Table 6.

Summary of current efforts on ML applications for dynamic function splitting in O-RAN.

Year	Ref	Contribution
2021	[41]	Applied RL to dynamically perform function split decisions for DUs and CU in a virtualized O-RAN architecture.
2023	[42]	Applied DRL to propose an energy-efficient RAN disaggregation and virtualization approach across edge sites, including DUs, and a cloud site, including CUs.
2022	[43]	Proposed a DRL-based algorithm to jointly solve the optimal placement of network functions between the CU, DU, and user RU in an O-RAN architecture.
2023	[59]	Proposed a DRL-based algorithm for a multi-objective optimization to minimize computational costs and the overhead associated with periodically reconfiguring dynamic VNFs splitting.