Invention content
The object of the present invention is to provide the extracting methods and extraction system of a kind of ambient noise, can accurately extract background and make an uproarSound.
To achieve the above object, the present invention provides following schemes：
A kind of extracting method of ambient noise, the extracting method include：
Level sample data set is obtained, the level sample data set includes multiple sample points, and the sample point is divided into letterNumber point and noise spot；
Determining multiple sample sets, the sample set includes N number of level sample data and concentrates continuous sample point,First sample point in the sample set is tested point；The two neighboring sample set includes identical sample point；
The matched curve of each sample set is determined according to least square method；
The slope characteristics value of each tested point is determined according to the matched curve；
Multiple starting points and multiple end points, the slope characteristics value of the starting point are determined according to each slope characteristics valueMore than or equal to T, the slope characteristics value of the nearest tested point of starting point described in the front distance positioned at the starting point is less than T, describedThe slope characteristics value of end point is more than-T, and the slope of the nearest tested point of end point described in the front distance positioned at the end point is specialValue indicative is less than or equal to-T, the time sequencing of tested point is referred to before described, T is given threshold；
Determine that the level sample data is concentrated in sample point all between the first signaling point and second signal pointFor signaling point, first signaling point is starting point, and the second signal point is end point, and first signaling point is located at describedBefore second signal point, first signaling point is the starting point nearest apart from the second signal point；
The level numerical value of the signaling point is replaced with to the level numerical value of adjacent noise point, the adjacent noise sample point isThe noise spot nearest apart from first signaling point or second signal point.
Optionally, it before determining multiple sample sets, further includes：
Obtain the most narrow signal bandwidth and resolution bandwidth in frequency domain service signal；
The maximum value N of fitting points is determined according to the most narrow signal bandwidth and resolution bandwidthmax；
Determine the fitting points N, wherein, 3<N<Nmax。
Optionally, most narrow signal bandwidth and resolution bandwidth described in the basis determine the maximum value N of fitting pointsmax, toolBody is：
According to formula Nmax=BWmin/ (2 × RBW) determines the maximum value N of fitting pointsmax；
Wherein, BWminFor most narrow signal bandwidth, RBW is resolution bandwidth.
Optionally, after the level numerical value that the level numerical value of the signaling point is replaced with to adjacent noise point, alsoIncluding：
Calculate the level difference value of two neighboring sample point；
Judge whether the absolute value of the level difference value is more than 5, obtain the first judging result；
When first judging result represents that the absolute value of the level difference value of two neighboring sample point is more than 5, by numerical valueThe value of larger sample point replaces with the value of the smaller sample point of numerical value.
Optionally, it after level sample data set to be measured is obtained, further includes：
The level numerical value of first sample point of the level sample data set is replaced with apart from the first sample pointThe level numerical value of nearest noise spot.
Optionally, the slope characteristics value that each tested point is determined according to the matched curve, specifically includes：
According to formula：Determine tested point at the matched curveSlope characteristics value；
Wherein, KpThe slope characteristics value of tested point for p-th of sample set, (xn, yn) for the in the sample setN sample point, xn=n, ynLevel numerical value for n-th of sample point.
Optionally, 5≤N≤9.
Optionally, N 6.
Optionally, T 0.3.
A kind of extraction system of ambient noise, the extraction system include：
Level sample data set acquisition module, for obtaining level sample data set to be measured, the level sample data setIncluding multiple sample points, the sample point is divided into signaling point and noise spot；
Sample set determining module, for determining multiple sample sets, the sample set includes N number of level sampleContinuous sample point in data set, first sample point in the sample set are tested point；Two neighboring sampleCollection includes identical sample point；
Matched curve determining module, for determining the matched curve of each sample set according to least square method；
Slope characteristics value determining module, for determining the slope characteristics value of each tested point according to the matched curve；
Starting points and end point determining module, for determining multiple starting points and multiple knots according to each slope characteristics valueSpot, the slope characteristics value of the starting point are more than or equal to T, and starting point described in the front distance positioned at the starting point is nearest to be treatedThe slope characteristics value of measuring point is less than T, and the slope characteristics value of the end point is more than-T, described in the front distance positioned at the end pointThe slope characteristics value of the nearest tested point of end point is less than or equal to-T, and the time sequencing of tested point is referred to before described, and T is settingThreshold value；
Signaling point determining module, for determining that the level sample data is concentrated in the first signaling point and second signal pointBetween all sample point be signaling point, first signaling point is starting point, and the second signal point is end point, describedFirst signaling point is located at before the second signal point, and first signaling point is the beginning nearest apart from the second signal pointPoint；
Replacement module, for the level numerical value of the signaling point to be replaced with to the level numerical value of adjacent noise point, the phaseAdjacent noise sample point is the noise spot nearest apart from first signaling point or second signal point.
According to specific embodiment provided by the invention, the invention discloses following technique effects：
The present invention proposes a kind of extracting method and extraction system of ambient noise, is calculated using least square linear curve matchingMethod handles the corresponding level data of frequency point each in frequency range, the slope characteristics value of sample is extracted, by comparing characteristic valueWith the signal determining thresholding of setting, SNR estimation and compensation is realized, the corresponding level value of signal frequency point value is replaced with into neighbouring noise frequency pointCorresponding level value, so as to accurately obtain the ambient noise of the frequency range.
Specific embodiment
Below in conjunction with the attached drawing in the embodiment of the present invention, the technical solution in the embodiment of the present invention is carried out clear, completeSite preparation describes, it is clear that described embodiment is only part of the embodiment of the present invention, instead of all the embodiments.It is based onEmbodiment in the present invention, those of ordinary skill in the art are obtained every other without making creative workEmbodiment shall fall within the protection scope of the present invention.
The object of the present invention is to provide a kind of ambient noise based on least square linear curve matching extracting method andExtraction system is handled the corresponding level data of frequency point each in frequency range using least square linear curve fitting algorithm,The slope characteristics value of sample is extracted, by comparing slope characteristics value and the signal determining thresholding of setting, SNR estimation and compensation is realized, will believeNumber corresponding level value of value of frequency point replaces with the corresponding level value of neighbouring noise frequency point, so as to which the background for accurately obtaining the frequency range is made an uproarSound.
In order to make the foregoing objectives, features and advantages of the present invention clearer and more comprehensible, it is below in conjunction with the accompanying drawings and specific realApplying mode, the present invention is described in further detail.
Fig. 1 is the flow chart of the extracting method of ambient noise of the present invention.As shown in Figure 1, the extracting method includes：
Step 101：Obtain level sample data set；
Step 102：Determine multiple sample sets, the sample set includes N number of level sample data and concentrates continuouslySample point, first sample point in the sample set is tested point；The two neighboring sample set includes identicalSample point.During concrete operations, the tested point of first sample set concentrates first sample point for level sample data, secondThe tested point of sample set concentrates second sample point for level sample data, and so on.
It is further included before step 102：
Step A1：Obtain the most narrow signal bandwidth and resolution bandwidth in frequency domain business；
The business Radio services include：Radio communication service, fixed service, fixed-satellite service, aviationFixed service, intersatellite service, space operation service, mobile service, mobile satellite service, land mobile service, satellite terrestrialMobile service, maritime mobile service, maritime mobile satellite service, port operation service, ship movement service, aviation movement industryBusiness, aviation movement (R) business, aviation movement (OR) business, aeronautical mobile satellite service, satellite aviation movement (R) business, satelliteIt is aviation movement (OR) business, broadcasting service, broadcasting-satellite service, radio determination service, radiodetermination-satellite service, wirelessElectric navigation service, radionavigation-satellite service, maritime radionavigation service, maritime radionavigation-satellite service, aviationAeronautical navigational radio service, aeronautical radionavigation-satellite service, radio location service, radiolocation-satellite service, meteorologyAuxiliary activities, earth exploration-satellite service, meteorological-satellite service, standard frequency and time signal servi, satellite standard frequency andTime signal business, space research service, amateur service, amateur-satellite service, radio astronomy service, safety service and special industryBusiness etc..
Step A2：The maximum value N of fitting points is determined according to the most narrow signal bandwidth and resolution bandwidthmax；SpecificallyFor：The maximum value N of fitting points is determined according to formula (1)max；
Nmax=BWmin/(2×RBW) (1)
Wherein, BWminFor most narrow signal bandwidth, RBW is resolution bandwidth.
Step A3：Determine the fitting points N, wherein, 3<N<Nmax.Preferably, 5≤N≤9.It is furthermore preferred that N=6.
Step 103：The matched curve of each sample set is determined according to least square method.
Step 104：The slope characteristics value of each tested point is determined according to the matched curve.It specifically includes：
Slope characteristics value of the tested point at the matched curve is determined according to formula (2)；
Wherein, KpThe slope characteristics value of tested point for p-th of sample set, (xn, yn) for the in the sample setN sample point, xn=n, ynLevel numerical value for n-th of sample point.
Step 105：Multiple starting points and multiple end points are determined according to each slope characteristics value, the starting point it is obliqueRate characteristic value is more than or equal to T, and the slope characteristics value of the nearest tested point of starting point described in the front distance of the starting point is less than T,The slope characteristics value of the end point is more than-T, and the slope of the nearest tested point of end point described in the front distance of the end point is specialValue indicative is less than or equal to-T, and the time sequencing of tested point is referred to before described；T is given threshold, as signal determining thresholding.
Step 106：It determines that the level sample data is concentrated and is in all between the first signaling point and second signal pointSample point for signaling point, first signaling point is starting point, and the second signal point is end point, first signaling pointBefore the second signal point, first signaling point is the starting point nearest apart from the second signal point；
Step 107：The level numerical value of the signaling point is replaced with to the level numerical value of adjacent noise point, the adjacent noiseSample point is the noise spot nearest apart from first signaling point or second signal point.
Optionally, after step 101, can also include：By first sample point of the level sample data setLevel numerical value replaces with the level numerical value of the noise spot nearest apart from the first sample point.
Optionally, it is further included after step 107：
Step B1：Calculate the level difference value of two neighboring sample point；
Step B2：Judge whether the absolute value of the level difference value is more than 5, obtain the first judging result；
Step B3：When first judging result represents that the absolute value of the level difference value of two neighboring sample point is more than 5,The value of the larger sample point of numerical value is replaced with to the value of the smaller sample point of numerical value.
Step B1- steps B3 is equivalent to filtering, and the signaling point for finding omission replaces with noise spot again.According to《Ultrashort wave frequencySection degree technical norms for measurement (tentative)》, in frequency spectrum occupancy rate measures work, degree statistics thresholding is defined as localityReceiver average noise power level or voltage value more than 5dB, so it is letter that difference, which is more than 5 absolutely there are one point, between consecutive pointsNumber point.
In engineering, the main noise source in monitoring system is thermal noise, and thermal noise is typical white Gaussian noise,Normal Distribution.Noise is to change at random at any time, and noise has piecewise continuity, with making an uproar for radio signal adjacent channelSound is considered continuous, and Signal averaging is on ambient noise, and the frequency spectrum of signal is mutation.Fig. 2 is wide for somewhere frequency modulationBroadcast frequency range spectrogram.
By the initial data of research and application system acquisition, the noise level of adjacent channel is although different and levelSize variation it is non-directional, but amplitude of variation very little.By counting the level value of synchronization on each channel, then drawInto a curve, it can be seen that noise section curve is flatter in entire frequency range, and signal section curve becomes precipitous suddenly, instituteTo be easy to resoluting signal and noise by human eye, and computer can not be according to this phenomenon identification signal and noise.Therefore, may be usedTo extract the feature of curve by some way, by differentiating that specific characteristic value makes computer have recognition capability.
By analysis it is found that signal has obvious characteristic on geometric figure in frequency domain, i.e., there are more apparent for signal both sidesRising edge and failing edge, as shown in Figure 3.Level value amplitude of variation between data sample point in signal rising edge and failing edgeIt is larger, and amplitude variation presents and significantly rises or falls trend.Assuming that in vertical coordinate system, in signal rising edge or declineN number of continuous data sample point is taken in, is carried out curve fitting by least square method linear curve fit algorithm, it can be seen thatThe oblique line inclination angle formed by the fitting of N number of data sample point is larger, such as the θ in Fig. 31；And N number of continuous data are taken in noise segmentAfter sample point is by least square method linear curve fit, the oblique line inclination angle being fitted is smaller or goes to zero, in Fig. 4θ2, therefore, data level sample point collected in the frequency range of ambient noise to be extracted can be divided into multiple length for NSample set, the slope characteristics value K of each sample set is then extracted with least square linear curve matchingi.Setting signalDecision threshold T, works as KiDuring >=T, it is believed that the subsample is signal data point, works as Ki<During T, it is believed that the subsample is noise dataPoint.In addition, as can be seen that a complete signal should have rising edge and failing edge simultaneously, to one section from Fig. 3 and Fig. 4When sample data comprising a signal carries out linear fit, if calculated for the first time to certain subsample { Si,Si+1,…,SN+i-1}Slope characteristics value Ki>=T, then it is assumed that the sample is signal data and belongs to the beginning of the rising edge of signal, marks i=m1, continue linear fit, as slope characteristics value Kj>-T(j>And K 1)j-1During≤- T, then it is assumed that this group of data are signal numberAccording to and belong to signal failing edge latter end, mark j=m2, by sample point S (m1) to S (m2) it is considered as signaling point, value is replacedIt is changed to the value of adjacent noise sample point.It, can be real after being handled using the method the original level sample data of a certain frequency rangeExisting signal frequency point detaches, and then extract ideal ambient noise sample with noise frequency point.
1st, the calculating of slope of curve characteristic value
If it is the set of a N number of point, wherein abscissaIt is to determine.Least square fitting curve y=f (x)=Kx+B is to meet root-mean-square error E2(f) minimum curve.The coefficient of least square curve is that K and B is following linear equation (3)Solution, these equations are known as normal equation:
The slope of place's matched curve can be calculated by linear equation (3)：
2nd, fitting points N is determining
If the noise of each frequency point takes Gaussian Profile, and the noise independent same distribution of continuous N number of frequency point, fixed according to big numberRule can obtain the noise of N number of frequency point also Normal Distribution；It can prove the minimum two of the noise of N number of frequency point of Normal DistributionThe slope characteristics value for multiplying curve matching levels off to zero.When the sample points N of selection is bigger, these samples more level off to normal state pointCloth.But the half that the points of selection will at least be less than signal points could provide ideal condition for the application of this method.It is falseIf the most narrow signal bandwidth in frequency range in all business is BWmin, resolution bandwidth RBW, the points of selection should be less than
Nmax=BWmin/(2*RBW).So the points N for suggesting choosing is finally in section (3, Nmax) in.It is being accounted for simultaneouslyBefore expenditure statistical work, point of receiver is rationally set preferably based on the occupied bandwidth of each service signal in the frequency range to be countedResolution bandwidth, to ensure Nmax>3。
Result extract to ambient noise of fitting points N is affected, and carries out curve fitting when extracting slope characteristics value, ifPoints may excessively lead to the slope characteristics value unobvious of the signaling point extracted, if points it is very few may cause to fluctuate it is largerThe slope characteristics value of adjacent noise spot is rung higher than thresholding feature to be judged, by practical application, fitting points generally take 5~9 effectsFruit is preferable.
3rd, the selection of signal determining thresholding
Signal determining thresholding T should rationally be set, if T settings are excessive, can be caused the relatively low sample of signal point of signal-to-noise ratioIt is determined as noise；If T settings are too small, calculation amount can be caused to increase, while also can be fluctuating larger noise sample number in frequency rangeAccording to being determined as signal.The present invention gives a kind of method of estimation of signal determining thresholding T according to actual monitoring situation.Assuming that prisonMain noise source in examining system is thermal noise, and thermal noise is typical white Gaussian noise, Normal Distribution.Pass through sightFrequency measurement modal data finds the level difference value between adjacent noise point substantially within 1dB.It therefore can be by emulating to signal determining doorLimit T is estimated.Present invention assumes that for the data Normal Distribution of fitting, when N=6 is counted in fitting, using Meng TekaLuo Fangfa has carried out the distribution situation of the slope of curve characteristic value after fitting 100,000 emulation, and simulation result is shown in Fig. 5.By Fig. 5It is found that the slope characteristics value approximation Normal Distribution after the fitting of 6 point curves, the probability that slope characteristics value is respectively less than 0.3 are80%.Therefore, when carrying out ambient noise extraction, it is contemplated that setting T=0.3.By actual tests, the N=6 that counts is being fitted,Decision threshold T=0.3, the ambient noise extracted meet degree statistical work requirement.
Further, since performance indicator has differences between different receiving devices, the signal determining gone out according to simulation estimateThresholding T is not necessarily optimum value, can be in the signal determining thresholding that simulation estimate goes out when practical application the method extracts ambient noiseThe size of T is suitably adjusted on the basis of T, is tested in advance, to determine to be suitble to the optimal decision threshold value of specific receiving device.
4th, the extracting method verification of ambient noise
In order to verify the feasibility of this method and extraction effect, the present invention is monitored using the ESMD of Rohde ＆ Schwarz Representative OfficeReceiver with reference to the monitoring system integrated based on ESMD, constructs a set of monitoring platform, and actual monitoring is carried out in different frequency range,The monitoring data of each frequency range are acquired and preserve, for method call proposed by the present invention processing.In order to which efficient process monitors numberAccording to the present invention has write ambient noise curve extraction procedure in MATLAB development environments, realizes to the automatic of frequency spectrum dataProcessing and visualization.In this experiment, fitting points N=6, signal determining thresholding used in ambient noise automatic Extraction AlgorithmT=0.3.
Fig. 6 for method using the present invention to Xinjiang Urumqi city somewhere collected 3~30MHz shortwaves frequency range dataFrequency spectrum data handled after obtained ambient noise curve.
As shown in fig. 6, the thicker curve in bottom is the ambient noise curve of method extraction.When acquiring this frequency range data,ESMD monitoring receivers are provided that panoramic scanning pattern (pscan), and scanning stepping (step) is set as 0.5kHz, signal levelUnit be dBuV.It can be seen from the figure that ambient noise fluctuations are larger in the frequency range, obtained using artificial interpretation modeAmbient noise difficulty is larger.It can be seen that the ambient noise curve extracted is substantially consistent with real background noise, error is veryIt is small.
Fig. 7 be method using the present invention to the frequency spectrum datas of collected 88~108MHz frequency modulation broadcastings frequency range data intoThe ambient noise curve obtained after row processing.
As shown in fig. 7, the thicker curve in bottom is the ambient noise curve of algorithm extraction.When acquiring this frequency range data,ESMD monitoring receivers are provided that panoramic scanning pattern (pscan), and scanning stepping (step) is set as 3.125kHz, signal electricityFlat unit is dBuV.
Fig. 8 be method using the present invention to the frequency spectrum datas of collected 108~1000MHz ultrashort wave frequency bands data intoThe ambient noise curve obtained after row processing, the thicker curve in bottom are the ambient noise curve of algorithm extraction.Acquire this frequency rangeDuring data, ESMD monitoring receivers are provided that panoramic scanning pattern (pscan), and scanning stepping (step) is set as 12.5kHz,The unit of signal level is dBuV.
Radio background noise curve extracting method proposed by the present invention can accurately extract entire receiver scanning frequencyAmbient noise in section, effect are influenced smaller and real by factors such as hardware device, the distribution of frequency range ambient noise and signal bandwidthsWith, conveniently.This method is had been supplied in our unit's daily radio monitoring work, and it is ideal to be proven effect, solvesThe problem of noise estimation value is relatively low compared with actual value in ITU-RP.372 methods improves ITU-RSM.1753 methods by artificialThe mode of noise sample value is extracted, and improves the degree of automation and survey that radio monitoring intermediate frequency spectrum degree statistics thresholding is setMeasure the accuracy of result.
The present invention also provides a kind of extraction system of ambient noise, Fig. 9 is the extraction system of ambient noise of the present inventionStructure connection figure.As shown in figure 9, the extraction system includes：
Level sample data set acquisition module 901, for obtaining level sample data set to be measured, the level sample dataCollection includes multiple sample points, and the sample point is divided into signaling point and noise spot；
Sample set determining module 902, for determining multiple sample sets, the sample set includes N number of levelSample data concentrates continuous sample point, and first sample point in the sample set is tested point；The two neighboring sampleThis subset includes identical sample point；
Matched curve determining module 903, for determining the matched curve of each sample set according to least square method；
Slope characteristics value determining module 904, for determining the slope characteristics value of each tested point according to the matched curve；
Starting points and end point determining module 905, for determining multiple starting points and more according to each slope characteristics valueA end point, the slope characteristics value of the starting point are more than or equal to T, and starting point described in the front distance positioned at the starting point is nearestThe slope characteristics value of tested point be less than T, the slope characteristics value of the end point is more than-T, positioned at the front distance of the end pointThe slope characteristics value of the nearest tested point of the end point is less than or equal to-T, and the time sequencing of tested point is referred to before described, and T isGiven threshold；
Signaling point determining module 906, for determining that the level sample data is concentrated in the first signaling point and the second letterAll sample points are signaling point between number point, and first signaling point is starting point, and the second signal point is end point,First signaling point is located at before the second signal point, and first signaling point is nearest apart from the second signal pointStarting point；
Replacement module 907, it is described for the level numerical value of the signaling point to be replaced with to the level numerical value of adjacent noise pointAdjacent noise sample point is the noise spot nearest apart from first signaling point or second signal point.
For system disclosed in embodiment, since it is corresponded to the methods disclosed in the examples, so the ratio of descriptionRelatively simple, reference may be made to the description of the method.
Specific case is applied in the present invention to be expounded the principle of the present invention and embodiment, above exampleIllustrate to be merely used to help understand method and its core concept of the invention；Meanwhile for those of ordinary skill in the art, according toAccording to the thought of the present invention, in specific embodiments and applications there will be changes.In conclusion the content of the present specificationIt should not be construed as limiting the invention.