Parameterization of nonideal quantizers for simultaneous estimation of quantizer and excitation signal parameters

This paper deals with the estimation of signal and quantizer parameters based on a nonideally sampled and quantized measurement record. It proposes solutions to reduce the inherently high-dimensional parameter space via approximation of the quantizer nonideality to make the computation possible in PC/MATLAB environment. The performance of the approximation techniques is examined as well. The computational demand of the proposed approximate maximum likelihood (AML) estimators is investigated and compared to the computational demand of the maximum likelihood (ML) estimators without approximation.