WO2008086735A1

WO2008086735A1 - Method for generating hybrid automatic request retransmission packet of low density parity check code

Info

Publication number: WO2008086735A1
Application number: PCT/CN2008/000078
Authority: WO
Inventors: Jun Xu; Zhifeng Yuan; Liujun Hu
Original assignee: Zte Corporation
Priority date: 2007-01-12
Filing date: 2008-01-11
Publication date: 2008-07-24
Also published as: CN101005334A; CN101005334B

Abstract

A method for generating hybrid automatic request retransmission packet of Low Density Parity Check code includes: A1) Low Density Parity Check code coder generates NFIR bit LDPC HARQ mother code codeword according to input K bit information packets; the codeword comprises bit information packets, extending check bit packets and puncturing check bit packets; the generated codeword is transmitted to the HARQ buffer (1); A2) the LDPC HARQ mother code codeword bit is rearranged in the HARQ buffer, the order of the information bit and the extending check bit is kept not changing, and the order of the puncturing check bit is changed (2); A3) the codeword bit from the rearranged HARQ mother code codeword is orderly selected; that is to say, the first transmission is started at the first system bit; then the starting position of every transmission follows the ending position of last transmission, in order to generate binary system sequence of the HARQ packet (3).

Description

Hybrid automatic request retransmission packet generation method for low density parity check code

Technical field

The present invention relates to the field of digital communications, and more particularly to a hybrid automatic request retransmission or HARQ packet generation method for low density parity check codes.

Background technique

Figure 1 is a typical digital communication system in which all digital communication systems require a channel coder. The channel encoder is designed to combat a wide variety of noises and disturbances during transmission. By artificially adding redundant information, the system has the ability to automatically correct errors, thus ensuring the reliability of digital transmission. Low-density parity check codes are a class of linear block codes that can be defined with very sparse parity check matrices or bipartite graphs, originally discovered by Gallager, so called Gallager codes. After decades of silence, with the development of computer hardware and related theories, MacKay and Neal rediscovered it and proved its ability to approach the Shannon limit. Recent studies have shown that low-density parity codes have the following characteristics: low decoding complexity, linear time coding, performance close to Shannon's limit, parallel decoding, and better performance than Turbo codes under long code length conditions.

Low-density parity check code (LDPC) is a linear block code based on sparse check matrix. It is the sparsity of its check matrix that can realize low-complexity coding and decoding. Therefore, the LDPC code is put into practical use. The aforementioned Gallager code is a regular LDPC code (regular LDPCC), and Luby and Mitzemnacher et al. generalize the Gallager code and propose a non-regular LDPC code (irregular LDPCC). LDPC codes have many decoding algorithms. Among them, Message Passing algorithm or Belief Propagation algorithm (BP algorithm) is the mainstream and basic algorithm of LDPC codes. At present, there are many improved effective decoding algorithms. .

The graphical representation of the LDPC parity check matrix is a bipartite graph. The bipartite graph and the check matrix have a corresponding relationship, and an M*N parity check matrix H defines a constraint that each N-bit codeword satisfies M parity sets. A bipartite graph includes N variable nodes and M parity nodes. When the mth check involves the nth bit, that is, the element H _m of the mth row and the nth column in H, _n = l, there will be a wire connecting the check node m and the variable node n. In the bipartite graph, There will be no connections between any nodes of the same class, and the total number of edges in the bipartite graph is equal to the number of non-zero elements in the check matrix.

A special type of LDPC code has become a mainstream application due to its structured features. Let the parity check matrix H of this LDPC code be a (Mxz)x(Nxz) matrix, which is composed of MxN block matrices, each block matrix is a different power of the basic permutation matrix of ζχζ, basic permutation When the matrix is a unit matrix, they are the cyclic shift matrix of the unit matrix (the default is right shift in the text).

The check matrix H has the following form:

The basic permutation matrix has the following form

The basic matrix element of the block square matrix means: If it is not an integer greater than or equal to zero, there is = 0, which can be represented by a negative one. If it is equal to 0, P is a unit matrix of zxz. If ⁷ ^ is an integer greater than or equal to 0, define ⁵ = ( ^pj , where P is a basic permutation matrix of ZXZ. By such a power, each block matrix can be uniquely identified, and the power of the identity matrix can be used. It is shown that the power of the zero matrix is generally represented by -1. Thus, if each block matrix of H is replaced by its power, a " ^x? 3⁄4 power matrix is obtained. Here, the definition is H. The base matrix, H is called the extension matrix. In actual coding, z = code length / number of columns of the base cell matrix n _b , called the spreading factor.

For example, the matrix H can be extended with the following expansion factor z=3 and a cornerstone out matrix 3⁄4 of m ₆ x = 2x 4:

0 1 0 - 1

z 2 1 2 1

If the LDPC code uses a basic matrix for each different spreading factor, then for each different code length, the LDPC code codec needs to store a basic matrix. When the code length is large, Store a lot of ^ matrix, which will cause representation and storage problems. Therefore, when it is necessary to implement the variable code length, a low-density parity check code of a plurality of code lengths within a certain range of the same code rate uses a basic matrix of a form, which is defined as a unified basic matrix Hr. When the code length is different, if the Hr is modified and extended, the parity check matrix Η can be obtained, so that the generated codec can be applied to the case where the code length is variable.

The correction is to use a spreading factor of different code lengths to correct the non-negative values in the basic matrix. The corrected element value should be smaller than the spreading factor value under the code length. There are many correction algorithms. For example, you can use mod, scale+floor, or scale+round to set the non-negative 1 element of the i-th row and j-th column of the base matrix. Correct the non-negative 1 element of the i-th row and j-th column of ^m ° ^d afterwards, with:

For the mod method:

, N

≡ P _v mod z≡ P _y mod―

Nb

For the rounding/floor method:

For rounding (round+round) methods: P' = Round

Where z is the expansion factor corresponding to the current code length, that is, the number of rows or columns of the block square matrix; ² is the expansion factor corresponding to the maximum supported code length. Mod is the modulo operation, L" is the bottom rounding operation, and Round is the rounding operation.

For example, for a low-density parity check code with a code length of 1152 bits, set a non-negative element of the base matrix to 93, a maximum supported code length of 2304, and a base matrix size of 12*24. The corrected result is:

For the mod method:

1 152

93mod^^ = 93 mod 48 = 45

twenty four

For the rounding (scale+f loor) method:

1152

93 χ- = |_46·5" = 46

2304- For the rounding (scale+round) method:

1152

Round(93 x^^-) = Round(46.5) = 47 This low-density parity check code with a variable code length of a specific code rate can use an encoder/decoder with the same form of the base matrix. .

Variable Rate The construction of the LDPC code guarantees the flexibility of the LDPC code to meet the requirements of the HARQ implementation. In order to construct a variable bit rate LDPC code, two methods are required:

Traditional puncturing methods and special extension methods. Studies have shown that puncturing applies to high code rates (eg, code rates greater than 1/2-2/3), and expansion is appropriate for low bit rates (eg, code rates less than 1/2-2/3). The combination of puncturing and expansion provides a low-density parity-check code structure with a low-to-high code rate, as shown in Figure 2, which provides near-optimal performance.

The following explains "puncturing": puncturing of BCH codes, convolutional codes, and turbo codes can be widely applied to obtain a flexible code rate. It is also possible to apply puncturing to the LDPC code. By appropriate puncturing, a series of high code rate codes can be obtained by removing (puncturing) the low code rate and some bits of the mother code code word. The encoder generates a complete set of parity face bits, but some bits are not sent (punctured). Decoder insertion wipe In addition to the location (erasures) to the punctured parity bits and then decoding code rate R _m. It should be noted that erasure is generally represented by zero, and insertion erasure is to place a zero at the position of the punctured parity bit. The variable nodes of those punctured locations can be considered as erased. Here, we define the removed codeword bits that satisfy the above description as "punctured parity bits", as shown in Figure 2.

The following explains "expansion": unlike puncturing, expansion is to construct a low-density parity check code with variable code rate from high bit rate to low bit rate by adding more parity. For a certain bit rate

LDPC code of R _m =, adding Δ parity, then the (M, N) parity of the above LDPC code

- increase the Δ row and increase the Δ column on the matrix, then the matrix size after expansion is

(Μ + ΑΜ) χ (Ν + ΑΜ) , the code rate of the extended code is changed to T' ⁱ . " = ~ - ~ , by adjusting Δ

Ν+ ΔΜ

A series of low code rate codes are obtained, where M = N - K. Here, we define an "expanded check bit" that satisfies the above-described increase of parity bits, as shown in FIG. In FIG. 2, the code defining the above code rate R _m is "base code", and the punctured mother code code word can obtain a low-density parity check code with a high code rate, and the mother code code word can be expanded. The low-density parity check code with low code rate, the combination of puncturing and expansion guarantees that the designed low-density parity check code has a sufficiently flexible code rate and ensures sufficient performance, and supports incremental redundant HARQ. The compiler is guaranteed to have as little coding error as possible. Hybrid Automatic Request Retransmission HARQ is an extremely important link adaptation technique in wireless communication systems, which is a combination of coding (FEC) technology and automatic request retransmission (ARQ) technology.

The following is a brief introduction to the hybrid automatic request retransmission HARQ method, which includes the transmission method in the transmitter and the reception method in the receiver. The method should be universal.

The sending method of the transmitter (Transmitter) includes:

a. The transmitter sends the first HARQ packet on the designated HARQ channel, a new HARQ transmission starts, and the retransmission times counter is set to 1;

b. The transmitter waits for the acknowledgement/negation sent by the receiver (ACK/NAK);

c. After a certain delay (depending on the synchronous mode or the asynchronous mode, the delay is determined by the system, etc.), if the transmitter receives the ACK, the packet has been correctly received, and the HARQ transmission ends; No Bay ij, retransmission count counter Add 1 to determine whether the number of retransmissions exceeds the maximum number of allowed transmissions. If it exceeds, the current HARQ transmission ends. If not, go to step; d. The transmitter makes another HARQ attempt again on the HARQ channel of the current HARQ transmission, and the transmission time can be selected by the system, and the process goes to step b-.

The receiving method of the receiver (Reiciver) includes:

a. Receive the HARQ packet, confirm whether the received packet is a HARQ attempt, and if yes, go to step b.; otherwise, go to step d.;

b. Discard all previously received HARQ attempts under this HARQ channel;

c. Decode the packet and go to step e.;

d. The receiver decoder combines the original erroneous and retransmitted coded packets according to a specific combining method; e. If correctly decoded, after a certain delay, sends an ACK in the feedback channel, which channel is allocated by the system; otherwise, sends a NAK , store this HARQ package.

In step d of the above receiving method, the merged method may include full incremental redundancy, partial incremental redundancy, and Chase combining.

The hybrid automatic request retransmission HARQ implementation structure, a large number of documents and books are introduced. Regardless of the method or structure, the generation of HARQ packets is critical and determines performance and complexity.

Since HARQ is a very important link adaptation technique, a good low density parity check code must support the implementation of HARQ, which is also the most important object of the present invention. The present invention will provide a HARQ packet generation method based on the low density parity check code design of the present invention. Summary of the invention

The technical problem to be solved by the present invention is to provide a low density parity check code HARQ packet generation method.

To achieve the above object, the present invention provides a hybrid automatic request retransmission packet generation method for a low density parity check code, the method comprising the following steps:

A1), the low density parity check code encoder generates an N _FIR bit LDPC HARQ mother code code word according to the input bit information group, the code word including a bit information packet, an extended parity bit packet, and a punctured parity bit packet, The generated codeword is sent to the HARQ buffer;

A2), rearrange the LDPC HARQ mother code codeword bits in the HARQ buffer, and maintain the information ratio The order of the expansion check bits is unchanged, and the order of the punctured check bits is changed;

A3), sequentially selecting codeword bits from the rearranged HARQ mother code codeword, that is, the first transmission starts from the first systematic bit, and then the start position of each transmission follows the end of the previous transmission, generating The binary sequence of the HARQ packet. According to the method of the present invention, in step A1, the low density parity check code has a code rate basic code, and the basic code is uniquely described by a base matrix m ₆ x H _3⁄4 and an expansion factor z; the punctured basic code obtains a code higher than the code, the parity bits need to substantially code codeword puncturing; expansion rate base code to obtain code below, the increase in m _b rows above the last row and last _Η έ and an increased

Am _b column, obtain a low density parity check code with a code rate of R _F » ~^ ""

m _b + Am _b

(m _b +Am _b )x (n _b + Δ»3⁄4 ) maximum expansion base matrix.

According to the method of the present invention, in step A1, the LDPC HARQ mother code codeword generating method comprises the following steps:

a) The system provides the maximum expansion basis matrix required for coding ("3⁄4 + Am _b ^max ) x (n _b + Am _b ^mm )

H ^sion (Am _b ^max ), spreading factor Z and basic matrix size parameters k _b , n _b + Am _b ;

b) If the input information group iT<^z, it is also necessary to fill a known bit (such as zero) to form a message group;

C) The encoder is based on '. "(Am ₆ ^max ), spreading factor z, encoding the input ^ bit information packet, generating

Bit codeword

d) If /^ has padding bits, remove the padding bits, and finally generate the code

N _FIR = K + (m _b + Am _3⁄4 ^max ) z bit codeword.

According to the method of the present invention, in step A2, the LDPC HARQ mother code codeword bit arrangement is performed based on the following formula, and the data of the i-th position before the arrangement is placed at the jth position after the arrangement.

^i≥m _h 'Z

κ other

Where L " represents the next rounding, i, j and 1 are all starting from o; i is the bit position index of the pre-arranged code words, j is the bit position index of the aligned code words, ≤ i, j < K + (m _b +Am _b )z,' x is the number of padding bits. P is a puncturing pattern vector with "¾ elements, is from to" ⁶ constituting a integer of ^1, is punctured index vector p 1 elements. According to the method of the present invention, the selection pattern P is produced by pressing - v _mt _ ₂ →... → P. The order produced. The known {^,.,., ^}, generation method includes the following steps:

Select an element in the set Re mainSet = ,... , " _δ - 1} / . ₊₁ · · , }, and the remaining elements form a new set Re mainSet\{j);

Determine the part of the check matrix corresponding to Remi»3⁄43⁄4tl{_/} as H^Z), and determine H _∞ , /) the minimum line weight and weight are ^=( ) lines ^= ( ;

, then determine the element to get ^

According to the method of the present invention, in step A3, the codeword bit selection rule is selected based on the following formula, and the bit of the selected result position is placed in the i-th bit position of the kth HARQ packet.

S _kJ = ( ^L _t o _tal (kr) + i) mod(N )

Where k is the index of the HARQ packet, reflecting the number of transmissions of the current HARQ packet, k=0 for the first transmission, k=l for the second transmission, and 1 for the next transmission k;

K-l

(&) is the sum of the lengths of the first k HARQ packets: L _total {k)^L Let 1^. , (- 1) = 0

7=0

Is the length of the HARQ mother code codeword; '' indicates the position index of the kth HARQ packet, which is used to identify the bit position of the current HARQ packet with the index i, and the first bit corresponds. '· ⁼⁰ (i=0 ) , where i starts at 0;

Indicates the length of the kth HARQ packet. The present invention also provides a hybrid automatic request retransmission packet generation method for another low density parity check code, the method comprising the following steps:

B1), the low density parity check code encoder generates an N _FIR bit LDPC HARQ mother code code word according to the input bit information group, the code word includes a bit information packet, an extended check bit packet, and the deletion The remaining check bits are parsed, and the generated codeword is sent to the HARQ buffer;

B2), sequentially selecting an LDPC HARQ codeword bit sequence from the HARQ buffer, that is, the first transmission starts from the first system bit, and the start position of each transmission follows the end of the previous transmission, and is generated. The binary sequence of the HARQ packet. The method according to the present invention, in the step B1, the low density parity check code with a code rate R _m basic code, the base matrix with the basic code and the expansion factor z H _fc unique description; substantially punctured code codeword obtained A code whose code rate is higher than R „, requires puncturing of parity bits of the basic code code word; expanding the basic code to obtain a code having a lower code rate, and adding A ₆ lines in the last row and the last column of the above H _fc Add Δ«3⁄4 column to get m _b + Am _b of low density parity check code with code rate R _FI » -1

(m _b +Am _b )x (n _b + Δ»3⁄4 ) maximum expansion basis matrix. According to the method of the present invention, in step B1, the LDPC HARQ mother code codeword generating method comprises the following steps:

a) The system will provide the maximum expansion basis matrix (m ₆ +Amr ^x )x(n _b +Am, ^max ) required for coding

_H , «« _(Ao3⁄4 max ₎ , the spreading factor z and the base matrix size parameter, n _b +Am _b ; b) If the input information packet <^ _z , it is also necessary to fill the known bits (such as zero) to form the k _b z information packet ;

C) the encoder encodes the input bit information packet according to 11: - '^ ^), the spreading factor Z, and generates (+A ₆ ^max bit codeword;

d) If k _bZ has padding bits, remove the padding bits, and finally generate the code

N _FIR =K + {m _b + Am _b ^maii )z 匕 codeword.

According to the method of the present invention, in step B2, the codeword bit selection rule is selected based on the following formula, and the bit of the selected result position is placed in the i-th bit position of the kth HARQ packet.

S _kJ = (L _lota! (kl) + i) mod(N _m )

Where k is the index of the HARQ packet, reflecting the number of transmissions of the current HARQ packet, k-0 for the first transmission, k=l for the second transmission, and 1 for the next transmission k;

A. _To , W is the sum of the lengths of the first k HARQ packets: _{toto /} W = ∑ Z. ₅ set ^ (- 1) = 0 N The length of the HARQ mother code code word; , ' indicates the position index of the kth HARQ packet, which is used to identify the bit position of the current HARQ packet with index i, the first bit corresponds to ^{= 0} (i=0), where i starts from 0;

Indicates the length of the kth HARQ packet. The present invention also provides a hybrid automatic request retransmission packet generation method for a low density parity check code, the method comprising the following steps:

C1), the low density parity check code encoder generates an NTM bit LDPC HARQ mother code codeword according to the input bit information group, the codeword includes a bit information packet, an extended check bit packet, and a punctured check bit packet, and generates The codeword is sent to the HARQ buffer;

C2), selecting an LDPC HARQ codeword bit from the HARQ buffer according to the following formula, and placing the bit of the selected result position into the ith bit position of the kth HARQ packet to generate a HARQ packet,

(i) mod(N _m ) i ≤ K _i ≥ m _b -z

i + x

(L _total (A: 1) + P, xz + (( + x) mod z ) mod (N _IR ) and /: k≠0 ₉ m _h -z>i> K

Where k is the index of the HARQ packet, reflecting the number of transmissions of the current HARQ packet, k=0 for the first transmission, k=l for the second transmission, plus 1 for the next transmission k; L” means rounding down, Both i and 1 start from 0, i is the bit position index of the pre-arranged codeword; (^ + i^+A Jz; X is the number of padding bits.

A. _To , W is the sum of the lengths of ffk HARQ packets: 0

NFIR is the length of the HARQ mother code codeword; 'is the position index of the kth HARQ packet, which is used to identify the bit position of the current HARQ packet index i, the first bit corresponds, ' ⁼⁰ ( i=0) , where i starts at 0;

Indicates the length of the kth HARQ packet;

P is a punctured pattern vector having ^ elements, which is composed of an integer from "6-1", and _P/ is an index of the punctured vector P as one element. According to the method of the present invention, the selection pattern p is generated according to , _→ P _W _ ₂ _>... _→ ρ. The order produced. Known ^...^^^), the production method includes the following steps:

Select an element in the set Re wK^&t = -\}l{P _M ,-,P _ni }, and the remaining elements form a new set Re mainSet\{j);

Determine the part of the check matrix corresponding to Re azT^tll/} as H _∞ , C) , and determine the minimum line weight and weight of H _/ ( ) as the number of rows W = /);

, then determine the element to get.

According to the method of the present invention, in step C1, the low-density parity check code has a code rate R _m basic code, and the basic code can be uniquely described by a basic matrix "3⁄4>^ H ₆ and an expansion factor z; The basic code code word obtains a code with a higher code rate, and needs to _punctrate the check bits of the basic code code word; expand the basic code to obtain a code with a lower code rate, and add m _b in the last row and the last column of the above Η Line and increase the Am ₆ column to obtain a low density parity check code with a code rate of _Λ(Δ3⁄4 ) = ^^

m _b + Am _b

(m _b +Am _b )x (n _b + Δ»3⁄4 ) maximum expansion basal exit matrix. According to the method of the present invention, in step C1, the LDPC HARQ mother code codeword generating method comprises the following steps:

a) The system will provide the maximum expansion basis matrix required for coding (m ₆ +Am _3⁄4 ^max )x(3⁄4 +Am ₆ ^max ) °" (Am _3⁄4 ^max ) expansion factor Z and base matrix size parameters k _b , n _b + Am _b ;

b) If the input information group ^: < _z , it is also necessary to fill a known bit (such as zero) to constitute / ^ information packet;

C) The encoder encodes the input k _b z bit information packet according to W _b ^xtension (Am ₆ ^max ), spreading factor Z, and generates (3⁄4 +

Bit codeword

d) If k _b z has padding bits, the padding bits are removed and finally encoded to generate N _FIR =K + (m _b + Am _b ^mm ) zt 码 codewords.

The present invention also provides a low density HARQ mother code codeword generation method comprising the following steps: a) The system will provide the maximum expansion basis matrix required for coding („3⁄4 + Am ) x(n _b + Am ^x ) ΗΓ ^∞ '"(Δ ^), expansion factor ζ and base matrix size parameter, n _b +Am _b ; b) if the input information packet < Z, i - needs to be filled with a known bit (such as zero) to constitute / ^ information packet;

C) Encoder > according to w _b ^xtensh " (Am^ ) Z, encoding the input bit information packet to generate a 0 ₆ + Am, ^max ) z bit code word;

d) If k _bZ has padding bits, the padding bits are removed and finally encoded to generate N _FIR =K + (m _b + Am _fc ^max )z bit codewords. The present invention also provides another low-density HARQ mother code codeword generation method, comprising the following steps: a) According to the information packet length matching principle, the system will provide the maximum expansion basis matrix required for coding ("^+Δη^Χ +Δ ^ν^) U ' ^ension (Am _b ^mm ) , expansion factor Ζ and basis matrix size parameters k _b , n _b +Δ»3⁄4;

b) If the input information group < ζ, it is also necessary to fill a known bit (such as zero) to form an information packet;

C) According to the size of the encoder, the weight is divided into three cases:

a) If Α^≤« ₆ ·ζ, the encoder will select ^'°"(0) from the maximum expansion base matrix, ie

Η,; then _encode the bit information packet according to Η _έ , ζ to generate a N^=.z bit codeword; wherein, A. _To , W is the sum of the lengths of the first k HARQ packets: 0;

(2) If Ζ^, ≥( ₊ Δ Γ).ζ, the encoder encodes the ζ bit information group according to the maximum expansion base matrix Jl ^temion (Am _b ^mm ), Z, and generates N _{∞; r} = (n _b +Am ₆ ^max ). Bit codeword;

(3) In addition to the above cases (1) and (2), calculate the Am ₆ = ^toial ~ ^Z encoder

Ζ

It will be selected from the maximum expansion base matrix and then based on . "(△/3⁄4), z encodes the ^^ bit information packet to generate a (/3⁄4+Am).z bit codeword, puncturing from the above codeword (3⁄4 + Δ»3⁄4 - , ., bits) , finally get a codeword of length = ^, d) If there is a padding bit, remove the known padding bits added by b).

The method of the invention systematically solves the defects of incremental redundant HARQ support of LDPC codes, and proposes a HARQ packet generation method for various LDPC codes, which considers the expansion and deletion well. The feature of the residual structured low-density parity check code makes the extended punctured structured low-density parity check code have a very efficient and very complete HARQ packet generation method. In addition, the HARQ packet generation method of the present invention has two maximum advantages, one ensures that the performance of the high code rate LDPC code is as high as possible by making the extended puncturing structured low density parity check code have an optimal puncturing distribution. Optimal; another proposed to encode based on the actual code rate, significantly improving the efficiency of the coding. BRIEF abstract

Figure 1 is a typical digital communication system;

2 is an LDPC code based on variable rate of expansion and puncturing;

Figure 3 is a first method of the method of the present invention;

4 is a method for generating an LDPC HARQ mother code code word in the present invention;

Figure 5 is a second method of the method of the present invention. Preferred embodiment of the invention

The present invention will be described in detail below with reference to the accompanying drawings and specific embodiments.

As shown in FIG. ³ , the present invention provides a method A for generating a hybrid automatic repeat request HARQ packet of an Extending-and-Puncturing structured low-density parity check code, the method A comprising the following steps. :

A1), the low density parity check code encoder generates an N _FIR bit LDPC HARQ mother code code word according to the input bit information group, the code word including a bit information packet, an extended parity bit packet, and a punctured parity bit packet, The generated codeword is sent to a HARQ buffer (buffer);

A2), rearranging the LDPC HARQ mother code codeword bits in the HARQ buffer, keeping the information bits and the expansion check bit order unchanged, changing the punctured parity bit order, in order to obtain an optimal puncturing distribution ;

A3), sequentially selecting codeword bits from the rearranged HARQ mother codeword, that is, the first transmission starts from the first systematic bit, and then the start position of each transmission follows the last transmission end. Location, the binary sequence of the generated HARQ packets.

In step A1, the low density parity check code is a structured expansion/puncturing low density parity check code, having a code rate, a basic code, and a basic matrix m _3⁄4 x 3⁄4 Η and an expansion factor. ζ Unique description; after puncturing the basic codeword, the code with higher code rate is obtained; after the basic code is expanded, the code with lower code rate is obtained. In the case of expansion, by adding 1 row and increasing 1 column in the last row and the last column of the above Η _δ , the +1)x of the low-density parity check code of the code rate R _F!R (1) = ~^~ is obtained ( 3⁄4 +1) basic matrix

ΗΓ ^∞ '°"(1); Add 2 rows and 2 columns in the last row and the last column of Η _δ above, and get a low-density parity check code with a code rate of ^(2) = ("3⁄4 +2 )χ(« _έ +2) basic matrix ΗΓ ^∞ '·"(2) , m _b + 2

By analogy, considering the general case of expansion, the expansion code with the code rate ^^^«3⁄4) = ~^ ~ is obtained, that is, the line is increased and the A ₆ column is added in the last row and the last column of the above 11 _3⁄4 , and the code rate is obtained. The n _b + Am _b ) x (n _b + Δ» ^3⁄4 ) of the low-density parity check code of R _Fm (Am _b ) is the maximum expansion base matrix U ^ension (Am _b ), where Am _b = {1, 2 , ···, △ ··· Am Am^ m _b and " _δ are given by the system; Δ TM ^χ reflects the maximum extent of the base matrix expansion, the lowest possible bit rate for the above low density parity check code 7?^^«3⁄4 The above describes how to maximize the base matrix by puncturing expansion based on the base code. In the next step, we will further introduce how to group the input bit information based on the maximum expansion base matrix. Based on the basic code of the code rate, the mother code code word can be obtained after puncturing and expanding. More specifically, in step Al t, as shown in FIG. 4, the LDPCHARQ mother code code word generating method includes the following steps:

a) According to the principle of information packet length matching, the system will provide the maximum extended base matrix required for coding.

H "(Am ₆ ^max ) , expansion factor Ζ and basic matrix size parameter, n _b +Am _b ;

b) If the input information packet < ^, it also needs to be filled with a known bit (such as zero); ^ information packet;

c) The encoder is based on _Η ^'·. " _{(Δ ΜΧ)} , expansion factor _ζ , encodes the input ^ bit information packet to generate (? 3⁄4 + Δ" _ζ bit codeword; d) If /^ has a padding bit, the padding bit is removed, and the final encoding generates a N^^ + ^+A ^^ bit codeword, where the N^ bit codeword consists of K information bits and (m _b + Δ ^nax z consists of z-bits, and the ("3⁄4 + Am _3⁄4 ^max ) z-bit calibration packet includes A ° ^ax · ζ expansion check bits and ^ 2 punctured parity bits.

In step A1, the LDPC HARQ mother code codeword can also be generated by the following method, and the method includes the following steps:

a) According to the principle of information packet length matching, the system will provide the maximum expansion basis matrix (m _b + Am _b ^max ) x(n _b + Δ« ^3⁄4 (Am _b ^maK ), expansion factor Z and basic matrix size parameters required for coding. , n _b +Am _b ; b) If the input information packet <^z, it is also necessary to fill a known bit (such as zero) to form a yt information packet;

c) The encoder encodes in three cases according to the size of ^:

(1) ^L _{lotal ≤n} _b -z, the encoder will select ΗΓ*"(0), ie Η _δ from the maximum expansion base matrix; then encode the ^ bit information group according to Η ₄ , to generate „ = . 2-bit codeword; where is the sum of the lengths of the first k HARQ packets: L _total K)^L Let £^, (-1) = 0;

(2) If I^≥(n _{b +} Am )'z , the encoder will be based on the maximum expansion basis matrix

Z encodes a bit information packet to generate (n _b +Am _b ^max )-z bit codeword;

(3) In addition to the above cases (1) and (2), the calculation Δ/3⁄4 = Ltotal ~ K ^Z encoder will be selected from the maximum expansion base matrix. "(△»3⁄4); Then encode the bit information packet according to Η "·"(Δ»3⁄4), Ζ, generate (/3⁄4+Δ»ζ)· _Ζbit codeword, _punctate from the above codeword (3⁄4 + Δ«3⁄4) ζ - ^, bits, finally get a codeword of length N _rr = A. _te , d) If there is a padding bit, remove the known padding bits added by b).

In step A2, the LDPC HARQ mother code codeword bit arrangement is performed based on the following formula. The data of the i-th position before the arrangement is placed at the jth position after the arrangement.

i i<K Mi≥m _h -z

J = (i + x)

P _z z + ((z + ) mod and / = - other where L, represents the next rounding, i, j and / are all starting from 0; i is the bit position index of the pre-arranged codeword, j is the permutation The bit position index of the latter code word, O^J + im^+Am ^ x is the number of padding bits.

P is a selection pattern vector, having ^ elements, consisting of integers from ¹ to ¹ , which is a predefined arrangement of the above elements. Careful selection of the pattern P is required to ensure the best possible puncturing distribution and to ensure that the performance of the punctured code is as optimal as possible. P, is the index of the selection vector P as / elements.

P is determined by the check matrix portion of the base matrix of the basic code / _{δ δ} . According to the method

Α, code word bit selection is one of the main steps, select the pattern Ρ representation, first select the 基础 of the base matrix Η _δ . Column corresponding parity bits, if necessary, and then select the corresponding column of the basis matrix of parity bits _116, and so on. The selection pattern P is generated according to P _m6 _ _{1 →} P _mi _ ₂ → - → P. The order produced. Known ^..., ^, production methods include:

Select an element in the set Re mainSet = -1}/{ ₊₁ ,··.,Ρ }, and the remaining elements form a new set Rew wSetl; that is, Rem 'wSet has 3⁄4— ζ· + 1 Element, i own RemainSet = { a ₂ ,---, a _mi _ _M } , if an element is deleted from the set Rem 3⁄43⁄4t, the set of remaining elements is

{Re j = l, 2, ---, m _b -i + l , each j represents a choice. Where "," means the set is poor.

determine

The corresponding check matrix part is B _se! (J); determining H _W ) the minimum line weight and weight is the number of rows N = 0);

First determine the set JSet = " I dZ U) = m d U)}, that is, determine the minimum hour

J

Collection, minimum

When you make a collection, you can choose an element from it.

Wherein, according to the above method, the element P; can be generated; since the change from "3⁄4-1" to 0, Ρ^, Ρ, .Ά can be generated in sequence, that is, the selection vector 可以 can be obtained. When ί = 0, j is empty. In step A3, the codeword bit selection rule is selected based on the following formula. In the codeword after bit alignment, the bits of the obtained result according to the formula are selected and placed in the kth HARQ packet.笫i bit position,

S _kJ = (L _total (kD + i) mod(N _a )

A. _To , W is the sum of the lengths of the first k HARQ packets: ZW 0

Is the length of the HARQ mother code codeword;

, ' indicates the position index of the kth HARQ packet, which is used to identify the bit position of the current HARQ packet with index i, the first bit corresponds to, '· ^{= 0} (i=0), where i starts from 0;

Indicates the length of the kth HARQ packet. As shown in FIG. 5, the present invention further provides a method for generating a hybrid automatic retransmission request HARQ packet of another low density parity check code B. The method B includes the following steps:

B1), the low density parity check code encoder generates a 13⁄4 bit LDPC HARQ mother code codeword according to the input bit information group, the codeword includes a bit information packet, an extended face bit packet, and a punctured check bit packet, and the generated The codeword is sent to the HARQ buffer (buffer);

B2), sequentially selecting an LDPC HARQ codeword bit sequence from the HARQ buffer, that is, the first transmission starts from the first system bit, and the start position of each transmission follows the end of the previous transmission, and is generated. The binary sequence of the HARQ packet.

The LDPC HARQ mother code codeword generation method in the step B1) is the same as the LDPC HARQ mother code codeword generation method in the foregoing method A, and will not be described in detail herein. Same, in ^ no longer detailed "" description. , One ; , ', , , ,

In Method B, the calibration: bit portion of the base matrix >< Η _δ must have an optimal puncturing structure to ensure that the performance of the punctured code is sufficiently good. Here, by determining the selection of the pattern vector to determine how to obtain the optimal puncturing structure, the performance of the so-called puncturing code is optimal, that is, the puncturing code has the best BLER performance on the AWGN channel at each code rate. Further, the values of the selected pattern vectors are selected by different puncturing degrees through simulation. The present invention does not limit the manner in which the optimal performance is specifically determined, and may be determined in other ways. The present invention also provides a method C for generating a hybrid automatic repeat request HARQ packet of an Extending-and-Puncturing 氐 density parity check code, the method C comprising the following steps:

C1), the low density parity check code encoder generates an N _FIR bit LDPCHARQ mother code code word according to the input bit information group, the code word includes a bit information packet, an extended check bit packet, and a punctured check bit packet, and the generated The codeword is sent to the HARQ buffer (buffer);

The LDPC HARQ mother code codeword generation method is the same as the LDPC HARQ mother code codeword generation method in the foregoing method A, and is not described in detail herein;

C2), selecting an LDPC HARQ codeword bit from a HARQ buffer, and placing a bit of the selected result position into an ith bit position of the 笫k HARQ packets to generate a HARQ packet,

≤K i≥m _b -z

-tk≠Q,m _h -z>i> K

Where k is the index of the HARQ packet, reflecting the number of transmissions of the current HARQ packet, k=0 for the first transmission, k=l for the second transmission, plus 1 for the next transmission k; L” means rounding down, Both i and / start from 0, i is the bit position index of the pre-arranged codeword; 0 ≤ i < K + (m _b + Am _b ) _Z ; X is the number of padding bits.

K-\

Is the sum of the lengths of the first k HARQ packets: L; ^^ , Let ^, (- 1) = 0

N ' ^R is the length of the HARQ mother code word;

k, i represents the position index of the 笫k HARQ packets, and is used to identify the bit position of the current HARQ packet with index i, the first bit corresponds to, '· ^{= 0} (i=0), where i starts from 0;

^ represents the length of the kth HARQ packet;

P is the selection pattern vector, which has "3⁄4 elements, which are composed of integers from to, which are predefined arrangements of the above elements. You need to carefully select the pattern P to ensure the best possible puncturing distribution, guarantee puncturing The performance of the code is as optimal as possible. P, is the index of the selection vector P is / element P is determined by the check matrix portion of the basic matrix of the basic code 3⁄4χ H ₆ . According to the method A, the code word bit selection is one of the main steps, and the selection pattern P indicates that P of the basic matrix H ₆ is first selected. The check bits corresponding to the columns, if necessary, then select the check bits corresponding to the columns of the base matrix ^, and so on. The selection pattern P is generated according to P _m4 — ! P^ ₂ →~→P. The order produced. Known as { ₊₁ ,...,P _mi — J, the production methods include: From the collection

1}/{J) ₊₁ , select an element, the remaining elements form a new set Re .wSetl; that is, Rem z3⁄43⁄4t has "3⁄4 - ζ · +1 elements, i has been RemainSet = {a _{v 2} , ---, a _mb _ _M } , if an element α is deleted from the set Remiw'"Set, the set of remaining elements is

_/· = 1,2,···,/3⁄4 - ί + 1 , each j represents a choice. Where "/" means the set difference.

Determine the part of the check matrix corresponding to Re a _Z '"»3⁄4tl/} as H _{ie /} ( ); determine B ^ /) minimum line weight d ⁿ /) and weight as d = /) number of lines = (· / ·) ;

First determine the set JSet = I dZ (j) = mm d U)}, that is, determine the minimum hour

j

Set, then determine the minimum

When the ^ constitutes a collection, then select an element from R5fet. Wherein, according to the above method, the element can be generated due to the change from "3⁄4- 1 to 0", and Ρ, Ρ^ ₁₅ , .., Ρ can be generated in sequence. , that is, the selection vector 可以 can be obtained. When = 0, j is empty.

Based on structured puncturing expansion low density parity check code, the present invention provides method A and method

B and method C are three methods; where method A and method C have the same usage scenario, which is different from method B. Method B has requirements for the structure of the basic matrix of the structured low-density parity check code, and method B gives a specific puncturing method. The designed base matrix (or parity check matrix) should consider this method, that is, the corresponding deletion is required. The base matrix (or parity check matrix) of the residual check node has an optimal puncturing distribution, which ensures that the performance of the punctured code is excellent enough. However, Method A and Method C are not required for the structure of the underlying matrix of the structured low density parity check code.

The present invention has been described in detail by way of specific embodiments thereof, and the description of the above embodiments is provided to enable those skilled in the art to make or use the invention. Modifications will be readily apparent to those skilled in the art. The invention is not limited to these examples, or some aspects thereof. The scope of the invention is to be described in detail by the appended claims. Industrial applicability

The hybrid automatic request retransmission packet generation method of the low density parity check code of the invention systematically solves the defect of the incremental redundant HARQ support of the LDPC code, and realizes the support of the low density parity check code for HARQ, and gives A HARQ packet generation method based on the low density parity check code design of the present invention. On the one hand, by making the extended puncturing structured low-density parity check code have the optimal puncturing distribution, the performance of the high-rate LDPC code is guaranteed to be as optimal as possible; on the other hand, based on the actual code rate, the encoding is performed. Significantly improved the efficiency of the coding.

Claims

Claim

A hybrid automatic request retransmission packet generation method for a low density parity check code, the method comprising the steps of:

A1), the low density parity check code encoder generates a hybrid automatic request retransmission of the N _FIR bit low density parity check code according to the input bit information group, that is, an LDPC HARQ mother code code word, the code word includes a bit information packet, and an expansion Correcting a bit packet and a puncturing check bit packet, the generated codeword being sent to a hybrid automatic request retransmission or HARQ buffer;

A2) reordering the LDPC HARQ mother code codeword bits in the HARQ buffer, keeping the information bits and the expansion check bit order unchanged, and changing the punctured check bit order;

A3), sequentially selecting codeword bits from the rearranged HARQ mother codewords, that is, the first transmission starts from the first systematic bit, and then the start position of each transmission follows the last transmission end. Location, the binary sequence of the generated HARQ packets.

2. The method according to claim 1, wherein: in step A1, the low density parity check code has a basic code rate, and the basic code uses a basic matrix m _b xn _b Η _δ and an expansion Factor z unique description;

The punctured basic code obtains a code whose code rate is higher than R _m , and needs to punctify the check bits of the basic code code word;

Expanding the basic code to obtain a code with a code rate lower than R _m , adding _3⁄4 lines and increasing the A ₆ column in the last row and the last column of the above Η _{δ to} obtain a low-density parity check with a code rate of R _FI Am _b ~^ " Code m _b + Am _b

Ο3⁄4+Δ/3⁄4)χ 3⁄4+Δ»3⁄4) Maximum expansion base matrix.

The method according to claim 2, wherein in the step A1, the LDPC HARQ mother code codeword generating method comprises the following steps:

a) The system provides the maximum expansion basis matrix (m _b + Am ^x ) x ( _b + Am ₆ ^max ) H - (Am ₆ ^max ), the spreading factor z and the base matrix size parameter, n _b + Am _b ;

b) if the information packet K z is input, it is also necessary to fill X known bits to form a Α _δ ζ information packet;

C) The encoder is based on ΗΓ ^∞ '. "(Δ ^), expansion factor ζ, 'group the input ^ bit information into Line coding, generating 03⁄4 + Am^z bit code words;

d) If there are padding bits, the padding bits are removed and the final encoding yields a N^ = K + (m _b + Am ₆ ^ra3X ) z bit mother codeword.

4. Method according to claim 3, characterized in that the known bit is zero.

The method according to any one of claims 1 to 4, wherein in step A2, the LDPC HARQ mother code codeword bit arrangement is performed based on the following formula, and the data of the ith position before the arrangement is arranged. Put it in the j position after the arrangement,

Wherein, L" represents the next rounding, i, j, and / all start from 0; i is the bit position index of the pre-arranged codeword, j is the bit position index of the aligned codeword, and X is the number of padding bits, 0 ≤ i, j < K + (m _b + Am _b ) z; x is the number of padding bits;

P is a punctured pattern vector, which has ^ elements, which are composed of integers from " ^{6 to} ¹ ," and P, which is an index of punctured vectors P.

6. The method according to claim 5, wherein: the selection pattern P is generated according to " _{1→ΡΒ3⁄4} _ ₂ +.· _→ Ρ. The order of the generated;. known ^..., /^^, the production method includes the following steps

Select an element in the set Re mainSet = 1} / { · ₊₁ , ·.·, }, and the remaining elements form a new set Re mainSetl{j};

H _∞ , () , determine H _∞ , /) minimum line weight and weight for the number of rows N = /);, then select an element from RSet

You can get it.

The method according to any one of claims 1 to 4, wherein in step A3, the codeword bit selection rule is selected based on the following formula, and the bit of the selected result position is placed. To the ith bit position of the kth HARQ packet, S _k = (L _lotal (kl) + i) mod(N _IR )

Where k is the index of the HARQ packet, reflecting the number of transmissions of the current HARQ packet, k = 0 at the first transmission, k = 1 at the second transmission, and 1 plus 1 for the next transmission;

A. , _a , W is the sum of the lengths of the first k HARQ packets: 0

NFIR is the length of the HARQ mother code codeword;

, ' indicates the position index of the kth HARQ packet, which is used to identify the bit position of the current HARQ packet with index i, the first bit corresponds to . '· ^{= 0} (i=0), where i starts from 0;

Indicates the length of the kth HARQ packet.

8. A hybrid automatic request retransmission packet generation method for a low density parity check code, the method comprising the steps of:

B1), the low density parity check code encoder generates a hybrid automatic request retransmission of the 1^ bit low density parity check code according to the input bit information group, that is, the LDPC HARQ mother code code word, the code word includes the bit information packet, and the expansion Checking the bit packet and the punctured check bit packet, and the generated codeword is sent to the hybrid automatic request retransmission or HARQ buffer;

B2), sequentially selecting the LDPC HARQ codeword bit sequence from the HARQ buffer, that is, the first transmission starts from the first system bit, and then the start position of each transmission follows the last transmission end. Location, the binary sequence of the generated HARQ packets.

9. The method of claim 8, wherein: in step B1, the low density parity check code with a code rate R _m basic code, base code group matrix with m _{_b} X n _b

唯一_δ and expansion factor z are the only description;

The punctured basic code code word obtains a code whose code rate is higher than R _m , and the parity bit of the basic code code word needs to be punctured;

Expanding the basic code to obtain a code with a code rate lower than R _m , adding a line and increasing the Δ«3⁄4 column in the last row and the last column of the above Η _{δ to} obtain a low-density parity check code having a code rate R _F » ~ ^ " m _b + Am _b

(m _b + Am _b )x (n _b + AJ maximum expansion base matrix.

The method according to claim 9, wherein in step B1, the LDPC HARQ mother code codeword generating method comprises the following steps:

a) The system will provide the maximum expansion basis matrix required for encoding ("3⁄4 + Am ₆ ^max )x (3⁄4 + Am _i ^max )

, expansion factor z and base matrix size parameters, n _b + Am _b ;

b) If the input information group <^z, it is also necessary to fill the known bits to form an information packet;

C) the encoder is based on

Spreading factor z, encoding the input bit information packet, and generating (" ₆ + Am^z bit codeword;

d) If /^ has padding bits, remove the padding bits, and finally encode the generated

Bit code word.

11. The method of claim 10 wherein: said known bit is zero.

The method according to any one of claims 8 to 11, wherein in step B2, the codeword bit selection rule is selected based on the following formula, and the bit of the selected result position is placed. To the ith bit position of the kth HARQ packet,

S _k = (L _iotal (kl) + i) mod{N _m )

K-l

Is the sum of the lengths of the first k HARQ packets: L _tatal (k) = ∑Lj , set, (- 1) = 0

7=0

^N FIR is the length of the HARQ mother code word;

, 'represents the location index of the 笫k HARQ packets, which is used to identify the bit position of the current HARQ packet index i, where i starts from 0, the first bit corresponds, ' ^{= 0} , where i=0;

Indicates the length of the kth HARQ packet.

13. A hybrid automatic request retransmission packet generation method for a low density parity check code, the method comprising the steps of:

C1), the low density parity check code encoder generates a hybrid automatic request retransmission of the N _FIR bit low density parity check code according to the input ^: bit information packet, that is, an LDPC HARQ mother code code word, the code word including the bit information packet , expansion check bit packet and punctured check bit packet, generated codeword Send to the hybrid automatic request retransmission or HARQ buffer;

C2), selecting the LDPC HARQ codeword bit from the HARQ buffer according to the following formula, and placing the bit of the selected result position into the ith bit position of the kth HARQ packet to generate a HARQ packet ,

≤K i≥m _b -z

K≠ 0,m _h · z > i > K

Where k is the index of the HARQ packet, reflecting the number of transmissions of the current HARQ packet, k=0 for the first transmission, k=l for the second transmission, plus 1 for the next transmission k; L” means rounding down, Both i and / start from 0, i is the bit position index of the pre-arranged codeword; 0 ≤ i < + ( ₆ + A ₆ ) z; X is the number of padding bits; is the length of the first k HARQ packets And: 0

NFIR is the length of the HARQ mother code codeword;

, ' indicates the position index of the kth HARQ packet, which is used to identify the bit position of the current HARQ packet index i, where i starts from 0, and the first bit corresponds to ¹ ^ '· ⁼⁰ , where i=0;

Indicates the length of the kth HARQ packet;

P is a punctured pattern vector, which has "3⁄4 elements, which are composed of integers from -1, and _P , which is an index of the punctured vector P.

14. The method of claim 13 wherein:

The selection pattern P is generated in the order of p _mb — _{l → mb} — _{2 →} .. _{→ 0} ; the known {P^-, P _m ^} , generating the S method includes the following steps:

Select an element in the set Re a"Set = - ..., corpse }, and the remaining elements form a new collection Re mainSet\{j);

Determine the part of the check matrix corresponding to Rem VzSetlW as H _∞ , ( ), and determine the minimum line weight and weight of H _i (_/) as d = /) the number of rows N = /);

15. The method according to claim 13, wherein: in step C1, the low density parity check code has a code rate basic code, and the basic code can use a basic matrix m _b X n _b ≡ _b and The expansion factor Z is the only description;

Expanding the base code to obtain a code with a code rate lower than R _m , adding 1⁄23⁄4 lines and increasing Δ3⁄4 columns in the last row and the last column of the above Η _δ , resulting in a low density of the code rate H _FJR (Am _b ) = one "^ ~ · Parity code m _b + Am _b

(m _b +Am _b )x(n _b +Am ₆ ) maximum expansion base matrix.

The method according to claim 15, wherein in step C1, the LDPC HARQ mother code codeword generating method comprises the following steps:

a) The system will provide the maximum expansion of the base matrix ^{(/ ¾ + Am, max)} x (¾ + Am 6 max) B. tensi0 "(Am b mm) ^ Z spreading factor and basic matrix size required for the encoding parameters, n _b +Am _b ; b) If the input information packet <;^, it is also necessary to fill the known bits to form the ^ information packet;

C) The encoder is based on 11^. "^ ^", a spreading factor z, encoding the input bit information packet to generate a 03⁄4+Am ₆ ^max )z bit codeword;

d) If there are padding bits, the padding bits are removed and the final encoding yields a N^ = K + (m _b + Am _b ^max ) z bit codeword.

17. The method of claim 16 wherein: said known bit is zero.

18. A low-density hybrid automatic request retransmission method, that is, a HARQ mother code codeword generation method, including the following steps: a) The system will provide the maximum expansion basis matrix required for encoding ("3⁄4 + Am ^x ) x ( _b + m ^x ) W" ( ), expansion factor _z and base matrix size parameter, n _{b +} Am _b ; b) if input information grouping

Information grouping

C) the encoder is based on

, ζ, encoding the input bit information packet to generate (3⁄4 + Am ₆ ^max ) z-bit codeword; d) if ^^ has padding bits, remove the padding bits, and finally encode to generate N = K + (m _b + Am _i r ^x bit codeword.

19. A low-density hybrid automatic request retransmission method, that is, a HARQ mother code codeword generation method, including the following steps

a) According to the principle of information packet length matching, the system will provide the maximum expansion basis matrix required for coding ( ₆₎

Expansion factor ζ and basic matrix size parameters, n _b + Am _b ;

b) if the input information packet also needs to be filled with known bits to constitute / ^ information packet;

C) The encoder encodes according to the size of k _b z in three cases:

(1) If JL _total ≤n _b . _Z , the encoder will choose from the largest expansion base matrix

, ie Η ₆ ; then encode the ^ bit information packet according to Η _& ζ to generate a N^-^z bit codeword; kl

Among them, A. _To ) is the sum of the lengths of the first k HARQ packets: L _t . _Tal ^ =∑L Let £^, (- 1) = 0 _;

7=0

(2) If

Encoder will encode the _2- bit information packet according to the maximum expansion base matrix Η^««· , to generate (n _b +Am ₆ ^max )-z bit codeword;

(3) In addition to the above cases (1) and (2), calculate Δ»3⁄4, encoder

Η ^Μ '·"(Δ»3⁄4) will be selected from the maximum expansion base matrix; then ("△3⁄4", z is used to encode the bit information packet to generate (% + Am).z bit codeword, Pruning from the above codewords (/3⁄4 + Δ«3⁄4 - bits, resulting in a codeword of length L =; d) if .2 has padding bits, remove the known padding bits added by b).