CN106330385B - A kind of interference type recognition methods - Google Patents
A kind of interference type recognition methods Download PDFInfo
- Publication number
- CN106330385B CN106330385B CN201610743463.6A CN201610743463A CN106330385B CN 106330385 B CN106330385 B CN 106330385B CN 201610743463 A CN201610743463 A CN 201610743463A CN 106330385 B CN106330385 B CN 106330385B
- Authority
- CN
- China
- Prior art keywords
- interference
- frequency
- signal
- spectrum
- time domain
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
Classifications
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04K—SECRET COMMUNICATION; JAMMING OF COMMUNICATION
- H04K3/00—Jamming of communication; Counter-measures
- H04K3/20—Countermeasures against jamming
- H04K3/22—Countermeasures against jamming including jamming detection and monitoring
- H04K3/224—Countermeasures against jamming including jamming detection and monitoring with countermeasures at transmission and/or reception of the jammed signal, e.g. stopping operation of transmitter or receiver, nulling or enhancing transmitted power in direction of or at frequency of jammer
Landscapes
- Engineering & Computer Science (AREA)
- Radar, Positioning & Navigation (AREA)
- Remote Sensing (AREA)
- Computer Networks & Wireless Communication (AREA)
- Signal Processing (AREA)
- Noise Elimination (AREA)
Abstract
The invention belongs to fields of communication technology, are related to a kind of interference type recognition methods.The present invention can interfere the hostility in the wireless communication system of complex electromagnetic environment and do Interference Detection and identification, and then joint decision goes out the situation of current electromagnetic environment, and to there is the interference type under interference environment to classify；Interference Detection mainly determines the existence of interference in terms of the time domain and frequency domain two；Disturbance ecology, which is mainly extracted characteristic parameter from time domain, frequency domain, time-frequency domain and will be interfered using decision tree, to be divided into time domain interference, frequency domain interference, frequency sweep three categories and detailed identification has been done in part interference；The interference characteristic parameter that the present invention extracts has preferable robustness and makes an uproar than less sensitive to dry, can also have preferable disturbance ecology performance in the case of decision tree decision method lower using algorithm complexity.
Description
Technical field
The invention belongs to fields of communication technology, are related to a kind of interference type recognition methods.
Background technology
With the fast development of electronic information technology, current communication system is more and more, and communication system is also more and more multiple
It is miscellaneous, and communicate residing electromagnetic environment also increasingly complexity etc..This proposes to various communication systems at anti-interference aspect huge
Challenge, especially in military communication, the interference free performance of communication system may directly control the trend of war, therefore generation
Various countries of boundary are coincidentally studied and are explored to national military communication Anti-Jamming Technique, so as under complicated international-style
Information Superiority is obtained, ensures the military security of this country.Important component of the disturbance ecology as communication countermeasure system, a side
Face can provide foundation for cognition current electromagnetic environment；On the other hand, can be that communication control processor is right for disturbance pattern
After select suitable antijamming measure to provide support, to make antijamming measure more targetedly and preferably improve communication system
Interference free performance.
There are many interference signal type in communication countermeasure, distinguished from different domains, can be divided into time domain interference, frequency domain is done
Disturb, Sweeping nonlinearity, spatial domain interference.On the one hand, due to different types of interference, suppressing method is different, and it is therefore necessary to will not
Congener interference distinguishes to use more targeted disturbance restraining method；On the other hand, when a system includes
When a variety of different types of interference, also there is ratio to the performance of system in the sequencing for the disturbance restraining method that receiving terminal is taken
Large effect (such as：One frequency domain interference, if carrying out frequency domain inhibition again after first passing around time domain AF panel, having very much can
The time domain AF panel that can be first passed through can inhibit to have an impact to subsequent frequency domain interference so that AF panel degradation) institute
To classify to different types of interference, and identify interference type, be avoided that the generation of the above situation.
For the automatic identification of interference type, study at present more mainly in Direct Sequence Spread Spectrum (Direct
Sequence Spread Spectrum, DSSS) or frequency hopping spread spectrum (Frequency-Hopping Spread Spectrum,
FHSS) in system and the identification in relation to radar emitter signal.Broadband interference is common with frequency-hopping system for spreading
Therefore effective conflicting mode mainly identifies various broadband interferences, such as broadband noise interference, broadband for interference type identification
Pectination interference, broadband Sweeping nonlinearity etc.；In radar emitter signal identification, mainly in the arteries and veins of the pulse signal received
Respective radiation source is analyzed and identified to personal feature in modulation signature and arteries and veins and (distance is taken advantage of to common cheating interference
Deceive, angle deception, velocity gate deception) classification.It can be seen that disturbance ecology algorithm is both for a certain certain types of dry at present
Specific application scenarios are disturbed or are directed to, interference type is fewer, is not suitable for complicated electromagnetic interference environment.On the other hand, it does
It includes interference characteristic extraction and two processes of classification of disturbance to disturb identification, and wherein interference characteristic parameter extraction is the core of disturbance ecology
The heart is also varied in the selection of characteristic parameter extraction algorithm, for example, with wavelet package transforms extraction characteristic parameter, use high-order
Cumulant, using Fourier Transform of Fractional Order scheduling algorithm, although above-mentioned algorithm can accurately extract the characteristic parameter of interference signal,
But the complexity of calculating is greatly increased, it is unfavorable for the realization of engineering；In various sorting algorithms, neural network classifier classification
When need a large amount of training sample, be susceptible to and owe study, spend the problems such as learning and being absorbed in local extremum, support vector cassification
Device solving the problems, such as high dimensional pattern identification, having prodigious advantage in identification of nonlinearity and small sample identification, but there is also
The problems such as selection of kernel function, determination of nuclear parameter, complexity is also relatively high.
Invention content
It is to be solved by this invention, it, should just there is provided a kind of interference identification method of suitable complicated electromagnetic interference environment
Method uses complexity by extracting one group without interfering prior information and making an uproar than insensitive characteristic of division parameter to dry
Interference signal common in wireless communication system is identified in lower Decision Tree Algorithm, and knows with preferable
Other performance.
In order to easily describe present disclosure, the interference signal to be identified is illustrated first：For channel radio
The common interference type of letter system, is divided into three categories：
1) time domain is interfered：Impulse noise interference, common radar pulse signal；
2) frequency domain interference：Partial-band jamming, spot jamming (AM interference, FM interference, BPSK interference, BPSK interference,
QPSK interference, CW interference, single tone jamming) Multi-tone jamming (pectination interference, random Multi-tone jamming)；
3) Sweeping nonlinearity：Chirp waveform, niose-modulating-frenquency jamming, single-tone/multitone Sweeping nonlinearity, sawtooth wave frequency sweep are dry
It disturbs.
The technical scheme is that：A kind of interference type recognition methods, which is characterized in that include the following steps：
A. the interference characteristic parameter of signal is extracted；The interference characteristic parameter includes at least：Time domain peak-to-average force ratio Rpm, time domain square
Kurtosis a4, normalization spectrum normal of impact difference δps；
B. judge the time domain peak-to-average force ratio R of extractionpmWhether preset time domain peak-to-average force ratio threshold value is more than, if so, judging letter
Number interference type be pulse jamming and to terminate deterministic process；If it is not, then entering step c；
C. frequency domain interference detection is carried out to signal, specific method is：Signal is done into NFFTThe FFT transform of point, then does frequency domain
Interference Detection processing obtains interference N frequency pointsJamWith interference frequency point positionJamming powerWith noise power PNoise；
D. according to the testing result of step c, judge NJam<Whether 1 is true, if so, judgement current demand signal is not interfered with and deposited
?；If it is not, then entering step e, specially：
To the interference frequency point position detected in step cIt is front and back to subtract each other to obtain frequency point blank vectorBy frequency point interval △ PiTreat as the same cluster less than 6, sub-clustering number is NCluster, each
A cluster center frequency point position is
E. sub-clustering is carried out to the interference detected in step c, if there is multiple clusters exist, then calculates frequency point where all clusters
The standard deviation C of positionvar；
F. Short Time Fourier Transform is done to signal, extracts the time-frequency figure of signal, and by time-frequency figure binaryzation, then calculate every
The LZ complexities of binaryzation sequence in a period, while marking pair of the new model quantity and new model appearance of each sequence
The position for answering frequency point finally obtains the average value of the average LZ complexities of signal and the standard deviation of time-frequency figure interference frequency point position
Var；
G. according to the characteristic parameter extracted in step a and step f, using the decision tree in sorting technique by frequency domain interference and
Sweeping nonlinearity is classified.
Further, the specific method of the interference characteristic parameter of extraction signal described in step a is：
A1. the envelope A=[A of signal are extracted1,A2,…,AL]T, ascending sequence obtains vector A '=[A '1,A
′2,…,A′L]T, wherein A '1≥A′2≥…≥A′LIfIt indicates downward rounding, takes preceding L in A 'mIt is a
Average value be signal peak valueEnvelope mean value isThe then time domain peak of signal
Than forWherein L is input signal length；
A2. time domain square kurtosis is extracted；It is S=[S specially to take the real part of time-domain signal or imaginary part1,S2,...,SL]T, μ
For the mean value of S, σ is the standard deviation of S, and the time domain square kurtosis of signal isWherein E [] is to average；
A3. extraction normalization spectrum normal of impact is poor；N first is done to signalFFTIt is P (n) that the FFT transform of point, which obtains signal spectrum,
Then normalizing spectrum isE [] is to average；In the normalized spatial spectrum for obtaining signalAfterwards, selection length
Spend LnSliding window frequency spectrum is normalized, obtain the flat of frequency spectrum；It is describedWith
Subtract the impulse part in the flat extraction normalized spatial spectrum of frequency spectrum
There are shock-free interference signal, definition normalization spectrum impact part of standards difference δ for distinguishing frequency spectrumpsFor：
Further, the specific method of the step f is：
F1. Short Time Fourier Transform is calculated：If input signal length is L, a length of N of Short Time Fourier Transform windowSTFT,
LOverlapIt counts for the overlapping of Short Time Fourier Transform adding window, frequency points are FFT points NsSTFT, segments is in time domain：For downward rounding；So obtaining a N after Short Time Fourier TransformSTFT
The three-dimensional time-frequency figure of × K × P, wherein P are the vector of energy value corresponding with time and frequency composition；Energy binaryzation thresholdingFor the maximum value of i-th of period self-energy, TpowerValue is preset as 10dB；Pass through two
Value thresholding TH obtains a NSTFTThe two-dimensional array of × K, this two-dimensional array is just the binaryzation time-frequency figure of signal, wherein having
The partial value of interference is 1, and the partial value that do not interfere is 0；
F2. search sequence new model：[N1, N2..., NK] be binaryzation time-frequency figure K period, Nk=[c1,
c2...,cN] be length it is NSTFT0,1 sequence string, wherein 1≤k≤K, ciTake 0 or 1, NkIn 1 position mean that k-th
The position of frequency point where period interference；From N1Start to do following processing to each period respectively：
F21. NkSequence is converted into character string forms { h (n) }, then defines the subsequence S and Q of { h (n) }, SQ expressions S
With the collating sequence of Q, SQ<D>Remaining sequence after deleting the last one binary sequence in SQ is indicated, if original complex degree C
(0)=1, S=h (1), Q=h (2), SQ=h (1) h (2)；
If f22. (1) S=h, h (2) ... .h (i), Q=h (i+1) SQ<D>=h (1), h (2) ... .h (i)；Judging Q is
No is SQ<D>Subset, if Q belongs to SQ<D>, complexity c (n) is constant, and S is constant, Q=h (i+1) h (i+2), repeats step f22
Until Q gets the last one binary character of sequence { h (n) }；If Q is not belonging to SQ<D>, then c=c+1 records new model
The position of frequency pointS=SQ=h (1), h (2), h (3) ..h (i+j) is enabled to repeat step f22 until Q gets sequence { h (n) }
The last one binary character；
F3. C is enabledk=c；To keep complexity value unrelated with sequence length, CkIt needs to be normalized；Sequence { h (n) }
Length be NSTFT, then normalized complexity be:New model frequency point position vector is
F4. judge whether all periods all calculate to finish, if not having been calculated, return to step f21；If calculating finishes,
Then enter step f5；
F5. model number vector C=[C are obtained1,C2,...,CK], calculate the average LZ complexities of signal
F6. it is poor to calculate interference frequency point location criteria：M=[M1,M2,..MK] be K period new model frequency point position square
Battle array, since each period obtains pattern quantity, not necessarily whole is equal, and also partial dot extraction is inaccurate, it is necessary to reject
Unusual frequency point, M ' are the matrix rejected after unusual frequency point：
Wherein c ' is C=[C1,C2,...,CK] in minimum value, K '≤K；Standard deviation is asked to obtain V=[V every a line of M '1,
V2,...,Vc′]T, interference frequency point location criteria difference be
Further, frequency domain interference and Sweeping nonlinearity are classified using the decision tree in sorting technique in step g
Specific method is：
G1. judge whether preset complexity threshold T is more than by the complexity lz obtained in step f5lz=0.06, if
It is noise FM in Partial band noise jammer in then judging interference type for frequency domain interference, random multitone, pectination interference or steady interference
One kind of interference, enters step g2；If it is not, spot jamming or steadily interfering center line in then judging interference type for frequency domain interference
Property frequency modulation interference, mono-/multi- sound Sweeping nonlinearity, sawtooth wave Sweeping nonlinearity one kind, enter step g3；
G2. judge whether preset normalization spectrum impact mark is more than by the normalization spectrum normal of impact difference obtained in step a3
Quasi- difference threshold Tδps=1.0, if so, random multitone, pectination interference in judging interference type for frequency domain interference, enter step
g4；If it is not, in then judging interference type for frequency domain interference in Partial band noise jammer or nonstationary interference niose-modulating-frenquency jamming one kind,
Enter step g5；
Whether the time-frequency domain interference frequency point location criteria difference Var g3. obtained in judgment step f6 is more than preset time-frequency domain
Interfere frequency point location criteria difference thresholding Tvar=14, if so, linear frequency sweep, mono-/multi- in judging interference type for nonstationary interference
Sound frequency sweep, sawtooth wave frequency sweep；If it is not, spot jamming, pectination interference in then judging interference type for frequency domain interference；
G4. the interference cluster frequency point interval variance C obtained in judgment step evarWhether preset interference cluster frequency point interval is more than
Variance threshold Tcvar=5, if so, random multitone in judging interference type for frequency domain interference；If it is not, then judging interference type
It is interfered for pectination in frequency domain interference；
Whether the time domain square kurtosis that g5 judgment steps a2 is obtained is more than preset time domain square kurtosis thresholding Ta4=2.85, if
It is Partial band noise jammer in then judging interference type for frequency domain interference, if it is not, noise in then judging interference type for nonstationary interference
Frequency modulation interferes.
Beneficial effects of the present invention are that the present invention can be to the hostility in the wireless communication system of complex electromagnetic environment
Interference Detection and identification are done in interference, and then joint decision goes out the situation of current electromagnetic environment, and to there is the interference under interference environment
Type is classified；Interference Detection mainly determines the existence of interference in terms of the time domain and frequency domain two；Disturbance ecology is main
It is that from time domain, frequency domain, time-frequency domain extraction characteristic parameter and will be interfered using decision tree and be divided into time domain interference, frequency domain interference, sweep
Part interference has simultaneously been done detailed identification by frequency three categories；The interference characteristic parameter that the present invention extracts have preferable robustness and
It makes an uproar than less sensitive, can also have in the case of decision tree decision method lower using algorithm complexity preferable dry to dry
Disturb recognition performance.
Description of the drawings
Fig. 1 is the detail flowchart of interference type identification of the present invention；
Fig. 2 is the extraction flow chart of interference characteristic parameter in step a；
Fig. 3 is the extraction flow chart of interference characteristic parameter in step f；
Fig. 4 is decision tree judgement flow chart；
Fig. 5 is disturbance ecology emulation of the present invention about various interference type recognition performance figures.
Specific implementation mode
With reference to the accompanying drawings and examples, detailed description of the present invention technical solution：
The present embodiment is to carry out Interference Detection and knowledge to the signal that data length is L=30720 × 20 in quiet period
Not, the long N of window has been added to signal in systemFFTFor 2048 Hamming window, therefore FFT length NFFTIt is 2048；Frequency domain FCME interference inspections
Power scale accumulation length NsumIt is 64, interference detection threshold factor T is thus calculatedfactorIt is 1.64, other in system interfere
Thresholding used in characteristic parameter is the empirical value taken after simulation analysis.
The setting of S1, parameter：Theory and simulation analysis according to systematic parameter and various Interference Detections and recognizer is true
Determine interference detection threshold factor TfactorFor 1.64, time domain peak-to-average force ratio thresholding TrpmFor 3.5, LZ complexity thresholdings TlzFor 0.06, when
The standard deviation thresholding T of frequency domain interference frequency point positionvarFor 14, normalization normal of impact difference thresholding TδpsFor 1.0, time domain square kurtosis door
Limit Ta4For 2.85, the standard deviation thresholding T at interference cluster frequency point intervalcvarIt is 5；
S2, interference characteristic parameter extraction a：Extract the time domain peak-to-average force ratio R of signalpm, time domain square kurtosis a4, normalization spectrum impact
Standard deviation δps；
S21, extraction time domain peak-to-average force ratio：Processing signal total length is the envelope A=[A that L extracts signal first1,A2,...,
AL]T, ascending sequence obtains vector A '=[A '1,A′2,...,A′L]T, wherein A '1≥A′2≥…≥A′LIfIt indicates downward rounding, takes preceding L in A 'mA average value is the peak value of signal
Envelope mean value isIt is by the time domain peak-to-average force ratio of signal then
S22, extraction time domain square kurtosis：It is S=[S to take the real part of time-domain signal or imaginary part1,S2,…,SL]T,For the mean value of S,Time domain square kurtosis for the standard deviation of S, signal is
S23, extraction normalization spectrum normal of impact are poor：It is N to be first divided into k=300 sections per segment length to signalFFT=2048, point
Data are r=[r after section0,r1,...,rk-1], wherein m sections of time domain datas areTo each section
Time domain data carries out FFT transform respectively, obtains k sections of frequency domain data R=[R0,R1,…,Rk-1], wherein m sections of frequency domain datas areThen it carries out mould side to each section of identical frequency point data to average operation, what is obtained is averaged
PowerWherein, the mean power of n-th of frequency point isPower is asked out
The amplitude spectrum that square operation obtains signal is P (n)=[P (1), P (2) ... .P (NFFT)], signal normalization spectrum isE [] is to average.In the normalized spatial spectrum for obtaining signalAfterwards,
Select length LnIt (is taken as) sliding window frequency spectrum is normalized, obtain the flat of frequency spectrum.
WithImpulse part in normalized spatial spectrum can be extracted by subtracting the flat of frequency spectrum
ByFrequency spectrum can be distinguished very well shock-free interference signal, in order to extract this parameter, defines normalizing
Change spectrum impact part of standards variance δps：
S3, judge time domain peak-to-average force ratio whether greatly and thresholding：The time domain peak-to-average force ratio R that will be extracted in S2pmIt makes comparisons with threshold value,
Then it is determined as pulse jamming if it is greater than thresholding；Otherwise S4；
S4, frequency domain interference detection：Take the part for receiving data, continuous NFFT=2048 sampled points are one section, are divided into
K=64 sections of r=[r0,r1,...,rk-1], wherein m sections of time domain datas areTo each section of time domain
Data carry out FFT transform respectively, obtain k sections of frequency domain data R=[R0,R1,...,Rk-1], wherein m sections of frequency domain datas areThen it carries out mould side to each section of identical frequency point data to average operation, obtains each frequency
The mean power of pointWherein, the mean power of n-th of frequency point isSo
It in block length be 64 false-alarm probabilities is P to be calculated afterwards according to adaptive threshold detecting algorithm principlefaIt is detected in the case of=5e-6
Threshold factor Tfactor=1.64.Then the processing of frequency domain FCME Interference Detections is done to frequency-region signal, obtains interference N frequency pointsJam
With interference frequency point positionJamming powerNoise power PNoiseEqual interference letter
Breath；
S5, judge whether frequency domain detects interference：According to testing result in S4, if NJam<1 judges in current system
Presence is not interfered with, S6 is otherwise turned；
S6, interference signal processing：To the interference frequency point position detected in S4It is front and back to subtract each other to obtain
Frequency point blank vectorBy frequency point interval △ PiTreat as the same cluster, sub-clustering number less than 6
For NCluster, each cluster center frequency point position isIf NCluster>1, then calculate all clusters places
The standard deviation of frequency point positionWherein
S7, interference characteristic parameter extraction b：Short Time Fourier Transform is done to signal, extracts the time-frequency figure of signal, and by time-frequency
Figure binaryzation, then calculates the LZ complexities of binaryzation sequence in each period, while marking the new model of each sequence
The position for the correspondence frequency point that quantity and new model occur finally obtains average LZ complexities and the time-frequency figure interference frequency point position of signal
The average value Var for the standard deviation set, turns S8；
S71, initiation parameter：Binaryzation energy relative threshold T is setpowerValue be 10dB, original complex angle value lz=0,
Pattern initial value c=1 etc.；
S72, Short Time Fourier Transform is calculated：Input signal length is L=30720 × 20, and Short Time Fourier Transform window is long
For NSTFT=1024, LOverlap=512 count for the overlapping of Short Time Fourier Transform adding window, take 1/2 overlapping adding window；Thus it calculates
Segments is in time domain：For downward rounding.So Short Time Fourier Transform
After obtain a NSTFTThe three-dimensional time-frequency figure of × K × P, wherein P are the three-dimensional of energy value corresponding with time and frequency composition
Vector.Energy binaryzation thresholdingFor the maximum value of i-th of period self-energy.Pass through
Binaryzation thresholding TH just obtains a NSTFTThe two-dimensional array of × K, this two-dimensional array are just the binaryzation time-frequency figure of signal, often
Noisy value of frequency point is 1 in one segmentation, and the partial value that do not interfere is 0；
S72, search sequence new model：[N1, N2..., NK] be binaryzation time-frequency figure K period, Nk=[c1,
c2...,cN] be length it is the 0 of N, 1 sequence string, wherein 1≤k≤K, ciTake 0 or 1, NkIn 1 position when meaning that k-th
Between section interference where frequency point position.From N1Start to do following processing to each period respectively：
S721, NkSequence biography changes character string forms { h (n) } into, then defines the subsequence S and Q of { h (n) }, SQ expressions
The collating sequence of S and Q, SQ<D>Remaining sequence after deleting the last one binary sequence in SQ is indicated, if original complex degree
C (0)=1, S=h (1), Q=h (2), SQ=h (1) h (2)
If S722, S=h (1), h (2) ... .h (i), Q=h (i+1) SQ<D>=h (1) .h (i) judges that Q is to h (2) ...
No is SQ<The subset of D
If S723, Q belong to SQ<D>, complexity c (n) is constant, and S is constant, Q=h (i+1) h (i+2), continues S722.
If S724, Q are not belonging to SQ<D>, c=c+1, then record new model frequency point position Mk c, enable S=SQ=h (1), h
(2), h (3) ..h (i+j) continues S722
The last one binary character that S73, the above process get sequence { h (n) } until Q, then enables Ck=c.To make
Complexity value is unrelated with sequence length, CkIt needs to be normalized.The length of sequence { h (n) } is NSTFT, then normalized
Complexity can be written as:New model frequency point position vector is
Indicate the frequency point position of c-th of pattern of k-th of time segment mark；
S74, judge that all periods, whether all calculating finished, if not having been calculated, turned S721；If calculating finishes, turn step
S75；
S75, model number vector C=[C are obtained by above-mentioned steps1,C2,...,CK], CkIt indicates to search in k-th of period
Then the pattern quantity that rope arrives calculates the average LZ complexities of signal
S76, it calculates and interferes frequency point location criteria poor：M=[M1,M2,..MK] be K period new model frequency point position
Matrix, since each period obtains pattern quantity, not necessarily whole is equal, and also partial dot extraction is inaccurate, it is necessary to pick
Except unusual frequency point, M ' is the matrix rejected after unusual frequency point：
Wherein c ' is C=[C1,C2,...,CK] in minimum value, K '≤K；Standard deviation is asked to obtain V=[V every a line of M '1,
V2,...,Vc′]T, interference frequency point location criteria difference be
S8, decision tree judgement：According to the characteristic parameter that S2 and S7 kinds are extracted, using the decision tree in sorting technique by frequency domain
Interference and nonstationary interference carry out exhaustive division；
S81, judge whether the complexity lz calculated by S75 is more than threshold Tlz=0.06, it is then interfered if it is greater than thresholding
Type may be in frequency domain interference (Partial band noise jammer, random multitone, pectination interference) or steady interference (niose-modulating-frenquency jamming)
One kind turning S82；If it is less than thresholding, then interference type may be that frequency domain interference (spot jamming) or steady interference are (linear to adjust
Frequency interference, mono-/multi- sound Sweeping nonlinearity, sawtooth wave Sweeping nonlinearity) one kind, turn S83；
S82, judge whether the normalization spectrum normal of impact difference calculated by S23 is more than thresholding Tδps=1.0, if it is greater than door
Then interference type may be frequency domain interference (random multitone, pectination interference) to limit, turn S84；If it is less than thresholding, then interference type can
It can be frequency domain interference (Partial band noise jammer) or one kind of nonstationary interference (niose-modulating-frenquency jamming), turn S85；
S83, judge whether the time-frequency domain interference frequency point location criteria difference Var calculated by S76 is more than thresholding Tvar=14, such as
Fruit is more than thresholding, and then interference type is nonstationary interference (linear frequency sweep, mono-/multi- sound frequency sweep, sawtooth wave frequency sweep)；If it is less than door
Then interference type is frequency domain interference (spot jamming, pectination interference) to limit；
S84, judge the interference cluster frequency point interval variance C calculated by S6varWhether threshold T is more thancvar=5, if greatly with
Thresholding interference type is frequency domain interference (random multitone), and otherwise interference type is frequency domain interference (pectination interference)；
S85, judge whether the time domain square kurtosis calculated by S22 is more than thresholding Ta4=2.85, it is then interfered if it is greater than thresholding
Type is frequency domain interference (Partial band noise jammer), and otherwise interference type is nonstationary interference (niose-modulating-frenquency jamming)；
Fig. 5 is the simulation performance that the interference type of the present embodiment identifies.
Interference type is impulse noise interference, partial-band Gaussian noise jamming, pectination interference, random Multi-tone jamming, noise tune
Frequency interference, spot jamming (BPSK interference), Sweeping nonlinearity (broadband sawtooth wave interference), wherein spot jamming and Sweeping nonlinearity
One of which, interference design parameter has respectively only been taken to be shown in Table 1.Channel is the awgn channel of standard, and disturbance ecology process is quiet
What the silent phase carried out, i.e., interference and noise are only received, a service subframe sampled data output known to system frame format is
30720, wherein the quiet period time is 20 service subframes, therefore total sampling number is 30720 × 20.Carrying out frequency domain interference
When detection, it is L=2048 × 64 only to take a portion data, sampled data block length per treatment, and N is on frequency domainFFT=
2048 points of FFT, and make Nsum=64 points of power is cumulative, and the frequency domain interference detection threshold factor being calculated is Tfactor=
1.64；Adding window length is N in Fourier in short-term calculatesSTFT=1024, adding window type is Hamming window, overlapping adding window points
LOverlapIt is 512；
1 interference simulation parameter of table
In Figure 5, JNR is from -10dB to 50dB, and using 3dB as interval value, each JNR values emulate 100 times and obtain different dry
Disturb the disturbance ecology performance of type, wherein what red line indicated is the identity of time domain interference (impulse noise interference) in Fig. 5
Can, it can correctly be identified substantially in JNR >=-7dB；What blue line indicated is frequency domain interference recognition performance, partial-band
Noise jamming and spot jamming (BPSK) can accurately identify out substantially in JNR >=-3dB, and pectination is interfered and random more
Sound interference in JNR >=5dB discrimination 95% or more；Black line indicate be Sweeping nonlinearity recognition performance, JNR >=
When 1dB, being substantially all to accurately identify out.Generally speaking, in JNR >=5dB, all disturbance ecology performances all 95% with
On, there is preferable recognition performance, and recognition performance is not made an uproar by dry than being influenced substantially in the case of high s/n ratio, is had preferable
Robustness.Therefore, the present invention has preferable classifying quality to disturbance type, has preferable application value.
Claims (2)
1. a kind of interference type recognition methods, which is characterized in that include the following steps：
A. the interference characteristic parameter of signal is extracted；The interference characteristic parameter includes at least：Time domain peak-to-average force ratio Rpm, time domain square kurtosis
a4, normalization spectrum normal of impact difference δps；
B. judge the time domain peak-to-average force ratio R of extractionpmWhether preset time domain peak-to-average force ratio threshold value is more than, if so, judging signal
Interference type is pulse jamming and terminates deterministic process；If it is not, then entering step c；
C. frequency domain interference detection is carried out to signal, specific method is：Signal is done into NFFTThe FFT transform of point, then does frequency domain interference
Detection process obtains interference N frequency pointsJamWith interference frequency point positionJamming powerWith noise power PNoise；
D. according to the testing result of step c, judge NJam<Whether 1 is true, if so, judgement current demand signal does not interfere with presence；
If it is not, then entering step e；
E. sub-clustering is carried out to the interference detected in step c, if there is multiple clusters exist, then frequency point position where calculating all clusters
Standard deviation Cvar, specially：
To the interference frequency point position detected in step cIt is front and back to subtract each other to obtain frequency point blank vectorBy frequency point interval △ PiTreat as the same cluster less than 6, sub-clustering number is NCluster, each
A cluster center frequency point position is
F. Short Time Fourier Transform is done to signal, the time-frequency figure of signal is extracted, and by time-frequency figure binaryzation, when then calculating each
Between in section binaryzation sequence LZ complexities, while marking the new model quantity of each sequence and corresponding frequency that new model occurs
The position of point finally obtains the average value Var of the average LZ complexities of signal and the standard deviation of time-frequency figure interference frequency point position；
G. according to the characteristic parameter extracted in step a and step f, using the decision tree in sorting technique by frequency domain interference and frequency sweep
Interference is classified.
2. a kind of interference type recognition methods according to claim 1, which is characterized in that extract signal described in step a
The specific method of interference characteristic parameter be：
A1. the envelope A=[A of signal are extracted1,A2,...,AL]T, ascending sequence obtains vector A '=[A '1,A′2,...,
A′L]T, wherein A '1≥A′2≥…≥A′LIf It indicates downward rounding, takes preceding L in A 'mA average value
For the peak value of signalEnvelope mean value isThen the time domain peak-to-average force ratio of signal isWherein L is input signal length；
A2. time domain square kurtosis is extracted；It is S=[S specially to take the real part of time-domain signal or imaginary part1,S2,...,SL]T, μ is S's
Mean value, δ are the variance of S, and the time domain square kurtosis of signal is
A3. extraction normalization spectrum normal of impact is poor；N first is done to signalFFTIt is P (n) that the FFT transform of point, which obtains signal spectrum, then returns
One, which changes spectrum, isE [] is to average；In the normalized spatial spectrum for obtaining signalAfterwards, selection length Ln's
Frequency spectrum is normalized in sliding window, obtains the flat of frequency spectrum；It is describedWithSubtract frequency
Impulse part in the flat extraction normalized spatial spectrum of spectrum
There are shock-free interference signal, definition normalization spectrum impact part of standards variance δ for distinguishing frequency spectrumpsFor：
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201610743463.6A CN106330385B (en) | 2016-08-29 | 2016-08-29 | A kind of interference type recognition methods |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201610743463.6A CN106330385B (en) | 2016-08-29 | 2016-08-29 | A kind of interference type recognition methods |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN106330385A CN106330385A (en) | 2017-01-11 |
| CN106330385B true CN106330385B (en) | 2018-11-09 |
Family
ID=57791352
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201610743463.6A Active CN106330385B (en) | 2016-08-29 | 2016-08-29 | A kind of interference type recognition methods |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN106330385B (en) |
Families Citing this family (27)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN106910507B (en) * | 2017-01-23 | 2020-04-24 | 中国科学院声学研究所 | Detection and identification method and system |
| CN108631893B (en) * | 2017-03-17 | 2020-12-29 | 成都信息工程大学 | GMSK channel interference detection method |
| CN109495133B (en) * | 2017-09-11 | 2021-09-07 | 大连军通科技有限公司 | Periodic pulse interference sensing method |
| CN107682109B (en) * | 2017-10-11 | 2019-07-30 | 北京航空航天大学 | A kind of interference signal classifying identification method suitable for UAV Communication system |
| CN108089162A (en) * | 2017-12-29 | 2018-05-29 | 中国电子科技集团公司第二十研究所 | A kind of detection of pulse interference signal and suppressing method |
| CN109995447A (en) * | 2017-12-31 | 2019-07-09 | ***通信集团湖北有限公司 | Interference type recognition methods, device, equipment and medium |
| CN110022182A (en) * | 2018-01-10 | 2019-07-16 | 富士通株式会社 | Interference source prediction model method for building up, method for interference source identification, device and terminal device |
| CN108257617B (en) * | 2018-01-11 | 2021-01-19 | 会听声学科技（北京）有限公司 | Noise scene recognition system and method |
| CN108764077B (en) * | 2018-05-15 | 2021-03-19 | 重庆邮电大学 | Digital signal modulation classification method based on convolutional neural network |
| CN108919203B (en) * | 2018-05-21 | 2022-04-22 | 西安电子科技大学 | Radar active interference identification method and system |
| CN110659656A (en) * | 2018-06-29 | 2020-01-07 | 中兴通讯股份有限公司 | Method and device for detecting interference |
| CN109283522B (en) * | 2018-10-29 | 2023-03-14 | 电子科技大学 | Co-location MIMO radar target tracking method based on joint time-space resource management |
| CN110146855B (en) * | 2019-06-11 | 2020-10-23 | 北京无线电测量研究所 | Radar intermittent interference suppression threshold calculation method and device |
| CN111092632B (en) * | 2019-12-03 | 2022-04-05 | 成都众志天成科技有限公司 | Interference suppression method and interference suppressor |
| CN110927706B (en) * | 2019-12-10 | 2022-05-24 | 电子科技大学 | Convolutional neural network-based radar interference detection and identification method |
| CN111245455A (en) * | 2020-02-19 | 2020-06-05 | 北京紫光展锐通信技术有限公司 | Dynamic interference suppression method for receiver, receiver system and storage medium |
| CN111382803B (en) * | 2020-03-18 | 2022-06-03 | 电子科技大学 | Communication signal feature fusion method based on deep learning |
| CN112083448B (en) * | 2020-09-04 | 2023-08-15 | 哈尔滨工程大学 | Satellite navigation system-oriented interference signal classification recognition feature extraction method and system |
| CN112162245B (en) * | 2020-09-29 | 2021-06-15 | 中国船舶重工集团公司第七二四研究所 | Radar broadband interference identification method based on time-frequency power spectrum projection |
| CN113219417B (en) * | 2020-10-21 | 2022-09-20 | 中国人民解放军空军预警学院 | Airborne radar interference type identification method based on support vector machine |
| CN112636781B (en) * | 2020-12-23 | 2022-03-15 | 电子科技大学 | Method for detecting digital modulation optimal interference signal |
| CN113302511B (en) * | 2021-03-31 | 2022-05-24 | 华为技术有限公司 | Interference processing method and device |
| CN113556774A (en) * | 2021-07-14 | 2021-10-26 | 中国电信股份有限公司 | Interference positioning method, device, equipment and storage medium of mobile network |
| CN113616213B (en) * | 2021-07-29 | 2023-02-28 | 山东大学 | Electrocardiosignal denoising method, electrocardiosignal denoising device and electrocardiosignal denoising storage medium based on BP neural network and improved EMD method |
| CN115079123B (en) * | 2022-08-22 | 2023-06-20 | 北京大有半导体有限责任公司 | Detection method, detection device, and computer-readable storage medium |
| CN115694690B (en) * | 2022-10-19 | 2024-06-04 | ***通信集团四川有限公司 | Method and device for determining radio interference type and electronic equipment |
| CN117714245B (en) * | 2024-02-06 | 2024-04-26 | 山东浪潮数据库技术有限公司 | Interference suppression system, method, equipment and medium in wireless ad hoc network system |
Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN104158611A (en) * | 2014-08-20 | 2014-11-19 | 西安电子科技大学 | System and method of detecting interference of wireless signal based on spectral analysis |
| US9130871B2 (en) * | 2010-01-29 | 2015-09-08 | Aruba Networks, Inc. | Interference classification with minimal or incomplete information |
-
2016
- 2016-08-29 CN CN201610743463.6A patent/CN106330385B/en active Active
Patent Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US9130871B2 (en) * | 2010-01-29 | 2015-09-08 | Aruba Networks, Inc. | Interference classification with minimal or incomplete information |
| CN104158611A (en) * | 2014-08-20 | 2014-11-19 | 西安电子科技大学 | System and method of detecting interference of wireless signal based on spectral analysis |
Non-Patent Citations (2)
| Title |
|---|
| 基于频域复杂度分析的频谱感知算法;罗孟宇等;《计算机工程与科学》;20130630;第35卷(第6期);第52-56页 * |
| 干扰识别技术研究与实现;李林甫;《中国优秀硕士学位论文全文数据库》;20141105(第11期);第5-55页 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN106330385A (en) | 2017-01-11 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN106330385B (en) | A kind of interference type recognition methods | |
| CN106789788B (en) | A kind of wireless digital signal Modulation Mode Recognition method and device | |
| CN101587186B (en) | Characteristic extraction method of radar in-pulse modulation signals | |
| Liu et al. | Biologically inspired covert underwater acoustic communication by mimicking dolphin whistles | |
| US11581967B2 (en) | Wireless channel scenario identification method and system | |
| CN102760444B (en) | Support vector machine based classification method of base-band time-domain voice-frequency signal | |
| CN106772331B (en) | Target identification method and Target Identification Unit | |
| CN107305774A (en) | Speech detection method and device | |
| CN104135327A (en) | Spectrum sensing method based on support vector machine | |
| Chen et al. | Interrupted-sampling repeater jamming suppression based on stacked bidirectional gated recurrent unit network and infinite training | |
| CN107682109B (en) | A kind of interference signal classifying identification method suitable for UAV Communication system | |
| CN105678273A (en) | Initial point detection algorithm of transient signal in radio frequency fingerprint identification technology | |
| CN108919203B (en) | Radar active interference identification method and system | |
| CN112083393B (en) | Intermittent sampling forwarding interference identification method based on spectrogram average time characteristics | |
| Akhtar et al. | A neural network target detector with partial CA-CFAR supervised training | |
| CN106936514A (en) | The underwater sound targeting signal detection method concentrated based on energy under condition of sparse channel | |
| CN103413134A (en) | Ground moving target micro-tremor signal characteristic extraction based on sparse decomposition | |
| CN110276256A (en) | Based on the low signal-to-noise ratio Modulation Recognition of Communication Signal method and device for adjusting ginseng accidental resonance | |
| Rigling et al. | Acf-based classification of phase modulated waveforms | |
| CN106972895A (en) | Underwater sound targeting signal detection method based on accumulation coefficient correlation under condition of sparse channel | |
| CN104596369A (en) | SVM-based continuous wave Doppler radio fuse frequency-swept jamming prevention method | |
| CN115061094B (en) | Radar target recognition method based on neural network and SVM | |
| Nguyen et al. | A parallel neural network-based scheme for radar emitter recognition | |
| CN113534059B (en) | Radar active interference identification method based on deep convolutional network under open set scene | |
| CN114676721A (en) | Radar active suppression interference identification method and system based on radial basis function neural network |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |