Intelligent Dynamic Resource Allocation for Network Slicing Based on Actor-Critic Algorithm

Network slicing has been widely acknowledged as a promising architectural technology to meet the different service requirements of diversified applications for future mobile networks. Since there are significant dynamics in traffic and resource availability, resource isolation must be realized throu...

Full description

Saved in:
Bibliographic Details
Main Authors: Yang, Bei, Li, Qiyao, She, Xiaoming, Zhu, Jianchi, Wei, Fengsheng, Chen, Peng, Wang, Jianxiu
Format: Conference Proceeding
Language:English
Subjects:
AC algorithm
Dynamic scheduling
dynamic slice resources allocation
Heuristic algorithms
Markov processes
Network slicing
Quality of service
Real-time systems
Reinforcement learning
user satisfaction
Online Access:Request full text
Tags: Add Tag
No Tags, Be the first to tag this record!
Description
Summary:Network slicing has been widely acknowledged as a promising architectural technology to meet the different service requirements of diversified applications for future mobile networks. Since there are significant dynamics in traffic and resource availability, resource isolation must be realized through agile re-provisioning of slices, to guarantee the Quality of Service (QoS) of individual network slices. We thus need to dynamically adjust the amount of resources allocated to different data flows carried in a slice. In this paper, we investigate the dynamic resource allocation mechanism in a slice with the aim of improving the user's service quality. We first model the dynamic resources allocation problem as a Markov Decision Process (MDP), and employ Actor-Critic (AC) algorithm, a specific reinforcement learning algorithm, to pursue the long-term optimal resources allocation. We conduct simulations to evaluate the performance of our proposed algorithm and the numerical results demonstrate its effectiveness.
ISSN:2770-1603
DOI:10.1109/ICAIT56197.2022.9862788