An Energy Aware Algorithm for Edge Task Offloading

• DOI: 10.32604/iasc.2022.018881
• Publication Date: 2021-10-10T09:22:15Z
• Authors: Ao Xiong; Meng Chen; Shaoyong Guo; Yongjie Li; Yujing Zhao; Qinghai Ou; Chuan Liu; Siwen Xu; Xiangang Liu
• Keywords: Mobile Edge Computing

To solve the problem of energy consumption optimization edge servers in process task unloading, we propose a unloading algorithm based on reinforcement learning this paper. The observes and analyzes current environment state, selects deployment location tasks according to states, realizes oriented optimization. achieve above goals, first construct network model including servers' link transmission consumption, which improves accuracy evaluation. Because complexity variability environment, paper designs Proximal Policy Optimization (PPO), besides use Dijkstra determine connection path between where adjacent are deployed. Finally, lots simulation experiments verify effectiveness proposed method unloading. Compared with contrast algorithms, average saving can reach 22.69%.