Mobility-aware Service Migration in MEC System

• Published in: 2024 International Conference on Information Networking (ICOIN) pp. 653 - 656
• Main Authors: Tran, Tuan Phong, Yoo, Myungsik
• Format: Conference Proceeding
• Language: English
• Published: IEEE 17.01.2024
• Subjects: Deep reinforcement learning; Delays; Energy consumption; Markov decision process; Markov decision processes; MEC; Multi-access edge computing; Predictive models; Recurrent neural network; Recurrent neural networks; service migration
• DOI: 10.1109/ICOIN59985.2024.10572184

Multi-access edge computing (MEC) has emerged as an effective approach for enhancing system quality. Nevertheless, the movement of users and variations in demand for the service might lead to an increase in system delays. This article investigates the issue of service migration, with a particular focus on the factors of mobility and service availability. Specifically, we model the Markov Decision Process (MDP) problem. To make effective service migration decisions, we propose a deep reinforcement learning (DRL) model. Furthermore, a recurrent neural network (RNN) is implemented in order to enhance model performance by predicting user movement. The experimental results demonstrate the effectiveness of the proposed method in reducing the system delay.