Abstract:We study the problem of representing a discrete tensor that comes from finite uniform samplings of a multi-dimensional and multiband analog signal. Particularly, we consider two typical cases in which the shape of the subbands is cubic or parallelepipedic. For the cubic case, by examining the spectrum of its corresponding time- and band-limited operators, we obtain a low-dimensional optimal dictionary to represent the original tensor. We further prove that the optimal dictionary can be approximated by the famous \ac{dpss} with certain modulation, leading to an efficient constructing method. For the parallelepipedic case, we show that there also exists a low-dimensional dictionary to represent the original tensor. We present rigorous proof that the numbers of atoms in both dictionaries are approximately equal to the dot of the total number of samplings and the total volume of the subbands. Our derivations are mainly focused on the \ac{2d} scenarios but can be naturally extended to high dimensions.
What problem does this paper attempt to address?
This paper mainly discusses the approximate representation of multi-dimensional, multi-band signals. In signal processing, a fundamental task is to represent or approximate a signal using a few atoms from a dictionary. For one-dimensional multi-band signals, existing research has utilized the Discrete Prolate Spheroidal Sequences (DPSS) to construct low-dimensional dictionaries for approximation. However, there is limited research on multi-dimensional signals, especially for subbands with a cubic or parallelepiped shape.
The paper first reviews the theory of one-dimensional multi-band signals, including band-limited, bandpass, and multi-band signals, particularly their eigenvalue distributions of covariance matrices. For band-limited signals, the eigenvalues exhibit obvious clustering phenomena, which allows for the construction of low-dimensional approximations using the first few eigenvectors. Although similar clustering properties exist for bandpass and multi-band signals, the analysis is more complex.
Next, the paper extends to multi-dimensional signals, with particular focus on subbands with a cubic or parallelepiped shape. In the two-dimensional scenario, the paper analyzes the analogous properties of the "covariance matrix" of multi-dimensional, multi-band signals and proves the existence of low-dimensional dictionaries for approximating the original signals. For subbands with a cubic shape, the paper demonstrates how to construct this dictionary by modulating the baseband DPSS, thereby simplifying practical applications. Although the analysis is more complex for subbands with a parallelepiped shape, it also suggests the existence of similar eigenvalue clustering phenomena.
The main contributions of the paper are as follows:
1. Determine the number of atoms in the low-dimensional dictionary for subbands with a cubic shape in multi-dimensional, multi-band signals, as well as the eigenvalue distributions of these signals.
2. Provide a method for approximating multi-dimensional, multi-band dictionaries using modulated baseband DPSS to reduce computational complexity.
3. Analyze the multi-dimensional, multi-band signals of subbands with a parallelepiped shape and prove the existence of a similar eigenvalue clustering phenomenon.
Based on these results, the paper provides a theoretical foundation for the efficient representation and approximation of multi-dimensional, multi-band signals.