Intelligent Congestion Control in QUIC for Reliable E2E Communication Network: A Digital Twin-based Approach

In this paper, we propose a digital twin-based intelligent congestion control approach to enhance the packet transmission adaptability and efficiency of the QUIC protocol within dynamic network environments. Specifically, by integrating actor-critic learning methods, the proposed approach adaptively tailors the pre-defined congestion control parameter settings to suit various application scenarios, thereby optimizing QUIC's performance under fluctuating network conditions. In addition, utilizing digital twin technology, the algorithm constructs a simulation model that mirrors the physical network within a digital framework. This model facilitates the rapid collection of network data, precise monitoring of congestion levels, and detailed analysis of historical transmission records and network state changes. These capabilities enable the predictive adjustment of model parameters and timely modifications to the congestion control strategy, ensuring optimal performance. Simulation results demonstrate that the proposed digital twin-based intelligent approach can enhance QUIC's transmission capabilities across diverse network conditions. Notably, it increases data throughput, reduces transmission delays, improves the convergence speed of the intelligence algorithm, and ensures reliable end-to-end transmissions within 5G networks.