基于RF-GBDT的无人机空对地信道参数预测算法  

A Parameter Prediction Algorithm for UAV Air-to-ground Channel Based on RF-GBDT

在线阅读下载全文

作  者:苏覃 江虹[1] 汪文斌 SU Qin;JIANG Hong;WANG Wenbin(School of Information Engineering,Southwest University of Science and Technology,Mianyang 621010,China)

机构地区:[1]西南科技大学信息工程学院,四川绵阳621010

出  处:《火力与指挥控制》2025年第3期26-33,40,共9页Fire Control & Command Control

基  金:国家自然科学基金(61771410);四川省科技厅基金资助项目(2023NSFSC1373)。

摘  要:针对经验模型实用性低,传统射线跟踪算法复杂度高、仿真效率低的问题,提出一种基于随机森林(RF)-梯度提升(GBDT)的无人机(UAV)空对地信道参数预测算法。RF-GBDT算法通过采用集成学习思想,构建多个模块,考虑收发机位置、发射频率与建筑物覆盖率,引入残差序列进行模型训练与超参数优化,以实现高效、准确及高泛化性的信道参数预测。实验结果表明,与传统RF模型和GBDT模型相比,RF-GBDT模型具有更小的时间复杂度、更高的预测精度、更快的收敛速度、更低的收敛误差,RF-GBDT模型预测RMSE至少降低了21.8%,收敛RMSE至少降低了27.6%,有利于通信系统的设计、部署与优化。Aiming at the inefficiency of empirical model,high complexity and low simulation efficiency associated with traditional ray-tracing algorithm,a parameter prediction algorithm for Unmanned Aerial Vehicle(UAV)air-to-ground channels based on Random Forest(RF)-Gradient Boosting(GBDT)is proposed.The RF-GBDT algorithm utilizes integrated learning ideas to build multiple modules,the transmitter and receiver positions,transmission frequency,and building coverage rate are taken into account.Residual sequences are introduced for model training and hyperparameter optimization to achieve efficient,accurate,and highly generalized channel parameter prediction.The experimental results show that compared with traditional RF models and GBDT models,the RF-GBDT model has smaller time complexity,higher prediction accuracy,faster convergence speed,and lower convergence errors.The RF-GBDT model predicts an RMSE reduction of at least 21.8%and converges an RMSE reduction of at least 27.6%,which is beneficial for the design,deployment,and optimization of communication systems.

关 键 词:射线跟踪 UAV 空对地信道参数预测 RF-GBDT 集成学习 

分 类 号:E96[军事—军事通信学] TN911[电子电信—通信与信息系统] TP391[电子电信—信息与通信工程]

 

参考文献:

正在载入数据...

 

二级参考文献:

正在载入数据...

 

耦合文献:

正在载入数据...

 

引证文献:

正在载入数据...

 

二级引证文献:

正在载入数据...

 

同被引文献:

正在载入数据...

 

相关期刊文献:

正在载入数据...

相关的主题
相关的作者对象
相关的机构对象