6G Radio Access Implementation: Challenges and Technologies

Abstract

Open radio access network (RAN) solutions are defined to be RAN implementations constructed from multiple vendor intellectual property by a system integrator (SI). They allow optimization during construction by the SI to meet specific, local network needs. This is an extremely attractive goal for operators and enterprise who wish to differentiate in power consumption of equipment cost or who may have specific needs for latency for certain types of traffic. This differentiation will vary from deployment to deployment and from network endpoint to endpoint. The O-RAN Alliance has focused on developing interfaces to allow multivendor interoperability as well as a RAN intelligent controller (RIC) that can control the use of the distributed and centralized unit (DU and CU) components of O-RAN. However, this effort has not achieved a dramatic shift to open RAN solution so far, and we argue that there is a critical need for a framework that supports automation of the construction of the RAN, within which the definition of the vendor components can also be modified at the SI level. Without this, hand optimization by the SI for each network deployment will be crushingly expensive. For 6G we see machine learning (ML) increasing the options and flexibility of the RAN dramatically and also allowing the RAN functionality to adapt at a finer granular level after deployment. 6G’s challenge to RAN deployment is therefore that the construction of the RAN within the network will constantly evolve at a local level. Mapping functionality to hardware and software while achieving real-time latency, synchronization, and power goals with fixed installed equipment will itself require automation and ML. In this chapter, we describe the problem of RAN deployment in some detail and describe how O-RAN must be expanded to allow automation and ML to be successfully applied for 6G.