基于深度神经网络的多传感器联合信号检测方法

针对多传感器分布式接收中的信号检测问题，提出了一种最小错误概率准则下的联合检测方法。所提方法采用分布式软信息融合处理策略，将多传感器信号检测视为二元假设检验，借助深度神经网络优异的函数逼近能力，在对神经网络结构、目标函数和网络输入输出进行分析基础上，给出了基于深度神经网络的假设检验后验概率求解方法。各独立接收单元利用深度神经网络估计信号有无两种假设的后验概率，然后送入融合中心，计算联合后验概率分布，并做出判决。与传统处理过程依赖严密的数学推导不同，所提方法参数解析和特征提取无需人工解算。最后，通过仿真实验对所提方法有效性进行了验证，并与现有方法进行了对比。结果表明，所提方法能够实现多个传感器信号有效融合，随着接收单元数目增加，能够显著提升信号检测概率，并降低虚警概率；与当前典型的S/K融合方法相比，所提方法在低信噪比下具有明显优势。 

Multi-sensor Joint Signal Detection Method Based on Deep Neural Network

? ?Aiming at the problem of signal detection in multi-sensor distributed reception， a joint detection method under the minimum error probability criterion is proposed. The proposed method adopts the distributed soft information fusion processing strategy， and regards multi-sensor signal detection as binary hypothesis test. With the help of the excellent function approximation ability of deep neural network（DNN）， based on the analysis of neural network structure， objective function and network input and output， a posteriori probability solution method of hypothesis test based on DNN is given. Each independent receiving unit uses the DNN to estimate the posterior probability of the signal with or without two assumptions， and then sends it to the fusion center to calculate the joint posterior probability distribution and make a decision. Different from the strict mathematical derivation of the traditional processing process， the parameter analysis and feature extraction of the proposed method do not need to be solved manually. Finally， the effectiveness of the proposed method is verified by simulation experiments， and compared with the existing methods. The results show that the proposed method can effectively fuse multiple sensor signals. With the increase of the number of receiving units， the signal detection probability can be significantly improved and the false alarm probability can be reduced. Compared with the current typical S/ K fusion methods， the proposed method has obvious advantages at low signal-to-noise ratio （SNR） values.