A MLE-based blind signal separation method for time–frequency overlapped signal using neural network
A MLE-based blind signal separation method for time–frequency overlapped signal using neural network
复制标题
基于MLE的神经网络时频重叠信号盲信号分离方法
DOI：
10.1186/s13634-022-00956-2
复制
发表时间：
2022-12
影响因子：
1.9
通讯作者：
Bin Yang
中科院分区：
文献类型：
--
作者：
Lihui Pang;Yilong Tang;Qingyi Tan;Yulang Liu;Bin Yang
盲信号分离（BSS）算法是从接收天线或传感器采集的观测信号中分离出各个原始/源信号。目标/损失/代价函数和优化方法是盲源分离算法的两个关键部分。从神经网络的角度对目标函数进行修正和优化是盲源分离领域的一个新概念。$$L_2$$。L. 2.正则化被采用作为基于最大似然估计（MLE）的目标函数的项，如Liu等人。（Sensors 21（3）：973，2021）;然而，我们修改了目标函数的概率密度函数（PDF）项，并将核密度估计方法用于时频重叠的数字通信信号。本文研究了多个优化器，并为我们的应用场景找到了正确的优化器。实验结果表明，该方法收敛速度快，在信噪比不小于10 dB的情况下，分离性能指数（PI）小于0.02。实验结果表明，该方法的性能优于传统的分离算法FastICA，特别是在低信噪比环境下，并且该方法对源信号的频率重叠水平（FOL）不敏感，即使FOL高达100%。100. %.它仍然可以得到高精度的分离结果与$$textrm{PI}。Pi. 0.02. .
AbstractThe blind signal separation (BSS) algorithm obtains each original/source signal from the observed signal collected by the receiving antenna or sensor. Objective/loss/cost function and optimization method are two key parts of BSS algorithm. Modifying the objective function and optimization from the perspective of neural network (NN) is a novel concept in BSS domain. $$L_2$$. L. 2. regularization is adopted as a term of maximum likelihood estimation (MLE)-based objective function like in Liu et al.(Sensors 21(3):973, 2021); however, we modified the probability density function (PDF) term of the objective function and used the kernel density estimation method for time–frequency overlapped digital communication signal. Multiple optimizers are studied in this paper, and we figure out the right optimizer for our application scenario. A varies of comparison experiments—whoseseparation results will be providedinforms of correlation coefficient and performance index—are carried out, which indicate our method can converge quickly and achieve satisfactory separation results with performance index (PI) lower than 0.02 when signal-to-noise ratio (SNR) no less than 10dB. Additionally, it demonstrates performance of our method is better than that of typical separation—FastICA, especially for the lower SNR environment, and it shows that our method is not sensitive to the frequency overlap level (FOL) of the source signal, even FOL as high as $$100%$$. 100. %. ; it still can get high-precision separation results with $$textrm{PI}. PI. 0.02. .
登录
查看更多内容
DOI：
--
发表时间：
2010-03
期刊：
--
影响因子：
--
作者：
P. Comon;C. Jutten
通讯作者：
P. Comon;C. Jutten
DOI：
10.3390/s21030973
发表时间：
2021-02-01
期刊：
Sensors (Basel, Switzerland)
影响因子：
--
作者：
Liu S;Wang B;Zhang L
通讯作者：
Zhang L
DOI：
10.1049/ip-vis:19941555
发表时间：
1994-12
期刊：
--
影响因子：
--
作者：
A. Bos
通讯作者：
A. Bos
影响因子：
1.3
作者：
Lihui Pang;Zhilong Qi;Shuai Li;B. Tang
通讯作者：
Lihui Pang;Zhilong Qi;Shuai Li;B. Tang
DOI：
10.5555/2627435.2638582
发表时间：
2014
期刊：
J. Mach. Learn. Res.
影响因子：
--
作者：
M. Gupta;Samy Bengio;J. Weston
通讯作者：
M. Gupta;Samy Bengio;J. Weston