Optimal measurements of linear approximation to Volterra TITO systems

Nonlinear system described with Volterra series, when excited with random multisines, can be viewed as a linear (best approximation) frequency response function covered by an additive nonlinear noise source. This model can tie easily extended to two-input, two-output systems, however the usual measurement procedure can produce very large variance. Using the analogy to the process noise we show, that the optimized measurements proposed for linear multidimensional systems work well also for the nonlinearity induced noise. Such measurement procedure is proposed as a universal tool to get rid of any disturbing effects.