On a Cooperative Deep Reinforcement Learning-Based Multi-Objective Routing Strategy for Diversified 6 G Metaverse Services

Metaverse has been widely recognized as an important 6 G application with increasingly stringent, detailed, and diversified requirements for multiple Quality of Service (QoS) metrics. However, traditional routing strategies are usually based on unified weights and select the paths independently, which neglects the service diversity and resource constraints. In this paper, we focus on the diversified metaverse service requirements and propose the Deep Reinforcement Learning (DRL) based multiple objective routing strategy for different services. Multiple agents in the DRL model cooperatively select the paths to improve the resource utilization efficiency. Simulation results illustrate that our proposal outperforms traditional strategies.