Energy-Efficient Anti-Jamming Metaverse Resource Allocation Based on Reinforcement Learning
Yiwen Zhan, Jinming Zhang, Zhiping Lin, Liang Xiao · 2024
Metaverse system facilitated by Extended Reality (XR) requires extensive computing and communication resources to provide seamless services for users and has to resist jamming attacks. In this paper, we propose a reinforcement learning based Metaverse resource allocation scheme against jamming, which optimizes the rendering mode, transmit channel, and power to fulfill the quality of experience (QoE) requirements for Metaverse services. This scheme incorporates both the background resolution, determined by the linear model of visual acuity decline, and the background correlation into the state formulation and rendering process, with both factors influencing the data size of the rendering task. Based on the required resolution, the background resolution, the data size of foreground and background, the background correlation, and the radio channel gain, the 2-level hierarchical architecture evaluates the expected utility of the rendering policy in the first level, and the transmission policy in the second level, and estimates the risk value depending on the latency to mitigate the risk of user experience disruption. Simulation results show that our proposed scheme reduces the service latency, conserves energy consumption, and enhances the QoE of Metaverse users.