基于DNN的无人机数据OFDM传输技术

针对无人机信道含多径、多普勒频移,还易受到外来干扰和高功率放大带来的非线性失真影响的问题,提出一种基于深度神经网络(deep neural network,DNN)的无人机正交频分复用(orthogonal frequency divi-sion multiplexing,OFDM)数据传输技术.下行链路采用OFDM系...

OFDM data transmission technology of UAV based on deep neural network

The unmmand aerial vehicle(UAV)channel contains multipath,Doppler frequency shift.It is easily affected by external interference and nonlinear distortion caused by high power amplification.To solve those problems a data transmission technology of UAV orthogonal frequency division multiplexing(OFDM)based on deep neural network(DNN)is proposed.In the downlink,OFDM system is used.After demodulation and borrowing at the receiver,the least square(LS)channel estimation and zero forcing(ZF)algorithm are used for preliminary signal detection,and then input into the DNN composed of BiLSTM and full connected layer for channel estimation and signal detection optimization,and data stream recovery.Simulation results show that compared with the traditional interpolation method,the proposed method has obvious advantages in channel estimation performance under three states of UAV takeoff and landing,flight and hovering,and the bit error rate performance is improved by at least one order of magnitude;under the influence of nonlinear distortion and external interference,the proposed method still has significant performance advantages,which not only simplifies the processing module of nonlinear distortion and interference,but also improves the stability of the system qualitative.