Dynamic 5G RAN slice adjustment and migration based on traffic prediction in WDM metro-aggregation networks

5G radio access network (RAN) slicing has recently been proposed to adapt service provision over substrate network resources to the diversified requirements of 5G services in terms of bandwidth, computation, latency, and reliability. 5G RAN slicing allows baseband processing functions, including centralized units and distributed units, to be deployed in the substrate network according to the different service requirements. Even though existing studies have shown how to perform 5G RAN slicing to flexibly provision service according to their initial requirements, a proper dynamic RAN slice adjustment and migration strategy is also required to accommodate the highly dynamic nature of mobile traffic while maintaining lower service blocking. In this paper, we provide a traffic-prediction-based strategy for dynamic 5G RAN slice adjustment and migration in wavelength division multiplexing (WDM) metro-aggregation networks, which targets the minimization of (1) the slice degradation penalty, which accounts for both the degradation degree and slice priority, and (2) the migrated traffic during slice migration. Our numerical results show that the proposed strategy can effectively reduce the penalty and slice migration.