Quantization for spectral super-resolution

TL;DR

This paper demonstrates that distributed noise-shaping beta-quantization significantly improves spectral super-resolution accuracy over naive methods, especially with measurement redundancy, by providing explicit error bounds for various algorithms.

Contribution

It introduces a novel quantization approach for spectral super-resolution that achieves superior reconstruction guarantees compared to traditional scalar quantization methods.

Findings

Quantization method guarantees reconstruction accuracy of order $O(M^{1/4}\lambda^{5/4} K^{- \lambda/2})$ with TV-min/BLASSO.

Alternative guarantees of order $O(M^{3/2} \lambda^{1/2} K^{- \lambda})$ with ESPRIT.

Naive rounding offers only $O(M^{-1}K^{-1})$ accuracy, regardless of algorithms.

Abstract

We show that the method of distributed noise-shaping beta-quantization offers superior performance for the problem of spectral super-resolution with quantization whenever there is redundancy in the number of measurements. More precisely, we define the oversampling ratio $\lambda$ as the largest integer such that $\lfloor M/\lambda\rfloor - 1\geq 4/\Delta$, where $M$ denotes the number of Fourier measurements and $\Delta$ is the minimum separation distance associated with the atomic measure to be resolved. We prove that for any number $K\geq 2$ of quantization levels available for the real and imaginary parts of the measurements, our quantization method combined with either TV-min/BLASSO or ESPRIT guarantees reconstruction accuracy of order $O(M^{1/4}\lambda^{5/4} K^{- \lambda/2})$ and $O(M^{3/2} \lambda^{1/2} K^{- \lambda})$ respectively, where the implicit constants are independent of $M$, $K$ and $\lambda$. In contrast, naive rounding or memoryless scalar quantization for the same alphabet offers a guarantee of order $O(M^{-1}K^{-1})$ only, regardless of the reconstruction algorithm.