Optimization of Quality of AI Service in 6G Native AI Wireless Networks

ELECTRONICS(2023)

引用 0|浏览6
暂无评分
摘要
To comply with the trend of ubiquitous intelligence in 6G, native AI wireless networks are proposed to orchestrate and control communication, computing, data, and AI model resources according to network status, and efficiently provide users with quality-guaranteed AI services. In addition to the quality of communication services, the quality of AI services (QoAISs) includes multiple dimensions, such as AI model accuracy, overhead, and data privacy. This paper proposes a QoAIS optimization method for AI training services in 6G native AI wireless networks. To improve the accuracy and reduce the delay of AI services, we formulate an integer programming problem to obtain proper task scheduling and resource allocation decisions. To quickly obtain decisions that meet the requirements of each dimension of QoAIS, we further transform the single-objective optimization problem into a multi-objective format to facilitate the QoAIS configuration of network protocols. Considering the computational complexity, we propose G-TSRA and NSG-TSRA heuristic algorithms to solve the proposed problems. Finally, the feasibility and performance of QoAIS optimization are verified by simulation.
更多
查看译文
关键词
ai service,optimization
AI 理解论文
溯源树
样例
生成溯源树,研究论文发展脉络
Chat Paper
正在生成论文摘要