Modeling MTC and HTC Radio Access in a Sliced 5G Base Station

In this article, we develop a modeling frameworkto describe the uplink behavior of radio access in a sliced cell,including most features of the standard 3GPP multiple accessprocedures. Our model allows evaluating throughput and latencyof each slice, as a function of cell parameters, when resourcesare in part dedicated to individual slices and in part shared.The availability of an accurate model is extremely importantfor the automated run time management of the cell and for thecorrect setting of its parameters. Indeed, our model considersmost details of the behavior of sliced 5G cells, including AccessClass Barring (ACB) and Random Access CHannel (RACH) pro-cedures, preamble decoding, Random Access Response (RAR),and Radio Resource Control (RRC) procedures.To cope with a number of slices devoted to serve various co-deployed tenants, we derive a multi-class queueing model ofthe network processor. We then present(i)an accurate andcomputationally efficient technique to derive the performancemeasures of interest using continuous-time Markov chains, whichscales up to a few slices only, and(ii)tight performance bounds,which are useful to tackle the case of more than a fistful ofslices. We prove the accuracy of the model by comparisonagainst a detailed simulator. Eventually, with our performanceevaluation study, we show that our model is very effective inproviding insight and guidelines for allocation and managementof resources in cells hosting slices for services with differentcharacteristics and performance requirements, such as machinetype communications and human type communications