Federated learning resource management for energy-constrained industrial IoT devices
Given the impact of limited wireless resources, a dynamic multi-dimensional resource joint management algorithm was proposed, which intended to tackle the problem of device failure and training interruption caused by the limited battery energy in federated learning network in industrial Internet of...
Saved in:
| Main Authors: | , , |
|---|---|
| Format: | Article |
| Language: | zho |
| Published: |
Editorial Department of Journal on Communications
2022-08-01
|
| Series: | Tongxin xuebao |
| Subjects: | |
| Online Access: | http://www.joconline.com.cn/zh/article/doi/10.11959/j.issn.1000-436x.2022126/ |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1841540003519266816 |
|---|---|
| author | Shaoshuai FAN Jianbo WU Hui TIAN |
| author_facet | Shaoshuai FAN Jianbo WU Hui TIAN |
| author_sort | Shaoshuai FAN |
| collection | DOAJ |
| description | Given the impact of limited wireless resources, a dynamic multi-dimensional resource joint management algorithm was proposed, which intended to tackle the problem of device failure and training interruption caused by the limited battery energy in federated learning network in industrial Internet of things (IIoT).Firstly, the optimization problem was decoupled into battery energy allocation, equipment resource allocation and communication resource allocation sub-problems which were interdependent with the goal of maximizing the fixed-time learning accuracy.Then, the equipment transmission and computing resource allocation problem were solved based on particle swarm optimization algorithm under the given energy budget.Thereafter, the resource block iterative matching algorithm was proposed to optimize the optimal communication resource allocation strategy.Finally, the online energy allocation algorithm was proposed to adjust the energy budget allocation.Simulation results validate the proposed algorithm can improve the model learning accuracy compared with other benchmarks, and can perform better in energy shortage scenarios. |
| format | Article |
| id | doaj-art-f21f0d9354ab44b7bf10b03a4ea5bc62 |
| institution | Kabale University |
| issn | 1000-436X |
| language | zho |
| publishDate | 2022-08-01 |
| publisher | Editorial Department of Journal on Communications |
| record_format | Article |
| series | Tongxin xuebao |
| spelling | doaj-art-f21f0d9354ab44b7bf10b03a4ea5bc622025-01-14T06:28:55ZzhoEditorial Department of Journal on CommunicationsTongxin xuebao1000-436X2022-08-0143657759392271Federated learning resource management for energy-constrained industrial IoT devicesShaoshuai FANJianbo WUHui TIANGiven the impact of limited wireless resources, a dynamic multi-dimensional resource joint management algorithm was proposed, which intended to tackle the problem of device failure and training interruption caused by the limited battery energy in federated learning network in industrial Internet of things (IIoT).Firstly, the optimization problem was decoupled into battery energy allocation, equipment resource allocation and communication resource allocation sub-problems which were interdependent with the goal of maximizing the fixed-time learning accuracy.Then, the equipment transmission and computing resource allocation problem were solved based on particle swarm optimization algorithm under the given energy budget.Thereafter, the resource block iterative matching algorithm was proposed to optimize the optimal communication resource allocation strategy.Finally, the online energy allocation algorithm was proposed to adjust the energy budget allocation.Simulation results validate the proposed algorithm can improve the model learning accuracy compared with other benchmarks, and can perform better in energy shortage scenarios.http://www.joconline.com.cn/zh/article/doi/10.11959/j.issn.1000-436x.2022126/federated learningbattery-poweredresource allocationlearning efficiency |
| spellingShingle | Shaoshuai FAN Jianbo WU Hui TIAN Federated learning resource management for energy-constrained industrial IoT devices Tongxin xuebao federated learning battery-powered resource allocation learning efficiency |
| title | Federated learning resource management for energy-constrained industrial IoT devices |
| title_full | Federated learning resource management for energy-constrained industrial IoT devices |
| title_fullStr | Federated learning resource management for energy-constrained industrial IoT devices |
| title_full_unstemmed | Federated learning resource management for energy-constrained industrial IoT devices |
| title_short | Federated learning resource management for energy-constrained industrial IoT devices |
| title_sort | federated learning resource management for energy constrained industrial iot devices |
| topic | federated learning battery-powered resource allocation learning efficiency |
| url | http://www.joconline.com.cn/zh/article/doi/10.11959/j.issn.1000-436x.2022126/ |
| work_keys_str_mv | AT shaoshuaifan federatedlearningresourcemanagementforenergyconstrainedindustrialiotdevices AT jianbowu federatedlearningresourcemanagementforenergyconstrainedindustrialiotdevices AT huitian federatedlearningresourcemanagementforenergyconstrainedindustrialiotdevices |