computer screen, the step of this recognition methods is as follows in the result display of identification:

(1) first define the Binary Zero of code word to be identified, 1 Bit data is code stream sequence, read in code stream sequence by computer and the length choosing recognition sequence is not less than (130 × 10) × 130, the length value of this recognition sequence is recognition matrix size maximum in a series of recognition matrixes set up in step 2;

(2) utilize recognition sequence to set up the recognition matrix of p × q size, wherein the size of matrix columns q gets 130 from 3 successively, and matrix line number p=10q;

(3) elementary transformation is carried out to recognition matrix and obtain the simplest matrix of row, calculate row the simplest rank of matrix r and upper left corner unit matrix dimension α thereof, the concrete grammar of elementary transformation is: to the from left to right column based processing of the data in recognition matrix, if the element on diagonal is 1, then the row at diagonal entry place is set to standard row, by same column, other have the row of the nonzero element first prime modulus two replaced with in this row and standard row to add the result of rear gained, if the element on diagonal is zero, then find the row at the nonzero element place below this row diagonal entry as standard row, by same column, other have the row of the nonzero element first prime modulus two replaced with in this row and standard row to add the result of rear gained, if the element of diagonal lower position is 0 entirely, then no longer carry out abbreviation,

(4) if row the simplest rank of matrix r and its columns q is unequal, then these four numerical value of difference q-r of columns q, the order r of current matrix, upper left corner unit matrix dimension α and columns and order are recorded, and with [q _i, r _i, α _i, q _i-r _i] form represents, subscript i represents i-th group of data, if the simplest rank of matrix r of row is equal with its columns q, goes to step 5;

(5) columns q is added 1, if q≤130, then return step 2, otherwise go to step 6;

(6) data [q of statistic record _i, r _i, α _i, q _i-r _i] group number scale is ReservedNum, if ReservedNum equals 0, then judge that code stream sequence is Turbo code, code-word type blind recognition terminates, if ReservedNum is not equal to 0, goes to step 7;

(7) the ReservedNum group data [q added up in comparison step 6 _i, r _i, α _i, q _i-r _i] in all upper left corner unit matrix dimension α _isize, obtain α _iminimum value be designated as α _min;

(8) these ReservedNum group data [q is observed _i, r _i, α _i, q _i-r _i] in all upper left corner unit matrix dimension α _isize, if there is a certain α _i> α _min, then this group [q is left out _i, r _i, α _i, q _i-r _i], ReservedNum subtracts 1 simultaneously, constantly repeats this operation, until remaining all data [q _i, r _i, α _i, q _i-r _i] in α _iall meet α _i=α _min, then the subscript of remaining data is renumberd by order from small to large;

(9) after step 8 data screening, more existing ReservedNum group data [q _i, r _i, α _i, q _i-r _i] in the simplest rank of matrix r of all row _isize, if there is a certain r _i>=r _i+1, then this group [q is left out _i, r _i, α _i, q _i-r _i], ReservedNum subtracts 1 simultaneously, and the subscript of remaining data is renumberd by order from small to large, constantly repeats this operation, until remaining all data [q _i, r _i, α _i, q _i-r _i] in r _iall meet r _i< r _i+1;

(10) after step 9 data screening, more existing ReservedNum group data [q _i, r _i, α _i, q _i-r _i] in the simplest matrix columns of all row and order difference q _i-r _isize, if there is a certain q _i-r _i>=q _i+1-r _i+1, then data [q is left out _i+1, r _i+1, α _i+1, q _i+1-r _i+1], ReservedNum subtracts 1 simultaneously, and the subscript of remaining data is renumberd by order from small to large, constantly repeats this operation, until remaining all data [q _i, r _i, α _i, q _i-r _i] in q _i-r _iall meet q _i-r _i< q _i+1-r _i+1;

(11) to remaining ReservedNum group data [q after previous step screening _i, r _i, α _i, q _i-r _i] in q _iask greatest common divisor, be code length ng;

(12) if code length ng > 8, then judge that code stream sequence is Turbo code, code-word type blind recognition terminates, otherwise goes to step 13;

(13) by code stream sequence by being less than 13 times of code lengths and setting up a recognition matrix p × q for the columns of code length multiple, and p > q, carry out elementary transformation to recognition matrix and obtain the simplest matrix of row, the concrete grammar of wherein elementary transformation is with in step 3;

(14) the simplest matrix diagonals line element of row is observed, see along the matrix lower right corner to angular direction, upper left, occur that element 0 and 1 is adjacent when first time on diagonal, and during the train value of the 0 column situation less than the train value of 1 column, calculate with code length ng with the train value q ' of this 0 element column, code word starting point can be obtained namely

(15) by the sequence after shiftg position in code stream sequence by being greater than 13 times of code lengths and setting up n abundant p for the columns of code length multiple _j× q _jthe recognition matrix of (1≤j≤n) size, and p _j> q _jif this n matrix column value is q ₁, q ₂, Lq _n, after the elementary transformation through describing in step 3, all abbreviation is the simplest matrix of row, and its order is respectively r ₁, r ₂, Lr _n;

(16) when n is tending towards infinite, if meet one of following two conditions, then judge that code stream sequence is Turbo code, code-word type blind recognition terminates, if following two conditions all do not meet, go to step 17, these two conditions are specially:

A) there is r _j=q _j(1≤j≤n);

b) $\underset{n\&RightArrow;\&infin;}{\lim }\frac{r_2-r_1}{q_2-q_1}=\frac{r_3-r_2}{q_3-q_2}=L=\frac{r_n-r_{n-1}}{q_n-q_{n-1}}$ Be false;

(17) series arrangement after shiftg position in code stream sequence is become the recognition matrix p × q being columns with 30 times of code lengths, and p > q, carry out elementary transformation to recognition matrix and obtain the simplest matrix of row, the concrete grammar of wherein elementary transformation is with in step 3;

(18) see along the matrix upper left corner to direction, the lower right corner, observe the matrix sub block in units of code length in the simplest matrix of row, if the number of 1 is not equal to code length and subtracts 1 in sub-block matrix diagonals line element, then judge that code stream sequence is as (n, k, m) systematic convolutional code, code-word type blind recognition terminates, otherwise goes to step 19;

(19) the full zero row row value of note row first, the simplest matrix is q ₀if, train value q ₀q from top to bottom in column ₀individual element is at q ₀+ d × ng (d ∈ N ^*) arrange and repeat, then judge that code stream sequence is as (n, k, m) systematic convolutional code, otherwise judge that code stream sequence is as deleting convolution code, code-word type blind recognition terminates.

Claims (1)

1. the code-word type blind-identification method of an error control coding, take computer as hardware platform, realize that data to be identified are read in, the Presentation Function of identifying and recognition result, wherein data to be identified are mainly read in calculator memory by computer by identifying, then identified by this blind-identification method, on the computer screen, the step of this recognition methods is as follows in the result display identified:

(1) first define the Binary Zero of code word to be identified, 1 Bit data is code stream sequence, read in code stream sequence by computer and the length choosing recognition sequence is not less than (130 × 10) × 130, the length value of this recognition sequence is recognition matrix size maximum in a series of recognition matrixes set up in step 2;

(2) utilize recognition sequence to set up the recognition matrix of p × q size, wherein the size of matrix columns q gets 130 from 3 successively, and matrix line number p=10q;

(3) elementary transformation is carried out to recognition matrix and obtain the simplest matrix of row, calculate row the simplest rank of matrix r and upper left corner unit matrix dimension α thereof, the concrete grammar of elementary transformation is: to the from left to right column based processing of the data in recognition matrix, if the element on diagonal is 1, then the row at diagonal entry place is set to standard row, by same column, other have the row of the nonzero element first prime modulus two replaced with in this row and standard row to add the result of rear gained, if the element on diagonal is zero, then find the row at the nonzero element place below this row diagonal entry as standard row, by same column, other have the row of the nonzero element first prime modulus two replaced with in this row and standard row to add the result of rear gained, if the element of diagonal lower position is 0 entirely, then no longer carry out abbreviation,

(4) if row the simplest rank of matrix r and its columns q is unequal, then these four numerical value of difference q-r of columns q, the order r of current matrix, upper left corner unit matrix dimension α and columns and order are recorded, and with [q _i, r _i, α _i, q _i-r _i] form represents, subscript i represents i-th group of data, if the simplest rank of matrix r of row is equal with its columns q, goes to step 5;

(5) columns q is added 1, if q≤130, then return step 2, otherwise go to step 6;

(6) data [q of statistic record _i, r _i, α _i, q _i-r _i] group number scale is ReservedNum, if ReservedNum equals 0, then judge that code stream sequence is Turbo code, code-word type blind recognition terminates, if ReservedNum is not equal to 0, goes to step 7;

(7) the ReservedNum group data [q added up in comparison step 6 _i, r _i, α _i, q _i-r _i] in all upper left corner unit matrix dimension α _isize, obtain α _iminimum value be designated as α _min;

(8) these ReservedNum group data [q is observed _i, r _i, α _i, q _i-r _i] in all upper left corner unit matrix dimension α _isize, if there is a certain α _i> α _min, then this group [q is left out _i, r _i, α _i, q _i-r _i], ReservedNum subtracts 1 simultaneously, constantly repeats this operation, until remaining all data [q _i, r _i, α _i, q _i-r _i] in α _iall meet α _i=α _min, then the subscript of remaining data is renumberd by order from small to large;

(9) after step 8 data screening, more existing ReservedNum group data [q _i, r _i, α _i, q _i-r _i] in the simplest rank of matrix r of all row _isize, if there is a certain r _i>=r _i+1, then this group [q is left out _i, r _i, α _i, q _i-r _i], ReservedNum subtracts 1 simultaneously, and the subscript of remaining data is renumberd by order from small to large, constantly repeats this operation, until remaining all data [q _i, r _i, α _i, q _i-r _i] in r _iall meet r _i< r _i+1;

(10) after step 9 data screening, more existing ReservedNum group data [q _i, r _i, α _i, q _i-r _i] in the simplest matrix columns of all row and order difference q _i-r _isize, if there is a certain q _i-r _i>=q _i+1-r _i+1, then data [q is left out _i+1, r _i+1, α _i+1, q _i+1-r _i+1], ReservedNum subtracts 1 simultaneously, and the subscript of remaining data is renumberd by order from small to large, constantly repeats this operation, until remaining all data [q _i, r _i, α _i, q _i-r _i] in q _i-r _iall meet q _i-r _i< q _i+1-r _i+1;

(11) to remaining ReservedNum group data [q after previous step screening _i, r _i, α _i, q _i-r _i] in q _iask greatest common divisor, be code length ng;

(12) if code length ng > 8, then judge that code stream sequence is Turbo code, code-word type blind recognition terminates, otherwise goes to step 13;

(13) by code stream sequence by being less than 13 times of code lengths and setting up a recognition matrix p × q for the columns of code length multiple, and p > q, carry out elementary transformation to recognition matrix and obtain the simplest matrix of row, the concrete grammar of wherein elementary transformation is with in step 3;

(14) the simplest matrix diagonals line element of row is observed, see along the matrix lower right corner to angular direction, upper left, occur that element 0 and 1 is adjacent when first time on diagonal, and during the train value of the 0 column situation less than the train value of 1 column, calculate with code length ng with the train value q ' of this 0 element column, code word starting point can be obtained namely

(15) by the sequence after shiftg position in code stream sequence by being greater than 13 times of code lengths and setting up n abundant p for the columns of code length multiple _j× q _jthe recognition matrix of (1≤j≤n) size, and p _j> q _jif this n matrix column value is q ₁, q ₂, L q _n, after the elementary transformation through describing in step 3, all abbreviation is the simplest matrix of row, and its order is respectively r ₁, r ₂, Lr _n;

(16) when n is tending towards infinite, if meet one of following two conditions, then judge that code stream sequence is Turbo code, code-word type blind recognition terminates, if following two conditions all do not meet, go to step 17, these two conditions are specially:

A) there is r _j=q _j(1≤j≤n);

b) $\underset{n\&RightArrow;\&infin;}{\lim }\frac{r_2-r_1}{q_2-q_1}=\frac{r_3-r_2}{q_3-q_2}=L=\frac{r_n-r_{n-1}}{q_n-q_{n-1}}$ Be false;

(17) series arrangement after shiftg position in code stream sequence is become the recognition matrix p × q being columns with 30 times of code lengths, and p > q, carry out elementary transformation to recognition matrix and obtain the simplest matrix of row, the concrete grammar of wherein elementary transformation is with in step 3;

(18) see along the matrix upper left corner to direction, the lower right corner, observe the matrix sub block in units of code length in the simplest matrix of row, if the number of 1 is not equal to code length and subtracts 1 in sub-block matrix diagonals line element, then judge that code stream sequence is as (n, k, m) systematic convolutional code, code-word type blind recognition terminates, otherwise goes to step 19;

(19) the full zero row row value of note row first, the simplest matrix is q ₀if, train value q ₀q from top to bottom in column ₀individual element is at q ₀+ d × ng (d ∈ N ^*) arrange and repeat, then judge that code stream sequence is as (n, k, m) systematic convolutional code, otherwise judge that code stream sequence is as deleting convolution code, code-word type blind recognition terminates.