Handover Management in a Sliced 5G Network Using Deep Reinforcement Learning

Unlike 4G, 5G aspires to offer distinct services such as voice and video (video streaming) communications, ehealth, and vehicular communication while utilizing the same 4G infrastructure. The cutting-edge technology of network slicing into the Radio Access Network (RAN) can be used to achieve this challenging goal. However, mobility management in a sliced Fifth Generation (5G network presents new and complex problems. User mobility in a network-sliced environment needs to be handled across not just various base stations or access methods but also various slices. In particular, base station (BS) physical resource constraints and network slice (NS) logical connection constraints should be taken into account while choosing a handover (HO) strategy. Here, we present a solution to improve network performance in a sliced 5G network by minimizing HO’s number. To this end, we applied a Deep Reinforcement Learning (DRL) method: Proximal Policy Optimization (PPO), a policy gradient reinforcement learning method. The simulation results show that PPO outperforms other algorithms, and the HO’s number is reduced.