Abstract: The method is founded on the GUAZZO algorithm and is adaptive as it does not require any prior statistical knowledge of source of message to be coded. The message is comprised of a sequence of symbols of the source which are represented by the indication of the direction "0" or "1" chosen at each node of a binary tree. According to the invention, the adaptiveness is obtained by associating with each node two finite-sized counters that count the "0"s and "1"s according to the routing of the tree. The contents of these counters enable the conditional probability of each of the elements "0" and "1" to be estimated at the node.

Abstract: A decoding process suitable for space links decodes a sequence of symbols received from a disturbed channel resulting from the convolutional coding of a sequence of information symbols to be transmitted. The trellis of the states of the coder is defined and, as a function of the symbols received, the most likely path in the trellis is determined and this defines the most probable sequence of symbols transmitted. At each node of the trellis, the most probable path or survivor is determined by a sequence of algebraic values indicating a sequence of transmitted symbols and the likelihood of each of these symbols and by measuring the cumulative likelihood of the sequence.

Abstract: In order to decode the bits of a message transmitted by an array of channels in parallel which have low reliability on account of the many disturbances which interfere with transmission, the binary decoding device comprises means for calculating the probability of error in each channel as a function of the coefficients of logarithmic likelihood of the decoded values and of the binary values received, as well as means for weighting the binary values received on each channel as a function of the calculated probability of error. In accordance with an optimum decision rule applied to the weighted values, a weighted decoding device delivers the binary values which are transmitted with the largest likelihood and the coefficients of logarithmic likelihood of these values.

Abstract: The transmitting station comprises coding means, including a shift-register having L stages, where L is the length of the code, for delivering continuation and repetition bit trains, each of which comprises information bits including at least L message bits as well as the parity bits associated with the information bits. In the case of a continuation bit train, the L message bits are supplied by a data source, in the case of a repetition bit train, they are fed back to the coding means from the shift-register. In the receiving station, the decoding of messages bits is a conventional decoding in accordance with the code where a continuation bit train is concerned, but also takes into account the previously decoded value where a repetition bit train is concerned. The trains may be identified by mean of prefixes of N.sub.1 bits, in which case, each bit train comprises (N.sub.1 + L) information bits.