WO2014194485A1 - 一种以低于奈奎斯特频率的采样频率进行信号采集方法 - Google Patents
一种以低于奈奎斯特频率的采样频率进行信号采集方法 Download PDFInfo
- Publication number
- WO2014194485A1 WO2014194485A1 PCT/CN2013/076762 CN2013076762W WO2014194485A1 WO 2014194485 A1 WO2014194485 A1 WO 2014194485A1 CN 2013076762 W CN2013076762 W CN 2013076762W WO 2014194485 A1 WO2014194485 A1 WO 2014194485A1
- Authority
- WO
- WIPO (PCT)
- Prior art keywords
- signal
- original input
- input signal
- sampling
- sparsity
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Ceased
Links
Classifications
-
- H—ELECTRICITY
- H03—ELECTRONIC CIRCUITRY
- H03M—CODING; DECODING; CODE CONVERSION IN GENERAL
- H03M7/00—Conversion of a code where information is represented by a given sequence or number of digits to a code where the same, similar or subset of information is represented by a different sequence or number of digits
- H03M7/30—Compression; Expansion; Suppression of unnecessary data, e.g. redundancy reduction
- H03M7/3059—Digital compression and data reduction techniques where the original information is represented by a subset or similar information, e.g. lossy compression
- H03M7/3062—Compressive sampling or sensing
-
- H—ELECTRICITY
- H03—ELECTRONIC CIRCUITRY
- H03M—CODING; DECODING; CODE CONVERSION IN GENERAL
- H03M7/00—Conversion of a code where information is represented by a given sequence or number of digits to a code where the same, similar or subset of information is represented by a different sequence or number of digits
- H03M7/30—Compression; Expansion; Suppression of unnecessary data, e.g. redundancy reduction
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
- H04L27/00—Modulated-carrier systems
- H04L27/26—Systems using multi-frequency codes
- H04L27/2601—Multicarrier modulation systems
- H04L27/2626—Arrangements specific to the transmitter only
- H04L27/2627—Modulators
- H04L27/2642—Wavelet transform modulators
Definitions
- the present invention relates to a signal acquisition method at a sampling frequency lower than a Nyqmst frequency, capable of compressing and sampling a sparse signal or a sparsely representable signal at a sampling frequency far below the Nyqmst frequency, and Reconstruct the original signal.
- the Nyqmst sampling theorem is the basis of modern communication and signal processing.
- the theorem states that in the process of analog/digital signal conversion, when the sampling frequency is greater than or equal to 2 times the highest frequency of the signal, the digital signal after sampling completely retains the original signal. Information, in general practical applications, ensure that the sampling frequency is 5 to 10 times the highest frequency of the signal.
- the Nyqmst sampling theorem since the Nyqmst sampling theorem only develops the least a priori information of the acquired signal, that is, the bandwidth of the signal, it does not utilize some structural features of the signal itself, such as redundancy.
- Compressed sensing uses the sparse nature of the signal to sample the signal well below the Nyqmst frequency.
- Compressed sensing theory states that as long as the signal can be sparsely represented in a certain transform domain, it can be used without a phase matrix.
- the closed observation matrix projects the high-dimensional signal into a low-dimensional space, and then reconstructs the original high-dimensional signal from these small projections by solving the optimization problem.
- an object of the present invention is to provide a signal acquisition method with a sampling frequency lower than the Nyquist frequency, which can sample and reconstruct a signal at a sampling frequency far below the Nyqmst frequency, breaking through the classical Nyquist sampling.
- the theorem limits the sampling frequency and reduces the pressure on the acquisition, storage, transmission, and processing of wideband signals.
- the present invention provides a signal acquisition method for sampling frequencies lower than the Nyquist frequency, comprising: determining, by the data transmitting end, the sparsity k of the original input signal X, and calculating the compression according to the sparsity k
- the number of samples M sampled, using the M channels to sample the original input signal X at a frequency f NYQ /M, where f NYQ is the Nyquist frequency of the original input signal X, and the sample values of each channel are integrated Obtaining M measurement values y, and then transmitting the M measurement values y to the data receiving end; and the data receiving end performs orthogonal matching tracking algorithm on the M measurement values y, and restores the received measurement value y to original
- the sparse representation of the input signal X is obtained, and the original input signal ⁇ is obtained by inverse transform.
- the data transmitting end determines the sparsity k of the original input signal X, including: the data transmitting end selects a transform base matrix according to the feature of the original input signal X, and uses the transform base matrix to sparsely represent the original input signal X. The sparsity k of the original input signal X under the transform basis matrix is then determined.
- the data transmitting end selects a transform base matrix according to the feature of the original input signal X, and uses a Fourier transform base matrix for the smooth original input signal; and uses a wavelet for the single point singular signal. Transform the base matrix.
- the smooth original input signal comprises a continuously varying sinusoidal signal, a cosine signal, an exponential signal or a sampling pulse;
- the single point singular signal comprises an impulse signal whose signal is discontinuously varying, a rectangular pulse Or spike signal.
- the data transmitting end calculates the sampling frequency M of the compressed sampling according to the sparsity k, and the sampling number M of the compressed sampling satisfies the following two conditions:
- N is the dimension of the high-dimensional original input signal
- k is the sparsity of the original input signal under the relevant transform basis
- N AX M
- A is an integer greater than or equal to 1, that is, the dimension of the high-dimensional original input signal is an integer multiple of the number of samples M of the compressed sample.
- the data transmitting end integrates the sampling values of each channel to obtain M measured values y, and the integration period is A times of the sampling period of each channel.
- the orthogonal matching tracking algorithm is a method for obtaining the original input signal X by the measured value y and the observation matrix ⁇ by multiple iterations.
- the data receiving end obtains the original input signal X by inverse transform, and includes: performing inverse on the original signal sparse representation obtained by solving the orthogonal matching pursuit algorithm according to the transform base matrix selected when the original input signal X is sparsely represented.
- the transform base matrix selected when the original input signal X is sparsely represented is a Fourier transform base matrix
- the inverse transform is a discrete Fourier transform
- the selected transform base matrix is a wavelet
- the transform base matrix is inverse transformed into a discrete wavelet transform.
- the method for collecting signals at a sampling frequency lower than the Nyquist frequency provided by the present invention is based on the theory of compressed sensing, and fully utilizes the sparse characteristics of the signal in the transform domain, and can be sparse at frequencies far below the Nyqmst frequency.
- Signal or signal that can be sparsely represented Compressed sampling breaks through the limitation of the sampling frequency by the classical Nyquist sampling theorem.
- this method only uses two steps of low-speed sampling and integration, which is simple to implement and reduces the pressure of data acquisition, storage, transmission and processing.
- FIG. 1 is a flow chart of a method for signal acquisition at a sampling frequency lower than a Nyquist frequency provided by the present invention
- FIG. 2 is a schematic diagram of signal acquisition in accordance with an embodiment of the present invention.
- FIG. 3 is a schematic diagram of signal reconstruction in accordance with an embodiment of the present invention.
- FIG. 4 is a sinusoidal signal of 50 Hz, 100 Hz, 150 Hz, and 200 Hz and a frequency domain representation thereof according to an embodiment of the present invention, wherein the left side is the original signal picture, and the right side is the sparse representation picture after the Fourier transform;
- FIG. 5 is a diagram showing the sampling and reconstruction effects of a signal according to an embodiment of the present invention, (a) being an original signal, (b) a sampled value, (c reconstructed signal, (d) an error of the reconstructed signal and the original signal.
- FIG. 1 is a flow chart of a method for signal acquisition at a sampling frequency lower than a Nyquist frequency provided by the present invention.
- the method is based on a compression sensing theory and has a frequency much lower than the Nyquist frequency.
- the acquisition of the probability realization signal includes the following steps:
- the data transmitting end determines the sparsity k of the original input signal X, calculates the sampling frequency M of the compressed sampling according to the sparsity k, and samples the original input signal X by the frequency ⁇ by using M channels, where f NYQ is the original input signal The Nyquist frequency of X, and integrate the sampled values of each channel to obtain M measured values y, and then the M measurements The value y is sent to the data receiving end;
- the data receiving end performs orthogonal matching tracking algorithm on the M measurement values y, restores the received measurement value y to the sparse representation of the original input signal X, and obtains the original input signal x through inverse transformation.
- the data transmitting end determines the sparsity k of the original input signal X, including: the data transmitting end selects a transform base matrix according to the feature of the original input signal X, and uses the transform base matrix to sparsely represent the original input signal X, and then seeks The sparsity k of the original input signal X under the transform basis matrix is derived.
- the data transmitting end selects a transform base matrix according to the characteristics of the original input signal X, and uses a Fourier transform base matrix for the smooth original input signal; and uses a wavelet transform base matrix for the single point singular signal.
- the smooth original input signal comprises continuously varying signals such as a sinusoidal signal, a cosine signal, an exponential signal or a sampling pulse; the single point singular signal comprises a non-continuously varying signal such as an impulse signal, a rectangular pulse or a spike signal.
- the data transmitting end calculates the sampling number M of the compressed sampling according to the sparsity k, and the sampling number M of the compressed sampling satisfies the following two conditions: a, M ⁇ k. ⁇ n ⁇ Nlk) , where N is high
- the data transmitting end integrates the sampled values of each channel to obtain M measured values y, and the integration period is A times of the sampling period of each channel.
- the original input signal X can be obtained by solving the optimization /r norm, and the orthogonal matching tracking algorithm is a kind A method of obtaining the original input signal X by the measured value y and the observation matrix ⁇ by multiple iterations.
- the data receiving end obtains the original input signal X by inverse transform, and includes: inversely changing the original signal sparse representation obtained by solving the orthogonal matching tracking algorithm according to the transform base matrix selected when the original input signal X is sparsely represented, if When the original input signal X is sparsely represented, the transform base matrix selected is a Fourier transform base matrix, and the inverse transform is a discrete Fourier transform. If the original input signal X is sparsely represented, the selected transform base matrix is a wavelet transform base matrix. Then the inverse transform is a discrete wavelet transform.
- 2 is a schematic diagram of signal acquisition according to an embodiment of the present invention
- FIG. 3 is a schematic diagram of signal reconstruction according to an embodiment of the present invention, which is specifically divided into the following three steps:
- Step 1 The data transmitting end sparsely indicates the original input signal X;
- the data transmitting end uses the Fourier transform base matrix to sparsely represent the original input signal.
- the left side is the original input signal picture
- Step 2 The data transmitting end samples the original input signal X at a sampling frequency of f NYQ /M;
- Step 3 The data receiving end reconstructs the original input signal X;
- the data receiving end reconstructs the received low-dimensional data (ie, the measured value y) into the original high-dimensional data by using the orthogonal matching pursuit algorithm, and the measured value y obtained after the steps 1 and 2 is equivalent to the following mathematical expression.
- the original input signal X can be obtained by solving the optimization/r norm, which is a method for obtaining the original input signal X by the measurement value y and the observation matrix ⁇ by multiple iterations.
- the reconstructed signal has an error of 10-14 , which is very small
- FIG. 5(d) shows the difference between the reconstructed signal and the original signal, and the magnitude is 10 – 13 , It can be seen that the method can achieve compressed sampling of the signal at a frequency much lower than the frequency of Nyqmst, and can recover the original high-dimensional signal very well from very few samples.
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Computer Networks & Wireless Communication (AREA)
- Signal Processing (AREA)
- Measurement Of Velocity Or Position Using Acoustic Or Ultrasonic Waves (AREA)
- Compression, Expansion, Code Conversion, And Decoders (AREA)
Abstract
本发明公开了一种低于Nyquist频率的信号采集方法,首先针对输入信号选择合适的变换基矩阵,用该变换基矩阵对信号进行稀疏表示以确定信号的稀疏度k,然后由稀疏度计算出压缩采样的采样次数M,然后分M个信道分别对信号以fNYQ/M进行采样,接着对每个信道的采样值进行积分,即得到M个测量值,然后重构端通过求解优化问题重构出原始信号。本发明以压缩感知理论为基础,能够以远低于Nyquist频率的频率对稀疏信号或者可稀疏表示的信号进行压缩采样,突破了经典的Nyquist采样定理对采样频率的限制,同时该方法实施简单,降低了数据采集、存储、传输以及处理的压力。
Description
一种以低于奈奎斯特频率的采样频率进行信号采集方法
技术领域 本发明涉及一种以低于奈奎斯特(Nyqmst)频率的采样频率进行信 号采集方法, 能够以远低于 Nyqmst频率的采样频率对稀疏信号或者可 稀疏表示的信号进行压缩采样, 并重构出原始信号。
背景技术
Nyqmst采样定理是现代通讯与信号处理的基础,该定理指出在进行 模拟 /数字信号转换过程中, 当采样频率大于等于信号最高频率 2倍时, 采样之后的数字信号完整地保留了原始信号中的信息, 一般实际应用中 保证采样频率为信号最高频率的 5~10倍。
对于宽带信号, 由于其本身频率比较高, 如果采用 Nyqmst采样定 理直接对信号进行采样, 则需要的采样频率就非常高, 现有的模数转换 器件的最高频率为 1GHz, 所以很难满足高频率的要求。 另外信号经模 数转换器之后, 一般先要存储, 然后再进行数据通信和处理, 高速采样 就对存储器的存储速度、 数字信号处理器的处理速度要求很高, 所以宽 带信号采用 Nyqmst采样定理进行采样将给宽带信号的采集、 存储、 传 输以及处理带来巨大的压力。
另外, 由于 Nyqmst采样定理只开发利用了被采集信号的最少的先 验信息, 即信号的带宽, 并没有利用信号本身具有的一些结构特点, 如 冗余等。
通常所处理的信号, 虽然信号本身并不稀疏, 但是在某些变换基下 可以稀疏表示。 压缩感知作为一个新兴理论, 就是利用信号的这种稀疏 特性以远低于 Nyqmst频率对信号进行采样。 压缩感知理论指出, 只要 信号可以在某个变换域下稀疏表示, 就可以用一个与该变换基矩阵不相
关的观测矩阵将高维信号投影到一个低维空间, 然后通过求解优化问题 来从这些少量的投影中重构出原始高维信号。
发明内容
(一) 要解决的技术问题
有鉴于此, 本发明的目的是提供一种以低于奈奎斯特频率的采样频 率进行信号采集方法, 能够以远低于 Nyqmst频率的采样频率对信号进 行采样和重构, 突破经典 Nyquist采样定理对采样频率的限制, 降低给 宽带信号的采集、 存储、 传输以及处理带来的压力。
(二) 技术方案
为达到上述目的, 本发明提供了一种以低于奈奎斯特频率的采样频 率进行信号采集方法, 包括: 数据发送端确定原始输入信号 X的稀疏度 k, 根据该稀疏度 k计算出压缩采样的采样次数 M, 利用 M个信道分别 对原始输入信号 X以频率 fNYQ/M进行采样,其中 fNYQ为原始输入信号 X 的奈奎斯特频率, 并对每个信道的采样值进行积分得到 M个测量值 y, 然后将这 M个测量值 y发送到数据接收端; 以及数据接收端对这 M个 测量值 y进行正交匹配追踪算法求解, 将接收到的测量值 y恢复为原始 输入信号 X的稀疏表示, 再通过反变换得到原始输入信号 χ。
上述方案中, 所述数据发送端确定原始输入信号 X的稀疏度 k, 包 括: 数据发送端根据原始输入信号 X的特征选择变换基矩阵, 用该变换 基矩阵对原始输入信号 X进行稀疏表示, 然后求出该原始输入信号 X在 该变换基矩阵下的稀疏度 k。
上述方案中, 所述数据发送端根据原始输入信号 X的特征选择变换 基矩阵的歩骤中, 对于光滑的原始输入信号, 采用傅里叶变换基矩阵; 对于单一点状的奇异信号, 采用小波变换基矩阵。 所述光滑的原始输入 信号包括信号是连续变化的正弦信号、余弦信号、指数信号或抽样脉冲; 所述单一点状的奇异信号包括信号是非连续变化的冲激信号、 矩形脉冲
或尖峰信号。
上述方案中, 所述数据发送端根据稀疏度 k计算出压缩采样的采样 次数 M的歩骤中, 压缩采样的采样次数 M满足以下两个条件:
a、 M≥k. NIk、, 其中 N为高维的原始输入信号的维数, k为原始 输入信号在相关变换基下的稀疏度;
b、 N=AX M, A为大于等于 1的整数, 即高维的原始输入信号的维 数是压缩采样的采样次数 M的整数倍。
上述方案中, 所述数据发送端对每个信道的采样值进行积分得到 M 个测量值 y的歩骤中, 积分周期是每个信道采样周期的 A倍。
上述方案中,所述数据接收端对 M个测量值 y进行正交匹配追踪算 法求解,将接收到的测量值 y恢复为原始输入信号 X的稀疏表示,包括: 根据数学运算式 二^^, 由原始输入信号 X求解得到的测量值 y, 其中原 始输入信号 X是一个 N X 1的向量, 测量值 y是一个 MX 1的向量, 观 测矩阵 Φ是一个 MX N的托普利兹矩阵, 该矩阵由 A=N/M个 MX M的 单位矩阵按行的方向合并而成; 已知测量值 y和观测矩阵 Φ, 要求解原 始输入信号 X, 由于原始输入信号 X是稀疏的, 所以能够通过求解最优 化 /r范数来得到原始输入信号 X, 正交匹配追踪算法就是一种通过多次 迭代由测量值 y和观测矩阵 Φ求解得到原始输入信号 X的方法。
上述方案中, 所述数据接收端通过反变换得到原始输入信号 X, 包 括: 根据对原始输入信号 X稀疏表示时选择的变换基矩阵, 对求解正交 匹配追踪算法得到的原始信号稀疏表示进行反变化, 若对原始输入信号 X稀疏表示时选择的变换基矩阵是傅里叶变换基矩阵, 则反变换为离散 傅里叶变换, 若对原始输入信号 X稀疏表示时选择的变换基矩阵是小波 变换基矩阵, 则反变换为离散小波变换。 (三) 有益效果
本发明提供的这种以低于奈奎斯特频率的采样频率进行信号采集 方法, 以压缩感知理论为基础, 充分利用信号在变换域的稀疏特性, 能 够以远低于 Nyqmst频率的频率对稀疏信号或者可稀疏表示的信号进行
压缩采样, 突破了经典的 Nyquist采样定理对采样频率的限制, 同时该 方法只用低速采样和积分两个歩骤,实施简单,降低了数据采集、存储、 传输以及处理的压力。
附图说明 为了更进一歩说明本发明的内容, 以下结合附图及实施例子, 对本 发明做详细描述, 其中:
图 1是本发明提供的以低于奈奎斯特频率的采样频率进行信号采集 的方法流程图;
图 2是依照本发明实施例进行信号采集的示意图;
图 3是依照本发明实施例进行信号重构的示意图;
图 4是依照本发明实施例 50Hz、 100Hz、 150Hz和 200Hz的正弦信 号及其频域表示, 其中左边为原始信号图, 右边为傅里叶变换后的稀疏 表示图;
图 5是依照本发明实施例信号的采样及重构效果图, (a)为原始信号、 (b)采样值、 (c重构信号、 (d)重构信号与原始信号的误差。
具体实施方式 为使本发明的目的、 技术方案和优点更加清楚明白, 以下结合具体 实施例, 并参照附图, 对本发明进一歩详细说明。
如图 1所示, 图 1是本发明提供的以低于奈奎斯特频率的采样频率 进行信号采集的方法流程图, 该方法在压缩感知理论基础上, 以远低于 Nyquist频率的频率高概率的实现信号的采集, 包括以下歩骤:
数据发送端确定原始输入信号 X的稀疏度 k, 根据该稀疏度 k计算 出压缩采样的采样次数 M, 利用 M个信道分别对原始输入信号 X以频 率 ^ 进行采样, 其中 fNYQ为原始输入信号 X的奈奎斯特频率, 并 对每个信道的采样值进行积分得到 M个测量值 y, 然后将这 M个测量
值 y发送到数据接收端; 以及
数据接收端对这 M个测量值 y进行正交匹配追踪算法求解,将接收 到的测量值 y恢复为原始输入信号 X的稀疏表示, 再通过反变换得到原 始输入信号 x。
其中, 所述数据发送端确定原始输入信号 X的稀疏度 k, 包括: 数 据发送端根据原始输入信号 X的特征选择变换基矩阵, 用该变换基矩阵 对原始输入信号 X进行稀疏表示, 然后求出该原始输入信号 X在该变换 基矩阵下的稀疏度 k。 所述数据发送端根据原始输入信号 X的特征选择 变换基矩阵的歩骤中, 对于光滑的原始输入信号, 采用傅里叶变换基矩 阵; 对于单一点状的奇异信号, 采用小波变换基矩阵。 所述光滑的原始 输入信号包括正弦信号、 余弦信号、 指数信号或抽样脉冲等连续变化的 信号; 所述单一点状的奇异信号包括冲激信号、 矩形脉冲或尖峰信号等 非连续变化的信号。
所述数据发送端根据稀疏度 k计算出压缩采样的采样次数 M的歩骤 中, 压缩采样的采样次数 M满足以下两个条件: a、 M≥k . \n{Nlk) , 其 中 N为高维的原始输入信号的维数, k为原始输入信号在相关变换基下 的稀疏度; b、 N=AX M, A为大于等于 1的整数, 即高维的原始输入信 号的维数是压缩采样的采样次数 M的整数倍。所述数据发送端对每个信 道的采样值进行积分得到 M个测量值 y的歩骤中,积分周期是每个信道 采样周期的 A倍。
所述数据接收端对 M个测量值 y进行正交匹配追踪算法求解,将接 收到的测量值 y恢复为原始输入信号 X的稀疏表示, 是根据数学表达式 γ = Φ ,由测量值 y和观测矩阵 Φ通过多次迭代求解原始输入信号 X的过 程, 包括: 根据数学运算式 = ^ , 由原始输入信号 X求解得到的测量 值 y, 其中原始输入信号 X是一个 NX 1的向量, 测量值 y是一个 MX 1 的向量, 观测矩阵 Φ是一个 MXN的托普利兹矩阵, 该矩阵由 A=N/M 个 MX M的单位矩阵按行的方向合并而成;已知测量值 y和观测矩阵 Φ , 要求解原始输入信号 X, 由于原始输入信号 X是稀疏的, 所以能够通过 求解最优化 /r范数来得到原始输入信号 X, 正交匹配追踪算法就是一种
通过多次迭代由测量值 y和观测矩阵 Φ求解得到原始输入信号 X的方法。 所述数据接收端通过反变换得到原始输入信号 X, 包括: 根据对原 始输入信号 X稀疏表示时选择的变换基矩阵, 对求解正交匹配追踪算法 得到的原始信号稀疏表示进行反变化, 若对原始输入信号 X稀疏表示时 选择的变换基矩阵是傅里叶变换基矩阵, 则反变换为离散傅里叶变换, 若对原始输入信号 X稀疏表示时选择的变换基矩阵是小波变换基矩阵, 则反变换为离散小波变换。
基于图 1所示的本发明提供的以低于奈奎斯特频率的采样频率进行 信号采集的方法流程图, 图 2至图 5以频率为 50Hz、 100Hz、 150Hz和 200Hz的正弦信号 (信号长度为 N=256 ) 为例, 对本发明进一歩详细说 明。 其中, 图 2是依照本发明实施例进行信号采集的示意图, 图 3是依 照本发明实施例进行信号重构的示意图, 具体分为以下三个歩骤:
歩骤 1: 数据发送端对原始输入信号 X进行稀疏表示;
由于是正弦信号, 所以数据发送端采用傅里叶变换基矩阵对原始输 入信号进行稀疏表示, 如图 4所示, 左边为原始输入信号图, 右边为信 号傅里叶变换后的稀疏表示图, 可以发现, 该原始输入信号只有 8个非 零值, 其他值都为零, 即该信号的稀疏度 k=8;
歩骤 2: 数据发送端对原始输入信号 X在 fNYQ/M的采样频率下进行 采样;
首先确定压缩采样的采样次数 M的值, 由于 M要满足两个条件: ( a) M≥k. N/k、, 其中 N为原始高维信号的维数, k为信号在相关 变换基下的稀疏度; (b ) N=AX M, A为大于等于 1的整数, 所以这里 取 M=32, A=8 ; 接着分 32 个信道分别对原始输入信号 x 在 fNYQ/32 fNYQ=400HZ;)采样频率下进行采样, 每个信道得到的采样值为 8 个, 并对每个信道的 8个采样值进行积分, 积分周期是每个信道采样周 期的 A倍, 这样就可以得到 32个测量值 y, 直接把这个 32个测量值 y 发送到数据接收端; 如图 5 ( a)所示, 为原始信号图; 如图 5 (b)所示, 为得到的 32个采样值。
歩骤 3 : 数据接收端重构出原始输入信号 X;
数据接收端利用正交匹配追踪算法, 将接收到的低维数据 (即测量 值 y) 重构为原始高维数据, 经过歩骤 1和 2之后得到的测量值 y等价 于如下数学运算式 Υ = Χ, 由原始输入信号 X求解得到的测量值 y,其中 原始输入信号 X是一个 NX 1的向量, 测量值 y是一个 M X 1的向量, 观测矩阵 Φ是一个 MX N的托普利兹矩阵, 该矩阵由 A=N/M个 MX M 的单位矩阵按行的方向合并而成; 已知测量值 y和观测矩阵 Φ, 要求解 原始输入信号 X, 由于原始输入信号 X是稀疏的, 所以能够通过求解最 优化 /r范数来得到原始输入信号 X, 正交匹配追踪算法就是一种通过多 次迭代由测量值 y和观测矩阵 Φ求解得到原始输入信号 X的方法。
如图 5的(c)所示为重构出的信号, 误差为 10— 14数量级, 非常小, 图 5 ( d)所示为重构出的信号与原始信号的差值, 量级为 10— 13, 可见该 方法可以以远低于 Nyqmst频率的频率实现信号的压缩采样, 并能非常 好的由极少采样值恢复出原始高维信号。
以上所述的具体实施例, 对本发明的目的、 技术方案和有益效果进 行了进一歩详细说明, 所应理解的是, 以上所述仅为本发明的具体实施 例而已, 并不用于限制本发明, 凡在本发明的精神和原则之内, 所做的 任何修改、 等同替换、 改进等, 均应包含在本发明的保护范围之内。
Claims
1、 一种以低于奈奎斯特频率的采样频率进行信号采集方法, 其特 征在于, 包括:
数据发送端确定原始输入信号 X的稀疏度 k, 根据该稀疏度 k计算 出压缩采样的采样次数 M, 利用 M个信道分别对原始输入信号 X以频 率 ^ 进行采样, 其中 fNYQ为原始输入信号 X的奈奎斯特频率, 并 对每个信道的采样值进行积分得到 M个测量值 y, 然后将这 M个测量 值 y发送到数据接收端; 以及
数据接收端对这 M个测量值 y进行正交匹配追踪算法求解,将接收 到的测量值 y恢复为原始输入信号 X的稀疏表示, 再通过反变换得到原 始输入信号 x。
2、 根据权利要求 1 所述的以低于奈奎斯特频率的采样频率进行信 号采集方法, 其特征在于, 所述数据发送端确定原始输入信号 X的稀疏 度 k, 包括:
数据发送端根据原始输入信号 X的特征选择变换基矩阵, 用该变换 基矩阵对原始输入信号 X进行稀疏表示, 然后求出该原始输入信号 X在 该变换基矩阵下的稀疏度 k。
3、 根据权利要求 2所述的以低于奈奎斯特频率的采样频率进行信 号采集方法, 其特征在于, 所述数据发送端根据原始输入信号 X的特征 选择变换基矩阵的歩骤中, 对于光滑的原始输入信号, 采用傅里叶变换 基矩阵; 对于单一点状的奇异信号, 采用小波变换基矩阵。
4、 根据权利要求 3 所述的以低于奈奎斯特频率的采样频率进行信 号采集方法, 其特征在于, 所述光滑的原始输入信号包括信号是连续变 化的正弦信号、 余弦信号、 指数信号或抽样脉冲; 所述单一点状的奇异 信号包括信号是非连续变化的冲激信号、 矩形脉冲或尖峰信号。
5、 根据权利要求 1 所述的以低于奈奎斯特频率的采样频率进行信 号采集方法, 其特征在于, 所述数据发送端根据稀疏度 k计算出压缩采 样的采样次数 M的歩骤中,压缩采样的采样次数 M满足以下两个条件:
a、 M≥ NIk、, 其中 N为高维的原始输入信号的维数, k为原始 输入信号在相关变换基下的稀疏度;
b、 N=AX M, A为大于等于 1的整数, 即高维的原始输入信号的维 数是压缩采样的采样次数 M的整数倍。
6、 根据权利要求 5 所述的以低于奈奎斯特频率的采样频率进行信 号采集方法, 其特征在于, 所述数据发送端对每个信道的采样值进行积 分得到 M个测量值 y的歩骤中,积分周期是每个信道采样周期的 A倍。
7、 根据权利要求 1 所述的以低于奈奎斯特频率的采样频率进行信 号采集方法,其特征在于,所述数据接收端对 M个测量值 y进行正交匹 配追踪算法求解, 将接收到的测量值 y恢复为原始输入信号 X的稀疏表 示, 包括:
根据数学运算式 二^^, 由原始输入信号 X求解得到的测量值 y, 其中原始输入信号 X是一个 N X 1的向量, 测量值 y是一个 MX 1的向 量, 观测矩阵 Φ是一个 MX N的托普利兹矩阵, 该矩阵由A=N/M个 M X M的单位矩阵按行的方向合并而成; 已知测量值 y和观测矩阵 Φ, 要 求解原始输入信号 X, 由于原始输入信号 X是稀疏的, 所以能够通过求 解最优化 /r范数来得到原始输入信号 X, 正交匹配追踪算法就是一种通 过多次迭代由测量值 y和观测矩阵 Φ求解得到原始输入信号 X的方法。
8、 根据权利要求 7 所述的以低于奈奎斯特频率的采样频率进行信 号采集方法, 其特征在于, 所述数据接收端通过反变换得到原始输入信 号 X, 包括:
根据对原始输入信号 X稀疏表示时选择的变换基矩阵, 对求解正交 匹配追踪算法得到的原始信号稀疏表示进行反变化, 若对原始输入信号 X稀疏表示时选择的变换基矩阵是傅里叶变换基矩阵, 则反变换为离散 傅里叶变换, 若对原始输入信号 X稀疏表示时选择的变换基矩阵是小波 变换基矩阵, 则反变换为离散小波变换。
Priority Applications (2)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| PCT/CN2013/076762 WO2014194485A1 (zh) | 2013-06-05 | 2013-06-05 | 一种以低于奈奎斯特频率的采样频率进行信号采集方法 |
| US14/805,868 US9455741B2 (en) | 2013-06-05 | 2015-07-22 | Method for collecting signal with sampling frequency lower than Nyquist frequency |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| PCT/CN2013/076762 WO2014194485A1 (zh) | 2013-06-05 | 2013-06-05 | 一种以低于奈奎斯特频率的采样频率进行信号采集方法 |
Related Child Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| US14/805,868 Continuation US9455741B2 (en) | 2013-06-05 | 2015-07-22 | Method for collecting signal with sampling frequency lower than Nyquist frequency |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| WO2014194485A1 true WO2014194485A1 (zh) | 2014-12-11 |
Family
ID=52007414
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| PCT/CN2013/076762 Ceased WO2014194485A1 (zh) | 2013-06-05 | 2013-06-05 | 一种以低于奈奎斯特频率的采样频率进行信号采集方法 |
Country Status (2)
| Country | Link |
|---|---|
| US (1) | US9455741B2 (zh) |
| WO (1) | WO2014194485A1 (zh) |
Families Citing this family (10)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US10812790B2 (en) * | 2016-04-19 | 2020-10-20 | Sony Corporation | Data processing apparatus and data processing method |
| CN107196661B (zh) * | 2017-05-16 | 2019-11-12 | 中国人民解放军国防科学技术大学 | 用于转子系统的振动信号处理的信号重构方法 |
| CN110048724A (zh) * | 2019-04-11 | 2019-07-23 | 池州学院 | 一种电能质量信号压缩采样重构方法 |
| CN110133628A (zh) * | 2019-06-13 | 2019-08-16 | 哈尔滨工业大学 | 一种短时强干扰下lfm信号的恢复方法 |
| CN113346908A (zh) * | 2021-05-11 | 2021-09-03 | 中国电力科学研究院有限公司 | 一种用于对高压电缆量测网络数据预压缩的方法及系统 |
| CN114221672B (zh) * | 2021-12-15 | 2023-05-05 | 西安电子科技大学 | 一种基于ifft的频域稀疏信号收发系统实现方法 |
| CN115620734A (zh) * | 2021-12-30 | 2023-01-17 | 西安交通大学 | 航空发动风扇单音噪声周向声模态识别方法及装置 |
| CN114325197B (zh) * | 2021-12-31 | 2024-09-06 | 杭州电力设备制造有限公司 | 一种电能质量扰动检测方法及装置 |
| CN114925726B (zh) * | 2022-05-13 | 2024-07-26 | 华侨大学 | 亚采样工作模态参数的识别方法、装置、设备和存储介质 |
| CN114997226B (zh) * | 2022-05-30 | 2025-03-14 | 国网湖南省电力有限公司 | 典型用电场景的源荷特征信号压缩感知方法及系统 |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US7289049B1 (en) * | 2006-08-21 | 2007-10-30 | L3 Communications Integrated Systems L.P. | Method and apparatus for compressed sensing |
| CN101895297A (zh) * | 2010-07-30 | 2010-11-24 | 哈尔滨工业大学 | 一种面向压缩感知的块稀疏信号重构方法 |
| CN102253117A (zh) * | 2011-03-31 | 2011-11-23 | 浙江大学 | 一种基于压缩感知的新型声学信号采集方法 |
| CN103346798A (zh) * | 2013-06-05 | 2013-10-09 | 中国科学院微电子研究所 | 一种以低于奈奎斯特频率的采样频率进行信号采集方法 |
Family Cites Families (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US7345603B1 (en) | 2006-11-07 | 2008-03-18 | L3 Communications Integrated Systems, L.P. | Method and apparatus for compressed sensing using analog projection |
| WO2010095083A1 (en) * | 2009-02-18 | 2010-08-26 | Technion Research & Development Foundation Ltd | Efficient sampling and reconstruction of sparse multi-band signals |
| US8836557B2 (en) * | 2010-10-13 | 2014-09-16 | Technion Research & Development Foundation Ltd. | Sub-Nyquist sampling of short pulses |
| US8890589B2 (en) * | 2012-04-23 | 2014-11-18 | Samsung Electronics Co., Ltd. | Apparatuses for measuring high speed signals and methods thereof |
| US8823573B1 (en) * | 2013-02-20 | 2014-09-02 | Raytheon Company | System and method for reconstruction of sparse frequency spectrum from ambiguous under-sampled time domain data |
-
2013
- 2013-06-05 WO PCT/CN2013/076762 patent/WO2014194485A1/zh not_active Ceased
-
2015
- 2015-07-22 US US14/805,868 patent/US9455741B2/en active Active
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US7289049B1 (en) * | 2006-08-21 | 2007-10-30 | L3 Communications Integrated Systems L.P. | Method and apparatus for compressed sensing |
| CN101895297A (zh) * | 2010-07-30 | 2010-11-24 | 哈尔滨工业大学 | 一种面向压缩感知的块稀疏信号重构方法 |
| CN102253117A (zh) * | 2011-03-31 | 2011-11-23 | 浙江大学 | 一种基于压缩感知的新型声学信号采集方法 |
| CN103346798A (zh) * | 2013-06-05 | 2013-10-09 | 中国科学院微电子研究所 | 一种以低于奈奎斯特频率的采样频率进行信号采集方法 |
Non-Patent Citations (1)
| Title |
|---|
| WANG, TIANJING ET AL.: "Filter-Based Compressed Sensing Signal Collection Scheme", CHINESE JOURNAL OF SCIENTIFIC INSTRUMENT, vol. 34, no. 3, March 2013 (2013-03-01), pages 573 - 581 * |
Also Published As
| Publication number | Publication date |
|---|---|
| US20150326246A1 (en) | 2015-11-12 |
| US9455741B2 (en) | 2016-09-27 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN103346798B (zh) | 一种以低于奈奎斯特频率的采样频率进行信号采集方法 | |
| US9455741B2 (en) | Method for collecting signal with sampling frequency lower than Nyquist frequency | |
| Wei et al. | Generalized wavelet transform based on the convolution operator in the linear canonical transform domain | |
| CN102253117B (zh) | 一种基于压缩感知的声学信号采集方法 | |
| CN106130563B (zh) | 一种基于压缩感知信号的阈值收缩迭代差分重构方法 | |
| CN107561367B (zh) | 一种基于压缩感知理论的宽频谱阻抗测量方法 | |
| CN102624399B (zh) | 一种压缩感知信号的重构方法 | |
| CN104779960B (zh) | 一种基于分块压缩感知的信号重构方法 | |
| CN103542261B (zh) | 基于压缩感知和掩膜信号法hht的管道泄漏声发射信号处理方法 | |
| CN109613473A (zh) | 基于稀疏性的展开互质线阵角度估计方法 | |
| CN110311686A (zh) | 一种压缩感知的伪随机等效采样信号重构方法 | |
| CN113884996A (zh) | 一种雷达专用测试设备多时宽脉冲信号校正方法和系统 | |
| Angrisani et al. | Frequency domain measurement node based on compressive sampling for sensors networks | |
| CN108132383B (zh) | 一种时间交错的多谐波信号欠采样方法 | |
| CN106872778B (zh) | 基于压缩感知的微弱信号检测方法 | |
| WO2020080143A1 (ja) | 計算機、センシングシステム、及びデータ通信方法 | |
| CN103391099B (zh) | 适用于一维缓变信号的随机采样器 | |
| CN103281087A (zh) | 基于多级压缩感知的信号采样系统及方法 | |
| CN103412189B (zh) | 一种用于电学层析成像系统的信息滤波解调方法 | |
| CN107493106A (zh) | 一种基于压缩感知的频率和角度联合估计的方法 | |
| CN103955956B (zh) | 一种面向压缩感知的图像联合重构方法 | |
| CN107612555A (zh) | 一种基于二分法的改进稀疏度自适应匹配追踪算法 | |
| Duarte | Localization and bearing estimation via structured sparsity models | |
| CN104391040A (zh) | 一种相控阵结构健康监测Lamb波压缩及重构方法 | |
| CN104242947A (zh) | 一种基于Haar小波树的SAMP重构方法 |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| 121 | Ep: the epo has been informed by wipo that ep was designated in this application |
Ref document number: 13886623 Country of ref document: EP Kind code of ref document: A1 |
|
| NENP | Non-entry into the national phase |
Ref country code: DE |
|
| 122 | Ep: pct application non-entry in european phase |
Ref document number: 13886623 Country of ref document: EP Kind code of ref document: A1 |