Parameter estimation and waveform design for cognitive radar by minimal free-energy principle

In this paper we develop a new framework for Bayesian parameter estimation using adaptive waveforms by the minimal free energy (FE) principle in the context of cognitive radar. Unlike conventional approaches, the new method utilizes the minimal FE principle as a unifying criterion for optimal estimator design and waveform design. The FE principle seeks to approximate the true density of the unknown parameters in response to sequential measurement data. In the case of a single unknown parameter we show that the estimators based on the FE principle and the conventional Bayesian estimator are identical. Moreover, the waveform design based on the FE principle results in similar water-filling solution as the traditional mutual information method.