The LSTM-Based Advantage Actor-Critic Learning for Resource Management in Network Slicing With User Mobility

Network slicing aims to efficiently provision diversified services with distinct requirements over the same physical infrastructure. Therein, in order to efficiently allocate resources across slices, demand-aware inter-slice resource management is of significant importance. In this letter, we consider a scenario that contains several slices in a radio access network with base stations that share the same physical resources (e.g., bandwidth or slots). We primarily leverage advantage actor-critic (A2C), one typical deep reinforcement learning (DRL) algorithm, to solve this problem by considering the varying service demands as the environment state and the allocated resources as the environment action. However, given that the user mobility toughens the difficulty to perceive the environment, we further incorporate the long short-term memory (LSTM) into A2C, and put forward an LSTM-A2C algorithm to track the user mobility and improve the system utility. We verify the performance of the proposed LSTM-A2C through extensive simulations.