Iterative algorithm for identification of third order Volterra systems

An iterative procedure for the identification of third-order finite-memory time-invariant Volterra nonlinear systems is presented. The data formulation is similar to the covariance method used in linear signal modeling problems, and the resulting correlation matrix is a block matrix with a mix of higher-order auto- and cross-correlation elements. The procedure takes advantage of the block structure of the correlation matrix to achieve computational reduction. The method is suitable for short data record lengths. Simulation results are presented to exhibit the iteration convergence performance of the proposed algorithm.<<ETX>>