Cloud-based convergence of heterogeneous RANs in 5G disaggregated architectures

Cloud-RAN based architectures are widely considered a fundamental part of 5G networks. As a consequence, in the upcoming standards for 5G RAN, disaggregating the RAN functionality between a Central Unit (CU) and multiple Distributed Units (DUs) is considered, addressing the splitting of the 5G protocol stack at the PDCP/RLC point. This split is expected to bring numerous advantages to mobile network operators, as through the isolation of the stack from the PDCP layer and upwards, the CU will be able to act as the Cloud-based convergence point among multiple heterogeneous technologies in the provisioned networks and hence able to serve multiple heterogeneous DUs. Moreover, data rate requirements for this type of split are not very demanding, thus allowing the IPbased transferring of data from the DU to CU and vice-versa.

In this work, we propose, implement and evaluate a protocol for a Cloud-RAN based architecture allowing the selection and dynamic switching of different heterogeneous networks in the RAN. We rely on the open source OpenAirInterface platform and

extend it to support data plane splitting of the LTE functionality, and the subsequent data injection to WiFi networks. We evaluate the platform using a real network setup, under several scenarios of network selection and different delay settings.