Dynamic, Reconfigurable and Green Network Slice Admission Control and Resource Allocation in the O-RAN Using Model Predictive Control

In the context of 5G, virtualization has transformed Radio Access Network (RAN) architectures, enabling efficient resource utilization, flexibility and scalability of RAN deployments and operations. Within this paradigm, network slicing has emerged as a pivotal technique, enabling the creation of tailored virtual network instances to meet diverse service requirements. This study performs joint slice request admission control and optimal Virtual Network Functions (VNFs) placement in O-RAN-enabled networks, subject to infrastructure and Quality-of-Service constraints. Contrary to existing schemes, emphasis is placed on two not yet deeply studied directions. The first is about handling future uncertainties on slice requests arrivals for which an iterative Model Predictive Control (MPC)-based approach is proposed that leverages updated traffic forecasts for dynamic adaptation. The second relates to VNFs migration in the O-RAN modules to increase the slice acceptance ratio in an energy efficient way. Through performance evaluations and comparisons, we demonstrate the efficacy of the proposed MPC-based solution compared to other approaches.