首页|基于DDPG的边云协同计算卸载方法

基于DDPG的边云协同计算卸载方法

扫码查看
移动设备的容量有限以及传统卸载算法仅考虑移动设备和边缘服务器计算资源,使单独的边缘计算面临资源有限和成本高的问题.为此,将云计算(Cloud Computing)、边缘计算(Edge Computing)与深度确定策略性梯度算法(Deep Deterministic Policy Gradient,DDPG)相结合,提出了一种基于DDPG的边云协同计算卸载方法(DDPG-ECC).将时延和能耗作为优化目标,利用边缘服务器和云服务器之间的协作,最小化计算卸载系统的时延和能耗,实现了计算卸载的优化分配.仿真结果表明,DDPG-ECC 性能良好,对于不同的工作负载具有很好的适应性和泛化能力.
Edge-cloud Collaborative Computation Offloading Method Based on DDPG
Mobile devices have limited cupacity,mobile devices and edge server computing were only con-sidered in the traditional offloading algorithms.Edge computing still faces problems of limited resources and high costs.Therefore,an edge-cloud collaborative computing offloading solution(DDPG-ECC)was proposed based on the Deep Deterministic Policy Gradient(DDPG)algorithm.It integrated cloud compu-ting,edge computing,and DDPG.The DDPG-ECC method strategically prioritized minimizing both laten-cy and energy consumption as optimization goals.By fostering collaboration between edge servers and cloud servers,it effectively reduced the latency and energy consumption of the computation offloading sys-tem,achieving an optimized allocation for computation offloading.Simulation results show that DDPG-ECC performs well and it is excellent adaptability and generalization capabilities for different workloads.

edge computingedge-cloud collaborationcomputing offloadingDDPG-ECC

徐炜鹏、李英、李建波

展开 >

青岛大学计算机科学技术学院,青岛 266071

边缘计算 边云协同 计算卸载

国家自然科学基金

61802216

2024

青岛大学学报(自然科学版)
青岛大学

青岛大学学报(自然科学版)

影响因子:0.248
ISSN:1006-1037
年,卷(期):2024.37(1)
  • 20