基于流形学习的数字信号识别算法与噪声环境下的性能
50 浏览量
更新于2024-08-26
收藏 252KB PDF 举报
本文探讨了基于流形学习的数字信号识别方法，发表在《国际信号处理、图像处理与模式识别》（International Journal of Signal Processing, Image Processing, and Pattern Recognition）2016年第9卷第2期，页码127-134。该研究论文的doi是<http://dx.doi.org/10.14257/ijsip.2016.9.2.11>，ISSN号为2005-4254。作者团队由Qingbo Ji、Boyang Feng、Yun Lin*、Zheng Dou*和Zhiqiang Wu及Zhiping Zhang组成，分别来自哈尔滨工程大学信息与通信工程学院和美国怀特州立大学电气工程系。
信号识别中的调制类型是关键特性之一，论文提出了一种新的算法来实现信号的调制识别。首先，研究者采用了小波变换和短时傅立叶变换（STFT）对信号进行预处理，以捕捉信号的时频特征。接着，利用流形学习技术，有效地降低高维数据的复杂性，提取出有助于识别的特征。这种方法的主要目标是在高斯白噪声环境中准确识别四种常见信号类型：MASK、MFSK（移相键控）、MPSK（多相键控）和QAM（正交幅度调制）。
流形学习在这里扮演了降维和特征提取的关键角色，通过保持数据局部结构的保真度，能够更好地分离不同类别的信号。设置合适的阈值作为分类器，使得算法能够在噪声背景下区分这些复杂的调制模式。这种结合信号处理技术与非线性数据分析的方法，提高了识别的精度和鲁棒性，对于实际信号处理应用具有重要意义。
这篇研究论文提供了将流形学习应用于数字信号识别的有效策略，展示了如何通过结合传统信号分析工具和现代机器学习方法，优化信号识别性能，适用于通信、雷达和信号监控等领域。同时，它也表明了在解决实际问题时，选择适当的技术组合的重要性，以及对噪声环境下的信号处理能力的需求。
International Journal of Signal Processing, Image Processing and Pattern Recognition
Vol.9, No.2 (2016), pp.127-134
http://dx.doi.org/10.14257/ijsip.2016.9.2.11
ISSN: 2005-4254 IJSIP
Copyright ⓒ 2016 SERSC
Identifying of Digital Signals Based on Manifold Learning
Qingbo Ji
1
, Boyang Feng
1
, Yun Lin
1*
, Zheng Dou
1*
, Zhiqiang Wu
2
and Zhiping Zhang
2
1
College of Information and Communication Engineering
Harbin Engineering University, Harbin, China
linyun@hrbeu.edu.cn
2
Department of Electrical Engineering
Wright State University, Dayton, Ohio, U.S.
zhiqiang.wu@wright.edu
Abstract
Modulation type is one of the most important characteristics used in signal recognition.
An algorithm to realize signal modulation identification is proposed in this paper. We
applied wavelet transformation and STFT to the signal, and then used manifold learning
method to reduce the high dimension and extracted the recognition feature. The proper
threshold value was set as the classifier to achieve the purpose of recognizing 4 kinds of
signals (MASK, MFSK, MPSK,QAM) in Gauss white noise environment. The algorithm
requires priori signal information no other than signal-to-noise rate. Simulation result
indicates the algorithm achieves good performance.
Keywords: Digital signals identification, Feature extraction, Manifold learning
method, Isomap
1. Introduction
Digital signals are widely used both in commercial and military fields. To analysis the
information transferred by the source, the signal mode needs to be figured out by
selecting the appropriate features. Feature selection is the process of choosing a subset of
the original predictive variables by eliminating redundant and uninformative ones. By
extracting as much information as possible from a given data set while using the smallest
number of features, we can save significant computing time and often build models that
generalize better to unseen points.
Among all the signal parameters, in-pulse characteristics have very special effects.
Many in-pulse characteristics have been used on signal recognition such as entropy
analysis, short time Fourier transformation, wavelet transformation, complexity feature
and so on. For example, Swami and Sadler [3] proposed a wavelet transform-based signal
identification method, with which the success rate of 98% at signal-to-noise ratio (SNR) 4
dB was reported. Zhang [4] proposed a support vector machine-based classifier to
classify the signals according to the proposed features. The types of the signals have been
identified with a success rate of about 90% for 0<SNR<5 dB. A digital modulation
classification system was proposed by Xu et al. [5] for CR using only temporal waveform
features. They reported a success rate of 95% at SNRs ranging from 10 to 80 dB. In [6],
the authors presented a high-performance multi-layer perception neural network with
resilient back propagation learning algorithm. In [7], a signal classification approach
based on neural network ensembles was proposed, which enables dynamic spectrum
access. From the research works mentioned above, it can be found that: (a) most of the
proposed methods can only recognise low-order and limited digital signals; (b) most of
*
Corresponding Author
下载后可阅读完整内容，剩余7页未读，立即下载
2023-04-19 上传
2021-04-30 上传
2021-03-20 上传
2021-09-23 上传
188 浏览量
2021-12-31 上传
2021-03-06 上传
点击了解资源详情
点击了解资源详情
