基于移动边缘计算的多流自适应卸载方案  

Multi-stream adaptive offloading scheme based on mobile edge computing

作　　者：胡叠丽 杨哲铭 纪雯[1,2] HU Dieli;YANG Zheming;JI Wen(Institute of Computing Technology,Chinese Academy of Sciences,Beijing 100190,China;Peng Cheng Laboratory,Shenzhen 518055,China;School of Computer Science and Technology,University of Chinese Academy of Sciences,Beijing 101408,China)

机构地区：[1]中国科学院计算技术研究所,北京100190 [2]鹏城实验室,广东深圳518055 [3]中国科学院大学计算机科学与技术学院,北京101408

出　　处：《电信科学》2024年第11期1-15,共15页Telecommunications Science

基　　金：国家重点研发计划项目(No.2023YFB4502805);国家自然科学基金资助项目(No.62072440);北京市自然科学基金资助项目(No.L221004)。

摘　　要：海量视频流的传输和分析需要大量的带宽和计算资源,这对当前基于移动边缘计算(mobile edge computing,MEC)的卸载方案提出了严峻挑战。对此,提出了一种基于多流协同优化框架的自适应卸载方案。首先,在满足长期MEC能量预算的约束条件下,通过联合优化数据流选择决策、服务器卸载决策、带宽资源分配和计算资源分配来最小化视频任务的处理成本。然后,基于李雅普诺夫优化方法,将长期优化问题转化为每个时隙独立的确定性子问题,并利用马尔可夫近似和KKT条件求解每个时隙的混合整数非线性规划问题。仿真结果表明,所提方案在满足长期MEC能量约束的同时,其成本性能显著优于已有基准研究方案。The transmission and analysis of massive video streams require significant edge bandwidth and computational resources,posing severe challenges to the current multimedia frameworks based on mobile edge computing(MEC).To address this issue,an adaptive offloading scheme based on a multi-stream collaborative optimization framework was proposed.Firstly,under the constraint of the long-term MEC energy budget,the processing cost of video tasks was minimized by optimizing the data stream selection decisions,server offloading decisions,bandwidth resource allocation,and computing resource allocation.Then,based on the Lyapunov optimization method,the longterm optimization problem was transformed into independent deterministic subproblems for each time slot, and themixed-integer nonlinear programming problems for each time slot were solved by Markov approximation and KKTconditions. Simulation results indicate that the proposed scheme not only meets the long-term MEC energy constraint,but also significantly outperforms existing benchmark schemes in terms of cost performance.

关 键 词：移动边缘计算 多流自适应卸载 视频处理 李雅普诺夫优化 

分 类 号：TP393[自动化与计算机技术—计算机应用技术]