Abstract: In a first data receiver (10), sampling circuitry (20, 25, 35) forms samples of a received data signal representing a succession of data symbols. The samples are formed at twice the symbol rate. Transversal filter circuitry (251, 261) operates on the samples by multiplying them by respective ones of a queue of coefficients. Further circuitry (55, 60, 65, 70, 252, 262), operative in response to the resulting products, forms decisions as to the values of the transmitted symbols and updates the values of the coefficients. Timing recovery circuitry (40) periodically identifies the largest of the coefficients in magnitude and either advances or retards the operation of the sampling circuitry by a fixed step size depending on whether that coefficient is or is not within a predetermined portion of the queue.

Abstract: In a multipoint data communication system using quadrature-amplitude modulation, a master modem (20) and a plurality of tributary modems (11a, 11b . . . 11n) are interconnected via respective transmission channels (13a, 13b . . . 13n, 16). Adaptive equalizer circuitry (55, 56) in the master modem equalizes the channel from a particular tributary by multiplying samples of signals received from the tributary by an ensemble of tap coefficients associated with the tributary. The tap coefficient ensembles for each tributary are stored in a memory (91) from which they are retrieved at the start of transmission from that tributary. Timing-acquisition circuitry (29) within the master modem adjusts the phase of the latter's sampling circuitry (23, 27) at the start of transmission from a given tributary so that the received signals are sampled at the correct time points.

Abstract: A quadrature amplitude modulated (QAM) data signal transmitted at T symbols per second is sampled in a data receiver at a rate of 2/T samples per second and applied to a transversal-type equalizer structure (25, 46, 34, 35) having taps spaced at T/2 second intervals. A demodulated equalized signal (a.sub.j, b.sub.j), generated once every T seconds, is quantized to form a decision (a.sub.j *, b.sub.j *) as to the value of the original modulating data symbol. An error signal (e.sub.j, e.sub.j) is formed in response to the pre- and post-quantized values of the demodulated equalized signal. Tap coefficients (c.sub.i (j), c.sub.i (j)) used in generating the equalized signals are updated in response to (a) a correction term which is a function of the error signal and (b) a predetermined tap leakage term which has a constant magnitude. The introduction of the tap leakage term maintains the coefficient values at minimum levels.