Adaptive Routing of Task in Computing Force Network by Integrating Graph Convolutional Network and Deep Q-Network

Computing Force Network (CFN) is an emerging technology that has become a hot research topic, with the explosive growth of ultra-low latency and real-time applications with specific computing and network requirements. The main challenge of CFN is to jointly use network and computing resources, and it is difficult to achieve real-time routing to meet the different requirements for different services. In this paper, an Adaptive Routing of task by integrating Graph Convolutional Network and Deep Q-Network (AR-DQN-GCN) is proposed for Computing Force Network, which combines the network and computing resources to provide fine-grained, real-time and dynamic adaptive routing decision. We integrate GCN into Deep Q-Network (DQN) model to obtain topological features from graph structure states. Meanwhile we design an adaptive mechanism, which can adaptively adjust the routing reward mode according to the types of services. We evaluate our model through experiments with representative network topologies.