Deep-Q-Network-Based Multimedia Multi-Service QoS Optimization for Mobile Edge Computing Systems

Boren Guo,Yaxin Wang,Hongwen Yang,Xin Zhang

doi:10.1109/access.2019.2951219

Abstract

By offloading storage and computing resources to the edge of networks, mobile edge computing (MEC) is emerged as a promising architecture to reduce the transmission delay and bandwidth waste for mobile multimedia services. This paper focuses on a multi-service scenario in the MEC systems, where the MEC server can provide three multimedia services including live streaming, buffered streaming and low latency enhanced mobile broadband applications for edge users at the same time. In order to satisfy various quality of service (QoS) requirements for different multimedia applications, the 5G QoS model is applied. Notably, the packets from the multimedia applications with the same or similar requirements are mapped into the same QoS flow, and each QoS flow is processed individually. Therefore, how to effectively schedule the limited radio resource for QoS flows is an intractable problem. To address the problem above, a QoS evaluation model is designed, and a QoS maximization problem is formulated. Furthermore, a deep reinforcement learning method, deep-Q-network, is adopted to make decisions to allocate radio resource dynamically. Compared with round-robin and priority-based scheduling algorithms, the simulation results validate that the proposed algorithm outperforms other resource scheduling algorithms for multi-service scenario.

Highlights

Triggered by the rapid advance of diverse smart devices, the mobile data traffic is experiencing a tremendous growth
DQN algorithm is considered as a deep reinforcement learning (DRL) algorithm, which focuses on how to interact with the environment to attain maximum cumulative reward
SCENARIO CONFIGURATION To evaluate the performance of the proposed DQN based resource allocation algorithm, simulations are performed on the Python platform

Summary

INTRODUCTION

Triggered by the rapid advance of diverse smart devices, the mobile data traffic is experiencing a tremendous growth. Various multimedia applications from massive mobile devices pose diverse requirements (e.g., ultra-low latency, high bitrate, etc.) on Radio Access Network (RAN). Based the QoS evaluation model, the resource allocation problem above can be formulated as a multimedia multi-service QoS optimization problem. The simulation results validate that the proposed DQN based algorithm can efficiently allocate RAN resource to different QoS flows to meet diverse QoS requirements, especially in high bitrate scenarios. To the best of our knowledge, this paper is the first research which focuses on the 5G QoS model [29], where the packets from various applications are mapped to different QoS flows with different QoS characteristics.

RELATED WORKS

TRAFFIC MODEL

DEEP-Q-NETWORK

DRL FRAMEWORK DESIGN

DEEP-Q-NETWORK BASED RESOURCE ALLOCATION ALGORITHM

CONCLUSION AND FUTURE WORKS