Slice Admission and Deployment Strategies in Resource-Constrained 5G Network Slices using an Actor-Critic Approach

The advent of 5G mobile communications networks has enabled the emergence of various services and applications. However, the diverse and complex requirements of these new services and applications have further intensified the recurring problem in network resource management for mobile network operators. As such, effective and intelligent resource management methods have become more crucial. Therefore, in this work, we propose an Actor-Critic Slice Admission Approach to 5G Network Slicing with the aim of improving the quality of service and user experience in 5G networks. The proposed approach implements reinforcement learning to create network slicing decisions for the admission network slice requests. In addition, slice deployment policies were formulated to ensure the efficient allocation of network resources based on the three 5G usage scenarios. The simulation has shown favorable results based on the network's resource efficiency, acceptance rate, and utilization.