MMD-ARMA approximation to the Volterra series expansion

Nonlinear filtering based on the Volterra series expansion is a powerful universal tool in signal processing. Due to the problem of increased complexity for higher orders and filter lengths, approximations up to third order nonlinearities using linear FIR-filters and multipliers have been developed earlier called multimemory decomposition (MMD). In our paper we go a step further in this approach using ARMA-filters instead which leads to a reduction in the number of coefficients to about 50% for similar system functions. The good performance of this new approach is demonstrated by means of a processor designed for identification of nonlinear loudspeaker distortions.