Electrical Engineering-Systems Dept.

סמינר מחלקתי

You are invited to attend a lecture by

Dr. Hagai Kirshner

(École Polytechnique Fédérale de Lausanne)

on the subject:

On Continuous-Time ARMA Identification

Continuous-time ARMA (Auto Regressive Moving Average) modeling provides a useful parameterization for describing phenomena in biology, astronomy and economy, to name a few. Within the context of signal processing, it will be shown that these stochastic models are also related to deterministic regularization. This relation can, in turn, be exploited for adapting the regularization operator to the given data. To this aim, I will address the problem of estimating continuous-time ARMA parameters from sampled data while imposing no restrictions on sampling interval values. I will consider the following aspects: uniqueness of the sampled process, continuous-to-discrete mapping of the parameters, and maximum-likelihood estimation. The proposed estimator is designed to exploit, rather than avoid, the aliasing of the continuous-time power-spectrum, which is a key point for designing sampling-invariant signal processing algorithms.