Hybrid Model-Data Driven Network Slice Reconfiguration by Exploiting Prediction Interval and Robust Optimization

Proactive reconfiguration of network slices according to uncertain traffic demands is essential improve resource utilization while ensuring service quality in 5G-and-beyond systems. Existing researches on slice are either model-driven or data-driven methods. However, methods may cause over-provisioning due a lack prediction mechanism, unrealistic inter-slice that involves costly and time-consuming operations such as VNF migration. To address these issues, this paper, we propose Hybrid Model-Data driven (HMD) framework intelligently performs by leveraging interval robust optimization. We design Prediction Interval-oriented Predictor (PIP) produce can bracket the future demand with prespecified probability. Based interval, an scheme (named box optimizer) perform fast reconfigurations. tackle over-conservativeness optimizer, further ellipsoid optimizer better optimality at cost increased complexity. Numerical results demonstrate proposed provide high robustness low power consumption. Meanwhile, trade-off between consumption realized be flexibly adjusted type level fluctuations.