强化学习无人机通信系统中的信息年龄优化

Reinforcement Learning-Based Age of Information Optimization in UAV-Enabled Communication System

  • 摘要: 针对6G移动通信系统中信息新鲜度表征和优化问题,提出基于信息年龄的信息新鲜度表征方法,并形成无人机能耗约束下的最小化信息年龄优化问题。而离散的信息年龄优化目标和复杂能耗约束使得非凸优化问题难以求解,因此提出基于强化学习(RL)的无人机轨迹方法。该方法构建与信息年龄相关的奖励函数以快速实现智能化的无人机轨迹决策,从而降低无人机通信系统的信息年龄。仿真结果表明,相比于基准方案能提高8.51%~21.82%的系统信息新鲜度,同时具有更优的收敛性。

     

    Abstract: Aiming at solving the characterization and optimization of information freshness in the sixth generation (6G) communication system, we firstly model information freshness based on the age of information (AoI) in the unmanned aerial vehicle (UAV) communication system and formulate an AoI minimization problem subjected to the energy consumption. However, the nonconvex problem is difficult to solve due to discreteness of AoI optimization and the complicated energy consumption expression. A reinforcement learning-based scheme is proposed to design the UAV’s trajectory, in which the reward function related to AoI is constructed to realize a fast and intelligent UAV trajectory decision, thus reducing the AoI of UAV communication system. The simulation results show that, compared with the benchmark schemes, the proposed trajectory design scheme can improve the information freshness by 8.51%~21.82%. In addition, the proposed scheme has a superior convergence.

     

/

返回文章
返回