Digital Twin-Assisted Satellite-Ground Cooperative Edge Network Resource Optimization Method

Shuopeng Li, Ran Han
Proceedings of 2025 2nd International Conference on Machine Learning and Intelligent Computing, PMLR 278:8-22, 2025.

Abstract

To address the challenge of efficiently processing data on resource-constrained IoT devices, this paper proposes a digital twin architecture for satellite-terrestrial collaborative edge networks and introduces Coordinated Constrained DDPG (CC_DDPG), a deep reinforcement learning (DRL)-based task offloading and resource allocation algorithm tailored for model training tasks. First, digital twin models for UAVs and satellites are constructed to enable real-time network state monitoring and decision support. Second, the joint optimization of task offloading, communication, and computing resources is formulated as a Markov Decision Process (MDP). By enhancing the actor network in the conventional DDPG method, the proposed algorithm dynamically balances training latency and energy consumption. Simulation results demonstrate that CC_DDPG significantly outperforms benchmark heuristic algorithms in convergence stability and multi-objective optimization performance.

Cite this Paper

BibTeX
@InProceedings{pmlr-v278-li25b, title = {Digital Twin-Assisted Satellite-Ground Cooperative Edge Network Resource Optimization Method}, author = {Li, Shuopeng and Han, Ran}, booktitle = {Proceedings of 2025 2nd International Conference on Machine Learning and Intelligent Computing}, pages = {8--22}, year = {2025}, editor = {Zeng, Nianyin and Pachori, Ram Bilas and Wang, Dongshu}, volume = {278}, series = {Proceedings of Machine Learning Research}, month = {25--27 Apr}, publisher = {PMLR}, pdf = {https://raw.githubusercontent.com/mlresearch/v278/main/assets/li25b/li25b.pdf}, url = {https://proceedings.mlr.press/v278/li25b.html}, abstract = {To address the challenge of efficiently processing data on resource-constrained IoT devices, this paper proposes a digital twin architecture for satellite-terrestrial collaborative edge networks and introduces Coordinated Constrained DDPG (CC_DDPG), a deep reinforcement learning (DRL)-based task offloading and resource allocation algorithm tailored for model training tasks. First, digital twin models for UAVs and satellites are constructed to enable real-time network state monitoring and decision support. Second, the joint optimization of task offloading, communication, and computing resources is formulated as a Markov Decision Process (MDP). By enhancing the actor network in the conventional DDPG method, the proposed algorithm dynamically balances training latency and energy consumption. Simulation results demonstrate that CC_DDPG significantly outperforms benchmark heuristic algorithms in convergence stability and multi-objective optimization performance.} }
Endnote
%0 Conference Paper %T Digital Twin-Assisted Satellite-Ground Cooperative Edge Network Resource Optimization Method %A Shuopeng Li %A Ran Han %B Proceedings of 2025 2nd International Conference on Machine Learning and Intelligent Computing %C Proceedings of Machine Learning Research %D 2025 %E Nianyin Zeng %E Ram Bilas Pachori %E Dongshu Wang %F pmlr-v278-li25b %I PMLR %P 8--22 %U https://proceedings.mlr.press/v278/li25b.html %V 278 %X To address the challenge of efficiently processing data on resource-constrained IoT devices, this paper proposes a digital twin architecture for satellite-terrestrial collaborative edge networks and introduces Coordinated Constrained DDPG (CC_DDPG), a deep reinforcement learning (DRL)-based task offloading and resource allocation algorithm tailored for model training tasks. First, digital twin models for UAVs and satellites are constructed to enable real-time network state monitoring and decision support. Second, the joint optimization of task offloading, communication, and computing resources is formulated as a Markov Decision Process (MDP). By enhancing the actor network in the conventional DDPG method, the proposed algorithm dynamically balances training latency and energy consumption. Simulation results demonstrate that CC_DDPG significantly outperforms benchmark heuristic algorithms in convergence stability and multi-objective optimization performance.
APA
Li, S. & Han, R.. (2025). Digital Twin-Assisted Satellite-Ground Cooperative Edge Network Resource Optimization Method. Proceedings of 2025 2nd International Conference on Machine Learning and Intelligent Computing, in Proceedings of Machine Learning Research 278:8-22 Available from https://proceedings.mlr.press/v278/li25b.html.

Related Material