In *2015 23rd European Signal Processing Conference (EUSIPCO)*, pages 2221-2225, Aug, 2015.

doi abstract bibtex

doi abstract bibtex

We derive the Cramér-Rao lower bounds (CRB) for parametric estimation of the number-weighted particle size distribution (PSD) from multiangle Dynamic Light Scattering (DLS) measurements. The CRB is a useful statistical tool to investigate the optimality of the PSD estimators. In the present paper, a Gaussian mixture (GM) model of the multimodal PSD is assumed and the associated Fisher information matrix (FIM) is determined. The usefulness of multiangle DLS in significantly decreasing the CRB is demonstrated. The mean square error (MSE) of the PSD GM model parameters estimation by the Bayesian inference method proposed in [1] is compared to the derived CRB for a simulated monomodal PSD. Results show that the MSE achieves the derived CRBs for the unbiased estimators of the PSD GM model parameters.

@InProceedings{7362779, author = {A. Boualem and M. Jabloun and P. Ravier and M. Naiim and A. Jalocha}, booktitle = {2015 23rd European Signal Processing Conference (EUSIPCO)}, title = {Cramér-Rao bounds for particle size distribution estimation from multiangle dynamic light scattering}, year = {2015}, pages = {2221-2225}, abstract = {We derive the Cramér-Rao lower bounds (CRB) for parametric estimation of the number-weighted particle size distribution (PSD) from multiangle Dynamic Light Scattering (DLS) measurements. The CRB is a useful statistical tool to investigate the optimality of the PSD estimators. In the present paper, a Gaussian mixture (GM) model of the multimodal PSD is assumed and the associated Fisher information matrix (FIM) is determined. The usefulness of multiangle DLS in significantly decreasing the CRB is demonstrated. The mean square error (MSE) of the PSD GM model parameters estimation by the Bayesian inference method proposed in [1] is compared to the derived CRB for a simulated monomodal PSD. Results show that the MSE achieves the derived CRBs for the unbiased estimators of the PSD GM model parameters.}, keywords = {Bayes methods;Gaussian processes;inference mechanisms;light scattering;mean square error methods;mixture models;parameter estimation;particle size;Cramér-Rao bounds;particle size distribution estimation;multiangle dynamic light scattering;CRB;parametric estimation;number-weighted particle size distribution;PSD;DLS;Gaussian mixture model;Fisher information matrix;FIM;mean square error;MSE;Bayesian inference method;Estimation;Light scattering;Mean square error methods;Bayes methods;Robustness;Europe;Signal processing;Particle Size Distribution;Multiangle Dynamic Light Scattering;Cramér-Rao Bound;Inverse Problem;Bayesian Inference}, doi = {10.1109/EUSIPCO.2015.7362779}, issn = {2076-1465}, month = {Aug}, }

Downloads: 0