CN112257529B - Automatic extraction system and method for radio signal characteristics - Google Patents
Automatic extraction system and method for radio signal characteristics Download PDFInfo
- Publication number
- CN112257529B CN112257529B CN202011084025.6A CN202011084025A CN112257529B CN 112257529 B CN112257529 B CN 112257529B CN 202011084025 A CN202011084025 A CN 202011084025A CN 112257529 B CN112257529 B CN 112257529B
- Authority
- CN
- China
- Prior art keywords
- signal
- level
- data
- frequency
- time
- 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
- 238000000605 extraction Methods 0.000 title claims abstract description 101
- 238000000034 method Methods 0.000 title claims abstract description 25
- 238000012544 monitoring process Methods 0.000 claims abstract description 88
- 238000012545 processing Methods 0.000 claims abstract description 36
- 230000006855 networking Effects 0.000 claims abstract description 32
- 238000013500 data storage Methods 0.000 claims abstract description 28
- 238000004891 communication Methods 0.000 claims abstract description 14
- 238000007726 management method Methods 0.000 claims abstract description 9
- 230000005540 biological transmission Effects 0.000 claims abstract description 7
- 238000004422 calculation algorithm Methods 0.000 claims abstract description 5
- 230000003595 spectral effect Effects 0.000 claims description 75
- 239000011159 matrix material Substances 0.000 claims description 65
- 238000004364 calculation method Methods 0.000 claims description 54
- 238000004220 aggregation Methods 0.000 claims description 46
- 230000002776 aggregation Effects 0.000 claims description 44
- 238000001228 spectrum Methods 0.000 claims description 39
- 238000005259 measurement Methods 0.000 claims description 12
- 238000000354 decomposition reaction Methods 0.000 claims description 9
- 230000005236 sound signal Effects 0.000 claims description 9
- 239000000284 extract Substances 0.000 claims description 7
- 230000007797 corrosion Effects 0.000 claims description 6
- 238000005260 corrosion Methods 0.000 claims description 6
- 238000004088 simulation Methods 0.000 claims description 6
- 230000001174 ascending effect Effects 0.000 claims description 4
- 238000010606 normalization Methods 0.000 claims description 3
- 238000010801 machine learning Methods 0.000 abstract description 2
- 238000010586 diagram Methods 0.000 description 2
- 230000007547 defect Effects 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 238000005516 engineering process Methods 0.000 description 1
- 238000012806 monitoring device Methods 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/23—Clustering techniques
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T5/00—Image enhancement or restoration
- G06T5/20—Image enhancement or restoration by the use of local operators
- G06T5/30—Erosion or dilatation, e.g. thinning
-
- G06T5/70—
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/90—Determination of colour characteristics
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04W—WIRELESS COMMUNICATION NETWORKS
- H04W24/00—Supervisory, monitoring or testing arrangements
- H04W24/02—Arrangements for optimising operational condition
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04W—WIRELESS COMMUNICATION NETWORKS
- H04W24/00—Supervisory, monitoring or testing arrangements
- H04W24/06—Testing, supervising or monitoring using simulated traffic
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F2218/00—Aspects of pattern recognition specially adapted for signal processing
- G06F2218/08—Feature extraction
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F2218/00—Aspects of pattern recognition specially adapted for signal processing
- G06F2218/12—Classification; Matching
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02D—CLIMATE CHANGE MITIGATION TECHNOLOGIES IN INFORMATION AND COMMUNICATION TECHNOLOGIES [ICT], I.E. INFORMATION AND COMMUNICATION TECHNOLOGIES AIMING AT THE REDUCTION OF THEIR OWN ENERGY USE
- Y02D30/00—Reducing energy consumption in communication networks
- Y02D30/70—Reducing energy consumption in communication networks in wireless communication networks
Abstract
A radio signal characteristic automatic extraction system and its extraction method, it is the technical field of wireless communication, including monitoring the networking application system, data storage deframes module, signal extraction module, apparatus signal library; the monitoring networking application system is internally provided with a data acquisition management and control platform; the monitoring networking application system realizes networking of the full-province monitoring equipment, gathers data of the issued tasks and provides a data transmission interface for the data storage frame-decoding module; the data storage frame-decoding module comprises a real-time processing cluster, an off-line processing cluster and a big data platform. The invention develops and designs a method for extracting radio signal characteristics by utilizing a big data platform, a radio network monitoring and machine learning algorithm, and comprehensively realizes acquisition of fine images such as signal frequency, bandwidth, time distribution, position distribution, energy distribution and the like, thereby better grasping panoramic information of signals.
Description
Technical Field
The invention belongs to the technical field of wireless communication, and particularly relates to an automatic extraction system and an extraction method for radio signal characteristics.
Background
With the rapid development of radio monitoring technology and radio service, more and more electromagnetic signals exist in space, so that the signals obtained by monitoring are more and more, and the radio mechanism has great challenges on the extraction, identification and comparison of the wireless electromagnetic spectrum signals due to the complexity of the electromagnetic signals. At present, monitoring staff of a radio mechanism mainly compares characteristic values such as level amplitude values, bandwidths, center frequency points and the like of signals by mainly using the comparison of wireless electromagnetic signals, and can not effectively identify and classify the signals due to the fact that the characteristic values are fewer, and meanwhile, the signals can not be effectively analyzed and identified due to the fact that the monitoring time is long, the monitoring data size is large, the spectrum consistency of an air signal and a standard signal is poor, the experience and the energy of monitoring staff are limited and the like.
The method for extracting the signal characteristics in the current industry is an effective method for solving the requirement by considering the networking monitoring condition and extracting the characteristics of the signals from the frequency domain, the energy domain, the time domain, the position domain, the content domain and the modulation domain of the signals.
Disclosure of Invention
Aiming at the defects of the prior art, the invention aims to provide an automatic radio signal characteristic extraction system, which solves the problem that signals cannot be effectively analyzed and identified due to too small characteristic value and large monitoring data quantity.
Another object of the present invention is to provide an automatic radio signal feature extraction method.
In order to achieve the above object, the present invention adopts the following technical scheme.
A radio signal characteristic automatic extraction system comprises a monitoring networking application system, a data storage frame-decoding module, a signal extraction module and a device signal library;
the monitoring networking application system is internally provided with a data acquisition management and control platform; the monitoring networking application system realizes networking of the full-province monitoring equipment, gathers data of the issued tasks and provides a data transmission interface for the data storage frame-decoding module;
the data storage frame-decoding module comprises a real-time processing cluster, an off-line processing cluster and a big data platform; the real-time processing cluster is connected with a monitoring data interface of the data acquisition management platform and is used for frame decomposition, back noise calculation and data aggregation; the off-line processing cluster performs frame decomposition, back noise calculation and data aggregation on the monitoring data acquired and uploaded by the internet, and the big data platform stores massive monitoring data;
the signal extraction module is in signal connection with the data storage frame-decoding module and the equipment signal library; the signal extraction module receives the frame-decoding aggregation data of the data storage frame-decoding module and acquires the dimensional data of a position domain, a content domain, a modulation domain, a frequency domain, an energy domain and a time domain from the frame-decoding aggregation data;
The equipment signal library is used for extracting signals, storing the signals according to dimensions of the day, equipment and the like, and comparing and classifying the signals subsequently.
A radio signal characteristic automatic extraction method applying the system comprises the following steps:
step S1, the monitoring networking application system realizes networking of the full-province monitoring equipment, can issue a monitoring task according to requirements, and simultaneously provides a monitoring task interface and offline monitoring data for the data storage frame-decoding module, and the offline monitoring data is sent to a big data platform through a transmission tool;
step S2, the real-time processing cluster of the data storage frame-decoding module is connected with a monitoring data interface of a monitoring networking application system according to the signal extraction requirement, and after the received data stream is frame-decoded, aggregation is carried out according to the minute level; the offline processing cluster reads the data stored in the HDFS, and after frame decomposition, aggregation is carried out according to the minute level; the minute-level aggregated data acquired after the real-time processing cluster and the off-line processing cluster are processed waits for signal extraction by a signal extraction module;
step S3, the signal extraction module forms a gray scale image according to the corresponding background noise margin from the aggregated minute-level aggregated data, and extracts signals through corrosion and expansion of the gray scale image, wherein the characteristic values of the extracted signals comprise: frequency domain features, energy domain features, time domain features, location domain features, content domain features, and modulation domain features;
And S4, the equipment signal library stores the signals extracted in the step according to equipment and dates for subsequent signal classification and signal alarming.
Specifically, in step S3, the following steps are included:
step S301, signal extraction condition setting; the signal extraction module is used for extracting signals from the aggregated minute-level data, wherein the conditions of signal extraction are < equipment number, signal bandwidth, signal duration interval, signal extraction frequency interval, signal extraction time interval and back noise margin >;
step S302, corresponding scanning data are obtained from the aggregated minute-level data according to the equipment number, the signal duration interval and the signal extraction interval, a level matrix is formed for the data, and meanwhile, the level matrix is compared with a background noise margin according to a level value to form a gray level map matrix;
step S303, extracting device signal range information: extracting signals from a gray scale graph according to the signal bandwidth and the signal duration, wherein gray scale rectangles with extraction value columns larger than 2 and rows larger than 30 are used as signals in the gray scale graph, and the signal range information of the extracting equipment is as follows: < signal start time t1, signal end time tn, signal start bandwidth fre_start, signal end bandwidth fre_end >.
Further, in step S302, the signal extraction module extracts a device number, a signal extraction time interval, and a signal extraction frequency interval from the aggregated minute-level data, and forms a level matrix for the portion of data: the data frequency points in each minute are arranged in an ascending order to form a matrix with frequency points as columns and time as rows; comparing the level value d (i, j) corresponding to the frequency point in the matrix with the background noise margin, if the level value d (i, j) is larger than the background noise margin, taking the value of the unit of the gray matrix as 1, otherwise, taking the value as 0, namely, if the level value d (i, j) is larger than the background noise margin, taking the value of the gray map corresponding to the frequency point and the corresponding time as 1, otherwise, taking the value as 0, and finally forming the gray matrix;
closing the gray matrix, and performing expansion operation according to the matrix unit of 1*3; and (3) performing corrosion operation on the expanded gray matrix according to a matrix unit of 1*3, removing burrs and jitter error values, and forming a gray image of smooth and noiseless data.
Further, the method comprises the following steps:
step S304, extracting frequency domain features: according to the equipment signal range, combining the corresponding minute-level aggregation data, extracting frequency domain features, wherein the extracted features comprise: < center frequency, signal bandwidth, spectral line first peak, spectral line second peak, spectral line third peak, spectral short-time energy, spectral short-time average energy, spectral amplitude greater than threshold continuous point energy, spectral peak number ratio >;
The center frequency calculation formula: fefr= (fre_end+fre_start)/2;
the signal bandwidth calculation formula: f_ebd=fre_end-fre_start;
wherein, F_EFR represents the center frequency of the signal, F_EBD represents the bandwidth of the signal, fre_end represents the end frequency of the signal, fre_start represents the start frequency of the signal;
the first peak value of the spectral line refers to the maximum extreme value of the level amplitude value in the signal duration and bandwidth, namely, the maximum level value is selected from the data of which the time range is from the signal start time t1 to the signal end time t2 and the frequency range is from the signal start frequency fre_start to the signal end frequency fre_end in the minute-level aggregation table of the equipment, and is the first peak value of the spectral line;
the calculation formula of the first peak value of the spectral line:
；
wherein F_SP1 represents a first peak value of a spectral line, f_max represents a maximum value of a level value, the matrix represents a level matrix of a signal, and amp (fi, tj) represents a level value of signal time tj and signal frequency fi;
the second peak value of the spectral line refers to the extreme value of the second level of the extreme value of the level amplitude value in the signal duration and bandwidth, namely, the second extreme value of the level value is selected from the data of the signal starting time t1 to the signal ending time t2 and the signal starting frequency fre_start to the signal ending frequency fre_end in the frequency range from the minute-level aggregation table of the equipment, and the second extreme value of the spectral line is obtained;
The calculation formula of the second peak of the spectral line:
；
wherein F_SP2 represents a second peak value of the spectral line, f_max represents a maximum value of the level value, the matrix represents a level matrix of the signal, amp (fi, tj) represents a level value of which the signal time is tj and the signal frequency is fi;
the third peak value of the spectral line refers to the extremum of the level amplitude in the signal duration and bandwidth, namely, the third extremum is ranked from the minute-level aggregation table of the equipment, the data from the signal start time t1 to the signal end time t2 and the data from the signal start frequency fre_start to the signal end frequency fre_end in the frequency range are obtained, and the third highest level value is selected from the data to be the third peak value of the spectral line;
the calculation formula of the third peak of the spectral line:
；
wherein F_SP3 represents a third peak value of the spectral line, f_max represents a maximum value of the level value, the matrix represents a level matrix of the signal, amp (fi, tj) represents a level value of which the signal time is tj and the signal frequency is fi;
the short-time energy index of frequency spectrum refers to the square sum of level in the bandwidth and duration of the signal, namely, the data from the signal starting time t1 to the signal ending time t2 in the time range and the data from the signal starting frequency fre_start to the signal ending frequency fre_end in the frequency range are obtained from a minute-level aggregation table of equipment, and all the square sums of the level values of frequency points are added to obtain the short-time energy index of frequency spectrum;
The calculation formula of the spectrum short-time energy index is as follows:
；
wherein F_SSOTt represents a spectrum short-time energy index, amp (fi, tj) represents a level value with signal time tj and signal frequency fi;
the short-time average energy of the frequency spectrum refers to the quotient of the sum of squares of the levels and the total number of frequency points in the bandwidth and duration of the signal;
the calculation formula of the spectrum short-time average energy is as follows:
；
wherein F_SASE represents the spectrum short-time average energy, amp (fi, tj) represents the signal time tj and the signal frequency fi, and M represents the total number of frequency points in the signal bandwidth range;
the continuous point energy index with the frequency spectrum amplitude larger than the threshold value refers to the total number of the level values of the frequency points in the statistical signal duration and the bandwidth, and the level values are larger than the sum of the back noise and the margin;
the calculation formula of the continuous point energy index with the spectrum amplitude larger than the threshold value is as follows:
wherein F_SPNR represents a continuous point energy index with the spectral amplitude larger than a threshold value, noi represents the sum of back noise and margin, amp (fi, tj) represents a level value with signal time tj and signal frequency fi;
the ratio of the number of the spectrum peaks refers to the ratio of the sum of the numbers of the first peak value of the spectrum lines, the second peak value of the spectrum lines and the third peak value of the spectrum lines in the signal duration and bandwidth range to the total number of levels;
The calculation formula of the spectrum peak number ratio is as follows:
where f_spnr_r represents the ratio of the number of spectral peaks, f_sp1 represents the first peak of spectral lines, f_sp2 represents the second peak of spectral lines, f_sp3 represents the third peak of spectral lines, and N represents the total number of levels in the signal duration and bandwidth range.
Specifically, the method further comprises the following steps:
step S305, extracting energy domain features: according to the equipment signal range, combining the corresponding minute-level aggregation data to extract the energy domain characteristics, wherein the extracted characteristics comprise: < mean signal level in duration, maximum signal level in duration, median signal level in duration, variance of level in duration >;
the average value of the signal level in the duration represents the average value of the signal duration and the level in the bandwidth, and the calculation formula is as follows:
；
wherein amp (fi, tj) represents a level value of signal time tj and signal frequency fi, and N represents the total level number in the range;
duration signal level maximum, which refers to the signal duration and the level maximum within the bandwidth range;
the median value of the signal level in the duration refers to the median value of the level in the signal duration and bandwidth range, and the median value of the level is estimated by adopting an approximation method due to the relatively large number of level values, and the formula is as follows:
；
Wherein M is e Level median, L represents medianThe lower limit level value of the group where the numbers are located, S m-1 Representing the number of occurrences of the level below the group of median levels, f m Representing the number of times the median of the levels is in the group, d representing the group spacing,indicating the total number of level occurrences.
The level variance in duration refers to the level variance in the signal duration and bandwidth range, and the calculation formula is as follows:
；
where E_cov represents the level variance, avg represents the average of the levels over the duration and bandwidth of the signal, and M represents the total number of levels.
Specifically, the method further comprises the following steps:
step S306, extracting time domain features: according to the equipment signal range, the time domain characteristics are extracted by combining the corresponding minute-level aggregation data, and the extracted characteristics comprise: < signal duration, signal occupancy, signal emission time law curve, signal emission time probability >;
the signal duration refers to the duration of the current signal, and the calculation formula is as follows:
；
wherein t_dura refers to the duration of the signal, t_end refers to the end time of the current signal, and t_start refers to the start time of the current signal;
the signal occupancy index refers to the occupancy of the time/day/week period of the signal in the measurement period; the signal occupancy rate is obtained by exploring a signal similar to the signal in the past week, counting the total signal duration of each hour, each day and each week, and dividing the signal duration by the corresponding time range;
According to each dimension of the signals, obtaining all similar signals in the past week by a similar algorithm, and adding the durations of the signals, so as to calculate the occupancy rate of time/day/week;
the signal emission time rule curve index refers to the time occupation degree of similar signals in the past week.
Specifically, the method further comprises the following steps:
step S307, extracting the location domain features: according to the equipment signal range, combining the corresponding minute-level aggregation data to extract the position domain features, wherein the extracted features comprise: < Direction finding intersection, direction finding indicator, signal coverage monitoring station >;
the direction finding intersection index refers to a longitude and latitude intersection point obtained by using two or more stations to measure the direction of the frequency point, and the specific calculation steps are as follows:
s307a, selecting the frequency point measuring station and the adjacent station starting side, wherein the number of the adjacent stations is more than or equal to one station, and starting single-frequency direction finding for the frequency point for the stations, and the direction finding time is about 5 minutes;
s307b, counting the number of times of each direction indicating degree, and carrying out hierarchical clustering to obtain the maximum possible direction indicating degree of the signal;
s307c, intersecting the direction-finding ray earth sphere of each site to obtain geographic intersection points, and performing unsupervised clustering on the intersection points to obtain the position of the maximum probability; taking the centroid of the first largest clustering result as the direction finding intersection of the frequency point;
S307d, acquiring an associated monitoring station of the signal according to station coverage;
the direction finding indicator is the direction finding degree of the frequency point at a direction finding site, and the calculation steps are as follows:
s307e, selecting the frequency point measuring station and starting single-frequency direction finding of the frequency point, wherein the direction finding time is more than 5 minutes;
s307f, clustering the direction-finding degree of the direction-finding result, controlling the clustering angle deviation within 10 degrees, and taking the largest clustering result as the direction-finding degree;
the signal monitoring coverage station index refers to an associated station capable of monitoring the signal, the station position is determined through the direction finding point of the signal, and the coverage monitoring station in the simulation range is obtained through station simulation.
Specifically, the method further comprises the following steps:
step S308, extracting content domain features: according to the equipment signal range, the content domain characteristics are extracted by combining the corresponding minute-level aggregation data, and the extracted characteristics comprise: < illegal content judgment factor, audio signal normalized kurtosis >;
the illegal content judgment factor is used for judging the validity of the demodulated content;
the audio signal normalized kurtosis index refers to that the demodulated audio signal is divided into two sections of silence and sound, and normalization calculation is carried out.
Specifically, the method further comprises the following steps:
step S309, extraction of modulation domain features:
and extracting modulation domain features according to the extracted signal information and the minute-level aggregation data, wherein the extracted features comprise the following steps: < modulation scheme, signal baud rate, communication scheme >;
the signal modulation mode index is used for acquiring the modulation mode of the signal through IQ measurement data;
the signal baud rate index is calculated to obtain the number of code element symbols transmitted by each second of signal through IQ measurement data;
and the communication system index is used for judging a common communication system through IQ measurement data, wherein the common communication system comprises a conventional AM/FM.
The invention develops and designs a method for extracting radio signal characteristics by utilizing a big data platform, a radio network monitoring and machine learning algorithm, and comprehensively realizes acquisition of fine images such as signal frequency, bandwidth, time distribution, position distribution, energy distribution and the like, thereby better grasping panoramic information of signals.
Drawings
FIG. 1 is a schematic diagram of the structure of the present invention;
FIG. 2 is a flow chart of the present invention;
FIG. 3 is a flow chart of frequency domain extraction of the present invention;
FIG. 4 is a flow chart of energy domain extraction of the present invention;
FIG. 5 is a time domain extraction flow chart of the present invention;
FIG. 6 is a flow chart of the location domain extraction of the present invention;
in the figure: the system comprises a monitoring networking application system 1, a data acquisition management platform 1a, a data storage frame-decoding module 2, a real-time processing cluster 2a, an offline processing cluster 2b, a big data platform 2c, a signal extraction module 3 and a device signal library 4.
Detailed Description
The invention is further described below by way of examples with reference to the accompanying drawings.
A radio signal characteristic automatic extraction system comprises a monitoring networking application system 1, a data storage frame-decoding module 2, a signal extraction module 3 and a device signal library 4.
The monitoring networking application system 1 is internally provided with a data acquisition management and control platform 1a; the monitoring networking application system 1 realizes networking of the full-province monitoring equipment, gathers data of the issued tasks and provides an interface for data transmission for the data storage frame-decoding module 2.
The data storage frame-decoding module 2 comprises a real-time processing cluster 2a, an off-line processing cluster 2b and a big data platform 2c; the real-time processing cluster 2a performs frame decomposition, back noise calculation and data aggregation by connecting a monitoring data interface of the data acquisition management platform 1a and using a real-time data processing cluster (flink, store) and the like; the offline processing cluster 2b performs frame decomposition, back noise calculation and data aggregation on the monitoring data acquired and uploaded in a networking manner through tools such as MR, spark and the like, and the big data platform 2c stores mass monitoring data.
The signal extraction module 3 is in signal connection with the data storage frame-decoding module 2 and the equipment signal library 4; the signal extraction module 3 receives the data of the deframed aggregation of the data storage deframed module 2 and obtains the data of the dimensions of the position domain, the content domain, the modulation domain, the frequency domain, the energy domain and the time domain from the data.
The device signal library 4 is used for extracting signals, storing the extracted signals according to dimensions of days, devices and the like, and comparing and classifying the signals subsequently.
A method for automatically extracting radio signal characteristics, comprising the steps of:
in step S1, the monitoring networking application system 1 realizes networking of the full-province monitoring device, and can issue a monitoring task according to requirements, and simultaneously provides a monitoring task interface and offline monitoring data for the data storage frame-decoding module 2, and the offline monitoring data is sent to the big data platform 2c through a transmission tool.
A monitoring task is issued in the monitoring networking application system 1, and for the real-time sweep monitoring task being executed, the monitoring networking application system 1 exposes a socket interface to a real-time processing cluster 2a of the data storage frame-decoding module 2; meanwhile, the monitoring networking application system 1 also supports the storage of monitoring data and periodically uploads the monitoring data to the big data platform 2c of the data storage and deframer module 2 for storage.
Step S2, the real-time processing cluster 2a of the data storage frame-decoding module 2 is connected with a monitoring data interface of the monitoring networking application system 1 according to the signal extraction requirement, and after the received data stream is frame-decoded, aggregation is carried out according to the minute level; the offline processing cluster 2b reads the data stored in the HDFS, and after the frame is removed, the data is aggregated according to the minute level; the minute-level aggregated data obtained after the processing of the real-time processing cluster 2a and the offline processing cluster 2b waits for the signal extraction by the signal extraction module 3.
The data storage frame-decoding module 2 carries out frame decoding on the monitoring data according to an atomic protocol, and divides the monitoring data into real-time processing and off-line processing, wherein the real-time processing mainly connects an IP address and a task ID of a socket interface provided by a connection application, and receives a data stream in real time; and the off-line processing is to mainly acquire monitoring data from the stored big data platform and perform frame decomposition.
The data are aggregated according to the minute level after the frame is removed, and under the condition that the signal extraction is not affected, the original data samples are aggregated to a certain extent, so that the calculated amount can be effectively reduced, the calculated time is saved, and the faster signal extraction and discovery are realized. The aggregated data is stored and saved according to the equipment classification, and is divided into: time, frequency, level value, back noise, device ID.
Step S3, the signal extraction module 3 forms a gray scale map from the aggregated minute-level aggregated data according to the corresponding background noise margin, and extracts signals by corroding and expanding the gray scale map, wherein the characteristic values of the extracted signals include: frequency domain features, energy domain features, time domain features, location domain features, content domain features, and modulation domain features.
Specifically, step S3 includes the following steps:
step S301, signal extraction condition setting; the signal extraction module 3 performs signal extraction from the aggregated minute-level data, and the signal extraction conditions are < equipment number, signal bandwidth, signal duration interval, signal extraction frequency interval, signal extraction time interval, and background noise margin >.
For example, the signal extraction module 3 performs signal extraction from the aggregated minute-level data, according to the conditions of signal extraction: < equ1, 100KHZ, 0.5h, 87-137, 20200620-20200621, 3>.
Wherein equ1 represents a device number, 100KHZ represents a minimum bandwidth of an extracted signal, 0.5h represents a minimum duration of the signal, 87 to 137 represent a signal extraction frequency interval, 20200620 to 20200621 represent a signal extraction time interval, and 3 represents a backnoise margin.
Step S302, corresponding scanning data are obtained from the aggregated minute-level data according to the equipment number, the signal duration time interval and the signal extraction interval, a level matrix is formed for the data, and meanwhile, the level matrix is formed by comparing the level value with the background noise margin, so that a gray level diagram matrix is formed, wherein the comparison rule is as follows:
The signal extraction module 3 extracts a device number, a signal extraction time interval and a signal extraction frequency interval from the aggregated minute-level data, and forms a level matrix for the data: the data frequency points in each minute are arranged in an ascending order to form a matrix with frequency points as columns and time as rows; comparing the level value d (i, j) corresponding to the frequency point in the matrix with the background noise margin, if the level value d (i, j) is larger than the background noise margin, taking the value of the unit of the gray matrix as 1, otherwise, taking the value as 0, namely, if the level value d (i, j) is larger than the background noise margin, taking the value of the gray map corresponding to the frequency point and the corresponding time as 1, otherwise, taking the value as 0, and finally forming the gray matrix;
closing the gray matrix, and performing expansion operation according to the matrix unit of 1*3; and (3) performing corrosion operation on the expanded gray matrix according to a matrix unit of 1*3, removing burrs and jitter error values, and forming a gray image of smooth and noiseless data.
Step S303, extracting device signal range information: extracting signals from a gray scale graph according to the signal bandwidth and the signal duration, wherein gray scale rectangles with extraction value columns larger than 2 and rows larger than 30 are used as signals in the gray scale graph, and the signal range information of the extracting equipment is as follows: < signal start time t1, signal end time tn, signal start bandwidth fre_start, signal end bandwidth fre_end >.
Step S304, extracting frequency domain features: according to the equipment signal range, combining the corresponding minute-level aggregation data, extracting frequency domain features, wherein the extracted features comprise: < center frequency, signal bandwidth, spectral line first peak, spectral line second peak, spectral line third peak, spectral short time energy, spectral short time average energy, spectral amplitude greater than threshold continuous point energy, spectral peak number ratio >.
The center frequency calculation formula: fefr= (fre_end+fre_start)/2.
The signal bandwidth calculation formula: f_ebd=fre_end-fre_start.
Where f_efr represents the center frequency of the signal, f_ebd represents the bandwidth of the signal, fre_end represents the end frequency of the signal, and fre_start represents the start frequency of the signal.
The first peak of the spectral line refers to the maximum extreme value of the level amplitude value in the signal duration and the bandwidth, namely, the maximum level value is selected from the data with the time range from the signal start time t1 to the signal end time t2 and the frequency range from the signal start frequency fre_start to the signal end frequency fre_end in the minute-level aggregation table of the device, and is the first peak of the spectral line.
The calculation formula of the first peak value of the spectral line:
；
Wherein f_sp1 represents the first peak of the spectral line, f_max represents the maximum value of the level value, the matrix represents the level matrix of the signal, and amp (fi, tj) represents the level value of the signal with signal time tj and signal frequency fi.
The second peak of the spectral line refers to the extremum of the second level of the signal duration and the level amplitude value in the bandwidth, namely, the second maximum level value is selected from the data of the signal starting time t1 to the signal ending time t2 and the frequency range from the signal starting frequency fre_start to the signal ending frequency fre_end in the minute-level aggregation table of the equipment.
The calculation formula of the second peak of the spectral line:
；
wherein f_sp2 represents the second peak of the spectral line, f_max represents the maximum value of the level value, the matrix represents the level matrix of the signal, and amp (fi, tj) represents the level value of the signal with signal time tj and signal frequency fi.
The third peak value of the spectral line refers to the extremum of the third level of the level amplitude in the signal duration and bandwidth, namely, the third level value is selected from the data of the signal starting time t1 to the signal ending time t2 and the frequency range from the signal starting frequency fre_start to the signal ending frequency fre_end in the minute-level aggregation table of the equipment.
The calculation formula of the third peak of the spectral line:
；
wherein f_sp3 represents the third peak of the spectral line, f_max represents the maximum value of the level value, the matrix represents the level matrix of the signal, and amp (fi, tj) represents the level value of the signal with signal time tj and signal frequency fi.
The short-time energy index of the frequency spectrum refers to the square sum of the level in the bandwidth and the duration of the signal, namely, the data from the signal starting time t1 to the signal ending time t2 in the time range and the data from the signal starting frequency fre_start to the signal ending frequency fre_end in the frequency range are obtained from a minute-level aggregation table of equipment, and all the square sums of the level values of the frequency points are added, namely, the short-time energy index of the frequency spectrum is obtained.
The calculation formula of the spectrum short-time energy index is as follows:
wherein, F_SSOTt represents a spectrum short-time energy index, and amp (fi, tj) represents a level value with signal time tj and signal frequency fi.
The short-time average energy of the spectrum refers to the sum of squares of the levels and the total number of frequency bins within the signal bandwidth and duration.
The calculation formula of the spectrum short-time average energy is as follows:
wherein, F_SASE represents the spectrum short-time average energy, amp (fi, tj) represents the signal time tj, the signal frequency is the level value of fi, and M represents the total number of frequency points in the signal bandwidth range.
The continuous point energy index with the spectral amplitude greater than the threshold value refers to the total number of the statistical signal duration and the level value of the frequency point in the bandwidth, wherein the level value is greater than the sum of the back noise and the margin.
The calculation formula of the continuous point energy index with the spectrum amplitude larger than the threshold value is as follows:
wherein F_SPNR represents a continuous point energy index with a spectral amplitude greater than a threshold, noi represents the sum of the back noise and the margin, amp (fi, tj) represents a level value with signal time tj and signal frequency fi.
The ratio of the number of spectral peaks refers to the ratio of the sum of the numbers of the first peak of spectral lines, the second peak of spectral lines and the third peak of spectral lines to the total number of levels in the signal duration and bandwidth range.
The calculation formula of the spectrum peak number ratio is as follows:
where f_spnr_r represents the ratio of the number of spectral peaks, f_sp1 represents the first peak of spectral lines, f_sp2 represents the second peak of spectral lines, f_sp3 represents the third peak of spectral lines, and N represents the total number of levels in the signal duration and bandwidth range.
Step S305, extracting energy domain features: according to the equipment signal range, combining the corresponding minute-level aggregation data to extract the energy domain characteristics, wherein the extracted characteristics comprise: < mean signal level over duration, maximum signal level over duration, median signal level over duration, variance of level over duration >.
The average value of the signal level in the duration represents the average value of the signal duration and the level in the bandwidth, and the calculation formula is as follows:
wherein amp (fi, tj) represents a level value of signal time tj and signal frequency fi, and N represents the total number of levels in the range.
Duration signal level maximum refers to the signal duration and the level maximum within the bandwidth.
The median value of the signal level in the duration refers to the median value of the level in the signal duration and bandwidth range, and the median value of the level is estimated by adopting an approximation method due to the relatively large number of level values, and the formula is as follows:
wherein M is e Level median, L represents the lower limit level value of the group in which the median is located, S m-1 Representing the number of occurrences of the level below the group of median levels, f m Representing the number of times the median of the levels is in the group, d representing the group spacing,indicating the total number of level occurrences.
The level variance in duration refers to the level variance in the signal duration and bandwidth range, and the calculation formula is as follows:
where E_cov represents the level variance, avg represents the average of the levels over the duration and bandwidth of the signal, and M represents the total number of levels.
Step S306, extracting time domain features: according to the equipment signal range, the time domain characteristics are extracted by combining the corresponding minute-level aggregation data, and the extracted characteristics comprise: < signal duration, signal occupancy, signal emission time law curve, signal emission time probability >.
The signal duration refers to the duration of the current signal, and the calculation formula is as follows:
where t_dura refers to the duration of the signal, t_end refers to the end time of the current signal, and t_start refers to the start time of the current signal.
The signal occupancy index refers to the occupancy of the period of time/day/week of the signal during the measurement period. Signal occupancy measures are obtained by exploring a signal similar to this signal from the past week and counting its total signal duration per hour, day, week, and dividing the signal duration by the corresponding time range (e.g., hour, day, week).
Based on the dimensions of this signal, all similar signals for the past week are obtained with a similar algorithm, and the durations of these signals are added together to calculate the occupancy of time/day/week.
The signal emission time law curve index refers to the time occupation degree of similar signals in the past week, and the calculation mode is as follows: according to each dimension of the signals, all similar signals in the past week are obtained according to the signal similarity, the duration of the signals is taken as a unit of minutes, and the occurrence time of the signals is obtained, so that a schedule occupied in the week is obtained.
The signal emission time probability is the same as the signal occupancy calculation method.
Step S307, extracting the location domain features: according to the equipment signal range, combining the corresponding minute-level aggregation data to extract the position domain features, wherein the extracted features comprise: < Direction finding intersection, direction finding indicator, signal coverage monitoring station >.
The direction finding intersection index refers to a longitude and latitude intersection point obtained by using two or more stations to measure the direction of the frequency point, and the specific calculation steps are as follows:
s307a, selecting the frequency point measuring station and the adjacent station starting side, wherein the number of the adjacent stations is more than or equal to one station, and starting the single-frequency direction finding of the frequency point for the stations, and the direction finding time is about 5 minutes.
S307b, counting the number of times of each direction indicating degree, and carrying out hierarchical clustering to obtain the maximum possible direction indicating degree of the signal.
S307c, intersecting the direction-finding ray spherical surfaces of each site to obtain geographic intersection points, and performing unsupervised clustering on the intersection points to obtain the position of the maximum probability. The clustering range can be controlled as required, and 500 meters are generally recommended. And taking the centroid of the first largest clustering result as the direction finding intersection of the frequency point.
S307d, according to the station coverage, acquiring the associated monitoring station of the signal (namely, the monitoring station capable of monitoring the signal).
The direction finding indicator is the direction finding degree of the frequency point at a direction finding site, and the calculation steps are as follows:
s307e, selecting the frequency point measuring station and starting single-frequency direction finding of the frequency point, wherein the direction finding time is more than 5 minutes.
S307f, clustering the direction-finding degree of the direction-finding result, controlling the clustering angle deviation within 10 degrees, and taking the largest clustering result as the direction-finding degree.
The signal monitoring coverage station index refers to an associated station capable of monitoring the signal, the station position is determined through the direction finding point of the signal, and the coverage monitoring station in the simulation range is obtained through station simulation.
Step S308, extracting content domain features: according to the equipment signal range, the content domain characteristics are extracted by combining the corresponding minute-level aggregation data, and the extracted characteristics comprise: < illegal content judgment factor, audio signal normalized kurtosis >.
The illegal content judgment factor is mainly used for judging the legitimacy of the demodulated content.
The audio signal normalized kurtosis index refers to that the demodulated audio signal is divided into two sections of silence and sound, and normalization calculation is carried out.
Step S309, extraction of modulation domain features: and extracting modulation domain features according to the extracted signal information and the minute-level aggregation data, wherein the extracted features comprise the following steps: < modulation scheme, signal baud rate, communication scheme >.
And acquiring a signal modulation mode index through IQ measurement data.
And calculating the signal baud rate index through IQ measurement data to obtain the number of symbol symbols transmitted by each second of signal.
And the communication system index is used for judging a common communication system through IQ measurement data, wherein the common communication system comprises a conventional AM/FM.
And S4, the equipment signal library 4 stores the signals extracted in the step 3 according to equipment and dates for subsequent signal classification and signal alarming.
Through experimental calculation, the automatic extraction method of the radio signal characteristics of the invention supports the automatic discovery of the monitoring task signals under the condition of a networking monitoring platform, and realizes the description of each dimension of a frequency domain, an energy domain, a time domain, a position domain, a content domain and a modulation domain on the signals, finally saves the identified signals, and the description of the signals is more comprehensive and careful, thereby greatly improving the efficiency and the fineness of signal management of staff.
Claims (10)
1. The automatic radio signal characteristic extraction system is characterized by comprising a monitoring networking application system (1), a data storage frame-decoding module (2), a signal extraction module (3) and a device signal library (4);
the monitoring networking application system (1) is internally provided with a data acquisition management and control platform (1 a); the monitoring networking application system (1) realizes networking of the full-province monitoring equipment, gathers data of an issued task and provides a data transmission interface for the data storage frame-decoding module (2);
the data storage frame-decoding module (2) comprises a real-time processing cluster (2 a), an offline processing cluster (2 b) and a big data platform (2 c); the real-time processing cluster (2 a) is connected with a monitoring data interface of the data acquisition management and control platform (1 a) and is used for frame decomposition, back noise calculation and data aggregation; the off-line processing cluster (2 b) carries out frame decomposition, back noise calculation and data aggregation on the monitoring data acquired and uploaded by the internet, and the big data platform (2 c) stores massive monitoring data;
the signal extraction module (3) is in signal connection with the data storage frame-decoding module (2) and the equipment signal library (4); the signal extraction module (3) receives the data of the frame de-aggregation of the data storage frame de-aggregation module (2) and obtains dimension data of a position domain, a content domain, a modulation domain, a frequency domain, an energy domain and a time domain from the data; comprises the following steps:
Step 1, setting signal extraction conditions; the signal extraction module (3) performs signal extraction from the aggregated minute-level data, wherein the signal extraction conditions are < equipment number, signal bandwidth, signal duration interval, signal extraction frequency interval, signal extraction time interval and background noise margin >;
step 2, acquiring corresponding scanning data from the aggregated minute-level data according to the equipment number, the signal duration interval and the signal extraction interval, forming a level matrix for the data, and simultaneously comparing the level value with a background noise margin to form a gray level map matrix;
in step 2, the method comprises the following steps: the signal extraction module (3) extracts equipment numbers, signal extraction time intervals and signal extraction frequency intervals from the aggregated minute-level data, and forms a level matrix for the data: the data frequency points in each minute are arranged in an ascending order to form a matrix with frequency points as columns and time as rows; comparing the level value d (i, j) corresponding to the frequency point in the matrix with the background noise margin, if the level value d (i, j) is larger than the background noise margin, taking the value of the unit of the gray matrix as 1, otherwise, taking the value as 0, namely, if the level value d (i, j) is larger than the background noise margin, taking the value of the gray map corresponding to the frequency point and the corresponding time as 1, otherwise, taking the value as 0, and finally forming the gray matrix;
Closing the gray matrix, and performing expansion operation according to the matrix unit of 1*3; performing corrosion operation on the expanded gray matrix according to a 1*3 matrix unit, removing burrs and jitter error values, and forming a gray image of smooth and noiseless data;
step 3, extracting equipment signal range information: extracting signals from a gray scale graph according to the signal bandwidth and the signal duration, wherein gray scale rectangles with extraction value columns larger than 2 and rows larger than 30 are used as signals in the gray scale graph, and the signal range information of the extracting equipment is as follows: < signal start time t1, signal end time tn, signal start bandwidth fre_start, signal end bandwidth fre_end >;
the equipment signal library (4) is used for extracting signals, storing the signals according to the dimensions of the equipment and the days, and comparing and classifying the signals subsequently.
2. A method for automatically extracting characteristics of a radio signal using the system of claim 1, comprising the steps of:
step S1, the monitoring networking application system (1) realizes networking of full-province monitoring equipment, can issue monitoring tasks according to requirements, and simultaneously provides a monitoring task interface and offline monitoring data for the data storage frame-releasing module (2), and the offline monitoring data is sent to the big data platform (2 c) through a transmission tool;
Step S2, the real-time processing cluster (2 a) of the data storage frame-decoding module (2) is connected with a monitoring data interface of the monitoring networking application system (1) according to the signal extraction requirement, and after the received data stream is frame-decoded, aggregation is carried out according to the minute level; the offline processing cluster (2 b) reads the data stored in the HDFS, and performs aggregation according to a minute level after frame decomposition; the minute-level aggregated data obtained after the processing of the real-time processing cluster (2 a) and the offline processing cluster (2 b) waits for signal extraction by the signal extraction module (3);
step S3, the signal extraction module (3) forms a gray scale graph according to the corresponding background noise margin from the aggregated minute-level aggregated data, and extracts signals through corrosion and expansion of the gray scale graph, wherein the characteristic values of the extracted signals comprise: frequency domain features, energy domain features, time domain features, location domain features, content domain features, and modulation domain features;
and S4, the equipment signal library (4) stores the signals extracted in the step S3 according to equipment and dates for subsequent signal classification and signal alarming.
3. The automatic radio signal feature extraction method according to claim 2, characterized in that in step S3, it comprises the steps of:
Step S301, signal extraction condition setting; the signal extraction module (3) performs signal extraction from the aggregated minute-level data, wherein the signal extraction conditions are < equipment number, signal bandwidth, signal duration interval, signal extraction frequency interval, signal extraction time interval and background noise margin >;
step S302, corresponding scanning data are obtained from the aggregated minute-level data according to the equipment number, the signal duration interval and the signal extraction interval, a level matrix is formed for the data, and meanwhile, the level matrix is compared with a background noise margin according to a level value to form a gray level map matrix;
step S303, extracting device signal range information: extracting signals from a gray scale graph according to the signal bandwidth and the signal duration, wherein gray scale rectangles with extraction value columns larger than 2 and rows larger than 30 are used as signals in the gray scale graph, and the signal range information of the extracting equipment is as follows: < signal start time t1, signal end time tn, signal start bandwidth fre_start, signal end bandwidth fre_end >.
4. A method for automatically extracting features of a radio signal as claimed in claim 3, comprising the steps of: the signal extraction module (3) extracts equipment numbers, signal extraction time intervals and signal extraction frequency intervals from the aggregated minute-level data, and forms a level matrix for the data: the data frequency points in each minute are arranged in an ascending order to form a matrix with frequency points as columns and time as rows; comparing the level value d (i, j) corresponding to the frequency point in the matrix with the background noise margin, if the level value d (i, j) is larger than the background noise margin, taking the value of the unit of the gray matrix as 1, otherwise, taking the value as 0, namely, if the level value d (i, j) is larger than the background noise margin, taking the value of the gray map corresponding to the frequency point and the corresponding time as 1, otherwise, taking the value as 0, and finally forming the gray matrix;
Closing the gray matrix, and performing expansion operation according to the matrix unit of 1*3; and (3) performing corrosion operation on the expanded gray matrix according to a matrix unit of 1*3, removing burrs and jitter error values, and forming a gray image of smooth and noiseless data.
5. A method for automatic extraction of radio signal characteristics according to claim 3 or 4, further comprising the steps of:
step S304, extracting frequency domain features: according to the equipment signal range, combining the corresponding minute-level aggregation data, extracting frequency domain features, wherein the extracted features comprise: < center frequency, signal bandwidth, spectral line first peak, spectral line second peak, spectral line third peak, spectral short-time energy, spectral short-time average energy, spectral amplitude greater than threshold continuous point energy, spectral peak number ratio >;
the center frequency calculation formula: fefr= (fre_end+fre_start)/2;
the signal bandwidth calculation formula: f_ebd=fre_end-fre_start;
wherein, F_EFR represents the center frequency of the signal, F_EBD represents the bandwidth of the signal, fre_end represents the end frequency of the signal, fre_start represents the start frequency of the signal;
the first peak value of the spectral line refers to the maximum extreme value of the level amplitude value in the signal duration and bandwidth, namely, the maximum level value is selected from the data of which the time range is from the signal start time t1 to the signal end time t2 and the frequency range is from the signal start frequency fre_start to the signal end frequency fre_end in the minute-level aggregation table of the equipment, and is the first peak value of the spectral line;
The calculation formula of the first peak value of the spectral line:
wherein F_SP1 represents a first peak value of a spectral line, f_max represents a maximum value of a level value, the matrix represents a level matrix of a signal, and amp (fi, tj) represents a level value of signal time tj and signal frequency fi;
the second peak value of the spectral line refers to the extreme value of the second level of the extreme value of the level amplitude value in the signal duration and bandwidth, namely, the second extreme value of the level value is selected from the data of the signal starting time t1 to the signal ending time t2 and the signal starting frequency fre_start to the signal ending frequency fre_end in the frequency range from the minute-level aggregation table of the equipment, and the second extreme value of the spectral line is obtained;
the calculation formula of the second peak of the spectral line:
；
wherein F_SP2 represents a second peak value of the spectral line, f_max represents a maximum value of the level value, the matrix represents a level matrix of the signal, amp (fi, tj) represents a level value of which the signal time is tj and the signal frequency is fi;
the third peak value of the spectral line refers to the extremum of the level amplitude in the signal duration and bandwidth, namely, the third extremum is ranked from the minute-level aggregation table of the equipment, the data from the signal start time t1 to the signal end time t2 and the data from the signal start frequency fre_start to the signal end frequency fre_end in the frequency range are obtained, and the third highest level value is selected from the data to be the third peak value of the spectral line;
The calculation formula of the third peak of the spectral line:
；
wherein F_SP3 represents a third peak value of the spectral line, f_max represents a maximum value of the level value, the matrix represents a level matrix of the signal, amp (fi, tj) represents a level value of which the signal time is tj and the signal frequency is fi;
the short-time energy index of frequency spectrum refers to the square sum of level in the bandwidth and duration of the signal, namely, the data from the signal starting time t1 to the signal ending time t2 in the time range and the data from the signal starting frequency fre_start to the signal ending frequency fre_end in the frequency range are obtained from a minute-level aggregation table of equipment, and all the square sums of the level values of frequency points are added to obtain the short-time energy index of frequency spectrum;
the calculation formula of the spectrum short-time energy index is as follows:
；
wherein F_SSOTt represents a spectrum short-time energy index, amp (fi, tj) represents a level value with signal time tj and signal frequency fi;
the short-time average energy of the frequency spectrum refers to the quotient of the sum of squares of the levels and the total number of frequency points in the bandwidth and duration of the signal;
the calculation formula of the spectrum short-time average energy is as follows:
；
wherein F_SASE represents the spectrum short-time average energy, amp (fi, tj) represents the signal time tj and the signal frequency fi, and M represents the total number of frequency points in the signal bandwidth range;
The continuous point energy index with the frequency spectrum amplitude larger than the threshold value refers to the total number of the level values of the frequency points in the statistical signal duration and the bandwidth, and the level values are larger than the sum of the back noise and the margin;
the calculation formula of the continuous point energy index with the spectrum amplitude larger than the threshold value is as follows:
wherein F_SPNR represents a continuous point energy index with the spectral amplitude larger than a threshold value, noi represents the sum of back noise and margin, amp (fi, tj) represents a level value with signal time tj and signal frequency fi;
the ratio of the number of the spectrum peaks refers to the ratio of the sum of the numbers of the first peak value of the spectrum lines, the second peak value of the spectrum lines and the third peak value of the spectrum lines in the signal duration and bandwidth range to the total number of levels;
the calculation formula of the spectrum peak number ratio is as follows:
；
where f_spnr_r represents the ratio of the number of spectral peaks, f_sp1 represents the first peak of spectral lines, f_sp2 represents the second peak of spectral lines, f_sp3 represents the third peak of spectral lines, and N represents the total number of levels in the signal duration and bandwidth range.
6. A method for automatic extraction of radio signal characteristics according to claim 3 or 4, further comprising the steps of:
step S305, extracting energy domain features: according to the equipment signal range, combining the corresponding minute-level aggregation data to extract the energy domain characteristics, wherein the extracted characteristics comprise: < mean signal level in duration, maximum signal level in duration, median signal level in duration, variance of level in duration >;
The average value of the signal level in the duration represents the average value of the signal duration and the level in the bandwidth, and the calculation formula is as follows:
；
wherein amp (fi, tj) represents a level value of signal time tj and signal frequency fi, and N represents the total level number in the range;
duration signal level maximum, which refers to the signal duration and the level maximum within the bandwidth range;
the median value of the signal level in the duration refers to the median value of the level in the signal duration and bandwidth range, and the median value of the level is estimated by adopting an approximation method due to the relatively large number of level values, and the formula is as follows:
；
wherein M is e Level median, L represents the lower limit level value of the group in which the median is located, S m-1 Representing the levelThe number of occurrences of the level below the group of median, f m Representing the number of times the median of the levels is in the group, d representing the group spacing,representing the total number of level occurrences;
the level variance in duration refers to the level variance in the signal duration and bandwidth range, and the calculation formula is as follows:
；
where E_cov represents the level variance, avg represents the average of the levels over the duration and bandwidth of the signal, and M represents the total number of levels.
7. A method for automatic extraction of radio signal characteristics according to claim 3 or 4, further comprising the steps of:
Step S306, extracting time domain features: according to the equipment signal range, the time domain characteristics are extracted by combining the corresponding minute-level aggregation data, and the extracted characteristics comprise: < signal duration, signal occupancy, signal emission time law curve, signal emission time probability >;
the signal duration refers to the duration of the current signal, and the calculation formula is as follows:
；
wherein t_dura refers to the duration of the signal, t_end refers to the end time of the current signal, and t_start refers to the start time of the current signal;
the signal occupancy index refers to the occupancy of the time/day/week period of the signal in the measurement period; the signal occupancy rate is obtained by exploring a signal similar to the signal in the past week, counting the total signal duration of each hour, each day and each week, and dividing the signal duration by the corresponding time range;
according to each dimension of the signals, obtaining all similar signals in the past week by a similar algorithm, and adding the durations of the signals, so as to calculate the occupancy rate of time/day/week;
the signal emission time rule curve index refers to the time occupation degree of similar signals in the past week.
8. A method for automatic extraction of radio signal characteristics according to claim 3 or 4, further comprising the steps of:
step S307, extracting the location domain features: according to the equipment signal range, combining the corresponding minute-level aggregation data to extract the position domain features, wherein the extracted features comprise: < Direction finding intersection, direction finding indicator, signal coverage monitoring station >;
the direction finding intersection index refers to a longitude and latitude intersection point obtained by using two or more stations to measure the direction of the frequency point, and the specific calculation steps are as follows:
s307a, selecting the frequency point measuring station and the adjacent station starting side, wherein the number of the adjacent stations is more than or equal to one station, and starting single-frequency direction finding for the frequency point for the stations, and the direction finding time is about 5 minutes;
s307b, counting the number of times of each direction indicating degree, and carrying out hierarchical clustering to obtain the maximum possible direction indicating degree of the signal;
s307c, intersecting the direction-finding ray earth sphere of each site to obtain geographic intersection points, and performing unsupervised clustering on the intersection points to obtain the position of the maximum probability; taking the centroid of the first largest clustering result as the direction finding intersection of the frequency point;
S307d, acquiring an associated monitoring station of the signal according to station coverage;
the direction finding indicator is the direction finding degree of the frequency point at a direction finding site, and the calculation steps are as follows:
s307e, selecting the frequency point measuring station and starting single-frequency direction finding of the frequency point, wherein the direction finding time is more than 5 minutes;
s307f, clustering the direction-finding degree of the direction-finding result, controlling the clustering angle deviation within 10 degrees, and taking the largest clustering result as the direction-finding degree;
the signal monitoring coverage station index refers to an associated station capable of monitoring the signal, the station position is determined through the direction finding point of the signal, and the coverage monitoring station in the simulation range is obtained through station simulation.
9. A method for automatic extraction of radio signal characteristics according to claim 3 or 4, further comprising the steps of:
step S308, extracting content domain features: according to the equipment signal range, the content domain characteristics are extracted by combining the corresponding minute-level aggregation data, and the extracted characteristics comprise: < illegal content judgment factor, audio signal normalized kurtosis >;
the illegal content judgment factor is used for judging the validity of the demodulated content;
the audio signal normalized kurtosis index refers to that the demodulated audio signal is divided into two sections of silence and sound, and normalization calculation is carried out.
10. A method for automatic extraction of radio signal characteristics according to claim 3 or 4, further comprising the steps of:
step S309, extraction of modulation domain features:
and extracting modulation domain features according to the extracted signal information and the minute-level aggregation data, wherein the extracted features comprise the following steps: < modulation scheme, signal baud rate, communication scheme >;
the signal modulation mode index is used for acquiring the modulation mode of the signal through IQ measurement data;
the signal baud rate index is calculated to obtain the number of code element symbols transmitted by each second of signal through IQ measurement data;
and the communication system index is used for judging a common communication system through IQ measurement data, wherein the common communication system comprises a conventional AM/FM.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202011084025.6A CN112257529B (en) | 2020-10-12 | 2020-10-12 | Automatic extraction system and method for radio signal characteristics |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202011084025.6A CN112257529B (en) | 2020-10-12 | 2020-10-12 | Automatic extraction system and method for radio signal characteristics |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN112257529A CN112257529A (en) | 2021-01-22 |
| CN112257529B true CN112257529B (en) | 2023-11-28 |
Family
ID=74243528
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202011084025.6A Active CN112257529B (en) | 2020-10-12 | 2020-10-12 | Automatic extraction system and method for radio signal characteristics |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN112257529B (en) |
Families Citing this family (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113194011B (en) * | 2021-04-29 | 2022-08-26 | 浙江原初数据科技有限公司 | Automatic establishment method and device for radio electromagnetic signal environment |
| CN113378678B (en) * | 2021-06-01 | 2022-04-22 | 华中科技大学 | Multi-domain fault feature extraction method and system for mechanical motion system |
Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110048977A (en) * | 2019-03-14 | 2019-07-23 | 中国人民解放军战略支援部队信息工程大学 | The recognition methods of short-wave signal system and device based on the detection of gray level co-occurrence matrixes textural characteristics |
| CN110334591A (en) * | 2019-05-24 | 2019-10-15 | 西华大学 | A kind of detection of unmanned plane Frequency Hopping Signal and recognition methods based on clustering |
| CN110943882A (en) * | 2019-11-12 | 2020-03-31 | 浙江原初数据科技有限公司 | Real-time black broadcast identification system based on networking monitoring and identification method thereof |
Family Cites Families (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| KR101640074B1 (en) * | 2014-04-03 | 2016-07-15 | 한국전자통신연구원 | Apparatus and method for collecting radio frequency feature of wireless device in wireless communication apparatus |
-
2020
- 2020-10-12 CN CN202011084025.6A patent/CN112257529B/en active Active
Patent Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110048977A (en) * | 2019-03-14 | 2019-07-23 | 中国人民解放军战略支援部队信息工程大学 | The recognition methods of short-wave signal system and device based on the detection of gray level co-occurrence matrixes textural characteristics |
| CN110334591A (en) * | 2019-05-24 | 2019-10-15 | 西华大学 | A kind of detection of unmanned plane Frequency Hopping Signal and recognition methods based on clustering |
| CN110943882A (en) * | 2019-11-12 | 2020-03-31 | 浙江原初数据科技有限公司 | Real-time black broadcast identification system based on networking monitoring and identification method thereof |
Also Published As
| Publication number | Publication date |
|---|---|
| CN112257529A (en) | 2021-01-22 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN112257529B (en) | Automatic extraction system and method for radio signal characteristics | |
| CN110879989B (en) | Ads-b signal target identification method based on small sample local machine learning model | |
| CN109640269B (en) | Fingerprint positioning method based on CSI and time domain fusion algorithm | |
| CN109447067B (en) | Bill direction detection and correction method and automatic bill checking system | |
| CN104700099A (en) | Method and device for recognizing traffic signs | |
| US20220091275A1 (en) | A method and a system for assessing aspects of an electromagnetic signal | |
| CN115187927B (en) | Remote monitoring and management method and system for sightseeing seat | |
| CN108289302B (en) | Method and system for positioning atmospheric waveguide interference of TD-LTE network | |
| CN102073867B (en) | Sorting method and device for remote sensing images | |
| JP2023512905A (en) | Railway-related data analysis system and method | |
| CN106982357A (en) | A kind of intelligent camera system based on distribution clouds | |
| CN116108491B (en) | Data leakage early warning method, device and system based on semi-supervised federal learning | |
| EP1724755B1 (en) | Method and system for comparing audio signals and identifying an audio source | |
| CN114708545A (en) | Image-based object detection method, device, equipment and storage medium | |
| CN104392101B (en) | Data sharing method and device | |
| CN115496097A (en) | Non-line-of-sight signal identification method based on wavelet gram convolution neural network | |
| Colak et al. | Automatic sunspot classification for real-time forecasting of solar activities | |
| Cui et al. | Exploiting spatial signatures of power ENF signal for measurement source authentication | |
| CN116304560B (en) | Track characterization model training method, characterization method and device | |
| CN114422049B (en) | Frequency spectrum monitoring big data cleaning method and system based on deep learning detection | |
| CN115184744B (en) | GIS ultrahigh frequency discharge signal detection device and method based on fast-RCNN | |
| CN113194011B (en) | Automatic establishment method and device for radio electromagnetic signal environment | |
| CN117746342B (en) | Method for identifying road ponding by utilizing public video | |
| CN110807549A (en) | Generation method, generation device, generation system and electronic equipment of meteorological prediction model | |
| CN114252889A (en) | Image control point acquisition method and device based on CORS station |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| CB03 | Change of inventor or designer information |
Inventor after: Tu Yongsheng Inventor after: Ma Gaofeng Inventor after: Fu Zuoming Inventor after: Zhang Pengcheng Inventor after: Li Jiangmin Inventor before: Ma Gaofeng Inventor before: Fu Zuoming Inventor before: Zhang Pengcheng Inventor before: Tu Yongsheng Inventor before: Li Jiangmin |
|
| CB03 | Change of inventor or designer information | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |