This dissertation proposes a systematic approach, based on a probabilistic graphical model, to infer missing observations in wireless sensor networks (WSNs) for sustaining environmental monitoring. This enables us to effectively address two critical challenges in WSNs: (1) energy-efficient data gathering through planned communication disruptions resulting from energy-saving sleep cycles, and (2) sensor-node failure tolerance in harsh environments. In our approach, we develop a pairwise Markov Random Field (MRF) to model the spatial correlations in a sensor network. Our MRF model is first constructed through automatic learning from historical sensed data, by using Iterative Proportional Fitting (IPF). When the MRF model is constructed, Loopy Belief Propagation (LBP) is then employed to perform information inference to estimate the missing data given incomplete network observations. The proposed approach is then improved in terms of energy-efficiency and robustness from three aspects: model building, inference and parameter learning. The model and methods are empirically evaluated using multiple real-world sensor network data sets. The results demonstrate the merits of our proposed approaches.
","abstract_html":"<p>This dissertation proposes a systematic approach, based on a probabilistic graphical model, to infer missing observations in wireless sensor networks (WSNs) for sustaining environmental monitoring. This enables us to effectively address two critical challenges in WSNs: (1) energy-efficient data gathering through planned communication disruptions resulting from energy-saving sleep cycles, and (2) sensor-node failure tolerance in harsh environments. In our approach, we develop a pairwise Markov Random Field (MRF) to model the spatial correlations in a sensor network. Our MRF model is first constructed through automatic learning from historical sensed data, by using Iterative Proportional Fitting (IPF). When the MRF model is constructed, Loopy Belief Propagation (LBP) is then employed to perform information inference to estimate the missing data given incomplete network observations. The proposed approach is then improved in terms of energy-efficiency and robustness from three aspects: model building, inference and parameter learning. The model and methods are empirically evaluated using multiple real-world sensor network data sets. The results demonstrate the merits of our proposed approaches.</p>","abstract_has_math":false,"creators":["Zhao, Wei"],"institution":null,"degree_name":"Doctor of Philosophy (PhD)","degree_level":"Dissertation","degree_discipline":"Computer Science","degree_department":null,"school":null,"contributors":["Yao Liang","Arjan Durresi","Xukai Zou","Luo Si","David K. Y. Yau"],"advisors":[],"committee_chairs":[],"committee_members":[],"year":2014,"date_issued":"2014-10-01T07:00:00Z","date_published":"2014-10-01T07:00:00Z","updated_at":"2026-07-24T03:53:41Z","subjects":["Computer Sciences"],"languages":[],"rights":[],"rights_urls":[],"identifier_entries":[]},"links":{"outbound_url":"https://docs.lib.purdue.edu/open_access_dissertations/402","outbound_label":"Repository record","outbound_source":"dc:identifier"},"metadata_groups":[{"id":"people","label":"People","entries":[{"key":"dc:contributor","label":"Contributor","values":["Yao Liang","Arjan Durresi","Xukai Zou","Luo Si","David K. Y. Yau"]},{"key":"dc:creator","label":"Author","values":["Zhao, Wei"]}]},{"id":"academic_context","label":"Academic Context","entries":[{"key":"thesis:degree_discipline","label":"Discipline","values":["Computer Science"]},{"key":"thesis:degree_level","label":"Degree Level","values":["Dissertation"]},{"key":"thesis:degree_name","label":"Degree Name","values":["Doctor of Philosophy (PhD)"]}]},{"id":"subjects_keywords","label":"Subjects and Keywords","entries":[{"key":"dc:subject","label":"Dc Subject","values":["Computer Sciences"]}]},{"id":"identifiers","label":"Identifiers","entries":[{"key":"dc:identifier","label":"Identifier","values":["https://docs.lib.purdue.edu/open_access_dissertations/402"]}]},{"id":"additional","label":"Additional Metadata","entries":[{"key":"dc:description.abstract","label":"Abstract","values":["This dissertation proposes a systematic approach, based on a probabilistic graphical model, to infer missing observations in wireless sensor networks (WSNs) for sustaining environmental monitoring. This enables us to effectively address two critical challenges in WSNs: (1) energy-efficient data gathering through planned communication disruptions resulting from energy-saving sleep cycles, and (2) sensor-node failure tolerance in harsh environments. In our approach, we develop a pairwise Markov Random Field (MRF) to model the spatial correlations in a sensor network. Our MRF model is first constructed through automatic learning from historical sensed data, by using Iterative Proportional Fitting (IPF). When the MRF model is constructed, Loopy Belief Propagation (LBP) is then employed to perform information inference to estimate the missing data given incomplete network observations. The proposed approach is then improved in terms of energy-efficiency and robustness from three aspects: model building, inference and parameter learning. The model and methods are empirically evaluated using multiple real-world sensor network data sets. The results demonstrate the merits of our proposed approaches.
"]},{"key":"dc:title","label":"Title","values":["Energy-efficient information inference in wireless sensor networks based on graphical modeling"]}]}],"canonical_facts":{"dc:contributor":["Yao Liang","Arjan Durresi","Xukai Zou","Luo Si","David K. Y. Yau"],"dc:creator":["Zhao, Wei"],"dc:description.abstract":["This dissertation proposes a systematic approach, based on a probabilistic graphical model, to infer missing observations in wireless sensor networks (WSNs) for sustaining environmental monitoring. This enables us to effectively address two critical challenges in WSNs: (1) energy-efficient data gathering through planned communication disruptions resulting from energy-saving sleep cycles, and (2) sensor-node failure tolerance in harsh environments. In our approach, we develop a pairwise Markov Random Field (MRF) to model the spatial correlations in a sensor network. Our MRF model is first constructed through automatic learning from historical sensed data, by using Iterative Proportional Fitting (IPF). When the MRF model is constructed, Loopy Belief Propagation (LBP) is then employed to perform information inference to estimate the missing data given incomplete network observations. The proposed approach is then improved in terms of energy-efficiency and robustness from three aspects: model building, inference and parameter learning. The model and methods are empirically evaluated using multiple real-world sensor network data sets. The results demonstrate the merits of our proposed approaches.
"],"dc:identifier":["https://docs.lib.purdue.edu/open_access_dissertations/402"],"dc:subject":["Computer Sciences"],"dc:title":["Energy-efficient information inference in wireless sensor networks based on graphical modeling"],"thesis:degree_discipline":["Computer Science"],"thesis:degree_level":["Dissertation"],"thesis:degree_name":["Doctor of Philosophy (PhD)"]},"updated_at":"2026-07-24T03:53:41Z"}