New orthogonalised IIR adaptive filter

A new IIR-ADF algorithm is proposed in which the discrete cosine transform, a fairly good approach of KLT for a large number of signal classes, is used to split the input signal into a set of nearly orthogonal transform coefficients. Subsequently these coefficients are filtered by a bank of IIR-ADFs whose coefficients are independently updated to minimise a common error. Computer simulations show that the proposed algorithm provides a faster convergence rate than the Gauss-Newton algorithm with a much reduced computational load