Routing and Resource Scheduling Algorithm Driven by Mixture of Experts in Large-scale Heterogeneous Local Power Communication Network

JING Chuanfang; ZHU Xiaorong

doi:10.11999/JEIT251176

Article Contents

Article Navigation > Journal of Electronics & Information Technology > 2025 >

JING Chuanfang, ZHU Xiaorong. Routing and Resource Scheduling Algorithm Driven by Mixture of Experts in Large-scale Heterogeneous Local Power Communication Network[J]. Journal of Electronics & Information Technology. doi: 10.11999/JEIT251176

Citation:

JING Chuanfang, ZHU Xiaorong. Routing and Resource Scheduling Algorithm Driven by Mixture of Experts in Large-scale Heterogeneous Local Power Communication Network[J]. Journal of Electronics & Information Technology. doi: 10.11999/JEIT251176

Citation:

PDF( 2828 KB)

Routing and Resource Scheduling Algorithm Driven by Mixture of Experts in Large-scale Heterogeneous Local Power Communication Network

doi: 10.11999/JEIT251176 cstr: 32379.14.JEIT251176

JING Chuanfang,
ZHU Xiaorong^,

School of Communication and Information Engineering, Nanjing University of Posts and Telecommunications, Nanjing 210003, China

Funds: The National Science and Technology Major Project (2024ZD1300400), The Natural Science Foundation of China (92367102), The Postgraduate Research & Practice Innovation Program of Jiangsu Province (KYCX22_0944)

Received Date: 2025-11-10
Accepted Date: 2026-01-22
Rev Recd Date: 2026-01-22

Available Online: 2026-02-01

Abstract

Abstract

Objective Emerging power services, such as distributed energy consumption, place stringent performance requirements on Large-Scale Heterogeneous Local Power Communication Networks (LHLPCNs). Limited communication resources and increasing service demands make it challenging to provide on-demand services and improve network capacity while ensuring Quality of Service (QoS). Conventional routing and resource scheduling algorithms based on optimization or heuristics depend on precise mathematical models and parameters, and their computational cost increases as network size and variables grow. These limitations reduce their adaptability to expanding power application scenarios. Advances in Mixture-of-Experts (MoE) frameworks offer a promising direction because they reduce the need to train task-specific models by using an ensemble of specialized AI experts. Motivated by these challenges, this study proposes an MoE-based routing and resource scheduling algorithm (RASMoE) for LHLPCNs integrating High-Power Line Carrier (HPLC) and Radio Frequency (RF). RASMoE is designed to meet personalized QoS requirements and support more power services within limited resources. Methods An optimization problem that minimizes the difference between QoS supply and demand in LHLPCNs is formulated as a 0–1 integer linear programming model considering multimodal links, channels, and modulation methods. To solve this NP-hard problem, a new MoE framework comprising expert networks and gated networks is designed. The framework supports personalized service requirements in terms of data rate, delay, and reliability, while improving convergence. The expert networks include shared and QoS-specific experts that generate optimal next hops and compute allocation strategies for links, channels, and modulation modes between node pairs. The gated networks dynamically combine and reuse these experts to support known and unforeseen service types. Extensive comparative experiments are conducted, and RASMoE shows improved resource utilization, reduced delay, and higher reliability relative to multiple baselines. Results and Discussions The performance supply-demand differences of five algorithms under varying service numbers are compared (Fig. 3). RASMoE consistently achieves the smallest differences across scenarios due to its gating network, which combines QoS-specific experts to align resource allocation with service requirements. Because control and compute-intensive services have strict delay requirements, their average End-to-End (E2E) latency under different service numbers is evaluated (Fig. 4). The proposed algorithm achieves the lowest average E2E latency because its GAT-enhanced expert networks extract node load states and interact with the network environment in real time through a Multi-Armed Bandit (MAB) mechanism. This supports adaptive allocation strategies. The average reliability of E2E paths for different numbers of control, compute-intensive, and acquisition services is also illustrated (Fig. 5). Conclusions This study proposes a MoE-driven routing and resource scheduling algorithm for LHLPCNs. The framework integrates expert networks and a gating network. The expert networks include GAT-based shared experts for E2E path selection and MAB-based QoS-specific experts for adaptive allocation of links, channels, and modulation schemes according to QoS demands and link states. The gated networks orchestrate and reuse these experts to support services with single or multiple QoS requirements, including previously unseen service types. Theoretical analysis shows that the method improves resource utilization in LHLPCNs, with notable advantages in multi-service scenarios characterized by diverse QoS demands. Future work will examine integrating the MoE framework with domain-specific models, including power load forecasting and predictive analytics, to enhance the use of renewable energy sources.
- Local power communication networks,
- Differentiated QoS,
- On-demand services,
- Routing and resource scheduling,
- Mixture of experts

FullText(HTML)

References(20)

References

[1]	CALERO F, CAÑIZARES C A, BHATTACHARYA K, et al. A review of modeling and applications of energy storage systems in power grids[J]. Proceedings of the IEEE, 2023, 111(7): 806–831. doi: 10.1109/JPROC.2022.3158607.
[2]	WU Dan, SEO G S, XU Lie, et al. Grid integration of offshore wind power: Standards, control, power quality and transmission[J]. IEEE Open Journal of Power Electronics, 2024, 5: 583–604. doi: 10.1109/OJPEL.2024.3390417.
[3]	LI Yuanzheng, YU Chaofan, SHAHIDEHPOUR M, et al. Deep reinforcement learning for smart grid operations: Algorithms, applications, and prospects[J]. Proceedings of the IEEE, 2023, 111(9): 1055–1096. doi: 10.1109/JPROC.2023.3303358.
[4]	LI Jun, SUN Haiyang, DENG Xiumei, et al. Mobility-aware user scheduling in wireless federated learning with contextual multi-armed bandit[J]. China Communications, 2025, 22(11): 256–272. doi: 10.23919/JCC.fa.2023-0427.202511.
[5]	LI Hongbo and DUAN Lingjie. Competitive multi-armed bandit games for resource sharing[J]. IEEE Transactions on Mobile Computing, 2025, 24(9): 8393–8404. doi: 10.1109/TMC.2025.3555971.
[6]	SALAH M M, SAAD R S, ZAKI R M, et al. Multi-armed bandits for resource allocation in UAV-assisted LoRa networks[J]. IEEE Internet of Things Magazine, 2025, 8(2): 40–45. doi: 10.1109/IOTM.001.2400088.
[7]	GALLI A, MOSCATO V, ROMANO S P, et al. Playing with a multi armed bandit to optimize resource allocation in satellite-enabled 5G networks[J]. IEEE Transactions on Network and Service Management, 2024, 21(1): 341–354. doi: 10.1109/TNSM.2023.3302064.
[8]	朱晓荣, 贺楚闳. 基于强化学习的大规模多模Mesh网络联合路由选择及资源调度算法[J]. 电子与信息学报, 2024, 46(7): 2773–2782. doi: 10.11999/JEIT231103. ZHU Xiaorong and HE Chuhong. Joint routing and resource scheduling algorithm for large-scale multi-mode mesh networks based on reinforcement learning[J]. Journal of Electronics & Information Technology, 2024, 46(7): 2773–2782. doi: 10.11999/JEIT231103.
[9]	LI Guiyi, OUYANG Yuxiang, and LINLONG Siyu. En route congestion prediction method for air route network based on spatiotemporal graph convolution network and attention[J]. IEEE Intelligent Transportation Systems Magazine, 2025, 17(3): 33–57. doi: 10.1109/MITS.2024.3450953.
[10]	JI Han, WU Xiping, ZENG Zhihong, et al. Graph attention networks-enabled load balancing for 6G heterogeneous networks with parallel transmission[J]. IEEE Transactions on Vehicular Technology, 2025, 1–14. Early Access. doi: 10.1109/TVT.2025.3636175.
[11]	CHEN Chongpu, CHEN Xinbo, YANG Yi, et al. Sparse attention graph convolution network for vehicle trajectory prediction[J]. IEEE Transactions on Vehicular Technology, 2024, 73(12): 18294–18306. doi: 10.1109/TVT.2024.3443850.
[12]	PARK J and LIM S. LEHAN: Link-feature enhanced heterogeneous graph attention network[J]. IEEE Access, 2022, 10: 86248–86255. doi: 10.1109/ACCESS.2022.3198941.
[13]	LI Chenxu, FENG Lei, LI Wenjing, et al. Long-term traffic flow prediction: A knowledge-driven graph attention spatio-temporal network[J]. IEEE Transactions on Network and Service Management, 2025, 22(5): 4206–4221. doi: 10.1109/TNSM.2025.3595919.
[14]	ZHOU Jingxuan, XIAO Wenhua, ZHANG Dayu, et al. Multi-task model fusion with mixture of experts structure[C]. 2023 9th International Conference on Big Data and Information Analytics (BigDIA), Haikou, China, 2023: 387–391. doi: 10.1109/BigDIA60676.2023.10429752.
[15]	CHENG Guangran, DONG Lu, CAI Wenzhe, et al. Multi-task reinforcement learning with attention-based mixture of experts[J]. IEEE Robotics and Automation Letters, 2023, 8(6): 3812–3819. doi: 10.1109/LRA.2023.3271445.
[16]	YUAN Renteng, ABDEL-ATY M, XIANG Qiaojun, et al. A temporal multi-gate mixture-of-experts approach for vehicle trajectory and driving intention prediction[J]. IEEE Transactions on Intelligent Vehicles, 2024, 9(1): 1204–1216. doi: 10.1109/TIV.2023.3336310.
[17]	RAMBABU R, VADAKKEPAT P, TAN K C, et al. A mixture-of-experts prediction framework for evolutionary dynamic multiobjective optimization[J]. IEEE Transactions on Cybernetics, 2020, 50(12): 5099–5112. doi: 10.1109/TCYB.2019.2909806.
[18]	DU Hongyang, LIU Guangyuan, LIN Yijing, et al. Mixture of experts for intelligent networks: A large language model-enabled approach[C]. 2024 International Wireless Communications and Mobile Computing (IWCMC), Ayia Napa, Cyprus, 2024: 531–536. doi: 10.1109/IWCMC61514.2024.10592370.
[19]	YANG Liang, YAN Xiaoqin, LI Sai, et al. Performance analysis of dual-hop mixed PLC/RF communication systems[J]. IEEE Systems Journal, 2022, 16(2): 2867–2878. doi: 10.1109/JSYST.2021.3088096.
[20]	JING Chuanfang, ZHU Xiaorong, and LIU Xu. Data/model jointly driven routing and resource allocation algorithms for large-scale self-organizing networks for new power systems[C]. 2nd Future Industrial Internet (FII 2024), Shenzhen, China, 2024: 55–76. doi: 10.1007/978-981-96-6736-9_5.