Breaking the Performance Gap of Fully and Semi-Supervised Learning in Electromagnetic Signature Recognition

dc.authorscopusid 57218794711
dc.authorscopusid 57064024200
dc.authorscopusid 57218796050
dc.authorscopusid 58487868600
dc.authorscopusid 57216218600
dc.authorscopusid 55336250100
dc.authorscopusid 7005179513
dc.contributor.author Panayırcı, Erdal
dc.contributor.author Wang, Q.
dc.contributor.author Chen, L.
dc.contributor.author Fu, G.
dc.contributor.author Liu, X.
dc.contributor.author Dong, Z.
dc.contributor.author Panayırcı, Erdal
dc.contributor.other Electrical-Electronics Engineering
dc.date.accessioned 2023-10-19T15:05:24Z
dc.date.available 2023-10-19T15:05:24Z
dc.date.issued 2023
dc.department-temp Wang, H., School of Electrical and Information Engineering, Tianjin University, Tianjin, China; Wang, Q., School of Electrical and Information Engineering, Tianjin University, Tianjin, China; Chen, L., School of Electrical and Information Engineering, Tianjin University, Tianjin, China; Fu, G., School of Electrical and Information Engineering, Tianjin University, Tianjin, China; Liu, X., School of Electrical and Information Engineering, Tianjin University, Tianjin, China; Dong, Z., School of Information Science and Technology, Tibet University, Lasha, China; Panayirci, E., Department of Electrical and Electronics Engineering, Kadir Has University, Istanbul, Turkey en_US
dc.description.abstract Intelligent electromagnetic signature recognition is one of the key technologies in Internet-of-Things (IoT) device connection, which can improve system security and speed up the authentication process. In practical scenarios, as the number of IoT devices increases, electromagnetic features such as fingerprint and modulation signals also increase substantially. However, since intelligent recognition technology, such as Automatic Modulation Classification (AMC), requires a large amount of labeled data to train the neural network classifier, it is challenging to collect so much labeled data. To address the performance degradation challenges with small training data, we propose an efficient semi-supervised electromagnetic recognition framework to break the performance gap with the fully supervised learning scheme. This framework can fully use the unlabeled electromagnetic data collected during the authentication process for self-training to improve the classifier’s performance. According to the idea of consistency regularization, we design a signal augmentation method and propose an ensemble pseudo-label design algorithm to improve confidence. Moreover, we perform a convex combination of electromagnetic features to smooth the model decision boundary while generalizing to unknown data distribution regions. Experimental results on the modulated data demonstrate the performance superiority of the proposed algorithm, i.e., use less than 5% of data with no more than 10% performance drop. IEEE en_US
dc.identifier.citationcount 0
dc.identifier.doi 10.1109/JIOT.2023.3295397 en_US
dc.identifier.endpage 1 en_US
dc.identifier.issn 2327-4662
dc.identifier.scopus 2-s2.0-85164779053 en_US
dc.identifier.scopusquality Q1
dc.identifier.startpage 1 en_US
dc.identifier.uri https://doi.org/10.1109/JIOT.2023.3295397
dc.identifier.uri https://hdl.handle.net/20.500.12469/4870
dc.identifier.wosquality Q1
dc.khas 20231019-Scopus en_US
dc.language.iso en en_US
dc.publisher Institute of Electrical and Electronics Engineers Inc. en_US
dc.relation.ispartof IEEE Internet of Things Journal en_US
dc.relation.publicationcategory Makale - Uluslararası Hakemli Dergi - Kurum Öğretim Elemanı en_US
dc.rights info:eu-repo/semantics/closedAccess en_US
dc.scopus.citedbyCount 1
dc.subject convex combination en_US
dc.subject Data models en_US
dc.subject electromagnetic signature recognition en_US
dc.subject Electromagnetics en_US
dc.subject ensemble pseudo-label en_US
dc.subject Feature extraction en_US
dc.subject Internet of Things en_US
dc.subject Modulation en_US
dc.subject Semi-supervised learning en_US
dc.subject Semisupervised learning en_US
dc.subject signal augmentation methods en_US
dc.subject Training en_US
dc.subject Classification (of information) en_US
dc.subject Internet of things en_US
dc.subject Supervised learning en_US
dc.subject Augmentation methods en_US
dc.subject Convex combinations en_US
dc.subject Electromagnetic signature recognition en_US
dc.subject Electromagnetic signatures en_US
dc.subject Electromagnetics en_US
dc.subject Ensemble pseudo-label en_US
dc.subject Features extraction en_US
dc.subject Semi-supervised learning en_US
dc.subject Signal augmentation method en_US
dc.subject Signature recognition en_US
dc.subject Authentication en_US
dc.title Breaking the Performance Gap of Fully and Semi-Supervised Learning in Electromagnetic Signature Recognition en_US
dc.type Article en_US
dspace.entity.type Publication
relation.isAuthorOfPublication 5371ab5d-9cd9-4d1f-8681-a65b3d5d6add
relation.isAuthorOfPublication.latestForDiscovery 5371ab5d-9cd9-4d1f-8681-a65b3d5d6add
relation.isOrgUnitOfPublication 12b0068e-33e6-48db-b92a-a213070c3a8d
relation.isOrgUnitOfPublication.latestForDiscovery 12b0068e-33e6-48db-b92a-a213070c3a8d

Files