Novel Evasion Attacks Against Adversarial Training Defense for Smart Grid Federated Learning

<h3>Abstract</h3><p dir="ltr">In the advanced metering infrastructure (AMI) of the smart grid, smart meters (SMs) are deployed to collect fine-grained electricity consumption data, enabling billing, load monitoring, and efficient energy management. However, some consumers...

وصف كامل

محفوظ في:
التفاصيل البيبلوغرافية
المؤلف الرئيسي: Atef H. Bondok (19482352) (author)
مؤلفون آخرون: Mohamed Mahmoud (4544233) (author), Mahmoud M. Badr (19482355) (author), Mostafa M. Fouda (14768509) (author), Mohamed Abdallah (3073191) (author), Maazen Alsabaan (17714529) (author)
منشور في: 2023
الموضوعات:
الوسوم: إضافة وسم
لا توجد وسوم, كن أول من يضع وسما على هذه التسجيلة!
_version_ 1864513507129556992
author Atef H. Bondok (19482352)
author2 Mohamed Mahmoud (4544233)
Mahmoud M. Badr (19482355)
Mostafa M. Fouda (14768509)
Mohamed Abdallah (3073191)
Maazen Alsabaan (17714529)
author2_role author
author
author
author
author
author_facet Atef H. Bondok (19482352)
Mohamed Mahmoud (4544233)
Mahmoud M. Badr (19482355)
Mostafa M. Fouda (14768509)
Mohamed Abdallah (3073191)
Maazen Alsabaan (17714529)
author_role author
dc.creator.none.fl_str_mv Atef H. Bondok (19482352)
Mohamed Mahmoud (4544233)
Mahmoud M. Badr (19482355)
Mostafa M. Fouda (14768509)
Mohamed Abdallah (3073191)
Maazen Alsabaan (17714529)
dc.date.none.fl_str_mv 2023-10-11T12:00:00Z
dc.identifier.none.fl_str_mv 10.1109/access.2023.3323617
dc.relation.none.fl_str_mv https://figshare.com/articles/journal_contribution/Novel_Evasion_Attacks_Against_Adversarial_Training_Defense_for_Smart_Grid_Federated_Learning/26830168
dc.rights.none.fl_str_mv CC BY 4.0
info:eu-repo/semantics/openAccess
dc.subject.none.fl_str_mv Engineering
Electrical engineering
Information and computing sciences
Cybersecurity and privacy
Data management and data science
Machine learning
Security
evasion attacks
federated learning
and smart power grid
Training
Data models
Smart grids
Detectors
Machine learning
Smart meters
Servers
Federated learning
Power grids
dc.title.none.fl_str_mv Novel Evasion Attacks Against Adversarial Training Defense for Smart Grid Federated Learning
dc.type.none.fl_str_mv Text
Journal contribution
info:eu-repo/semantics/publishedVersion
text
contribution to journal
description <h3>Abstract</h3><p dir="ltr">In the advanced metering infrastructure (AMI) of the smart grid, smart meters (SMs) are deployed to collect fine-grained electricity consumption data, enabling billing, load monitoring, and efficient energy management. However, some consumers engage in fraudulent behavior by hacking their meters, leading to either traditional electricity theft or more sophisticated evasion attacks (EAs). EAs aim to illegally reduce electricity bills while deceiving theft detection mechanisms. The current methods for identifying such attacks raise privacy concerns due to the need for access to consumers’ detailed consumption data to train detection mechanisms. To address privacy concerns, federated learning (FL) is proposed as a collaborative training approach across multiple consumers. Adversarial training (AT) has shown promise in countering evasion threats on machine learning models. This paper, first, investigates the susceptibility of traditional electricity theft classifiers trained by FL to EAs for both independent and identically distributed (IID) and Non-IID consumption data. Then, it investigates the effectiveness of AT in securing the global electricity theft detector against EAs, assuming no misbehavior from the participant consumers in the FL process. After that, we introduce three novel attacks, namely Distillation, No-Adversarial-Sample-Training, and False-Labeling, which can be launched during the AT process to make the global model susceptible to evasion at inference time. Finally, extensive experiments are conducted to validate the severity of these proposed attacks. Our findings reveal that the AT can counter EAs effectively when the FL participants are honest, but it fails when they act maliciously and launch our attacks. This work lays the foundation for future endeavors in exploring additional countermeasures, in conjunction with AT, to bolster the security and resilience of FL machine learning models against adversarial attacks in the context of electricity theft detection.</p><h2>Other Information</h2><p dir="ltr">Published in: IEEE Access<br>License: <a href="https://creativecommons.org/licenses/by/4.0" target="_blank">https://creativecommons.org/licenses/by/4.0</a><br>See article on publisher's website: <a href="https://dx.doi.org/10.1109/access.2023.3323617" target="_blank">https://dx.doi.org/10.1109/access.2023.3323617</a></p>
eu_rights_str_mv openAccess
id Manara2_5559b4c7b000f604b8f7e25e5cca7568
identifier_str_mv 10.1109/access.2023.3323617
network_acronym_str Manara2
network_name_str Manara2
oai_identifier_str oai:figshare.com:article/26830168
publishDate 2023
repository.mail.fl_str_mv
repository.name.fl_str_mv
repository_id_str
rights_invalid_str_mv CC BY 4.0
spelling Novel Evasion Attacks Against Adversarial Training Defense for Smart Grid Federated LearningAtef H. Bondok (19482352)Mohamed Mahmoud (4544233)Mahmoud M. Badr (19482355)Mostafa M. Fouda (14768509)Mohamed Abdallah (3073191)Maazen Alsabaan (17714529)EngineeringElectrical engineeringInformation and computing sciencesCybersecurity and privacyData management and data scienceMachine learningSecurityevasion attacksfederated learningand smart power gridTrainingData modelsSmart gridsDetectorsMachine learningSmart metersServersFederated learningPower grids<h3>Abstract</h3><p dir="ltr">In the advanced metering infrastructure (AMI) of the smart grid, smart meters (SMs) are deployed to collect fine-grained electricity consumption data, enabling billing, load monitoring, and efficient energy management. However, some consumers engage in fraudulent behavior by hacking their meters, leading to either traditional electricity theft or more sophisticated evasion attacks (EAs). EAs aim to illegally reduce electricity bills while deceiving theft detection mechanisms. The current methods for identifying such attacks raise privacy concerns due to the need for access to consumers’ detailed consumption data to train detection mechanisms. To address privacy concerns, federated learning (FL) is proposed as a collaborative training approach across multiple consumers. Adversarial training (AT) has shown promise in countering evasion threats on machine learning models. This paper, first, investigates the susceptibility of traditional electricity theft classifiers trained by FL to EAs for both independent and identically distributed (IID) and Non-IID consumption data. Then, it investigates the effectiveness of AT in securing the global electricity theft detector against EAs, assuming no misbehavior from the participant consumers in the FL process. After that, we introduce three novel attacks, namely Distillation, No-Adversarial-Sample-Training, and False-Labeling, which can be launched during the AT process to make the global model susceptible to evasion at inference time. Finally, extensive experiments are conducted to validate the severity of these proposed attacks. Our findings reveal that the AT can counter EAs effectively when the FL participants are honest, but it fails when they act maliciously and launch our attacks. This work lays the foundation for future endeavors in exploring additional countermeasures, in conjunction with AT, to bolster the security and resilience of FL machine learning models against adversarial attacks in the context of electricity theft detection.</p><h2>Other Information</h2><p dir="ltr">Published in: IEEE Access<br>License: <a href="https://creativecommons.org/licenses/by/4.0" target="_blank">https://creativecommons.org/licenses/by/4.0</a><br>See article on publisher's website: <a href="https://dx.doi.org/10.1109/access.2023.3323617" target="_blank">https://dx.doi.org/10.1109/access.2023.3323617</a></p>2023-10-11T12:00:00ZTextJournal contributioninfo:eu-repo/semantics/publishedVersiontextcontribution to journal10.1109/access.2023.3323617https://figshare.com/articles/journal_contribution/Novel_Evasion_Attacks_Against_Adversarial_Training_Defense_for_Smart_Grid_Federated_Learning/26830168CC BY 4.0info:eu-repo/semantics/openAccessoai:figshare.com:article/268301682023-10-11T12:00:00Z
spellingShingle Novel Evasion Attacks Against Adversarial Training Defense for Smart Grid Federated Learning
Atef H. Bondok (19482352)
Engineering
Electrical engineering
Information and computing sciences
Cybersecurity and privacy
Data management and data science
Machine learning
Security
evasion attacks
federated learning
and smart power grid
Training
Data models
Smart grids
Detectors
Machine learning
Smart meters
Servers
Federated learning
Power grids
status_str publishedVersion
title Novel Evasion Attacks Against Adversarial Training Defense for Smart Grid Federated Learning
title_full Novel Evasion Attacks Against Adversarial Training Defense for Smart Grid Federated Learning
title_fullStr Novel Evasion Attacks Against Adversarial Training Defense for Smart Grid Federated Learning
title_full_unstemmed Novel Evasion Attacks Against Adversarial Training Defense for Smart Grid Federated Learning
title_short Novel Evasion Attacks Against Adversarial Training Defense for Smart Grid Federated Learning
title_sort Novel Evasion Attacks Against Adversarial Training Defense for Smart Grid Federated Learning
topic Engineering
Electrical engineering
Information and computing sciences
Cybersecurity and privacy
Data management and data science
Machine learning
Security
evasion attacks
federated learning
and smart power grid
Training
Data models
Smart grids
Detectors
Machine learning
Smart meters
Servers
Federated learning
Power grids