Deep Reinforcement Learning Based xApp for RAN Slice Management Using OpenAirInterface

Network slicing is a key enabler for providing differentiated service support to heterogeneous use cases and applications in 5G and beyond networks through creating multiple logical slices. Resource management for satisfying diverse requirements of slices is a highly challenging task under time-varying traffic and wireless channel conditions. This paper presents a deep reinforcement learning (DRL) based xApp for dynamically managing the service level agreements (SLAs) of network slices, eliminating the need for human-in-the-loop decision-making. The proposed xApp is implemented within an open source mobile network emulator to create an O-RAN compliant end-to-end 5G network capable of dynamic resource management capabilities. The intelligent resource management xApp operates on the RAN Intelligent Controller (RIC), enabling monitoring and dynamic resource control of the gNodeB through the E2 interface. The proof-of-concept experiment results demonstrate that the trained DRL model deployed on the RIC platform is successfully utilized to manage the key performance indicators of the network slices.