{"id":{"repo_id":"colostate","oai_identifier":"oai:mountainscholar.org:10217/242675"},"canonical_url":"https://search.dev.ndltd.org/etd/colostate/oai:mountainscholar.org:10217/242675","repository":{"repo_id":"colostate","name":"Colorado State University","base_url":"https://api.mountainscholar.org/server/oai/request"},"display":{"title":"Deep learning for IoT fingerprinting","abstract":"The rapid growth of the Internet-of-Things (IoT) industry has introduced new attack surfaces in home and enterprise networks due to insufficient built-in security measures in many IoT devices. IoT network fingerprinting, the process of identifying IoT devices from their network communication patterns, can be used to select appropriate access controls for vulnerable IoT devices, offering a promising solution to this security problem. Typical state-of-the-art IoT fingerprinting approaches identify devices by applying deep learning models to samples of their network traffic. This work addresses 5 challenges of using deep learning for IoT fingerprinting: (1) traffic collected by a single observer may be insufficient for training an accurate fingerprinting model; (2) variations in communication rates among different IoT devices results in imbalanced datasets, which can lead to biased models; (3) it is difficult for a model to capture the relationships between packets when some packets belong to separate flows; (4) it is inefficient to adapt an existing IoT fingerprinting model to identify new devices; and (5) relying on fixed-length samples of traffic can result in arbitrarily long fingerprinting times. For the first challenge, we propose a federated learning approach for training a fingerprinting model using traffic collected by multiple separate observers. For the second challenge, we propose a hierarchical Mixture-of-Experts that balances training data by grouping devices based on their communication rates before identifying them. For the third challenge, we propose a relative encoding technique for packet endpoints that preserves relationships across flows. For the fourth challenge, we propose a bi-component fingerprinting architecture that efficiently adapts to new devices by reusing a portion of its parameters. Finally, for the fifth challenge, we propose a fixed-time traffic sampling approach. We evaluate our proposed approaches through a series of experiments and demonstrate how each one overcomes its associated challenge in an experimental setting.","abstract_html":"The rapid growth of the Internet-of-Things (IoT) industry has introduced new attack surfaces in home and enterprise networks due to insufficient built-in security measures in many IoT devices. IoT network fingerprinting, the process of identifying IoT devices from their network communication patterns, can be used to select appropriate access controls for vulnerable IoT devices, offering a promising solution to this security problem. Typical state-of-the-art IoT fingerprinting approaches identify devices by applying deep learning models to samples of their network traffic. This work addresses 5 challenges of using deep learning for IoT fingerprinting: (1) traffic collected by a single observer may be insufficient for training an accurate fingerprinting model; (2) variations in communication rates among different IoT devices results in imbalanced datasets, which can lead to biased models; (3) it is difficult for a model to capture the relationships between packets when some packets belong to separate flows; (4) it is inefficient to adapt an existing IoT fingerprinting model to identify new devices; and (5) relying on fixed-length samples of traffic can result in arbitrarily long fingerprinting times. For the first challenge, we propose a federated learning approach for training a fingerprinting model using traffic collected by multiple separate observers. For the second challenge, we propose a hierarchical Mixture-of-Experts that balances training data by grouping devices based on their communication rates before identifying them. For the third challenge, we propose a relative encoding technique for packet endpoints that preserves relationships across flows. For the fourth challenge, we propose a bi-component fingerprinting architecture that efficiently adapts to new devices by reusing a portion of its parameters. Finally, for the fifth challenge, we propose a fixed-time traffic sampling approach. We evaluate our proposed approaches through a series of experiments and demonstrate how each one overcomes its associated challenge in an experimental setting.","abstract_has_math":false,"creators":["Bar-on, Maxwel A., author","Ray, Indrakshi, advisor","Bezawada, Bruhadeshwar, committee member","Ray, Indrajit, committee member","Jayasumana, Anura, committee member"],"institution":"Colorado State University. Libraries","degree_name":"Master of Science (M.S.)","degree_level":"Masters","degree_discipline":"Computer Science","degree_department":null,"school":null,"contributors":[],"advisors":[],"committee_chairs":[],"committee_members":[],"year":2025,"date_issued":"2025","date_published":"2025","updated_at":"2026-08-21T22:21:56Z","subjects":["federated learning","Internet of Things","transfer learning","fingerprinting","deep learning","network security"],"languages":["eng","English"],"rights":["Copyright and other restrictions may apply. User is responsible for compliance with all applicable laws. For information about copyright law, please see https://libguides.colostate.edu/copyright."],"rights_urls":[],"identifier_entries":[{"key":"dc:identifier.uri","label":"Identifier URI","values":["https://doi.org/10.25675/3.025567"],"render_values":[{"text":"https://doi.org/10.25675/3.025567","href":"https://doi.org/10.25675/3.025567","code":true}]}]},"links":{"outbound_url":"https://hdl.handle.net/10217/242675","outbound_label":"Handle","outbound_source":"dc:identifier.uri"},"source_record":{"url":"https://api.mountainscholar.org/server/oai/request?verb=GetRecord&metadataPrefix=dim&identifier=oai%3Amountainscholar.org%3A10217%2F242675","prefix":"dim"},"metadata_groups":[{"id":"people","label":"People","entries":[{"key":"dc:creator","label":"Author","values":["Bar-on, Maxwel A., author","Ray, Indrakshi, advisor","Bezawada, Bruhadeshwar, committee member","Ray, Indrajit, committee member","Jayasumana, Anura, committee member"]}]},{"id":"academic_context","label":"Academic Context","entries":[{"key":"dc:date.accessioned","label":"Dc Date Accessioned","values":["2026-01-12T11:27:40Z"]},{"key":"dc:date.issued","label":"Date","values":["2025"]},{"key":"dc:publisher","label":"Institution","values":["Colorado State University. Libraries"]},{"key":"dc:type","label":"Dc Type","values":["Text","Image"]},{"key":"thesis:degree_discipline","label":"Discipline","values":["Computer Science"]},{"key":"thesis:degree_level","label":"Degree Level","values":["Masters"]},{"key":"thesis:degree_name","label":"Degree Name","values":["Master of Science (M.S.)"]},{"key":"thesis:institution_name","label":"Thesis Institution Name","values":["Colorado State University"]}]},{"id":"subjects_keywords","label":"Subjects and Keywords","entries":[{"key":"dc:subject","label":"Dc Subject","values":["federated learning","Internet of Things","transfer learning","fingerprinting","deep learning","network security"]}]},{"id":"language_rights","label":"Language and Rights","entries":[{"key":"dc:language","label":"Dc Language","values":["English"]},{"key":"dc:language.iso","label":"Language (ISO)","values":["eng"]},{"key":"dc:rights","label":"Dc Rights","values":["Copyright and other restrictions may apply. User is responsible for compliance with all applicable laws. For information about copyright law, please see https://libguides.colostate.edu/copyright."]}]},{"id":"identifiers","label":"Identifiers","entries":[{"key":"dc:identifier","label":"Identifier","values":["Baron_colostate_0053N_19301.pdf"]},{"key":"dc:identifier.uri","label":"Identifier URI","values":["https://hdl.handle.net/10217/242675","https://doi.org/10.25675/3.025567"]}]},{"id":"additional","label":"Additional Metadata","entries":[{"key":"dc:description","label":"Description","values":["Zip file contains thesis presentation."]},{"key":"dc:description.abstract","label":"Abstract","values":["The rapid growth of the Internet-of-Things (IoT) industry has introduced new attack surfaces in home and enterprise networks due to insufficient built-in security measures in many IoT devices. IoT network fingerprinting, the process of identifying IoT devices from their network communication patterns, can be used to select appropriate access controls for vulnerable IoT devices, offering a promising solution to this security problem. Typical state-of-the-art IoT fingerprinting approaches identify devices by applying deep learning models to samples of their network traffic. This work addresses 5 challenges of using deep learning for IoT fingerprinting: (1) traffic collected by a single observer may be insufficient for training an accurate fingerprinting model; (2) variations in communication rates among different IoT devices results in imbalanced datasets, which can lead to biased models; (3) it is difficult for a model to capture the relationships between packets when some packets belong to separate flows; (4) it is inefficient to adapt an existing IoT fingerprinting model to identify new devices; and (5) relying on fixed-length samples of traffic can result in arbitrarily long fingerprinting times. For the first challenge, we propose a federated learning approach for training a fingerprinting model using traffic collected by multiple separate observers. For the second challenge, we propose a hierarchical Mixture-of-Experts that balances training data by grouping devices based on their communication rates before identifying them. For the third challenge, we propose a relative encoding technique for packet endpoints that preserves relationships across flows. For the fourth challenge, we propose a bi-component fingerprinting architecture that efficiently adapts to new devices by reusing a portion of its parameters. Finally, for the fifth challenge, we propose a fixed-time traffic sampling approach. We evaluate our proposed approaches through a series of experiments and demonstrate how each one overcomes its associated challenge in an experimental setting."]},{"key":"dc:format.medium","label":"Dc Format Medium","values":["born digital","masters theses","ZIP","PDF"]},{"key":"dc:title","label":"Title","values":["Deep learning for IoT fingerprinting"]}]}],"canonical_facts":{"dc:creator":["Bar-on, Maxwel A., author","Ray, Indrakshi, advisor","Bezawada, Bruhadeshwar, committee member","Ray, Indrajit, committee member","Jayasumana, Anura, committee member"],"dc:date.accessioned":["2026-01-12T11:27:40Z"],"dc:date.issued":["2025"],"dc:description":["Zip file contains thesis presentation."],"dc:description.abstract":["The rapid growth of the Internet-of-Things (IoT) industry has introduced new attack surfaces in home and enterprise networks due to insufficient built-in security measures in many IoT devices. IoT network fingerprinting, the process of identifying IoT devices from their network communication patterns, can be used to select appropriate access controls for vulnerable IoT devices, offering a promising solution to this security problem. Typical state-of-the-art IoT fingerprinting approaches identify devices by applying deep learning models to samples of their network traffic. This work addresses 5 challenges of using deep learning for IoT fingerprinting: (1) traffic collected by a single observer may be insufficient for training an accurate fingerprinting model; (2) variations in communication rates among different IoT devices results in imbalanced datasets, which can lead to biased models; (3) it is difficult for a model to capture the relationships between packets when some packets belong to separate flows; (4) it is inefficient to adapt an existing IoT fingerprinting model to identify new devices; and (5) relying on fixed-length samples of traffic can result in arbitrarily long fingerprinting times. For the first challenge, we propose a federated learning approach for training a fingerprinting model using traffic collected by multiple separate observers. For the second challenge, we propose a hierarchical Mixture-of-Experts that balances training data by grouping devices based on their communication rates before identifying them. For the third challenge, we propose a relative encoding technique for packet endpoints that preserves relationships across flows. For the fourth challenge, we propose a bi-component fingerprinting architecture that efficiently adapts to new devices by reusing a portion of its parameters. Finally, for the fifth challenge, we propose a fixed-time traffic sampling approach. We evaluate our proposed approaches through a series of experiments and demonstrate how each one overcomes its associated challenge in an experimental setting."],"dc:format.medium":["born digital","masters theses","ZIP","PDF"],"dc:identifier":["Baron_colostate_0053N_19301.pdf"],"dc:identifier.uri":["https://hdl.handle.net/10217/242675","https://doi.org/10.25675/3.025567"],"dc:language":["English"],"dc:language.iso":["eng"],"dc:publisher":["Colorado State University. Libraries"],"dc:rights":["Copyright and other restrictions may apply. User is responsible for compliance with all applicable laws. For information about copyright law, please see https://libguides.colostate.edu/copyright."],"dc:subject":["federated learning","Internet of Things","transfer learning","fingerprinting","deep learning","network security"],"dc:title":["Deep learning for IoT fingerprinting"],"dc:type":["Text","Image"],"thesis:degree_discipline":["Computer Science"],"thesis:degree_level":["Masters"],"thesis:degree_name":["Master of Science (M.S.)"],"thesis:institution_name":["Colorado State University"]},"updated_at":"2026-08-21T22:21:56Z"}