Abstract: A learning and decoding technique is provided for a neural network. The technique involves using a set of neurons, referred to as beacons, wherein said beacons are binary neurons capable of assuming only two states, an on state and an off state. The beacons are distributed in blocks of a predetermined number of beacons and being allocated for processing a sub-message. Each beacon is associated with a specific occurrence of the sub-message. Learning includes splitting a message into B sub-messages to be learned, where B is greater than or equal to two; activating, for a sub-message, a single beacon in each block to be in the on state, all of the other beacons of the block being in the off state; and activating binary connections between the on beacons of each of the block for a message to be learned, which assume only connected and disconnected states.

Abstract: A scheme for coding digital source data, implementing, in parallel, at least two elementary coding steps each taking into account all of the source data, and including a permutation step modifying a sequence of consideration of the source data between the elementary coding steps. In one example, the source data is organized in sequences of non-binary symbols, each non-binary symbol including n binary source elements, n being greater than or equal to 2. The example permutation step includes reversibly modifying contents of at least some of the non-binary symbols, delivering modified non-binary symbols, and permutating the sequence of the source or modified non-binary symbols.

Abstract: Disclosed are a method and a device for the convolutive encoding of blocks each formed by a predetermined number N of source data elements, wherein each of said source data elements is introduced twice into one and the same convolutive encoder implementing a generating polynomial with a period L in an order such that the two instances of introduction of one and same source data element d.sub.i are separated by the introduction of (p.sub.i.L)-1 other source data elements, p.sub.i being a non-zero integer. Also disclosed are a corresponding decoding method and device that can be applied, in particular, to the transmission of short messages, for example in radiotelephony, for satellite communications or gain computer telecommunications (Internet for example). FIG. 2.

Abstract: The invention relates to a device for the reception of signals formed by a series of digital symbols corresponding to the convolutive encoding of items of source digital data comprising p cascade-connected detection and decoding modules M.sub.1 to M.sub.p, p being greater than or equal to 2, each of said modules M.sub.i including: (a) inter-symbol interference correction means supplied by a symbol input R.sub.i and delivering estimated symbols A.sub.i,1 with weighted values, (b) means for the decoding of said estimated symbols A.sub.i,1, performing operations symmetrical to said convolutive encoding and delivering decoded symbols A.sub.i,2 with weighted values, and (c) means for the computation of an item of correction information Z.sub.i+1 from said estimated symbol A.sub.i,1 and said decoded symbol A.sub.i,2, (d) said means for the correction of each of said modules M.sub.i, except for the first module M.sub.1, taking account of at least one item of correction information Z.sub.

Abstract: The transmitted bits are coded according to the product of at least two systematic block codes. An iterative decoding is applied in order to determine, at each code word search step, a data matrix and a decision matrix which are used for the following step. The new decision matrix is determined at each step by decoding the rows or the columns of the input matrix, and the new data matrix is determined by taking account of correction terms which increase the reliability of the decoding at each iteration. The method is especially suited for use with high-efficiency block codes.

Abstract: An error-correction method for the coding of source digital data elements to be transmitted or broadcast, notably in the presence of high transmission noise, the method comprising at least two independent steps of systematic convolutional coding, each of the coding steps taking account of all of the source data elements, the method comprising at least one step for the temporal interleaving of the source data elements, modifying the order in which the source data elements are taken into account for each of the coding steps, and a corresponding iterative decoding method using, for the decoding, at each iteration, an intermediate data element obtained by the combination of the received data element with a data element estimated during the previous iteration.

Abstract: A method for the decoding of digital data obtained according to a convolutional coding and transmitted on a noise-ridden channel aimed at improving the methods of maximum likelihood decoding such as the Viterbi algorithm consisting in associating, with each estimation s(t) of the corresponding source digital element a(t), a weighting coefficient m(t) representing the reliability of the estimation s(t), the estimation s(t) taking account of the values of a first series of L received samples, and the weighting m(t) taking account of the values of a second series of N samples received, N being strictly greater than L. In a preferred embodiment, a second trellis is associated with the standard first decoding trellis, a trace-back being done in the second trellis on two concurrent paths, used to revise the weighting coefficients obtained at output of the first trellis.