Multi-agent reinforcement learning based 5G bi-level multi-slice resource allocation

With the complexity of application scenarios and the wide application of network slicing, the base station faces different requirements of different slicing in resource allocation. The base station can separate the centralized unit (CU)-distributed unit (DU) according to the slicing demand and carry out different charges according to different separation schemes. We build a more efficient bi-level model for resource allocation in the context of CU-DU separation, and for practical consideration, we use the deterministic reinforcement learning algorithm to solve this bi-level model. The results show that the reinforcement learning method is effective in solving the bi-level resource allocation model.