Table 13.

Research opportunities for ML applications in O-RAN.

	Category	Issues
Section 4.1	Proactive Maintenance	- How to evolve the O-RAN framework in conjunction with ML system design; - To develop ML training approaches across the network irrespective of equipment vendor or site location (i.e., different configurations of multiple vendor equipment and unharmonized data across multiple sites).
Section 4.2	xApps, rApps, dApps Operation	- Orchestration of xApps, rApps, and dApps in the O-RAN RIC when they are simultaneously operated for network automation; - Orchestration across the domain with SON functions in a core network and any newly added xApps, rApps, and dApps to the RAN.
Section 4.3	Satellite NTN	- To optimize the resource allocation for capacity enhancement; - To mitigate network interference sources such as adjacent satellite interference and inter-system interference.
Section 4.4	Massive MIMO	- To provide right interfaces to integrate multi-antenna processing in O-RAN to maximize the spectral efficiency; - To dynamically adjust the level of coordination/cooperation between DUs and to efficiently perform the RU clustering; - To effectively distribute the channel state information between the split baseband functions.
Section 4.5	Mobility Management	- To jointly optimize the trajectory of UAV and the task offloading among diverse O-RAN elements.
Section 4.6	Network Management	- To optimize the flexible functional split of RAN slices dynamically to respond to changing network environments.
Section 4.7	Data Privacy Security	- To make ML models to access and utilize the data without compromising user privacy; - To secure ML models against adversarial attacks and to develop the measure indicating protection against potential data breaches or cyberattacks.
Section 4.8	Big Data Collection for ML	- To improve data collection by using O-RAN, including collection and consolidation of hybrid empirical and synthetic data.