Abstract: The echo canceller for bidirectional digital transmission systems cancels the echo queues of a desired number N of digital symbols, transmitted on the line in the last N signalling periods. It consists of a part which effects very simply the summation of the last N transmitted symbols, so as to ease the system implementation by an integrated circuit, and of a part which computes the estimated sample of the echo-channel pulse response. The results of the two parts are multiplied by one another and the estimated echo signal is obtained.

Abstract: A bidirectional digital transmission system with echo-cancellation allows the measurement of the received data signal and the automatic adaptation of the attenuator and of the line equalizer. They are implemented by using circuits already present in a conventional system, in addition to auxiliary circuits.

Abstract: A bidirectional transmission system with echo cancellation allows simultaneous interchange of digital information between an exchange and subscribers on a single transmission line. Data flows are transmitted by packets separated by short silence intervals within which clock signals are extracted and echo-cancellation adjustment is accomplished. Use of an adaptive equalizer and a single sampling operation per sampling period are possible.

Abstract: The digital circuit extracts synchronism signals from a serial flow of coded data according to a process which uses a phaselocked loop where the different level transitions of the signal received are separately considered and a phase is determined for each type, the latter being compared with a specific reference phase, established in the initialization step. The error signal obtained from the comparison is used for correcting synchronism signal frequency.

Abstract: In order to measure the amplitude of a periodic signal of known frequency which may be heavily contaminated by noise and whose phase is not known at the point of measurement, e.g. at an output end of an optical fiber, two samples are periodically taken in every cycle at instants spaced apart by a quarter-cycle. The two samples are accumulated over N cycles and the signs of the resulting algebraic sums are compared at every N.sup.th cycle to determine whether the sampling phases should be advanced or retarded within a cycle by a predetermined increment .DELTA. in order to let the first and the second sample of a pair respectively approach the peak and the zero crossing of a half-cycle of the signal wave at which point, with the noise more or less balanced out, the second sum is substantially zero while the first sum represents N times the actual signal amplitude.