초록 ▼

The invention concerns a method and a device for error correction coding associating with a data source series a coded data block, to be transmitted to at least one receiver comprising at least two coding stages (21i) each comprising at least two basic coding modules (22i,j), each of said coding sta

The invention concerns a method and a device for error correction coding associating with a data source series a coded data block, to be transmitted to at least one receiver comprising at least two coding stages (21i) each comprising at least two basic coding modules (22i,j), each of said coding stages receiving a series of data to be processed, distributed between said basic coding modules, and delivering a series of processed data, derived from said basic coding modules and at least one branching stage (23i), said branching stage being inserted between two successive coding stages, a first coding stage and a second coding stage, and distributing the processed data derived from each basic coding module of said first coding stage between at least two basic coding modules of said second stage. The invention also concerns the corresponding decoding method and device, based on path likelihood.

대표청구항 ▼

1. Error correction coding method of the type associating a series of source data with a block of coded data to be transmitted to at least one receiver, comprising: at least two coding steps ( 21i) each performed by at least two basic coding modules (22i,j), each of the said coding steps receivin

1. Error correction coding method of the type associating a series of source data with a block of coded data to be transmitted to at least one receiver, comprising: at least two coding steps ( 21i) each performed by at least two basic coding modules (22i,j), each of the said coding steps receiving a series of data to be processed, distributed between the said basic coding modules, and outputting a series of processed data from the said basic coding modules; and at least one mixing step ( 23i), the said mixing step being inserted between two successive coding steps, a first coding step and a second coding step, distributing the processed data output from each basic coding module in the said first coding step between at least two basic coding modules in the said second step; said block of coded data to be transmitted comprising at least some of said source data and at least some processed data output from the last coding stage. 2. Coding method according to claim 1, characterized in that the coded data block to be transmitted comprises all the said source data.3. Coding method according to claim 1, characterized in that the said coded data block comprises processed data output from at least two coding steps.4. Coding method according to claim 1, characterized in that at least one of the said mixing steps uses at least one swap matrix.5. Coding method according to claim 1, characterized in that at least one of the said mixing steps distributes the processed data to at least two distinct coding steps.6. Coding method according to claim 1, characterized in that at least one item of the said source data and/or at least one item of the said processed data may be duplicated (51) at least once, in order to form at least two items of data to be processed.7. Coding method according to claim 1, characterized in that at least one of the said source data is directly input (52,57) into a coding step other than the first coding step.8. Coding method according to claim 1, characterized in that it includes puncturing means applied to at least some of the said data to be processed and/or on at least some of the said processed data.9. Coding method according to claim 1, characterized in that the said coding modules (22i,j) use a redundancy code with length n−k less than or equal to 12.10. Coding method according to claim 1, characterized in that the said basic coding modules are built using elementary coding blocks associating the (x0, x0+x1) pair with the (x0, x1) pair.11. Coding method according to claim 1, characterized in that at least two of the said coding modules (411,452,453,454,471) are different.12. Coding method according to claim 1, characterized in that at least some coding modules (22i,j) use at least one code belonging to the group containing: a (4,2) code, for example that associates the values (x 0, x1, x0, x0+x1) with the (x0, x1) pair; a trivial code that associates x 0 with x0.13. Transmission and/or reception method comprising a step in which at least one coded signal is transmitted according to the method in claim 1.14. Application of the coding method according to claim 1, to at least one of the techniques belonging to the group comprising: transmission and/or broadcasting of digital signals; speech recognition and/or processing. 15. Coding method according to claim 1, characterized in that it uses at least two coding units each performing at least two of the said coding steps and at least one mixing step inserted between two successive coding steps, and in that the said source data are input into each of the said coding units in different input orders. 16. Coding method according to claim 15, characterized in that it uses at least one mixing unit, modifying the order in which the said source data are input into one of the said coding units.17. Coding method according to claim 1, characterized in that all the said coding modules are identical (22i,j).18. Coding method according to claim 17, character ized in that each of the said coding modules (22i,j) uses a Reed-Müller code.19. Coding method according to claim 1, characterized in that at least one of the said mixing steps uses a “parity” swap that groups inputs with even indexes, and inputs with odd indexes, at the output.20. Coding method according to claim 19, characterized in that the swap performs the following operations for a swap length k: inputs with even indexes i=2p are sent to outputs j=p, p=0, 1, . . . int[k/2]−1; inputs with odd indexes i=2p+1 are sent to outputs j=p int[k/2], where int[x] is the function giving the integer part of x. 21. Coding method according to either of claim 19, characterized in that it uses mixing steps involving a parity swap, and that it comprises: three coding steps each performed by three basic coding modules; three coding steps each performed by four basic coding modules; five coding steps each performed by nine basic coding modules, the said basic coding modules using an extended Hamming code [8,4]. 22. Coding method according to claim 1, characterized in that at least one of the said mixing steps generates at least one output that is a function of at least two inputs to the said mixing step.23. Coding method according to claim 22, characterized in that the said function can use at least one summation, a multiplication and/or a weighting.24. Coding method according to claim 1, characterized in that it comprises a step (48) in which at least one of the following elements is checked and/or adjusted: coding type and/or characteristic used by at least one of the basic coding modules; mixing used by at least one of the said mixing steps; punching used on at least some of the said data to be processed and/or on at least some of the said processed data; number of coding steps. 25. Coding method according to claim 24, characterized in that the said check and/or the said adjustment acts systematically on a given period and/or as a function of at least one item of information representative of at least one of the aspects belonging to the group including: at least one characteristic of the transmission channel; at least one characteristic of the receiver; at least one characteristic of the source signal. 26. Coding method according to claim 1, characterized in that at least one of the said basic coding modules (22i,j) is used in a trellis defining a set of paths bijectively associated with a set of code words, and outputting optimum path metrics according to a predetermined criterion.27. Coding method according to claim 26, characterized in that the said inputs to the said function are path metrics.28. Coding method according to claim 26, characterized in that the said coding and mixing steps are iterated at least twice.29. Coding method according to claim 26, characterized in that the said coding and mixing steps are identical to the decoding and mixing steps used during decoding, the coding being initialized by forcing the redundancy bits to 0, and the coding consisting of only keeping the metrics for the maximum path of each coding module.30. Error correction coding device of the type that associates a coded data block with a series of source data, the coded data block to be transmitted to at least one receiver, comprising: at least two coding stages each using at least two basic codings, each of the said coding stages receiving a series of data to be processed, distributed between the said basic codings and outputting a series of processed data corresponding to the said basic codings; and at least one mixing stage, the said mixing stage being inserted between two successive coding stages, a first coding stage and a second coding stage, and distributing the processed data corresponding to each basic coding of the said first coding stage between at least two basic codings of the second stage; said block of coded data to be transmitted comprising at least some of said source data and at least some processed data output from the last coding stage. 31. Coded signal produced by an error correction coding method, of the type associating a coded data block with a series of source data, the coded data block to be transmitted to at least one receiver, comprising: at least two coding steps ( 21i) each performed by at least two basic coding modules (22i,j), each of the said coding steps receiving a series of data to be processed distributed between the said basic coding modules, and outputting a series of processed data corresponding to the said basic coding modules; and at least one mixing step ( 23j), the said mixing step being inserted between two successive coding steps, a first coding step and a second coding step, and distributing the processed data output from each basic coding module performing the said first coding step between at least two basic coding modules of the said second step, the said coded signal comprising at least some of the said source data and at least some processed data output by the last coding stage. 32. Method for decoding a data block coded by an error correction coding method, of the type associating a block of coded data with a series of source data, the coded data block to be transmitted to at least one receiver, comprising: at least two coding steps ( 21i) each performed by at least two basic coding modules (22i,j), each of the said coding steps receiving a series of data to be processed distributed between the said basic coding modules, and outputting a series of processed data corresponding to the said basic coding modules; and at least one mixing step ( 23j), the said mixing step being inserted between two successive coding steps, a first coding step and a second coding step, and distributing the processed data output from to each basic coding module of the said first coding step between at least two basic coding modules of the said second step, the said data block performed by at least some of the said source data and at least some processed data output by the last coding stage. 33. Decoding method according to claim 32, characterized in that it is iterative.34. Decoding method according to claim 32, characterized in that it is used as a coding method with the redundancy inputs being forced to 0 and the information inputs corresponding to the data to be coded.35. Decoding method according to claim 34, characterized in that it is used alternately firstly for coding and secondly for decoding data.36. Decoding method according to claim 32, characterized in that it uses at least one of the techniques belonging to the group comprising: decoding using a structure symmetrical to the structure used during coding; exhaustive decoding, according to which all possible code words are considered and the best is selected according to a predetermined selection criterion; Viterbi type decoding, using a decoding trellis; Chase decoding. 37. Decoding method according to claim 36, characterized in that it uses at least two decoding steps, performed by at least two decoding modules, and at least one swap step inserted between two consecutive decoding steps, the said steps being symmetric with the corresponding coding steps, and in that each of the said decoding modules comprises twice as many inputs and outputs as the corresponding coding module, to assure propagation of probability values from the input to the output side of the decoder, and secondly from the output to the input side of the said decoder.38. Decoding method according to claim 36, characterized in that it includes at least one iteration of the following operations: propagation of all the said probability values towards the input side of the said decoder, producing a first set of estimated values; propagation of all the said probability values towards the output side of the said decoder, producing a second set of estimated values. 39. Decoding method according to claim 38, characterized in that it forms a summation of the estimated values of each of the said sets of estimated values, the next iteration being made on the results of the said summation.40. Decoding method according to claim 39, characterized in that, for each output from one of the said decoding modules, it takes account of a Boolean function depending on at least two inputs to the corresponding coding module, and in that a decoding constraint imposes that an output side output from a given decoding stage must be equal to the corresponding input side output of the next decoding stage.41. Decoding method according to claim 32, characterized in that it determines path probabilities.42. Decoding method according to claim 41, characterized in that the said probabilities are determined from probability laws.43. Decoding method according to claim 41, characterized in that at least one of the said basic decoding modules use a trellis with nodes in which addition, comparison and/or selection operations are carried out and that output optimum path metrics and in that additions, multiplications and/or weightings of at least two of the said optimum path metrics are carried out in at least one of the said mixing steps.44. Decoding method according to claim 43, characterized in that the said coding method is a coding method in which the said coding and mixing steps are iterated at least twice, the first and last mixing steps forming a single step.45. Decoding method according to claim 41, characterized in that in each iteration, each of the said elementary decoding operations determines a state probability for each possible corresponding coding state, and the state with the highest probability can be selected.46. Decoding method according to claim 45, characterized in that the said state probability corresponds to the sum of a first probability value obtained starting from information bits and a second probability value obtained from redundancy bits.47. Decoding method according to claim 46, characterized in that the said first (or second) probability associated with a given output from one of the said elementary decodings may be obtained by determining the logarithm of the sum of the exponentials of the probabilities of states connected to the said output, minus the logarithm of the sum of the exponentials of the probabilities received at the input, as information bits (or redundancy bits) for states connected to the said output.48. Decoding method according to claim 47, characterized in that when the decoding has reached a predetermined convergence criterion, possibly independent for each coding module, an a posteriori probability is associated with each bit:V ″(xj)=Log(&Sgr; exp(Ei/(xj=0)→Ei)−Log(&Sgr; exp(Ei/(xj=1)→Ei))49. Decoding method according to claim 47, characterized in that it comprises the following steps: calculate state probabilities: “forwards”: Fi, based on information bits;“backwards”: Bi, based on redundancy bits; calculate global state probabilities: E i=Fi+Bi; detect the maximum state probability E i; calculate the following probability values: V ′(xj)=2 Log[&Sgr; exp−Ei/xj−Ei)]−Log[&Sgr; exp(Ei+Fi/xj−Ei)]  (values propagated backwards, thus giving the information bits xj) V ′(rj)=2 Log[&Sgr; exp−Ei/rj−Ei)]−Log[&Sgr; exp(Ei+Bi/rj−Ei)]  (values propagated forwards, thus giving the redundancy bits rj) where (Ei/xj→Ei) represents the probability of state Ei connected to output xj.50. Decoding method according to claim 49, characterized in that the said step to calculate the probability values is simplified by only including the state with the greatest probability in each of the exponential sums.51. Method for processing data that can be used for coding and for decoding data, comprising: at least two coding/decoding steps ( 21j) each performed by at least two basic coding modules (22i,j), each of the said coding/decoding steps receiving a series of data to be proc essed distributed between the said basic coding modules, and outputting a series of processed data, corresponding to the said basic coding modules; and at least one mixing step ( 23j), the said mixing step being inserted between two successive coding/decoding steps, a first coding/decoding step and a second coding/decoding step, and distributing the processed data output from each basic coding/decoding module of the said first coding step between at least two basic coding modules of the said second step.52. Method for processing data according to claim 51, characterized in that at least some of the said calculations are made using preprogrammed tables.53. Method for processing data according to claim 51, characterized in that at least some of the said calculations are done using analog components.54. Method for processing data according to claim 51, characterized in that at least one of the said elementary codings and/or at least one or the said elementary decodings and/or at least one or the said elementary mixings can be programmed.55. Method for processing data according to claim 54, characterized in that programmable decoding elements are programmed using at least one of the following operating methods: as a function of a predetermined command produced by coding; as a function of a reference sequence coded by coding; by blind learning starting from the received data. 56. Method for processing data according to claim 51, characterized in that the said coding performs combined source coding and channel coding, and the said decoding correspondingly performs combined channel decoding and source decoding.57. Coding and/or decoding method according to claim 56, characterized in that the said decoding also at least partially equalizes transmission channel effects.58. Method for processing data according to any claim 51, characterized in that at least one of the said basic coding and decoding modules use a trellis, and in that the said coding and mixing steps for coding, and the said decoding and mixing steps for decoding, are identical.59. Method for processing data according to claim 58, characterized in that the same structure is used for coding and decoding.60. Method for processing data according to claim 58, characterized in that said trellis is a trellis like that illustrated inFIG. 11, for which the output labels “01” and “10” are reversed.61. Device for decoding a coded data block of a coded signal comprising at least some source data and at least some processed data output by a last coding stage of a coding method, characterized in that it comprises: at least two decoding stages each using at least two basic decodings, each of the said decoding stages receiving a series of data to be processed, distributed between the said basic decodings and outputting a series of processed data corresponding to the said basic decodings; and at least one mixing stage, the said mixing stage being inserted between two successive decoding stages, a first decoding stage and a second decoding stage, and distributing the processed data corresponding to each basic decoding of the said first decoding stage between at least two basic decodings of the second stage; said block of decoded data to be transmitted comprising at least some of said source data and at least some processed data output from the last decoding stage. 62. Transmission and/or reception method comprising a step in which at least one decoded signal is transmitted according to the method in claim 61.63. Application of the decoding method according to claim 61, to at least one of the techniques belonging to the group comprising: transmission and/or broadcasting of digital signals; speech recognition and/or processing.