CN101217337A - A low density parity code encoding device and method supporting incremental redundancy hybrid automatic repeat - Google Patents

Abstract

The invention discloses an encoding method of a structured parity check code which has low density, supports any code rate and any code length, and is used for encoding any information bit sequence with length of K that is input by a channel encoder, outputting any bit sequence of a code word with length of N into a subsequent processing unit, and the length of the check bits is M, equal to the result of subtracting K from N, the code rate is r, equal to the result of K divided by N; furthermore, the method comprises the following steps: (1) a mother code set of the parity check code is determined with low density in a given code rate R0, and the mother code set consists of finite parity check codes that have low density, code rate R0 and different code length, and has a basis matrix Hb<uniform> that is in a unified form; (2) an extension factor is determined according to the length matching of information groups, the parameters and the matrixes which are required by the encoder are determined through shortening, removing or extending according to the extension factor and the basis matrix Hb<uniform>; (3) the encoder encodes the input information groups with K bits according to the parameters and the matrixes required by the encoder, generating a code word with N bits and outputting the code word.

Description

A kind of steadily increase redundancy of supporting mixes low density parity code encoding device and the method that retransmits automatically

Technical field

The present invention relates to, particularly the present invention relates to be used in a kind of digital communication system the code device and the method for transfer of data error correction, particularly relate to the code device and the method for the low density parity check code in the digital communicating field error correcting technique.Based on a female sign indicating number collection, by means of to the expansion (extending) of basis matrix, shorten (shortening) and delete surplus (puncturing), can obtain a code check code length flexibly, best performance, the effective low density parity check code encoder of hardware.The invention still further relates to a kind of method of HARQ packet generation of LDPC sign indicating number, above-mentioned new LDPC sign indicating number can be used for the method for this HARQ packet generation, has guaranteed that new LDPC sign indicating number can support steadily increase redundancy to mix automatic repeat requests IR HARQ well.The present invention gives the error detecting method of a kind of new LDPC, and this method can be applied to the error detection occurs of above-mentioned LDPC HARQ, has guaranteed the performance and overhead (CRC bit) best compromise of error detection occurs.

Background technology

All digital communication systems all need channel encoder, as shown in Figure 1.Channel encoder is in order to resist various noises and interference in the transmission course, and by increasing redundant information artificially, the system that makes has the ability of automatic correction mistake, thereby guarantees the reliability of Digital Transmission.Low density parity check code is that a class can be found by Gallager, so be called the Gallager sign indicating number at first with the linear block codes of very sparse parity matrix or bipartite graph definition.Through the silence of many decades, along with the development of computer hardware and correlation theory, MacKay and Neal have rediscovered it, and have proved that it has the performance of approaching shannon limit.Current research shows that the low-density parity check code sign indicating number has following characteristics: low decoding complexity, but linear time code has the performance of approaching shannon limit, but parallel decoding, and performance is better than Turbo code under the long code elongate member.

The LDPC sign indicating number is a kind of based on sparse parity check matrix, utilizes the sparse property of its check matrix just, could realize the coding and decoding of low complex degree, thereby make the LDPC sign indicating number move towards practicability.The LDPC sign indicating number that above-mentioned Gallager sign indicating number is a kind of canonical (regular LDPCC), and people such as Luby and Mitzenmacher promote the Gallager sign indicating number, propose non-regular LDPC sign indicating number (irregular LDPCC).The LDPC sign indicating number has a lot of decoding algorithms, wherein, information pass-algorithm (Message Passing algorithm) or confidence spread algorithm (Belief Propagationalgorithm, the BP algorithm) is the main flow and the basic algorithm of LDPC sign indicating number, occurred a lot of improved efficient coding algorithms at present.

The diagrammatic representation form of LDPC parity matrix is a bipartite graph.Have one to one relation between bipartite graph and the check matrix, the parity check matrix H of a M*N has defined the constraint that each code word with N bit satisfies M parity check collection.A bipartite graph comprises N variable node and M parity check node.When m verification relates to n bit, i.e. the element H of the capable n row of m among the H _{M, n}, will there be a line to connect check-node m and variable node n at=1 o'clock.In the bipartite graph, do not have connection between any of a sort node, and the number of nonzero element equates in total limit number in the bipartite graph and the check matrix.

The special LDPC sign indicating number of one class becomes mainstream applications gradually owing to have structurized feature.If the parity check matrix H of this LDPC sign indicating number is the (M * z) * (matrix of N * z), it is to be made of M * N matrix in block form, each matrix in block form all is the different powers of the basic permutation matrix of z * z, when basic permutation matrix was unit matrix, they all were the cyclic shift matrices (be defaulted as in the literary composition and move to right) of unit matrix.Have following form:

If $h_{\mathrm{ij}}^b=-1,$ Have $P^{h_{\mathrm{ij}}^b}=0.$

If h _Ij ^bBe to be greater than or equal to 0 integer, definition $P^{h_{\mathrm{ij}}^b}={\left(P\right)}^{h_{\mathrm{ij}}^b},$ Here P is a z * z

$P=\left[\begin{array}{ccccc}0& 1& 0& ...& 0\\ 0& 0& 1& ...& 0\\ ...& ...& ...& ...& ...\\ 0& 0& 0& ...& 1\\ 1& 0& 0& ...& 0\end{array}\right]$

The standard permutation matrix, as follows:

By such power time h _Ij ^bJust can each matrix in block form of unique identification, the power of unit matrix time can

$H=\left[\begin{array}{ccccc}{P}^{h_{00}^b}& P^{h_{01}^b}& P^{h_{02}^b}& ...& P^{h_{0n_b}^b}\\ P^{h_{10}^b}& P^{h_{11}^b}& P^{h_{12}^b}& ...& P^{h_{1n_b}^b}\\ ...& ...& ...& ...& ...\\ P^{h_{m_{b}0}^b}& P^{h_{m_{b}1}^b}& P^{h_{m_{b}2}^b}& ...& P^{h_{m_b{n}_b}^b}\end{array}\right]=P^{H_b}$

Represent that with 0 null matrix is generally represented with-1.Like this, if all use its power to replace each matrix in block form of H, just obtain a m _b* n _bPower submatrix H _bHere, definition H _bB is the basis matrix of H, and H is called H _bExtended matrix.When actual coding, the columns n of z=code length/basis matrix _b, be called spreading factor.

For example, matrix H can be with following parameter z and one 's 2 * 4 basis matrix H _bExpansion obtains:

Z=3 and $H_b=\left[\begin{array}{cccc}0& 1& 0& -1\\ 2& 1& 2& 1\end{array}\right]$

Therefore, encoder that also we can say LDPC sign indicating number of the present invention is by basis matrix H _b, spreading factor z and the unique generation of selected basic permutation matrix.

If all adopt a basis matrix for each different spreading factor LDPC sign indicating number, so, for each different code length, described LDPC code coding/decoding device all needs to store a basis matrix, when code length is a lot, will store a lot of basis matrixs, expression and storage problem can appear like this.Therefore, when needs were realized becoming code length, the low density parity check code of multiple code length can use the basis matrix of a unified form in the certain limit of same code check, and we are defined as unified basis matrix H _b ^UniformDuring different code length, if to H _b ^UniformRevise and expand, can obtain parity check matrix H, make to generate

Coder applicable to the variable occasion of code length.

Correction is to utilize the spreading factor of different code length to basis matrix H _bIn nonnegative value revise, revising the back element value should be less than the spreading factor value under this code length.Correction algorithm has a variety of, for example, can adopt delivery (mod), round (scale+floor) or round off (scale+round) etc., establishes P _IjBe basis matrix H _bNon-negative 1 element of the capable j row of i, P ' _IjBe H after revising _b ^ModifiedNon-negative 1 element of the capable j of i row, have:

For delivery (mod) method:

$P_{\mathrm{ij}}^{\&prime;}\&equiv;P_{\mathrm{ij}}\mathrm{mod}z\&equiv;P_{\mathrm{ij}}\mathrm{mod}\frac{N}{n_b}$

For rounding (scale+floor) method:

For (scale+round) method that rounds off:

$P_{\mathrm{ij}}^{\&prime;}=\mathrm{Round}(P_{\mathrm{ij}}\&times;\frac{z}{z_{\max }})=\mathrm{Round}(P_{\mathrm{ij}}\&times;\frac{N}{N_{\max }})$

Wherein, z is the spreading factor of current code length correspondence, i.e. the line number of piecemeal square formation or columns; z _MaxSupport the spreading factor of code length correspondence for maximum.Mod is a modulo operation,

For under round operation, Round is the operation that rounds up.

For example, for code length 1152 bit low density parity check codes, establishing certain non-negative element of its basis matrix is 93, and the maximum code length of establishing support is 2304, and the basis matrix size is 12*24, and then its correction result is:

For delivery (mod) method:

$93\mathrm{mod}\frac{1152}{24}=93\mathrm{mod}48=45$

For rounding (scale+floor) method:

For (scale+round) method that rounds off:

$\mathrm{Round}(93\&times;\frac{1152}{2304})=\mathrm{Round}\left(46.5\right)=47$

The low density parity check code of this specific code check variable code length is because basis matrix with same form, so fully can be with an encoder/decoder.In IEEE802.16e, the LDPC sign indicating number is supported code check 1/2, code check 2/3, code check 3/4, code check 5/6, this moment occurs four basis matrixs, realizes the coding and decoding of the different code checks comparison difficulty that will become with unique encoding device/decoder; If four code checks need four coders, hardware cost will become one and use bottleneck.

Automatically request retransmission HARQ is an extremely important link adaptation techniques in a kind of wireless communication system.

Below just automatic request retransmission HARQ method is done simple an introduction, this method comprises in the transmitter method of reseptance two parts in the sending method and receiver, this method should be general.

The sending method of transmitter (Transmitter) comprising:

A. transmitter sends first HARQ bag at the HARQ of appointment channel, a new HARQ transmission beginning, and retransmission times counter puts 1;

B. the affirmation of transmitter wait receiver transmission/deny (ACK/NAK);

C. in certain back (depend on the method for synchronization or asynchronous system, postpone by decisions such as systems) that postpones, if transmitter is received ACK, then this bag correctly receives this HARQ end of transmission; Otherwise retransmission times counter adds 1, judges whether number of retransmissions surpasses the maximum the number of transmissions that allows, and as if surpassing, then this HARQ end of transmission if do not surpass, then goes to d.

D. transmitter is done the trial of another HARQ once more at the HARQ channel of this HARQ transmission, and transmitting time can be selected by system, goes to b.

The method of reseptance of receiver (Reiciver) comprising:

A. receive HARQ bag, whether the bag that acknowledges receipt of is that first HARQ attempts, if go to b. otherwise go to d.

B. being discarded in the HARQ that receives before all under this HARQ channel attempts;

C. bag is decoded, go to e.

D. receiver decoder makes up former mistake and encoded packets that retransmit according to specific merging method;

If e. be correctly decoded, then after certain delay, in feedback channel, send ACK, this channel is by system assignment; Otherwise, send NAK, store this HARQ bag.

In the steps d of above-mentioned method of reseptance, the method for merging may comprise that complete steadily increase redundancy, part steadily increase redundancy and Chase merge.

To automatic request retransmission HARQ implementation structure, lot of documents and books all have introduction.No matter be method or structure, it is important that the pass is all put in the generation of HARQ bag, determined performance and complexity.

Because HARQ is a kind of very important link adaptation techniques, so good low density parity check code must be supported the realization of HARQ, this also is the most important purpose of the present invention.The present invention will provide the HARQ packet generation method based on low density parity check code design of the present invention.

For the HARQ system of low density parity check code, error detection occurs is necessary, and two kinds of typical methods are arranged.A kind of is directly to utilize the parity matrix of low density parity check code to detect, and the transposition that is about to codeword vector and parity matrix judges to taking advantage of whether multiplied result is null vector, is then correct again, is not mistake then.

Another kind is, before coding, add cyclic redundancy after the information bit sequence and detect (CRC) bit, the LDPC encoder is encoded to above-mentioned CRC code word, produce the LDPC code word and give the postorder processing unit, arrive receiver by Channel Transmission, the soft code word of LDPC of the processing unit output in the receiver before the decoder will be delivered to ldpc decoder and decipher, and the information block of decoding output is carried out CRC and detected.

Summary of the invention

The technical problem to be solved in the present invention provides the problem that LDPC sign indicating number that a kind of the object of the invention is to solve different code check different code lengths can not use identical coder, a kind of newly-designed coder and method are provided, improved the LDPC sign indicating number in performances such as the flexibility of code check/code length, decoding speed (throughput, decoding postpones), power loss, coding and decoding complexities.The present invention is particularly suitable for supporting steadily increase redundancy to mix and retransmits (IR_HARQ), can guarantee low code check realization by the unified basis matrix of expansion (extend), guaranteed to support complete steadily increase redundancy to mix that the LDPC sign indicating number has enough low code check when retransmitting IR_HARQ, as 1/3,1/5 etc., the decoder complexity is minimum as a result.

In order to solve the problems of the technologies described above, the invention provides a kind of structured low density parity check code coding method of supporting any code check/code length, random length to the channel encoder input is encoded for the K information bit sequence, exporting random length then is that N code word bits sequence is given follow-up processing unit, check digit length M=N-K, code check r=K/N, described method comprises the steps:

1) determines a specific code check R ₀The female sign indicating number collection of low density parity check code, described female sign indicating number collection is to be R by limited code check ₀The low density parity check code of different code length constitutes, the basis matrix H that described female sign indicating number collection has Unified Form _b ^Uniform

2) according to the information block length coupling, determine spreading factor; According to described spreading factor and basis matrix H _b ^Uniform,, determine encoder parameters needed and matrix by shortening or deleting surplus or expansion;

3) encoder is encoded to the K bit information grouping of input according to encoder parameters needed and matrix, produces the N bit codewords, the output code word.

The method according to this invention, in step 1),

The female sign indicating number collection of described low density parity check code has defined P same code rate R ₀(N=z * the n of different code length _b, K=z * k _b) low density parity check code, wherein, Zset is the spreading factor set, and spreading factor z can be any one element among the Zset, and Zset is greater than zero positive integer z by P ₁＜z ₂＜...＜z _K-1＜z _k＜...＜z _PThe set that the family constitutes, P is certain greater than 1 integer, and basis matrix H _b ^UniformCode check is always arranged ${\mathrm{<figure-callout\; id="0"\; label="R"\; filenames="A20071007280800392.png,A20071007280800401.png"\; state="\{\{state\}\}">R</figure-callout>}}_0=\frac{(n_b-m_b)}{n_b},$ Wherein, n _bBe definite integer greater than 2, m _bBe definite integer greater than 1, k _b=n _b-m _b

The method according to this invention is in step 2) in, determine that relatively code check R is less than R ₀, calculation code desired parameters and matrix comprise:

Step B1, utilization

Determine Δ m, at basis matrix H _b ^UniformLast column and to increase Δ m respectively capable and increase Δ m row in last row back, expansion forms (m _b+ Δ m) * (n _b+ Δ m) expansion basis matrix H _b ^{Uniform_extension}Here,

Expression rounds downwards;

Step B2, utilize specific spreading factor formula to determine spreading factor, determine z _tWith element among the Zset the following z of relation is arranged _K-1＜z _t≤ z _k, z wherein _K-1, z _kBe the adjacent element of size among the Zset, the required spreading factor of encoder is z _k

Step B3, based on the unified spread foundation matrix H in expansion back _b ^{Uniform_extension}With spreading factor z=z _k, according to specific correction algorithm to H _b ^{Uniform_extension}Revise, can calculate ((n _b+ Δ m) * z _k, k _b* z _k) the encode basis matrix H of needed correction of LDPC _b ^{Modifed_extension}Here, the basis matrix of the correction needed matrix of encoding exactly, spreading factor and the matrix size parameter parameters needed of encoding exactly.

The method according to this invention is in step 2) in, determine that relatively code check R is more than or equal to R ₀, calculation code desired parameters and matrix comprise:

Utilize spreading factor $z_t=\frac{K}{k_b}$ Determine spreading factor z _t, determine z _tWith element among the Zset the following z of relation is arranged _K-1＜z _t≤ z _k, z wherein _K-1, z _kBe the adjacent element of size among the Zset, the required spreading factor of encoder is z _kBased on unified basis matrix H _b ^UniformWith spreading factor z _k, can calculate (m according to specific correction algorithm and spreading factor z _b* z _k, n _b* z _k) the needed revised basis matrix of coding; Wherein, the basis matrix of the described correction needed matrix of encoding exactly, spreading factor and the matrix size parameter parameters needed of encoding exactly.

The method according to this invention, in step 3), the coding of information bit calculated comprise:

Step B1, interpolation x=k _bZ-K zero bit before K the information bit, the formation needed k that encodes _bZ _kInformation block; Carry out ((n then _b+ Δ m) * z _k, k _b* z _k) the LDPC coding, form (n _b+ Δ m) * z _kIndividual coding codeword bit;

Step B2, in above-mentioned coding codeword, the x that adds among the deletion step B1 zero bit if code word bits number after the deletion and needed grouping size N do not match, then behind deletion zero bit, continues to select correct position, y=(m in the deletion code word _b+ Δ m) * z _k-N+K bit, forming length is the coding codeword of N.

The method according to this invention, particular extension factor computing formula is among the step B2: $z_t=\frac{K}{k_b}.$

The method according to this invention, particular extension factor computing formula is among the step B2: $z_t=\frac{N-K}{m_b+\mathrm{\&Delta;m}}.$

The method according to this invention, in step B1, described basis matrix expansion method is:

For H _b ^ExtensionIn newly-increased matrix part, satisfy m for any _b+ 1≤i≤m _b+ Δ m line index i has

$H_b^{\mathrm{extension}}(i,i-m_b)=0$

$H_b^{\mathrm{extension}}(i,i-m_b+n_b-1)=0$

$H_b^{\mathrm{extension}}(i,i-m_b+n_b)=0$

Fill with the element of representing zero square formation other position, as-1;

Wherein, i is a matrix H _b ^ExtensionLine index, the ranks index is all since 1.

The method according to this invention, in step B1, described basis matrix expansion method is:

For H _b ^ExtensionIn newly-increased matrix part, for line index i=m _b+ 1, always have

$H_b^{\mathrm{extension}}(i,i-m_b)=0$

$H_b^{\mathrm{extension}}(i,i-m_b+1)=0$

$H_b^{\mathrm{extension}}(i,i-m_b+n_b)=0$

Fill with the element of representing zero square formation other position, as-1;

Satisfy m for any _b+ 2≤i≤m _b+ Δ m line index i has

$H_b^{\mathrm{extension}}(i,i-m_b+1)=0$

$H_b^{\mathrm{extension}}(i,i-m_b+n_b-1)=0$

$H_b^{\mathrm{extension}}(i,i-m_b+n_b)=0$

Fill with the element of representing zero square formation other position, as-1;

Wherein, i is a matrix H _b ^ExtensionLine index, the ranks index is all since 1.

The present invention also provides a kind of code device of supporting the low density parity check code of any code check/code length, channel encoder input information bits sequence is encoded, export the code word bit sequence then and give follow-up processing unit, it is characterized in that, described code device comprises: basis matrix memory, matrix parameter calculator, LDPC shorten code coder, deletion code word bit device, wherein

Described basis matrix memory is used to store unified m _b* n _bBasis matrix H _b ^Uniform, and the output basis matrix is to the matrix parameter calculator;

Described matrix parameter calculator produces LDPC and shortens the needed matrix of code coder and parameter and deletion code word bit device parameters needed;

LDPC shortens code coder, according to matrix and the parameter that described matrix parameter calculator produces, information block encoded, and generated codeword, and output to deletion code word bit device;

Deletion code word bit device is used for deleting the code word bits that shortens the code word some, and the N that will obtain after will deleting code word output.

According to device of the present invention, described matrix parameter calculator comprises relatively selector, first kind calculator, second a class calculator, wherein,

Relatively selector is used for the code check R that comparison actual bit rate R and female sign indicating number collect ₀If, R 〉=R ₀, select operation first kind calculator; If R＜R ₀, move the second class calculator;

First kind calculator is used to produce R 〉=R ₀The time LDPC shorten code coder 604 needed matrixes, spreading factor z;

The second class calculator is used to produce R＜R ₀The time LDPC shorten code coder 604 needed matrixes, spreading factor z.

According to device of the present invention, in described first kind calculator,

At first determine spreading factor $z_t=\frac{K}{k_b},$ Determine z then _tWith element among the Zset following magnitude relationship z is arranged _K-1＜z _t≤ z _k, z wherein _K-1, z _kBe the adjacent element of size, then z among the Zset _kNeed the spreading factor exported exactly;

Based on m ' _b=m _b, n ' _b=n _b, unified basis matrix H _b ^UniformWith spreading factor z _k,, determine H by revising _b ^Modified

With correction matrix H _b ^Modified, spreading factor z _kOutput to LDPC with the matrix size parameter and shorten code coder 604, perhaps to correction matrix H _b ^ModifiedExpand, form parity check matrix H, export parity check matrix H, spreading factor z then _k Shorten code coder 604 with the matrix size parameter to LDPC.

According to device of the present invention, in the described second class calculator,

At first definite

Wherein, Expression rounds up;

Determine z according to specific spreading factor computing formula _t

According to z _tMagnitude relationship z with element among the Zset _K-1＜z _t≤ z _k, determine the spreading factor z that exports _k, wherein, z _K-1, z _kIt is the adjacent element of size among the Zset;

Determine m ' _b=m _b+ Δ m, n ' _b=n _b+ Δ m is to (m _b, n _b) basis matrix H _b ^UniformExpand, at H _b ^UniformLast column and to increase Δ m respectively capable and increase Δ m row in last row back, expansion obtain (m ' _b, n ' _b) basis matrix H _b ^{Uniform_extension}

Based on basis matrix H _b ^{Uniform_extension}With spreading factor z _k, to basis matrix H _b ^{Uniform_extension}Revise, form correction matrix H _b ^{Modified_extension}

With correction matrix H _b ^{Modified_extension}, spreading factor z _kOutput to LDPC with the matrix size parameter and shorten code coder 604, perhaps to correction matrix H _b ^{Modified_extension}Expand, form parity check matrix H, export parity check matrix H, spreading factor z then _k Shorten code coder 604 with the matrix size parameter to LDPC.

According to device of the present invention, described calculating z _tParticular extension factor computing formula be $z_t=\frac{K}{k_b}.$

According to device of the present invention, described calculating z _tParticular extension factor computing formula be $z_t=\frac{N-K}{m_b+\mathrm{\&Delta;m}}.$

According to device of the present invention, described code device further comprises an input information packet buffer, is used for the information block of input is carried out buffer-stored and it is outputed to LDPC shortening code coder 604.

According to device of the present invention, described code device further comprises an output code word packet buffer, is used for buffer-stored and output are carried out in the code word grouping of deletion code word bit device 605 outputs.

According to device of the present invention, described deletion code word bit device 605 is that selection will be at y=m ' _b* z _kCode word bits deletion on-N+K delete position.

According to device of the present invention, described shortening code coder, comprise one (m ' _b* z, n ' _b* z) the female code coder of LDPC, one add zero bit device, one and remove zero bit device, establish input k bit information block and spreading factor z _k, wherein,

One is added zero bit device, is used for adding x=k before the grouping of k bit information _bZ _k-K zero bit obtains k _bThe coding information needed grouping of z bit outputs to the next stage encoder;

One (m ' _b* z, n ' _b* z) the female code coder of LDPC, (n ' _b* z _k, k _b* z _k) the female code coder of LDPC, according to the basis matrix and the spreading factor (perhaps parity matrix) of input above-mentioned coding information needed grouping is encoded, obtain n ' _bZ _kIndividual female sign indicating number code word bits outputs to and removes zero bit device;

Remove zero bit device for one, front x zero bit obtains shortening the sign indicating number code word in female sign indicating number code word of deletion input; Output is shortened the sign indicating number code word to deletion code word bit device, here k ' _b=n ' _b-m ' _b

The present invention also provides the production method of the automatic repeat requests HARQ bag of a kind of mixing of low density parity check code, said method comprising the steps of:

C1), grouping generates the female sign indicating number of LDPC HARQ code word according to input K bit information;

C2), the female sign indicating number of a LDPC HARQ code word bits is arranged, wherein, keep the order of information bit, change the check bit order;

C3), select LDPC HARQ code word bits, by to step 2) output arranges the back code word and sequentially selects bit sequence successively, promptly transmission for the first time is from first systematic bits, the starting position of each transmission is immediately following the position of the last end of transmission later on, the binary sequence of the HARQ bag that finally obtains outputs to next processing unit.

According to the method for device of the present invention, in step C1,

The K bit information grouping of input is that lowest bit rate is R under complete steadily increase redundancy condition _FIRSituation under import, wherein, R _FIRBe given by system, N _FIR=K/R _FIR, for above-mentioned N _FIRFemale sign indicating number code word, the front is a K information bit, the back is M _FIR=N _FIR-K check bit.

According to the method for device of the present invention, in step C1, the female sign indicating number of LDPC HARQ code word generates and may further comprise the steps:

1) determines a specific code check R ₀The female sign indicating number collection of low density parity check code, described female sign indicating number collection is to be R by limited code check ₀The low density parity check code of different code length constitutes, the basis matrix H that described female sign indicating number collection has Unified Form _b ^Uniform, wherein, the random length of channel encoder input is the K information bit sequence, the output random length is a N code word bits sequence, check digit length M=N-K, code check r=K/N;

2) according to the information block length coupling, determine spreading factor; According to described spreading factor and basis matrix H _b ^Uniform,, determine encoder parameters needed and matrix by shortening or deleting surplus or expansion;

3) encoder is encoded to the K bit information grouping of input according to encoder parameters needed and matrix, produces the N bit codewords, the output code word.

According to the method for device of the present invention, in step C2, the female sign indicating number of LDPC HARQ code word bits is arranged and is based on that following formula arranges, with the data of arranging preceding i position be put into arrange after the j position,

Wherein,

Round under the expression, i, j and l are since 0; I is the bit position indices of code word before arranging, and j is the bit position indices of arranging the back code word; P deletes surplus pattern, and size is m _bVector, be by from k _bTo n _b-1 integer constitutes, the predefined arrangement that above-mentioned element is formed.

According to the method for device of the present invention, in step C3, the code word bits selective rule is based on that following formula selects, code word after bit arrangement, select bit, be put into i bit position of k HARQ bag according to the resulting position as a result of described formula

$S_{k,i}=\left\{\begin{array}{cc}\left(i\right)\mathrm{mod}\left(N_{\mathrm{IR}}\right)& k=0\\ (\underset{j=0}{\overset{k-1}{\mathrm{\&Sigma;}}}{L}_j+i)\mathrm{mod}\left(N_{\mathrm{IR}}\right)& k\&NotEqual;0\end{array}\right.$

Wherein, k is the index of HARQ bag, reflects the number of transmissions of current HARQ bag, k=0 when transmitting for the first time, and k=1 when transmitting for the second time adds 1 for transmit k next time;

N _FIRBe the length of female sign indicating number code word;

S _{K, i}The location index of representing k HARQ bag, being used for identifying current HARQ bag index is the bit position of i, the corresponding S of first bit _{K, i}=0 (i=0), wherein i is since 0;

L _kThe length of representing k HARQ bag.

LDPC code coding/decoding apparatus and method provided by the invention have following advantage: the LDPC sign indicating number of different code check different code lengths has unified coder, and the LDPC sign indicating number of different code check different code lengths has a unified R0 basis matrix.The most important thing is, can effectively support IR_HARQ, and fully lowest bit rate can be much smaller than R0 steadily increase redundancy (Full IR) time, can be by being that the unified basis matrix of R0 is expanded (extension) and obtained to code check, the complexity of decoder depends on that code check is the complexity of the LDPC sign indicating number of R0 when R＞R0, and the complexity of decoder depends on that code check is the complexity of the LDPC sign indicating number of R when R＜R0.Surplus method is deleted in existing relatively shortening, has saved hardware cost and computation complexity significantly, has greatly improved the flexibility of code book body, and the scope of application of sign indicating number will obtain greatly expansion.

The invention solves the LDPC sign indicating number aspect variable code rate and the defective supported of steadily increase redundancy HARQ, make the LDPC sign indicating number not fail sign indicating number aspect the flexibility of code check/code length in turbo, and in decoding speed (throughput, the decoding delay), power loss, coding and decoding complexity aspect significantly are better than the turbo sign indicating number, quite (the turbo sign indicating number was dominant when short code was long at aspect of performance and turbo sign indicating number, the LDPC sign indicating number was dominant when long code was long, and medium code length is suitable).Generally speaking, it is an inevitable trend that the LDPC sign indicating number replaces the turbo sign indicating number, and the present invention will be devoted to promote this trend.

Description of drawings

Fig. 1 is the system schematic at apparatus of the present invention place;

Fig. 2 is (n _b* z, k _b* z) the code word of the female sign indicating number of structured LDPC;

Fig. 3 is that code check is less than R ₀The time (N, K) code word of shortening-deletion LDPC sign indicating number;

Fig. 4 is (n _b* z, k _b* z) the code word of the female sign indicating number of structured LDPC;

Fig. 5 is that code check is greater than R ₀The time (N, K) code word of deletion-shortening LDPC sign indicating number;

Fig. 6 is the present invention's (N, K) LDPC code device;

Fig. 7 is the matrix parameter calculator in apparatus of the present invention;

Fig. 8 is that LDPC of the present invention shortens code encoding device;

Fig. 9 is first kind of basis matrix expansion (extension) method that the present invention proposes;

Figure 10 is second kind of basis matrix expansion (extension) method that the present invention proposes;

Embodiment

The present invention is described in detail below in conjunction with drawings and the specific embodiments.

Basic ideas of the present invention are, structured LDPC code collection based on specific code check variable code length, be close to design principle as far as possible with structured LDPC information bit or check digit number as far as possible with actual coding information bit or check digit number, have optimum number of times with the LDPC sign indicating number and be distributed as condition, by expanding (extending), shorten (shortening) or deleting surplus (puncturing) operation, realize the coding and decoding of the low density parity check code of any code length arbitrary bit rate; Realized support, guaranteed that the LDPC sign indicating number has enough low code check under complete steadily increase redundancy merging mode HARQ.

Fig. 1 is the system schematic at apparatus of the present invention place.Code device of the present invention, i.e. channel encoder among Fig. 1.Channel encoder is realized chnnel coding: be exactly the reliability that guarantees communication, it will be encoded to the data after the above-mentioned source encoding, add redundant bit, guarantee receiving terminal, can correction process, and raising communication quality and reliability.Usually said error correcting code or chnnel coding are meant LDPC sign indicating number, turbo sign indicating number, convolution code, RS sign indicating number, BCH code.Source encoder has been realized source encoding, guarantees the validity of communication, and it is converted into binary digital signal with analog signal (as voice, image etc.), the line data of going forward side by side compression (as language compression, image compression etc.).In addition, need to prove that Code And Decode is corresponding, so adopted encoder of the present invention and coding method, its decoder and coding/decoding method also just have corresponding requirement, and the present invention seldom gives unnecessary details.

The invention provides a kind of low density parity check code encoding method of supporting any code check/code length, be used for K information bit of encoder input encoded, produce M check bit, information bit and check bit constitute N code word bits together, output to the next stage processing unit then.Wherein, M=N-K, code check R=K/N, K and N can be arbitrarily greater than zero integer.Described method specifically comprises the steps:

A, specific code check R of setting ₀And the female sign indicating number collection of the low density parity check code of limited code length is used for realizing (N=z * n _b, K=z * k _b) coding and decoding of low density parity check code, wherein, z is a spreading factor, and Zset is the spreading factor set, and z ∈ Zset, Zset are greater than zero positive integer z by P ₁＜z ₂＜...＜z _K-1＜z _k＜...＜z _PThe set that constitutes, P is certain greater than 1 integer.This yard collection has the (m of a Unified Form _b* n _b) basis matrix H _b ^Uniform, in order to realize (z * n _b, z * k _b) coding and decoding, revise unified basis matrix H with specific correction computing and spreading factor z _b ^Uniform, obtain revised basis matrix H _b ^Modified, again according to H _b ^ModifiedJust can expand with z and to obtain parity matrix; And basis matrix H _b ^UniformCode check is always arranged ${\mathrm{<figure-callout\; id="0"\; label="R"\; filenames="A20071007280800392.png,A20071007280800401.png"\; state="\{\{state\}\}">R</figure-callout>}}_0=\frac{(n_b-m_b)}{n_b},$ Wherein, n _bBe definite integer greater than 2, m _bBe definite integer greater than 1, k _b=n _b-m _b

B, as code check R＜R ₀The time, any information block length is K before the order coding, and coding back code word size arbitrarily is N, and z is the spreading factor of structured LDPC code, and the check digit number is M=N-K, code check r=K/N; Be used to finish (N, the coding method of LDPC sign indicating number K) is as described below:

Step B1, code check coupling: calculate

At H _b ^UniformLast column and to increase Δ m respectively capable and increase Δ m row in last row back, expansion obtains (m _b+ Δ m) * (n _b+ Δ m) basis matrix H _b ^{Uniform_extension}Here,

Expression rounds downwards.

Step B2, code length coupling: calculate spreading factor $z_t=\frac{K}{k_b}$ Perhaps $z_t=\frac{N-K}{m_b+\mathrm{\&Delta;m}},$ Determine z _tWith element among the Zset the following z of relation is arranged _K-1＜z _t≤ z _k, wherein, z _K-1, z _kBe the adjacent element of size among the Zset, the required spreading factor of encoder is z _kBased on the unified basis matrix H in expansion (extension) back _b ^{Uniform_extension}With the spreading factor z=z that calculates _k, can obtain ((n _b+ Δ m) * z _k, k _b* z _k) the female sign indicating number of LDPC; Need to prove, calculate z _tRecommend to use $z_t=\frac{K}{k_b},$ Because can prove $\frac{K}{k_b}\&GreaterEqual;\frac{N-K}{m_b+\mathrm{\&Delta;m}};$

Step B3, shortening coding: add x=(k _bZ-K) individual zero bit is before K the information bit, the formation needed k that encodes _bZ _kInformation block; Carry out ((n then _b+ Δ m) * z _k, k _b* z _k) the LDPC coding, obtain (n _b+ Δ m) * z _kIndividual coding codeword bit;

Step B4, symbol deletion: for above-mentioned coding codeword, the x that adds among the deletion step B3 zero bit is not if code word bits number after the deletion and needed grouping size N match, then after deletion zero bit, select correct position, y=(m in the deletion code word _b+ Δ m) * z _k-N+K bit finally obtains the code word that length is N.

Wherein, in above-mentioned steps B1, at H _b ^UniformLast column and to increase Δ m respectively capable and increase Δ m row and form (m in last row back _b+ Δ m) * (n _b+ Δ m) matrix H _b ^{Uniform_extension}Extended method have following two kinds, but the present invention is not limited to this two kinds of methods.These two kinds of methods are respectively:

Method 1: at H _b ^{Uniform_extension}In newly-increased matrix part, satisfy m for any _b+ 1≤i≤m _b+ Δ m line index i has

$H_b^{\mathrm{uniform}\_\mathrm{extension}}(i,i-m_b)=0$

$H_b^{\mathrm{uniform}\_\mathrm{extension}}(i,i-m_b+n_b-1)=0$

$H_b^{\mathrm{uniform}\_\mathrm{extension}}(i,i-m_b+n_b)=0$

Here, the ranks index is all since 1.

As shown in Figure 9, provided an instantiation of method 1.First formula has produced the from left to right null sequence of first one-tenth miter angle, and second formula produced from left to right second null sequence that becomes miter angle, and the 3rd formula produced from left to right the 3rd null sequence that becomes miter angle.Wherein, second null sequence and the 3rd null sequence are adjacent.

Method 2: for H _b ^{Uniform_extension}In newly-increased matrix part, for line index i=m _b+ 1, always have

$H_b^{\mathrm{uniform}\_\mathrm{extension}}(i,i-m_b)=0$

$H_b^{\mathrm{uniform}\_\mathrm{extension}}(i,i-m_b+1)=0$

$H_b^{\mathrm{uniform}\_\mathrm{extension}}(i,i-m_b+n_b)=0$

Satisfy m for any _b+ 2≤i≤m _b+ Δ m line index i has

$H_b^{\mathrm{uniform}\_\mathrm{extension}}(i,i-m_b+1)=0$

$H_b^{\mathrm{uniform}\_\mathrm{extension}}(i,i-m_b+n_b-1)=0$

$H_b^{\mathrm{uniform}\_\mathrm{extension}}(i,i-m_b+n_b)=0$

Here, the ranks index is all since 1.

As shown in figure 10, provided an instantiation of method 2.For line index i=m _b+ 1=5, according to the 1st, 2,3 formula of method 2, can expand unified basis matrix the 5th row the 1st, the 2nd and three elements of the 9th row is 0.For line index i 〉=(m _b+ 2) matrix part=6, three null sequences that become miter angle are from left to right arranged, first null sequence is that the 4th formula of method 2 calculates, and second null sequence is that the 5th formula of method 2 calculates, and the 3rd null sequence is that the 6th formula of method 2 calculates.

It may be noted that for above-mentioned two kinds of expansion methods, in the expansion basis matrix 0 element representation the unit matrix of parity matrix, other positions can be with the filling of the element of zero square formation of representing parity matrix, as-1 in the expansion basis matrix.

C, as code check R 〉=R ₀The time, any information block length is K before the order coding, and coding back code word size arbitrarily is N, and z is the spreading factor of structured LDPC code, and the check digit number is M=N-K; Be used to finish (N, the coding method of LDPC sign indicating number K) comprises:

Step C1, code check coupling: calculate spreading factor $z_t=\frac{K}{k_b},$ Determine z _tWith element among the Zset the following z of relation is arranged _K-1＜z _t≤ z _k, z wherein _K-1, z _kIt is the adjacent element of size among the Zset; Based on unified basis matrix H _b ^UniformWith the spreading factor z that calculates _k, obtain (m _b* z _k, n _b* z _k) the female sign indicating number of LDPC;

Step C2, shortening coding: add x=(k _bZ _k-K) individual zero bit before K the information bit, the formation needed k that encodes _b* z _kInformation block; Carry out (m then _b* z _k, n _b* z _k) the LDPC coding, obtain m _b* z _kIndividual check digit, information block and check digit grouping have constituted n _b* z _kShorten the sign indicating number code word;

Step C3, symbol deletion:,, behind above-mentioned deletion zero bit, continue deletion y=(m the code word if code word bits number after the deletion and needed grouping size N do not match to shortening x zero bit that adds among the sign indicating number code word deletion step C2 _b* z _k-N+K) individual bit finally obtains the code word that length is N.

Below come technology of the present invention as further instruction with a real example.

At first, earlier female sign indicating number collection of low density parity check code is revised, obtained the modified basis matrix.

A specific code check R ₀=1/2, have the female sign indicating number collection of the structurized LDPC sign indicating number of a plurality of code lengths, the unified basis matrix size of this mother's sign indicating number collection is m _b* n _b=16 * 32, the spreading factor of sign indicating number collection is to be that step-length is from z with 1 _Min=12 are increased to z _Max=640, note is made z ∈ Zset={z _Min: 1: z _Max.Because the corresponding particular extension factor z of specific code length N is so information block length is with k _bFor step-length from z _Min* k _bBe increased to z _Max* n _b, note is made K ∈ { z _Min* k _b: k _b: z _Max* k _b192: 16 }={: 10240}.Code check R ₀The unified basis matrix H of=1/2 female sign indicating number collection _b ^UniformAs follows:

$H_b^{\mathrm{uniform}}=\left[H_b^{\mathrm{system}}\right|H_b^{\mathrm{parity}}],$ Wherein,

$H_b^{\mathrm{system}}=$

$H_b^{\mathrm{parity}}=$

In order to realize (z * n _b, z * k _b) coding and decoding of low density parity check code, need with foregoing certain revise computing and spreading factor z revise unified basis matrix H _b ^Uniform, obtain revised basis matrix H _b ^Modified, according to H _b ^ModifiedJust can obtain parity matrix with z.This example has adopted and has rounded (Scale+floor) correction formula and revise unified basis matrix, and correction is to H _b ^UniformThe element (h of expression non-zero square formation _Ij ^b) _UniformCarry out, here z _Max=640:

By to above-mentioned H _b ^UniformRevise, can obtain the modified basis matrix of the low density parity check code correspondence of information block length K=288, code word size N=576, as follows:

$H_b^{\mathrm{mod}\mathrm{ified}}=$

As code check R＜R ₀The time, the female sign indicating number that provides based on embodiment collects, promptly $R_0=\frac{K}{N}=1/2,$ Utilize the LDPC coding method of any length/code check of support of the present invention, details are as follows:

Code check $R=\frac{K}{N}<{\mathrm{<figure-callout\; id="0"\; label="R"\; filenames="A20071007280800392.png,A20071007280800401.png"\; state="\{\{state\}\}">R</figure-callout>}}_0=\frac{1}{2},$ Information block length is K before the coding, and coding back code word size is N, and z is the spreading factor of structured LDPC code, and the check digit number that needs is M=N-K.

According to the low density parity check code encoding method of the above-mentioned any code check/code length of support, finish (step is as follows for N, the coding of LDPC sign indicating number K):

Step B1, code check coupling: calculate

At H _b ^UniformLast column and to increase Δ m respectively capable and increase Δ m row in last row back, expansion obtains (m _b+ Δ m) * (n _b+ Δ m)=(16+ Δ m) basis matrix H * (32+ Δ m) _b ^{Uniform_extension}Here,

Expression rounds up.

Step B2, code length coupling: calculate spreading factor $z_t=\frac{K}{k_b}=\frac{K}{16},$ Determine z _tWith element among the Zset the following z of relation is arranged _K-1＜z _t≤ z _k, wherein, z _K-1, z _kIt is the adjacent element of size among the Zset; z _kIt is exactly the spreading factor that we finally need; Based on the unified basis matrix H in expansion back _b ^{Uniform_extension}With the spreading factor z that calculates _k, can obtain ((n _b+ Δ m) * z _k, k _b* z _k)=((32+ Δ m) * z _k, (16+ Δ m) * z _k) the female sign indicating number of LDPC;

Step B3, shortening coding: add x=(k _bZ _k-K)=(16z _k-K) individual zero bit before K the information bit, the formation needed 16 * z that encodes _kInformation block; Carry out ((n then _b+ Δ m) * z _k, k _b* z _k)=((32+ Δ m) * z _k, (16+ Δ m) * z _k) the LDPC coding, obtain (n _b+ Δ m) * z _k=(32+ Δ m) * z _kBit codewords;

The deletion of step B4, symbol: to x zero bit that adds among female sign indicating number code word deletion step B3, if code word bits number after the deletion and needed grouping size N do not match, then behind deletion zero bit after, select correct position, y=(m in the deletion code word _b+ Δ m) * z _k-N+K=(16+ Δ m) * z _k-N+K bit finally obtains the code word that length is N.

Here provide a more concrete example, follow above-mentioned steps, realize that code check R=3/11 code length is the LDPC coding of K=300 for the N=1100 information block length, method is as follows:

Step 1, code check coupling: calculate At H _b ^UniformLast column and to increase Δ m=27 respectively capable and increase Δ m=27 row in last row back, expansion obtains m ' _b* n ' _b=(m _b+ Δ m) * (n _b+ Δ m)=43 * 59 basis matrix H _b ^{Uniform_extension}, above-mentioned expansion method is to adopt among the step B1 of coding method of the present invention method 2 to carry out;

Step 2, code length coupling: calculate spreading factor $z_t=\frac{K}{k_b}=\frac{K}{16}=18.75;$ Can determine z _tWith element among the Zset the following z of relation is arranged _K-1=18＜z _t≤ z _k=19, z wherein _K-1, z _kBe the adjacent element of size, then z=z among the Zset _k=19 is exactly the spreading factor that encoder needs;

According to z=19, to above-mentioned (m _b+ Δ m) * (n _b+ Δ m)=43 * 59 H _b ^{Uniform_extension}Round (scale+floor) downwards and revise, obtain (m _b+ Δ m) * (n _b+ Δ m)=43 * 59 H _b ^{Modified_extension}According to spreading factor z=19 and (m _b+ Δ m) * (n _b+ Δ m)=43 * 59 H _b ^{Modified_extension}, just can obtain a parity matrix ((m by the basis matrix expansion _b+ Δ m) * z _k, (n _b+ Δ m) * z _kThe female sign indicating number of the LDPC of)=817 * 1121 H.

Step 3, shortening coding: add x=k _bBefore * z-K=16*19-300=4 zero bit to 300 information bit, constitute the needed 16 * z=304 information block of coding;

Based on ((m _b+ Δ m) * z _k, (n _b+ Δ m) * z _kThe H of)=817 * 1121 is that 304 information block is encoded to length, just can obtain length and be (m _b+ Δ m) * z _k=817 verification grouping, information block and verification grouping have constituted length and have been (n _b+ Δ m) * z _k=1121 code words.

Step 4, symbol deletion: remove x=k in the above-mentioned 304 bit informations grouping _b* z-K=4 zero, and last information block length is K=300; Again from above-mentioned (m _b+ Δ m) * z _kSelect in=817 bit verifications groupings y=(m ' _b* z-N+K)=and 43 * 19-1100+300=17 position punches (Puncturing), and terminal check length is M=(m _b+ Δ m) * z _k-y=817-17=800; The verification grouping of above-mentioned K=300 bit information grouping and M=800 bit has constituted the N=1100 bit codewords that we need.

As code check R 〉=R ₀The time, the female sign indicating number that provides based on embodiment collects, promptly $(R=\frac{K}{N})\&GreaterEqual;({\mathrm{<figure-callout\; id="0"\; label="R"\; filenames="A20071007280800392.png,A20071007280800401.png"\; state="\{\{state\}\}">R</figure-callout>}}_0=\frac{1}{2}),$ Utilize the coding method of the LDPC sign indicating number of any length/code check of support of the present invention, details are as follows:

Code check $(R=\frac{K}{N})\&GreaterEqual;({\mathrm{<figure-callout\; id="0"\; label="R"\; filenames="A20071007280800392.png,A20071007280800401.png"\; state="\{\{state\}\}">R</figure-callout>}}_0=\frac{1}{2}),$ Information block length is K before the coding, and coding back code word size is N, and z is the spreading factor of structured LDPC code, and the check digit number that needs is M=N-K.

According to the low density parity check code encoding method of the above-mentioned any code check/code length of support, finish (step is as follows for N, the coding of LDPC sign indicating number K):

Step C1, code check coupling: calculate spreading factor

Based on unified basis matrix H _b ^UniformSpreading factor z with calculating obtains (n _b* z, m _b* z)=(32 * z, the female sign indicating number of 16 * z) LDPC.

Step C2, shortening coding: add x=(k _b* z-K)=(16 * z-K) individual zero bits before K the information bit, the formation needed k that encodes _b* z=16 * z bit information grouping; Carry out (N then _LdpC=n _b* z=32 * z, K _Ldpc=k _b* z=16 * z) LDPC encodes, and obtains n _b* z=32 * z bit codewords.

Step C3, symbol deletion: x zero bit to adding among female sign indicating number code word deletion step C2, if code word bits number and N after the deletion are unequal, continue deletion y=m the code word behind above-mentioned deletion zero bit _b* z-N+K=16 * z-N+K bit finally obtains the code word that length is N.

Require emphasis, the selection of delete position is very important, needs to guarantee the best performance of deletion back sign indicating number.Here, the deleted bit position not only can comprise the check digit part, and can comprise the information bit part.

Here provide a more concrete example, follow above-mentioned steps, realize the LDPC coding of code check R=3/4 code length N=400 information block length K=300, step is as follows:

Step 1, at first calculate spreading factor

Step 2, interpolation x=k _bBefore * z-K=16 * z-K=16*19-300=4 zero bit to 300 information bit, constitute the needed k of coding _b* z=16 * z=304 information block;

According to z=19, to above-mentioned m _b* n _b=16 * 32 H _b ^UniformCarry out above-mentioned rounding (Scale+Floor) correction, obtain m _b* n _b=16 * 32 H _b ^ModifiedAccording to spreading factor z=19 and H _b ^Modified, obtain a parity matrix m by the basis matrix expansion _bZ * n _bThe H of z=304 * 608;

Based on m _bZ * n _bThe H of z=304 * 608 is k to length _bThe information block of z=304 is encoded, and just can obtain length is m _bThe verification grouping of z=304, it is n that information block and verification grouping have constituted length _bThe z=608 code word;

Step 3, remove above-mentioned k _bX=k in the grouping of z=304 bit information _b* z-K=4 zero, and last information block length is K=300; Again from above-mentioned m _bZ _kSelect y=m in=304 bit verifications grouping _b* z-N+K=16 * 19-400+300=204 punches a position (Puncturing), and terminal check length is M=m _bZ-y=304-204=100; The verification grouping of above-mentioned K=300 bit information grouping and M=100 bit has constituted the N=400 bit codewords that we need.

The invention provides a kind of code device of supporting the low density parity check code of any code check/code length, it and above-mentioned example are consistent, be used for the input signal of code device is carried out chnnel coding, be transmitted to demodulator then, described code device comprises: basis matrix memory 601, matrix parameter calculator 602, input information packet buffer 603, LDPC shorten code coder 604, deletion code word bit device 605, and output code word packet buffer 606, as shown in Figure 6.Wherein:

Basis matrix memory 601 is used to store unified m _b* n _bBasis matrix H _b ^Uniform, the unified basis matrix of output is to matrix parameter calculator 602.

Matrix parameter calculator 602, input K, N, k _b, n _bAnd H _b ^Uniform, produce LDPC and shorten code coder 604 needed matrixes and parameter and deletion code word bit device 605 parameters needed.As shown in Figure 7, described matrix parameter calculator 602 comprises comparison selector 701, first kind calculator 702 and the second class calculator 703.The treatment step of described matrix parameter calculator 602 is as described below:

1), the code check R of the current actual coding of input: relatively selector 701 is finished the code check R of code check R=K/N with female sign indicating number ₀=1/2 compares, if R 〉=R ₀, enter first kind calculator 702; Otherwise enter the second class calculator 703.

2), in first kind calculator 702, at first calculate spreading factor $z_t=\frac{K}{k_b},$ Determine z then _tWith element among the Zset following magnitude relationship z is arranged _K-1＜z _t≤ z _k, z wherein _K-1, z _kBe the adjacent element of size, then z among the Zset _kNeed the spreading factor exported exactly; Calculate m ' _b=m _b, n ' _b=n _b, based on unified basis matrix H _b ^UniformWith the spreading factor z that calculates _k,, obtain H by revising _b ^ModifiedWith correction matrix H _b ^Modified, spreading factor z _kOutput to LDPC with the matrix size parameter and shorten code coder 604, perhaps to H _b ^ModifiedExpand, obtain parity check matrix H, export H, spreading factor z then _k Shorten code coder 604 with the matrix size parameter to LDPC;

3), in the second class calculator 703, at first calculate

Calculate then $z_t=\frac{K}{k_b}=\frac{K}{16};$ Determine z _tMagnitude relationship z with element among the Zset _K-1＜z _t≤ z _k, z here _K-1, z _kBe the adjacent element of size, then z among the Zset _kNeed the spreading factor exported exactly; Calculate m ' _b=m _b+ Δ m=16+ Δ m, n ' _b=n _b+ Δ m=32+ Δ m is for (m _b=16, n _b=32) basis matrix H _b ^UniformExpand, at H _b ^UniformLast column and to increase Δ m respectively capable and increase Δ m row in last row back, expansion obtain (m ' _b, n ' _b) basis matrix H _b ^{Uniform_extension}, based on basis matrix H _b ^{Uniform_extension}With the spreading factor z that calculates _k,, obtain H by revising _b ^{Modified_extension}With correction matrix H _b ^{Modified_extension}, spreading factor z _kOutput to LDPC with the matrix size parameter and shorten code coder 604, perhaps to H _b ^{Modified_extension}Expand, obtain parity check matrix H, export H, spreading factor z then _k Shorten code coder 604 with the matrix size parameter to LDPC; Here,

Expression rounds up.

Input information packet buffer 603 is used for the information block of input is carried out buffer-stored and it is outputed to LDPC shortening code coder 604.

LDPC shortens code coder 604, according to basis matrix, spreading factor and matrix size parameter from the needed correction of matrix parameter calculation element 602 input codings, to encoding from the information block of information block buffer input, generated codeword outputs to next stage deletion code word bit device 605.As Fig. 8, LDPC shortens code coder 604 and comprises: interpolation zero bit device 801, (m ' _b* z, n ' _b* z) the female code coder 802 of LDPC, remove zero bit device 803.The treatment step of described LDPC shortening code coder 604 is as described below:

Input k bit information is grouped into and adds zero bit device 801, input spreading factor z _kTo adding zero bit device 801, the female code coder 802 of LDPC and removing zero bit device 803; Before the grouping of k bit information, add x=16z _k-K zero bit obtains female yard information block of 16z bit; Above-mentioned female sign indicating number information block is delivered to (n ' _b* z _k, 16 * z _k) the female code coder 802 of LDPC encodes, and obtains n ' _bZ _kIndividual female sign indicating number code word bits; Remove x zero bit that adds before female sign indicating number coding, obtain shortening code coder coding back data; Output is shortened the sign indicating number code word to deletion code word bit device 605, here k ' _b=n ' _b-m ' _b=k _b=16

Deletion code word bit device 605 is used for input and shortens the sign indicating number code word; Under the condition of performance that guarantees to delete code, select y=m ' _b* z _kThe bit that shortens code word on these positions is deleted in-N+K delete position; With N the code word output that obtains after the deletion.

Output code word packet buffer 606 is used for the code word grouping of deletion code word bit device 605 outputs is carried out buffer-stored, final output.

The present invention also provides a kind of production method of LDPC HARQ bag.In the communication technology, request retransmission HARQ technology comprises the sending method of transmitter and method of reseptance two parts of receiver automatically.In the sending method of transmitter, each transmission all relates to the HARQ bag and generates, the production method of LDPC HARQ of the present invention bag provides the platform of practical application for low density parity check code design flexibly, can mix automatic repeat requests HARQ by fine support steadily increase redundancy.

The production method of LDPC HARQ bag of the present invention may further comprise the steps:

1, grouping generates the female sign indicating number of LDPC HARQ code word (mother codewordgeneration) according to input K bit information

In this step, the grouping of input K bit information, lowest bit rate is R under complete steadily increase redundancy condition _FIR=1/5, calculate N=K/R _FIRIf above-mentioned N value is not an integer, also need to round up or round downwards, obtain N _FIR, above-mentioned N _FIRA female sign indicating number code word comprise: K information bit, and M _FIR=N _FIR-K check bit.

Use the low density parity check code encoding method and the device of any code check/code length of support of the present invention,, produce N input K bit information grouping carrying out LDPC coding _FIRThe bit codewords grouping.

2, the female sign indicating number of LDPC HARQ code word bits is arranged (mother codeword bits permutation)

In this step, the order of information bit will remain unchanged, and need the change check bit in proper order, and purpose is to guarantee that bit is selected to produce the surplus distribution of deleting that code word has optimum in the step 3, guarantees that promptly the bit selection produces the codeword performance optimum.

The size that surplus pattern P is deleted in definition is m _b, it is by from m _bTo n _b-1 integer constitutes, a specific arrangement of above-mentioned element in the vector.For example, H _bSize be m _b* n _b=16 * 32, arrange vectorial P and be defined as P=[17,19,21,23,25,27,29,31,18,24,22,28,30,20,26,16], P _lIndex is the element of l among the vectorial P of expression arrangement, and l is since 0.

If i is for the bit position indices of code word before arranging, in order to realize arrangement mappings, correspondingly arranging back code word bits institute index is j.The data that following formulate will be arranged preceding i position are put into j position, arrangement back:

Wherein, arrange vectorial P index

Round under the expression, i, j and l are since 0.

3, LDPC HARQ code word bits is selected (codeword bits selection)

Code word bits is selected to be used for producing the HARQ bag, has to delete function surplus and that repeat.

It is by to step 2 that code word bits is selected) output arranges the back code word and sequentially selects bit sequence, and the binary sequence of the HARQ bag that obtains outputs to next processing unit (as modulator).

Make that k is the index of HARQ bag, reflect the number of transmissions of current HARQ bag, k=0 when transmitting for the first time, k=1 when transmitting for the second time adds 1 for transmit k next time.

N _FIRBe the length of female sign indicating number code word.

S _{K, i}The location index of representing k HARQ bag, being used for identifying current HARQ bag index is the position of i bit, the corresponding S of first bit _{K, i}=0, wherein index i is since 0.

L _kThe length of representing k HARQ bag.

Following formula has been described the code word bits selective rule, and right formula result of calculation has identified step 2) output code word the bit position.Following formulate code word after bit arrangement, is selected the bit of right formula result of calculation position, is put into i bit position of k HARQ bag.

$S_{k,i}=\left\{\begin{array}{cc}\left(i\right)\mathrm{mod}\left(N_{\mathrm{FIR}}\right)& k=0\\ (\underset{j=0}{\overset{k-1}{\mathrm{\&Sigma;}}}{L}_j+i)\mathrm{mod}\left(N_{\mathrm{FIR}}\right)& k\&NotEqual;0\end{array}\right.$

In a word, the present invention has provided the LDPC code encoding method and the device of any code length/code check of support that uses unified coder.New design of the present invention has solved the fatal shortcoming of LDPC sign indicating number aspect change code length and variable code rate, makes the LDPC sign indicating number not fail the sign indicating number in turbo aspect the flexibility of code check/code length, and has kept excellent properties.The present invention gives a kind of method of HARQ packet generation of LDPC sign indicating number, female sign indicating number code word that the LDPC code encoding method of any code length/code check of above-mentioned support and device can be used for this HARQ packet generation method generates, and has guaranteed that the female sign indicating number of a new LDPC code word can support complete steadily increase redundancy HARQ well.

Here the present invention is described in detail by specific embodiment, provide the description of the foregoing description to make or be suitable for the present invention in order to make those skilled in the art, the various modifications of these embodiment are to understand easily for a person skilled in the art.The present invention is not limited to these examples, or some aspect wherein.Scope of the present invention is elaborated by additional claim.

Claims (26)

1. structured low density parity check code coding method of supporting any code check/code length, random length to the channel encoder input is encoded for the K information bit sequence, exporting random length then is that N code word bits sequence is given follow-up processing unit, check digit length M=N-K, code check r=K/N, described method comprises the steps:

1) determines a specific code check R ₀The female sign indicating number collection of low density parity check code, described female sign indicating number collection is to be R by limited code check ₀The low density parity check code of different code length constitutes, the basis matrix H that described female sign indicating number collection has Unified Form _b ^Uniform

2) according to the information block length coupling, determine spreading factor; According to described spreading factor and basis matrix H _b ^Uniform,, determine encoder parameters needed and matrix by shortening or deleting surplus or expansion;

3) encoder is encoded to the K bit information grouping of input according to encoder parameters needed and matrix, produces the N bit codewords, the output code word.

2. the structured low density parity check code coding method of any code check/code length of support according to claim 1 is characterized in that: in step 1),

The female sign indicating number collection of described low density parity check code has defined P same code rate R ₀(N=z * the n of different code length _b, K=z * k _b) low density parity check code, wherein, Zset is the spreading factor set, and spreading factor z can be any one element among the Zset, and Zset is greater than zero positive integer z by P ₁＜z ₂＜...＜z _K-1＜z _k＜...＜z _PThe set that constitutes, P is certain greater than 1 integer, and basis matrix H _b ^UniformCode check is always arranged $R_0=\frac{(n_b-m_b)}{n_b}$ , wherein, n _bBe definite integer greater than 2, m _bBe definite integer greater than 1, k _b=n _b-m _b

3. the structured low density parity check code coding method of any code check/code length of support according to claim 2 is characterized in that: in step 2) in, determine that relatively code check R is less than R ₀, calculation code desired parameters and matrix comprise:

Step B1, utilization Determine Δ m, at basis matrix H _b ^UniformLast column and to increase Δ m respectively capable and increase Δ m row in last row back, expansion forms (m _b+ Δ m) * (n _b+ Δ m) expansion basis matrix H _b ^{Uniform_extension}Here,

Expression rounds downwards;

Step B2, utilize specific spreading factor formula to determine spreading factor, determine z _tWith element among the Zset the following z of relation is arranged _K-1＜z _t≤ z _k, z wherein _K-1, z _kBe the adjacent element of size among the Zset, the required spreading factor of encoder is z _k

Step B3, based on the unified spread foundation matrix H in expansion back _b ^{Uniform_extension}With spreading factor z=z _k, according to specific correction algorithm to H _b ^{Uniform_extension}Revise, can calculate ((n _b+ Δ m) * z _k, k _b* z _k) the encode basis matrix H of needed correction of LDPC _b ^{Modifed_extension}Here, the basis matrix of the correction needed matrix of encoding exactly, spreading factor and the matrix size parameter parameters needed of encoding exactly.

4. the structured low density parity check code coding method of any code check/code length of support according to claim 2 is characterized in that: in step 2) in, determine that relatively code check R is more than or equal to R ₀, calculation code desired parameters and matrix comprise:

Utilize spreading factor $z_t=\frac{K}{k_b}$ Determine spreading factor z _t, determine z _tWith element among the Zset the following z of relation is arranged _K-1＜z _t≤ z _k, z wherein _K-1, z _kBe the adjacent element of size among the Zset, the required spreading factor of encoder is z _kBased on unified basis matrix H _b ^UniformWith spreading factor z _k, can calculate (m according to specific correction algorithm and spreading factor z _b* z _k, n _b* z _k) the needed revised basis matrix of coding; Wherein, the basis matrix of the described correction needed matrix of encoding exactly, spreading factor and the matrix size parameter parameters needed of encoding exactly.

5. according to the structured low density parity check code coding method of claim 3 or any code check/code length of 4 described supports, it is characterized in that: in step 3), the coding of information bit calculated comprise:

Step B1, interpolation x=k _bZ-K zero bit before K the information bit, the formation needed k that encodes _bZ _kInformation block; Carry out ((n then _b+ Δ m) * z _k, k _b* z _k) the LDPC coding, form (n _b+ Δ m) * z _kIndividual coding codeword bit;

Step B2, in above-mentioned coding codeword, the x that adds among the deletion step B1 zero bit if code word bits number after the deletion and needed grouping size N do not match, then behind deletion zero bit, continues to select correct position, y=(m in the deletion code word _b+ Δ m) * z _k-N+K bit, forming length is the coding codeword of N.

6. the structured low density parity check code coding method of any code check/code length of support according to claim 3 is characterized in that: particular extension factor computing formula is among the step B2: $z_t=\frac{K}{k_b}.$

7. according to claim 3 the structured low density parity check code coding method of supporting any code check/code length, it is characterized in that: particular extension factor computing formula is among the step B2: $z_t=\frac{N-K}{m_b+\mathrm{\&Delta;m}}.$

8. the structured low density parity check code coding method of any code check/code length of support according to claim 3 is characterized in that: in step B1, described basis matrix expansion method is:

For H _b ^ExtensionIn newly-increased matrix part, satisfy m for any _b+ 1≤i≤m _b+ Δ m line index i has

$H_b^{\mathrm{extension}}(i,i-m_b)=0$

$H_b^{\mathrm{extension}}(i,i-m_b+n_b-1)=0$

$H_b^{\mathrm{extension}}(i,i-m_b+n_b)=0$

Fill with the element of representing zero square formation other position, as-1;

Wherein, i is a matrix H _b ^ExtensionLine index, the ranks index is all since 1.

9. the structured low density parity check code coding method of any code check/code length of support according to claim 3 is characterized in that: in step B1, described basis matrix expansion method is:

For H _b ^ExtensionIn newly-increased matrix part, for line index i=m _b+ 1, always have

$H_b^{\mathrm{extension}}(i,i-m_b)=0$

$H_b^{\mathrm{extension}}(i,i-m_b+1)=0$

$H_b^{\mathrm{extension}}(i,i-m_b+n_b)=0$

Fill with the element of representing zero square formation other position, as-1;

Satisfy m for any _b+ 2≤i≤m _b+ Δ m line index i has

$H_b^{\mathrm{extension}}(i,i-m_b+1)=0$

$H_b^{\mathrm{extension}}(i,i-m_b+n_b-1)=0$

$H_b^{\mathrm{extension}}(i,i-m_b+n_b)=0$

Fill with the element of representing zero square formation other position, as-1;

Wherein, i is a matrix H _b ^ExtensionLine index, the ranks index is all since 1.

10. code device of supporting the low density parity check code of any code check/code length, channel encoder input information bits sequence is encoded, export the code word bit sequence then and give follow-up processing unit, it is characterized in that, described code device comprises: basis matrix memory, matrix parameter calculator, LDPC shorten code coder, deletion code word bit device, wherein

Described basis matrix memory is used to store unified m _b* n _bBasis matrix H _b ^Uniform, and the output basis matrix is to the matrix parameter calculator;

Described matrix parameter calculator produces LDPC and shortens the needed matrix of code coder and parameter and deletion code word bit device parameters needed;

LDPC shortens code coder, according to matrix and the parameter that described matrix parameter calculator produces, information block encoded, and generated codeword, and output to deletion code word bit device;

Deletion code word bit device is used for deleting the code word bits that shortens the code word some, and the N that will obtain after will deleting code word output.

11. the code device of the low density parity check code of any code check/code length of support according to claim 10 is characterized in that: described matrix parameter calculator, comprise relatively selector, first kind calculator, second a class calculator, wherein,

Relatively selector is used for the code check R that comparison actual bit rate R and female sign indicating number collect ₀If, R 〉=R ₀, select operation first kind calculator; If R＜R ₀, move the second class calculator;

First kind calculator is used to produce R 〉=R ₀The time LDPC shorten code coder 604 needed matrixes, spreading factor z;

The second class calculator is used to produce R＜R ₀The time LDPC shorten code coder 604 needed matrixes, spreading factor z.

12. the code device of the low density parity check code of any code check/code length of support according to claim 11 is characterized in that: in described first kind calculator,

At first determine spreading factor $z_t=\frac{K}{k_b}$ , determine z then _tWith element among the Zset following magnitude relationship z is arranged _K-1＜z _t≤ z _k, z wherein _K-1, z _kBe the adjacent element of size, then z among the Zset _kNeed the spreading factor exported exactly;

Based on m ' b=m _b, n ' _b=n _b, unified basis matrix H _b ^UniformWith spreading factor z _k,, determine H by revising _b ^Modified

With correction matrix H _b ^Modified, spreading factor z _kOutput to LDPC with the matrix size parameter and shorten code coder 604, perhaps to correction matrix H _b ^ModifiedExpand, form parity check matrix H, export parity check matrix H, spreading factor z then _kShorten code coder 604 with the matrix size parameter to LDPC.

13. the code device of the low density parity check code of any code check/code length of support according to claim 11 is characterized in that: in the described second class calculator,

At first definite

, wherein, Expression rounds up;

Determine z according to specific spreading factor computing formula _t

According to z _tMagnitude relationship z with element among the Zset _K-1＜z _t≤ z _k, determine the spreading factor z that exports _k, wherein, z _K-1, z _kIt is the adjacent element of size among the Zset;

Determine m ' _b=m _b+ Δ m, n ' _b=n _b+ Δ m is to (m _b, n _b) basis matrix H _b ^UniformExpand, at H _b ^UniformLast column and to increase Δ m respectively capable and increase Δ m row in last row back, expansion obtain (m ' _b, n ' _b) basis matrix H _b ^{Uniform_extension}

Based on basis matrix H _b ^{Uniform_extension}With spreading factor z _k, to basis matrix H _b ^{Uniform_extension}Revise, form correction matrix H _b ^{Modified_extension}

With correction matrix H _b ^{Modified_estension}, spreading factor z _kOutput to LDPC with the matrix size parameter and shorten code coder 604, perhaps to correction matrix H _b ^{Modified_extension}Expand, form parity check matrix H, export parity check matrix H, spreading factor z then _kShorten code coder 604 with the matrix size parameter to LDPC.

14. the code device of the low density parity check code of any code check/code length of support according to claim 13 is characterized in that: described calculating z _tParticular extension factor computing formula be $z_t=\frac{K}{k_b}.$

15. the code device of the low density parity check code of any code check/code length of support according to claim 13 is characterized in that: described calculating z _tParticular extension factor computing formula be $z_t=\frac{N-K}{m_b+\mathrm{\&Delta;m}}.$

16. the code device of the low density parity check code of any code check/code length of support according to claim 10, it is characterized in that, described code device further comprises an input information packet buffer, is used for the information block of input is carried out buffer-stored and it is outputed to LDPC shortening code coder 604.

17. the code device of the low density parity check code of any code check/code length of support according to claim 10, it is characterized in that, described code device further comprises an output code word packet buffer, is used for buffer-stored and output are carried out in the code word grouping of deletion code word bit device 605 outputs.

18. the code device of the low density parity check code of any code check/code length of support according to claim 10 is characterized in that, described deletion code word bit device 605 is that selection will be at y=m ' _b* z _kCode word bits deletion on-N+K delete position.

19. the code device of the low density parity check code of any code check/code length of support according to claim 10 is characterized in that: described shortening code coder, comprise one (m ' _b* z, n ' _b* z) the female code coder of LDPC, one add zero bit device, one and remove zero bit device, establish input k bit information block and spreading factor z _k, wherein,

One is added zero bit device, is used for adding x=k before the grouping of k bit information _bZ _k-K zero bit obtains k _bThe coding information needed grouping of z bit outputs to the next stage encoder;

One (m ' _b* z, n ' _b* z) the female code coder of LDPC, (n ' _b* z _k, k _b* z _k) the female code coder of LDPC, according to the basis matrix and the spreading factor (perhaps parity matrix) of input above-mentioned coding information needed grouping is encoded, obtain n ' _bZ _kIndividual female sign indicating number code word bits outputs to and removes zero bit device;

Remove zero bit device for one, front x zero bit obtains shortening the sign indicating number code word in female sign indicating number code word of deletion input; Output is shortened the sign indicating number code word to deletion code word bit device, here k ' _b=n ' _b-m ' _b

20. an application rights requires 1 or the coding method of claim 10 or the method for the automatic repeat requests HARQ bag of mixing that device produces low density parity check code, said method comprising the steps of:

C1), grouping generates the female sign indicating number of LDPC HARQ code word according to input K bit information;

C2), the female sign indicating number of a LDPC HARQ code word bits is arranged, wherein, keep the order of information bit, change the check bit order;

C3), select LDPC HARQ code word bits, by to step 2) output arranges the back code word and sequentially selects bit sequence successively, promptly transmission for the first time is from first systematic bits, the starting position of each transmission is immediately following the position of the last end of transmission later on, the binary sequence of the HARQ bag that finally obtains outputs to next processing unit.

21. method according to claim 20 is characterized in that, in step C1,

The K bit information grouping of input is that lowest bit rate is R under complete steadily increase redundancy condition _FIRSituation under import, wherein, R _FIRBe given by system, N _FIR=K/R _FIR, for above-mentioned N _FIRFemale sign indicating number code word, the front is a K information bit, the back is M _FIR=N _FIR-K check bit.

22. method according to claim 20 is characterized in that, in step C1, the female sign indicating number of LDPC HARQ code word generates and may further comprise the steps:

1) determines a specific code check R ₀The female sign indicating number collection of low density parity check code, described female sign indicating number collection is to be R by limited code check ₀The low density parity check code of different code length constitutes, the basis matrix H that described female sign indicating number collection has Unified Form _b ^Uniform, wherein, the random length of channel encoder input is the K information bit sequence, the output random length is a N code word bits sequence, check digit length M=N-K, code check r=K/N;

2) according to the information block length coupling, determine spreading factor; According to described spreading factor and basis matrix H _b ^Uniform,, determine encoder parameters needed and matrix by shortening or deleting surplus or expansion;

3) encoder is encoded to the K bit information grouping of input according to encoder parameters needed and matrix, produces the N bit codewords, the output code word.

23. method according to claim 20 is characterized in that, in step C2, the female sign indicating number of LDPC HARQ code word bits is arranged and is based on that following formula arranges, with the data of arranging preceding i position be put into arrange after the j position,

Wherein, Round under the expression, i, j and 1 are since 0; I is the bit position indices of code word before arranging, and j is the bit position indices of arranging the back code word; P deletes surplus pattern, and size is m _bVector, be by from k _bTo n _b-1 integer constitutes, the predefined arrangement that above-mentioned element is formed.

24. method according to claim 20, it is characterized in that, in step C3, the code word bits selective rule is based on that following formula selects, code word after bit arrangement, select bit, be put into i bit position of k HARQ bag according to the resulting position as a result of described formula

$S_{k,i}=\left\{\begin{array}{cc}\left(i\right)\mathrm{mod}\left(N_{\mathrm{IR}}\right)& k=0\\ (\underset{j=0}{\overset{k-1}{\mathrm{\&Sigma;}}}{L}_j+i)\mathrm{mod}\left(N_{\mathrm{IR}}\right)& k\&NotEqual;0\end{array}\right.$

Wherein, k is the index of HARQ bag, reflects the number of transmissions of current HARQ bag, k=0 when transmitting for the first time, and k=1 when transmitting for the second time adds 1 for transmit k next time;

N _FIRBe the length of female sign indicating number code word;

S _{K, i}The location index of representing k HARQ bag, being used for identifying current HARQ bag index is the bit position of i, the corresponding S of first bit _{K, i}=0 (i=0), wherein i is since 0;

L _kThe length of representing k HARQ bag.

25. the basis matrix expansion method of a structured LDPC code, described basis matrix H _bSize be m _b* n _b, at H _bLast column and to increase Δ m respectively capable and increase Δ m row in last row back, form (m _b+ Δ m) * (n _b+ Δ m) matrix H _b ^Extension, described method comprises:

For H _b ^ExtensionIn newly-increased matrix part, satisfy m for any _b+ 1≤i≤m _b+ Δ m line index i has

$H_b^{\mathrm{extension}}(i,i-m_b)=0$

$H_b^{\mathrm{extension}}(i,i-m_b+n_b-1)=0$

$H_b^{\mathrm{extension}}(i,i-m_b+n_b)=0$

Here, the ranks index is all since 1;

Fill with the element of representing zero square formation other position, as-1.

26. the basis matrix expansion method of a structured LDPC code, described basis matrix H _bSize be m _b* n _b, at H _bLast column and to increase Δ m respectively capable and increase Δ m row in last row back, form (m _b+ Δ m) * (n _b+ Δ m) matrix H _b ^Extension, described method comprises:

For H _b ^ExtensionIn newly-increased matrix part, for line index i=m _b+ 1, always have

$H_b^{\mathrm{extension}}(i,i-m_b)=0$

$H_b^{\mathrm{extension}}(i.i-m_b+1)=0$

$H_b^{\mathrm{extension}}(i,i-m_b+n_b)=0$

Fill with the element of representing zero square formation other position, as-1;

Satisfy m for any _b+ 2≤i≤m _b+ Δ m line index i has

$H_b^{\mathrm{extension}}(i,i-m_b+1)=0,H_b^{\mathrm{extension}}(m_b+i,i+1)=0$

$H_b^{\mathrm{extension}}(i,i-m_b+n_b-1)=0$

$H_b^{\mathrm{extension}}(i,i-m_b+n_b)=0,H_b^{\mathrm{extension}}(m_b+i,n_b+i)=0$

Wherein, i is a matrix H _b ^ExtensionLine index, the ranks index is all since 1;

Fill with the element of representing zero square formation other position, as-1.

Images