Title :
Sigma-Delta Quantization of Geometrically Uniform Finite Frames
Author :
Abdelkefi, Fatma
Author_Institution :
Lab. of Commun. Technol., Swiss Fed. Inst. of Technol., Zurich, Switzerland
Abstract :
We consider sigma-delta quantization of cyclic geometrically uniform (CGU) finite frames, family of frames containing finite harmonic frames (both in CM and RM). For first- and second-order sigma-delta quantizers, we establish that the reconstruction minimum squares error (MSE) behaves as 1/r2 where r denotes the frame redundancy. This result is shown to be true both under the quantization model used in J.J. Benedetto et al. (2004) as well as under the widely used additive white quantization noise assumption. For the widely used L-th order noise shaping filter G(z) = (1 - z-1)L, we show that the MSE behaves as 1/r2 irrespectively of the filter order L. More importantly, we prove also that in the case of tight and normalized CGU frame, when the frame length is too large compared to the filter order, the reconstruction MSE can decay as faster as O(1/r2L+1).
Keywords :
filtering theory; mean square error methods; quantisation (signal); white noise; L-th order noise shaping filter; MSE; additive white quantization noise; geometrically uniform finite frames; minimum squares error; sigma-delta quantization; Additive noise; Additive white noise; Analog-digital conversion; Bit rate; Communications technology; Delta-sigma modulation; Laboratories; Noise shaping; Quantization; Redundancy; CGU frames; Overcomplete representations; Sigma-Delta quantization;
Conference_Titel :
Acoustics, Speech and Signal Processing, 2007. ICASSP 2007. IEEE International Conference on
Conference_Location :
Honolulu, HI
Print_ISBN :
1-4244-0727-3
DOI :
10.1109/ICASSP.2007.367130