In Chap. 7 , we used DDPG as an RL approach, where it showed a performance close to the NI, PSO-based solution, while reducing the runtime by 31.5% in dynamic environments, by taking advantage of transfer learning. Although the runtime reduction is substantial, this might not be sufficient in dynamic scenarios, where users are constantly on the move, and the UAV needs to change its location frequently. This brings us to introduce a novel low-complexity DDPG codebook-based approach for the UAV deployment problem.

错误:搜索内容不能为空,请输入英文关键词
错误:关键词超出字数限制,请精简
高级检索

A Codebook-Based DRL Approach for UAV Deployment and Trajectory Design

  • Tho Le-Ngoc,
  • MohammadMahdi Ghadaksaz,
  • Mobeen Mahmood

摘要

In Chap. 7 , we used DDPG as an RL approach, where it showed a performance close to the NI, PSO-based solution, while reducing the runtime by 31.5% in dynamic environments, by taking advantage of transfer learning. Although the runtime reduction is substantial, this might not be sufficient in dynamic scenarios, where users are constantly on the move, and the UAV needs to change its location frequently. This brings us to introduce a novel low-complexity DDPG codebook-based approach for the UAV deployment problem.