%0 Journal Article
%A LU Zhao-ming
%A ZHANG Hai-jun
%A GUAN Wan-qing
%T Intelligent Resource Allocation Algorithm for 6G Multi-Tenant Network Slicing Based on Deep Reinforcement Learning
%D 2020
%R 10.13190/j.jbupt.2020-211
%J Journal of Beijing University of Posts and Telecommunications
%P 132-139
%V 43
%N 6
%X In the future, the sixth generation of mobile communications system (6G) network services merge reality and virtual reality, and support real-time interaction. It is urgent to quickly match the personalized service requirements of multiple tenants, therefore a two-layer hierarchical intelligent management scheme for network slicing is proposed, including the global resource manager at the upper level and the local resource managers for different tenants at the lower level. Firstly, based on the real-time status description of end-to-end slice, a service quality evaluation model is established considering the difference of multi-type slice requests from different tenants. With the service quality feedback, deep reinforcement learning (DRL) algorithm is adopted to optimize the global resource allocation and local resource adjustment. Hence, utilization efficiency of multi-dimensional resources in different domains are improved and tenants are able to customize resource usage. The simulation results show that the proposed scheme can optimize the long-term revenue of resource providers while guaranteeing the service quality.
%U https://journal.bupt.edu.cn/EN/10.13190/j.jbupt.2020-211