Some improvements of the linear subspace algorithm swede for time delay estimation

In this paper, some improvements on the linear subspace algorithm SWEDE are proposed in order to adapt this algorithm to the estimation of the time delay from radar data in microwave range. At first, the radar pulse is taken into account in the algorithm. Then, the noise whitening technique in [2] is improved by a full block diagonalization of the noise covariance matrix. Finally, SWEDE is used in conjunction with ESPRIT to make the time delay estimation more efficient. Both algorithms are tested on simulated data to measure the layer thickness of materials for civil engineering applications.