CN104280611A  Highfrequency signal spectrum sensing method based on stochastic resonance systems  Google Patents
Highfrequency signal spectrum sensing method based on stochastic resonance systems Download PDFInfo
 Publication number
 CN104280611A CN104280611A CN201410444311.7A CN201410444311A CN104280611A CN 104280611 A CN104280611 A CN 104280611A CN 201410444311 A CN201410444311 A CN 201410444311A CN 104280611 A CN104280611 A CN 104280611A
 Authority
 CN
 China
 Prior art keywords
 stochastic resonance
 signal
 resonance system
 frequency
 primary user
 Prior art date
Links
 238000001228 spectrum Methods 0 abstract title 2
 238000006243 chemical reaction Methods 0 abstract 1
 230000001965 increased Effects 0 abstract 1
 230000002829 reduced Effects 0 abstract 1
Abstract
Description
Technical field
The highfrequency signal frequency spectrum sensing method of a kind of cascade stochastic resonance system based on change of scale that the present invention proposes, is applicable to the detection probability and the success ratio that improve primary user's signal, belongs to cognition wireless electrical domain.
Background technology
Recent years, input science is in ceaselessly progress and development, and input refers to the extraction be in feeble signal under strong neighbourhood noise, detection.At present, for input, domestic and international each researcher is proposed oneself method, such as noise filtering, signal amplifies and optimum matching detects etc.But the detection for feeble signal also has signal to be in detection under strong noise environment, and traditional signal detecting method is just often felt simply helpless.At present, correlative study shows to use nonlinear theory and technology more efficiently can process the test problems of feeble signal, particularly Stochastic Resonance Theory (stochastic resonance, SR) theoretical and chaos (chaos) theory, the detection for input especially feeble signal provides new method.Stochastic Resonance Theory describes a kind of nonlinear principle, namely when a comparatively faint signal is surrounded compared with very noisy, if by a nonlinear system, and feeble signal, neighbourhood noise and nonlinear system three are when reaching certain matching relationship, the energy of neighbourhood noise can to the energy trasfer of feeble signal, in other words, the energy dropoff of neighbourhood noise while the energy of feeble signal strengthens, this just can improve the signal to noise ratio (S/N ratio) of mixed signal.
When Stochastic Resonance Theory is used for Detection of Weak Signals, process and the common ground of other signal detecting methods signal to noise ratio (S/N ratio) that to be final goals be all in order to improve mixed signal of sampling, be convenient to detect feeble signal.And maximum difference is Stochastic Resonance Theory utilizes noise, and noise energy is shifted, and on the contrary, traditional signal detecting method major part is the mode by attenuating noise or noise isolation.The Stochastic Resonance Theory Land use systems different to noise is that the test problems of feeble signal provides new thinking.
But, according to linear response theory and adiabatic approximation theory, traditional stochastic resonance system is only applicable to the detection of lowfrequency cycle primary user signal, although there is adaptive algorithm, stochastic resonance system is made to be applicable to the environment of different noise intensity, but once the frequency of primary user's signal becomes higherfrequency (frequency is greater than 1Hz), Stochastic Resonance Theory is just no longer applicable to the detection of signal.This just needs the method introducing change of scale, makes traditional stochastic resonance system be applicable to the detection of high frequency primary user signal.But, traditional scale transformation method can reduce the sampling step length of cognitive user, this just has higher requirement to sampling precision and sampling rate, and meanwhile, the less sampling period means that frequency spectrum perception more easily can be subject to the impact of stochastic resonance system output signal highfrequency noises burr.
Summary of the invention
Technical scheme: based on stochastic resonance system highfrequency signal frequency spectrum sensing method detect high frequency primary user's signal time key step as follows:
1). cognitive user carries out perception to surrounding environment, and the mixed signal sampled is sent to fusion center;
2). fusion center is estimated primary user's signal frequency that may exist in mixed signal;
3). according to the flow process of change of scale, mixed signal is amplified, and the parameter of stochastic resonance system at different levels is adjusted;
4). sampling mixed signal is sent to the stochastic resonance system of cascade by fusion center, processes;
5). fast fourier transform is carried out to the output of second level stochastic resonance system, and uses traditional frequency spectrum perception algorithm to detect primary user's signal.
Beneficial effect: the present invention compared with prior art, has the following advantages:
Be not suitable for the problem detecting highfrequency signal for traditional stochastic resonance system, first the method utilizes change of scale, makes stochastic resonance system be applicable to high frequency master and detects, decrease detecting period simultaneously, improve perception efficiency.Then be further processed for stochastic resonance system output signal, high frequency component signal reduced further, makes output signal " smooth " more, improve the detection probability to primary user's signal and success ratio.
Accompanying drawing explanation
Fig. 1 n level of the present invention cascade stochastic resonance system model.
Fig. 2 highfrequency signal frequency spectrum sensing method flow process based on accidental resonance of the present invention.
Specific embodiments
1. change of scale
The method of mathematics mesoscale conversion is incorporated in Stochastic Resonance Theory, makes traditional stochastic resonance system be applicable to the detection and treatment of actual wireless communication environment mediumhigh frequency primary user signal.
If be applied to by Stochastic Resonance Theory in frequency spectrum perception, can represent with s (t)=Acos ω t the weak periodical primary user signal treating perception, wherein A represents the amplitude of primary user's signal, and ω represents the frequency of primary user's signal.Γ (t) represents that around cognitive user, average is 0, noise intensity is the white Gaussian noise of D, and x (t) represents the sampling mixed signal after stochastic resonance system process, and a, b are two variable elements of stochastic resonance system.
Corresponding stochastic resonance system model is as follows:
In above formula, δ (t) is unit impulse function, if a>0 and b>0, can the mode of replacing variable be used to be normalized conversion abovementioned stochastic resonance system model, that is:
τ＝at??(3)
Formula (2) and formula (3) are substituted into formula (1) can obtain:
Wherein, noise meet:
Therefore, derivation can obtain:
Wherein, < ξ (τ) >=0, < ξ (τ), ξ (0) >=δ (τ), namely ξ (τ) represents that average is 0, and variance is the white Gaussian noise of 1.
Formula (6) is substituted into formula (4) can obtain:
By above formula abbreviation, can obtain:
Contrast (8) and formula (1), can discoverable type (8) be the normalized form of formula (1), these two formulas be of equal value in form.Meanwhile, in formula (8), the frequency of primary user's signal is ω/a, compared with before converting with normalization, and the 1/a before having become.This explanation, if want to utilize stochastic resonance system to detect primary user's signal of high frequency, can according to the principle of the conversion of normalization above, in the mode of change of scale, proportionally adjust, amplify parameter a and the b of original stochastic resonance system, make high frequency primary user signal normalization become the signal of low frequency, Stochastic Resonance Theory so just can be made to be applicable to the detection of highfrequency signal, adjust as follows:
First, the reference foundation of one group of best stochastic resonance system parameter as change of scale is chosen.Each parameter is as follows: primary user's signal amplitude A _{0}, frequency f _{0}; White Gaussian noise expects E=0, variance stochastic resonance system parameter a _{0}, b _{0}; Stochastic resonance system sampling step length h _{0}.
Suppose to treat that perception high frequency primary user signal frequency is f, the mixing sampled signal treating perception carries out fast fourier transform, uses following formula to estimate the frequency f of primary user's signal that may exist in mixed signal:
In above formula, L represents cognitive user total sampling time, and kmax represents sampling mixed signal fast fourier transform maximum amplitude spectrum position, and Gmax represents the amplitude of amplitude spectrum in kmax position.Wherein,  Gmax+1 >= during Gmax1, r=1, otherwise r=1.
After estimating the primary user signal frequency f and may exist, by following parameter a, b and sampling step length h method of adjustment as follows:
When carrying out matlab emulation, because sampling is discrete, the noise variance σ therefore in environment ^{2}relevant with the noise intensity size of reality and sampling step length h two factors:
Be D before change of scale _{0}, σ _{0}; After conversion is D, σ; Noise intensity D=2D after change of scale _{0}b/a ^{2}, in addition, because the frequency of primary user's signal reduces after normalization, therefore sampling step length h also will amplify accordingly, becomes original a doubly, so, through change of scale, the variances sigma of Gaussian noise ^{2}=4D _{0}b/ (a ^{3}h), after change of scale with change of scale before the ratio of variance be:
Can push away Normalized Scale conversion is equivalent to and becomes original by primary user's signal amplitude times, and the variance of neighbourhood noise becomes original 2b/a ^{3}doubly:
According to the analysis of change of scale principle and derivation, the high frequency primary user signal of perception is treated in the inverse transformation utilizing Normalized Scale to convert and stochastic resonance system parameter carries out parameter adjustment, makes stochastic resonance system be applicable to the detection of high frequency primary user signal.
2. cascade stochastic resonance system model
The stochastic resonance system of n level cascade can regard the series connection of n traditional stochastic resonance system as, and the output of upper level stochastic resonance system, as the output of next stage stochastic resonance system, can be expressed as Fig. 1.
The every onelevel of cascade stochastic resonance system is all optimizing again upper level, the accidental resonance characteristic of performance at different levels is identical, and in the signal processing applications of reality, twostage cascade stochastic resonance system structure is relatively simple, effect of optimization is comparatively obvious, therefore applies the most general.Therefore, the present invention starts with from the simplest twolevel concatenation stochastic resonance system, is studied the characteristic of cascade stochastic resonance system and feasibility.Twolevel concatenation stochastic resonance system model can represent with formula below:
Wherein, s (t)=Acos ω t represents the weak periodical primary user signal treating perception; Γ (t) represents that in cognitive user surrounding environment, average is 0, variance is σ ^{2}white Gaussian noise, and d represents noise intensity, and represent that average is 0, variance is the white Gaussian noise of 1; x _{1}with x _{2}represent the sampling mixed signal (x after the first order and second level stochastic resonance system process respectively _{1}input signal as second level stochastic resonance system); A, b are two variable elements of stochastic resonance system, and k is the enlargement factor of cascade system.
Cascade stochastic resonance system is formed by two identical traditional stochastic resonance system cascades, the method coefficient of intermediate amplifier is set to K=1 usually, if the sampling mixed signal of input is very faint, when second level stochastic resonance system exports and is still difficult to detect primary user's signal, K can be set to larger numerical value.
By above to the analysis of stochastic resonance system principle, can know that the mixing sampled signal of cognitive user is after the process that have passed through first order stochastic resonance system, no matter can stochastic resonance system there is best accidental resonance, and the FFT amplitude spectrum of the treated mixed signal that stochastic resonance system exports is made up of two parts: one is the S relevant with cycle primary user signal to be detected _{1}(f), S _{1}f () is that two is Ss relevant with Gaussian noise with primary user's signal frequency in input mixed signal with frequently _{2}(f), and S _{2}f () meets Lorentz distribution:
Wherein, u is relevant with two parameter a, b of stochastic resonance system, and meets u=a/b.F _{0}for the frequency of primary user's signal in cognitive user sampled signal.D is environmental noise level.
Lorentz distribution represents that the FFT amplitude spectrum concentration of energy of output signal is in the lower region of frequency, and it is less at the energy of highfrequency region, this is because Stochastic Resonance Phenomenon not only can make noise energy shift to primary user's signal energy, also can make noise high frequency energy in mixed signal that transfer to a certain degree occurs to low frequency energy direction simultaneously, this transfer can make original Gauss's flat noise be converted into " look " noise, namely compose structure and there occurs change, be no longer uniformly distributed.
By by two traditional stochastic resonance system cascades, the energy of the highfrequency signal in first order stochastic resonance system can being outputed signal shifts to low frequency signal further, high fdrequency component is made also to be that " burr " greatly reduces, this just can make output signal become more smooth, further reduction frequency glitches is on the impact of testing result, and make the comparatively intensive low frequency region of noise energy become narrower, improve the detection probability to primary user's signal.
Claims (3)
Priority Applications (1)
Application Number  Priority Date  Filing Date  Title 

CN201410444311.7A CN104280611A (en)  20140902  20140902  Highfrequency signal spectrum sensing method based on stochastic resonance systems 
Applications Claiming Priority (1)
Application Number  Priority Date  Filing Date  Title 

CN201410444311.7A CN104280611A (en)  20140902  20140902  Highfrequency signal spectrum sensing method based on stochastic resonance systems 
Publications (1)
Publication Number  Publication Date 

CN104280611A true CN104280611A (en)  20150114 
Family
ID=52255686
Family Applications (1)
Application Number  Title  Priority Date  Filing Date 

CN201410444311.7A CN104280611A (en)  20140902  20140902  Highfrequency signal spectrum sensing method based on stochastic resonance systems 
Country Status (1)
Country  Link 

CN (1)  CN104280611A (en) 
Citations (4)
Publication number  Priority date  Publication date  Assignee  Title 

US4659982A (en) *  19840621  19870421  Centre National De La Recherches Scientifique  Microwave apparatus and method of operation to determine position and/or speed of a movable body or discontinuity or change in a material 
CN101191804A (en) *  20071203  20080604  中国人民解放军国防科学技术大学  Selfadapting random resonant weak signal detection method 
CN101825665A (en) *  20100528  20100908  桂林电子科技大学  Method for detecting stochastic resonance transient electromagnetic weak signals 
CN103969505A (en) *  20140506  20140806  四川大学  Stochastic resonance highfrequency weak signal detection method based on interpolation 

2014
 20140902 CN CN201410444311.7A patent/CN104280611A/en not_active Application Discontinuation
Patent Citations (4)
Publication number  Priority date  Publication date  Assignee  Title 

US4659982A (en) *  19840621  19870421  Centre National De La Recherches Scientifique  Microwave apparatus and method of operation to determine position and/or speed of a movable body or discontinuity or change in a material 
CN101191804A (en) *  20071203  20080604  中国人民解放军国防科学技术大学  Selfadapting random resonant weak signal detection method 
CN101825665A (en) *  20100528  20100908  桂林电子科技大学  Method for detecting stochastic resonance transient electromagnetic weak signals 
CN103969505A (en) *  20140506  20140806  四川大学  Stochastic resonance highfrequency weak signal detection method based on interpolation 
NonPatent Citations (1)
Title 

邓辉等: "嵌入式变尺度级联随机共振系统的实现", 《哈尔滨工程大学学报》 * 
Similar Documents
Publication  Publication Date  Title 

Eldar  Compressed sensing of analog signals in shiftinvariant spaces  
Gedalyahu et al.  Timedelay estimation from lowrate samples: A union of subspaces approach  
CN100483139C (en)  Selfadapting random resonant weak signal detection method  
US20040252850A1 (en)  System and method for spectral enhancement employing compression and expansion  
Han et al.  Multifrequency weak signal detection based on wavelet transform and parameter compensation bandpass multistable stochastic resonance  
Zhou et al.  Spectral analysis of polynomial nonlinearity with applications to RF power amplifiers  
Dehghan et al.  Chebyshev finite difference method for Fredholm integrodifferential equation  
RRL  Estimation of reliability in multicomponent stressstrength model: Loglogistic distribution  
Hwang et al.  Use of twostage leastsquares algorithms for identification of continuous systems with time delay based on pulse responses  
Willsey et al.  Quasiorthogonal wideband radar waveforms based on chaotic systems  
Bhrawy et al.  New solutions for (1+ 1)dimensional and (2+ 1)dimensional Kaup–Kupershmidt equations  
CN102141403B (en)  Realtime mixed denoising method based on wavelet threshold, median filtering and mean filtering  
Samuels  Theory of stochastic linear systems with Gaussian parameter variations  
Zhang et al.  Robust filtering under stochastic parametric uncertainties  
Qian  Fractional Brownian motion and fractional Gaussian noise  
CN101527698B (en)  Nonstationary interference suppression method based on HilbertHuang transformation and adaptive notch  
Jiang et al.  Parameterinduced stochastic resonance in an overdamped linear system  
CN101917363B (en)  Method and device for estimating Doppler frequency shift  
US20150207557A1 (en)  Signal reproduction apparatus and signal reproduction method  
Safi et al.  Blind identification in noisy environment of nonminimum phase finite impulse response (FIR) system using higher order statistics  
Chen  Defining urban and rural regions by multifractal spectrums of urbanization  
Mosavi et al.  Jamming mitigation in global positioning system receivers using wavelet packet coefficients thresholding  
CN103414504A (en)  Adaptive beamforming method based on convex combination  
Moore et al.  On nonlinear filters involving transformation of the time variable  
Matsuo et al.  Design of UWB pulses based on Bsplines 
Legal Events
Date  Code  Title  Description 

C06  Publication  
PB01  Publication  
C10  Entry into substantive examination  
SE01  Entry into force of request for substantive examination  
WD01  Invention patent application deemed withdrawn after publication 
Application publication date: 20150114 