Alpha稳定分布噪声和多径干扰下的无人机集群MIMO信号调制识别

Modulation recognition of unmanned aerial vehicle swarm MIMO signals under Alpha stable distribution noise and multipath interference

针对具有多径效应、大气噪声等复杂因素的无人机(unmanned aerial vehicle,UAV)集群多输入多输出(multiple-input multiple-output,MIMO)信道的信号调制方式识别问题,提出基于循环谱特征和高阶累积量特征的调制识别方法。首先,根据UAV集群复杂通信信道环境,建立Alpha稳定分布噪声干扰和多径干扰下的UAV集群MIMO信道。其次,分析MIMO接收信号的高阶累积量特征和循环谱特征,提取出判别调制识别方式能力强的特征值,构造集群信号特征样本。最后,将特征样本输入深度稀疏自编码网络,实现6种调制方式的识别。仿真结果表明,该调制识别方法在UAV集群复杂通信环境下是有效的,当识别准确率为90%时,深度稀疏自编码网络识别性能优于多层感知机识别性能约1 dB。在存在直射径的MIMO多径信道中,当混合信噪比为0 dB时,识别准确率均能达到96%,在低信噪比下有较高的识别准确率,对复杂的信道环境下的MIMO信号识别具有鲁棒性。

Aiming at the problem of signal modulation recognition of multiple-input multiple-output(MIMO)channel of unmanned aerial vehicle(UAV)swarm with multipath effect,atmospheric noise and other interference factors,a modulation recognition method based on cyclic spectral features and high-order cumulant features is proposed.Firstly,according to the characteristics of the complex communication channel of the UAV swarm,the UAV swarm channel with Alpha stable distribution noise interference and multipath interference is established.Secondly,the high-order cumulants and cyclic spectral features of MIMO received signals are analyzed,and the feature values with strong discriminative ability are extracted to construct swarm signal samples.Finally,the samples are fed into the deep sparse autoencoder network to realize recognition of six modulation types.The simulation results show that this modulation recognition method is feasible in complex channel environment of UAV swarm.When the accuracy is 90%,the recognition performance of the deep sparse autoencoder network is about 1 dB better than that of the multilayer perceptron.The accuracy of the method can reach 96% in the MIMO multipath channel with line of sight path when the mixed signal-to-noise ratio is 0 dB,indicating that it has a high recognition accuracy at low signal-to-noise ratio,and it is robust to modulation recognition in complex MIMO communication channels.