CN113572555A - Black broadcast monitoring method based on zero sample learning - Google Patents
Black broadcast monitoring method based on zero sample learning Download PDFInfo
- Publication number
- CN113572555A CN113572555A CN202110943763.XA CN202110943763A CN113572555A CN 113572555 A CN113572555 A CN 113572555A CN 202110943763 A CN202110943763 A CN 202110943763A CN 113572555 A CN113572555 A CN 113572555A
- Authority
- CN
- China
- Prior art keywords
- attribute
- signal
- channel
- legal
- characteristic
- 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.)
- Granted
Links
- 238000000034 method Methods 0.000 title claims abstract description 74
- 238000012544 monitoring process Methods 0.000 title claims abstract description 57
- 230000010287 polarization Effects 0.000 claims abstract description 16
- 238000001228 spectrum Methods 0.000 claims abstract description 11
- 238000005259 measurement Methods 0.000 claims description 30
- 238000005516 engineering process Methods 0.000 claims description 9
- 238000005070 sampling Methods 0.000 claims description 6
- 238000003058 natural language processing Methods 0.000 claims description 3
- 238000000354 decomposition reaction Methods 0.000 claims description 2
- 238000012552 review Methods 0.000 claims description 2
- 238000012806 monitoring device Methods 0.000 claims 1
- 238000006243 chemical reaction Methods 0.000 abstract description 18
- 230000008569 process Effects 0.000 abstract description 7
- 238000010801 machine learning Methods 0.000 abstract description 4
- 238000000691 measurement method Methods 0.000 abstract 2
- 230000007613 environmental effect Effects 0.000 description 8
- 238000001514 detection method Methods 0.000 description 5
- 238000013473 artificial intelligence Methods 0.000 description 4
- 238000004891 communication Methods 0.000 description 4
- 238000010586 diagram Methods 0.000 description 4
- 238000007726 management method Methods 0.000 description 4
- 238000004364 calculation method Methods 0.000 description 3
- 230000006870 function Effects 0.000 description 3
- 238000012549 training Methods 0.000 description 2
- 230000004075 alteration Effects 0.000 description 1
- 238000004458 analytical method Methods 0.000 description 1
- 230000006399 behavior Effects 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 230000033228 biological regulation Effects 0.000 description 1
- 238000012512 characterization method Methods 0.000 description 1
- 238000013479 data entry Methods 0.000 description 1
- 238000013500 data storage Methods 0.000 description 1
- 238000013135 deep learning Methods 0.000 description 1
- 230000007547 defect Effects 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 239000003814 drug Substances 0.000 description 1
- 229940079593 drug Drugs 0.000 description 1
- 230000005670 electromagnetic radiation Effects 0.000 description 1
- 238000007667 floating Methods 0.000 description 1
- 230000036541 health Effects 0.000 description 1
- 238000007689 inspection Methods 0.000 description 1
- 238000013507 mapping Methods 0.000 description 1
- 230000007246 mechanism Effects 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 238000003062 neural network model Methods 0.000 description 1
- 238000005457 optimization Methods 0.000 description 1
- 230000008520 organization Effects 0.000 description 1
- 238000012545 processing Methods 0.000 description 1
- 230000035945 sensitivity Effects 0.000 description 1
- 238000000926 separation method Methods 0.000 description 1
- 238000006467 substitution reaction Methods 0.000 description 1
Images
Classifications
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04H—BROADCAST COMMUNICATION
- H04H60/00—Arrangements for broadcast applications with a direct linking to broadcast information or broadcast space-time; Broadcast-related systems
- H04H60/29—Arrangements for monitoring broadcast services or broadcast-related services
-
- 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
The invention relates to a black broadcast monitoring method based on zero sample learning, relates to the field of radio monitoring, and aims to solve the technical problems of lack of intelligent degree, low accuracy and efficiency and the like in the prior art. The invention designs a black broadcast identification process according to a zero sample learning principle in machine learning. The method comprises the steps of identifying the frequency spectrum characteristic, the bandwidth characteristic, the carrier-to-noise ratio characteristic, the power characteristic, the polarization mode characteristic and the voice attribute characteristic of a broadcast signal at a specific moment, converting the identified signal characteristic from the characteristic space into whether legal attributes exist in the attribute space according to a preset error threshold, and finally identifying the signal as a legal or illegal signal in the attribute space according to the attributes of the broadcast signal and a defined distance measurement method. On the basis of signal feature identification, the invention introduces an attribute conversion and measurement method, improves the intelligent level of black broadcast signal identification, and can effectively improve the efficiency and accuracy of black broadcast signal identification.
Description
Technical Field
The invention relates to the technical field of communication, in particular to a black broadcast monitoring method based on zero sample learning.
Background
Radio monitoring is an important means for maintaining radio order, and technical monitoring is carried out on radio by utilizing monitoring equipment, so that characteristic measurement, direction finding positioning, monitoring and recording of radio signals are obtained, and acquisition, recording and analysis of the characteristics are important means for realizing radio frequency spectrum order management. The black broadcasting station is illegal broadcasting station, including broadcasting station which is not authorized by related organization to increase frequency, private setting up broadcasting illegal drugs, health products advertisement. In recent years, the black broadcasting of FM broadcasting stations established by lawbreakers privately becomes rampant, the black broadcasting disturbs the normal radio broadcasting order, is very easy to interfere civil aviation communication navigation, exceeds the standard of electromagnetic radiation and infringes the lawful rights and interests of people, and is very easy to become a tool for lawbreakers to issue bad information and launch propaganda, so that the national security and social stability are directly influenced, severe social influence is caused, and the black broadcasting becomes a key point and a difficult point for governing relevant law enforcement departments such as radio and television in recent years.
In the radio management regulation of the people's republic of China implemented in 2016 and the explanation on the problems of law application of criminal cases such as disturbing radio communication management order published in 2017, it is clear to strengthen the protection and detection of special frequency of important industries and to severely penalize black broadcasting illegal criminal behaviors. Therefore, efficient identification finds black broadcasts to be of great significance for radio spectrum order management.
Zero sample learning is a class of training scenarios and solutions in the machine learning category. Zero-sample learning techniques were originally derived from the field of image recognition and are now being specifically studied and applied in a number of artificial intelligence sub-fields. In the field of machine learning, a classical supervised learning model needs a large amount of sample data with classification labels to train a learner, and experience is solidified in the learning model through multiple iterations of data, so that automatic classification from a feature space to a classification space is realized. In reality, the types of objects obey long-tail distribution, a large number of types of objects have few samples or no samples are collected, and zero-sample learning or attribute learning aims to realize zero-sample classification by learning the mapping from a feature space to an attribute space and by a specific attribute classification method.
With the rapid development of electronic communication technology, the cost of black broadcasting equipment is continuously reduced, the volume of the equipment is continuously reduced, and the performances of starting at any time, flexible concealment and the like can be realized. The existing manual monitoring and identifying method needs to consume a large amount of labor cost and energy, meanwhile, the supervision and the reporting of the masses are relied on, and professionals repeatedly measure, record and collect evidence in the field have certain hysteresis, so that the method can be used by lawless persons in practice. At present, the technical means of realizing automatic black broadcast discovery and alarm recording by the technical means has a great progress, and the technical means of discovering and identifying black broadcasts by an artificial intelligence method is also emerging. However, the black broadcast has characteristics of simulating and encroaching legal broadcast signals, a machine learning model in artificial intelligence needs a large amount of sample data to supervise a training model, meanwhile, signal attenuation, measurement errors and the like exist in an actual measurement channel, and the monitoring efficiency and the identification accuracy of the conventional artificial intelligence method need to be further improved.
Disclosure of Invention
Technical problem to be solved
Aiming at the defects of the prior art, the invention provides a black broadcast monitoring method based on zero sample learning, and solves the problems of low efficiency, low recognition accuracy and low intelligence degree of the prior monitoring technology.
(II) technical scheme
The invention provides the following technical scheme: a black broadcast monitoring method based on zero sample learning comprises the following steps: the method comprises the following steps:
s1, using the radio monitoring equipment to perform full-band spectrum scanning on the broadcast signal of the specific monitoring place at the specific time t, and generating a corresponding channel setThereinRepresenting a particular channel signal.
The real-time spectrum scanning is carried out on the broadcast signals at a specific moment t, the moment t is consistent with the sampling moment of legal channel signals corresponding to the region in the database, and if the site or the moment t is not recorded in the database, manual legal channel data entry needs to be carried out at the site and the moment in a manual mode.
S2, for each channel signal in the channel setObtaining its frequency characteristic FfBandwidth characteristic FWCarrier to noise ratio characteristic FRPower characteristic FPPolarization mode characteristic FPMVoice attribute feature Ftext。
And S3, comparing the corresponding characteristic records in the database one by one, and judging whether the signal has legal attributes according to the definition and the error range of the characteristics.
Wherein the frequency characteristic FfConverting into attribute A of channel to be judged according to equipment identification precision and allowable error rangefOptionally, the frequency error range is set to ± 5% of the legal channel bandwidth. The specific conversion method is that if the error range is within the error range, the error range is connected with a legal record S in the databasefDBIf the frequency is matched, the channel S is markedfProperty A offIf the channel signal is not recorded in the database, the monitored channel S is marked 1fAll attributes A ofi=0。
Wherein the monitored channel bandwidth characteristic FWAttribute converted into bandwidth characteristic according to equipment identification precision and allowable error rangeOptionally, if the bandwidth error range is set to ± 5% of the legal channel bandwidth, the specific conversion method is that, if the bandwidth error range is within the error range, the legal record of the channel with the same frequency in the database is recordedMarking the channel S to be judged if the bandwidth is accordantfProperty (2) ofIf the error range does not match the legal record of the same frequency channel in the database, marking the channel S to be judgedfProperty (2) of
Wherein the monitored channel carrier-to-noise ratio characteristic FRAttribute converted into carrier-to-noise ratio characteristic according to equipment identification precision and allowable error rangeOptionally, if the carrier-to-noise ratio error range is set to ± 5% of the carrier-to-noise ratio of the legal channel, the specific conversion method is that if the carrier-to-noise ratio error range is within the error range, the legal record of the same frequency channel in the database is recordedMarking the channel S to be judged if the carrier-to-noise ratio is accordantfProperty A ofRIf the legal record of the same frequency channel in the database does not match within the error range, marking the channel S to be judgedfProperty A ofR=0。
Wherein the monitored channel power characteristic FPAttribute converted into power characteristic according to equipment identification precision and allowable error rangeOptionally, if the power error range is set to ± 5% of the legal channel power, the specific conversion method is to set the power error range within the error range and match the legal record of the same frequency channel in the databaseIf the power is matched, marking the channel S to be judgedfProperty (2) ofIf the error range does not match the legal record of the same frequency channel in the database, marking the channel S to be judgedfProperty (2) of
Wherein the monitored polarization mode characteristic F of the channelPMAnd converted into an attribute A of polarization mode characteristicfPMThe specific conversion method is that if and the databaseMiddle common frequency channel SfPMDBIf the polarization is the same, the channel S to be determined is markedfProperty A offPMIf not, marking the channel S to be judged as 1fProperty A offPM=0。
Wherein the monitored speech characteristics FtextConverting the recorded speech into a text in natural language by using a speech recognition technology, optionally converting the recorded speech into the text by using a deep learning technology (DNN acoustic modeling technology + RNNLM language modeling technology), optionally extracting keywords from the converted text by using a natural language processing algorithm such as a TextRank algorithm, wherein the specific conversion method is that if the converted text is in an error range, if the converted text optionally contains 5 keywords, the converted text is regarded as consistent with the same frequency channel SfTextDBIf the keyword (S) matches, the channel S to be determined is markedfProperty (2) ofIf the error range does not match the legal record in the database, marking the channel S to be judgedfProperty (2) of
S4, calculating the distance between the channel signal to be determined and the legal channel signal in the database according to the distance measurement formula, wherein the distance measurement formula should include the euclidean distance but is not limited to the euclidean distance.
The distance between the channel to be determined and the legal channel in the attribute space can be specifically expressed as:
wherein A isfDBRepresenting a channel S to be determinedfThe frequency attribute database of (a) records a value,representing a channel S to be determinedfThe bandwidth attribute database of (2) records a value,representing a channel S to be determinedfIs recorded in the carrier-to-noise ratio attribute database, AfDBRepresenting a channel S to be determinedfIs recorded in the power attribute database, AfPMDBPolarization mode attribute database entry, A, representing a signal to be determinedfTextDBA speech attribute database record value representing the signal to be determined.
And judging the legality of the channel to be judged according to the distance measurement result in the attribute space, if the distance measurement result is smaller than a distance threshold, the signal is a legal signal, otherwise, if the measured distance legal signal exceeds the distance threshold, alarming and obtaining evidence for manual checking.
S5, the legal distance threshold is DflegalDBObtaining storage, preferably D, from actual monitoringflegalDBCan be formulated according to the equipment precision and the environment average level in the actual monitoring, can properly improve the robustness of the method, if D isf<DflegalDBThen the signal is judged to be legitimate, if Df>DflegalDBThen the alarm recording reminds the manual review.
According to the above-mentioned process, completing monitoring frequency channel setAll signals inThe recording of illegal channels in the database can be increased in actual use, the distance between the channel to be judged and the illegal sample channel in the attribute space is compared, if the distance is close to the illegal channel, the channel is the illegal channel, and if the distance is close to the legal channel, the channel is legal.
Compared with the prior art, the invention provides a black broadcast monitoring method based on zero sample learning, which has the following beneficial effects:
1. the invention separates the physical characteristic acquisition and the attribute judgment of the signal by introducing the concept of the attribute, has simple separation judgment process and high efficiency, and simultaneously increases the measurement error and the consideration of the environment actual measurement factor in the conversion from the characteristic to the attribute, thereby achieving the purpose of monitoring automation and increasing the robustness of the monitoring method.
2. According to the invention, the attribute space distance between the monitoring channel and the legal channel is measured by introducing the attribute decomposition of the legal channel in the attribute space, so that the zero sample black broadcasting channel with the distance from the legal channel can be rapidly and accurately found, meanwhile, the fault tolerance of the method to the attribute identification stage can be improved by setting the threshold value of the attribute space distance, the monitoring identification mechanism is further optimized, and the intelligent level of the monitoring technology is improved.
3. The invention can extract the voice attribute in the signal at a specific moment by introducing the voice recognition technology, effectively discover the physical characteristic disguised as the black broadcast of a legal channel, and solve a technical difficulty in the monitoring of the black broadcast.
It is to be understood that both the foregoing general description and the following detailed description are exemplary and explanatory only and are not restrictive of the invention, as claimed.
Drawings
Fig. 1 is a schematic flow chart of a black broadcast monitoring method based on zero sample learning according to the present invention;
fig. 2 is a schematic diagram of a method for converting a channel feature space into an attribute space of a black broadcast monitoring method based on zero sample learning according to the present invention;
fig. 3 is a flow chart illustrating a conversion process from a channel feature space to an attribute space of a black broadcast monitoring method based on zero sample learning according to the present invention;
fig. 4 is a schematic diagram of determining an attribute spatial distance between an unknown channel and a legal channel according to a distance metric function in the black broadcast monitoring method based on zero sample learning according to the present invention.
Detailed Description
The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the drawings in the embodiments of the present invention, and it is obvious that the described embodiments are only a part of the embodiments of the present invention, and not all of the embodiments.
Examples of which are illustrated in the accompanying drawings, wherein like reference numerals refer to the same or similar elements or elements having the same or similar function throughout. The embodiments described below with reference to the drawings are illustrative and intended to be illustrative of the invention and are not to be construed as limiting the invention.
In the description of the present invention, it is to be understood that the terms "central," "longitudinal," "lateral," "length," "width," "thickness," "upper," "lower," "front," "rear," "left," "right," "vertical," "horizontal," "top," "bottom," "inner," "outer," "clockwise," "counterclockwise," "axial," "radial," "circumferential," and the like are used in the orientations and positional relationships indicated in the drawings for convenience in describing the invention and to simplify the description, and are not intended to indicate or imply that the referenced devices or elements must have a particular orientation, be constructed and operated in a particular orientation, and are therefore not to be considered limiting of the invention.
In the present invention, unless otherwise expressly stated or limited, the terms "mounted," "connected," "secured," and the like are to be construed broadly and can, for example, be fixedly connected, detachably connected, or integrally formed; can be mechanically or electrically connected; either directly or indirectly through intervening media, either internally or in any other relationship. The specific meanings of the above terms in the present invention can be understood by those skilled in the art according to specific situations.
As shown in fig. 1-4, the present invention provides a black broadcast monitoring method based on zero sample learning:
referring to fig. 1, which is a schematic flow chart of a black broadcast monitoring method based on zero sample learning according to the present invention, it can be seen from fig. 1 that the broadcast content monitoring method provided by the present invention includes the following steps:
s1, using radio monitoring equipment to monitor a specific location at a specific time tThe broadcast signal is subjected to full-band spectrum scanning to generate a corresponding channel setThereinRepresenting a particular signal.
In the technical scheme of the invention, the broadcast content monitoring method can be applied to a common black broadcast monitoring system, the common black broadcast monitoring system is provided with necessary antennas and processing devices, has the capabilities of discovering, identifying and measuring the performance of signals of the place, and meanwhile, the common black broadcast monitoring system is provided with common data storage and calculation capabilities.
In the technical scheme provided by the invention, all channel sets needing to be detected and compared can be determined after the full-band spectrum scanning is carried out on the monitored receiving signals at a specific time tThereinRepresenting a particular channel signal. In the step, a frequency list of the location time t is found and determined to serve as a comparison basis of subsequent signals one by one.
In the invention, the characteristics of all legal channel signals at the moment t of the monitoring place can be determined by manually monitoring the signals at the place at the moment t, the attributes of all signals in a legal channel set are automatically obtained by using a characteristic-attribute conversion method, and the characteristics and the attributes of the legal channels are stored. The manual listening to obtain the characteristics and attributes of the legal channels belongs to the initialization process of the method. This process is not necessary if the legal channel characteristics and attributes of the monitoring location at time t are known conditions.
S2: for each channel signal in the channel setObtaining its frequency characteristic FfBandwidth characteristic FWCarrier to noise ratio characteristic FCPower characteristic FPPolarization mode characteristic FPMVoice attribute feature Ftext。
In the present invention, the channel set is traversedEach signal ofAnd obtaining its frequency characteristic FfBandwidth characteristic FWCarrier to noise ratio characteristic FCPower characteristic FPPolarization mode characteristic FPMVoice attribute feature FtextAs a feature space for channel signals in a channel set. In practical applications, the use of the above-described basic features as feature representations in the feature space may not be limited.
In the present invention, the frequency characteristic FfLegitimacy is a necessary prerequisite for the normal legitimacy of other features. Bandwidth characteristic FWCarrier to noise ratio characteristic FCPower characteristic FPPolarization mode characteristic FPMDescribing the detected channel signal from a different physical characteristic point of view, and the voice attribute characteristic FtextThe channel signal to be decided is described from the content.
In the present invention, the frequency characteristic FfBandwidth characteristic FWCarrier to noise ratio characteristic FCPower characteristic FPPolarization mode characteristic FPMThe voice attribute feature F can be quickly calculated by hardware equipment and a calculation methodtextThe identified channel signal needs to be recorded and sampled at time t, and in practice, the sampling time may be set to 60 s.
S3: and comparing whether the same-frequency channel exists in the database, if so, comparing other characteristics of the same-frequency legal channel in the database, and judging whether the channel signal characteristics have legal attributes according to the definition and the error range of the corresponding characteristics.
Referring to fig. 2, a schematic diagram of a method for converting a channel signal feature space into an attribute space according to a black broadcast monitoring method based on zero sample learning of the present invention, and referring to fig. 3, a flowchart for converting a channel feature space into an attribute space of the present invention, specifically, the conversion from a feature to an attribute, includes the following steps:
s301, in actual measurement, frequency characteristic F of channel to be determinedfProperty A converted into frequency characteristics according to equipment precision, environmental influence and allowable error rangefIn the actual measurement, the frequency error range can be set to be +/-5% of the legal channel bandwidth in consideration of the detection precision, the environmental influence and the measurement error of the equipment, the error can be further optimized and adjusted in combination with the actual measurement, and if the frequency of the monitored channel signal is within the error range, the frequency of the monitored channel signal and a legal record F in a database are set to be within the error rangefDBComparing one by one, if the identified signal channel frequency exists in the database, marking the attribute A of the frequency f channelfWhen the signal to be determined has legal frequency attribute, if the frequency signal is not recorded in the database, the basic characteristic frequency of the channel is illegal, the channel is determined as an illegal channel, namely black broadcast, and all other attributes A are markediAnd (5) omitting the conversion step of other characteristics to attributes, and directly marking the channel as an illegal channel.
S302, in actual measurement, bandwidth characteristics F of the channel to be judgedWAttribute converted into bandwidth characterization according to device identification accuracy, environmental impact and allowable error rangeIn the actual measurement, the bandwidth error range can be set to be +/-5% of the legal channel bandwidth in consideration of the detection precision, the environmental influence and the measurement error of the equipment, the error can be further optimized and adjusted in combination with the reality, and the specific conversion method is that if the bandwidth error range is within, the bandwidth F is recorded with the same-frequency channel in the databaseWDBIf yes, marking the property of the frequency f channelI.e. the channel to be determined has legal frequencyAttribute, if the bandwidth of the same frequency channel is not consistent with the bandwidth of the same frequency channel in the database within the error range, marking the attribute of the channelI.e. the channel to be determined does not have legal frequency properties.
S303, carrying out actual measurement on carrier-to-noise ratio characteristic F of channel to be determinedSAttribute converted into carrier-to-noise ratio characteristic according to equipment identification precision and allowable error rangeIn the actual measurement, considering the detection precision, environmental influence and measurement error of the equipment, the carrier-to-noise ratio error range can be set to be +/-5% of the carrier-to-noise ratio of a legal channel, the actual optimization and adjustment can be further carried out, and the specific conversion method is that if the carrier-to-noise ratio is within the error range, the carrier-to-noise ratio F is recorded with the same frequency channel in the databaseSDBIf yes, marking the attribute A of the frequency f channelRIf the channel to be judged has legal carrier-to-noise ratio attribute and the carrier-to-noise ratio of the same frequency channel in the database are not consistent within the error range, marking the attribute A of the channelRAnd when the channel to be judged is 0, the channel to be judged does not have the legal carrier-to-noise ratio attribute.
S304, in actual measurement, the power characteristic F of the channel to be determinedPAttribute converted into power characteristic according to equipment identification precision and allowable error rangeIn the actual measurement, the detection precision, the environmental influence and the measurement error of the equipment are considered, the power error range can be set to be +/-5% of the legal channel power, and the error can be further optimized and adjusted in combination with the actual situation. The specific conversion method is that if the error range is within, and the recording power F of the same frequency channel in the databasePDBIf yes, marking the property of the frequency f channelI.e. the channel to be determined has legal power attributeIf the power of the same frequency channel in the error range is inconsistent with the power of the same frequency channel in the database, marking the attribute of the channelI.e. the channel to be determined does not have legal power properties.
S305, in actual measurement, the channel polarization mode characteristic F of the channel to be determinedPMProperty of conversion into polarization mode characteristicThe specific conversion method is the polarization mode F if the same frequency channel is selectedPMDBIf the channel is the same, then the attribute of the channel is markedThat is, the channels to be determined have legal polarization mode attribute, if they are not the same, the attribute of the channel is markedThat is, the channel to be determined does not have the legal polarization mode attribute.
S306, in actual measurement, sound characteristics F of a channel to be judged are obtained by carrying out sound recording sampling on channel content at specific time ttextIn practice, the sampling time can be set to 60S, and the voice feature F of the channel to be judged can be obtainedtextConversion to natural language text using speech recognition techniques, in practice, deep neural network models such as: the DNN acoustic modeling technique + RNNLM language modeling technique converts the recorded speech into text, and optionally, uses a natural language processing algorithm such as: extracting keywords by a TextRank algorithm, and performing F on the keywords of the channel to be judged and the speech characteristic keywords of the legal channel in the database at the same frequency and the momenttextDBComparing, if the keywords are matched, marking the attribute of the channelThat is, the channel to be determined has legal voice attribute, if the keywords can not be matched,then the attributes of the channel are markedI.e. the channel to be determined does not have legal speech properties. In practice, optionally, a keyword blacklist may be added, such as hitting a word direct alarm record in the blacklist for manual checking.
S4, after the features in the channel to be determined are converted into corresponding attributes, referring to the schematic diagram of determining the attribute distance between the unknown channel and the legal channel by the distance function in the invention of fig. 3, the attribute space distance between the unknown channel and the legal channel can be calculated.
In practice, the attribute space distance from the channel to be determined to the legal channel is calculated by the calculating device according to a distance measurement formula, wherein the attribute space distance including the euclidean distance but not limited to the euclidean distance may be selected to measure the attribute distance between the channel to be determined and the legal channel.
The specific calculation method of the euclidean distance may be represented as:
the distance measurement result is compared with a legal distance threshold D stored in a databaseflegalDBI.e. the decision threshold value is compared, if the distance measurement result is less than the distance threshold Df≤DflegalDBIf the measured distance exceeds the threshold value D, the channel is legal, otherwise, if the measured distance exceeds the threshold value Df>DflegalDBAnd alarming and obtaining evidence records for manual checking.
Wherein D isflegalDBThe method can be determined according to the equipment precision and the average level of environmental interference in actual monitoring, the selection of the threshold value can be determined according to actual conditions, the result floating range caused by the equipment precision, the measurement error and the environmental factors can be obtained by calculating the spatial distance between the legal channel and the attribute space of the threshold value at different moments for multiple times, in practice, the robustness and the sensitivity of the method can be adjusted by flexibly adjusting the threshold distance, and if the threshold distance is set to be smallNamely, the requirement on the compliance of the channel in the attribute space is strict, and the false alarm rate of the method is increased and the robustness is reduced. If the threshold distance is set to be large, the requirement on the compliance of the channel in the attribute space is loose, the robustness of the method is high, the false alarm rate is reduced, and the method can be adjusted according to the requirement in practice.
According to the above-mentioned process, completing monitoring frequency channel setAll signals inThe polling judgment is carried out, in practice, a program can be set to complete 24-hour uninterrupted monitoring or set for sampling inspection at a specific moment, and the high-efficiency, accurate and intelligent identification of black broadcasting is achieved while manpower is saved automatically.
Although embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that changes, modifications, substitutions and alterations can be made in these embodiments without departing from the principles and spirit of the invention, the scope of which is defined in the appended claims and their equivalents.
It should be noted that the embodiments and features of the embodiments may be combined with each other without conflict.
The technical solution of the present invention will be clearly and completely described below with reference to the accompanying drawings. It is to be understood that the described embodiments are merely exemplary of the invention, and not restrictive of the full scope of the invention. The following description of at least one exemplary embodiment is merely illustrative in nature and is in no way intended to limit the invention, its application, or uses. All other embodiments, which can be derived by a person skilled in the art from the embodiments given herein without making any creative effort, shall fall within the protection scope of the present invention.
Claims (10)
1. A black broadcast monitoring method based on zero sample learning is characterized in that: the method comprises the following steps:
s1, real-time spectrum scanning is carried out on the broadcast signals received by the monitoring point equipment at the specific time t to generate a channel set
S2, for each channel signal in the channel setAcquiring frequency spectrum characteristics, bandwidth characteristics, carrier-to-noise ratio characteristics, power characteristics, polarization mode characteristics and voice attribute characteristics of the voice signal;
s3, comparing the characteristics of the signal to be judged one by one with the corresponding characteristic records in the database, judging whether the signal has legal attributes or not according to the definition and the error range of the corresponding characteristics, and analyzing the corresponding characteristics after the judgment is finished;
s4, calculating the attribute decomposition distance between the channel signal and the legal signal in the database according to the distance measurement formula;
and S5, according to the distance measurement result, if the distance is smaller than the distance threshold, the signal is a legal signal, otherwise, if the distance exceeds the distance threshold, the alarm is given and evidence is taken for manual check.
2. The black broadcasting monitoring method based on zero sample learning as claimed in claim 1, wherein: in the step 1, the monitoring device performs real-time spectrum scanning on the received broadcast signal at a specific time t and at a monitoring position existing in the database, where the time t is required to be consistent with a time of sampling and recording a legal channel signal corresponding to the region in the database, and the characteristics of each channel signal s obtained in the step 2 include, but are not limited to, a spectrum characteristic, a bandwidth characteristic, a carrier-to-noise ratio characteristic, a power characteristic, a polarization mode characteristic, and a voice attribute characteristic.
3. The black broadcasting monitoring method based on zero sample learning as claimed in claim 1, wherein: in said step 3The corresponding characteristics include frequency characteristics, and when comparing, the attribute A of the channel signal is firstly determinedfIf it is legal, its legal attribute is determined by comparing the channel signal of said frequency with the database, if it is, the frequency signal SfDBIf the database does not have the signal of the frequency, all the attributes A are recorded as' 1fiAnd is noted as "0".
4. The black broadcasting monitoring method based on zero sample learning as claimed in claim 1, wherein: the corresponding characteristics in step 3 include bandwidth characteristics, legal attributes thereofThe determination method is to compare the bandwidth of the frequency signal in the databaseIf the bandwidths are consistent, the bandwidth attributeThe attribute is marked as "1" if the bandwidths are not consistent, and the attribute is marked as "0".
5. The black broadcasting monitoring method based on zero sample learning as claimed in claim 1, wherein: the corresponding characteristics in the step 3 comprise carrier-to-noise ratio characteristics and legal attributes A thereoffRThe determination method is to compare the carrier-to-noise ratio of the frequency signal in the databaseThe carrier-to-noise ratio attribute is described as "1" if the carrier-to-noise ratios are identical, and the attribute is described as "0" if the carrier-to-noise ratios are not identical.
6. The black broadcasting monitoring method based on zero sample learning as claimed in claim 1, wherein: the corresponding characteristics in step 3 include power characteristics, which are legalProperty of (2)The determination method is to compare the power of the frequency signal in the databaseThe power attribute is described as "1" if the powers match, and the attribute is described as "0" if the powers do not match.
7. The black broadcasting monitoring method based on zero sample learning as claimed in claim 1, wherein: the corresponding characteristics in the step 3 comprise polarization mode characteristics and legal attributes A thereoffPMThe determination method is to compare the polarization mode S of the frequency signal in the databasefPMDBIf the polarization modes are identical, the polarization mode attribute is set to "1", and if the polarization modes are not identical, the attribute is set to "0".
8. The black broadcasting monitoring method based on zero sample learning as claimed in claim 1, wherein: the corresponding features in the step 3 comprise voice attribute features, the voice features are obtained by intercepting a certain time period T (such as 1 minute) at the moment T, and legal attributes A of the voice featuresfTextThe determining method comprises extracting text corresponding to audio data by voice recognition technology, extracting keywords in text by natural language processing method such as TextRank method, and comparing voice attribute S of the frequency signal at the same time t in databasefTextDBIf there are more than a specific threshold (e.g., 5) of the same keyword, the attribute is described as "1", and if there are no same keywords, the attribute is described as "0".
9. The black broadcasting monitoring method based on zero sample learning as claimed in claim 1, wherein: the distance metric formula in step 4 includes but is not limited to euclidean distances,
wherein A isfDBA frequency attribute database record value representing the signal to be determined,a bandwidth attribute database record value representing the signal to be determined,a carrier-to-noise ratio attribute database record value representing the signal to be determined,record value of a power attribute database representing a signal to be determined, AfPMDBPolarization mode attribute database entry, A, representing a signal to be determinedfTextDBA voice attribute database record value representing the signal to be determined.
10. The black broadcasting monitoring method based on zero sample learning as claimed in claim 1, wherein: the legal distance threshold in the step 5 is DflegalDB,DflegalDBCan be adjusted and determined according to the error of the legal signal tested in actual condition, if Df<DflegalDBThen the signal of the channel is judged to be legal, if Df>DflegalDBThen the alarm recording reminds the manual review.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202110943763.XA CN113572555B (en) | 2021-08-17 | 2021-08-17 | Black broadcast monitoring method based on zero sample learning |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202110943763.XA CN113572555B (en) | 2021-08-17 | 2021-08-17 | Black broadcast monitoring method based on zero sample learning |
Publications (2)
Publication Number | Publication Date |
---|---|
CN113572555A true CN113572555A (en) | 2021-10-29 |
CN113572555B CN113572555B (en) | 2023-11-17 |
Family
ID=78171904
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN202110943763.XA Active CN113572555B (en) | 2021-08-17 | 2021-08-17 | Black broadcast monitoring method based on zero sample learning |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN113572555B (en) |
Cited By (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN115842597A (en) * | 2023-02-09 | 2023-03-24 | 成都华乾科技有限公司 | Black broadcast detection method and system based on multi-source data fusion |
Citations (7)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US20120021710A1 (en) * | 2010-07-26 | 2012-01-26 | Kenta Tsukamoto | Electromagnetic wave discrimination device, electromagnetic wave discrimination method, and electromagnetic wave discrimination program |
CN103929260A (en) * | 2014-03-12 | 2014-07-16 | 上海风格信息技术股份有限公司 | Analysis method for abnormal radio frequency points |
CN107276707A (en) * | 2017-06-08 | 2017-10-20 | 国家无线电监测中心 | A kind of black automation of broadcast continuity analysis method and Weigh sensor device based on multi-attribute analysis |
CN110321930A (en) * | 2019-06-05 | 2019-10-11 | 大连理工大学 | A kind of illegal broadcast singal classification method based on Recognition with Recurrent Neural Network |
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 |
CN110995374A (en) * | 2019-03-29 | 2020-04-10 | 国家无线电监测中心检测中心 | Black broadcast signal monitoring system |
CN111461025A (en) * | 2020-04-02 | 2020-07-28 | 同济大学 | Signal identification method for self-evolving zero-sample learning |
-
2021
- 2021-08-17 CN CN202110943763.XA patent/CN113572555B/en active Active
Patent Citations (7)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US20120021710A1 (en) * | 2010-07-26 | 2012-01-26 | Kenta Tsukamoto | Electromagnetic wave discrimination device, electromagnetic wave discrimination method, and electromagnetic wave discrimination program |
CN103929260A (en) * | 2014-03-12 | 2014-07-16 | 上海风格信息技术股份有限公司 | Analysis method for abnormal radio frequency points |
CN107276707A (en) * | 2017-06-08 | 2017-10-20 | 国家无线电监测中心 | A kind of black automation of broadcast continuity analysis method and Weigh sensor device based on multi-attribute analysis |
CN110995374A (en) * | 2019-03-29 | 2020-04-10 | 国家无线电监测中心检测中心 | Black broadcast signal monitoring system |
CN110321930A (en) * | 2019-06-05 | 2019-10-11 | 大连理工大学 | A kind of illegal broadcast singal classification method based on Recognition with Recurrent Neural Network |
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 |
CN111461025A (en) * | 2020-04-02 | 2020-07-28 | 同济大学 | Signal identification method for self-evolving zero-sample learning |
Non-Patent Citations (2)
Title |
---|
林秋海,龙 华: "基于贝叶斯判别的非法调频广播电台的识别", 《通信技术》 * |
浙江省宁波市无线电监测站等: "基于机器学习的无线电监测方法初探", 《CHINA RADIO》 * |
Cited By (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN115842597A (en) * | 2023-02-09 | 2023-03-24 | 成都华乾科技有限公司 | Black broadcast detection method and system based on multi-source data fusion |
CN115842597B (en) * | 2023-02-09 | 2023-05-02 | 成都华乾科技有限公司 | Black broadcast detection method and system based on multi-source data fusion |
Also Published As
Publication number | Publication date |
---|---|
CN113572555B (en) | 2023-11-17 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN107276707B (en) | Black broadcast automatic analysis method based on multi-attribute analysis and intelligent identification device | |
CN107528832B (en) | Baseline construction and unknown abnormal behavior detection method for system logs | |
US9930541B2 (en) | Maintaining the integrity of configuration information of a network of access points for use in positioning an apparatus | |
Liu et al. | A sound monitoring system for prevention of underground pipeline damage caused by construction | |
CN105679313A (en) | Audio recognition alarm system and method | |
CN108198561A (en) | A kind of pirate recordings speech detection method based on convolutional neural networks | |
CN116308958A (en) | Carbon emission online detection and early warning system and method based on mobile terminal | |
CN108831506A (en) | Digital audio based on GMM-BIC distorts point detecting method and system | |
CN113572555B (en) | Black broadcast monitoring method based on zero sample learning | |
Liguori et al. | Towards the evaluation of the measurement uncertainty of environmental acoustic noise | |
CN112532941A (en) | Vehicle source intensity monitoring method and device, electronic equipment and storage medium | |
CN103871405A (en) | AMR audio authenticating method | |
CN109995450B (en) | Cloud technology based voice recognition and intelligent black broadcast identification method | |
CN112637834B (en) | Fingerprint fusion identification method and device for wireless communication equipment | |
CN103426439B (en) | A kind of broadcast television audio signal content consistency detecting method | |
CN116318457B (en) | Radio signal monitoring method and system | |
CN111879456B (en) | Building curtain wall safety detection method and system | |
CN113191224A (en) | Unmanned aerial vehicle signal extraction and identification method and system | |
CN104200815B (en) | A kind of audio-frequency noise real-time detection method based on correlation analysis | |
CN112566117B (en) | Vehicle node identity recognition method and device based on metric learning | |
CN108766465A (en) | A kind of digital audio based on ENF universal background models distorts blind checking method | |
Rouniyar et al. | Channel response based multi-feature audio splicing forgery detection and localization | |
CN117150265B (en) | Robust radio frequency signal open set identification method under low signal-to-noise ratio condition | |
CN109960929B (en) | Regression model-based zero sample intrusion detection method | |
CN117275512B (en) | Switch cabinet state online monitoring method and system based on voiceprint recognition |
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 | ||
GR01 | Patent grant | ||
GR01 | Patent grant |