Optimal One-Bit Quantizers - Is the optimum quantization threshold symmetric or asymmetric?

Dr.-Ing. Gholamreza Alirezaei

RWTH Aachen University
Institute for Theoretical Information Technology

Abstract:

A fundamental problem in signal processing is the specification of optimal data quantization for noisy signals. Using information theoretic metrics, the optimal solution seems to be extremely difficult to achieve. Even for one-bit quantizers, determining the capacity-maximizing quantization-threshold is a challenging problem. Only for few particular noise distributions an optimal quantization-threshold can be deduced. In the presentation, an information theoretic approach for describing one-bit quantizers is discussed and the capacity-maximizing quantization-threshold is deduced for both uniformly and Gaussian distributed noise.

Biography:Biography:

Gholamreza Alirezaei is born in 1976 in Isfahan/IRAN. From 1996 to 2001 he studied electrical engineering with focus on telecommunications at RWTH University in Aachen/GERMANY. He also received his doctoral degree with distinction in electrical engineering from the same university. In addition he studied mathematics and business administration concurrently. He worked at different companies and was research and development leader from 2008 to 2011. Since 2011 he is working at the Institute for Theoretical Information Technology at RWTH University. Currently he is a lecturer for "Advanced Topics in Information Theory" at both universities RWTH and TUM. He is holder of several prestigious awards and recognitions for his contribution in signal processing, optimization and data communications. His research interests are widespread and include design and implementation of RF circuits, development of target estimation and classification algorithm, determination of mathematical bounds and inequalities, optimization of communication systems, and efficient processing of information.