A study of the business parameter matching model for backbone transmission networks and data networks based on reinforcement learning

Aiming at the dynamic optimization problem of matching service parameters between backbone transmission network and data network, this study proposes a reinforcement learning-based model to improve network resource utilization efficiency and performance. The model is based on deep Q network and realizes adaptive matching of bandwidth allocation, delay control and traffic demand by designing state space, action space and reward function. Simulation experiments show that the model outperforms traditional static methods in different load scenarios, significantly improves reward value, bandwidth utilization and reduces delay. By modeling the network state as a dynamic system and utilizing the learning ability of intelligent agents, the model effectively overcomes the shortcomings of traditional methods in dynamic environments and shows good convergence and engineering application potential. This study provides new ideas for the intelligent optimization of network parameters and provides technical support for the coordinated operation of backbone transmission network and data network.