无人机系统中基于能量效率的资源分配研究

Research on Resource Allocation Based on Energy Efficiency in UAV System

无人机作为一种新的通信基础设施,引起了学术界和工业界的广泛关注.由于其在空中部署的特点,广泛应用于军事侦查、监控、城市交通管理等场景.然而,无人机的部署面临以下挑战:①由于无人机信道的广播特性,无人机与地面通信时容易被窃听;②由于无人机采用蓄电池供电,其续航能力和传输数据量受限于电池容量.本文采用物理层安全容量来描述无人机通信速率,为了降低数据被窃听的风险,合法接收者即宏基站将辅助发射人工干扰噪声,以降低窃听者的通信信道质量;通过联合控制无人机的发射功率和宏基站的干扰功率,以最大化无人机系统的能量效率;考虑到无人机信道的动态特性,本文将此功率控制问题建模为马尔科夫决策过程,并采用深度Q学习网络算法来获得最优的功率策略.Python仿真实验证明:本文所提出算法的平均回报值和收敛性能都优于Policy Gradient算法.

Unmanned aerial vehicle as a new communication base infrastructure has caused great attentions in academia and industry,which is widely used in military reconnaissance,monitoring,and public traffic management scenarios.However,the deployment of UAV meets following challenges:(1)since the channel between the UAV and the legitimate receiver is public,the communication data between them is easy to be intercepted by eavesdroppers;(2)the flight endurance and communication is confined by the limitation of battery of the UAV.The study employs physical layer security capacity to measure the rate of the UAV.Furthermore,in order to decrease the risk of data intercepting,the legitimate receiver,i.e.,macro base station,transmits the artificial noise to reduce the quality of the eavesdropper's communication channel.The study jointly optimizes the transmit power of the UAV and the MBS to maximize the energy efficiency of the UAV system.Considering the dynamic characteristics of the wireless channel of the UAV,the above optimization problem is modeled as a Markov decision process,and we utilize deep Q-learning network to search the optimal power strategy.Python simulation results demonstrate that the proposed algorithm outperforms the Policy Gradient algorithm in terms of average rewards and convergence performance.