RL-PDNN: Reinforcement Learning for Privacy-Aware Distributed Neural Networks in IoT Systems

<p>Due to their high computational and memory demand, deep learning applications are mainly restricted to high-performance units, e.g., cloud and edge servers. Particularly, in Internet of Things (IoT) systems, the data acquired by pervasive devices is sent to the computing servers for classif...

وصف كامل

محفوظ في:
التفاصيل البيبلوغرافية
المؤلف الرئيسي: Emna Baccour (16896366) (author)
مؤلفون آخرون: Aiman Erbad (14150589) (author), Amr Mohamed (3508121) (author), Mounir Hamdi (14150652) (author), Mohsen Guizani (12580291) (author)
منشور في: 2021
الموضوعات:
الوسوم: إضافة وسم
لا توجد وسوم, كن أول من يضع وسما على هذه التسجيلة!
_version_ 1864513560383586304
author Emna Baccour (16896366)
author2 Aiman Erbad (14150589)
Amr Mohamed (3508121)
Mounir Hamdi (14150652)
Mohsen Guizani (12580291)
author2_role author
author
author
author
author_facet Emna Baccour (16896366)
Aiman Erbad (14150589)
Amr Mohamed (3508121)
Mounir Hamdi (14150652)
Mohsen Guizani (12580291)
author_role author
dc.creator.none.fl_str_mv Emna Baccour (16896366)
Aiman Erbad (14150589)
Amr Mohamed (3508121)
Mounir Hamdi (14150652)
Mohsen Guizani (12580291)
dc.date.none.fl_str_mv 2021-04-02T00:00:00Z
dc.identifier.none.fl_str_mv 10.1109/access.2021.3070627
dc.relation.none.fl_str_mv https://figshare.com/articles/journal_contribution/RL-PDNN_Reinforcement_Learning_for_Privacy-Aware_Distributed_Neural_Networks_in_IoT_Systems/24049251
dc.rights.none.fl_str_mv CC BY 4.0
info:eu-repo/semantics/openAccess
dc.subject.none.fl_str_mv Information and computing sciences
Artificial intelligence
Data management and data science
Distributed computing and systems software
Machine learning
Task analysis
Servers
Deep learning
Dynamic scheduling
Real-time systems
Performance evaluation
Biological neural networks
IoT devices
Distributed DNN
Privacy
White-box
Resource constraints
DQN
dc.title.none.fl_str_mv RL-PDNN: Reinforcement Learning for Privacy-Aware Distributed Neural Networks in IoT Systems
dc.type.none.fl_str_mv Text
Journal contribution
info:eu-repo/semantics/publishedVersion
text
contribution to journal
description <p>Due to their high computational and memory demand, deep learning applications are mainly restricted to high-performance units, e.g., cloud and edge servers. Particularly, in Internet of Things (IoT) systems, the data acquired by pervasive devices is sent to the computing servers for classification. However, this approach might not be always possible because of the limited bandwidth and the privacy issues. Furthermore, it presents uncertainty in terms of latency because of the unstable remote connectivity. To support resource and delay requirements of such paradigm, joint and real-time deep co-inference framework with IoT synergy was introduced. However, scheduling the distributed, dynamic and real-time Deep Neural Network (DNN) inference requests among resource-constrained devices has not been well explored in the literature. Additionally, the distribution of DNN has drawn the attention to the privacy protection of sensitive data. In this context, various threats have been presented, including white-box attacks, where malicious devices can accurately recover received inputs if the DNN model is fully exposed to participants. In this paper, we introduce a methodology aiming at distributing the DNN tasks onto the resource-constrained devices of the IoT system, while avoiding to reveal the model to participants. We formulate such an approach as an optimization problem, where we establish a trade-off between the latency of co-inference, the privacy of the data, and the limited resources of devices. Next, due to the NP-hardness of the problem, we shape our approach as a reinforcement learning design adequate for real-time applications and highly dynamic systems, namely RL-PDNN. Our system proved its ability to outperform existing static approaches and achieve close results compared to the optimal solution.<br></p><h2>Other Information</h2><p>Published in: IEEE Access<br>License: <a href="https://creativecommons.org/licenses/by/4.0/legalcode" 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.2021.3070627" target="_blank">https://dx.doi.org/10.1109/access.2021.3070627</a></p>
eu_rights_str_mv openAccess
id Manara2_07086691311788e2bc4365f98068178a
identifier_str_mv 10.1109/access.2021.3070627
network_acronym_str Manara2
network_name_str Manara2
oai_identifier_str oai:figshare.com:article/24049251
publishDate 2021
repository.mail.fl_str_mv
repository.name.fl_str_mv
repository_id_str
rights_invalid_str_mv CC BY 4.0
spelling RL-PDNN: Reinforcement Learning for Privacy-Aware Distributed Neural Networks in IoT SystemsEmna Baccour (16896366)Aiman Erbad (14150589)Amr Mohamed (3508121)Mounir Hamdi (14150652)Mohsen Guizani (12580291)Information and computing sciencesArtificial intelligenceData management and data scienceDistributed computing and systems softwareMachine learningTask analysisServersDeep learningDynamic schedulingReal-time systemsPerformance evaluationBiological neural networksIoT devicesDistributed DNNPrivacyWhite-boxResource constraintsDQN<p>Due to their high computational and memory demand, deep learning applications are mainly restricted to high-performance units, e.g., cloud and edge servers. Particularly, in Internet of Things (IoT) systems, the data acquired by pervasive devices is sent to the computing servers for classification. However, this approach might not be always possible because of the limited bandwidth and the privacy issues. Furthermore, it presents uncertainty in terms of latency because of the unstable remote connectivity. To support resource and delay requirements of such paradigm, joint and real-time deep co-inference framework with IoT synergy was introduced. However, scheduling the distributed, dynamic and real-time Deep Neural Network (DNN) inference requests among resource-constrained devices has not been well explored in the literature. Additionally, the distribution of DNN has drawn the attention to the privacy protection of sensitive data. In this context, various threats have been presented, including white-box attacks, where malicious devices can accurately recover received inputs if the DNN model is fully exposed to participants. In this paper, we introduce a methodology aiming at distributing the DNN tasks onto the resource-constrained devices of the IoT system, while avoiding to reveal the model to participants. We formulate such an approach as an optimization problem, where we establish a trade-off between the latency of co-inference, the privacy of the data, and the limited resources of devices. Next, due to the NP-hardness of the problem, we shape our approach as a reinforcement learning design adequate for real-time applications and highly dynamic systems, namely RL-PDNN. Our system proved its ability to outperform existing static approaches and achieve close results compared to the optimal solution.<br></p><h2>Other Information</h2><p>Published in: IEEE Access<br>License: <a href="https://creativecommons.org/licenses/by/4.0/legalcode" 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.2021.3070627" target="_blank">https://dx.doi.org/10.1109/access.2021.3070627</a></p>2021-04-02T00:00:00ZTextJournal contributioninfo:eu-repo/semantics/publishedVersiontextcontribution to journal10.1109/access.2021.3070627https://figshare.com/articles/journal_contribution/RL-PDNN_Reinforcement_Learning_for_Privacy-Aware_Distributed_Neural_Networks_in_IoT_Systems/24049251CC BY 4.0info:eu-repo/semantics/openAccessoai:figshare.com:article/240492512021-04-02T00:00:00Z
spellingShingle RL-PDNN: Reinforcement Learning for Privacy-Aware Distributed Neural Networks in IoT Systems
Emna Baccour (16896366)
Information and computing sciences
Artificial intelligence
Data management and data science
Distributed computing and systems software
Machine learning
Task analysis
Servers
Deep learning
Dynamic scheduling
Real-time systems
Performance evaluation
Biological neural networks
IoT devices
Distributed DNN
Privacy
White-box
Resource constraints
DQN
status_str publishedVersion
title RL-PDNN: Reinforcement Learning for Privacy-Aware Distributed Neural Networks in IoT Systems
title_full RL-PDNN: Reinforcement Learning for Privacy-Aware Distributed Neural Networks in IoT Systems
title_fullStr RL-PDNN: Reinforcement Learning for Privacy-Aware Distributed Neural Networks in IoT Systems
title_full_unstemmed RL-PDNN: Reinforcement Learning for Privacy-Aware Distributed Neural Networks in IoT Systems
title_short RL-PDNN: Reinforcement Learning for Privacy-Aware Distributed Neural Networks in IoT Systems
title_sort RL-PDNN: Reinforcement Learning for Privacy-Aware Distributed Neural Networks in IoT Systems
topic Information and computing sciences
Artificial intelligence
Data management and data science
Distributed computing and systems software
Machine learning
Task analysis
Servers
Deep learning
Dynamic scheduling
Real-time systems
Performance evaluation
Biological neural networks
IoT devices
Distributed DNN
Privacy
White-box
Resource constraints
DQN