Admission Control and Resource Reservation for Prioritized Slice Requests with Guaranteed SLA under Uncertainties
Network slicing has emerged as a key concept in 5G systems, allowing Mobile Network Operators (MNOs) to build isolated logical networks (slices) on top of shared infrastructure networks managed by Infrastructure Providers (InP). Network slicing requires the assignment of infrastructure network resources to virtual network components at slice activation time and the adjustment of resources for slices under operation. Performing these operations just-in-time, on a best-effort basis, comes with no guarantee on the availability of enough infrastructure resources to meet slice requirements. This paper proposes a prioritized admission control mechanism for concurrent slices based on an infrastructure resource reservation approach. The reservation accounts for the dynamic nature of slice requests while being robust to uncertainties in slice resource demands. Adopting the perspective of an InP, reservation schemes are proposed that maximize the number of slices for which infrastructure resources can be granted while minimizing the costs charged to the MNOs. This requires the solution of a max-min optimization problem with a non-linear cost function and non-linear constraints induced by the robustness to uncertainties of demands and the limitation of the impact of reservation on background services. The cost and the constraints are linearized and several reduced-complexity strategies are proposed to solve the slice admission control and resource reservation problem. Simulations show that the proportion of admitted slices of different priority levels can be adjusted by a differentiated selection of the delay between the reception and the processing instants of a slice resource request.