Abstract
In this dissertation, we investigate periodicity estimation with nested periodic dictionaries (NPDs), a newly introduced family of matrices that can capture periodicity in the data. We address two main problems. First, we study the detection of periodic signals using their representations in NPDs. To this end, the detection problem is posed as a composite hypothesis testing problem, where the signals are assumed to admit sparse representations in a Ramanujan Periodicity Transform (RPT) dictionary as an instance of NPDs. For the binary case, we develop a generalized likelihood ratio detector and obtain exact distributions of the test statistics in terms of confluent hypergeometric functions, along with flexible approximate distributions. Subsequently, we extend our approach to multi-hypothesis and multi-channel settings, where we account for spatial correlations between the different channels. We study the application of the proposed method in the detection of periodic brain responses to external visual stimuli, known as steady-state visually evoked potentials (SSVEPs), which is fundamental to the development of Brain Computer Interfaces. Results based on experiments with synthesized and real-data demonstrate that the RPT detector outperforms conventional spectral-based methods. Second, we address the problem of period estimation. Periodic signals composed of periodic mixtures admit sparse representations in NPDs. Therefore, their underlying hidden periods can be estimated by recovering the exact support of said representations. In this dissertation, we investigate support recovery guarantees of such signals in noise-free and noisy settings. While sufficient recovery conditions of sparse signals have been studied in the literature on compressive sensing, these conditions are of limited use for NPDs, because their analysis does not capture their intrinsic structures. Therefore, we establish new conditions based on a newly introduced notion of nested periodic coherence. Our results show significant improvement over generic recovery bounds as the conditions hold over a larger range of sparsity levels.
Notes
If this is your thesis or dissertation, and want to learn how to access it or for more information about readership statistics, contact us at STARS@ucf.edu
Graduation Date
2021
Semester
Fall
Advisor
Atia, George
Degree
Doctor of Philosophy (Ph.D.)
College
College of Engineering and Computer Science
Department
Electrical and Computer Engineering
Degree Program
Electrical Engineering
Identifier
CFE0009314; DP0026918
URL
https://purls.library.ucf.edu/go/DP0026918
Language
English
Release Date
June 2022
Length of Campus-only Access
None
Access Status
Doctoral Dissertation (Open Access)
STARS Citation
Saidi, Pouria, "Periodicity Estimation Using Representations in Nested Periodic Dictionaries with Applications in Neural Engineering" (2021). Electronic Theses and Dissertations, 2020-2023. 1343.
https://stars.library.ucf.edu/etd2020/1343