Abstract: A method of transmitting QPSK digital signals in which each 2-digit binary number, referred to as a symbol, is assigned a phase of a carrier, and symbols are added to enable error correction at the receiver. The error correction code is a product code. To obtain transparency to phase rotations, the I bits (I1, I3, etc.) and the Q bits (Q2, Q4, etc.) of a symbol to be transmitted are placed in different rows but in the same column, so that each row contains only one type of bit, I or Q. The error correction coding is effected: row by row (NC-KC), and by pairs of adjacent columns, and the code I (or Q) bits of two associated adjacent columns are deduced from the I (or Q) bits of the two columns.

Abstract: It is proposed to use channel coding involving a plurality of concatenated coders and decoders in series and in parallel for coded transmission in which the efficiency can be adapted as a function of the requirements of a system and an application. The coders of the invention are convolutional self-orthogonal code (CSOC) coders. It is shown that this enables perfect adaptation with circuits whose complexity and cost are low and which therefore support high bit rates.

Abstract: A method of soft input to soft output decoding of a word s of a block linear code of dimension k and length n received from a transmission channel is provided, including generating a list of firm words (ub) of the code close to the received code word (s) by coding a list of k-tuplets obtained by firmly approximating components of the received word and changing the least likely components, and calculating the jth component of the output soft word as the difference between the metrics of the closest generated code word and the closest generated code word having an opposite jth component, or, failing this, the farthest generated code word. Also provided is a method of iterative decoding of a product code word received on a transmission channel using the soft input to soft output decoding method. It rapidly and efficiently decodes product code words without an algebraic decoder.

Abstract: A method of transmitting QAM-16 modulated digital signals in which each of sixteen states represents a symbol with four bits and is represented in a complex plane by a point with particular coordinates. In each quadrant of the complex plane, four points represent four numbers each of four bits, in which numbers the last two bits are the same and the first two bits represent numbers that are all different. The numbers are divided into four subsets over the whole of the complex plane. Each subset is formed by the set of numbers having the same first two bits, so that the numbers are therefore in the four different quadrants. The coordinates of the symbols in the complex plane are chosen so that each subset takes the place of another subset after a rotation of +K&pgr;/2, where K is an integer.

Abstract: The invention relates to a method of transmitting QPSK digital signals in which each 2-digit binary number, referred to as a symbol, is assigned a phase of a carrier, and symbols are added to enable error correction at the receiver.

Abstract: The invention provides a block-encoded modulation scheme using multi-level partitioning techniques. This scheme is made transparent to phase ambiguities of ±&pgr;/2 and of &pgr;, by means of differential encoding and appropriate mapping, it is applicable to 16-QAM modulation, and it has theoretical encoding gain that is optimal for the rate of the code. The decoder associated with this scheme uses the Wagner algorithm which is much less complicated to implement than the Viterbi algorithm or than the Reed-Solomon algorithm.

Abstract: The invention relates to a method of encoding a data block comprising first and second zones, the method being of the type consisting in applying product encoding to the block. It comprises first block encoding applied to the first zone, second block encoding applied to the second zone, and third block encoding applied to the data obtained by the first two encodings, but in a direction orthogonal to the first two codes, the three encodings being performed using BCH codes, the first zone being protected by a code that is more powerful than the second zone. The block is constituted by an ATM cell, the first zone corresponding to the header of the ATM cell, and the second zone corresponding to the payload data field of said ATM cell.

Abstract: A branch synchronization device for multidimensional trellis coded digital data receivers using convolutional encoders includes a Viterbi decoder receiving two streams of symbols and supplying coded sequences and a time shifter for time shifting the symbol streams fed to the Viterbi decoder. The time shifter is controlled by a branch synchronization decision device generating a time shift control signal. The branch synchronization decision device cooperates with a calculator circuit receiving the coded sequences and supplying to the synchronization decision device an output signal of a first type if the coded sequences correspond to sequences such as would be obtained directly from the convolutional encoder and supplying an output signal of a second type if the coded sequences do not correspond to sequences that could have been obtained from the convolutional encoder. This output signal is fed to the branch synchronization device.

Abstract: The invention relates to an 8-PSK differential coder for trellis-coded modulation and to a corresponding pragmatic decoder, enabling 90.degree., 180.degree., and 270.degree. phase ambiguities to be lifted. The invention applies in particular to transmitting images at high rate by radio link.