CN103856972A - Wideband spectrum sensing method, device, corresponding user equipment and base station - Google Patents

Abstract

The invention relates to a wideband spectrum sensing method based on sub-Nyquist sampling and cycle performance detection and a device thereof, user equipment which is applied in a wireless communication network and contains the device and a base station. The method comprises the following steps: an input signal is converted to a first output signal through a sub-Nyquist sampling module; a first cycle cross-spectral density vector or first cycle cross-correlation function vector of the first output signal is calculated by a cycle cross-spectral density or cycle cross-correlation function generator; a second cycle cross-spectral density vector or second cycle cross-correlation function vector of a signal obtained after spectrum displacement and lowpass filtering of the input signal is calculated according to the first cycle cross-spectral density vector or first cycle cross-correlation function vector by a sparse signal recovery technology; and thresholding is conducted on the second cycle cross-spectral density vector or second cycle cross-correlation function vector to obtain an occupied sub-band.

Description

Broader frequency spectrum cognitive method and device and corresponding subscriber equipment and base station

Technical field

The present invention relates to radio network technique, particularly, relate to broader frequency spectrum cognitive method and device and corresponding subscriber equipment and base station based on Sub-nyquist sampling and cycle characteristics detection.

Background technology

Nowadays, cognitive radio (Cognitive Radio:CR) obtains more and more general concern because it can more effectively utilize limited frequency spectrum resource.A crucial technology that realizes cognitive radio is that broader frequency spectrum detects: cognitive radio users need to identify idle sub-band for it in a very large bandwidth range.Traditional broader frequency spectrum detection technique arranges one group of adjustable narrow band filter identifies the situation that takies of each sub-frequency bands, and whether such method once can only be identified a sub-frequency bands occupied, has brought larger time delay to frequency spectrum detection process.Whether occupied in order once to detect multiple sub-bands, a kind of method is that multiple narrow radio frequency passages are set, the system complexity that this has brought huge development cost and increased for system.Another kind method is to adopt the radio-frequency channel in a broadband to process the signal in whole bandwidth.According to Nyquist away from, the sample frequency that this method requires must be the more than 2 times of whole bandwidth.And cognitive radio need to detect very wide bandwidth conventionally, therefore required sample frequency can be very high, at this moment must use analog-to-digital conversion and digital signal processing appts at a high speed, for the realization of system has brought the adverse effect of cost aspect.

For fear of the adverse effect of bringing because meeting high sampling rate that Nyquist principle produces, the Sub-nyquist sampling technology based on compressed sensing (Compressive sensing:CS) is good selection.This technology is based on the following fact: taking of broader frequency spectrum is normally sparse.Utilize compressed sensing technology, sparse signal detects required sample rate can be significantly less than nyquist sampling rate.Therefore compressed sensing can reduce the sample rate that broader frequency spectrum detects significantly, thereby greatly reduces the requirement to analog-to-digital conversion and digital signal processing appts.In the Sub-nyquist sampling technology based on compressed sensing, modulation wide-band transducer (ModulatedWideband Converter:MWC) technology is wherein comparatively remarkable one.As shown in Figure 1, input signal generates one group of narrow band signal after modulation wide-band transducer is processed, and this group narrow band signal can be sampled by lower sample rate.(there is no noise effect) in the ideal case, we can recover original sparse broadband signal from the narrow band signal of modulation wide-band transducer output, thereby have realized the broader frequency spectrum perception under Sub-nyquist sampling rate.

But, in real system, due in input signal inevitably with the noise such as white Gaussian noise, and these noises due to modulation wide-band transducer spectral overlay effect accumulate, thereby make the signal to noise ratio of output signal much smaller than the signal to noise ratio of input signal.In order primary user not to be produced to interference, cognitive radio users need to have the ability of the very faint primary user's signal of identification conventionally.In the time that primary user's signal is very faint, the noise storage effect of modulation wide-band transducer will make situation further worsen, and make the primary user's signal in output signal completely be submerged among noise signal and become and can not identify.Therefore, modulation wide-band transducer technology cannot normally be worked because the accumulation of noise becomes in actual applications.

Summary of the invention

According to the understanding of the above-mentioned technical problem to background technology and existence, the present invention has designed a kind of impact that can remove again the noise stack of modulation wide-band transducer when use modulation wide-band transducer utilizes the advantage of its Sub-nyquist sampling, thereby effectively carry out the technology of broader frequency spectrum detection, i.e. broader frequency spectrum cognitive method and device based on Sub-nyquist sampling and cycle characteristics detection.

According to a first aspect of the invention, propose a kind of broader frequency spectrum cognitive method based on Sub-nyquist sampling and cycle characteristics detection, having comprised:

A. input signal is the first output signal through Sub-nyquist sampling module converts;

B. calculated the first recycling cross spectrum density vector or the first recycling cross correlation function vector of described the first output signal by recycling cross spectrum density or recycling cross correlation function generator;

C. utilize sparse signal recovery technology to go out the second recycling cross spectrum density vector or the second recycling cross correlation function vector of the signal after frequency spectrum shift and low-pass filtering of described input signal according to described the first recycling cross spectrum density vector or described the first recycling cross correlation function vector calculation; And

D. the valuation that described the second recycling cross spectrum density vector or described the second recycling cross correlation function vector draw occupied sub-band through thresholding processing.

In one embodiment, described Sub-nyquist sampling module is modulation wide-band transducer.Modulation wide-band transducer (Modulated Wideband Converter) is a kind of form of Sub-nyquist sampling technology, and this technology has low computation complexity and not mating and the advantage such as the robustness of the factor such as hardware is imperfect model.

In one embodiment, the signal to noise ratio of described input signal is lower than-3dB.Even method of the present invention all can be worked well under the even lower state of signal-to-noise of-10dB.

According to a second aspect of the invention, propose a kind of broader frequency spectrum sensing device based on Sub-nyquist sampling and cycle characteristics detection, having comprised:

Sub-nyquist sampling module, it is for being converted to input signal the first output signal;

Recycling cross spectrum density or recycling cross correlation function generator, it is for calculating the first recycling cross spectrum density vector or the first recycling cross correlation function vector of described the first output signal;

Sparse signal recovers module, and it is for utilizing sparse signal recovery technology to go out the second recycling cross spectrum density vector or the second recycling cross correlation function vector of the signal after frequency spectrum shift and low-pass filtering of described input signal according to described the first recycling cross spectrum density vector or described the first recycling cross correlation function vector calculation; And

Thresholding processing module, it is for estimating occupied sub-band according to described the second recycling cross spectrum density vector or described the second recycling cross correlation function vector.

In one embodiment, described Sub-nyquist sampling module is modulation wide-band transducer.Modulation wide-band transducer is a kind of form of Sub-nyquist sampling technology, and this technology has low computation complexity and not mating and the advantage such as the robustness of the factor such as hardware is imperfect model.

In one embodiment, the signal to noise ratio of described input signal is lower than-3dB.Even method of the present invention all can be worked well under the even lower state of signal-to-noise of-10dB.

According to a third aspect of the invention we, proposed subscriber equipment in a kind of cordless communication network, described subscriber equipment comprises according to the broader frequency spectrum sensing device based on Sub-nyquist sampling and cycle characteristics detection described in second aspect present invention.

According to a forth aspect of the invention, provide the base station in a kind of cordless communication network, described base station comprises according to the broader frequency spectrum sensing device based on Sub-nyquist sampling and cycle characteristics detection described in second aspect present invention.

On the whole, the invention provides a kind of broader frequency spectrum cognition technology being operated under Sub-nyquist sampling rate, this technology has low sampling rate simultaneously, high accuracy, the advantages such as low computation complexity and the robustness to the factor such as hardware is imperfect, and can under low signal-to-noise ratio environment, normally work, there is stronger practicality.In addition, the present invention has solved the intrinsic noise stack problem of modulation wide-band transducer technology effectively, makes to modulate wide-band transducer and can really be applied in actual broader frequency spectrum sensory perceptual system.

Brief description of the drawings

By reading the following detailed description to non-limiting example with reference to accompanying drawing, it is more obvious that other features, objects and advantages of the present invention will become.

Fig. 1 shows the schematic diagram of modulation wide-band transducer MWC technology;

Fig. 2 shows the schematic diagram of traditional broader frequency spectrum sensory perceptual system based on modulation wide-band transducer;

Fig. 3 shows the flow chart according to the broader frequency spectrum cognitive method based on Sub-nyquist sampling and cycle characteristics detection of the present invention;

Fig. 4 shows the schematic diagram according to the broader frequency spectrum sensing device based on Sub-nyquist sampling and cycle characteristics detection of the present invention;

Fig. 5 shows the design sketch according to an embodiment of the present invention; And

Fig. 6 shows the design sketch according to another embodiment of the present invention.

In the drawings, run through different diagrams, same or similar Reference numeral represents same or analogous device (module) or step.

Embodiment

For method and apparatus of the present invention being described better and using according to method of the present invention or comprise the subscriber equipment that is applied to cordless communication network and the base station according to device of the present invention, following paper two important prior aries that arrive used herein, are respectively modulation wide-band transducer technology and cycle characteristics detection technique.Do simple introduction referring to accompanying drawing and correlation formula.

Modulation wide-band transducer technology (Modulated Wideband Converter:MWC)

Fig. 1 shows the schematic diagram of modulation wide-band transducer MWC technology.As seen from the figure, first, this transducer comprises M passage, and input signal x (t) is first multiplied by the signal waveform p of one-period at m passage _m(t), the cycle of this signal waveform is T _p=1/f _p, then pass through the low-pass filtering of h (t) and with T _s=1/f _sfrequency sampling obtain output signal.Due to p _m(t) be periodic signal, can be expressed as note

fourier transform be:

${\tilde{X}}_m\left(f\right)={\∫}_{-\∞}^{\∞}{\tilde{x}}_m\left(t\right)e^{-j2\mathrm{\πf\τ}}\mathrm{dt}={\∫}_{-\∞}^{\∞}{x}_m\left(t\right)\left(\underset{l=-\∞}{\overset{\∞}{\mathrm{\Σ}}}{c}_{m,l}{e}^{j\frac{2\mathrm{\π}}{T_p}\mathrm{lt}}\right)e^{-j2\mathrm{\πf\τ}}\mathrm{dt}=\underset{l=-\∞}{\overset{\∞}{\mathrm{\Σ}}}{c}_{m,l}X(f-{\mathrm{lf}}_p)---\left(1\right)$

Note y _m(t) be the output of low pass filter in m passage, its Fourier transform is:

$Y_m\left(f\right)=\underset{l=-L_0}{\overset{L_0}{\mathrm{\Σ}}}{c}_{m,l}X(f-{\mathrm{lf}}_p),$ -f _s/2≤f≤f _s/2 (2)

The output of the low pass filter of M passage is write as to vector form, is obtained:

$\stackrel{\‾}{Y}\left(f\right)=A\stackrel{\‾}{Z}\left(f\right),$ for-f _s/2≤f≤f _s/2 (3)

Wherein and Z _l(f)=X (f+ (l-L ₀-1) f _p) be the Fourier transform of the signal of input signal x (t) after frequency spectrum shift and low-pass filtering, the element of matrix A is by { cm, l} determines.

In the situation that there is noise, the time-domain representation of (3) formula is:

$\stackrel{\‾}{y}\left(t\right)=A(\stackrel{\‾}{z}\left(t\right)+\stackrel{\‾}{n}\left(t\right)),---\left(4\right)$

Because x (t) has sparse property at frequency domain,

a sparse vector, Ke Yicong

low speed sample in recover.The existence of noise will reduce

the accuracy of recovering.When noise by force to a certain extent, signal

to be submerged in noise Wu Facong completely middle recovery.And modulation wide-band transducer itself has the effect of noise stack, make its sensitiveness to noise stronger, this can find out from (2) formula.As shown in (2) formula, the output signal of modulation wide-band transducer is the stack of input signal different spectral.In this course, the noise on different spectral is also applied.Because signal has frequency-domain sparse, noise does not possess, and the result of spectral overlay makes to modulate the signal to noise ratio of wide-band transducer output will be far below the signal to noise ratio of input.The decline of signal to noise ratio makes to modulate wide-band transducer and is difficult to eventually normal work at a lot of real systems.

In order to address this problem, must introduce new signal processing technology, i.e. cycle characteristics detection technique (Cyclic Feature Detection:CFD).Be defined as follows function:

The Cyclic Autocorrelation Function of signal x (t):

${\hat{R}}_x(\mathrm{\τ},\mathrm{\α})=\underset{T\→\∞}{\lim }\frac{1}{T}{\∫}_{-T/2}^{T/2}E(x(t+\mathrm{\τ}/2)x*(t-\mathrm{\τ}/2))e^{-j2\mathrm{\π\αt}}\mathrm{dt}---\left(5\right)$

The Cyclic Spectrum density of signal x (t):

${\hat{S}}_x(f,\mathrm{\α})={\∫}_{-\∞}^{\∞}{\hat{R}}_x(\mathrm{\τ},\mathrm{\α})e^{-j2\mathrm{\πf\τ}}\mathrm{d\τ}---\left(6\right)$

For cyclo-stationary signal, exist non-vanishing α to make Cyclic Spectrum density or the Cyclic Autocorrelation Function of this signal non-vanishing; And for the stationary signal including white noise, only have in the time that α equals zero, its Cyclic Spectrum density or Cyclic Autocorrelation Function are just non-vanishing.This characteristic detects to very robust of noise cycle characteristics.In foundation method of the present invention, just use this characteristic of Cyclic Spectrum density or Cyclic Autocorrelation Function to resist the noise in real system.Its effect is very obvious.

Although cycle characteristics detects, noise is had to stronger robustness, this technology can not simply and be modulated wide-band transducer in conjunction with the noise stack problem that solves the latter, need first from the output of modulation wide-band transducer, to recover signal x (t) because calculate Cyclic Autocorrelation Function or the Cyclic Spectrum density of signal x (t), and noise stack problem makes this recovery process be difficult to realize.

The present invention proposes a kind of technology, effectively combine modulation wide-band transducer and cycle characteristics and detect.First, Cyclic Autocorrelation Function and cyclic spectral density function are expanded between two signals:

Signal x ₁and x (t) ₂(t) the recycling cross correlation function between:

${\hat{R}}_{x_1,x_2}(\mathrm{\τ},\mathrm{\α})=\underset{T\→\∞}{\lim }\frac{1}{T}{\∫}_{-T/2}^{T/2}E\left(x_1(t+\mathrm{\τ}/2)\right)x_2^{*}(t-\mathrm{\τ}/2)e^{-j2\mathrm{\π\αt}}\mathrm{dt}---\left(7\right)$

Signal x ₁and x (t) ₂(t) the recycling cross spectrum density between:

${\hat{S}}_{x_1,x_2}(f,\mathrm{\α})={\∫}_{-\∞}^{\∞}{\hat{R}}_{x_1,x_2}(\mathrm{\τ},\mathrm{\α})e^{-j2\mathrm{\πf\τ}}\mathrm{d\τ}---\left(8\right)$

In above formula (7) and (8), subscript x can be replaced by y/z/n etc. too.

The technology that the present invention proposes is not directly recovered input signal x (t), but calculate the first recycling cross correlation function or the first recycling cross spectrum density of modulating wide-band transducer output signal, then by the second recycling cross correlation function or the second recycling cross spectrum density of certain distortion of estimating input signal x (t) in described the first recycling cross correlation function or the first recycling cross spectrum density, finally the second recycling cross correlation function or the second recycling cross spectrum density are compared and drawn the valuation of occupied sub-band with a predetermined threshold.Due to the recovery of having avoided input signal x (t), this technology, to very robust of noise, can be operated under very low signal to noise ratio environment.

Fig. 3 shows the flow chart according to the broader frequency spectrum cognitive method based on Sub-nyquist sampling and cycle characteristics detection of the present invention.As shown in the figure, according to the broader frequency spectrum cognitive method based on Sub-nyquist sampling and cycle characteristics detection of the present invention, comprising:

A. input signal is the first output signal through Sub-nyquist sampling module converts;

B. calculated the first recycling cross spectrum density vector or the first recycling cross correlation function vector of described the first output signal by recycling cross spectrum density or recycling cross correlation function generator;

C. utilize sparse signal recovery technology to go out the second recycling cross spectrum density vector or the second recycling cross correlation function vector of the signal after frequency spectrum shift and low-pass filtering of described input signal according to described the first recycling cross spectrum density vector or described the first recycling cross correlation function vector calculation; And

D. the valuation that described the second recycling cross spectrum density vector or described the second recycling cross correlation function vector draw occupied sub-band through thresholding processing.

In order to realize above-mentioned broader frequency spectrum cognitive method, need to walk around the y from containing noise signal _m[n] recovers x (t), but need to utilize sparse signal recovery technology according to y _mthe first recycling cross spectrum density vector of [n] or the first recycling cross correlation function vector calculation go out the signal after frequency spectrum shift and low-pass filtering of described input signal, in (4) formula

, the second recycling cross spectrum density vector or the second recycling cross correlation function vector.At this, need to be by means of the linear relationship that next will set up.

Particularly, y _mand y (t) _n(t) the recycling cross spectrum density between is:

${\hat{S}}_{y_m,y_n}(f,\mathrm{\α})={\∫}_{-\∞}^{\∞}{\hat{R}}_{y_m,y_n}{(t,\mathrm{\τ})e}^{-j2\mathrm{\πf\τ}}\mathrm{d\τ}=\underset{T\→\∞}{\lim }\frac{1}{T}{\∫}_{-\∞}^{\∞}{\∫}_{-T/2}^{T/2}E(y_m(t+\mathrm{\τ}/2)y_n*(t-\mathrm{\τ}/2))e^{-j2\mathrm{\π}(\mathrm{\αt}+\mathrm{f\τ})}\mathrm{dtd\τ}$

$=\underset{T\→\∞}{\lim }\frac{1}{T}{\∫}_{-\∞}^{\∞}{\∫}_{-T/2}^{T/2}{R}_{y_m,y_n}{(t,\mathrm{\τ})e}^{-j2\mathrm{\π}(\mathrm{\αt}+\mathrm{f\τ})}\mathrm{dtd\τ}---\left(9\right)$

Wherein,

$R_{y_m,y_n}(t,\mathrm{\τ})=E(y_m(t+\mathrm{\τ}/2)y_n*(t-\mathrm{\τ}/2))---\left(10\right)$

For traditional cross-correlation function.Can be obtained by above formula (4), (9) and (10):

${\hat{S}}_{y_m,y_n}(f,\mathrm{\α})=\underset{T\→\∞}{\lim }\frac{1}{T}{\∫}_{-\∞}^{\∞}{\∫}_{-T/2}^{T/2}\underset{i,j}{\mathrm{\Σ}}{a}_{m,i}{a}_{n,j}^{*}(R_{z_i,z_j}(t,\mathrm{\τ})+R_{n_i,n_j}(t,\mathrm{\τ}))e^{-j2\mathrm{\π}(\mathrm{\αt}+\mathrm{f\τ})}\mathrm{dtd\τ}$

$=\underset{i,j}{\mathrm{\Σ}}{a}_{m,i}{a}_{n,j}^{*}\underset{T\→\∞}{\lim }\frac{1}{T}{\∫}_{-\∞}^{\∞}{\∫}_{-T/2}^{T/2}(R_{z_i,z_j}(t,\mathrm{\τ})+R_{n_i,n_j}(t,\mathrm{\τ}))e^{-j2\mathrm{\π}(\mathrm{\αt}+\mathrm{f\τ})}\mathrm{dtd\τ}$

$=\underset{i,j}{\mathrm{\Σ}}{a}_{m,i}{a}_{n,j}^{*}({\hat{S}}_{z_i,z_j}(f,\mathrm{\α})+{\hat{S}}_{n_i,n_j}(f,\mathrm{\α}))---\left(11\right)$

Definition

wherein for matrix,

be

(i, j) individual element.Definition for

form after vectorization.Similar definition

with

(11) formula can be expressed as following matrix form:

${\stackrel{\underset{^}{\‾}}{S}}_y=\stackrel{\‾}{A}({\stackrel{\underset{^}{\‾}}{s}}_z+{\stackrel{\underset{^}{\‾}}{s}}_n)---\left(12\right)$

Notice and only have the Z of working as _i(f) ≠ 0 and when i=j, just be not equal to zero, therefore when input signal x (t) is during at frequency-domain sparse, sparse vector, can utilize sparse signal recovery technology from

in estimate.And noise vector in (12) formula power in statistical significance, be zero.Therefore need only observation time long enough, statistics is enough accurate, recovery just not affected by noise.

Thus, this programme is walked around the y from containing noise signal _m[n] recovers x (t), but calculates based on { y _m[n] } the first recycling cross spectrum density vector or the first recycling cross correlation function vector, then utilize the signal after frequency spectrum shift and low-pass filtering that sparse signal recovery technology estimates described input signal from described the first recycling cross spectrum density vector or described the first recycling cross correlation function vector (in (4) formula

) the second recycling cross spectrum density vector or the second recycling cross correlation function vector, next in thresholding treatment step by the second recycling cross spectrum density or the second recycling cross correlation function and a predetermined threshold.If the second recycling cross spectrum density or the second recycling cross correlation function that a certain sub-band is corresponding are greater than predetermined threshold value, this sub-band is identified as occupied state; Otherwise this sub-band is identified as idle condition, can use by perceived radio subscriber.Be formulated, if following formula meets, we will judge that l sub-frequency bands is in occupied state:

$\left\{Z_l\left(f\right)\right|\left|{\hat{S}}_{z_l,z_l}^{\left(m\right)}\right|>T_{\mathrm{Thred}}\}---\left(13\right)$

Introduced the noise vector in formula (12) above

power statistic on be zero,, in the time of observation time long enough, the solution of the present invention will be not affected by noise.But in real system, observation time is limited, therefore noise vector power can not be really zero.The size of noise power depends on the length of observation time, thereby the performance that broader frequency spectrum detects is also subject to the impact of observation time length.Observation time is shorter, adds up more insufficient, and noise power is also just larger, thereby the performance of frequency spectrum detection is also poorer.In actual applications, need appropriate balance detection time and accuracy of detection.

Fig. 4 shows the schematic diagram according to the broader frequency spectrum sensing device based on Sub-nyquist sampling and cycle characteristics detection of the present invention.As shown in the figure, comprise according to the broader frequency spectrum sensing device based on Sub-nyquist sampling and cycle characteristics detection of the present invention: Sub-nyquist sampling module 210, it is for being converted to input signal the first output signal;

Recycling cross spectrum density or recycling cross correlation function generator 220, it is for calculating the first recycling cross spectrum density vector or the first recycling cross correlation function vector of described the first output signal;

Sparse signal recovers module 230, and it is for utilizing sparse signal recovery technology to go out the second recycling cross spectrum density vector or the second recycling cross correlation function vector of the signal after frequency spectrum shift and low-pass filtering of described input signal according to described the first recycling cross spectrum density vector or described the first recycling cross correlation function vector calculation; And

Thresholding processing module 240, it is for estimating occupied sub-band according to described the second recycling cross spectrum density vector or described the second recycling cross correlation function vector.

Device illustrated in fig. 4 is the virtual bench of the method flow shown in execution graph 3, it should be appreciated by those skilled in the art: four parts of this device not necessarily must be divided and to be arranged, and it also can be positioned among a module that can realize above function.

Next will performance of the present invention be described with two embodiment.

In ensuing two embodiment, total bandwidth is 10GHz, and according to traditional Nyquist principle, sample frequency is at least 20GHz so.The bandwidth of every sub-frequency bands is set to B=1GHz, i.e. ten sub-frequency bands altogether, and only have a sub-frequency bands in occupied state at every turn.

Fig. 5 shows the design sketch according to an embodiment of the present invention.In this embodiment, the signal to noise ratio of input signal is-10dB (because the noise stack of MWC makes the signal to noise ratio of the output signal of MWC decline ten times ,-20dB).The port number of MWC is made as M=2, and the sample rate of each passage is f _s=5GHz, now total sample rate is 10GHz, is only the half of traditional nyquist sampling rate.In embodiment illustrated in fig. 5, in the time that observation time N is 100 OFDM symbols, its performance is extraordinary, and the alert rate of mistake and loss are all within 0.1.

Fig. 6 shows the design sketch according to another embodiment of the present invention.In this embodiment, input signal is-3dB, and the port number of MWC is made as M=1, and the sample rate of each passage is f _s=5GHz, therefore total sample rate is 5GHz, is only 25% of traditional nyquist sampling rate.In embodiment illustrated in fig. 6, in the time that observation time N is 100 OFDM symbols, also can obtain good performance.

Certainly,, if observation time N is larger, sampling precision is also just higher.From the embodiment of Fig. 5 and Fig. 6, be under extremely low signal to noise ratio, still normally to work and to there is good performance according to method of the present invention.

On the whole, the invention provides a kind of broader frequency spectrum cognition technology being operated under Sub-nyquist sampling rate, this technology has low sampling rate simultaneously, high accuracy, the advantages such as low computation complexity and the robustness to the factor such as hardware is imperfect, and can under low signal-to-noise ratio environment, normally work, there is stronger practicality.In addition, the present invention has solved the intrinsic noise stack problem of modulation wide-band transducer technology effectively, makes to modulate wide-band transducer and can really be applied in actual broader frequency spectrum sensory perceptual system.。

In the specific descriptions of following preferred embodiment, with reference to the appended accompanying drawing that forms a part of the present invention.Appended accompanying drawing shows by way of example can realize specific embodiment of the present invention.The embodiment of example is not intended to limit according to all embodiment of the present invention.Be appreciated that not departing under the prerequisite of scope of the present invention, can utilize other embodiment, also can carry out the amendment of structural or logicality.Therefore, following specific descriptions are also nonrestrictive, and scope of the present invention is limited by appended claim.

To those skilled in the art, obviously the invention is not restricted to the details of above-mentioned example embodiment, and in the situation that not deviating from spirit of the present invention or essential characteristic, can realize the present invention with other concrete form.Therefore, in any case, all should regard embodiment as exemplary, and be nonrestrictive.In addition, significantly, " comprising ", other elements and step do not got rid of in a word, and wording " one " is not got rid of plural number.Multiple elements of stating in device claim also can be realized by an element.The first, the second word such as grade is used for representing title, and does not represent any specific order.

Claims (8)

1. the broader frequency spectrum cognitive method (300) based on Sub-nyquist sampling and cycle characteristics detection, comprising:

A. input signal is the first output signal (310) through Sub-nyquist sampling module converts;

B. calculated the first recycling cross spectrum density vector or the first recycling cross correlation function vector (320) of described the first output signal by recycling cross spectrum density or recycling cross correlation function generator;

C. utilize sparse signal recovery technology to go out the second recycling cross spectrum density vector or the second recycling cross correlation function vector (330) of the signal after frequency spectrum shift and low-pass filtering of described input signal according to described the first recycling cross spectrum density vector or described the first recycling cross correlation function vector calculation; And

D. described the second recycling cross spectrum density vector or described the second recycling cross correlation function vector draw occupied sub-band (340) through thresholding processing.

2. method according to claim 1, wherein, described Sub-nyquist sampling module is modulation wide-band transducer.

3. method according to claim 1, wherein, the signal to noise ratio of described input signal is lower than-3dB.

4. the broader frequency spectrum sensing device (400) based on Sub-nyquist sampling and cycle characteristics detection, is characterized in that, comprising:

Sub-nyquist sampling module (410), it is for being converted to input signal the first output signal;

Recycling cross spectrum density or recycling cross correlation function generator (420), it is for calculating the first recycling cross spectrum density vector or the first recycling cross correlation function vector of described the first output signal;

Sparse signal recovers module (430), and it is for utilizing sparse signal recovery technology to go out the second recycling cross spectrum density vector or the second recycling cross correlation function vector of the signal after frequency spectrum shift and low-pass filtering of described input signal according to described the first recycling cross spectrum density vector or described the first recycling cross correlation function vector calculation; And

Thresholding processing module (440), it is for estimating occupied sub-band according to described the second recycling cross spectrum density vector or described the second recycling cross correlation function vector.

5. device according to claim 4, is characterized in that, described Sub-nyquist sampling module is modulation wide-band transducer.

6. device according to claim 4, is characterized in that, the signal to noise ratio of described input signal is lower than-3dB.

7. a subscriber equipment in cordless communication network, is characterized in that, described subscriber equipment comprises according to the broader frequency spectrum sensing device (400) based on Sub-nyquist sampling and cycle characteristics detection described in any one in claim 4 to 6.

8. the base station in cordless communication network, is characterized in that, described base station comprises according to the broader frequency spectrum sensing device (400) based on Sub-nyquist sampling and cycle characteristics detection described in any one in claim 4 to 6.

Images