On reduced complexity IIR adaptive filters

Reduced complexity IIR adaptive filters, like Laguerre adaptive filters with pole updating are being considered as an alternative to full adaptive IIR schemes. On the other hand, low complexity realizations exist that allow a better compromise between filter-order and modeling performance. We discuss, in this work, characteristics and properties of a reduced complexity IIR orthogonal realization that improve on those of the Laguerre and sub-optimal IIR counterparts. After reviewing the Laguerre and sub-optimal adaptive filters with pole updating, we introduce a reduced complexity IIR adaptive filter based on the Steiglitz-McBride method. The particular realization proposed is based on a generalized orthonormal family of functions. The Steitglitz-McBride method is discussed mainly because it can avoid the local minima problem related to direct mean squared output error minimization. Computer simulations are given to illustrate the good performance of the proposed adaptive IIR filter with respect to computational complexity and modeling capability if compared with its Laguerre counterpart.