Native Network Digital Twin Architecture for 6G: From Design to Practice

Based on the operator's experience in actual large-scale network operations, it is essential to provide deep insight into complex and largescale networks from all perspectives. Meanwhile, future network evolution will also require comprehensive optimization regarding service experience and network operation. Network digital twin (NDT) is currently considered a potential solution to tackle those challenges, by which the physical network can be effectively monitored, analyzed, predicted, simulated, and optimized through its real-time replica. In this context, this article proposes a native network digital twin architecture for 6G, and elaborates on NDT's full life cycle process from design state to runtime state. Moreover, to address two major technical challenges regarding real-time data acquisition and digital twin modelling, digital acquisition of network functions in a native mode, new data pipeline mechanism, and gray-box modelling are introduced. Finally, an NDT has been designed and implemented for the rehearsal and evaluation of network signaling storms to tackle the challenges that they present to network operators, and has been compared with real network testing results. The experimental results show that the mean absolute percentage error (MAPE) values of the successful registration count and registration arrival rate on NDT reach 25 percent and 3 percent, respectively, convincingly demonstrating the effectiveness of the proposed architecture and digital twin modelling method.