CN113162879B - Modulation signal identification method combining feature extraction - Google Patents
Modulation signal identification method combining feature extraction Download PDFInfo
- Publication number
- CN113162879B CN113162879B CN202110479674.4A CN202110479674A CN113162879B CN 113162879 B CN113162879 B CN 113162879B CN 202110479674 A CN202110479674 A CN 202110479674A CN 113162879 B CN113162879 B CN 113162879B
- Authority
- CN
- China
- Prior art keywords
- signal
- order
- fractional
- signals
- wavelet
- 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.)
- Active
Links
Images
Classifications
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
- H04L27/00—Modulated-carrier systems
- H04L27/0012—Modulated-carrier systems arrangements for identifying the type of modulation
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F2218/00—Aspects of pattern recognition specially adapted for signal processing
- G06F2218/02—Preprocessing
- G06F2218/04—Denoising
- G06F2218/06—Denoising by applying a scale-space analysis, e.g. using wavelet analysis
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F2218/00—Aspects of pattern recognition specially adapted for signal processing
- G06F2218/08—Feature extraction
Abstract
The invention discloses a modulation signal identification method combining feature extraction, which is characterized by comprising the following steps of: 1) receiving a signal; 2) signal preprocessing; 3) extracting characteristic parameters; 4) setting a judgment threshold; 5) and (5) classification and identification. The method has the advantages of less required characteristic parameters, concise steps, low complexity, high recognition rate under low signal-to-noise ratio, capability of making up the limitation of single characteristic of high-order cumulative quantity, and suitability for recognition of various different types of modulation signals.
Description
Technical Field
The invention relates to the technical field of communication signal processing, in particular to the technologies of signal receiving, processing, feature extraction, classification and identification and the like, and particularly relates to a modulation signal identification method combining feature extraction.
Background
Nowadays, electromagnetic environments are increasingly complex, modulation types of different frequency bands are increasingly increased, effective modulation type identification can provide more user information for spectrum sensing, identification of communication signals has important status and effect in the fields of communication electronic countermeasure, radio signal management and the like, and research topics which are commonly concerned with in the field of uncooperative communication have been known for many years. In an actual environment, an electromagnetic environment is complex, and electromagnetic waves such as multipath effects, random noise, burst signals and the like may be interwoven together, which brings difficulty to signal identification. Therefore, how to effectively inhibit interference and improve the identification accuracy rate in a low signal-to-noise ratio environment is an important work of the existing non-cooperative communication signal identification.
Currently, modulation identification technologies are mainly divided into two types of methods: maximum likelihood based decision theory recognition and feature extraction based pattern recognition. Based on a decision theory method, the likelihood ratio is compared with a threshold value determined by Bayesian standard to make a decision, so that an optimal solution in Bayesian sense can be obtained, although the method can obtain high accuracy, the calculation complexity is high, complete priori knowledge needs to be obtained, the engineering is difficult to realize, and the efficiency is low; however, the existing common feature-based identification methods include transient features, high-order cumulant, entropy features and the like, but the above features all have some defects, such as that the transient features are susceptible to noise, the high-order cumulant is insensitive to noise, but part of signals cannot be identified through the high-order cumulant. Fractional order wavelet transform is used as a new time-frequency analysis tool, a traditional wavelet transform time-frequency domain signal analysis method is popularized to a time-fractional frequency domain, and signal time-fractional frequency domain localization characteristics are embodied.
Disclosure of Invention
The invention aims to provide a modulation signal identification method combining feature extraction aiming at the defects of the prior art. The method has the advantages of less required characteristic parameters, concise steps, low complexity, high recognition rate under low signal-to-noise ratio, capability of making up the limitation of single characteristic of high-order cumulative quantity, and suitability for recognition of various different types of modulation signals.
The technical scheme for realizing the purpose of the invention is as follows:
a modulation signal identification method combining feature extraction comprises the following steps:
1) signal receiving: the receiver carries out down-conversion, timing synchronization, matched filtering and discretization processing on the received signal to obtain a digital baseband signal sequence, wherein a signal model of the digital baseband signal sequence is shown as a formula (1):
wherein s (n) is a transmission symbol sequence, θ is a phase frequency offset, h (n) is a channel response function, g (n) is additive noise, and P is an amplitude;
2) signal preprocessing: normalizing the maximum value and the minimum value of the baseband signal sequence obtained in the step 1) to obtain a normalized signal sequence, and removing direct current components, as shown in a formula (2):
wherein s is a digital baseband signal sequence received via a receiver endNormalizing the signal energy by the power normalization processing of the formula (2) to obtain
3) Characteristic parameter extraction: performing high-order cumulant calculation and fractional order wavelet transformation on the normalized signals in the step 2), and realizing the method by the following steps:
3-1) extracting high-order cumulant parameters of the baseband signals: first by computing the mixing moment M of order p of the symbol sequencepqAnd further extracting second, fourth and sixth order cumulant features of the signals, wherein for the complex random sequence X (t) with zero mean value independent and same distribution, the p-order mixing moment of X (t) is defined as shown in a formula (3):
Mpq＝E{[x(t)p-qx*(t)q]} (3)，
wherein, p represents the conjugation of the sequence, p is the order, q represents the number of sequences for taking conjugation, E represents the mean value, and for the zero-mean stationary complex random process x (t), the high-order cumulant of the signal according to formula (3) can be represented as:
second order cumulant:
C20＝Cum(x，x)＝M20，
C21＝Cum(x，x*)＝M21，
fourth order cumulant:
sixth-order cumulant:
since the mean gaussian white noise is 0 and the cumulative quantity higher than the second order is 0, the cumulative quantity of the received signal is less affected by noise, and two characteristic parameters based on the high-order cumulative quantity are respectively selected as shown in formula (4) and formula (5):
C1＝|C42|/|C40| (4)，
C2＝|C63|2/|C42|3 (5)；
3-2) Fractional Wavelet Transform (FRWT): the main parameter of FRWT is the transformation order p, when p is 0, FRWT is degenerated into wavelet transformation, i.e. WT; when p is 1, FRWT is a dual frequency transform, i.e., a combination of fractional fourier transform and WT; when p is between 0 and 1, the transform is in a fractional domain, and the current implementation method of the fractional wavelet transform mainly includes two ideas: one is to fuse the wavelet transform and the fractional Fourier transform so as to realize the fractional wavelet transform; the second is time-frequency transformation based on fractional order wavelet packet, and adopts fractional order wavelet transformation method which fuses wavelet transformation and fractional order Fourier transformation, firstly, the Hilbert transformation is carried out on the received digital baseband signal X (t) to obtain analytic signal as shown in formula (6):
in equation (6), x (t) represents a received baseband signal,
expressing impulse response, t is time, tau is time delay, then performing fractional order wavelet transform on r (t), and the implementation process is shown in formula (7):
wherein R (t) is an analytic function, Rp(u) denotes a fractional Fourier transform function, p is the order of the transform, Kp(t, u) is a transformation kernel, which can be expressed as shown in equation (8):
in the formula (8), n represents the value of a natural number, δ (t) represents an impulse function,
the final output of the fractional wavelet transform is shown in equation (9):
wherein S isp(v, τ) represents a fractional order wavelet transform function,
is the mother wavelet function of the wavelet transform;
3-3) extracting the characteristic value of the fractional wavelet transform: obtaining a wavelet coefficient S after fractional wavelet transform, wherein if the length of S is L, the detail component of the signal is shown in a formula (10):
wherein, { h0,h1,...,hNIs the low pass filter coefficient of a fractional wavelet function, dmFor the detail component at the decomposition level m layers, N is the number of decomposition layers, and furthermore, L and N represent natural numbers, and when L is an even number,
when the number of L is an odd number,
constructing characteristic values by using detail components, and performing N-layer fractional wavelet decomposition on the signals, wherein the characteristic value is constructed as shown in a formula (11):
in the formula, d '(m) is a module value of the maximum value in each layer of detail components, and V is a characteristic value obtained by adding d' (m) of each layer after N layers of decomposition;
the eigenvalue obtained by the formula (7) -formula (11) depends on the decomposition layer number, wavelet basis function and transformation order p of the wavelet transformation, and the haar wavelet decomposition layer number is selected as 7 layers, and the order p is 1;
4) setting a decision threshold: calculating characteristic parameters which can be used for identifying the type of a debugging signal through the step 3), then identifying the signal by setting a judgment threshold, wherein theoretical values of high-order cumulant C1 and C2 of the signals can set judgment thresholds th1 to 0.36, th2 to 0.1375, th3 to 24 and th4 to 15, while partial signals such as 2ASK and BPSK, MFSK and 8PSK signals can not be distinguished through high-order cumulant characteristics, so classification is realized by using fractional order wavelet transform characteristics V, Monte Carlo simulation is performed every 2db in a signal-to-noise ratio range from-5 to 15 to obtain characteristic parameter curves of the two groups of signals, and a median of the characteristic parameter curves of two adjacent signals is set as the judgment threshold of the fractional order wavelet transform;
5) classification and identification: the ten signals to be identified are classified as follows:
5-1) dividing BPSK, QPSK, 8PSK, 2FSK, 4FSK, 8FSK, 16QAM, 32QAM, 2ASK and 4ASK signals into signal set 1, signal set 2 and 32QAM signals by adopting characteristic parameters C1, wherein the signal set 1 comprises 8PSK, 2FSK, 4FSK and 8FSK signals, and the signal set 2 comprises BPSK, QPSK, 16QAM, 2ASK and 4ASK signals;
5-2) sequentially identifying 8PSK, 2FSK, 4FSK and 8FSK signals in a signal set 1 by adopting a fractional wavelet transformation characteristic V, and classifying the signal set 2 into 2ASK, 4ASK and a signal set 3, wherein the signal set 3 comprises BPSK, QPSK and 16QAM signals;
5-3) sequentially identifying the signals in the signal set 3 by using the characteristic parameter C2 to finish all ten kinds of signal classification identification.
The technical scheme provides that high-order cumulant features are combined with fractional wavelet transform coefficient features in modulation identification based on combined feature extraction of high-order cumulant and fractional wavelet transform, wherein the fractional wavelet transform is a novel time-frequency analysis method, the time-fractional frequency domain localization features of signals can be extracted, the features are not easily interfered by noise and have strong robustness, the modulation signal identification rate under low signal-to-noise ratio is effectively improved by combining the high-order cumulant and the fractional wavelet transform features, the traditional single modulation identification method based on feature extraction is improved, the identification type number is increased, and ten signals of BPSK, QPSK, 8PSK, 2FSK, 4FSK, 8FSK, 16QAM, 32QAM, 2ASK and 4ASK can be effectively identified.
The method has the advantages of less required characteristic parameters, concise steps, low complexity, high recognition rate under low signal-to-noise ratio, capability of making up the limitation of single characteristic of high-order cumulative quantity, and suitability for recognition of various different types of modulation signals.
Description of the drawings:
FIG. 1 is a schematic flow diagram of an example method;
FIG. 2 is a graph of characteristic values of fractional wavelet transform of the signal set 1 in the embodiment;
FIG. 3 is a graph of characteristic values of fractional wavelet transform of the signal set 2 in the embodiment;
FIG. 4 is a flow chart of classification recognition in the embodiment;
FIG. 5 is a graph showing the comparison of the recognition rates in the examples.
Detailed Description
The invention will be further illustrated, but not limited, by the following description of the embodiments with reference to the accompanying drawings.
Example (b):
referring to fig. 1, a method for identifying a modulation signal by combining feature extraction includes the following steps:
1) signal receiving: the receiver carries out down-conversion, timing synchronization, matched filtering and discretization processing on the received signal to obtain a digital baseband signal sequence, wherein a signal model of the digital baseband signal sequence is shown as a formula (1):
wherein s (n) is a transmission symbol sequence, θ is a phase frequency offset, h (n) is a channel response function, g (n) is additive noise, and P is an amplitude;
2) signal preprocessing: normalizing the maximum value and the minimum value of the baseband signal sequence obtained in the step 1) to obtain a normalized signal sequence, and removing direct current components, as shown in a formula (2):
wherein s is a digital baseband signal sequence received by a receiver end, and the signal energy is normalized by the power normalization processing of the formula (2)
3) Characteristic parameter extraction: performing high-order cumulant calculation and fractional order wavelet transformation on the normalized signals in the step 2), and realizing the method by the following steps:
3-1) extracting high-order cumulant parameters of the baseband signals: first by computing the mixing moment M of order p of the symbol sequencepqAnd further extracting second, fourth and sixth order cumulant features of the signals, wherein for the complex random sequence X (t) with zero mean value independent and same distribution, the p-order mixing moment of X (t) is defined as shown in a formula (3):
Mpq＝E{[x(t)p-qx*(t)q]} (3)，
wherein, p represents the conjugation of the sequence, p is the order, q represents the number of sequences for taking conjugation, E represents the mean value, and for the zero-mean stationary complex random process x (t), the high-order cumulant of the signal according to formula (3) can be represented as:
second-order cumulant:
C20＝Cum(x，x)＝M20，
C21＝Cum(x，x*)＝M21，
fourth order cumulant:
sixth-order cumulant:
since the mean gaussian white noise is 0 and the cumulative quantity higher than the second order is 0, the cumulative quantity of the received signal is less affected by noise, and in this example, two characteristic parameters based on the high-order cumulative quantity are selected as shown in formula (4) and formula (5):
C1＝|C42|/|C40| (4)，
C2＝|C63|2/|C42|3 (5)；
3-2) fractional order wavelet transform FRWT: the main parameter of FRWT is the transformation order p, when p is 0, FRWT is degenerated into wavelet transformation, i.e. WT; when p is 1, FRWT is a dual frequency transform, i.e., a combination of fractional fourier transform and WT; when p is between 0 and 1, the transform is in a fractional domain, and the current implementation method of the fractional wavelet transform mainly includes two ideas: one is to fuse the wavelet transform and the fractional Fourier transform so as to realize the fractional wavelet transform; the second is time-frequency transformation based on fractional order wavelet packet, in this example, a fractional order wavelet transformation method which fuses wavelet transformation and fractional order fourier transformation is adopted, and first hilbert transformation is performed on a received digital baseband signal x (t) to obtain an analytic signal as shown in formula (6):
in equation (6), x (t) represents a received baseband signal,
expressing impulse response, t is time, tau is time delay, then performing fractional order wavelet transform on r (t), and the implementation process is shown in formula (7):
wherein R (t) is an analytic function, Rp(u) denotes a fractional Fourier transform function, p is the order of the transform, Kp(t, u) is a transformation kernel, which can be expressed as shown in equation (8):
in the formula (8), n represents a natural number value, δ (t) represents an impulse function,
the final output of the fractional wavelet transform is shown in equation (9):
wherein S isp(v, τ) represents a fractional order wavelet transform function,
is the mother wavelet function of the wavelet transform;
3-3) extracting the characteristic value of the fractional wavelet transform: obtaining a wavelet coefficient S after fractional wavelet transform, wherein if the length of S is L, the detail component of the signal is shown in a formula (10):
wherein, { h0,h1,...,hNIs the low pass filter coefficient of a fractional wavelet function, dmFor the detail component at the decomposition level m layers, N is the number of decomposition layers, and furthermore, L and N represent natural numbers, and when L is an even number,
when the number of L is an odd number,
constructing characteristic values by using detail components, and performing N-layer fractional wavelet decomposition on the signals, wherein the characteristic value is constructed as shown in a formula (11):
in the formula, d '(m) is a module value of the maximum value in each layer of detail components, and V is a characteristic value obtained by adding d' (m) of each layer after N layers of decomposition;
from the formula (7) -formula (11), it can be seen that the feature value depends on the decomposition level, wavelet basis function, and transformation order p of the wavelet transform, in this example, the haar wavelet decomposition level is 7, and the order p is 1;
4) setting a judgment threshold: the characteristic parameters that can be used for identifying the type of the debug signal are calculated by step 3), wherein the theoretical values of the characteristic parameters C1 and C2 of each signal in the signal set are shown in table 1 below, and then the signals are identified by setting the decision threshold, the theoretical values of the high-order cumulative quantities C1 and C2 of each signal can set the decision threshold th1 to 0.36, th2 to 0.1375, th3 to 24, and th4 to 15, it can be seen from table 1 that partial signals such as 2ASK and BPSK, MFSK and 8PSK signals cannot be distinguished by high order cumulative features, therefore, the fractional order wavelet transform characteristic V is adopted to realize classification, 100 Monte Carlo simulations are carried out every 2db within the range of signal-to-noise ratio from-5 to 15 to obtain characteristic parameter curves of two groups of signals, as shown in figures 2 and 3, setting the median of two adjacent signal characteristic parameter curves as a judgment threshold of fractional order wavelet transformation according to the simulation curve:
TABLE 1 high-order cumulant characteristic parameter theoretical values
5) Classification and identification: referring to fig. 4, the classification process of ten signals to be recognized is as follows:
5-1) dividing BPSK, QPSK, 8PSK, 2FSK, 4FSK, 8FSK, 16QAM, 32QAM, 2ASK and 4ASK signals into signal set 1, signal set 2 and 32QAM signals by adopting characteristic parameters C1, wherein the signal set 1 comprises 8PSK, 2FSK, 4FSK and 8FSK signals, and the signal set 2 comprises BPSK, QPSK, 16QAM, 2ASK and 4ASK signals;
5-2) sequentially identifying 8PSK, 2FSK, 4FSK and 8FSK signals in a signal set 1 by adopting a fractional wavelet transformation characteristic V, and classifying the signal set 2 into 2ASK, 4ASK and a signal set 3, wherein the signal set 3 comprises BPSK, QPSK and 16QAM signals;
5-3) sequentially identifying the signals in the signal set 3 by using the characteristic parameter C2 to finish all ten kinds of signal classification identification. The effectiveness of the method of this example can be verified by the following simulation:
firstly, simulation conditions:
the simulation parameters are set as follows: under the condition of white Gaussian noise, ten signals of BPSK, QPSK, 8PSK, 2FSK, 4FSK, 8FSK, 16QAM, 32QAM, 2ASK and 4ASK are adopted as a signal set to be classified, and the parameters of the ten signals are set as follows:
sampling frequency: fs-120 khz carrier frequency: fc-10 khz symbol rate: rb 2000 bit/s.
Secondly, simulation result analysis:
as shown in fig. 5, in which the classification of the characteristic parameters used in the conventional method divided by 2PSK and 2ASK uses instantaneous amplitude characteristics, MFSK and 8PSK are high-order cumulative characteristics after fourier transform, the classification of other signals is consistent with the high-order accumulative quantity characteristics used by the method of the embodiment, so that the recognition rate is obviously improved by adopting the method of the embodiment, the method is more obvious under low signal-to-noise ratio, the recognition rate of the method is higher than that of the traditional method when the signal-to-noise ratio is lower than 4db, the recognition rate of the method is about ten percent higher than that of the traditional method when the signal-to-noise ratio is-10 and-8 db, the method is twenty percent higher than the traditional method when the signal-to-noise ratio is-6 and-4 db, and then the increase gradually flattens until the increase reaches 4db, and the two methods are close to one hundred percent, thereby showing that the fractional order wavelet transformation characteristic can improve the recognition rate under low signal-to-noise ratio.
Claims (1)
1. A modulation signal identification method combining feature extraction is characterized by comprising the following steps:
1) signal receiving: the receiver carries out down-conversion, timing synchronization, matched filtering and discretization processing on the received signal to obtain a digital baseband signal sequence, wherein a signal model of the digital baseband signal sequence is shown as a formula (1):
wherein s (n) is a transmission symbol sequence, theta is a phase frequency offset, h (n) is a channel response function, g (n) is additive noise, and P is amplitude;
2) signal preprocessing: normalizing the maximum value and the minimum value of the baseband signal sequence obtained in the step 1) to obtain a normalized signal sequence, and removing direct current components, as shown in a formula (2):
wherein s is a digital baseband signal sequence received by a receiver end, and the signal energy is normalized by the power normalization processing of formula (2)To obtain
3) Characteristic parameter extraction: performing high-order cumulant calculation and fractional order wavelet transform on the normalized signals in the step 2), and realizing the method by the following steps:
3-1) extracting high-order cumulant parameters of the baseband signals: first by computing the mixing moment M of order p of the symbol sequencepqAnd further extracting second, fourth and sixth order cumulant features of the signals, wherein for a complex random sequence X (t) with zero mean value independent and same distribution, the p-order mixing moment of X (t) is defined as shown in a formula (3):
Mpq＝E{[x(t)p-qx*(t)q]} (3)，
where p denotes the conjugation of the sequence, p is the order, q is the number of sequences to be conjugated, E is the mean, and for a zero-mean stationary complex random process x (t), the high-order cumulant of the signal according to equation (3) can be expressed as:
second-order cumulant:
C20＝Cum(x，x)＝M20，
C21＝Cum(x，x*)＝M21，
fourth order cumulant:
sixth-order cumulant:
the mean value of the white gaussian noise is 0, the cumulant higher than the second order is 0, and two characteristic parameters based on the high-order cumulant are selected and respectively shown in a formula (4) and a formula (5):
C1＝|C42|/|C40| (4)，
C2＝|C63|2/|C42|3 (5)；
3-2) fractional order wavelet transform FRWT: the main parameter of FRWT is the transformation order p, when p is 0, FRWT is degenerated into wavelet transformation, i.e. WT; when p is 1, FRWT is a dual frequency transform, i.e., a combination of fractional fourier transform and WT; when p is between 0 and 1, the transformation is in a fractional domain, a fractional wavelet transform method which fuses wavelet transform and fractional Fourier transform is adopted, and Hilbert transform is firstly carried out on a received digital baseband signal X (t) to obtain an analytic signal as shown in a formula (6):
in equation (6), x (t) represents a received baseband signal,
expressing impulse response, t is time, tau is time delay, then performing fractional order wavelet transform on r (t), and the implementation process is shown in formula (7):
wherein R (t) is an analytic function, Rp(u) denotes a fractional Fourier transform function, p is the order of the transform, Kp(t, u) is a transformation kernel, which can be expressed as shown in equation (8)：
In the formula (8), n represents a natural number value, δ (t) represents an impulse function,
the final output of the fractional wavelet transform is shown in equation (9):
wherein S isp(v, τ) represents a fractional order wavelet transform function,
is the mother wavelet function of the wavelet transform;
3-3) extracting the characteristic value of fractional wavelet transform: obtaining a wavelet coefficient S after fractional wavelet transform, wherein if the length of S is L, the detail component of the signal is shown in a formula (10):
wherein, { h0,h1,...,hNIs the low-pass filter coefficient of a fractional order wavelet function, dmFor the detail component at the decomposition level m layers, N is the number of decomposition layers, and furthermore, L and N represent natural numbers, and when L is an even number,
when the number of L is an odd number,
and (3) constructing a characteristic value by using the detail component, and performing N-layer fractional order wavelet decomposition on the signal, wherein the characteristic value is constructed as shown in formula (11):
in the formula, d '(m) is a module value of a maximum value in each layer of detail component, V is a characteristic value obtained by adding d' (m) of each layer after N layers of decomposition, the characteristic value obtained by the formula (7) -formula (11) depends on the decomposition layer number, the wavelet basis function and the transformation order p of wavelet transformation, the haar wavelet decomposition layer number is selected to be 7 layers, and the order p is 1;
4) setting a decision threshold: calculating characteristic parameters which can be used for identifying the type of a debugging signal by the step 3), then identifying the signal by setting a judgment threshold, wherein the theoretical values of high-order cumulant C1 and C2 of each signal can be set with judgment thresholds th1 being 0.36, th2 being 0.1375, th3 being 24 and th4 being 15, while signals 2ASK, BPSK, MFSK and 8PSK cannot be distinguished by the characteristics of the high-order cumulant, adopting fractional order wavelet transformation characteristics V to realize classification, carrying out 100 Monte Carlo simulations every 2db in the range of signal-to-noise ratio from-5 to 15 to obtain characteristic parameter curves of two groups of signals, and setting the median of the characteristic parameter curves of two adjacent signals as the threshold of the fractional order wavelet transformation;
5) classification and identification: the ten signals to be identified are classified as follows:
5-1) dividing BPSK, QPSK, 8PSK, 2FSK, 4FSK, 8FSK, 16QAM, 32QAM, 2ASK and 4ASK signals into signal set 1, signal set 2 and 32QAM signals by adopting characteristic parameters C1, wherein the signal set 1 comprises 8PSK, 2FSK, 4FSK and 8FSK signals, and the signal set 2 comprises BPSK, QPSK, 16QAM, 2ASK and 4ASK signals;
5-2) sequentially identifying 8PSK, 2FSK, 4FSK and 8FSK signals in a signal set 1 by adopting a fractional wavelet transformation characteristic V, and classifying the signal set 2 into 2ASK, 4ASK and a signal set 3, wherein the signal set 3 comprises BPSK, QPSK and 16QAM signals;
5-3) sequentially identifying the signals in the signal set 3 by using the characteristic parameter C2 to finish all ten kinds of signal classification identification.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110479674.4A CN113162879B (en) | 2021-04-30 | 2021-04-30 | Modulation signal identification method combining feature extraction |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110479674.4A CN113162879B (en) | 2021-04-30 | 2021-04-30 | Modulation signal identification method combining feature extraction |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN113162879A CN113162879A (en) | 2021-07-23 |
| CN113162879B true CN113162879B (en) | 2022-06-21 |
Family
ID=76872693
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202110479674.4A Active CN113162879B (en) | 2021-04-30 | 2021-04-30 | Modulation signal identification method combining feature extraction |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN113162879B (en) |
Citations (9)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101655834A (en) * | 2009-09-17 | 2010-02-24 | 哈尔滨工业大学 | Signal separation method based on fractional wavelet transform |
| US9768874B1 (en) * | 2013-03-20 | 2017-09-19 | Georgia Tech Research Corporation | System and methods for autonomous signal modulation format identification |
| CN107948107A (en) * | 2017-11-16 | 2018-04-20 | 成都玖锦科技有限公司 | A kind of digital modulation signals sorting technique based on union feature |
| EP3435246A1 (en) * | 2017-07-24 | 2019-01-30 | Tata Consultancy Services Limited | System and method for signal analysis |
| CN110035025A (en) * | 2019-04-22 | 2019-07-19 | 桂林电子科技大学 | A kind of detection recognition method of the multicarrier mixed signal based on direct feature extraction |
| CN110046591A (en) * | 2019-04-22 | 2019-07-23 | 桂林电子科技大学 | A kind of unmanned plane figure communication number based on fractional wavelet transform and remote signal recognition methods |
| CN110059639A (en) * | 2019-04-22 | 2019-07-26 | 桂林电子科技大学 | A kind of Frequency Hopping Signal detection method based on fractional wavelet transform and Hilbert-Huang transform |
| CN111049770A (en) * | 2019-12-06 | 2020-04-21 | 西安电子科技大学 | Modulation signal identification method based on high-order cumulant |
| CN111327554A (en) * | 2020-02-27 | 2020-06-23 | 电子科技大学 | Feature extraction method for digital modulation signal identification |
-
2021
- 2021-04-30 CN CN202110479674.4A patent/CN113162879B/en active Active
Patent Citations (9)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101655834A (en) * | 2009-09-17 | 2010-02-24 | 哈尔滨工业大学 | Signal separation method based on fractional wavelet transform |
| US9768874B1 (en) * | 2013-03-20 | 2017-09-19 | Georgia Tech Research Corporation | System and methods for autonomous signal modulation format identification |
| EP3435246A1 (en) * | 2017-07-24 | 2019-01-30 | Tata Consultancy Services Limited | System and method for signal analysis |
| CN107948107A (en) * | 2017-11-16 | 2018-04-20 | 成都玖锦科技有限公司 | A kind of digital modulation signals sorting technique based on union feature |
| CN110035025A (en) * | 2019-04-22 | 2019-07-19 | 桂林电子科技大学 | A kind of detection recognition method of the multicarrier mixed signal based on direct feature extraction |
| CN110046591A (en) * | 2019-04-22 | 2019-07-23 | 桂林电子科技大学 | A kind of unmanned plane figure communication number based on fractional wavelet transform and remote signal recognition methods |
| CN110059639A (en) * | 2019-04-22 | 2019-07-26 | 桂林电子科技大学 | A kind of Frequency Hopping Signal detection method based on fractional wavelet transform and Hilbert-Huang transform |
| CN111049770A (en) * | 2019-12-06 | 2020-04-21 | 西安电子科技大学 | Modulation signal identification method based on high-order cumulant |
| CN111327554A (en) * | 2020-02-27 | 2020-06-23 | 电子科技大学 | Feature extraction method for digital modulation signal identification |
Non-Patent Citations (7)
| Title |
|---|
| "A novel modulation classification algorithm based on daubechies5 wavelet and fractional fourier transform in cognitive radio";Jian Liu;《2012 IEEE 14th International Conference on Communication Technology》;20130502;全文 * |
| "基于分数阶小波变换的OFDM混合信号识别方法";吴娟;《电视技术》;20201231;第1-9页 * |
| "基于高阶累积量和小波变换的调制识别算法";谭晓衡;《系统工程与电子技术》;20180131;第40卷(第1期);第1-7页 * |
| "基于高阶累积量的调制方式识别研究";李跃;《西南科技大学学报》;20180930;第33卷(第3期);第1-5页 * |
| "多天线多载波信号识别技术";易国顺;《万方学位论文数据库》;20210214;第1-62页 * |
| "多载波混合信号识别与参数估计技术研究";吴娟;《万方学位论文数据库》;20210204;第1-69页 * |
| "跳频信号的检测与识别";吕国裴;《万方学位论文数据库》;20210204;第1-61页 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN113162879A (en) | 2021-07-23 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Chan et al. | Symbol rate estimation by the wavelet transform | |
| CN108134753B (en) | Method for identifying modulation mode of broadband signal | |
| Sills | Maximum-likelihood modulation classification for PSK/QAM | |
| CN108512791B (en) | Satellite-borne AIS demodulation method based on timing frequency offset compensation | |
| CN104618277B (en) | A kind of short-wave OFDM channel estimation methods under arrowband interference | |
| CN103199945B (en) | The recognition methods of cognitive radio signal modulation system when a kind of low signal-to-noise ratio | |
| CN103780462A (en) | Satellite communication signal modulation identification method based on high-order cumulants and spectrum characteristics | |
| CN111935046B (en) | Low-complexity frequency shift keying signal symbol rate estimation method | |
| CN104363194A (en) | PSK (phase shift keying) modulation recognition method based on wave form transformation | |
| CN114422311B (en) | Signal modulation recognition method and system combining deep neural network and expert priori features | |
| CN112702080B (en) | Direct sequence spread spectrum signal and pseudo code estimation method based on K-means algorithm | |
| CN108737302B (en) | Symbol rate estimation method and device for stochastic resonance combined wavelet transform under low signal-to-noise ratio condition | |
| CN106526620B (en) | Narrow-band interference rejection method based on transform domain filtering | |
| CN105516036A (en) | CPFSK modulation identification method | |
| CN113162879B (en) | Modulation signal identification method combining feature extraction | |
| CN104883328A (en) | Self-adaptive shortwave frequency hopping system channel estimation method | |
| CN109039379B (en) | Power line communication signal filtering method | |
| CN112800863B (en) | Time-frequency feature extraction method and system for communication signal modulation pattern recognition | |
| Yu et al. | Modulation recognition of co-channel OQPSK and MSK based on cyclostationarity | |
| Hazza et al. | Robustness of digitally modulated signal features against variation in HF noise model | |
| CN105763499A (en) | CPFSK modulation recognizing method based on signal instantaneous frequency moving average | |
| Sun | MPSK signals modulation classification using sixth-order cumulants | |
| CN112600784B (en) | Large frequency offset bit synchronization method based on quadratic differential correlation | |
| CN115378776A (en) | MFSK modulation identification method based on cyclic spectrum parameters | |
| Sun et al. | Symbol rate estimation using cyclic correlation and Haar wavelet transform |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |