ID	原文	译文
53967	针对免调度非正交多址接入(Non-Orthogonal Multiple Access，NOMA)系统，多用户传输场景的上行信道估计(Channel Estimation，CE)与活动用户检测(Active User Detection，AUD)问题可被建模为压缩感知重建问题。	For scheduling-free non-orthogonal multiple access(NOMA) systems, the uplink channel estimation(CE) and active user detection(AUD) issues in multi-user transmission scenarios can be modeled as the reconstruction problem of compressed sensing.
53968	本文提出了一种改进的近似消息传递(Approximate Message Passing，AMP)算法——阈值自适应-加约束重加权-近似消息传递(Threshold Adaptive Constrained Reweighted Approximate Message Passing，TA-CR-AMP)算法来联合解决CE和AUD问题。	In view of the shortcomings of the existing compressed sensing algorithms, this paper proposes an improved Approximate Message Passing(AMP) algorithm-Threshold Adaptive Constrained Reweighted Approximate Message Passing(TA-CR-AMP) algorithm to combine Solve CE and AUD problems.
53969	该算法在合适的迭代终止准则下，对AMP算法加入更新稀疏信号稀疏结构的操作，在此基础上对算法引入加约束的重加权，并令阈值自适应变化。	The algorithm adds the operation of updating the sparse structure of the sparse signal to the AMP algorithm under a suitable iteration termination criterion, and on this basis introduces a constrained heavy weighting to the algorithm, and makes the threshold adaptively change.
53970	仿真结果表明，与AMP算法相比，本文提出的算法以较低的复杂度获得了更加优越的信道估计和活跃用户检测性能。	The simulation results show that, compared with the AMP algorithm, the algorithm proposed in this paper obtains more superior channel estimation and active user detection performance with lower complexity.
53971	另外，本算法获得了与更加复杂的期望最大-贝叶斯AMP(Expectation Maximization Bayesian Approximate Message Passing，EM-B-AMP)算法相近的性能。	In addition, the algorithm proposed in this paper has achieved similar performance to the more complex Expectation Maximization Bayesian Approximate Message Passing(EM-B-AMP) algorithm.
53972	针对现有基于深度学习理论的信号智能检测方法大多只能对单信号或时频域不重叠的信号进行检测，本文提出了一种基于掩膜区域卷积神经网络(Mask R-CNN)与Criminisi算法的时频重叠多信号智能检测新方法。	In view of the existing intelligent signal detection methods based on deep learning theory, most of them can only detect single signal or signals which don't overlap in time-frequency domains. This paper proposes a new intelligent detection method based on Mask R-CNN and Criminisi algorithm for time-frequency overlapping multi-signals.
53973	首先将一维时域信号通过时频变换得到二维时频图像。	First, the signal in the time domain is transformed into a time-frequency image.
53974	然后针对时频图中多信号重叠部分像素位置信息缺失这一问题，提出了利用Criminisi算法对信号重叠部分像素位置信息进行恢复。	Then, to solve the problem of missing pixels' position information in the overlapping part of multiple signals in the time-frequency domain, the Criminisi algorithm to repair and fill the information is applied.
53975	最后，基于缺失信息恢复后的图像使用Mask R-CNN进行训练，再用训练后的网络对未知信号进行检测。	Finally, Mask R-CNN is used for training the restored image, and used for detecting the unknown signals.
53976	实验结果表明，该方法在信噪比(SNR)为-3 dB时，时频域重叠信号的平均检测率达92%，相比基于卷积神经网络的信号检测方法，在SNR大于-3 dB时检测率平均提高20%以上。	Experimental results show that when the SNR is 0 dB, the average detection rate of overlapping signals in the time-frequency domain reaches 99%. Compared with the method based on convolutional neural network, the average detection rate is increased by more than 20%.