Method and apparatus for decision feedback equalization with reduced convergence time

A decision feedback equalization apparatus for performing multiple iterations of a blind mode decision feedback equalization on a transmitted signal x(n) resulting from transmission of an input signal s(n) through a channel having a z-domain transfer function A(z), said apparatus including:

• means for generating a replica input signal s'(n) during each of the iterations;
• means for generating a feedback coefficient a'.sub.0 (n) in response to a feedback signal Ke(n)s'(n) during each of the iterations, where K is a convergence scaling factor;
• means for generating an error signal e(n) in response to the feedback coefficient a'.sub.0 (n) during each of the iterations;
• and
• adaptive filter means for processing the replica input signal s'(n) to generate a replica transmitted signal x'(n) in response to a set of adaptively generated coefficients a'.sub.j (n), where j=1, 2, . . . , N, and N is an integer, and wherein the adaptive filter means includes a means for generating each of the coefficients a'.sub.j (n) in response to a feedback signal 2Ke(n)s'(n-j), where K is said convergence scaling factor.

A decision feedback equalization apparatus for performing multiple iterations of a blind mode decision feedback equalization on a transmitted signal x(n) resulting from transmission of an input signal s(n) through a channel having a z-domain transfer function A(z), said apparatus including:

• means for generating a replica input signal s'(n) during each of the iterations;
• means for generating an error signal e(n) during each of the iterations;
• and
• means for generating an updated vector A'(n) of replica coefficient signals a'.sub.j (n) during each of the iterations, where A'(n)=A'(n-1)+KWS'(n-1)e(n-1), A'(n) is a transposed vector [a'.sub.0 (n) a'.sub.1 (n) . . . a'.sub.N (n)].sup.T of said replica coefficient signals a'.sub.j (n), with j=0, 1, . . . , N, A'(n-1) is a transposed vector [a'.sub.0 (n-1) a'.sub.1 (n-1) . . . a'.sub.N (n-1)].sup.T of coefficient signals generated during an iteration immediately preceding said each of the iterations, S'(n-1) is a transposed vector [s'(n) s'(n-1) . . . s'(n-N)].sup.T of said replica input signal s'(n) and phase-delayed replica input signals s'(n-m), where m=1, 2, . . . ,N, where e(n-1) is an error signal generated during an iteration immediately preceding said each of the iterations, K is a convergence scaling factor, and W is a diagonal (N+1).times.(N+1) matrix whose first diagonal coefficient signal has a value equal to one, whose other diagonal coefficient signals each have a value equal to two, and whose off-diagonal coefficient signals each have a value equal to zero.