CN107276629B - Hop cycle estimation method based on time frequency analysis center of gravity of frequency - Google Patents
Hop cycle estimation method based on time frequency analysis center of gravity of frequency Download PDFInfo
- Publication number
- CN107276629B CN107276629B CN201710468758.1A CN201710468758A CN107276629B CN 107276629 B CN107276629 B CN 107276629B CN 201710468758 A CN201710468758 A CN 201710468758A CN 107276629 B CN107276629 B CN 107276629B
- Authority
- CN
- China
- Prior art keywords
- frequency
- time
- gravity
- frequency analysis
- center
- 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.)
- Expired - Fee Related
Links
Classifications
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04B—TRANSMISSION
- H04B1/00—Details of transmission systems, not covered by a single one of groups H04B3/00 - H04B13/00; Details of transmission systems not characterised by the medium used for transmission
- H04B1/69—Spread spectrum techniques
- H04B1/713—Spread spectrum techniques using frequency hopping
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
- H04L25/00—Baseband systems
- H04L25/02—Details ; arrangements for supplying electrical power along data transmission lines
- H04L25/0202—Channel estimation
- H04L25/022—Channel estimation of frequency response
Landscapes
- Engineering & Computer Science (AREA)
- Computer Networks & Wireless Communication (AREA)
- Signal Processing (AREA)
- Power Engineering (AREA)
- Measuring Frequencies, Analyzing Spectra (AREA)
Abstract
The invention discloses a kind of hop cycle estimation methods based on time frequency analysis center of gravity of frequency comprising the step of calculating the time frequency analysis value of n-th of time frequency analysis moment, k-th of frequency interval point of the ultrashort wave frequency hopping signal after raised cosine low-pass filtering calculates the step of center of gravity of frequency of n-th of time frequency analysis window, calculate the step of center of gravity of frequency difference of adjacent time frequency analysis window, calculate the step of estimated value of hop cycle；The present invention characterizes the time frequency analysis result of the window with the energy value of the center of gravity of frequency of time frequency analysis window, with the time frequency analysis aggregation and cross term contradiction for overcoming conventional method, the optimal of the two is reached, and base band molding filtration is reduced significantly to the spread spectrum and flat influence of Frequency Hopping Signal, it avoids Energy maximum value method and divergence problem is estimated for the carrier frequency of base band molding filtration, obtained preferable estimation performance.The present invention is more practical and accurate for the hop cycle estimation of Frequency Hopping Signal, while providing reference for the scouting of Frequency Hopping Signal estimation.
Description
Technical field
The invention belongs to Frequency-hopped Signal Reconnaissance technical fields, and in particular to a kind of cycle-skipping based on time frequency analysis center of gravity of frequency
Phase estimation method.
Background technique
Frequency Hopping Signal belongs to typical non-stationary signal, and domestic and foreign scholars mostly use Time-Frequency Analysis Method to obtain clearly greatly
After time-frequency figure, then estimate its frequency parameter.But current Time-Frequency Analysis Method works as time-frequency there are the contradiction of aggregation and cross term
When aggregation is higher, more cross term can be generated；When cross term is less, time-frequency locality is lower；Both be unable to reach
It is optimal simultaneously.Meanwhile being transmitted for adaptive channel, Frequency Hopping Signal passes through the low passes molding filtration such as raised cosine, base-band signal frequency more
Waveform changes.It is directed to the hop cycle estimation method of Frequency Hopping Signal at present, mostly uses the Energy maximum value in time frequency analysis window
Place frequency carries out fast Fourier (Fast Fourier Transform, FFT) to the carrier frequency sequence and becomes as the window carrier frequency
It changes, obtains hop cycle estimated value；This method estimates performance for the hop cycle of the Frequency Hopping Signal after raised cosine low-pass filtering
It is poor.
Frequency hopping communications is answered due to its excellent low probability of intercept and anti-interference in military communication confrontation field extensively
With；Wherein, the parameter Estimation of Frequency Hopping Signal, especially hop cycle are estimated to have become the emphasis of research.Frequency Hopping Signal belongs to typical case
Non-stationary signal, after domestic and foreign scholars mostly use greatly Time-Frequency Analysis Method to obtain clearly time-frequency figure, then estimate its frequency hopping join
Number.Time-Frequency Analysis Method mainly include with Short Time Fourier Transform (Short Time Fourier Transform, STFT),
Gabor transformation, wavelet transformation and S-transformation etc. are the linear transformation of representative and are distributed (Wigner-Ville with Wigner Weir
Distribution, WVD), puppet WVD (Pseudo Wigner-Ville Distribution, PWVD), smooth-switch method
(Smoothed Pseudo Wigner-Ville Distribution, SPWVD) and using Cohen class as the Cohen class of kernel function
Nonlinear transformation；But there are the contradictions of aggregation and cross term for two alanysis methods.S.Barbarossa believes for single frequency hopping
Number, time-frequency figure is obtained using WVD Time-Frequency Analysis Method, is carried using the frequency in the time frequency window at time-frequency Energy maximum value as the window
Frequently, estimation method when and giving the hop cycle of Frequency Hopping Signal, frequency hopping rate and jump, there is estimation property well when frequency point is less
Energy.On this Research foundation, lot of domestic and foreign scholar is the influence for weakening WVD distribution cross term to Signal parameter estimation, is used
PWVD, SPWVD and Cohen class etc. by when, frequency domain smoothing, adding window inhibit Nonlinear and crossing item.Based on linear transformation when
Frequency analysis method can completely eliminate the influence of cross term, and domestic scholars Chen Li tiger etc. is adopted in " time frequency analysis of Frequency Hopping Signal "
Become in " the Frequency Hopping Signal time frequency analysis based on short-time Hartley transform " using Hartley in short-term with STFT method and sunlight etc.
(Short Time Hartley Transform, STHT) method is changed, but there is a problem of that aggregation is not high.For aggregation and
The contradiction of cross term, some scholars reach the two resultant effect using combination time frequency analysis and rearrangement Time-Frequency Analysis Method, but same
When introduce computationally intensive problem.On the basis of obtaining Frequency Hopping Signal time-frequency figure, mostly use greatly in the time frequency analysis window
Frequency where time-frequency Energy maximum value carries out fast Fourier (Fast Fourier as the window carrier frequency, to the carrier frequency sequence
Transform, FFT) transformation, obtain hop cycle estimated value；Chen Lihu is in its degree doctoral thesis " scouting of Frequency Hopping Signal simultaneously
Technical research " in point out to characterize the carrier frequency of the window with the center of gravity of frequency of different window in same jump for slow frequency hopping system, put down
Its equal center of gravity difference, obtains the estimated value of hop cycle.Feng Tao etc. is " a kind of combination time-frequency distributions are in Frequency Hopping Signal parameter Estimation
Application " in each frequency component of Frequency Hopping Signal is taken out by one group of bandpass filter, calculate separately each signal component
WVD obtain new time-frequency distributions and parameter estimation, key is how to determine bandpass filtering and by the cumulative summation of each WVD
Device number simultaneously designs the bandpass filter for matching each component.Zhang Chaoyang etc. is in " blind separations and parameter blind estimation of multiple frequency-hopping signals "
For the Frequency Hopping Signal of the multiple unknown any Study firsts received, propose a kind of when first separating each signal and carrying out respectively again
Frequency analysis estimates the method for frequency parameter, first using eigenmatrix joint near-optimization (JADE) algorithm separation frequency hopping letter
Number, the parameters such as hop cycle for recycling the SPWVD of multiwindow overlapping to estimate Frequency Hopping Signal.Sha Zhi is superfine " to be based on dilute comb
The Frequency Hopping Signal Time-Frequency Analysis Method of reconstruct " and " the Frequency Hopping Signal time-frequency figure modification method based on time-frequency sparsity " in propose adopt
The time-frequency figure of Frequency Hopping Signal is obtained with the method for sparse reconstruct, then carries out parameter Estimation, but calculation amount is larger.Meanwhile it is above
The Frequency Hopping Signal model that is directed to of time frequency analysis be all ideal rectangle Frequency Hopping Signal model, but the radio frequency frequency of practical Frequency Hopping Signal
Spectrum inevitably will receive baseband signal influence.But Sha Zhichao is in " the Frequency Hopping Signal time-frequency figure amendment based on time-frequency sparsity
Method " in only give Frequency Hopping Signal model including baseband signal, analyse in depth base-band signal spectrum to frequency hopping
The influence of Signal parameter estimation.
Summary of the invention
Technical problem to be solved by the invention is to provide one kind can weaken significantly base-band signal spectrum influence based on
The hop cycle estimation method of time frequency analysis center of gravity of frequency.
Used technical solution is to solve above-mentioned technical problem: a kind of hop cycle based on time frequency analysis center of gravity of frequency
Estimation method comprising following steps:
(1) the ultrashort wave frequency hopping signal s after raised cosine low-pass filtering is calculatednN-th of time frequency analysis moment kth of (τ)
The time frequency analysis value STFT of a frequency interval pointn(n,k)；
In (formula 1),
S is Frequency Hopping Signal power；THFor hop cycle；TrFor the frequency switching time of hop cycle；T0For the take-off moment；fnFor
Frequency hopping instantaneous frequency；θ is Frequency Hopping Signal phase；Δ T is time sampling interval；Δ f is frequency sampling interval；N=1,2,3 ...；
K=1,2,3 ...；g(τ-k(TH-Tr)-T0) it is equivalent low-pass signal after base band molding filtration, g*(τ-i Δ T) is g
The conjugation of (τ-i Δ T)；
In (formula 1), if base band molding filtration is raised cosine low pass molding filtration, after raised cosine low pass molding filtration
Equivalent low-pass signal expression formula be
So (formula 1) is
In (formula 3), B is baseband signal bandwidth, and α is the raised cosine roll off factor；
(2) n-th of time frequency analysis moment were n-th of time frequency analysis window, calculated the center of gravity of frequency of n-th of time frequency analysis window
In (formula 4), fsFor time-frequency sample rate；N is that time frequency analysis window is long, i.e. the time-frequency sampling number of STFT；STFTn 2
(n, k) is the time frequency analysis value of n-th of time frequency analysis moment, k-th of frequency interval point, i.e. energy value；
(3) center of gravity of frequency of n-th of time frequency analysis window is calculatedWith the center of gravity of frequency of (n-1)th time frequency analysis window
Difference, i.e., the center of gravity of frequency difference DELTA f of adjacent time frequency analysis windown-1:
In (formula 5), n=2,3 ...；
(4) the time frequency analysis window number N for including in a jump is calculatedh:
In (formula 6), Δ F is to jump frequency interval, NhInitial value is 0；
If the center of gravity of frequency difference DELTA f of adjacent time frequency analysis windown-1Less than stepped-frequency interval Δ F, then this adjacent time frequency analysis window
It is in same hop cycle；If otherwise the center of gravity of frequency difference DELTA f of adjacent time frequency analysis windown-1Greater than stepped-frequency interval Δ F, then go out
Frequency hopping is showed；
(5) hop cycle T is calculatedHEstimated value
In (formula 7),For NhAssembly average, T be window time span,
Further, the value range of the long N of time frequency analysis window is 16,32,64,128,256,512,1024 ...,
32768。
Further, the time-frequency sample rate fsValue range be 200MHz to 500MHz.Further, the time-frequency
Sample rate fsFor 250MHz.
The beneficial effects of the present invention are: the present invention is directed to the hop cycle estimation method pair based on time frequency analysis Energy maximum value
Frequency Hopping Signal through raised cosine filter estimates degradation problem, proposes the hop cycle estimation based on time frequency analysis center of gravity of frequency
Method, the present invention characterize the time frequency analysis of the window as a result, overcoming tradition side with the energy value of the center of gravity of frequency of time frequency analysis window
Time frequency analysis aggregation defect contradictory with cross term, has reached the optimal of the two in method, and reduces base band molding filter significantly
Wave avoids Energy maximum value method and the carrier frequency of base band molding filtration is estimated to the spread spectrum and flat influence of Frequency Hopping Signal
Divergence problem is counted, preferable estimation performance has been obtained, the present invention is more practical and accurate for the hop cycle estimation of Frequency Hopping Signal,
Reference is provided simultaneously for the scouting estimation of Frequency Hopping Signal.
Detailed description of the invention
Fig. 1 is the time domain waveform of ideal rectangle baseband signal.
Fig. 2 is the time domain waveform of the baseband signal after raised cosine filter.
Fig. 3 is the frequency-domain waveform of baseband signal after ideal rectangle baseband signal and raised cosine filter.
Fig. 4 is using center of gravity of frequency method to the time frequency analysis result of raised cosine filter Frequency Hopping Signal.
Fig. 5 is using Energy maximum value method to the time frequency analysis result of raised cosine filter Frequency Hopping Signal.
Fig. 6 is the T of different time frequency analysis estimation methodsHEstimate performance comparison.
Specific embodiment
The present invention is for the hop cycle estimation method based on time frequency analysis Energy maximum value to the frequency hopping after raised cosine filter
Signal estimates degradation problem, proposes the hop cycle estimation method based on time frequency analysis center of gravity of frequency.Frequency hopping letter is provided first
Number mathematical model, spectral contributions and Energy maximum value estimation method performance of the detailed analysis base band molding filtration to Frequency Hopping Signal
The reason of decline, and according to common Time-Frequency Analysis Method, propose hop cycle algorithm for estimating based on time frequency analysis center of gravity of frequency and
Step finally carries out simulating, verifying, obtains for the faster slow frequency hopping signal of hop rate through raised cosine filter, is based on STFT frequency
The more applicable conclusion of the hop cycle estimation method of center of gravity.
For the Frequency Hopping Signal through the base band molding filtration such as raised cosine, the time-frequency energy value at center carrier frequence is not necessarily most
Greatly, the true carrier frequency of the window is unable to characterize with frequency where time frequency window self-energy maximum value, therefore the present invention is with time frequency analysis
Center of gravity of frequency characterizes the carrier frequency value of the window, by the center of gravity of frequency difference and jump frequency interval of comparison front and back window, determines in a jump
The time frequency window number for including；The product of statistical average window number and each window time span, obtains hop cycle estimated value.Based on when
Steps are as follows for the hop cycle algorithm for estimating of frequency analysis center of gravity of frequency:
(1) signal s is calculatednThe time frequency analysis value STFT at n-th of moment of (τ)n(n,k)；
In formula, S is Frequency Hopping Signal power；THFor hop cycle, THValue range be between 0.5 millisecond to 100 milliseconds, it is right
In ultrashort wave frequency hopping signal THIt is 5 milliseconds；TrFor the frequency switching time of hop cycle, TrIt is related with the frequency synthesizer in radio station, it is right
In the signal T that hop rate is slowerrIt is 0.5 millisecond；T0For the take-off moment, it is determined by available machine time and hop cycle relationship, 0 < T0<
TH；fnFor frequency hopping instantaneous frequency, it is determined by the working frequency in radio station, for ultrashort wave radio set, fnValue range be 30MHz
~88MHz；θ is Frequency Hopping Signal phase, 0 < θ <, 2 π；Δ T is time sampling interval, it is determined by the sample rate of STFT；Δ f is
Frequency sampling interval, it is determined by the sample rate and number of sampling of STFT；N=1,2,3 ... it is n-th of time frequency analysis moment；k
=1,2,3 ... it is k-th of frequency interval point；g(τ-k(TH-Tr)-T0) it is equivalent low communication after base band molding filtration
Number, g*(τ-i Δ T) is the conjugation of g (τ-i Δ T)；Equivalent low-pass signal table after raised cosine low pass molding filtration signal
It is up to formula
The time frequency analysis value STFT at n-th of momentn(n, k) is expressed as
In formula, B is baseband signal bandwidth, and α is the raised cosine roll off factor.
(2) the center of gravity formula of plane lamina
Being located in xOy plane has n particle, they are located at point (x1,y1),(x2,y2),…,(xn,yn) at, quality
Respectively M1,M2,…,Mn, by mechanical knowledge it is found that the barycentric coodinates of the system of material points are
Wherein,For the gross mass of the system of material points；
In plane center of gravity formulaIn, xiFor the x-axis coordinate i=1,2 of i-th piece of infinitesimal ..., n；MiIt is
The weight of i block infinitesimal；
Plane center of gravity formula is used for reference, the center of gravity of frequency of n-th of time frequency analysis window is calculated
In formula, fsFor time-frequency sample rate, N is the long instant frequency sampling points of time frequency analysis window, STFTn 2When (n, k) is n-th
The energy value of k-th of frequency interval point of frequency analysis window.
(3) the center of gravity of frequency difference of (n-1)th time frequency analysis window and n-th of time frequency analysis window is calculated
In formula, n=2,3 ...
(4) according to center of gravity of frequency difference DELTA fn-1, calculate the time frequency analysis window number N for including in a jumph
In formula, Δ F is to jump frequency interval, NhInitial value is 0.
That is, if the center of gravity of frequency difference DELTA f of adjacent time frequency windown-1Less than stepped-frequency interval Δ F, then at this adjacent windows
In same hop cycle；If otherwise the center of gravity of frequency difference DELTA f of adjacent time frequency windown-1Greater than stepped-frequency interval Δ F, there is frequency
Jump.
(5) to NhStatistical average is done, and it is asked to obtain hop cycle estimated value with the product of window time span T
In formula,For NhFor assembly average, window time span T is
In formula, N is the time-frequency sampling number of STFT, generally takes 16,32,64,128,256 ...；fsFor time-frequency sample rate,
250MHz is generally for the time frequency analysis system of ultrashort wave frequency hopping, is specifically determined by system performance.
It should be noted that if based on other time frequency analysis center of gravity of frequency methods estimation hop cycle, it only need to be by algorithm
STFTn(n, k) value is changed to corresponding time frequency analysis as a result, algorithm subsequent processing is constant.
The time frequency analysis of Frequency Hopping Signal:
Frequency Hopping Signal is the signal of carrier wave random jump, and mathematical model is
Wherein, S is Frequency Hopping Signal power, THFor hop cycle, TrFor the frequency switching time of hop cycle, T0For the take-off moment,
fkFor frequency hopping instantaneous frequency, θ is Frequency Hopping Signal phase, g (τ-k (TH-Tr)-T0) it is equivalent low after base band molding filtration
Messenger.
The parameter set of Frequency Hopping Signal is hop cycle, frequency switching time, take-off moment and instantaneous frequency, to Frequency Hopping Signal
Parameter Estimation is estimation parameter set { TH,Tr,fk,T0, but THCorrect estimation be other parameters estimation premise.Meanwhile it jumping
The parameter Estimation of frequency signal depends on the time-frequency figure of Frequency Hopping Signal, but the time-frequency figure of Frequency Hopping Signal can be inevitably by base band
The spectral contributions of signal；Because of the radio spectrum of Frequency Hopping Signal, the frequency spectrum of baseband signal is only subjected to radio frequency and has been moved, shape
Shape is determined by the spectral shape of baseband signal completely, if being not seeking to eliminate the spectral contributions of baseband signal, only with time frequency analysis
Carrier frequency value of the frequency values as the window at window Energy maximum value, it is fine to the ideal signal without base band molding filtration, but nothing
Method correctly estimates the Frequency Hopping Signal after base band molding filtration.
The baseband frequency spectrum of Frequency Hopping Signal:
The common base band formed filter of frequency hopping communications is raised cosine filter at present, and Model in Time Domain is
Wherein, B is baseband signal bandwidth, and α is the raised cosine roll off factor.
Raised cosine molding filtration can be regarded as carrying out rectangular filter to after base band symbol interpolation and raised cosine weighting,
Time response consists of two parts: the time domain of time domain factor sinc (the 4B τ) and raised cosine roll off of ideal rectangle formed filter
The factorThe time domain waveform of ideal rectangle baseband signal is as shown in Figure 1；After raised cosine filter baseband signal when
Domain waveform is as shown in Figure 2；Ideal baseband signal and the frequency-domain waveform of baseband signal after raised cosine filter are as shown in Figure 3.
It can be seen from Fig. 1-3 after raised cosine filter, time domain waveform distorts, spectral sidelobes sharp fall.
This is because front and back symbol association is got up by the weighted superposition to time domain waveform, destroys symbol by raised cosine filter structure
Independence, so that time domain waveform is distorted, and introduce intersymbol interference (Inter Symbol Interference,
ISI)；With the crosstalk of time domain waveform, exchanges for and reduce the excessive problem of ideal rectangle filtering side-lobe energy.Ideal rectangle is believed
Number, spectrum energy concentrates on zero-frequency (for radiofrequency signal, concentrating on carrier frequency), so now for the time-frequency of Frequency Hopping Signal
Analysis result is mostly mountain peak shape (three-dimensional) or contour threadiness (two dimension), can be characterized with frequency where time frequency window self-energy maximum value
The window carrier frequency；But practical Frequency Hopping Signal will necessarily pass through base band molding filtration, and after raised cosine filter, baseband frequency spectrum becomes flat
It is smooth, so that different at passband self-energy maximum value be surely directed at zero-frequency (for radiofrequency signal, different to be surely directed at carrier frequency), institute
To obtain hop cycle estimation by being FFT to frequency where time-frequency Energy maximum value for the Frequency Hopping Signal after raised cosine filter
The method performance of value can sharp fall.
Common Time-Frequency Analysis Method: Frequency Hopping Signal is that typical non-stationary signal should be simultaneously for the frequency spectrum of Frequency Hopping Signal
From time domain and frequency domain；Time frequency analysis can in terms of time and frequency two simultaneously observation signal energy density size,
And they are with the variation of time and frequency；So time frequency analysis is more and more important as the analysis tool of Frequency Hopping Signal, under
Simply introduce common Time-Frequency Analysis Method in face.
Short Time Fourier Transform: the basic thought of STFT is in the frame of conventional Fourier Transform, non-stationary signal
Regard a series of superposition of short-term stationarity signals as, and short-time characteristic is then realized by the adding window in time domain, and passes through translation ginseng
Number is to cover entire time domain.It is defined as
Wherein, h*(τ-t) is the conjugation of window function.
The meaning of STFT may be interpreted as: in time domain window function h*(τ-t) intercept signal s (τ) carries out interception result
The spread spectrum scenarios in the τ moment segment signal can be obtained in FFT；Constantly move window function h*The center of (τ-t)
Obtain the spectrum value not being somebody's turn to do simultaneously；These frequency spectrum value sets, are STFTs(τ,f)。
Cohen class time frequency analysis: numerous Bilinear TFDs are unified into a kind of form by Cohen class time frequency analysis, can be changed
The time-frequency locality of the Linear Time-Frequency Analysis methods such as kind STFT, is defined as
Wherein, φ (ν, τ) is the kernel function of distribution.
Kernel function φ (ν, τ) determines that Cohen is distributed the performance of possible form and time frequency analysis；As φ (ν, τ)=1
When, Cohen distribution becomes WVD distribution；As φ (ν, τ)=h (ν), h (ν) is smooth window function, and Cohen distribution at this time becomes
PWVD distribution, i.e., the convolution of the WVD and window function of non-adding window in frequency domain；As φ (ν, τ)=h (ν) l (τ), it is equivalent to WVD and exists
Time domain and frequency domain do smothing filtering simultaneously, become SPWVD distribution.
Because Cohen class time frequency analysis is converted using quadratic nonlinearity, cross term is introduced while improving aggregation
Interference, how selecting performance, more preferably kernel function is the emphasis for weakening Cohen class time frequency analysis cross term and influencing.
Performance comparison: Chen Lihu is in its academic dissertation " Intelligence Technology of Frequency Hopping Signal is studied " and Xiong Liangcai etc. in " Choi-
The optimization of Williams distribution parameter and its application " in using comentropy be index quantification analyze the aggregations of several time frequency analysis with
Cross term performance, as shown in table 1.
The comentropy of several time frequency analysis of table 1
In table 1, N is the sampling number analyzed in the period, and H, G are corresponding window length.
The entropy of STFT is maximum as shown in Table 1, and aggregation is relatively poor；Frequency division whens for WVD, PWVD and SPWVD etc.
Analysis method when due to increasing, the relevant treatments such as frequency domain filtering, improves its time-frequency locality.
Preferable time frequency analysis would generally consume more operands simultaneously, and Chen Lihu is in " the Intelligence Technology of Frequency Hopping Signal
Research " and " time frequency analysis of Frequency Hopping Signal " while giving the operands of various time frequency analysis and compare, as shown in table 2.
The operand of 2 time frequency analysis of table compares
In table 2, N, H and P definition are consistent with table 1.
As shown in Table 2, long for the sampling number of equal length and time frequency window, the operand of WVD is minimum, but its cross term
It interferes larger, when frequency hopping frequency point and more number of signals, clearly time-frequency figure cannot be obtained.The operand of STFT is relatively
It is small；By increasing time frequency analysis window length, the preferable time-frequency figure of aggregation can be obtained.Simultaneously for Frequency Hopping Signal, need one
The operations such as time frequency analysis and the parameter Estimation of signal are completed in a hop cycle time, operand is major consideration；So fortune
The lesser STFT method of calculation amount is using more.
Hop cycle algorithm for estimating:
For the Frequency Hopping Signal through the base band molding filtration such as raised cosine, the time-frequency energy value at center carrier frequence is not necessarily most
Greatly, it so the true carrier frequency of the window may be unable to characterize with frequency where time frequency window self-energy maximum value, proposes with time frequency analysis
Center of gravity of frequency characterize the carrier frequency value of the window；Center of gravity of frequency difference and Hopping frequencies interval by comparison front and back window, determine one
The time frequency window number for including in jump；The product of statistical average window number and each window time span, obtains hop cycle estimated value.Base
In the hop cycle algorithm for estimating of time frequency analysis center of gravity of frequency, steps are as follows:
(1) the time frequency analysis value STFT of signal s (τ) is calculatedn 2(n,k)；
(2) plane center of gravity formula is used for reference, the center of gravity of frequency of n-th of time frequency analysis window is calculated
Wherein, fsFor time-frequency sample rate, N is that time frequency analysis window is long, STFTn 2(n, k) is n-th of time window, k-th of frequency
Energy value (i.e. spectrogram value) at component.
(3) the center of gravity of frequency difference of former and later two windows is calculated
(4) the time frequency analysis window number N for including in a jump is calculatedh
In formula, Δ F is to jump frequency interval.If the formula meaning is the center of gravity of frequency difference DELTA f of adjacent time frequency windown-1Less than jump
Frequency interval Δ F, then this adjacent windows is in same hop cycle；Conversely, there is frequency hopping.
(5) statistical average NhWith the product of window time span T, hop cycle estimated value is obtained
It should be noted that if based on other time frequency analysis center of gravity of frequency methods estimation hop cycle, it only need to be by STFTn 2
(n, k) value is changed to corresponding time frequency analysis as a result, algorithm subsequent processing is constant.
Emulation experiment:
For the correctness for verifying hop cycle algorithm for estimating, simulink emulation experiment, simulation parameter such as 3 institute of table are built
Show.
3 simulation parameter of table
Testing 1 center of gravity of frequency method and Energy maximum value method influences the time frequency analysis result of raised cosine filter Frequency Hopping Signal:
Center of gravity of frequency method is to the time frequency analysis result of raised cosine filter Frequency Hopping Signal as shown in figure 4, Energy maximum value method pair
The time frequency analysis result of raised cosine filter Frequency Hopping Signal is as shown in Figure 5.As can be seen that the Frequency Hopping Signal after raised cosine filter and
Speech, center of gravity of frequency method specific energy maximum value process have better time-frequency focusing.Center of gravity of frequency method is with the frequency of current time frequency analysis window
Exactly because frequency where rate center of gravity as the time frequency analysis window time frequency analysis as a result, center of gravity relative stability, ensure that
The stability of the analysis result of each time frequency analysis window in identical hop cycle.Energy maximum method is with the energy of each time frequency analysis window
Measure maximum where frequency as the time frequency analysis window time frequency analysis as a result, and because raised cosine filter after, Frequency Hopping Signal frequency
Spectrum becomes flat, and diffusion (as shown in Figures 2 and 3), the different load for being surely directed at the window at time-frequency Energy maximum value occur for frequency spectrum
Frequently, so unstable state, i.e. frequency spread condition occurs in the analysis result of each time frequency analysis window in identical hop cycle.
Carrier frequency of the center of gravity of frequency method using carrier frequency where center of gravity of frequency in the window as the window can weaken spread spectrum and flat bring
Influence.For center of gravity of frequency method, since statistical considerations also will appear the center of gravity calculation deviation of individual windows, but the deviation is much
Less than frequency interval is jumped, it can be classified as a jump signal when carrier frequency judges, do not influence subsequent THEstimation, while from center of gravity of frequency
Time-frequency energy value at place it can also be seen that, carrier frequency where center of gravity is not to also demonstrate and answer where Energy maximum value at carrier frequency
With center of gravity of frequency method specific energy maximum value process more suitable for the Frequency Hopping Signal after analysis raised cosine filter.
The T of the different time frequency analysis estimation methods of experiment 2HEstimate performance:
The T of different time frequency analysis estimation methodsHEstimate that performance comparison is as shown in Figure 6.
The time frequency analysis using STFT time-frequency energy (i.e. spectrogram), WVD and SPWVD is illustrated in figure 6 as a result, then applying
The normalization variance of center of gravity of frequency method estimation hop cycle.As seen from Figure 6, as signal-to-noise ratio (Signal to noise
Ratio, SNR) it is smaller when, center of gravity of frequency method based on STFT estimates better performances, when SNR is larger, is based on WVD and SPWVD
Center of gravity of frequency method estimate better performances.This is because main based on estimation performance of the time frequency analysis center of gravity of frequency method to hop cycle
It is restricted by three factors: first is that the correct position of noise jamming center of gravity of frequency, second is that the cross term of time frequency analysis, third is that time-frequency
The aggregation of analysis.When SNR is smaller, the first two is major influence factors, and STFT is not influenced by cross term interference, this
When its estimate better performances；And SPWVD is when passing through, window adding in frequency domain, eliminates the interference of partial intersection item, it estimates performance time at this time
It；And WVD method cross term interference is serious, it estimates that performance is poor at this time.When SNR is larger, latter two be it is main influence because
Element, and after SNR is sufficiently big, it is negligible with the cross term interference of noise, estimate that performance is mainly time-frequency locality shadow at this time
It rings, and reaches limit circle of time-frequency locality；The time-frequency locality of STFT is unable to reach the performance bound of uncertainty principle, when secondary
Frequency analysis such as WVD and SPWVD can achieve the performance bound, so when SNR is larger, the estimation performance based on STFT center of gravity of frequency
It is worse than the estimation performance based on WVD and SPWVD center of gravity of frequency.
The computing overhead of the different time frequency analysis estimation methods of experiment 3:
By the hop cycle algorithm for estimating step based on time frequency analysis center of gravity of frequency of front it is found that for it is different based on when
Frequency analysis center of gravity of frequency method estimates that the algorithm of hop cycle, the computing overhead after obtaining time frequency analysis result are equally, all to pass through
H multiplication, H+3N/H sub-addition are crossed.The operand of different time frequency analysis estimation methods is different, depends primarily on various time-frequencies
The operand of analysis is different.As shown in Table 2, the operand of WVD will be far smaller than the operand of SPWVD, for slow frequency hopping system
(sampling number is less than sampling number in hop cycle in time frequency window), the sampling number N of hop cycle are greater than the sampling number of time frequency window
H, so the operand for STFT is far smaller than WVD, this is because H is remote for the temporal resolution for guaranteeing slow frequency hopping system
Much smaller than N.So the slow frequency hopping system fast for hop rate, the center of gravity of frequency estimation hop cycle based on STFT is more applicable in.
The present invention is for the hop cycle estimation method based on time frequency analysis Energy maximum value to the frequency hopping through raised cosine filter
Signal estimates degradation problem, proposes the hop cycle estimation method based on time frequency analysis center of gravity of frequency.This method subtracts significantly
Weak spread spectrum and flat influence of the base band molding filtration to Frequency Hopping Signal, with the center of gravity of frequency table in certain time frequency analysis window
The carrier frequency for levying the window avoids Energy maximum value method and estimates divergence problem for the carrier frequency of base band molding filtration, obtained preferably
Estimation performance；By comparing the estimation performance and operand of different time frequency analysis, slow frequency hopping system faster for hop rate is obtained
System, the more applicable conclusion of the algorithm based on STFT center of gravity of frequency estimation hop cycle.Mentioned algorithm and conclusion are to form through base band
The parameter Estimation of the Frequency Hopping Signal of filtering and practical Frequency Hopping Signal provides reference.
Embodiment described above is merely a preferred embodiment of the present invention, and the simultaneously exhaustion of non-present invention possible embodiments.
It is any aobvious to made by it under the premise of without departing substantially from the principle of the invention and spirit for persons skilled in the art
And the change being clear to, it should all be contemplated as falling within claims of the invention.
Claims (4)
1. a kind of hop cycle estimation method based on time frequency analysis center of gravity of frequency, it is characterised in that include the following steps:
(1) the ultrashort wave frequency hopping signal s after raised cosine low-pass filtering is calculatednN-th of time frequency analysis moment, k-th of frequency of (τ)
The time frequency analysis value STFT of spaced pointsn(n,k)；
In (formula 1),
S is Frequency Hopping Signal power；THFor hop cycle；TrFor the frequency switching time of hop cycle；T0For the take-off moment；fnFor frequency hopping wink
When frequency；θ is Frequency Hopping Signal phase；Δ T is time sampling interval；Δ f is frequency sampling interval；N=1,2,3 ...；K=1,
2,3,…；g(τ-k(TH-Tr)-T0) it is equivalent low-pass signal after base band molding filtration, g*(τ-i Δ T) is g (τ-
I Δ T) conjugation；
In (formula 1), if base band molding filtration is raised cosine low pass molding filtration, g (τ-k (TH-Tr)-T0) expression formula be
So (formula 1) is then
In (formula 3), B is baseband signal bandwidth, and α is the raised cosine roll off factor；
(2) n-th of time frequency analysis moment were n-th of time frequency analysis window, calculated the center of gravity of frequency of n-th of time frequency analysis window:
In (formula 4), fsFor time-frequency sample rate；N is that time frequency analysis window is long, i.e. the time-frequency sampling number of STFT；STFTn 2(n, k) is
The time frequency analysis value of n-th of time frequency analysis moment, k-th of frequency interval point, i.e. energy value；
(3) center of gravity of frequency of n-th of time frequency analysis window is calculatedWith the center of gravity of frequency of (n-1)th time frequency analysis windowDifference
Value, i.e., the center of gravity of frequency difference DELTA f of adjacent time frequency analysis windown-1:
In (formula 5), n=2,3 ...；
(4) the time frequency analysis window number N for including in a jump is calculatedh:
In (formula 6), Δ F is to jump frequency interval, NhInitial value is 0；
If the center of gravity of frequency difference DELTA f of adjacent time frequency analysis windown-1Less than stepped-frequency interval Δ F, then this adjacent time frequency analysis window is in
In same hop cycle；Conversely, if analyzing the center of gravity of frequency difference DELTA f of frequency window when adjacentn-1Greater than stepped-frequency interval Δ F, then occur
Frequency hopping；
(5) hop cycle T is calculatedHEstimated value
In (formula 7),For NhAssembly average, T be window time span,
2. the hop cycle estimation method according to claim 1 based on time frequency analysis center of gravity of frequency, it is characterised in that described
The value range of the long N of time frequency analysis window is 16,32,64,128,256,512,1024 ..., 32768.
3. the hop cycle estimation method according to claim 1 based on time frequency analysis center of gravity of frequency, it is characterised in that: described
Time-frequency sample rate fsValue range be 200MHz to 500MHz.
4. the hop cycle estimation method according to claim 1 based on time frequency analysis center of gravity of frequency, it is characterised in that: described
Time-frequency sample rate fsFor 250MHz.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201710468758.1A CN107276629B (en) | 2017-06-20 | 2017-06-20 | Hop cycle estimation method based on time frequency analysis center of gravity of frequency |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201710468758.1A CN107276629B (en) | 2017-06-20 | 2017-06-20 | Hop cycle estimation method based on time frequency analysis center of gravity of frequency |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN107276629A CN107276629A (en) | 2017-10-20 |
| CN107276629B true CN107276629B (en) | 2019-07-16 |
Family
ID=60069504
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201710468758.1A Expired - Fee Related CN107276629B (en) | 2017-06-20 | 2017-06-20 | Hop cycle estimation method based on time frequency analysis center of gravity of frequency |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN107276629B (en) |
Families Citing this family (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112929053B (en) * | 2021-03-10 | 2022-02-22 | 吉林大学 | Frequency hopping signal feature extraction and parameter estimation method |
| CN112994741B (en) * | 2021-05-11 | 2021-07-23 | 成都天锐星通科技有限公司 | Frequency hopping signal parameter measuring method and device and electronic equipment |
| CN114050951B (en) * | 2021-10-30 | 2023-05-02 | 西南电子技术研究所（中国电子科技集团公司第十研究所） | Blind estimation method for parameters of ultrashort wave frequency hopping signals |
Family Cites Families (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| FR2903828B1 (en) * | 2006-07-13 | 2009-03-27 | Thales Sa | METHOD FOR TEMPORALLY SYNCHRONIZING AN ORTHOGONAL FREQUENCY EVACUATION WAVY FORM AND DECENTRALIZED TRANSIT TIME MANAGEMENT SATELLITE RADIOCOMMUNICATION SYSTEM. |
| CN103746722B (en) * | 2014-01-02 | 2015-07-01 | 东南大学 | Method for estimating jump cycle and take-off time of frequency hopping signal |
| CN104158561B (en) * | 2014-08-12 | 2016-08-24 | 西安烽火电子科技有限责任公司 | A kind of HF frequency hopping processing method |
| CN105337636B (en) * | 2015-10-08 | 2017-11-21 | 西安电子科技大学 | Asynchronous Frequency Hopping Signal Blind Parameter Estimation based on frequency splicing |
-
2017
- 2017-06-20 CN CN201710468758.1A patent/CN107276629B/en not_active Expired - Fee Related
Also Published As
| Publication number | Publication date |
|---|---|
| CN107276629A (en) | 2017-10-20 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN108469560B (en) | Electromagnetic interference objective complexity evaluation method based on rapid S-transform time-frequency space model | |
| Sejdić et al. | A window width optimized S-transform | |
| Ouyang et al. | Short-time Fourier transform receiver for nonstationary interference excision in direct sequence spread spectrum communications | |
| Chilukuri et al. | Multiresolution S-transform-based fuzzy recognition system for power quality events | |
| CN107276629B (en) | Hop cycle estimation method based on time frequency analysis center of gravity of frequency | |
| Wang et al. | Selection of window length for singular spectrum analysis | |
| CN103746722B (en) | Method for estimating jump cycle and take-off time of frequency hopping signal | |
| CN110336587B (en) | Method for acquiring combined time-frequency distribution in multi-frequency-hopping signal reconnaissance | |
| CN103308804A (en) | Method for extracting time-frequency parameters of power quality disturbance signals on basis of fast K-S (Kaiser-S) transformation | |
| CN108594177A (en) | Based on radar signal modulation system analysis method, the signal processing system for improving HHT | |
| CN106209703A (en) | A kind of Frequency Hopping Signal Blind Parameter Estimation and device | |
| Enserink et al. | On detection of cyclostationary signals | |
| CN103139828A (en) | Broadband spectrum sensing device and method | |
| CN109004996A (en) | Signal detecting method based on more sinusoidal windows spectrum peaks | |
| Nayak et al. | Advanced signal processing techniques for feature extraction in data mining | |
| CN102122972A (en) | Transform-domain-based narrowband interference inhibiting method in shortwave spread spectrum communication | |
| Wang et al. | An adaptive time-frequency filtering algorithm for multi-component LFM signals based on generalized S-transform | |
| CN103078661A (en) | Spread spectrum system interference inhibition method based on iteration threshold | |
| Liang et al. | An individual emitter recognition method combining bispectrum with wavelet entropy | |
| Li et al. | Interference classification and identification of TDCS based on improved convolutional neural network | |
| Yang et al. | Parameter estimation of frequency hopping signals based on time frequency analysis | |
| Yang et al. | A parameter estimation algorithm for frequency-hopping signals with a stable noise | |
| Al-Khafaji et al. | Frequency estimation of FM signals under non-Gaussian and colored noise | |
| Ma et al. | An adaptive method of parameters selection for SPWVD | |
| Shen et al. | Radar Waveform Recognition Based on Time-frequency Images Under Deteriorating Channel Conditions |
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 | ||
| CF01 | Termination of patent right due to non-payment of annual fee | ||
| CF01 | Termination of patent right due to non-payment of annual fee |
Granted publication date: 20190716 Termination date: 20200620 |