首页|Time-Varying Channel Estimation Based on Air-Ground Channel Modelling and Modulated Learning Networks
Time-Varying Channel Estimation Based on Air-Ground Channel Modelling and Modulated Learning Networks
扫码查看
点击上方二维码区域，可以放大扫码查看
原文链接
万方数据
To improve the time-varying channel es-timation accuracy of orthogonal frequency division multi-plexing air-ground datalink in complex environment,this paper proposes a time-varying air-ground channel estima-tion algorithm based on the modulated learning networks,termed as MB-ChanEst-TV.The algorithm integrates the modulated convolutional neural networks(MCNN)with the bidirectional long short term memory(Bi-LSTM),where the MCNN subnetworks accomplish channel inter-polation in frequency domain and compress the network model while the Bi-LSTM subnetworks achieve channel prediction in time domain.Considering the unique char-acteristics of airframe shadowing for unmanned aircraft systems,we propose to combine the classical 2-ray chan-nel model with the tapped delay line model and present a more realistic channel impulse response samples genera-tion approach,whose code and dataset have been made publicly available.We demonstrate the effectiveness of our proposed approach on the generated dataset,where experimental results indicate that the MB-ChanEst-TV model outperforms existing state-of-the-art methods with a lower estimation error and better bit error ratio per-formance under different signal to noise ratio conditions.We also analyze the effect of roll angle of the aircraft and the duration percentage of the airframe shadow on the channel estimation.
万方数据
