A novel method for identifying cubically nonlinear systems using minimally bandpass sampled data

Identification of cubically nonlinear systems with a bandpass input signal is investigated in this paper. When excited by a bandpass signal, a nonlinear system may generate an output signal which occupies multiple frequency bands. Therefore, the output signal must be bandpass sampled with caution to avoid aliasing; i.e., for a cubically nonlinear system excited by a bandpass input, the bandpass sampling frequency for the output signal needs to be at least eighteen times the bandwidth of the input signal, otherwise the output signal could be aliased. In this paper, we propose a novel method for identifying the cubically nonlinear system using the input and output data sampled at twice the bandwidth of the input signal, even though aliasing exists in the sampled output under the circumstances. The significantly lower sampling rate makes the proposed method highly appreciated when high speed sampling is unattainable.