{"id":{"repo_id":"msu","oai_identifier":"oai:d.lib.msu.edu:etd_52774"},"canonical_url":"https://search.ndltd.org/etd/msu/oai:d.lib.msu.edu:etd_52774","repository":{"repo_id":"msu","name":"Michigan State University","base_url":"https://d.lib.msu.edu/oai"},"display":{"title":"On the Application of Matrix Concentration Inequalities to the Sensor Selection Problem for State Estimation","abstract":"Thesis (Ph.D.)--Michigan State University. Electrical and Computer Engineering - Doctor of Philosophy, 2025","abstract_html":"Thesis (Ph.D.)--Michigan State University. Electrical and Computer Engineering - Doctor of Philosophy, 2025","abstract_has_math":false,"creators":["Calle, Christopher"],"institution":null,"degree_name":null,"degree_level":null,"degree_discipline":null,"degree_department":null,"school":null,"contributors":["Bopardikar, Shaunak D","Tan, Xiaobo","Srivastava, Vaibhav","Li, Zhaojian"],"advisors":[],"committee_chairs":[],"committee_members":[],"year":2025,"date_issued":"2025","date_published":"2025","updated_at":"2026-11-18T18:59:28Z","subjects":["Operations research"],"languages":["eng"],"rights":["Attribution 4.0 International"],"rights_urls":[],"identifier_entries":[{"key":"dc:identifier","label":"Identifier","values":["etd:52774","local:Calle_grad.msu_0128D_21768"],"render_values":[{"text":"etd:52774","href":null,"code":true},{"text":"local:Calle_grad.msu_0128D_21768","href":null,"code":true}]}],"person_identifier_entries":[]},"links":{"outbound_url":"https://doi.org/doi:10.25335/k06f-dg11","outbound_label":"DOI","outbound_source":"dc:identifier","record_url":"https://doi.org/doi:10.25335/k06f-dg11","record_label":"Repository record","record_kind":"doi","record_source":"dc:identifier","download_url":null,"download_label":null,"download_source":null,"access_status":null,"embargo_until":null,"access_note":null},"source_record":{"url":"https://d.lib.msu.edu/oai?verb=GetRecord&metadataPrefix=oai_dc&identifier=oai%3Ad.lib.msu.edu%3Aetd_52774","prefix":"oai_dc"},"metadata_groups":[{"id":"people","label":"People","entries":[{"key":"dc:contributor","label":"Contributor","values":["Bopardikar, Shaunak D","Tan, Xiaobo","Srivastava, Vaibhav","Li, Zhaojian"]},{"key":"dc:creator","label":"Author","values":["Calle, Christopher"]}]},{"id":"academic_context","label":"Academic Context","entries":[{"key":"dc:date","label":"Dc Date","values":["2025"]},{"key":"dc:relation","label":"Dc Relation","values":["Electronic Theses & Dissertations"]},{"key":"dc:type","label":"Dc Type","values":["text","Electronic dissertations"]}]},{"id":"subjects_keywords","label":"Subjects and Keywords","entries":[{"key":"dc:subject","label":"Dc Subject","values":["Operations research"]}]},{"id":"language_rights","label":"Language and Rights","entries":[{"key":"dc:language","label":"Dc Language","values":["eng"]},{"key":"dc:rights","label":"Dc Rights","values":["Attribution 4.0 International"]}]},{"id":"identifiers","label":"Identifiers","entries":[{"key":"dc:identifier","label":"Identifier","values":["etd:52774","local:Calle_grad.msu_0128D_21768","https://doi.org/doi:10.25335/k06f-dg11"]}]},{"id":"additional","label":"Additional Metadata","entries":[{"key":"dc:description","label":"Description","values":["Thesis (Ph.D.)--Michigan State University. Electrical and Computer Engineering - Doctor of Philosophy, 2025","Sensors are fundamental in observing the latent state of a dynamical system of interest. Since sensors do not directly observe the latent state in practice, sensors become a necessity to partially or fully estimating the latent state in question. Further complexity arises when the measurements outputted by the sensors are noisy. Our uncertainty of the state estimate in this dissertation is quantified using the ubiquitous Kalman filter and its model-based assumptions. In an ideal scenario, all the sensors that can be deployed are deployed. However, this scenario is unrealistic in practice due to constraints on resources. Given that the number of sensors one can choose for deployment is a fixed quantity, a problem of great interest is to choose a selection of sensors that optimally reduces our uncertainty on the state estimate. In this dissertation, we address the aforementioned problem, succinctly stated as the problem of sensor selection for state estimation, by leveraging a class of statistical tools known as matrix concentration inequalities (CIs). Specifically, we investigate the a priori guarantees that exist for the chosen selection of sensors. A priori guarantees are of great importance since they inform us about the quality (of the state estimate) we should expect prior to sampling any selection of sensors for deployment. Many contemporary works formulate such guarantees for our problem of interest by either directly or indirectly requiring the submodularity property, a property of set functions, to hold. Unlike submodular-based approaches, matrix CIs are capable of providing a priori guarantees regardless of whether submodularity is present or not. We assume the sensors we choose for deployment are chosen via a sampling with replacement policy as a consequence of utilizing matrix CIs in this dissertation.In the first part of this dissertation, we investigate the myriad of settings that matrix CIs are applicable in the sensor selection problem for state estimation. Specifically, we investigate the a priori guarantees that are possible for three settings of practical interest. Since the a priori guarantees of interest inform us on the quality of our state estimate, the guarantees in this chapter are with respect to the state estimation error covariance of the Kalman filter. Our analysis first assumes that the sampling policy for choosing an appropriate sensor selection is fixed or specified. We extend our analysis by studying the problem of finding an optimum sampling policy when the policy is not specified. We demonstrate the guarantees that are possible for the three different scenarios of interest.In the second part of this dissertation, we significantly extend the utility of matrix CIs to improve the a priori guarantees associated with our problem of interest. First, we derive a novel statistical tool, i.e., a generalization to an existing matrix CI. Second, we perform a comprehensive analysis on the feasibility, sample complexity, and guarantees of our novel result to improve the a priori guarantees of matrix CIs in general. Third, we investigate the general optimization of sums of i.i.d. and p.s.d. random matrices. Also, we highlight the applicability of our work to a variety of model-based assumptions and sampling policies. Lastly, we demonstrate our improved guarantees as they relate to the sensor selection problem for state estimation.","Description based on online resource. Title from PDF t.p. (Michigan State University Fedora Repository, viewed ).","Includes bibliographical references."]},{"key":"dc:format","label":"Dc Format","values":["application/pdf","1 online resource (106 p.) : ill."]},{"key":"dc:title","label":"Title","values":["On the Application of Matrix Concentration Inequalities to the Sensor Selection Problem for State Estimation"]}]}],"canonical_facts":{"dc:contributor":["Bopardikar, Shaunak D","Tan, Xiaobo","Srivastava, Vaibhav","Li, Zhaojian"],"dc:creator":["Calle, Christopher"],"dc:date":["2025"],"dc:description":["Thesis (Ph.D.)--Michigan State University. Electrical and Computer Engineering - Doctor of Philosophy, 2025","Sensors are fundamental in observing the latent state of a dynamical system of interest. Since sensors do not directly observe the latent state in practice, sensors become a necessity to partially or fully estimating the latent state in question. Further complexity arises when the measurements outputted by the sensors are noisy. Our uncertainty of the state estimate in this dissertation is quantified using the ubiquitous Kalman filter and its model-based assumptions. In an ideal scenario, all the sensors that can be deployed are deployed. However, this scenario is unrealistic in practice due to constraints on resources. Given that the number of sensors one can choose for deployment is a fixed quantity, a problem of great interest is to choose a selection of sensors that optimally reduces our uncertainty on the state estimate. In this dissertation, we address the aforementioned problem, succinctly stated as the problem of sensor selection for state estimation, by leveraging a class of statistical tools known as matrix concentration inequalities (CIs). Specifically, we investigate the a priori guarantees that exist for the chosen selection of sensors. A priori guarantees are of great importance since they inform us about the quality (of the state estimate) we should expect prior to sampling any selection of sensors for deployment. Many contemporary works formulate such guarantees for our problem of interest by either directly or indirectly requiring the submodularity property, a property of set functions, to hold. Unlike submodular-based approaches, matrix CIs are capable of providing a priori guarantees regardless of whether submodularity is present or not. We assume the sensors we choose for deployment are chosen via a sampling with replacement policy as a consequence of utilizing matrix CIs in this dissertation.In the first part of this dissertation, we investigate the myriad of settings that matrix CIs are applicable in the sensor selection problem for state estimation. Specifically, we investigate the a priori guarantees that are possible for three settings of practical interest. Since the a priori guarantees of interest inform us on the quality of our state estimate, the guarantees in this chapter are with respect to the state estimation error covariance of the Kalman filter. Our analysis first assumes that the sampling policy for choosing an appropriate sensor selection is fixed or specified. We extend our analysis by studying the problem of finding an optimum sampling policy when the policy is not specified. We demonstrate the guarantees that are possible for the three different scenarios of interest.In the second part of this dissertation, we significantly extend the utility of matrix CIs to improve the a priori guarantees associated with our problem of interest. First, we derive a novel statistical tool, i.e., a generalization to an existing matrix CI. Second, we perform a comprehensive analysis on the feasibility, sample complexity, and guarantees of our novel result to improve the a priori guarantees of matrix CIs in general. Third, we investigate the general optimization of sums of i.i.d. and p.s.d. random matrices. Also, we highlight the applicability of our work to a variety of model-based assumptions and sampling policies. Lastly, we demonstrate our improved guarantees as they relate to the sensor selection problem for state estimation.","Description based on online resource. Title from PDF t.p. (Michigan State University Fedora Repository, viewed ).","Includes bibliographical references."],"dc:format":["application/pdf","1 online resource (106 p.) : ill."],"dc:identifier":["etd:52774","local:Calle_grad.msu_0128D_21768","https://doi.org/doi:10.25335/k06f-dg11"],"dc:language":["eng"],"dc:relation":["Electronic Theses & Dissertations"],"dc:rights":["Attribution 4.0 International"],"dc:subject":["Operations research"],"dc:title":["On the Application of Matrix Concentration Inequalities to the Sensor Selection Problem for State Estimation"],"dc:type":["text","Electronic dissertations"]},"updated_at":"2026-11-18T18:59:28Z"}