Abstract

In this paper, we address the system identification problem of a continuous-time stochastic system. We propose an approach that involves approximating the continuous-time system through continuous-time basis functions, utilizing sampled data. Furthermore, we assume access only to discrete-time output measurements. The estimation problem is formulated using the Maximum Likelihood approach in the frequency domain, taking into account an approximation of the output power spectrum with basis functions, called Tsai's approximation. The benefits of our proposal are illustrated via numerical simulations.