Deep Reinforcement Learning Based Multi-Access Edge Computing Schedule for Internet of Vehicle

TL;DR

This paper introduces a multi-agent deep reinforcement learning approach using UAVs to enhance wireless network services for Internet of Vehicles, improving QoE through collaborative policy learning.

Contribution

It proposes the M-AGCDRL algorithm that combines local observations with global maps and uses graph attention networks for effective multi-agent cooperation.

Findings

M-AGCDRL improves QoE for IoVs in simulations.

The method outperforms existing approaches in network performance.

Collaborative learning enhances policy effectiveness.

Abstract

As intelligent transportation systems been implemented broadly and unmanned arial vehicles (UAVs) can assist terrestrial base stations acting as multi-access edge computing (MEC) to provide a better wireless network communication for Internet of Vehicles (IoVs), we propose a UAVs-assisted approach to help provide a better wireless network service retaining the maximum Quality of Experience(QoE) of the IoVs on the lane. In the paper, we present a Multi-Agent Graph Convolutional Deep Reinforcement Learning (M-AGCDRL) algorithm which combines local observations of each agent with a low-resolution global map as input to learn a policy for each agent. The agents can share their information with others in graph attention networks, resulting in an effective joint policy. Simulation results show that the M-AGCDRL method enables a better QoE of IoTs and achieves good performance.