A Simulation on Resource Allocation Policies: URLLC Packets Loss and eMBB Packets Waiting Time

5G leverages the deployment of end-to-end virtual networks, with specific Quality of Service (QoS) profiles, on top of a common physical infrastructure. Slicing is the term used for the functionality that makes this coexistence possible. Each deployed virtual network is called a slice in 5G terminology. Slicing is particularly challenging to dimension in order to abide by the QoS constraints of slices. For instance, slices such as Enhanced Mobile Broadband (eMBB) and Ultra Reliable Low Latency Communications (URLLC) have conflicting QoS requirements. The deployment of these two types of slices requires accurate dimensioning of all resources in the physical network. Performance evaluation models are then needed to identify the functional limits of these new systems.This study addresses the requirements of Électricité de France (EDF) in utilizing a URLLC slice for the control, automation, and protection of their smart electrical network. Due to the inability to deploy an independent network, EDF is a client of a Mobile Network Operator (MNO) that serves its mass users with an eMBB slice that shares the same network as their URLLC slice. Given that 5G slicing has not yet been implemented, a simulation is developed to replicate real-world conditions of resource allocation and determine the optimal number of reserved Resource Blocks (RBs), named guard channels, for exclusively serving URLLC packets while maintaining a maximum loss rate.The results illustrate how the required number of guard channels varies with the total number of servers in the system and the arrival rate of URLLC packets. Since the value of the arrival rate of eMBB packets (λ_e) is not possible to obtain from the operator, the data is plotted by varying λ_e.