Abstract: A method of updating the coefficients of an adaptive filter, where the filter utilizes a polarity-coincidence correlator (PCC) as part of an iterative update equation. The expression for the standard PCC correlator is modified to correct for its inherent distortion by operating on the update term with a factor of the form of sin [.pi./2 (f)], where f represents the PCC correlator. This produces an update equation and correlator which behaves like the well known least mean square (LMS) update algorithm in many respects, but is simpler and more cost-effective to implement in many situations. Architectures for both a direct form and transpose form implementation of the present invention are also disclosed. An adaptive filter based on the present invention will exhibit a very robust convergence behavior, although it will be slower than that of a standard LMS based filter. In addition, the steady state noise of the filter can be made to approach the theoretical minimum.

Abstract: A method and apparatus for compensating for post-decision feedback equalization noise during decision feedback equalization. The method includes the steps of performing decision feedback equalization on a noisy signal, while isolating noise components of the noisy signal and filtering the isolated noise components to generate post-decision feedback equalization noise having desired spectrum and power. Preferably, the noisy signal is pre-filtered in a feedforward filter and the isolated noise components are filtered to compensate for noise enhancement that has resulted from filtration in the feedforward filter. The apparatus of the invention is a decision feedback equalizer including a quantizer and a noise filter circuitry for controlling the spectrum and power of the noise component of the signal received by the quantizer.

Abstract: A method and apparatus for performing blind mode decision feedback equalization with reduced convergence time. In operation, the decision feedback equalizer (DFE) of the invention receives a transmitted signal x(n) resulting from transmission of an input signal s(n) through a transmission channel and a receive filter. The apparatus includes a DFE feedback loop having an adaptive filter designed to converge rapidly and iteratively during blind mode decision feedback equalization to a best approximation A'(z) of the unknown, combined, transmission channel and receive filter transfer function A(z). In a preferred embodiment, the apparatus includes DFE circuitry for implementing a Widrow-Hoff least-mean-squared adaptation process, including circuitry for generating an input signal replica s'(n) and an error signal e(n) during each iteration of the adaptation process. In this embodiment, the adaptive filter adaptively generates N replica coefficients a'.sub.j (n), where j=1, 2, . . .

Abstract: The present invention provides a technique for updating the coefficients a(n) of an adaptive filter impulse response A(n), whereA'(n)=(a.sub.o (n) a.sub.l (n) . . . a.sub.m (n))is the adaptive filter impulse response at time n. The impulse response is utilized for tracking the output y(n) of a communications channel, where ##EQU1## where H is an unknown impulse response of the channel,H'=(h.sub.o h.sub.l . . . h.sub.m),the ' character denoting transposition. The channel is driven with a known input signal s(n), whereS'(n)=(s(n) s(n-l) . . . s(n-m)).According to the techinque, the output x(n) of an adaptive filter, wherex(n)=S'(n)A(n)is compared with the channel output y(n) according to the Least Mean Squares (LMS) algorithm to develop an LMS error prediction signal. At the same time, the adaptive filter output x(n) is compared with the channel output y(n) according to the Sign Algorithm (SA) to develop an SA error prediction signal.

Abstract: The present invention provides a technique for data detection signal processing for removing echo interference and/or distortion from a received 2B1Q coded signal. According to the preferred technique, the received 2B1Q signal Q, where q=(-3, -1, +1, +3), is shifted one bit to the left to generate a corresponding shifted symbol alphabet S, where S=(-2, 0, +2, +4). The shifted received signals are then processed through a transversal filter. That is, the signal is provided to a delay line which is tapped at intervals corresponding to the symbol width of the received signal. Each tap along the delay line is connected through an amplifier to a summing device that provides an output y(n), where ##EQU1## The tap gains, or coefficients a.sub.j, are set to subtract the effects of interference from symbols that are adjacent in time to the desired symbol. In an adaptive embodiment of the invention, the output y(n) resulting from the convolution between the transmitted symbols s(n) and the gain coefficients a.sub.j.