Fronthaul Modelling for 5G Systems

Abstract

5G networks aim to provide reliable, fast and service orientated communications to its users. One of its most attractive features, is its random access network (RAN) virtualization technologies for different vertical industries that can tailor the spectrum allocation based on the performance needs of the customers. This service pliancy brings new architectural challenges that are still being tackled both in the industry and academia. One of such, is finding a reliable, fast, flexible and cost effective fronthaul communication transport protocol via fiber optic links. Current network architectures favor Ethernet as a transport mechanism to meet the demands of fronthaul transport due to its upgradability to higher data rates, network switching, scalability and wide availability in the market. Furthermore, eCPRI and oRAN provide a set of layers that define encapsulating procedures for user and control plane data that is carried inside Ethernet frames.

This thesis work aims to build a model of a Nokia 5G subsystem IP that processes eCPRI and oRAN encapsulated IQ samples and is transported through Ethernet. The model was aimed to replicate the same processing procedures, configuration and register definitions as is done in the real implementation in the ASIC/FPGA. The motivation behind building the model is to recreate the ideal behaviour of the subsystem in order to have a tool that allows system designers and architects to test new features and algorithms, which can greatly reduce costs and time, and also provide verification engineers with a tool to generate ideal responses of the system when testing the real physical implementation of the system. Furthermore, the model can be integrated with other available subsystem IP models and thus have an orchestrated ideal behaviour of a complete system module.