.png){kind=icon}
Return
Collaborative Vehicular Edge Cloud Computing Task Offloading Optimization Scheme Based on Deep Reinforcement Learning
.png){kind=icon}
PRE
AI
DOI：10.1007/978-3-031-93251-9_12.png)
.png){kind=icon}
Original request for help.png){kind=icon}
Share.png){kind=icon}
Save
Abstract
En 中文
Edge server computing resource limitations are often ignored in traditional vehicular edge computing, failing to fully utilise the computing power of vehicles, edge servers and cloud servers. At the same time, data usage efficiency and exploration efficiency in complex environments are not fully considered when using deep reinforcement learning to solve the vehicle task offloading problem. Therefore, this paper proposes an offloading optimisation scheme for vehicular edge cloud computing based on Distributed Distributional Deep Deterministic Policy Gradient (D4PG). Aiming at the shortage of edge server resources, this paper constructs a framework of Vehicular Edge Cloud Computing (VECC) system, which makes full use of the computational resources of vehicles, edge servers, and clouds to perform vehicular tasks. Aiming at the problems of data usage efficiency and environment exploration efficiency of DRL in complex environments, this paper proposes a computational offloading optimisation algorithm based on D4PG, which makes use of distributed computational target value distribution to effectively use sampled data to improve learning efficiency. By introducing the value distribution, D4PG is able to better explore the environment and effectively utilise the experience to improve the exploration efficiency of the complex environment. In this study, the latency and energy consumption of the system are integrated into a cost function with the goal of minimising the system cost. Experimental results show that the method proposed in this paper outperforms traditional task offloading schemes in terms of stability performance.
Keywords:
Mobile Edge Computing
Mobile Cloud Computing
Internet of Vehicles
Task Offloading
Deep Reinforcement Learning
Journal
C
IF:
0
Papers:
18
Citations:
0
Organization
Cited Papers
.png){kind=icon}
ShareSaveShareSaveDeep Reinforcement Learning and Markov Decision Problem for Task Offloading in Mobile Edge Computing
PREAI
ShareSaveShareSaveShareSaveShareSave
View more