FR2838581A1 - Method for coding and/or decoding of the error-correction codes, corresponding devices and signal representative of coded data block - Google Patents

Abstract

The method for decoding the error-correction codes is of type associating a block of coded data to the data coded according to a global code comprising at least two constituent sub-codes (Ri). An irregular bipartite graph is associated with the global code, the decoding method is iterative, and each iteration produces a block of extrinsic data. A block of data to decode comprising the coded data and the extrinsic data is distributed in a set of disjointed memory banks (BMi,661,662,66i,66p) which are independently addressable. At each iteration a step of supplying in parallel at least two decoders (SISOi, 671, 672, 67i, 67p) with data to decode, where each corresponds respectively to at least one of the sub-codes. The data to decode are extracted in parallel from at least two of the memory banks (BMi) for supplying the decoders (SISOi), and each decoder is supplied sequentially with the correspondingly read data to decode. At least one of the sub-codes is a circular convolutional recursive code. At least one of the sub-codes is a systematic code. At least two of the sub-codes are m-binary codes, where m is an integer greater or equal to 2, and an intra-symbols permutation is applied between at least two of the sub-codes. The method also comprises a step of directing the data to code towards one of the codes, in parallel and simultaneously. The step of directing comprises a step of circular permutation of a set comprising at least a part of data to code. A signal representative of a block of coded data is to be decoded by the iterative decoding method. A device for decoding the error-correction code implements the means for storing a bloc of data to decode, and the means for supplying in parallel at least two decoders. A device for coding the error-correction code for transmitting towards a receiver and/or storing on a data support, comprises the means for supplying in parallel at least two coders.

Description

<Desc / Clms Page number 1>

 Method for coding and / or decoding error correcting codes, devices and corresponding signals

 The present invention relates to the field of encoding digital data intended to be transmitted and / or stored on a noisy channel and then decoded so as to correct the transmission and / or reading errors.

 More precisely, the invention relates to error correction coding based on the parallel concatenation of several codes, and a decoding method allowing high data rates.

 Even more specifically, the invention relates to an improvement of the coding method commonly called turbo-codes, as well as the associated decoding method.

 The general principle of turbo-codes (registered trademark) is presented in French patent FR-91 05280, entitled Error correction coding method to at least two systematic parallel convolutional encodings, iterative decoding method, decoding module and corresponding decoder.

 This document describes a first type of turbo-codes based on the use of systematic recursive convolutional codes.

 Turbo codes have the advantages of simplicity, consistency and flexibility of choice of performance for high performance.

 The use of turbo-codes requires iterative decoding whose performance increases during the iterations. To achieve near-optimal performance, it is necessary to perform the order of a dozen iterations which greatly penalizes the speed of decoding. Thus, a disadvantage of turbo codes of the prior art is that they have a high latency decoding, due to their nature.

 Another disadvantage of the turbo codes described above is that they are not suitable for a parallel decoding architecture.

 Parity codes are also known, including:

<Desc / Clms Page number 2>

 - LDPC codes described by R.G. Gallager in the document Low-density parity-check codes and published by MIT Press in 1963.

- generalized low density (GLD) codes described by RM Tanner in an article entitled A Recursive Approach to Low Complexity Codes (or, in French, a recursive approach to low complexity codes), published in IEEE Transaction On Information
Theory, Vol. IT-27 in September 1981.

 These codes have the same disadvantages as the turbo codes described above. In particular, they have a high decoding latency because they require a very large number of iterations. In addition, they are not suitable for parallel decoding of elementary codes constituents because they are irregular.

 Thus, for real-time applications such as voice and video, short frames are needed to decrease decoding latency. The decrease in the size of the frames causes a deterioration of the reliability of the decoded data. Indeed, it grows with the number of symbols taken into account in a frame. In addition, the algorithms used to decode a frame are based on a SISO type algorithm (from the English Soft Input Soft Output or Flexible Inputs and Flexible Outputs in French) (for example a MAP algorithm (or Maximum A Posteriori)) whose degree of parallelism is reduced and which therefore limits the maximum possible flow.

 Product turbo-codes consisting of several block concatenated codes according to at least two dimensions of a representative table of a corresponding product turbo-code are also known. Turbo-codes produced are presented in particular in the document Near Optimum Decoding of Product Codes (or Decoding close to the optimum of product codes) in French) written by R. Pyndiah, A. Glavieux, A. Picard and S. Jacq and published in the 1994 Globecom'94 colloquium in San Francisco (USA). They allow a greater degree of parallelism to decoding than

<Desc / Clms Page number 3>

 codes of the first type described above, several component codes can be decoded in parallel. Constituent codes used are linear algebraic codes. Nevertheless, the choice of a regular interleaver is not optimal in terms of performance.

 Furthermore, codes are known that make it possible to improve latency during decoding and described in the article Decoder-first code design (or code development first built on the basis of a decoder) written by E. Boutillon, J. Castura and F. Kschischang and published in Proceedings Proceedings of the 2nd International Symposium on Turbo Codes and Other Topics, pages 459 to 462 (Brest, France) in September 2000.

 In this paper, a decoder structure adapted to decode low density parity codes (LDPC) is described. According to the structure described, different memory banks containing data to be decoded supply in parallel decoders each associated with a parity code. The set of data corresponding to a parity codeword simultaneously enter into a basic decoder. In addition, elementary decoding is done in parallel. Thus, a low latency of decoding is obtained.

 Nevertheless, this technique of the prior art has the disadvantage of being relatively complex to implement.

 In addition, it requires many memory banks, the number of memory banks being equal to the number of data entering simultaneously in the elementary decoders, the product of the number of decoders by the size of a basic code word. Also, this technique is not suitable for large LDPC codes.

 This technique also has the disadvantage of not providing sufficient code quality for many applications.

 The invention in its various aspects is intended to overcome these disadvantages of the prior art.

 More precisely, an object of the invention is to provide a code in the form of a signal and an associated coding and / or decoding method as well as the

<Desc / Clms Page number 4>

 corresponding devices, which combine the advantages of a good bit error rate with low decoding latency and / or high throughput.

 More specifically, an object of the invention is to provide a signal code and an associated coding and / or decoding method and the corresponding devices, which combine the advantages of a good bit error rate with a low decoding latency and / or high throughput.

 Another object of the invention is to provide a code which has a high free distance for a given code length and therefore a large error correction capability.

 Yet another object of the invention is to provide a code and a method of coding and / or decoding associated and the corresponding devices, which are relatively simple to implement.

 For this purpose, the invention proposes an error correction code decoding method of the type associating a decoded data block with data encoded according to a global code comprising at least two constituent sub-codes, an irregular bipartite graph being associated to the global code, the decoding method being iterative and producing at each iteration a block of extrinsic data, each of the extrinsic data relating to one of the coded data, the method implementing a step of storing a block of data to decoding comprising said coded data and extrinsic data, the data block to be decoded being distributed in a plurality of disjoint memory banks, addressable independently, the method being remarkable in that: - at each iteration, it comprises, in addition, a step of feeding in parallel of at least two decoders among a plurality of decoders, each corresponding respectively to t to at least one of said sub-codes, with corresponding data to be decoded from the data block to be decoded,

<Desc / Clms Page number 5>

 data to be decoded being extracted in parallel from at least two of said memory banks to supply as many decoders, and each of the decoders being fed sequentially by the data to be decoded corresponding thereto.

 According to one particular characteristic, the method is remarkable in that at least one of the subcodes is a circular convolutional recursive code.

 According to one particular characteristic, the method is remarkable in that at least one of the sub-codes is a systematic code.

 Thus, the invention allows a use of short codes and thus a decrease in coding latency.

 In addition, the coding method allows improvement of the GLD codes using several constituent sub-codes.

 According to a particular characteristic, the method is remarkable in that at least two of the sub-codes are m-binary codes, m being an integer greater than or equal to 2 and in that an intra-symbol permutation is applied between minus two of the m-binary sub-codes.

 Thus, the invention applies not only to the binary component sub-codes, but also to the duo-binary component sub-codes or more generally to the m-binary component sub-codes.

 According to one particular characteristic, the method is remarkable in that it further comprises a step of switching each of the data to be decoded to one of the decoders, said data being switched in parallel and simultaneously so that each of between them feeds a decoder corresponding to him own.

 According to a particular characteristic, the method is remarkable in that said routing step itself comprises a step of circular permutation of a set comprising at least a part of the data to be decoded.

<Desc / Clms Page number 6>

 Thus, the invention makes it possible to improve the interleaving while using the maximum of parallelism related to the decoding in parallel of several component codes of the global code.

 According to one particular characteristic, the method is remarkable in that the circular permutation is a rotation which has a determined pitch as a function of the read rank of the data read in the memory banks.

 Thus, the interleaver is implemented simply and efficiently.

 According to a particular characteristic, the method is remarkable in that the switching step implements a step of addressing each of the memory banks so that data can be read in the memory bank in a predetermined order distinct from the writing order of said data in the memory bank.

 Thus, we have a wide variety of choices of interleavers and thus opportunities for optimization of the code obtained, while taking into account the maximum parallelism of different operations. This makes it possible to obtain a global code having very good performances.

 According to one particular characteristic, the method is remarkable in that - each line of a first data matrix representative of the data coded according to the global code is representative of data coded by a subcode constituting global code belonging to a first group subcodes; each line of at least one second matrix of data representative of the data coded according to the global code is representative of data encoded by a subcode constituting global code belonging to at least a second group of sub-codes; each matrix of the second data matrices being obtained from a transformation of said first matrix, the transformation comprising a permutation of at least a portion of the columns of the first matrix.

<Desc / Clms Page number 7>

 The notion of matrix is relatively theoretical and the person skilled in the art uses a matrix, in practice, in various forms and in particular by distinguishing sub-blocks in a data block, each sub-block corresponding to a matrix line being coded according to a subcode of the global code.

 According to a particular characteristic, the method is remarkable in that the transformation further comprises a permutation of at least a portion of the lines of the first matrix.

 Thus, the invention allows a large choice of interleavers (i.e., permutations) while maintaining a relatively simple implementation.

 According to one particular characteristic, the method is remarkable in that each line of a first matrix of data representative of the data coded according to the global code is representative of data coded by a subcode constituting global code belonging to a first group of sub-codes. -codes, some of the data of the first matrix, called non-significant data, not being significant; each line of a first subset of lines of at least one second matrix of data representative of the data coded according to the global code is representative of data coded by a subcode constituting global code belonging to at least a second group subcodes; a second subset of the rows of the second data matrix containing the insignificant data; each matrix of at least one second data matrix being obtained from a transformation of the first matrix, the transformation comprising a permutation of at least a portion of the columns of the first matrix.

 According to one particular characteristic, the method is remarkable in that it comprises a decoding test step implementing at least one stopping criterion so that when the at least one stopping criterion is verified

<Desc / Clms Page number 8>

 for at least one of the sub-codes, at least one of the decoders associated with the sub-code or codes stops decoding the sub-code or codes for which the stop criterion or criteria are checked.

 In this way, the definition of the global code allows the application of a stop criterion on the component codes in order to reduce over the iterations the number of component codes processed. This stopping criterion is, for example, based on the thresholding of the weighted decisions. Thus, the invention makes it possible to save calculating time of the decoding device and therefore of the energy, which is particularly useful when the decoding method is implemented within terminals operating on a battery.

 According to one particular characteristic, the method is remarkable in that it implements a step of reading coded data from an optical and / or magnetic medium and / or of transmitting said encoded data over an interference channel.

 According to one particular characteristic, the method is remarkable in that at least one of the sub-codes corresponds to interference between symbols representative of the decoded data block when the block is stored on the optical and / or magnetic medium and / or when the block is transmitted in an interference channel.

 Thus, the invention is particularly well suited to storage applications on an optical and / or magnetic medium, the storage channel itself being able to encode data stored according to a sub-code implementing interferences between symbols corresponding to the stored and / or transmitted data. One or more other sub-codes can then be implemented according to code techniques using, for example, circular recursive systematic codes.

 According to one particular characteristic, the method is remarkable in that it implements a step of receiving the coded data coming from a transmitter.

<Desc / Clms Page number 9>

Thus, the invention can advantageously be implemented for transmission-type applications on a noisy channel (for example, wireless radio channel, infra-red and / or acoustic)
According to a particular characteristic, the method is remarkable in that the data sets coded with each of the sub-codes are all different two by two.

Thus, the bipartite graph associated with the global code is irregular, in particular because the sub-codes of the global code are all different two by two.

 According to one particular characteristic, the method is remarkable in that at least two of the subcodes comprise at least two of the data to be decoded in common.

 Thus, the bipartite graph associated with the global code is irregular notably because two sub-codes of the global code have at least two data in common.

 According to one particular characteristic, the method is remarkable in that the global code is of code type produced with a non-uniform interleaver.

 Thus, the bipartite graph associated with the global code of the product code type is irregular, in particular because the associated interleaver is not uniform, that is, if the global code is symbolized by a two-dimensional matrix, the subnotes -codes do not just match the rows and columns of the matrix.

 According to one particular characteristic, the method is remarkable in that it further comprises a step of demultiplexing the decoded data block so as to supply at least two distinct recipients with data belonging to the demodulated data block.

 Thus, the invention can take into account data from different sources (for example, multimedia type). These data are grouped together to be coded to a larger size code than if the data from separate sources were coded separately. They thus benefit from a higher power of correction, each block of information corresponding to a

<Desc / Clms Page number 10>

 fraction or a subset of the rollers of the overall code. The larger size global code is therefore more efficient and its properties can therefore advantageously be exploited by the decoding method.

 According to a particular characteristic, the method is remarkable in that the degree of the information symbols of the decoded data block in the global code is not uniform.

 The degree of an information symbol in a global code is, here, by definition the number of arcs that are connected to this symbol in a representation of the code as a bipartite graph.

 Thus, the more information bits require a high level of protection, the higher their degree in the overall code will be. The protection of a symbol may, in addition, be improved by decreasing the yield of the constituent sub-codes to which the symbol is associated.

 The invention also relates to an error correction code coding method, of the type associating a source data block with data coded according to a global code comprising at least two component sub-codes, an irregular bipartite graph being associated with said global code. , the coded data block being intended to be transmitted to at least one receiver and / or stored on a data medium, remarkable in that the coded data block is intended to be decoded by the iterative decoding method.

 The invention further relates to an error correction code encoding method of the type associating a source data block with data encoded according to a global code comprising at least two constituent sub-codes, an irregular bipartite graph being associated global code, the coded data block being intended to be transmitted to at least one receiver and / or stored on a data medium, the coded data block being intended to be decoded by an iterative decoding method, remarkable in that the method further comprises a step of supplying in parallel at least two coders among a plurality of coders, each corresponding respectively to at least one of the sub-codes, with data to be coded corresponding to said block of data to be coded , data to

<Desc / Clms Page number 11>

 code being extracted in parallel from at least two memory banks to supply as many encoders.

 According to one particular characteristic, the method is remarkable in that each of the coders is fed sequentially by the data to be encoded corresponding to it.

 According to one particular characteristic, the method is remarkable in that at least one of said subcodes is a circular convolutional recursive code.

 According to one particular characteristic, the method is remarkable in that at least one of the sub-codes is a systematic code.

 According to a particular characteristic, the method is remarkable in that at least two of the sub-codes are m-binary codes, m being an integer greater than or equal to 2 and in that an intra-symbol permutation is applied between minus two of the m-binary sub-codes.

 According to a particular characteristic, the method is remarkable in that it further comprises a step of switching each of the data to be coded to one of the coders, the data being switched in parallel and simultaneously so that each of they feed an encoder corresponding to him own.

 According to a particular characteristic, the method is remarkable in that the routing step itself comprises a circular permutation step of a set comprising at least a part of the data to be coded.

 According to one particular characteristic, the method is remarkable in that the circular permutation is a rotation which has a determined pitch as a function of the read rank of the data read in the memory banks.

 According to a particular characteristic, the method is remarkable in that the switching step implements a step of addressing each of the memory banks so that data can be read in the memory bank in a predetermined order distinct from the writing order of the data in the memory bank.

 According to one particular characteristic, the method is remarkable in that each line of a first matrix of data representative of said data

<Desc / Clms Page number 12>

 encoded according to the global code is representative of data encoded by a subcode constituting global code belonging to a first group of subcodes; each line of at least a second data matrix representative of said global code-encoded data is representative of data encoded by a global code sub-code belonging to at least a second group of sub-codes; each matrix of one or both data matrices being obtained from a transformation of the first matrix, the transformation comprising a permutation of at least a portion of the columns of the first matrix.

 According to a particular characteristic, the method is remarkable in that the transformation further comprises a permutation of at least a portion of the lines of the first matrix.

 According to one particular characteristic, the method is remarkable in that each line of a first matrix of data representative of the data coded according to the global code is representative of data coded by a subcode constituting global code belonging to a first group of sub-codes. -codes, some of the data of the first matrix, called non-significant data, not being significant; each line of a first subset of lines of at least one second data matrix representative of the global code encoded data is representative of data encoded by a subcode constituting global code belonging to at least a second group of subcodes; a second subset of the rows of the second data matrix containing the insignificant data; and each matrix of one or both of the data matrices being obtained from a transformation of the first matrix, the transformation comprising a permutation of at least a portion of the columns of the first matrix.

 According to one particular characteristic, the method is remarkable in that it implements a step of writing said coded data on a medium

<Desc / Clms Page number 13>

 optical and / or magnetic and / or data transmission coded on an interference channel.

 According to one particular characteristic, the method is remarkable in that at least one of the sub-codes corresponds to interference between symbols representative of the decoded data block when the block is stored on the optical and / or magnetic medium and / or when the block is transmitted in an interference channel.

 According to one particular characteristic, the method is remarkable in that it implements a step of receiving the coded data coming from a transmitter.

 According to a particular characteristic, the method is remarkable in that the data sets coded with each of the sub-codes are all different two by two.

 According to one particular characteristic, the method is remarkable in that at least two of the sub-codes comprise at least two data coded in common.

 According to one particular characteristic, the method is remarkable in that the global code is of code type produced with a non-uniform interleaver.

 According to one particular characteristic, the method is remarkable in that it further comprises a step of multiplexing at least two data blocks each coming from two different sources so as to form the source data block.

 According to one particular characteristic, the method is remarkable in that the degree of the information symbols of said decoded data block in the global code is not uniform.

 The invention also relates to a signal representative of a block of data coded according to a global code comprising at least two constituent sub-codes, an irregular bipartite graph being associated with the global code, the coded data block being intended to be decoded by the iterative decoding method.

<Desc / Clms Page number 14>

 The invention also relates to an error correction code decoding device, of the type associating a decoded data block with data coded according to a global code comprising at least two constituent sub-codes, an irregular bipartite graph being associated global code audit, the associated decoding method being iterative and producing at each iteration a block of extrinsic data, each of the extrinsic data relating to one of the coded data, the device implementing storage means of a data block to decode comprising said coded data and said extrinsic data, the data block to be decoded being distributed in a plurality of disjoint memory banks, independently addressable, the device being remarkable in that at each iteration, it further comprises means supplying in parallel at least two decoders among a plurality of decoders, corresponding to each n respectively to at least one of the sub-codes, by corresponding data to be decoded from the data block to be decoded, data to be decoded being extracted in parallel from at least two of said memory banks to supply as many decoders, and each of the decoders being fed sequentially by the data to be decoded corresponding to it.

 In addition, the invention relates to an error correction code encoding device, of the type associating a source data block with data coded according to a global code comprising at least two constituent sub-codes, an irregular bipartite graph being associated with the global code, the coded data block being intended to be transmitted to at least one receiver and / or stored on a data medium, the coded data block being intended to be decoded by an iterative decoding device, the device being in this the device further comprises means for supplying in parallel at least two coders among a plurality of coders, each corresponding respectively to at least one of the sub-codes, with corresponding data to be coded from the data block to encode,

<Desc / Clms Page number 15>

 data to be coded being extracted in parallel from at least two of the memory banks to supply as many encoders.

 The advantages of the coding methods, the signal, the coding device and the decoding device are the same as those of the decoding method. They are not detailed further.

 Other features and advantages of the invention will appear more clearly on reading the following description of a preferred embodiment, given as a simple illustrative and nonlimiting example, and the appended drawings, among which: FIG. 1 shows a particular embodiment of the coding operation of the invention which consists in using four wheels making it possible to form a code (12, 8) according to the invention; FIGS. 2 and 3 illustrate two particular embodiments of the coding operation, according to the invention; FIG. 4 shows an embodiment of an encoder making it possible to construct irregular codes, according to the invention; FIG. 5 describes an encoder whose architecture makes it possible to implement the codes according to FIGS. 1 to 4; FIG. 6 illustrates a decoder adapted to decode roulette codes described with reference to FIGS. 1 to 4; and FIG. 7 illustrates an application of the invention to data storage.

 The general principle of the invention relies on codes allowing modular decodings associated with different levels of quality and different rates according to the number of modules implemented in a decoder.

 The invention applies in all cases where it is necessary to transmit and / or store digital information with a certain level of reliability.

A preferred field of application of the invention is that of digital transmission over a noisy channel. For example, the invention can be used for transmitting and receiving data whose reliability and throughput is important.

<Desc / Clms Page number 16>

 such as satellite, fiber optic, radiofrequency channel, cable or physical media information

 The invention is also applicable for applications for storing data on optical and / or magnetic media, for example.

 The invention also makes it possible to code and decode any signal, in particular a sound, video or data signal or the concatenation of several heterogeneous signals.

 More specifically, the general principle of the invention is based on a coding of a sequence of K information symbols, using R component codes making it possible to obtain a code word of length N strictly greater than the number K.

 The global code therefore comprises constituent codes called, here, casters by analogy to a possible embodiment using systematic and circular recursive convolutional codes.

 For each value of an integer i varying from 1 to R, the ith roulette R, of a global code is defined, in the general case, by the ordered set M, of the symbols of the constituent code word (subcode of the global code) considered. By definition, m, represents the cardinality of Mi. The order of the symbols of M, is given by the nature of the constituent code and / or by the order induced by the decoding algorithm.

 The global code is made from the wheels by a bipartite graph.

This graph comprises: a line comprising the N symbols of the global code word; - a line containing the R wheels; and a set of M arcs connecting the R rollers with the N symbols of the code word.

 According to the invention, each wheel has mi connections. It is therefore connected to m, symbols of the code word. The number of arcs M is therefore equal to the sum of the mi peri ranging from 1 to R.

<Desc / Clms Page number 17>

 It is recalled that many codes can be described by a general bipartite graph and called Tanner codes which consist of describing a code using a bipartite graph that associates constraints to a group of information symbols. When the constraint is a simple parity code, the LDPCs described by RG Gallager are obtained in the document Low-density parity-check codes and published by MIT Press in 1963. If constraint is a block code (n, k), so we obtain a generalized low density code (GLD) described by RM Tanner in an article entitled A recursive approach to low complexity codes (or, in French, a recursive approach to codes low complexity), published in IEEE Transaction On Information Theory, Vol. IT-27 in September 1981.

 A turbo-code (with two component codes) is generally described by a bipartite graph that contains two constraints, respectively associated with a non-interlaced encoder and an interleaved encoder.

 Note #i (Mi) the ordered set of symbols of the global code word connected to the wheel R ,. Note also, for the sake of simplicity #i (K), the number of the symbol of the codeword which is connected with the kth input of the wheel Ri. When decoding the wheel i, #i (k) will be the kth symbol used by the decoder. The set {#i (k)} represents the unordered set of symbols of the global code word connected to the wheel Ri.

According to the invention, the bipartite graph satisfies the following conditions: the sets #i (Mi) satisfy a particular constraint, in order to allow the decoding of the component codes with a degree
P parallelism; and - the bipartite graph is irregular.

 In order to satisfy the first of the above conditions, sets #i (Mi) are chosen to allow decoding iteration (processing of the R rollers) in one or more steps (the number of steps). being denoted T) and the rollers of each step t (t varying from 1 to T) being

<Desc / Clms Page number 18>

 decoded by at least two decoders in parallel (the number of decoders being denoted by P, with P being equal to the ratio R on 1).

T-1 ) comprend un décodage des roulettes Rt~l jp+1, ... , Rt,p chacune des P roulettes R(t-DP+n ### RtP. étant associées à un des P décodeurs. L'étape T comprend un décodage des dernières roulettes (RR-P+1, ... , RR). The first step of each iteration comprises a decoding of the first P rollers R1, R2 ... RP by P decoders in parallel, each of the first P casters being associated with one of the P decoders. The second step of each iteration includes a decoding of the following P rollers RP + 1, ..., R2P. In a general way, the ith step of each iteration (for t varying from 1 to

T-1) comprises a decoding of the rollers Rt ~ l jp + 1, ..., Rt, p each of the P rollers R (t-DP + n ### RtP) being associated with one of the P decoders. T includes a decoding of the last wheels (RR-P + 1, ..., RR).

 For each step t of the T steps of each iteration, P decoders are used in parallel, each associated with a memory bank BM1, BM2, .., BMP.

cycle symbole k de l'étape t, les P décodeurs traitant les P symboles (t~l,pT1(k) , (t-1).P+2lk)> ID,.p(k réalisent des accès mémoires en parallèle et indépendants dans P bancs distincts. Each of the P memory banks contains the data necessary for the execution of one of the P decoders as well as the data produced by one of the P decoders. Every

symbol cycle k of the step t, the P decoders processing the P symbols (t-1, pT1 (k), (t-1) .P + 2lk)> ID, .p (k realize memory accesses in parallel and independent in P separate benches.

les P symboles (t,.p+1(k) , (t I).P2(k)" <'PtAk)) aux P bancs (BM,, BM2,..., BMp) ; c'est-à-dire que l'on peut trouver une répartition des symboles dans les P bancs mémoires, compatible avec les ensembles #(Mi) et garantissant des accès mémoires en parallèle ; - soit, dans le cas contraire, si les symboles #i(k) et #j(k) correspondent au même banc mémoire BMx, alors ils sont égaux

D(k) = <'Pik)) ; la donnée lue dans un banc est alors unique et elle est dupliquée avant d'être transmise aux roulettes R, et Rj. The constraint on sets #i (Mi) is thus the following: at each symbol cycle k: either there exists a bijection which associates, for all the values of t ranging from 1 to T and for all the values of k ranging from 1 at M,

the P symbols (t, .p + 1 (k), (t I) .P2 (k) "<'PtAk)) at P banks (BM ,, BM2, ..., BMp); that is to say that we can find a distribution of symbols in P banks memory compatible with sets # (Mi) and guaranteeing access memories in parallel - or, if not, if the symbols #i (k) and #j (k) correspond to the same BMx memory bank, so they are equal

D (k) = (Pik)); the data read in a bank is then unique and it is duplicated before being transmitted to the wheels R, and Rj.

<Desc / Clms Page number 19>

 Thus, when using P decoders in parallel, one multiplies by a factor equal to P the surface of the circuit dedicated to the computation (computational part) with respect to a single decoder. On the other hand, the time of an iteration is divided by P.

The decoding latency is therefore divided by P. The bit rate is therefore multiplied in the same proportions. In summary, for a linear increase of the surface, one obtains a linear increase of the flow.

{t(Mz) {<t>i)} pour tout i différent de j, i et j variants de 1 à R) (alors que cette propriété n'est pas vérifiée pour les turbo-codes classiques tels que décrits dans le brevet français n FR-91 05280) ; et

- soit il existe deux ensembles {<t>,(M,)} et {<t>i)} pour deux valeurs i et j distinctes dont l'intersection contient au moins deux éléments (sous forme mathématique : il existe i différent

dey tels que c({<P,(M,)} rl {<t>i)}) 2, où card représente la fonction cardinal d'un ensemble) ; - soit le code vérifie la propriété : soitr l'ensemble des couples d'indice de roulette (i,j,k) vérifiant : - {#i(Mi)} n {#j(Mj)} = 0 (où 0 représente l'ensemble vide);

- card( { <t>,(M)} fil {èJ?iMk)}) = 1 ; et - c({0/M,)} n {<t>k(Mk)}) = 1 ; alors : - si {#i(Mi)} # {#k(Mk)} = {#i(h)} (où #i(h) représente un point unique d'indice h pour la roulette Ri; et Moreover, according to the invention, the bi-partite graph is irregular. This irregularity makes it possible to generate a pseudo-random code and thus to obtain good performances. It is characterized by the following two properties: - the sets {#i (Mi)} are all different two by two (either

{t (Mz) {<t> i)} for all i different from j, i and j variants from 1 to R) (while this property is not verified for conventional turbo-codes as described in the patent French FR-91 05280); and

- or there are two sets {<t>, (M,)} and {<t> i)} for two distinct values i and j whose intersection contains at least two elements (in mathematical form: there are i different

dey such that c ({<P, (M,)} rl {<t> i)}) 2, where card represents the cardinal function of a set); either the code checks the property: let all the pairs of roulette index (i, j, k) satisfying: - {#i (Mi)} n {#j (Mj)} = 0 (where 0 represents the empty set);

- card ({<t>, (M)} wire {èJ? iMk)}) = 1; and - c ({0 / M,)} n {<t> k (Mk)}) = 1; then: - if {#i (Mi)} # {#k (Mk)} = {#i (h)} (where #i (h) represents a single point of index h for the wheel Ri, and

<Desc / Clms Page number 20>

- si {<)}n{<(MJ} = 1 IDJ(h')l (où <Pih') représente un point unique d'indice h' pour la roulette Ri); il existe au moins un triplet (i,j,k) dans l'ensemble # tel que h est différent de h'.) (dans ce cas, on peut obtenir des codes ayant de meilleures performances que les codes à concaténation parallèles qui utilise un entrelaceur uniforme, connus en soi, pour lesquels cette dernière condition n'est pas vérifiée et qui sont notamment décrits dans l'article Code construction and decoding of Parallel concatenated tail-biting codes (ou en français construction de code et décodage de codes circulaires à concaténation parallèle ) écrit par C. Weiss, C.
Bettstetter et S. Riedel et paru dans la revue IEEE transaction on information theory , Vol. 47, N 1 de janvier 2001)
Selon un autre aspect de l' invention, pour la mise en #uvre du décodage, afin de limiter la latence de décodage, on applique un critère d'arrêt tel que décrit dans le document Stopping rules for turbo decoders (ou critère d'arrêt pour des turbo-décodeurs en français ) écrit par A. Matache, S. Dolinar et F ; dans le rapport TMO progress Report 42-142 en août 2000 et édité par le JPL (Jet Propulsion Laboratory). Néanmoins, selon des techniques connues, le critère d'arrêt est appliqué à partir d'un découpage d'une trame à décoder en fenêtres de taille égale. Le critère d'arrêt est alors appliqué sur les fenêtres de façon à arrêter le traitement des fenêtres ayant convergé et ainsi, diminuer le nombre de fenêtres traitées lors des itérations suivantes. Un tel dispositif présente deux inconvénients :

if {<)} n {<(MJ} = 1 IDJ (h ') 1 (where <Pih') represents a single point of index h 'for the wheel Ri); there is at least one triplet (i, j, k) in the set # such that h is different from h '.) (in this case, codes with better performance can be obtained than parallel concatenation codes that use a uniform interleaver, known per se, for which the latter condition is not verified and which are described in particular in the article Code construction and decoding of Parallel concatenated tail-biting codes (or in French code construction and decoding of circular codes parallel concatenation) written by C. Weiss, C.
Bettstetter and S. Riedel and published in the journal IEEE transaction on information theory, Vol. 47, No. 1 of January 2001)
According to another aspect of the invention, for the implementation of the decoding, in order to limit the decoding latency, a stopping criterion is applied as described in the document Stopping rules for turbo decoders (or stopping criterion for turbo-decoders in French) written by A. Matache, S. Dolinar and F; in report TMO Progress Report 42-142 in August 2000 and published by the JPL (Jet Propulsion Laboratory). Nevertheless, according to known techniques, the stopping criterion is applied from a division of a frame to be decoded into windows of equal size. The stop criterion is then applied to the windows so as to stop the processing of the converged windows and thus reduce the number of windows processed during the following iterations. Such a device has two disadvantages:

<Desc / Clms Page number 21>

 the window division is arbitrary and is not generally adapted to the configuration of the residual errors in the decoding process; and splitting a continuous window frame requires complex and suboptimal management of edge effects.

 According to the invention, the code, by construction, makes it possible to overcome these disadvantages of the prior art by improving the use of the weighted decision thresholding stop criterion which is applied for each of the R wheels.

Indeed, according to the invention, the window on which the stop criterion is applied is not chosen arbitrarily, but on the contrary, this window corresponds exactly to a constituent code.

 This stopping criterion can be used for low bit rate applications where the maximum of parallelism allowed by the code is not used. Note that the use as described below is not limited to the codes implementing a decoding method according to the invention but applies to any code comprising at least one type of subcode wheel and interlace information symbols that can be any. Thus, for example, one can choose to implement a single decoder that successively decodes the R rollers. Thus, during the decoding of a roulette wheel, a stop criterion is applied as described in the report cited above to each wheel (and not according to the state of the art only to the global code). If the criterion is checked for at least one roulette, this roulette will not be decoded during the next iteration, thus saving energy consumption and resources.

Similarly, if all the decoders used in parallel converged, we can go directly to the next decoding step, which allows a decoding latency and consumption gain. Moreover, if the criterion is verified for at least one roulette, an associated decoder can be used to decode a data block and, in particular, for another roulette for which a data block is waiting for decoding. Thus, the decoding resources can be distributed for optimization according to the needs.

<Desc / Clms Page number 22>

 The method thus defined therefore has three advantageous properties: the graph is irregular, which makes it possible to effectively emulate a random code, and thus to obtain good performances; a simple implementation of the high-speed decoding method is allowed by the simultaneous decoding of several wheels, the data being taken into account sequentially by each of the decoders associated with a wheel; and the method allows efficient use of stopping criteria based on the structure of the code and not on an arbitrary division of the frame into packets of equal size.

 Moreover, the coding and decoding methods have several variants, in particular: on the bipartite graph; and / or - on the wheels.

 According to different variants on the bipartite graph, the code can be: systematic; or - unsystematic (in this case, when the constituent codes are parity codes, low density parity code (or LDPC) codes are obtained).

 If the code is systematic, the rollers Ri of the code (for i varying from 1 to R) are then defined by the following parameters: the ordered set C, systematic information symbols of the roulette wheel; Moreover, c, represents the number of systematic information symbols of the roulette R, (or c, = card (Ci)). This number is called the circumference of the roulette R; the ordered set E, redundancy symbols of the roulette wheel; the number of redundancy symbol of the wheel Ri is denoted e, (or e, = card (E,)); and the quantity mi equal to the sum of the circumference c, and the number e, defines the size of the roulette R, (or m, = c, + e,).

<Desc / Clms Page number 23>

 Thus, an ordered set M, consisting of the ordered union of the sets Ç and E, (or M, = (Ci, Ei)) of m, (equal to the sum of ci and e,) symbols is a code word / ?, if the passage, in the order of c, C symbols, in the encoder associated with the wheel R, generates the e, redundancy symbols E ,.

 Note deg (s) the degree of the symbol number s of the codeword, that is, the number of arcs that are connected to this symbol.

 The global code dimension is the maximum of the degrees of the symbols of the code word. Preferably, the number of steps is greater than or equal to the size of the global code.

 Note #i (Ci) the ordered set of symbols of the code word connected to the information of the wheel Ri. Thus, #i (k) represents the symbol number of the codeword which is connected to the kth input of the wheel Ri. The set {#i (k)} represents the unordered set of symbols of the codeword connected to the information of the wheel Ri.

 According to variants of the invention applied to the rollers: the rollers may correspond to any block codes, for example of the BCH or Reed-Solomon type; the constituent codes may be convolutional and particularly circular recursive codes that solve the problem of closing the trellis associated with the corresponding decoder without deterioration of the spectral efficiency of the coding; in this case, we can associate with each wheel: the generating polynomials of the systematic recursive convolutional code associated with Ri; and a punching pattern associated with the wheel Ri making it possible to obtain the efficiency ri; the codes used may be based on binary, dumbinary or n-tuplets symbols; the codes (m1, ci) may be identical for each of the rollers R,; and or

<Desc / Clms Page number 24>

 punching may be identical for all rollers R, and thus allow a uniform yield ri.

 Alternative implementations of the bipartite graph are also possible: the number of rollers processed during a decoding step may be different for each step: for a given architecture, it is possible in particular to manage the effects of edges due to the use of words of variable length (for example, the number of wheels is not necessarily a multiple of P); - according to the constraints of the technology, the type of realization of roulette (insertion of pipe-line, type of realization of the algorithm of the decoder with soft input - flexible exit), and constraints of flows of the application, the number P 'of memory banks and the number P of wheels made in parallel may be different; the code must be constructed for values of P 'and P greater than or equal to 2.

 According to the invention, the code can be two-dimensional. Thus, to the generic code described above, the following specifications are added: the number of rollers R is even; - all the rollers have a circumference equal to C; the degree deg (k) of each of the K information symbols is two and each information symbol is connected to two separate wheels.

 The rollers can then be separated into two groups such that in each group, the degrees deg (k) of the information symbols are 1.

 According to a particular variant producing a parallel concatenation, the code words and sets IIi (Ci) are then defined as follows: the K symbols of the code word are written line by line in a matrix having R / 2 lines and C columns; the lines define the sets IIi (Ci) (i varying from 1 to R / 2) of the first group. They are elements of the corresponding constituent codes;

<Desc / Clms Page number 25>

then, for each of the lines, any permutation between the symbols of said line is carried out; - Any permutation is performed on each of the columns. In particular, the permutation on each of the columns may be a circular rotation; the lines obtained define the sets IIi (Ci) (i varying from R / 2 to
R) of the second group. They are coded by the corresponding constituent codes.

 This two-dimensional code requires P equal to R / 2 memory banks each being filled by a line of the non-permuted matrix. When the component codes are circular recursively systematic, one can use the organization of the following memory: at each address of the memory, the data relating to the systematic symbol (information symbol coming from the channel, extrinsic information produced by the decoders) as well as the redundancy symbols from the corresponding channel. Each group can then be decoded simultaneously by P decoders in parallel to T (equal to 2) steps. The method of permutation of the matrix ensures during the decoding of the second group of wheels that the memory accesses are performed on separate memory banks. The permutations performed on the lines correspond to a temporal interleaving of the data (that is to say, the permutations on the lines are carried out according to the decoding order by reading the symbols of each memory). The permutations associated with the columns correspond to a spatial interleaving of the data. This interleaving is performed physically by a permutation network between the P banks and the P decoders.

 According to the invention, from the code described above, it is possible to construct an irregular code. The 2P component codes used each have a circumference equal to the sum C + D.

 The sets IIi (Ci) defining the constituent codes are obtained from the rows of the non-permuted matrices and the permuted lines defined above: each constituent code is associated with the sequence of the symbols

<Desc / Clms Page number 26>

 information of a line to which the first D information symbols of the next line are added; for the last line, we add the D information symbols of the first line.

 The irregular code thus constructed has symbols of degree equal to three which are located at the ends of the rollers. The other symbols are all of degree equal to two. This coding scheme makes the code more robust to transmission errors and allows a faster convergence of the decoder.

According to another variant of the invention implementing a two-dimensional code by serial concatenation, with a different number of rollers for the first dimension and the second dimension, the following specifications are added: the number of rollers worth R which is equal to the sum of P1 and P2 with a product of P1 and P2 equal to N and P1 being greater than or equal to
P2; there are therefore PI wheels of circumference equal to P2 and P2 rollers of circumference equal to P1; all degrees deg (k) of the N information symbols are equal to 2 and therefore each information symbol is connected to two separate wheels.

 The casters are separated into two groups of respectively Pl and P2 casters.

The coding method and the sets #i (Mi) of the bipartite graph are then defined as follows: the N symbols of the code word are written line by line in a matrix having P, lines and P2 columns; to this matrix, we add a number of empty columns equal to the difference of P1 and P2 (either
P1-P2 in condensed form); the lines define the sets #i (Mi) (for i ranging from 1 to P) of the first group; they are then coded by the corresponding constituent codes of length equal to P2.

 for each of the lines, then, any permutation between the symbols of said line is carried out; these permutations do not

<Desc / Clms Page number 27>

 are not all identical (alternatively, these permutations are identical and can be advantageously removed); these permutations distribute uniformly over the columns, the symbols of the empty columns of the preceding matrix; a permutation Oj is performed on each of the P1 columns of the resulting matrix, respecting the following constraint: after permutation, the empty symbols of the preceding matrix are distributed over the P1-P2 lines; the non-empty lines obtained define the sets #i (Mi) (for i ranging from P to P1 + P2) of the second group; they are coded by constituent codes.

 The intra-line permutation presented in the preceding examples (dimension code 2, irregular codes and product codes) may in a preferred embodiment be chosen from the following permutations (the size of the memory, the line being C): the symbol k (k varying from 0 to C) of the wheel corresponds to the symbol located at the address l equal to the product ak modulo C (l = ak mod C) of the memory where a and C are prime between them. Preferably, in this case, a is close to the square root of C; the symbol k (k varying from 0 to C) of the wheel corresponds to the symbol located at the address l equal to the sum of the product ak and ss modulo C (l = ak + ss mod C) of the memory where a and C being prime between them.

 Preferably, in this case, a is close to the square root of C.

 The invention also relates to the coding, decoding and interleaving devices defining the sets #i (Mi) or Il, (Ç).

 These devices are presented with reference to FIGS. 5, 6 and 7.

 In relation to FIG. 1, the general principle of the code according to a particular embodiment of the invention is presented.

 The code is composed of four constituent codes 14, 15, 16 and 17, two codes of which can be coded and / or decoded simultaneously according to the invention.

<Desc / Clms Page number 28>

Each of these constituent codes is a circular systematic recursive convolutional code: - of circumference equal to three; - of size equal to four; and - defining a global code of size K equal to eight and length N equal to twelve.

 Each constituent code is represented by a wheel, which has: - three ordered inputs el, e2 and e3 each associated with an information symbol, and - an output associated with a redundancy symbol.

The degree of each of the symbols is not uniform: the degree of the redundancy symbols (symbols 9 to 12) is equal to one; and - the degree of information symbols is equal to one (for symbols
2, 3, 7 and 8) or two (for symbols 1, 4, 5 and 6).

 The set of symbols 1 to 12 of the code 13 is stored in two memory banks BM, and BM2 which respectively contain the following symbols: - 1, 2, 3, 4, 9 and 10 (memory bank BM,); and - 5, 6, 7, 8, 11 and 12 (BM2 memory bank).

 During the coding and / or decoding operation, the wheels are coded and / or decoded with a degree of parallelism equal to two. Indeed, for each of the four symbols used in the constituent code, the following memory accesses are made: - for a first group consisting of the wheels 14 and 15: - from the symbol el, we access the symbols 1 or 6; from symbol e2, symbols 5 or 4 are accessed; from symbol e3, symbols 3 or 8 are accessed; from symbol s, symbols 9 or 11 are accessed; - For a second group consisting of casters 16 and 17: - from the symbol el, we access the symbols 5 or 2;

<Desc / Clms Page number 29>

from the symbol e2, symbols 6 or 1 are accessed; from symbol e3, symbols 4 or 7 are accessed; from the symbol s, symbols 10 or 12 are accessed;
At each symbol cycle, separate memory banks are accessed and thus a degree of parallelism equal to two is used.

 According to an alternative embodiment of the invention described in FIG. 2, the overall coding operation is based on two distinct groups each comprising P rollers Ri applied globally to the same data.

 The circumference of each of the rollers Ri is equal to C. The K information symbols to be encoded are written line by line in a matrix 20 with P rows and C columns (K is equal to the product PxC).

 Each line is coded by a constituent code 21, for example of the circular systematic recursive convolutional code type. The ith line of the matrix 20 comprises the N given symbols of a codeword 21 (systematic information and redundancy) corresponding to a rollers R1. According to one variant, the ith line of the matrix 20 comprises the K systematic information symbols of a codeword 21 corresponding to a wheel Ri (the redundancy symbols not being represented).

 Then, a permutation of the symbols of each of the lines is performed by an interleaver (D 22. This interleaver is identical for each of the lines, According to a variant of the invention, this interleaver is different for each line.

 According to a variant not shown, the interleaver # is equal to the identity for all the lines and can therefore be deleted.

 On each of the columns of the matrix 23 is performed a permutation Oi 24 so that after permutation the C symbols of each row of the matrix 25 come from different lines of the matrix 23 (before permutation).

The permutation Oi 24 is circular, according to the preferred embodiment, or non-circular according to an alternative embodiment.

 The lines of the resulting matrix are encoded by P component codes 26 which may be circular recursive convolutional codes. The

<Desc / Clms Page number 30>

 The second row of the matrix 25 comprises the C systematic information symbols corresponding to a wheel Ri + 5.

 Thus, if the first (respectively second) line of the matrix 20 comprises a sequence that begins with the bits a, b and c (respectively d, e, f) as illustrated in FIG. 2, these bits will be interleaved by the permutation # 22. Thus, for example, the bits a, b and c will respectively be placed at the fourth, seventh and second positions of the first line of the matrix 23.

Similarly, the bits d, e and f will respectively be placed at the fourth, seventh and second positions of the second row of the matrix 23. The data of the fourth, seventh and second columns of the matrix 23 will then be respectively interlaced by the permutations. O4, O7 and O8.

 Thus, the matrix 25 is obtained in which, for example: the datum d is located at the fourth position of the first line; - The data e are respectively located in second and seventh position of the second line; - The data f and a are respectively located in second and fourth position of the fourth line; and the datum b is located in the seventh position of the last line.

 Thus, the data a, b and c (respectively d, e and /) which are placed in the same row of the matrices 20 and 23 are, after the permutations Oi, located in distinct lines of the matrix 25. In this way, while before permutation, they are coded with the same wheel R, (respectively R2), after permutation, the data a, b and c (respectively d, e and /) are coded with separate wheels.

 The association of the K information symbols and the redundancy symbols produced by the 2P constituent codes forms the code word of length N.

 According to another variant embodiment of the invention described in FIG. 3, the bipartite graph is separated into two groups. The first group contains P1

<Desc / Clms Page number 31>

 roulettes of circumference P2 and the second group comprises P2 wheels circumferences Pl, with P1 greater than or equal to Pz and the product Pl by P2 equal to N.

 The N information symbols of the global codeword are written line by line in a square matrix 30 on the side Pl of which only the first P2 columns are filled. The last P1- P2 columns remain empty with a value symbolized by V.

 Each of the P1 lines is coded by a constituent code 31, the ith line of the matrix 30 corresponding to a roulette R1 of length P2. Each of the codes 31 is, for example, a circular systematic recursive convolutional code (CRSC).

 Then, a permutation of the symbols of each of the lines is performed by an interleaver #i 32. Each of the P1 interleavers has the following property: after permutation, the symbols V of the matrix 30 are distributed uniformly over all the columns of the matrix 33. Thus, the same number of symbols V is obtained on each of the columns of the matrix 33.

 We perform a permutation Oj on each of the Pl column of the matrix 33. These rotations have the following property: after rotation, the symbols V of the matrix 33 are found in the Pl-PZ last rows of the matrix 33.

 The P2 non-empty lines of the resulting matrix 35 are then encoded by P2 constituent codes 36 of length P1.

 The P1 + P2 component codes define the N symbols of the global code word.

 This code requires P equal to Pl memory banks, each being filled by a line of the non-permuted matrix. The first (respectively second) group can then be decoded simultaneously by P1 (respectively P2 decoders in parallel) The method of permutation of the matrix ensures during the decoding of the second group of wheels that the memory accesses are performed on separate memory banks.

 According to another variant embodiment of the invention described in FIG. 4, the coding operation makes it possible to construct an irregular code from K (equal to 24) information symbols. This code consists of eight component codes Ri (i

<Desc / Clms Page number 32>

 being between 1 and 8) 411, 412, 413, 414, 441, 442, 443 and 444 of yield r. Each constituent code is of circumference equal to eight.

 The information symbols are stored in four memory banks 401 to 404.

 Interleaving 420 is defined in a manner similar to that presented with reference to FIG.

The code is then defined as follows: information symbols are written line by line in an M matrix with four rows and six columns; the lines of the matrix M are put end to end to form a circular frame 400, which will be coded by the four rollers R1 411,
R2 412, R3 413 and R4 414. The eight information symbols of the wheel R1 411 (respectively R2 412, R3 413 and R4 414) consist of the symbols of the line 401 (respectively 402, 403 and
404) of the matrix M and the first two symbols of the roulette R2
412 (respectively R3 413, R4 414 and R1411); - We perform a permutation 420 of the four lines and a permutation of the six columns of the matrix M to obtain a permuted matrix M '; - The lines of the matrix M 'obtained are put end to end to form a circular frame 430, which will be coded by the four wheels Rs
441, R6 442, R7 443 and R8 444. The eight information symbols of the wheel R5 415 (respectively R6 416, R7 417 and Rs 418) consist of the symbols of the line 431 (respectively 432.433 and
434) of the permuted matrix M and the first two symbols of the wheel R6 416 (respectively R7 417, R8 418 and R5 415).

 The permutation acting on the columns of the matrix M is chosen such that during the decoding of the rollers 441 to 444, the memory accesses can be performed simultaneously on separate memory banks. This permutation on the columns is for example the identity, a rotation on the

<Desc / Clms Page number 33>

 columns or, more generally, any permutation respecting the property related to memory access previously stated.

 After coding, the twenty-four information symbols and the redundancy symbols are transmitted on the channel.

 FIG. 5 presents an embodiment of a global encoder 500 using P recurrent circular recursive systematic convolutional encoders (CRSCx) 571, 572, 57i,..., 57P and whose architecture makes it possible to generate the codes which are notably described in FIG. FIGS. 1, 2 and 4.

 The global coder 500 accepts on an input 550 blocks of K information symbols which are also presented on an output 553 (the overall code is systematic) and supplies, in addition, on an output 590 blocks of redundancy symbols from the coding blocks of information.

The global encoder 500 comprises: the input 550 and the output 553 of information symbols; the output 590 of redundancy symbols; a control unit 504;
P memory banks (denoted BMJ 561 to 56P) - elementary encoders of the Recursive Convolution type
Circular Systematics (or CRSC) 571 to 57P; a PILk 510 permutation module; a circular rotation module RC (OJ 530; a control module Ok 520 of the module 530; a perforation module 540.

 The control unit 504 manages the operation of the entire device 500. This control unit produces the following control signals: a command 501 for zeroing the circular recursive convolutional encoders 571 to 57p; circular recursive convolutional code is obtained by a first pre-coding, with an initial state of the null encoder, which makes it possible to define the initial state of the coder during the coding operation (a description of the circular convolutional codes in FIG.

<Desc / Clms Page number 34>

the article written by C. Berrou, C. Douillard and M. Jézéquel, entitled Multiple Parallel Concatenation of Circular Recursive Systems, and published in the journal Annales des Télécommunications, Vol.
54, No. 3-4, pp 166-172, 1999); a memory access control 502 defining the address of the symbol to be processed and intended for the modules 510 and 520; a command 503 for controlling the memory banks 561 to 56P for reading and writing.

 The control unit 504 can be implemented in hard-wired form and / or in the form of a computer program running on a processor.

 The access addresses to the memory banks 561 to 56P are produced by the module 510, controlled via the signal 502 by the control unit 504. The module 510 controls a permutation of the data on each bank if necessary via a addressing signal 505.

 The data to be coded 550 are written in the memory banks 561, 562, ..., 56i,..., 56P through buses 551, 552,... 55i, 55P with the following partition for a block of K symbols. information: the memory bank 561 memorizes the symbols of index 0 to Cl; the memory bank 562 stores the symbols of index C at 2C-1; the memory bank 56i memorizes the index symbols (i-1) C to iC-1; and the memory bank 56P stores the index symbols (P-1) C to K-1.

 According to one variant, the data to be encoded comprise not only the information symbols but also redundancy symbols originating from a code constituting global code. The architecture of this variant being similar to that of the encoder 500, it will not be further detailed.

 The control unit 504 manages the write accesses of the symbols 550 or read symbols that are transferred to the output 553 when they are not read directly when they are presented on the input 550, via the signal 503 for the write and read commands and via the module 510 for the determination of the access addresses.

<Desc / Clms Page number 35>

 Each of the memory blocks respectively 561 to 56P is connected to the module 530 by a link respectively 531 to 53P enabling the data to be read by the module 530 in the corresponding memory block, this reading being also controlled by the signal 503 and the module 510. simultaneous reading of a symbol in each of the memory blocks, by the module 530 is possible.

The memory blocks 561 to 56P are single-port or dual-port RAM types depending on the bandwidth required.

 Based on the signal 502 from the control unit 504, the block 520 defines the circular rotation Ok to be applied to all the symbols presented at the inputs 531 to 53P and transmits the signal 506 to the module 530 to indicate the rotation Circular Ok to be made between the inputs and outputs of the 530 module.

 Each of the CRSC coders respectively 571 to 57P is connected to the module 530 by a link respectively 541 to 54P allowing the reading of a given symbol at the output of the module 530 by the corresponding CRSC coder. A symbol read by the module 530 is presented almost simultaneously to an output of the module 530 which switches this symbol to the appropriate output as a function of the signal 506. A simultaneous reading of a symbol by each of the CRSC coders in the module 530 is therefore possible. .

 Each of the circular recursive convolutional coders 571 to 57P respectively makes it possible to perform the pre-coding of each information sequence making it possible to determine the initial state of the coder such that its final state after coding of the sequence in question will be identical to that state. initial; and performing the recursive convolutional coding of each sequence of symbols presented on its input respectively 541 to 54P by multiplication by a multiplicative polynomial respectively /, at fp and by division by a divisive polynomial respectively g 1 to gp, making it possible to produce the redundancy bits (respectively Etp + 1 to

<Desc / Clms Page number 36>

EtP + P respectively transmitted on outputs 581 to 58P to the perforation module 540.

 To simplify the realization, the generating polynomials of the CRSC codes will be chosen identical for the P coders.

 According to one variant, for an adaptation to any global code, the generating polynomials of the CRSC codes will not all be identical.

 The redundancy bits produced by the CRSCs 571-57P are punched by the module 540 to provide the redundancy symbols 590 associated with the information symbols 550.

 In summary, at each read access controlled by the signals 502 and 503, the memory banks 561 to 56P provide symbols (in an order possibly different from the input order, which corresponds to a first interleaving) on which the Module 530 applies a circular rotation of Ok symbols. The symbols obtained are then used by circular recursive convolutional encoders 571 to 57P.

 In an implementation of the encoder 500 corresponding to the code illustrated with reference to FIG. 1, the data permutations are carried out by the control unit 504 and by the module 510; - two memory banks are used; the module 510 provides, in this embodiment, two memory addresses for the two memory banks via the signal 505; the spatial permutation between the memory banks is carried out by the control unit 504 and by the modules 520 and 530; and - the rollers 14 and 15 (respectively 16 and 17) are coded simultaneously using the coders 571 and 572 in the first step with T equal to 1 (respectively second step with T equal to 2), P being then equal to 2.

<Desc / Clms Page number 37>

 In an implementation of the encoder 500 corresponding to the code illustrated with reference to FIG. 2 (respectively 3), the permutation 22 (respectively #i 32) is carried out by the control unit 504 and by the module 510; the rotation 24 (respectively 34) is carried out by the control unit 504 and by the modules 520 and 530 and the first group of rollers 21 R1 to R5 (respectively 31 R1 to R5) and the second group of rollers 26 R6 to R10 (respectively R6 to R9) are successively coded using encoders 571 to 576, P being then 5 and the rollers being coded simultaneously within each group.

Similarly, in an embodiment of the encoder 500 corresponding to the code illustrated with reference to FIG. 4, the interleaver 420 is implemented by the control unit
504 and modules 510,520 and 530; the first group of rollers 411 R1 to 414 R4 and the second group of rollers 441 R5 to 444 R8 are successively coded using the coders 571 to 574, P being then equal to 4 and the rollers being coded simultaneously within each group.

The operation of encoding a frame of K information symbols by TP roulettes thus comprises the following steps: a writing of the information symbols 550 received in the memory banks 561 to 56P;
T coding steps (T is 2 in the previous examples illustrated with reference to Figures 1, 2 and 4) by P rollers for providing the corresponding redundancy symbols; punching (optional) the redundancy symbols (produced by the P rollers) by the module 540 making it possible to obtain the desired output, and

<Desc / Clms Page number 38>

 transmission on a channel of K systematic symbols (if the code is systematic) and N-K symbols of redundancy, if necessary punched.

 The organization of the memory is, according to the embodiment described, the following: at each memory address of a memory bank, are stored the data relating to a systematic symbol (information symbol from the channel and / or extrinsic information produced by the decoders) as well as the redundancy symbols from the corresponding channel.

 For each of the T encoding steps by the P rollers, the following successive operations are performed: initialization of the convolutive encoders in the null state via the signal 501; first reading of the C information symbols of the P rollers considered and pre-coding of each of the rollers; determination of the initial states of the coders; second reading of the C information symbols by the P rollers considered and coding of each of the rollers in order to produce the corresponding redundancy symbols.

 According to a variant of the encoder 500 not shown, an encoder similar to the encoder 500 implements non-convolutional systematic circular recursive constitutive codes.

 Likewise, according to another variant, an encoder adapted to the coding of the code illustrated with reference to FIG. 3 is implemented on the basis of the encoder 500 with the following modifications: the perforation of the code is not necessary; - the organization of the memory is adapted to the code; and if a constituent code is not systematic, the outputs 58i relate to all the symbols of the constituent code word.

 FIG. 6 shows a decoder 600 adapted to decode a global code produced by the encoder 500 illustrated with reference to FIG. 5. The decoder 600 uses

<Desc / Clms Page number 39>

 Parallel SISO P decoders 671, 672, ..., 67i, ..., 67P each corresponding to the encoders 571, 572, ..., 57i, ..., 57P.

 The decoder 600 accepts on an input 650 blocks of N soft information symbols (symbols from the channel and possibly prior information) and provides on an output 680 decoded symbols. The incoming and / or stored data in the decoder 600 are information and redundancy symbols from the channel and weighted information produced by the decoders.

 The global decoder 600 comprises: the input 650 of soft information symbols; the output 680 of decoded symbols; a control unit 604; - memory banks (BMx rated) 661 to 66P; elementary decoders of SISO type 671 to 67P; an intra-line permutation module, PILk 610, which is identical for the P memory banks; - a circular rotation module RC (Ok) 630; an OK 620 control module of the module 630; a decision module 690.

 The control unit 604 manages the operation of the entire device 600. This control unit produces the following control signals: a memory access control 602 defining the address of the symbol to be processed and intended for the modules 610 and 620; a control 603 for controlling the memory banks 661 to 66P in reading and writing, the memory banks being able to be addressed simultaneously and independently of one another.

 The control unit 604 also exchanges a control signal 601 with each of the SISO decoders; signal makes it possible in particular to implement the stopping criterion by defining which decoders are actually used during an iteration;

<Desc / Clms Page number 40>

The access addresses to the memory banks 661 to 66P are produced by the module 610 controlled via the signal 602 by the control unit 604. The module 610 controls a permutation of the data on each bank if necessary by means of a signal addressing 605.

 The data to be decoded 650 is written or read in the memory banks 661, 662,..., 66i,..., 66P through bidirectional buses 651, 652,..., 65i, 65P with the following partition for a block of data. K information symbols in the case where the constituent codes are circular recursive convolutional codes (for these codes the redundancy symbols and the associated systematic symbols are stored at the same memory address): the memory bank 661 stores the relative data symbols of index 0 to C-1; the memory bank 662 stores the data relating to the symbols of index C to 2C-1; the memory bank 66i stores the data relating to the index symbols (i-1) C to iC-1, and the memory bank 66P stores the data relating to the index symbols (P-1) C to K-1.

 In the general case of a constituting code that is not circular recursive convolutional circular, the memory organization is adapted to the code.

 The control unit 604 manages the write accesses of the symbols 650 or read symbols which are transferred to the output 680, via the signal 603 for the write and read commands and via the module 610 for the determination of the addresses access.

 Each of the memory blocks 661 to 66P respectively is connected to the module 630 via a link 631 to 63P, respectively, enabling the data to be read by the module 630 in the corresponding memory block, this reading being also controlled by the signal 603 and the module 610. The module 630 can therefore read or write simultaneously and in parallel several symbols, a symbol being read or written in each of the memory blocks. The memory blocks 661 to 66P are

<Desc / Clms Page number 41>

 Single-port or dual-port RAM type depending on the bandwidth required.

 Each of the SISO decoders 671 to 67P, respectively, is connected to the module 630 by a bidirectional link respectively 641 to 64P allowing the reading or writing of a given symbol at the output of the module 630 by the corresponding SISO decoder. Thus, several symbols are read in parallel in the corresponding memory banks and feed as many SISO decoders.

 As a function of the signal 602 emitted by the control unit 604, the block 620 applies, in this embodiment, the circular rotation Ok to all the symbols presented at the inputs / outputs 631 to 63P or 641 to 641P and transmits the signal 606 to module 630 to indicate the circular rotation Ok to be made between the inputs and outputs of module 630.

 A symbol read or written by the module 630 is presented together with an output of the module 630 which sets this symbol to the appropriate output as a function of the signal 606. A simultaneous reading or writing of a symbol by each of the SISO decoders in the module 630 is therefore possible.

 Each of the SISO decoders 671 to 67P uses a flexible input and flexible output algorithm, which produces weighted information in the first (respectively second) step, where T is 1 (respectively 2) and P is then 2. SISO decoders contain also modules allowing to implement the stopping criterion. Each of the modules returns information in particular on the convergence of the decoding associated with the corresponding decoder, the control unit via the signal 601. The control unit supports the implementation of the application of the criteria of 'stops.

 The weighted information is then written to memory symmetrically with the previous read operation.

 In summary, at each read access controlled by the signals 602 and 603, the memory banks 661 to 66P provide symbols (in an order possibly different from the input order, which corresponds to a first

<Desc / Clms Page number 42>

 interleaving) on which the module 630 applies a circular rotation of Ok symbols. The symbols obtained are then used by the SISO decoders 671 to 67P. After decoding, a symmetrical operation is carried out, the data being transmitted from each of the SISO decoders to a memory bank after a passage in the module 630.

 At the end of the iterative decoding, the module 690 determines the hard decisions, result of the decoding from the data present in the memory banks 661 to 66p. The hard decisions can then be read on an output 691 of the decoder 600.

 In an implementation of the decoder 600 corresponding to the code illustrated with reference to FIG. 1, two memory banks are used; the data permutations are carried out by the control unit 604 and by the module 610 which provides two different memory addresses for the two memory banks used via the signal 605; the spatial permutation between the memory banks is carried out by the control unit 604 and by the modules 620 and 630; and - the rollers 14 and 15 (respectively 16 and 17) are decoded simultaneously using the decoders 671 and 672, in the first step with T equal to 1 (respectively second step with T equal to 2), P being then equal to 2.

 In an implementation of the decoder 600 corresponding to the code illustrated with reference to FIG. 2 (respectively 3), the permutation 22 (respectively # 1 32) is carried out by the control unit 604 and by the module 610; the rotation 24 (respectively 34) is carried out by the control unit 604 and by the modules 620 and 630 and the first group of rollers 21 R1 to R5 (respectively 31R1 to R5) and the second group of rollers 26 R6 at R10

<Desc / Clms Page number 43>

 (respectively 36 R6 to R9) are successively decoded using the decoders 671 to 675, P being then 5 and the rollers within each group being decoded simultaneously in parallel.

Similarly, in an embodiment of the encoder 600 corresponding to the code illustrated with reference to FIG. 4, the interleaver 420 is implemented by the control unit
604 and modules 610, 620 and 630; the first group of rollers 411 R1 to 414 R4 and the second group of rollers 441 Rs to 444 R8 are successively decoded using the decoders 671 to 674, P being then 4 and the rollers inside each group being decoded simultaneously in parallel.

The iterative decoding operation of a frame of K information symbols by T. P rollers thus comprises the following steps: a writing of the data relating to the information symbols 650 received in the memory banks 661 to 66P (soft decisions at the output of the channel of systematic information symbols and redundancy symbols); a plurality of decoding iterations, each consisting of T decoding steps (T being, for example, 2 for the codes illustrated above with reference to FIGS. 1 to 4) of P rollers implementing a SISO type algorithm that requires reading; C systematic information symbols of the roulette considered and writing C weighted information symbols produced in the corresponding memory banks; and reading, in the memory banks 661 to 66P, data relating to
K information symbols to make the decision lasts on the symbols and produce the decoded frame that will be transmitted on the output
680.

<Desc / Clms Page number 44>

 Iterative decoding is controlled by the control unit 600 defining the iterative decoding algorithm to be used and in particular the use of the stopping criterion.

 To reduce the decoding time of a frame, a stop criterion is applied to data windows. Each of the windows is determined to exactly match a constituent code.

 Thus, the blocks of data on which the stopping criterion is checked during the current iteration are determined and the corresponding calculations are not carried out during the next iteration. This method offers the possibility of reducing the overall size of the data to be decoded as the iterations proceed at the cost of a limited performance degradation because the windows correspond to component codes.

Thus, the stopping criterion is used in several ways, in particular: if at least one of the P decoders used in parallel has converged (indication by each corresponding signal 601), the control unit
600 freezes this or these decoders; and / or if all the P decoders used in parallel have converged (indicated by the corresponding signals 601), the control unit 600 skips the corresponding decoding step t and goes directly to the next decoding step t + 1 , which allows a latency gain of decoding and consumption.

* For further details on the decoding algorithms used for SISO decoders, refer to the articles (describing respectively an algorithm of type BJCR and Viterbi): Optimal decoding of linear codes for minimizing symbol error rate, (or in French decoding optimal linear codes to minimize the symbol error rate) written by R. R Bahl, J. Cocke, F. Jelinek, J.

 Raviv, and published in the journal IEEE Transactions on Information Theory, pp. 284-287, March 1974.

<Desc / Clms Page number 45>

A Viterbi algorithm with soft decision outputs and their applications (or in French a Viterbi algorithm with flexible decision outputs and its applications) written by J. Hagenauer and P. Hoeher and appeared in the report of the IEEE Globecom conference, pp. 1680-
1686, November 1989.

 According to a variant of the decoder not shown, a decoder similar to the decoder 600 implements non-convolutional systematic circular recursive constitutive codes.

 Likewise, according to another variant, a decoder adapted to the coding of the code illustrated with reference to FIG. 3 is implemented on the basis of the decoder 600 with an organization of the memory adapted to the code. In particular, each symbol of the code word is identified by a specific memory address (whereas for the decoder 600, the same addresses can be used for a systematic symbol and a corresponding redundancy symbol).

 FIG. 7 illustrates an application of the coding and decoding method according to the invention to data storage on magnetic and / or optical media.

 The document Application of iterative decoding techniques to the correction of inter-symbol interference written by P. Didier, A. Picard, C.

Douillard and M. Jezequel, presented at the 15th GRETSI (France) symposium from 18 to 21 September 1995, proposes a mechanism in which an intersymbol interference channel is considered to be one of the constituent codes of a convolutional code concatenated in series. Similarly, M. Oberg and PH Siegel propose in an article titled performance analysis of turbo-equalised partial response channels (or in French performance analysis of turbo-equalized partial response channels) (published in the journal IEEE Transactions on Communications, Flight 49, No. 3, March 2001, pages 436-444) a use of an iterative turbo-decoding in the specific case of a magnetic channel with intersymbol interference.

<Desc / Clms Page number 46>

 These techniques have the drawbacks of the prior art coding mechanisms presented above. They are not detailed further.

 Nevertheless, the invention also makes it possible to overcome these disadvantages.

Thus, it can be considered that the channel 700 is a magnetic channel with intersymbol interference which corresponds to a convolutional code and makes it possible to code the data according to a roulette code.

 Due to its nature, the channel 700 is continuous. Its cutting into wheels (in essence finite size) therefore requires special treatment.

 According to the embodiment illustrated with reference to FIG. 7, particular symbols are inserted in the frame so as to obtain a periodic return to a predefined state (for example, the state 0). The number of added symbols is greater than or equal to the memory of the channel 700. The cutting of the rollers is then between two forced passages to the predefined state.

 Thus, according to the example illustrated with reference to FIG. 7, four blocks each comprising three information symbols are respectively coded by four wheels 721, 731, 741 and 751 to form four codewords 720, 730, 740 and 750. The data thus encoded are then interleaved (according to an interlacing 770) and recorded on a magnetic medium corresponding to the channel 700 (data 701 to 716). If the memory of the channel 700 is equal, for example, to three, three bits 717 to 719 equal to 0 are added after the symbols 701 to 716 recorded, which makes it possible to force the passage of a wheel 760 corresponding to the coding on the channel 700 of the registered symbols 701 to 719 in the predefined state equal to 0.

 According to an alternative embodiment, wheels are created by duplicating the last bits of the rollers so as to loop the information (for example, a message Il 12 13 14 will be transmitted in the channel 700 by a message completed by the last two symbols 13 14 placed at the head of a completed message 13 14 11 12 13 14 if the memory of the channel 700 is equal to 2, with only the four final symbols of the completed message which participates in a roulette).

<Desc / Clms Page number 47>

 According to another variant, the overlapping edge effects of the rollers are managed as illustrated with reference to FIG. 4, different blocks corresponding to a roulette encoded by the channel being treated independently, taking a context which makes it possible to ensure the convergence of the decoding algorithms, so that the code thus obtained is not suboptimal.

 According to yet another variant, a message to be coded is divided into blocks and edge effects are managed, on decoding, by an information transmission mechanism (for example, the state of the nodes of the trellis used by a trunk algorithm). type SISO) from one block to another.

 The decoding operation is carried out by a decoder entirely similar to that illustrated with reference to FIG. 6, with elementary decoders corresponding to sub-codes 700, 721, 731, 741 and 751, the codes 721, 731, 741 and 751 being decodable in FIG. parallel by separate decoders. Similarly, several blocks coded according to the code 700 may themselves be decoded in parallel by separate decoders It will therefore not be further detailed.

 Of course, the invention is not limited to the embodiments mentioned above.

 In particular, those skilled in the art can make any variant in the type of rollers used, as well as their number in a coding device and / or global code decoding.

 Those skilled in the art may also implement a number of coders and / or decoders as well as memory banks adapted to the needs of the coding and / or decoding operations depending in particular on the desired coding and / or decoding latency and / or the complexity of the coder and / or global code decoder. Thus, in certain applications aimed at optimizing latency, the number of memory banks and / or coders / decoders will be close to or even equal to the number of subcodes. In order to further improve the decoding speed, decoding modules implementing a decoding iteration can be cascaded. For applications aimed at optimizing complexity while

<Desc / Clms Page number 48>

 keeping a high coding / decoding speed, the number of memory banks as well as subcode encoders / decoders can be reduced (for example corresponding to a fraction of the number of subcodes belonging to the global code).

 The invention is not limited to the case where the number of steps T is equal to 2 but also extends to the cases where T is greater than 2, especially when the overall code can be modeled as a first matrix and second matrices, each second matrix being obtained by interleaving the columns and possibly rows of the first matrix, and each row of the first and second matrices corresponding to a subcode word of the global code. In this case, T can clearly take any value (for example, 3,4, 5, ...).

 The invention is also not limited to the case where the dimension of the code is 2 but also extends to cases where it is greater than 2.

 Note that the invention is not limited to the coding or decoding of data intended for or coming from a transmission channel but extends to any application of the error-correcting codes and, in particular to the storage of data on media. magnetic and / or optical.

 The invention furthermore relates to devices comprising one or more encoders or decoders according to the invention, and in particular mobiles or radio communication infrastructure equipment as well as data transmission and / or reception equipment, in particular wireless (eg radio, optical and / or acoustic) and broadband.

 The invention also relates to the codes generated by the encoders described above and the information signals encoded with such encoders.

 The invention is not limited to the code allowing a uniform protection of the information symbols but also extends to the case where a protection level according to the information symbols themselves (for example, from different sources or of unequal importance according to a source coding or according to their meaning (data or control symbols)). Thus, the constituent codes of a global code according to the invention may be different in yield depending on the desired level of protection. In addition, the symbols

<Desc / Clms Page number 49>

 information may have different degrees and be distributed over one or more constituent codes.

 It will be noted that the invention is not limited to a purely hardware implementation but that it can also be implemented in the form of a sequence of instructions of a computer program or any form mixing a hardware part and a software part. In the case where the invention is partially or totally implemented in software form, the corresponding instruction sequence can be stored in a removable storage means (such as for example a floppy disk, a CD-ROM or a DVD-ROM) or no, this storage means being partially or completely readable by a computer or a microprocessor.

Claims (44)

 1. A method for decoding error correction code, of the type associating a decoded data block with data encoded according to a global code comprising at least two constituent sub-codes (Ri), an irregular bipartite graph being associated with said global code , said decoding method being iterative and producing at each iteration a block of extrinsic data, each of the extrinsic data relating to one of said coded data, said method implementing a step of storing a block of data to be decoded comprising said coded data and said extrinsic data, said data block to be decoded being distributed in a plurality of disjoint memory banks (BMi), independently addressable, characterized in that at each iteration, said method further comprises a step of feeding in parallel with at least two decoders (671,672, 671, 67P) among a plurality of decoders, correspo each respectively denoting at least one of said sub-codes, with corresponding data to be decoded from said data block to be decoded, data to be decoded being extracted in parallel from at least two of said memory banks (BMi) to feed as many decoders, and each of said decoders being fed sequentially by said data to be decoded corresponding thereto. 2. Method according to claim 1, characterized in that at least one of said sub-codes is a circular convolutional recursive code. 3. Method according to any one of claims 1 and 2, characterized in that at least one of said sub-codes is a systematic code. 4. Method according to any one of claims 1 to 3, characterized in that at least two of said sub-codes are m-binary codes, m being an integer greater than or equal to 2 and in that an intra permutation -symbols is applied between at least two of said m-binary sub-codes. <Desc / Clms Page number 51> 5. Method according to any one of claims 1 to 4, characterized in that it further comprises a step of switching each of the data to be decoded to one of said decoders, said data being switched in parallel and simultaneously from so that each of them feeds a decoder corresponding to it own. 6. Method according to claim 5, characterized in that said step of switching itself comprises a circular permutation step of a set comprising at least a portion of the data to be decoded. 7. Method according to claim 6, characterized in that said circular permutation is a rotation which has a step determined according to the read rank of the data read in said banks memory. 8. Method according to any one of claims 5 to 7, characterized in that said step of switching implements a step of addressing each of said memory banks so that data can be read in said memory bank in a predetermined order distinct from the writing order of said data in said memory bank. 9. Method according to any one of claims 1 to 8, characterized in that each line of a first matrix of data representative of said data encoded according to said global code is representative of data encoded by a subcode constituent of said global code belonging to to a first group of subcodes; each line of at least a second matrix of data representative of said data encoded according to said global code is representative of data encoded by a constituent subcode of said global code belonging to at least a second group of sub-codes; each of said at least one second matrix of data being obtained from a transformation of said first matrix, said transformation comprising a permutation of at least a portion of the columns of said first matrix. <Desc / Clms Page number 52>  10. The method of claim 9, characterized in that said transformation further comprises a permutation of at least a portion of the lines of said first matrix. 11. Method according to any one of claims 1 to 10, characterized in that each line of a first matrix of data representative of said data encoded according to said global code is representative of data encoded by a subcode constituting said global code belonging to to a first group of sub-codes, a part of the data of said first matrix, said non-significant data, not being significant; each line of a first subset of lines of at least one second matrix of data representative of said data encoded according to said global code is representative of data encoded by a subcode constituting said global code belonging to at least a second group of subcodes; a second subset of lines of said second data matrix containing said non-significant data; and each of said at least one second data matrix being derived from a transformation of said first matrix, said transformation comprising a permutation of at least a portion of the columns of said first matrix. 12. Method according to any one of claims 1 to 11, characterized in that it comprises a decoding test step implementing at least one stopping criterion so that when said at least one criterion d stopping is verified for at least one of said sub-codes, at least one of said decoders associated with said at least one of said sub-codes stops decoding said at least one of said sub-codes for which said at least one stopping criterion is checked. 13. Method according to any one of claims 1 to 12, characterized in that it implements a step of reading said coded data from an optical and / or magnetic medium and / or transmission of said coded data. on an interference channel. <Desc / Clms Page number 53> 14. Method according to claim 13, characterized in that at least one of said sub-codes corresponds to interference between representative symbols of said decoded data block when said block is stored on said optical and / or magnetic medium and / or when said block is transmitted in an interference channel. 15. Method according to any one of claims 1 to 14, characterized in that it implements a step of receiving said coded data from a transmitter. 16. A method according to any one of claims 1 to 15, characterized in that the data sets coded with each of said sub-codes are all different two by two. 17. Method according to any one of claims 1 to 16, characterized in that at least two of said sub-codes comprise at least two of said data to be decoded in common. 18. Method according to any one of claims 1 to 17, characterized in that said global code is of product code type with a non-uniform interleaver. 19. Method according to any one of claims 1 to 18, characterized in that it further comprises a step of demultiplexing said decoded data block so as to supply at least two distinct recipients with data belonging to said block of data. decoded data demultiplexed. 20. Method according to any one of claims 1 to 19, characterized in that the degree of the information symbols of said decoded data block in said global code is not uniform. An error correction code encoding method, of the type associating a source data block with data encoded according to a global code comprising at least two constituent sub-codes, an irregular bipartite graph being associated with said global code, said block of coded data being intended to be transmitted to at least one receiver and / or stored on a data medium, <Desc / Clms Page number 54>  characterized in that said coded data block is for decoding by said iterative decoding method according to any one of claims 1 to 19. 22. An error correction code coding method, of the type associating a source data block with data coded according to a global code comprising at least two constituent sub-codes, an irregular bipartite graph being associated with said global code, said block of code coded data being intended to be transmitted to at least one receiver and / or stored on a data medium, said coded data block being intended to be decoded by an iterative decoding method, characterized in that said method further comprises a step of supplying in parallel at least two of a plurality of coders, each corresponding respectively to at least one of said sub-codes, by data to be coded corresponding to said block of data to be coded, data to be coded being extracted in parallel with at least two of said memory banks for supplying as many encoders. 23. The method of claim 22, characterized in that each of said encoders is fed sequentially by said data to be encoded him corresponding. 24. The method as claimed in claim 22, wherein at least one of said sub-codes is a circular convolutional recursive code. 25. Method according to any one of claims 22 to 24, characterized in that at least one of said sub-codes is a systematic code. 26. Method according to any one of claims 22 to 25, characterized in that at least two of said sub-codes are m-binary codes, m being an integer greater than or equal to 2 and in that an intra permutation -symbols is applied between at least two of said m-binary sub-codes. 27. Method according to any one of claims 22 to 26, characterized in that it further comprises a step of switching each of the data to be encoded to one of said encoders, said data being switched in parallel and <Desc / Clms Page number 55>  simultaneously so that each of them feeds a coder corresponding to it own. 28. The method of claim 27, characterized in that said routing step itself comprises a circular permutation step of a set comprising at least a portion of the data to be encoded. 29. The method of claim 28, characterized in that said circular permutation is a rotation which has a step determined according to the read rank of the data read in said banks memory. 30. Method according to any one of claims 27 to 29, characterized in that said step of switching implements a step of addressing each of said memory banks so that data can be read in said memory bank in a predetermined order distinct from the writing order of said data in said memory bank. 31 Method according to any one of claims 22 to 30, characterized in that each line of a first data matrix representative of said data encoded according to said global code is representative of data encoded by a constituent subcode of said global code belonging to a first group of sub-codes; each line of at least a second matrix of data representative of said data encoded according to said global code is representative of data encoded by a constituent subcode of said global code belonging to at least a second group of sub-codes; each of said at least one second matrix of data being obtained from a transformation of said first matrix, said transformation comprising a permutation of at least a portion of the columns of said first matrix. 32. The method of claim 31, characterized in that said transformation further comprises a permutation of at least a portion of the lines of said first matrix. <Desc / Clms Page number 56> 33. Method according to any one of claims 22 to 30, characterized in that each line of a first data matrix representative of said data encoded according to said global code is representative of data encoded by a subcode constituting said global code belonging to to a first group of sub-codes, a part of the data of said first matrix, said non-significant data, not being significant; each line of a first subset of lines of at least one second matrix of data representative of said data encoded according to said global code is representative of data encoded by a subcode constituting said global code belonging to at least a second group of subcodes; a second subset of lines of said second data matrix containing said non-significant data; and each of said at least one second data matrix being derived from a transformation of said first matrix, said transformation comprising a permutation of at least a portion of the columns of said first matrix. 34. Method according to any one of claims 22 to 33, characterized in that it implements a step of writing said encoded data on an optical and / or magnetic medium and / or of transmitting said coded data on a interference channel. 35. Method according to claim 34, characterized in that at least one of said sub-codes corresponds to interferences between representative symbols of said decoded data block when said block is stored on said optical and / or magnetic medium and / or when said block is transmitted in an interference channel. 36. Method according to any one of claims 22 to 35, characterized in that it implements a step of receiving said coded data from a transmitter. <Desc / Clms Page number 57>  37. A method according to any one of claims 22 to 36, characterized in that the data sets encoded with each of said sub-codes are all different two by two. 38. Method according to any one of claims 22 to 37, characterized in that at least two of said sub-codes comprise at least two data coded in common. 39. Method according to any one of claims 22 to 38, characterized in that said global code is of product code type with a non-uniform interleaver. 40. A method according to any one of claims 22 to 39, characterized in that it further comprises a step of multiplexing at least two data blocks from each of two separate sources so as to form said block. source data. 41. Method according to any one of claims 22 to 40, characterized in that the degree of the information symbols of said decoded data block in said global code is not uniform. 42. Signal representative of a block of data encoded according to a global code comprising at least two constituent sub-codes, an irregular bipartite graph being associated with said global code, characterized in that said coded data block is intended to be decoded by said iterative decoding method according to any one of claims 1 to 20. 43. An error correction code decoding device, of the type associating a decoded data block with data coded according to a global code comprising at least two constituent sub-codes, an irregular bipartite graph being associated with said global code, said method of decoding being iterative and producing at each iteration a block of extrinsic data, each of the extrinsic data relating to one of said coded data, said device implementing storage means of a block of data to be decoded comprising said coded data and said data <Desc / Clms Page number 58>  extrinsic, said data block to be decoded being distributed in a plurality of disjoint memory banks, independently addressable, characterized in that at each iteration, said device further comprises means for supplying in parallel of at least two decoders among a plurality of decoders, each corresponding respectively to at least one of said sub-codes, by corresponding data to be decoded from said data block to be decoded, data to be decoded being extracted in parallel from at least two of said memory banks to supply as much as decoders, and each of said decoders being fed sequentially by said data to be decoded corresponding thereto. 44. An error correction code encoding device, of the type associating a source data block with data coded according to a global code comprising at least two constituent sub-codes, an irregular bipartite graph being associated with said global code, said block of data coded data being intended to be transmitted to at least one receiver and / or stored on a data medium, said coded data block being intended to be decoded by an iterative decoding device, characterized in that said device further comprises means for supplying in parallel at least two of a plurality of encoders, each corresponding respectively to at least one of said sub-codes, by corresponding data to be coded from said data block to be coded, data to be coded being extracted in parallel with at least two of said memory banks for supplying as many encoders.