Moment Estimation and Dithered Quantization

This letter examines the influence of low-bit quantization on moment estimators with special emphasis on the 1-bit case. Moment estimators are especially useful if no prior knowledge on the distribution of the observations is available or if an ML approach is analytically intractable or computationally infeasible. In order to arrive at analytical results for this very general case, we focus on a dithered quantization scheme that allows us to specify and analyze its asymptotic behavior. We show that consistency can be retained under mild conditions, and furthermore, we quantify the asymptotic variance. Additionally, we illustrate how to find an estimator that achieves the best performance possible in this scenario. Finally, we bolster our analytical results with simulations for the illustrative case of an AR(1) process and provide a comparison with undithered schemes. A conclusion summarizes this letter's contribution and explores possible areas of application

[1]  Brian M. Sadler,et al.  Monobit digital receivers for ultrawideband communications , 2005, IEEE Transactions on Wireless Communications.

[2]  Helmut Berndt,et al.  Correlation function estimation by a polarity method using stochastic reference signals , 1968, IEEE Trans. Inf. Theory.

[3]  David L. Neuhoff,et al.  Quantization , 2022, IEEE Trans. Inf. Theory.

[4]  B. Porat,et al.  ARMA parameter estimation based on sample covariances, for missing data , 1986, ICASSP '86. IEEE International Conference on Acoustics, Speech, and Signal Processing.

[5]  J. B. Knowles,et al.  Correlating Devices and Their Estimation Errors , 1967 .

[6]  R. Gray,et al.  Dithered Quantizers , 1993, Proceedings. 1991 IEEE International Symposium on Information Theory.

[7]  Ananthram Swami,et al.  Minimax Quantization for Distributed Maximum Likelihood Estimation , 2006, 2006 IEEE International Conference on Acoustics Speech and Signal Processing Proceedings.

[8]  Daniel S. Ruchkin,et al.  Error of correlation coefficient estimates from polarity coincidences (Corresp.) , 1965, IEEE Trans. Inf. Theory.

[9]  Brian M. Sadler,et al.  Fast estimation of higher-order moments using sign bit and reference signals , 1996, 1996 IEEE Digital Signal Processing Workshop Proceedings.

[10]  John A. Gubner,et al.  Distributed estimation and quantization , 1993, IEEE Trans. Inf. Theory.

[11]  J. V. Vleck,et al.  The spectrum of clipped noise , 1966 .

[12]  Zhi-Quan Luo,et al.  Universal decentralized estimation in a bandwidth constrained sensor network , 2005, IEEE Transactions on Information Theory.