Service-Driven Resource Management in Vehicular Networks Based on Deep Reinforcement Learning

This paper studies a joint communication, computing and caching resource allocation problem in vehicular networks to improve user satisfaction and reduce costs. We propose a double-scale deep reinforcement learning (DSDRL) framework that combines on-policy strategy and off-policy strategy to enable...

Full description

Saved in:
Bibliographic Details
Published in2020 IEEE 31st Annual International Symposium on Personal, Indoor and Mobile Radio Communications pp. 1 - 6
Main Authors Lyu, Zhengwei, Wang, Ying, Liu, Man, Chen, Yuanbin
Format Conference Proceeding
LanguageEnglish
Published IEEE 01.08.2020
Subjects
caching
communication
computing
Delays
double-scale deep reinforcement learning(DSDRL)
Dynamic scheduling
Machine learning
Optimization
Resource management
Task analysis
Vehicle dynamics
Vehicular Network
Online AccessGet full text

Cover

More Information
Summary:This paper studies a joint communication, computing and caching resource allocation problem in vehicular networks to improve user satisfaction and reduce costs. We propose a double-scale deep reinforcement learning (DSDRL) framework that combines on-policy strategy and off-policy strategy to enable dynamic resources allocation, which considers not only the diversity and difference of services but also the costs of network operators. Simulation results show that the proposed scheme can effectively improve the long-term revenue of network operators.
ISSN:2166-9589
DOI:10.1109/PIMRC48278.2020.9217216