A Semi-Empirical Data-Rate Estimation Method of 5G RAN Slicing

Network Slicing is a novel 5G technique enabling a single net to meet differential needs from multiple tenants. In this paper, a semi-empirical method called linear reduction factor (LRF) for the estimation and prediction of data goodput rates is proposed for radio access network (RAN) slicing. The method is derived from our fundamental understanding of peak data throughput calculation, and a bold but ambitious assumption that the linearity of the RAN-sliced data goodput should hold with the amount of allocated resource, in units of resource block group (RBG). The paper presents theoretical derivation steps in a semi-empirical way toward how to estimate the slice goodput directly from the measured reduction factor of system throughput with the full frequency bandwidth. When combined the linearity with the number of allocated REGs, the proposed method can precisely predict the data goodput for a RAN slice with any given number of RBGs. This research also successfully ran a software-defined platform to conduct the verification experiments using the Mosaic5G FlexRAN real-time controller for configuring and monitoring the emulated RAN. The desired linearity and precision of the proposed LRF method are fully supported by two case studies, namely equal-slicing and prioritized-slicing.