AI Services-Oriented Dynamic Computing Resource Scheduling Algorithm Based on Distributed Data Parallelism in Edge Computing Network of Smart Grid
Abstract
:1. Introduction
2. Related Work
2.1. Heuristic for Resource Scheduling
2.2. Machine Learning-Based Approach for Resource Scheduling
3. Materials and Methods
- Splitting of AI training data
- 2.
- Task scheduling order
- 3.
- Edge data center selection
- 4.
- Edge computing resource scheduling
3.1. MILP Model
3.2. DRL-Based Dynamic Collaborative Resource Scheduling Algorithm
- (1)
- Initialize the parameters on edge data centers of different edge computational resource suppliers.
- (2)
- Input the request and the current state of edge data center in the t-th time frame into DNN, which consists of an M-layer convolutional neural network and a N-layer fully connected network.
- (3)
- Based on and , the corresponding Q value is calculated for each action, and then DNN outputs the set of Q values for all possible actions under the current strategy , where indicates the action in t-th time frame.
- (4)
- The action with the largest value is selected from the set of Q values to interact with the environment. The action can also be selected utilizing the -Greedy strategy. That is to say that the action is selected with the probability with the highest Q value. The action is randomly selected to fully explore the action space with (1-) probability. The value will gradually increase after each action selection.
- (5)
- After executing the action , the status of remaining computational resource capacity on the edge data center is updated to the next state and return the reward value .
- (6)
- The quadruple entry <,,,> is saved in the memory.
- (7)
- L sets of data entries from the memory are quantitatively and randomly extracted at regular intervals. In other words, a random mini-batch of data <> are sampled to update DNN parameters by .
- (8)
- Train DNN parameters using extracted data entries based on Bellman optimization equations, formulated as , where is a discount factor. The Bellman error, formulated as , can be minimized to update DNN parameters by the gradient descent method. Then, the value of actions is fit well by DNN if the Bellman error converges after updating the DNN parameters.
- (9)
- Repeat the above steps until the Q-value function converges. When using the model, the current computational resource status on the edge data center is also inputted into the model along with user requirements, and the model will return a dynamic, collaborative resource scheduling strategy.
4. Simulation Results and Analysis
4.1. Simulation Setup
4.2. Simulation Scenario I
- (a)
- Multi-objective optimization vs. number of AI services
- (b)
- Profit vs. number of AI services
- (c)
- Backlog vs. number of AI services
- (d)
- Running time vs. number of AI services
4.3. Simulation Scenario II
- (a)
- Multi-objective optimization vs. number of AI services
- (b)
- Profit vs. number of AI services
- (c)
- Backlog vs. number of AI services
- (d)
- Running time vs. number of AI services
5. Discussion and Conclusions
Author Contributions
Funding
Data Availability Statement
Acknowledgments
Conflicts of Interest
References
- Tang, W.; Lu, W.; Tan, Q.; Ma, J.; Liu, Z. Research on the status and development trend of the data center from the perspective of energy and power. In Proceedings of the IEEE International Conference on Advances in Electrical Engineering and Computer Applications (AEECA), Dalian, China, 20–21 August 2022; pp. 562–565. [Google Scholar]
- Luo, F.; Zhao, J.; Dong, Z.Y.; Chen, Y.; Xu, Y.; Zhang, X.; Wong, K.P. Cloud-Based Information Infrastructure for Next-Generation Power Grid: Conception, Architecture, and Applications. IEEE Trans. Smart Grid 2016, 7, 1896–1912. [Google Scholar] [CrossRef]
- Slama, S.B. Prosumer in smart grids based on intelligent edge computing: A review on Artificial Intelligence Scheduling Techniques. Ain Shams Eng. J. 2022, 13, 101504. [Google Scholar] [CrossRef]
- Yang, Q.; Liu, Y.; Chen, T.; Tong, Y. Federated machine learning: Concept and applications. ACM Trans. Intell. Syst. Technol. 2019, 10, 1–19. [Google Scholar] [CrossRef]
- Feng, C.; Wang, Y.; Chen, Q.; Ding, Y.; Strbac, G.; Kang, C. Smart grid encounters edge computing: Opportunities and applications. Adv. Appl. Energy 2021, 1, 100006. [Google Scholar] [CrossRef]
- Gao, H.; Xia, W.; Yan, F.; Shen, L. Optimal Cloud Resource Scheduling in Smart Grid: A Hierarchical Game Approach. In Proceedings of the 2020 IEEE 91st Vehicular Technology Conference (VTC2020-Spring), Online, 25 May–31 July 2020. [Google Scholar]
- Yao, J.; Li, Z.; Li, Y.; Bai, J.; Wang, J.; Lin, P. Cost-Efficient Tasks Scheduling for Smart Grid Communication Network with Edge Computing System. In Proceedings of the 15th International Wireless Communications & Mobile Computing Conference (IWCMC), Tangier, Morocco, 24–28 June 2019; pp. 272–277. [Google Scholar]
- Li, Z.; Liu, Y.; Xin, R.; Gao, L.; Ding, X.; Hu, Y. A Dynamic Game Model for Resource Allocation in Fog Computing for Ubiquitous Smart Grid. In Proceedings of the 2019 28th Wireless and Optical Communications Conference (WOCC), Beijing, China, 9–10 May 2019; pp. 1–5. [Google Scholar]
- Sharif, Z.; Jung, L.T.; Razzak, I.; Alazab, M. Adaptive and Priority-Based Resource Allocation for Efficient Resources Utilization in Mobile-Edge Computing. IEEE Internet Things J. 2023, 10, 3079–3093. [Google Scholar] [CrossRef]
- Jiang, J.; Xin, P.; Wang, Y.; Liu, L.; Chai, Y.; Zhang, Y.; Liu, S. Computing Resource Allocation in Mobile Edge Networks Based on Game Theory. In Proceedings of the 2021 IEEE 4th International Conference on Electronics and Communication Engineering (ICECE), Xi′an, China, 17–19 December 2021; pp. 179–183. [Google Scholar]
- Wang, G.; Xu, F.; Zhao, C. Multi-Access Edge Computing Based Vehicular Network: Joint Task Scheduling and Resource Allocation Strategy. In Proceedings of the 2020 IEEE International Conference on Communications Workshops (ICC Workshops), Dublin, Ireland, 7–11 June 2020; pp. 1–6. [Google Scholar]
- Yan, J.; Rui, L.; Yang, Y.; Chen, S.; Chen, X. Resource Scheduling Algorithms for Burst Network Flow in Edge Computing. In Lecture Notes in Electrical Engineering, Proceedings of the 11th International Conference on Computer Engineering and Networks, Hechi, China, 21–25 October 2021; Liu, Q., Liu, X., Chen, B., Zhang, Y., Peng, J., Eds.; Springer: Singapore, 2022; Volume 808, p. 808. [Google Scholar]
- Seah, W.K.G.; Lee, C.-H.; Lin, Y.-D.; Lai, Y.-C. Combined Communication and Computing Resource Scheduling in Sliced 5G Multi-Access Edge Computing Systems. IEEE Trans. Veh. Technol. 2021, 71, 3144–3154. [Google Scholar] [CrossRef]
- Jun, N. Research on cloud computing resource scheduling strategy based on genetic ant colony fusion algorithm. In Proceedings of the Conference on Computer Science and Communication Technology, Tianjin, China, 12–13 March 2022. [Google Scholar]
- Liu, Z.; Zhan, C.; Cui, Y.; Wu, C.; Hu, H. Robust Edge Computing in UAV Systems via Scalable Computing and Cooperative Computing. IEEE Wirel. Commun. 2021, 28, 36–42. [Google Scholar] [CrossRef]
- Shao, X.; Hasegawa, G.; Dong, M.; Liu, Z.; Masui, H.; Ji, Y. An Online Orchestration Mechanism for General-Purpose Edge Computing. IEEE Trans. Serv. Comput. 2023, 16, 927–940. [Google Scholar] [CrossRef]
- Xi, L.; Wang, Y.; Wang, Y.; Wang, Z.; Wang, X.; Chen, Y. Deep Reinforcement Learning-Based Service-Oriented Resource Allocation in Smart Grids. IEEE Access 2021, 9, 77637–77648. [Google Scholar] [CrossRef]
- Wang, T.; Lu, B.; Wang, W.; Wei, W.; Yuan, X.; Li, J. Reinforcement Learning-Based Optimization for Mobile Edge Computing Scheduling Game. IEEE Trans. Emerg. Top. Comput. Intell. 2023, 7, 55–64. [Google Scholar] [CrossRef]
- Vijayasekaran, G.; Duraipandian, M. Deep Q-learning based Resource Scheduling in IoT Edge Computing. In Proceedings of the 2024 International Conference on Inventive Computation Technologies (ICICT), Lalitpur, Nepal, 24–26 April 2024; pp. 359–364. [Google Scholar]
- Wang, S.; Wu, Y.-C.; Xia, M.; Wang, R.; Poor, H.V. Machine intelligence at the edge with learning centric power allocation. IEEE Trans. Wireless Commun. 2020, 19, 7293–7308. [Google Scholar] [CrossRef]
- Chi, Y.; Zhang, Y.; Liu, Y.; Zhu, H.; Zheng, Z.; Liu, R.; Zhang, P. Deep Reinforcement Learning Based Edge Computing Network Aided Resource Allocation Algorithm for Smart Grid. IEEE Access 2022, 11, 6541–6550. [Google Scholar] [CrossRef]
- Chen, S.; Wen, H.; Wu, J.; Lei, W.; Hou, W.; Liu, W.; Xu, A.; Jiang, Y. Internet of Things Based Smart Grids Supported by Intelligent Edge Computing. IEEE Access 2019, 7, 74089–74102. [Google Scholar] [CrossRef]
- Lin, Y.; Han, S.; Mao, H.; Wang, Y.; Dally, W.J. Deep gradient compression: Reducing the communication bandwidth for distributed training. In Proceedings of the Sixth International Conference on Learning Representations, Vancouver, BC, Canada, 30 April–3 May 2018; pp. 1–14. [Google Scholar]
- Tran, N.H.; Bao, W.; Zomaya, A.; Nguyen, M.N.H.; Hong, C.S. Federated learning over wireless networks: Optimization model design and analysis. In Proceedings of the IEEE Conference on Computer Communications, Paris, France, 29 April–2 May 2019; pp. 1387–1395. [Google Scholar]
- Xu, J.; Liu, X.; Zhu, X. Deep Reinforcement Learning Based Computing Offloading and Resource Allocation Algorithm for Mobile Edge Networks. In Proceedings of the 2020 IEEE 6th International Conference on Computer and Communications (ICCC), Chengdu, China, 11–14 December 2020; pp. 1542–1547. [Google Scholar]
- Liu, M.; Li, Y.; Zhao, Y.; Yang, H.; Zhang, J. Adaptive DNN model partition and deployment in edge computing-enabled metro optical interconnection network. In Proceedings of the Optical Fiber Communication (OFC) Conference, San Diego, CA, USA, 8–12 March 2020. [Google Scholar]
- Li, Y.; Zeng, Z.; Li, J.; Yan, B.; Zhao, Y.; Zhang, J. Distributed Model Training Based on Data Parallelism in Edge Computing-Enabled Elastic Optical Networks. IEEE Commun. Lett. 2020, 25, 1241–1244. [Google Scholar] [CrossRef]
- Wang, H.; Agarwal, S.; Papailiopoulos, D. Pufferfish: Communication-efficient Models at No Extra Cost. arXiv 2021, arXiv:2103.03936. [Google Scholar]
- Mansour, A.B.; Carenini, G.; Duplessis, A.; Naccache, D. Federated Learning Aggregation: New Robust Algorithms with Guarantees. In Proceedings of the 2022 21st IEEE International Conference on Machine Learning and Applications (ICMLA), Nassau, Bahamas, 12–14 December 2022; pp. 721–726. [Google Scholar]
Parameters | Descriptions |
---|---|
Set of edge computing resource suppliers, . | |
Set of edge data centers, j. | |
Set of computing resource request users, k. | |
Set of different types of AI services requested by computing service request users, . | |
Backlog threshold of AI-distributed model training tasks in the edge data center (GOPS). | |
A very large positive integer. |
Parameters | Descriptions |
---|---|
Computing resource processing capacity of j-th edge data center for i-th computing resource supplier (GOPS), , j. | |
Computational processing cost of j-th edge data center for i-th computing resource supplier (Yuan/GOPS), , j. | |
Edge computing service request for k-th user’s f-th type AI service request (GOPS), k, . | |
Payment for data computing service of f-th type of AI service for k-th computing service request user (Yuan/GOPS), k, . | |
Complaint rate of AI-distributed model training tasks for f-th type of AI service from k-th computing service request user (statistical value), k, . | |
The remaining amount of AI-distributed model training tasks can be complained about in the j-th edge data center of i-th computing resource supplier (GOPS), , j. |
Parameters | Descriptions |
---|---|
Boolean variable. The requesting computing resources of k-th user’s f-th type of AI service is assigned to the j-th edge data center of i-th edge computing resource supplier (GOPS) in t-th time frame. If it is, equals 1. | |
Backlog of AI-distributed model training tasks from j-th edge data centers of i-th edge computing resource supplier (GOPS) in t-th time frame. | |
The j-th edge data center of i-th edge computing resource supplier is used in t-th time frame. If it is, equals 1. |
Parameters | Values |
---|---|
Number of CNN layers | 2 |
Filters of CNN | 5 |
Kernel size of CNN | 2 |
Strides of CNN | 1 |
Number of training iterations | 3000 |
Learning rate | 0.01 |
Reward decay | 0.9 |
Memory size | 5000 |
Batch size | 32 |
Exploration rate | 0.9 |
Disclaimer/Publisher’s Note: The statements, opinions and data contained in all publications are solely those of the individual author(s) and contributor(s) and not of MDPI and/or the editor(s). MDPI and/or the editor(s) disclaim responsibility for any injury to people or property resulting from any ideas, methods, instructions or products referred to in the content. |
© 2024 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https://creativecommons.org/licenses/by/4.0/).
Share and Cite
Zou, J.; Xin, P.; Wang, C.; Zhang, H.; Wei, L.; Wang, Y. AI Services-Oriented Dynamic Computing Resource Scheduling Algorithm Based on Distributed Data Parallelism in Edge Computing Network of Smart Grid. Future Internet 2024, 16, 312. https://doi.org/10.3390/fi16090312
Zou J, Xin P, Wang C, Zhang H, Wei L, Wang Y. AI Services-Oriented Dynamic Computing Resource Scheduling Algorithm Based on Distributed Data Parallelism in Edge Computing Network of Smart Grid. Future Internet. 2024; 16(9):312. https://doi.org/10.3390/fi16090312
Chicago/Turabian StyleZou, Jing, Peizhe Xin, Chang Wang, Heli Zhang, Lei Wei, and Ying Wang. 2024. "AI Services-Oriented Dynamic Computing Resource Scheduling Algorithm Based on Distributed Data Parallelism in Edge Computing Network of Smart Grid" Future Internet 16, no. 9: 312. https://doi.org/10.3390/fi16090312