Identification of factorable Volterra systems

Identification of systems which can be represented as a finite sum of factorable Volterra kernals, each composed of individual linear dynamic subsystems, connected in parallel with outputs multiplied in the time domain, is considered. Both a multilevel-testing and a sequential-single-test procedure are derived to isolate each factorable kernel, and algorithms, which provide estimates of the individual linear subsystems associated with each kernal, are formulated using correlation techniques based on either Gaussian or pseudorandom excitation.