CN102891737A - Method and system for coding and decoding binary rateless codes - Google Patents

Abstract

The invention discloses a method and system for coding and decoding binary rateless codes and belongs to data transmission technologies in the field of communications. According to the method and the system, starting from lowering the inversion complexity of a BDRC (Binary Deterministic Rateless Code) decoder, original data packets are subdivided into a plurality of original data blocks of equal length before coding, the BDRC coding is then carried out, and then, the order of a code generation matrix is reduced, so that the complexity in matrix inversion during decoding is effectively reduced, and in addition, the complexity of a BDRC coding process is reduced, namely, the complexity of coding and decoding processes can be greatly reduced on the premise that the performance of BDRC is not affected.

Description

A kind of coding﹠decoding method and system of binary system no-rate codes

Technical field

The present invention relates to the data transmission technology of the communications field, particularly a kind of chnnel coding and interpretation method and system that reduces complexity.

Background technology

In recent years, along with the development of the communication technology, for the also constantly growth of demand of transmission capacity and transmission range.Data can be owing to the impact that is subject to various noises and decline produces mistake in transmission course, be to guarantee quality and the efficient of communicating by letter, and must control effectively and corrects these mistakes.Wherein more common method is that forward error correction (FEC) and feedback retransmit methods such as (ARQ).

In the coded system of forward error correction, decoder not only can be found mistake automatically, and can also carry out error correction in certain scope to the mistake that exists in the code word.Common forward error correction has Read-Solomon (Reed-Solomon) code, convolution code, Turbo code and low density parity check code (LDPC) etc.Turbo code and LDPC for the immediate development speed have one-way transmission, need not feed back, and can automatically correct a mistake according to the regularity of code, and error correction is rapid, the characteristics that error correcting capability is very strong.But owing to volume, the decoding architecture of forward error correction are complicated, efficient is low, generally only is used in communication physical layer as the bit-level error correction in the frame.

In order to guarantee the reliability of communication system transmission, only adopt the data after forward error correction can't guarantee to decipher whether entirely true, for this reason, need to adopt other error correcting technique of frame level, the easiest the most common method is the ARQ technology.The ARQ technology realizes comparatively simple, but shortcoming is to need feedback channel, and receives that whenever an erroneous frame just needs feedback once.And the other class that recent development goes out is called as the frame level code of " without speed " code, does not then need a large amount of feedbacks, and feeds back once after only need to receiving a series of Frames, thereby with respect to the ARQ technology, can reduce in a large number the feedback requirements amount.

Common no-rate codes comprises at present: LT (Luby Transform) code, Raptor code and binary system are determined no-rate codes (Binary Deterministic Rateless codes, BDRC) etc.LT, Raptor code prove in theory owing to adopting the random coded method, and the performance of these two kinds of codes only has when the frame number that participates in coding is very many, just can reach theoretical optimal performance, and when encoding frame number for the participation of general quantity, then poor-performing.No matter the BDRC code proves in theory that then its performance participates in the coding frame number what can both reach in theory optimal performance for.But the BDRC code is owing to will invert to a large sparse matrix in the interpretation method, although inversion process can utilize the sparse matrix characteristic to simplify to a certain extent, the complexity of its implementation procedure is still too high, is unfavorable for using in real system.

Summary of the invention

Goal of the invention of the present invention is: the problem for the coded system of above-mentioned existing BDRC code exists, a kind of improved BDRC coding method is proposed, and under the prerequisite that performance of BDRC code book body is not exerted an influence, can reduce widely the complexity of decode procedure.

BDRC coding method of the present invention comprises the following steps:

With T the original data block that be divided into of long K source code packet to be compiled for P, the length of each original data block is p=P/T, and described original data block press the capable K of T and is listed as placement;

Successively every row original data block is sent into binary system no-rate codes encoder and encode, every row produces M redundant block, and then every row has N encoding block, N=K+M, and p 〉=N-1;

Described encoding block is pressed the capable N row of T and is placed, and every row encoding block forms an encoded packets; Successively N encoded packets carried out framing and consist of coded frame, and add incoming frame identifier and Frame Check Sequence for each coded frame.

In the above-mentioned steps, the process of every row original data block being carried out the BDRC coding is existing conventional BDRC cataloged procedure, produce the process of M redundant block, detailed process can be described as: obtain one group of interim piece at the additional bit " 0 " of the afterbody of each original data block, then each interim piece is carried out cyclic shift in piece and obtain other one group of interim piece, shift count is that the original data block frame number subtracts 1 and subtracts 1 product and block length p with present encoding piece frame number and add the numerical value that obtains behind 1 the delivery, afterwards should interim piece remove last bit and obtain new one group of interim piece, at last whole up-to-date interim pieces being carried out XOR processes, obtain the redundant block of a current frame number, repeat this process, can obtain all M redundant block; Conventional BDRC cataloged procedure also can represent with matrix-style: K original data block form one long for the capable vector of K * p multiply by one by the definite behavior K * p of coding rule, classify the generator matrix of the two element field of M * p as, obtain one and longly be the column vector of M * p, this column vector has comprised that M is long to be the redundant block of p, and this generator matrix or is the matrix in block form that p * the p square formation forms by K * M piecemeal and each matrix in block form.

BDRC coding method of the present invention is from the invert angle of complexity of the decoder that reduces the BDRC code, be subdivided into a plurality of isometric original data blocks by the raw data packets before will encoding, and then carry out BDRC and encode, and then reduced the exponent number of coding generator matrix, thereby effectively reduce when decoding to the complexity of matrix inversion, add the complexity that has reduced BDRC cataloged procedure itself, namely in situation the about performance of BDRC code book body not being exerted an influence, can reduce widely the complexity of coding﹠decoding process.

Preferably, the afterbody that inserts frame identifier and/or the coded frame of each at each the head of coded frame inserts Frame Check Sequence.

The improved BDRC coding method of corresponding the present invention the invention also discloses a kind of interpretation method of binary system no-rate codes, it is characterized in that, comprises the following steps:

K the encoded packets of receiving is divided into T group coding piece, and every group comprises K encoding block;

Determine correct original data block and the redundant block number of first group coding piece, the number of correct original data block is taken as Ns, and the number of correct redundant block is taken as Nr, and then the number of wrong original data block is Nr, and Ns+Nr=K;

According to the redundant block create-rule of binary system no-rate codes coding side, obtain Nr capable hypermatrix A in generator matrix corresponding to correct redundant block; Matrix A is divided into matrix B, C, matrix B is the square formation of the Nr * Nr of row piecemeal composition corresponding to wrong original data block again, and Matrix C is the hypermatrix of the Nr * Ns of row piecemeal composition corresponding to correct original data block;

According to described matrix B, C, reach the individual correct raw data packets of Ns of every group coding piece and the individual correct redundant block of Nr recovers every group coding piece successively according to binary system no-rate codes coding rule wrong original data block.

Because BDRC interpretation method of the present invention, after initial data is carried out piecemeal, to encode and corresponding interpretation method again, it is that a depression of order of the generator matrix of BDRC code is processed in essence, thereby do not sacrificing in any performance and the efficiency of transmission situation, reduce widely the complexity of matrix inversion and solving equations, effectively reduced the decoding overall complexity.

Further, according to described matrix B, C, reach the individual correct raw data packets of Ns of every group coding piece and the individual correct redundant block of Nr recovers every group coding piece successively according to binary system no-rate codes coding rule wrong original data block, concrete steps are:

To be multiplied each other with Matrix C by the vector that the correct original data block of Ns forms obtains vectorial a, and the vector of the individual correctly redundant block of vectorial a and Nr composition is carried out behind the XOR and the vectorial a addition amount of obtaining b; According to formula x=B ^-1Bmod2 reverts to correct original data block with the individual wrong original block of the Ns of every group coding piece.

In the BDRC interpretation method of the present invention, in the recovery process that realizes according to the correct redundant block of each group coding piece its wrong original data block, only at first group coding piece compute matrix B, C, for other code set, can directly utilize known matrix B ^-1And C, only needing to obtain different b according to different Ns correct original data block with Nr correct redundant block, Nr wrong original data block with each group reverts to correct original data block fast.

Based on the improved BDRC coding﹠decoding of the present invention method, the invention discloses a kind of coding﹠decoding system of binary system no-rate codes, comprise transmitting end device and receiving end device, wherein:

Transmitting end device comprises coding processing module, signal processing module and transceiver module,

Described coding processing module is to the source code packet rear formation coded frame of encoding: will grow first T the data block that be divided into for K the source code packet to be compiled of P, the length of each data block is p=P/T, and original data block press the capable K of T and is listed as placement; Then successively every line number is carried out binary system no-rate codes coding, every row produces M redundant block, and then every row has N encoding block, N=K+M, and p 〉=N-1; Encoding block is pressed the capable N row of T and is placed, and every row encoding block forms an encoded packets; Successively N encoded packets carried out framing again and consist of coded frame, and add incoming frame identifier and Frame Check Sequence for each coded frame;

The coding processing module is delivered to signal processing module successively with N coded frame, and described signal processing module interweaves, delivers to transceiver module after the modulation treatment coded frame;

Described transceiver module carries out signal and sends, and intercepts simultaneously the signal of receiving end device feedback, then stops transmission if receive that the signal of " stopping to send " or N coded frame all are sent;

Receiving end device comprises transceiver module, signal processing module, frame check processing module, decoding processing module and processing data packets module;

Described transceiver module is sent the transmitting terminal signal of receiving into signal processing module, and signal processing module carries out demodulation, deinterleaving is processed and obtained coded frame, and described coded frame is delivered to frame check processing module;

Whether described frame check processing module is differentiated this coded frame and is received correctly, obtain encoded packets after correct coded frame removed frame identifier and Frame Check Sequence, preserve encoded packets to memory according to the frame identifier, correct coding bag number in memory is K, or correct coding bag number is less than K but total coding bag number when being N, then control transceiver module sends " stopping to send " to transmitting terminal signal, if the encoded packets number is K, then K encoded packets delivered to the decoding processing module, otherwise N encoded packets delivered to the processing data packets module;

Described decoding processing module is delivered to the processing data packets module with all correct original data blocks after the wrong original data block in each group coding bag is reverted to correct original data block:

K encoded packets is divided into T group coding piece, and every group comprises K encoding block; Determine correct original data block and the redundant block number of first group coding piece, the number of correct original data block is taken as Ns, and the number of correct redundant block is taken as Nr, and then the number of wrong original data block is Nr, and Ns+Nr=K; According to the redundant block create-rule of no-rate codes coding side, obtain Nr capable hypermatrix A in generator matrix corresponding to correct redundant block; Matrix A is divided into matrix B, C, matrix B is the square formation of the Nr * Nr of row piecemeal composition corresponding to wrong original data block again, and Matrix C is the hypermatrix of the Nr * Ns of row piecemeal composition corresponding to correct original data block; According to described matrix B, C, reach the individual correct raw data packets of Ns of every group coding piece and the individual correct redundant block of Nr recovers every group coding piece successively according to the no-rate codes coding rule wrong original data block again;

The original data block basis principle identical with transmitting end device that described processing data packets module is correct with all, every T original data block forms a raw data packets, if the number of raw data packets is K, then sends into system's stay of two nights, otherwise reports an error to system's stay of two nights.

In sum, owing to having adopted technique scheme, the invention has the beneficial effects as follows: from the invert angle of complexity of the decoder that reduces the BDRC code, be subdivided into a plurality of isometric original data blocks by the raw data packets before will encoding, and then carry out BDRC and encode, and then reduced the exponent number of coding generator matrix, thereby effectively reduce when decoding to the complexity of matrix inversion, add the complexity that has reduced BDRC cataloged procedure itself, namely under the prerequisite that performance of BDRC code book body is not exerted an influence, can reduce widely the complexity of coding﹠decoding process.

Embodiment

Disclosed all features in this specification, or the step in disclosed all methods or the process except mutually exclusive feature and/or step, all can make up by any way.

Disclosed arbitrary feature in this specification is unless special narration all can be replaced by other equivalences or the alternative features with similar purpose.That is, unless special narration, each feature is an example in a series of equivalences or the similar characteristics.

Embodiment 1

1. transmitting terminal

Transmitting end device comprises coding processing module, signal processing module and transceiver module, and the specific implementation process of each module is as follows:

(1) coding processing module:

Piecemeal: to 8 raw data packets that information source is sent, each raw data packets length is 1000 bits, and 1000 bits are divided into 100 data blocks, and each data block length is 10 bits, places according to row, has 100 row, 8 data blocks of every row;

Coding: 8 data blocks of above-mentioned every row are carried out conventional BDRC coding, and cataloged procedure is as follows: first at each original data block S _i, i=1,2 .., 8, the additional bit " 0 " of afterbody obtains S _i', then it is carried out cyclic shift in frame, cyclic process can be represented by the formula:

S _i=Φ (S _i" ', be (j-1) mod11 (i-1)) formula (1)

Wherein, k time cyclic shift is carried out in Φ (S, k) expression to S, with the primitive frame S after the process cyclic shift that obtains _i" remove the tail bit and obtain S _i" ', and with whole 8 S _iThe coded frame R that it is j that " ' carry out XOR obtains a frame number _j, j=1,1 ..., 16, can represent with following formula:

$R_j={\mathrm{\Σ}}_{i=1}^8{S}_i^{\′\′\′}\mathrm{mod}2$ Formula (2)

Also can be with the above-mentioned cataloged procedure of matrix notation: generator matrix be one 8 * 16 matrix in block form, wherein hypermatrix be 10 * 10 and each row and column the square formation of " 1 " is arranged at the most, per like this 8 original data blocks and 16 redundant blocks, form 24 encoding blocks, always have 100 groups, place according to row, have 100 row, 24 encoding blocks of every row; Then according to the order of Column Row all encoding blocks being carried out framing processes;

Framing: carry out framing by every row, be that per 100 encoding blocks are that unit carries out framing, every like this frame data length is 1000, totally 24 frames, then insert the frame number of unique this frame of identification in every frame header, and its afterbody insert one long be 16 cyclic redundancy check (CRC) sequence, 24 coded frame after handling are sequentially sent into signal processing module by front and back.

(2) signal processing module: coded frame is carried out that routine interweaves, delivered to transceiver module after the modulation treatment.

(3) transceiver module: settling signal sends, and waits for simultaneously the signal of receiving terminal feedback, in case the feedback signal of receiving " stop to send " then stop signal transmission processing process, otherwise till this process is repeated until that all 24 coded frame all send.

2, receiving terminal

Receiving end device comprises transceiver module, signal processing module, frame check processing module, decoding processing module and processing data packets module, and the specific implementation process of each module is as follows:

(1) transceiver module: the transmitting terminal signal of receiving is sent into signal processing module;

(2) signal processing module: the transmitting terminal signal of receiving is carried out demodulation, deinterleaving process and obtain coded frame, coded frame is delivered to frame check processing module;

(3) frame check processing module: differentiate this coded frame and whether receive correctly, (being encoded packets) was kept in the corresponding memory according to frame number after the correct coding frame removed frame number and Frame Check Sequence, correct coding bag number in memory is 8, or correct coding bag number is less than K but total coding bag number when being 24, then send the signal of " stopping to send " to transmitting terminal, if the encoded packets number is 8, then 8 encoded packets are delivered to the decoding processing module, otherwise 24 encoded packets are delivered to the processing data packets module.

In the present embodiment, when receiving the 6th redundant block, when correct frame number reached 8, the control transceiver module sent the signal of " stopping to send " to transmitting terminal, then 8 encoded packets in the memory are delivered to the decoding processing module;

(3) decoding processing module: at first, 8 encoded packets and transmitting terminal are divided into 100 row by capable accordingly, then 8 encoding blocks of every behavior are divided into original data block and redundant block with 8 encoding blocks of first group, wherein, have correct original data block 5 (to be assumed to be S ₂, S ₃, S ₆, S ₇And S ₈), 3 of mistake original data blocks (are assumed to be S ₁, S ₄And S ₅), correct redundant block is 3 and (is assumed to be R ₂, R ₃And R ₆), then find out 3 capable hypermatrix A in generator matrix corresponding to correct redundant block, represent with following formula

$\left[\begin{array}{cccc}{g}_{21}& g_{22}& \·\·\·& g_{28}\\ g_{31}& g_{32}& \·\·\·& g_{38}\\ g_{61}& g_{62}& \·\·\·& g_{68}\end{array}\right]$ Formula (3)

Wherein, g _IjBe 10 * 10 matrixs in block form.Again this matrix is divided into two hypermatrix, a block matrice of square B of 3 * 3 who is formed by 3 row piecemeal frames corresponding to wrong original data block, namely

$B=\left[\begin{array}{ccc}{g}_{21}& g_{24}& g_{25}\\ g_{31}& g_{34}& g_{35}\\ g_{61}& g_{64}& g_{65}\end{array}\right]$ Formula (4)

According to the BDRC characteristics, square formation B necessarily has unique inverse matrix B ^-1

3 * 5 the hypermatrix C that another is comprised of 5 row hypermatrix corresponding to correct original data block, namely

$C=\left[\begin{array}{ccccc}{g}_{22}& g_{23}& g_{26}& g_{27}& g_{28}\\ g_{32}& g_{32}& g_{36}& g_{37}& g_{38}\\ g_{62}& g_{62}& g_{66}& g_{67}& g_{68}\end{array}\right]$ Formula (5)

Then the vector that 5 correct original data blocks is formed and Matrix C are carried out two element field and are multiplied each other and obtain a vector, should vector and the addition of 3 correct redundant block composition of vector two element fields obtain vectorial b, namely

B=C * [S ₂S ₃S ₆S ₇S ₈] ^T+ [R ₂R ₃R ₆] mod2 formula (6)

Then 3 wrong original data blocks just can revert to correct original data block by following formula:

X=B ^-1Bmod2 formula (7)

And for the encoding block of other groups, need not to calculate in addition B ^-1And C, only need to calculate b according to formula (6), at last according to formula (7), 3 wrong original data blocks are reverted to correct original data block;

(4) processing data packets module: according to the principle identical with transmitting terminal, the packet that forms long 1000 by per 100 original blocks of row is sent into system's stay of two nights with 8 correct original data block pieces.

The present invention is not limited to aforesaid embodiment.The present invention expands to any new feature or any new combination that discloses in this manual, and the arbitrary new method that discloses or step or any new combination of process.

Claims (8)

1. the coding method of a binary system no-rate codes is characterized in that, comprises the following steps:

With T the original data block that be divided into of long K source code packet to be compiled for P, the length of each original data block is p=P/T, and described original data block press the capable K of T and is listed as placement;

Successively every row original data block is sent into binary system no-rate codes encoder and encode, every row produces M redundant block, and then every row has N encoding block, N=K+M, and p 〉=N-1; Described encoding block is pressed the capable N row of T and is placed, and every row encoding block forms an encoded packets;

Successively N encoded packets carried out framing and consist of coded frame, and add incoming frame identifier and Frame Check Sequence for each coded frame.

2. the coding method of binary system no-rate codes as claimed in claim 1 is characterized in that, inserts the afterbody insertion Frame Check Sequence of frame identifier and/or the coded frame of each at the head of each coded frame.

3. the coding method of binary system no-rate codes as claimed in claim 1 or 2 is characterized in that, described Frame Check Sequence is the cyclic redundancy check (CRC) sequence.

4. the interpretation method of a binary system no-rate codes is characterized in that, comprises the following steps:

K the encoded packets of receiving is divided into T group coding piece, and every group comprises K encoding block;

Determine correct original data block and the redundant block number of first group coding piece, the number of correct original data block is taken as Ns, and the number of correct redundant block is taken as Nr, and then the number of wrong original data block is Nr, and Ns+Nr=K;

According to the redundant block create-rule of binary system no-rate codes coding side, obtain Nr capable hypermatrix A in generator matrix corresponding to correct redundant block; Matrix A is divided into matrix B, C, matrix B is the square formation of the Nr * Nr of row piecemeal composition corresponding to wrong original data block again, and Matrix C is the hypermatrix of the Nr * Ns of row piecemeal composition corresponding to correct original data block;

According to described matrix B, C, reach the individual correct raw data packets of Ns of every group coding piece and the individual correct redundant block of Nr recovers every group coding piece successively according to binary system no-rate codes coding rule wrong original data block.

5. the interpretation method of binary system no-rate codes as claimed in claim 4, it is characterized in that, according to described matrix B, C, reach the individual correct raw data packets of Ns of every group coding piece and the individual correct redundant block of Nr recovers every group coding piece successively according to binary system no-rate codes coding rule wrong original data block, concrete steps are:

To be multiplied each other with Matrix C by the vector that the correct original data block of Ns forms obtains vectorial a, and the vector of the individual correctly redundant block of vectorial a and Nr composition is carried out behind the XOR and the vectorial a addition amount of obtaining b;

According to formula x=B ^-1Bmod2 reverts to correct original data block with the individual wrong original block of the Ns of every group coding piece.

6. the coding﹠decoding system of a binary system no-rate codes is characterized in that, comprises transmitting end device and receiving end device, wherein:

Transmitting end device comprises coding processing module, signal processing module and transceiver module,

Described coding processing module is to the source code packet rear formation coded frame of encoding: will grow first T the data block that be divided into for K the source code packet to be compiled of P, the length of each data block is p=P/T, and original data block press the capable K of T and is listed as placement; Then successively every line number is carried out binary system no-rate codes coding, every row produces M redundant block, and then every row has N encoding block, N=K+M, and p 〉=N-1; Encoding block is pressed the capable N row of T and is placed, and every row encoding block forms an encoded packets; Successively N encoded packets carried out framing again and consist of coded frame, and add incoming frame identifier and Frame Check Sequence for each coded frame;

The coding processing module is delivered to signal processing module successively with N coded frame, and described signal processing module interweaves, delivers to transceiver module after the modulation treatment coded frame;

Described transceiver module carries out signal and sends, and intercepts simultaneously the signal of receiving end device feedback, then stops transmission if receive that the signal of " stopping to send " or N coded frame all are sent;

Receiving end device comprises transceiver module, signal processing module, frame check processing module, decoding processing module and processing data packets module,

Described transceiver module is sent the transmitting terminal signal of receiving into signal processing module, and signal processing module carries out demodulation, deinterleaving is processed and obtained coded frame, and described coded frame is delivered to frame check processing module;

Whether described frame check processing module is differentiated this coded frame and is received correctly, obtain encoded packets after correct coded frame removed frame identifier and Frame Check Sequence, preserve encoded packets to memory according to the frame identifier, when the correct coding bag number in the memory is that K or correct coding bag number are less than K but total coding bag number when being N, then send the signal of " stopping to send " to transmitting terminal, if the encoded packets number is K, then K encoded packets delivered to the decoding processing module, otherwise N encoded packets delivered to the processing data packets module;

Described decoding processing module is delivered to the processing data packets module with all correct original data blocks after the wrong original data block in each group coding bag is reverted to correct original data block:

K encoded packets is divided into T group coding piece, and every group comprises K encoding block; Determine correct original data block and the redundant block number of first group coding piece, the number of correct original data block is taken as Ns, and the number of correct redundant block is taken as Nr, and then the number of wrong original data block is Nr, and Ns+Nr=K; According to the redundant block create-rule of no-rate codes coding side, obtain Nr capable hypermatrix A in generator matrix corresponding to correct redundant block; Matrix A is divided into matrix B, C, matrix B is the square formation of the Nr * Nr of row piecemeal composition corresponding to wrong original data block again, and Matrix C is the hypermatrix of the Nr * Ns of row piecemeal composition corresponding to correct original data block; According to described matrix B, C, reach the individual correct raw data packets of Ns of every group coding piece and the individual correct redundant block of Nr recovers every group coding piece successively according to the no-rate codes coding rule wrong original data block again;

The original data block basis principle identical with transmitting end device that described processing data packets module is correct with all, every T original data block forms a raw data packets, if the number of raw data packets is K, then sends into system's stay of two nights, otherwise reports an error to system's stay of two nights.

7. the coding﹠decoding system of binary system no-rate codes as claimed in claim 6 is characterized in that, described coding processing module is inserted the afterbody insertion Frame Check Sequence of frame identifier and/or the coded frame of each at the head of each coded frame.

8. such as the coding﹠decoding system of claim 6 or 7 described binary system no-rate codes, it is characterized in that, described decoding processing module is according to described matrix B, C, reach the individual correct raw data packets of Ns of every group coding piece and the individual correct redundant block of Nr recovers every group coding piece successively according to binary system no-rate codes coding rule wrong original data block, concrete steps are:

To be multiplied each other with Matrix C by the vector that the correct original data block of Ns forms obtains vectorial a, and the vector of the individual correctly redundant block of vectorial a and Nr composition is carried out behind the XOR and the vectorial a addition amount of obtaining b;

According to formula x=B ^-1Bmod2 reverts to correct original data block with the individual wrong original block of the Ns of every group coding piece.