基于深度强化学习的无人机小区间干扰抑制技术研究  

Research on UAV Intercell Interference Suppression Based on Deep Reinforcement Learning

作　　者：黎海涛 吕鑫 张帅 黄嘉伟 LI Haitao;LYU Xin;ZHANG Shuai;HUANG Jiawei(Faculty of Information Technology,Beijing University of Technology,Beijing 100124,China)

机构地区：[1]北京工业大学信息学部,北京100124

出　　处：《无线电通信技术》2023年第2期331-337,共7页Radio Communications Technology

基　　金：航空科学基金(2018ZC15003)。

摘　　要：无人机(UAV)辅助蜂窝网络的空中基站工作在频谱高度拥挤的场景中,会造成严重空中小区间干扰(Inter-Cell Interference, ICI)而大大降低网络性能。为解决该问题,研究了基于深度强化学习的无人机辅助蜂窝网络小区间干扰抑制技术。首先建立了无人机辅助蜂窝网络中基于联合波束成形与功率控制(Joint Beamforming and Power Control, JBPC)的抗干扰优化模型。然后提出了基于置信区间上界(Upper Confidence Bound, UCB) Dueling深度Q网络(Deep Q Network, DQN)的深度强化学习求解算法,并利用该算法设计了联合波束成形和功率控制的ICI抑制技术。仿真结果表明,基于UCB Dueling DQN学习的JBPC干扰抑制技术的收敛性优于DQN和Dueling DQN算法,且能达到穷举法的最优容量,有利于提高无人机辅助蜂窝网络性能。In UAV assisted cellular network,due to multiple aerial base stations working in a highly congested spectrum scenario,serious aerial Inter-Cell Interference(ICI)will occur,which greatly reduces network performance.In order to solve this problem,this paper studies the ICI suppression technology of UAV assisted cellular network based on deep reinforcement learning.First,an optimiza-tion model of Joint Beamforming and Power Control(JBPC)in UAV assisted cellular networks is established.Then,a deep reinforce-ment learning algorithm based on the Upper Confidence Bound(UCB)Dueling Deep Q Network(DQN)is proposed.We utilize the proposed learning algorithm to design the intercell interference suppression technology based on joint beamforming and power control.Simulation results show that the convergence performance of JBPC interference suppression technology using UCB Dueling DQN learning is better than general DQN and Dueling DQN algorithms,which is helpful to improve the capacity of UAV assisted cellular network.

关 键 词：无人机辅助蜂窝网络 深度强化学习 置信区间上界 小区间干扰抑制 

分 类 号：TN929.5[电子电信—通信与信息系统]