Combining graph neural network with deep reinforcement learning for resource allocation in computing force networks

• Published in: Frontiers of information technology & electronic engineering Vol. 25; no. 5; pp. 701 - 712
• Main Authors: Han, Xueying, Xie, Mingxi, Yu, Ke, Huang, Xiaohong, Du, Zongpeng, Yao, Huijuan
• Format: Journal Article
• Language: English
• Published: Hangzhou Zhejiang University Press 01.05.2024; Springer Nature B.V
• Subjects: Bandwidths; Cloud computing; Communications Engineering; Communications networks; Communications traffic; Computation; Computer Hardware; Computer Science; Computer Systems Organization and Communication Networks; Deep learning; Edge computing; Electrical Engineering; Electronics and Microelectronics; Graph neural networks; Instrumentation; Internet service providers; Machine learning; Network latency; Network management systems; Network topologies; Networks; Neural networks; Optimization; Power; Research Article; Resource allocation; Deep learning; 图神经网络; 资源分配; Computing force network; TP393; Graph neural network; 路由优化; Resource allocation; 深度学习; Routing optimization; 算力网络
• ISSN: 2095-9184; 2095-9230
• DOI: 10.1631/FITEE.2300009

Fueled by the explosive growth of ultra-low-latency and real-time applications with specific computing and network performance requirements, the computing force network (CFN) has become a hot research subject. The primary CFN challenge is to leverage network resources and computing resources. Although recent advances in deep reinforcement learning (DRL) have brought significant improvement in network optimization, these methods still suffer from topology changes and fail to generalize for those topologies not seen in training. This paper proposes a graph neural network (GNN) based DRL framework to accommodate network traffic and computing resources jointly and efficiently. By taking advantage of the generalization capability in GNN, the proposed method can operate over variable topologies and obtain higher performance than the other DRL methods.