CN113098640A - Frequency spectrum anomaly detection method based on channel occupancy prediction - Google Patents
Frequency spectrum anomaly detection method based on channel occupancy prediction Download PDFInfo
- Publication number
- CN113098640A CN113098640A CN202110323351.6A CN202110323351A CN113098640A CN 113098640 A CN113098640 A CN 113098640A CN 202110323351 A CN202110323351 A CN 202110323351A CN 113098640 A CN113098640 A CN 113098640A
- Authority
- CN
- China
- Prior art keywords
- sequence
- channel occupancy
- channel
- occupancy rate
- time period
- 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
- 238000001228 spectrum Methods 0.000 title claims abstract description 52
- 238000001514 detection method Methods 0.000 title claims abstract description 26
- 238000000034 method Methods 0.000 claims abstract description 37
- 238000003062 neural network model Methods 0.000 claims abstract description 17
- 238000012549 training Methods 0.000 claims abstract description 14
- 230000007787 long-term memory Effects 0.000 claims abstract description 7
- 230000006403 short-term memory Effects 0.000 claims abstract description 7
- 238000004364 calculation method Methods 0.000 claims description 25
- 230000005856 abnormality Effects 0.000 claims description 11
- 238000012935 Averaging Methods 0.000 claims description 6
- 229920006395 saturated elastomer Polymers 0.000 claims description 4
- 230000002159 abnormal effect Effects 0.000 abstract description 13
- 241000282414 Homo sapiens Species 0.000 description 3
- 238000004891 communication Methods 0.000 description 3
- 238000004458 analytical method Methods 0.000 description 2
- 238000013528 artificial neural network Methods 0.000 description 2
- 230000009286 beneficial effect Effects 0.000 description 2
- 238000010586 diagram Methods 0.000 description 2
- 238000005516 engineering process Methods 0.000 description 2
- 238000002474 experimental method Methods 0.000 description 2
- 238000005259 measurement Methods 0.000 description 2
- 238000012544 monitoring process Methods 0.000 description 2
- 238000006243 chemical reaction Methods 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 230000009545 invasion Effects 0.000 description 1
- 238000010801 machine learning Methods 0.000 description 1
- 230000015654 memory Effects 0.000 description 1
- 238000005070 sampling Methods 0.000 description 1
- 239000004576 sand Substances 0.000 description 1
- 238000011895 specific detection Methods 0.000 description 1
- 238000010183 spectrum analysis Methods 0.000 description 1
- 238000012706 support-vector machine Methods 0.000 description 1
Images
Classifications
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04B—TRANSMISSION
- H04B17/00—Monitoring; Testing
- H04B17/30—Monitoring; Testing of propagation channels
- H04B17/373—Predicting channel quality or other radio frequency [RF] parameters
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04B—TRANSMISSION
- H04B17/00—Monitoring; Testing
- H04B17/30—Monitoring; Testing of propagation channels
- H04B17/309—Measuring or estimating channel quality parameters
- H04B17/336—Signal-to-interference ratio [SIR] or carrier-to-interference ratio [CIR]
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04B—TRANSMISSION
- H04B17/00—Monitoring; Testing
- H04B17/30—Monitoring; Testing of propagation channels
- H04B17/391—Modelling the propagation channel
- H04B17/3913—Predictive models, e.g. based on neural network models
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- Signal Processing (AREA)
- Computer Networks & Wireless Communication (AREA)
- Electromagnetism (AREA)
- Evolutionary Computation (AREA)
- Artificial Intelligence (AREA)
- Quality & Reliability (AREA)
- Theoretical Computer Science (AREA)
- General Health & Medical Sciences (AREA)
- Software Systems (AREA)
- Molecular Biology (AREA)
- Computing Systems (AREA)
- General Engineering & Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Mathematical Physics (AREA)
- Data Mining & Analysis (AREA)
- Computational Linguistics (AREA)
- Biophysics (AREA)
- Biomedical Technology (AREA)
- Life Sciences & Earth Sciences (AREA)
- Health & Medical Sciences (AREA)
- Monitoring And Testing Of Transmission In General (AREA)
Abstract
The invention discloses a frequency spectrum anomaly detection method based on channel occupancy prediction, which comprises the following steps of: s1: acquiring frequency band scanning data in a historical time period to generate a first CSI sequence; s2: equally dividing the first CSI sequence to obtain a first channel occupancy rate sequence; s3: training a neural network model; s4: inputting the first channel occupancy rate sequence into a neural network model for prediction to obtain a second channel occupancy rate sequence and a third channel occupancy rate sequence; s5: acquiring frequency band scanning data in the current time period to generate a second CSI sequence; s6: equally dividing the second CSI sequence to obtain a fourth channel occupancy rate sequence; s7: calculating the percentage of deviation of the first and second channel occupancy sequences; s8: calculating the deviation percentage of the occupancy rate sequences of the third and fourth channels; s9: the percent deviation is compared. The method uses a long-term and short-term memory network to learn the law of the occupancy rate of the channel under the normal working condition, and classifies data which do not conform to the law as abnormal.
Description
Technical Field
The invention belongs to the technical field of spectrum detection, and particularly relates to a spectrum anomaly detection method based on channel occupancy prediction.
Background
With the rapid development of electromagnetic technology and wireless communication technology, the form of radio signals shows a diversified trend, the demand of human beings on radio spectrum resources is more and more intense, the radio spectrum is not inexhaustible, and the contradiction between the increasing demand and the limited spectrum resources increases the difficulty for the supervision of the electromagnetic spectrum and the safety guarantee of the electromagnetic space. In recent years, personal use conditions of amateur radio stations, unmanned aerial vehicles and wireless communication equipment are more and more common, and due to the lack of knowledge on electromagnetic space safety, cases of illegal invasion to other wireless communication frequency bands occur, even civil aviation radio receives interference, and safety accidents occur.
The existing frequency spectrum anomaly detection methods are mainly divided into two categories: one method is to determine whether the spectrum state is abnormal by analyzing the change of the spectrum characteristic parameters by using a spectrum analysis method. The other method is to use a supervised machine learning algorithm to perform two classifications to judge whether the frequency spectrum is abnormal or not, such as: support vector machines, naive bayes classification, etc.
On one hand, in an actual radio propagation frequency band, radio signals are in a normal working state in most of time, the probability of occurrence of abnormity is relatively low, and due to the fact that a radio system is complex, frequency spectrum signals at a detection end of the system are abnormal due to various reasons such as internal faults of the system, external interference signals and the like, the sample acquisition difficulty is high, the supervised detection method is difficult to fully master experience knowledge, and therefore the detection precision is influenced; on the other hand, the algorithm is based on specific tasks and scenes, and the type of the abnormal signal is defined by human standards, so that the algorithm has limitations.
Disclosure of Invention
The invention aims to solve the problem of large limitation of the existing spectrum detection, and provides a spectrum abnormity detection method based on channel occupancy rate prediction.
The technical scheme of the invention is as follows: a frequency spectrum abnormity detection method based on channel occupancy prediction comprises the following steps:
s1: acquiring frequency band scanning data in a historical time period and generating a first CSI sequence;
s2: averaging the first CSI sequences of each day to obtain a first channel occupancy rate sequence;
s3: training a neural network model by using a long-term and short-term memory network until the neural network model is saturated and stored;
s4: inputting the first channel occupancy rate sequence into a stored neural network model for prediction to obtain a second channel occupancy rate sequence and a third channel occupancy rate sequence;
s5: acquiring frequency band scanning data in the current time period and generating a second CSI sequence;
s6: averaging the second CSI sequences of each day to obtain a fourth channel occupancy rate sequence;
s7: calculating the deviation percentage of the first channel occupancy rate sequence and the second channel occupancy rate sequence;
s8: calculating the deviation percentage of the occupancy rate sequence of the third channel and the occupancy rate sequence of the fourth channel;
s9: and comparing whether the deviation percentage of the first channel occupancy rate sequence and the second channel occupancy rate sequence is smaller than the deviation percentage of the third channel occupancy rate sequence and the fourth channel occupancy rate sequence, if so, determining that the abnormality occurs in the current time period, otherwise, determining that the abnormality does not occur.
Further, the specific method of step S1 is: acquiring frequency band scanning data in a historical time period, taking out frequency spectrum data with the channel center frequency of XHz, and generating a first CSI sequence through a threshold method, wherein channel state information in the first CSI sequenceCSI1The calculation formula of (2) is as follows:
wherein 0 represents that the channel is in an idle state, 1 represents that the channel is in an occupied state, e1Representing measured level values in a historical period of time, E1Representing the noise threshold over the historical period of time and X representing the set frequency.
Further, the specific method of step S2 is: the first CSI sequence of each day is divided into M1 segments, and the calculation formula is as follows:
M1=W1/m1
wherein, W1 represents the total number of time slots of the frequency band scanning data in the historical time period, and m1 represents the number of time slots contained in each sequence in the historical time period;
the calculation formula of the first channel occupancy sequence S is:
wherein,the working time of the frequency point in the historical time period is shown,representing the observed time within the historical time period.
Further, step S4 includes the following sub-steps:
s41: equally dividing the first channel occupancy rate sequence S into Y subsequences with the length of time _ step1, and predicting frequency band scanning data in a historical time period by using a stored neural network model to obtain a second channel occupancy rate sequence Q;
s42: and taking out the last time-step 1-length subsequence in the first channel occupancy rate sequence S, and predicting the time-step 2-length data in the current time period by using the stored neural network model to obtain a third channel occupancy rate sequence T.
Further, the specific method of step S5 is: acquiring frequency band scanning data in the current time period, taking out frequency spectrum data with the channel center frequency of XHz, and generating a second CSI sequence through a threshold method, wherein the CSI sequence comprises Channel State Information (CSI)2The calculation formula of (2) is as follows:
wherein 0 represents that the channel is in an idle state, 1 represents that the channel is in an occupied state, e2Representing measured level values in the current time period, E2Representing the noise threshold for the current time period and X representing the set frequency.
Further, the specific method of step S6 is: the second CSI sequences of each day are equally divided into M2 segments, and the calculation formula is as follows:
M2=W2/m2
wherein, W2 represents the total number of time slots of the frequency band scanning data in the current time period, and m2 represents the number of time slots contained in each sequence of the current time period;
the calculation formula of the fourth channel occupancy rate sequence R is:
wherein,the working time of the frequency point in the current time period is shown,representing the observed time within the current time period.
Further, in step S7, the first channel occupancy sequence and the second channel occupancy sequence have a deviation percentage P1The calculation formula of (2) is as follows:
wherein s isiElement representing a first sequence S of occupancy rates of channels, qiElements, n, representing a second channel occupancy sequence Q1Representing the total number of corresponding channel occupancy sequence elements in the first channel occupancy sequence and the second channel occupancy sequence;
in step S8, the percentage of deviation P between the third channel occupancy sequence and the fourth channel occupancy sequence2The calculation formula of (2) is as follows:
wherein, tiElements, r, representing a third frequency channel occupancy sequence TiElement, n, representing a fourth channel occupancy sequence R2Representing the total number of corresponding channel occupancy sequence elements in the third channel occupancy sequence and the fourth channel occupancy sequence.
The invention has the beneficial effects that:
(1) the spectrum anomaly detection method of the invention is an obvious unsupervised detection scheme. In the process of establishing the detection model, data with abnormal frequency spectrum are not used, and the method mainly uses a long-term and short-term memory network to learn the rule of the occupancy rate of the channel under the normal working condition and classifies the data which do not conform to the rule as abnormal data.
(2) The invention simplifies the acquired complex frequency band scanning data into a frequency channel occupancy rate sequence by utilizing the concept of frequency channel occupancy rate, can effectively represent the use state of a frequency spectrum, and trains a network model only by learning the rule of the frequency channel occupancy rate sequence.
Drawings
FIG. 1 is a flow chart of a method of spectrum anomaly detection;
FIG. 2 is a diagram of changes in channel occupancy;
FIG. 3 is a graph of maximum, minimum and difference energy levels for a partially noisy channel;
fig. 4 is a frequency channel occupancy sequence chart with a channel center frequency of 943.8MHz in an embodiment of the invention.
Detailed Description
The embodiments of the present invention will be further described with reference to the accompanying drawings.
As shown in fig. 1, the present invention provides a spectrum anomaly detection method based on channel occupancy prediction, which includes the following steps:
s1: acquiring frequency band scanning data in a historical time period and generating a first CSI sequence;
s2: averaging the first CSI sequences of each day to obtain a first channel occupancy rate sequence;
s3: training a neural network model by using a long-term and short-term memory network until the neural network model is saturated and stored;
s4: inputting the first channel occupancy rate sequence into a stored neural network model for prediction to obtain a second channel occupancy rate sequence and a third channel occupancy rate sequence;
s5: acquiring frequency band scanning data in the current time period and generating a second CSI sequence;
s6: averaging the second CSI sequences of each day to obtain a fourth channel occupancy rate sequence;
s7: calculating the deviation percentage of the first channel occupancy rate sequence and the second channel occupancy rate sequence; obtaining the upper limit of allowable deviation of a predicted value of a channel occupancy rate sequence in a historical time period;
s8: calculating the deviation percentage of the occupancy rate sequence of the third channel and the occupancy rate sequence of the fourth channel; the predicted deviation value of the channel occupancy rate sequence of the next time period;
s9: and comparing whether the deviation percentage of the first channel occupancy rate sequence and the second channel occupancy rate sequence is smaller than the deviation percentage of the third channel occupancy rate sequence and the fourth channel occupancy rate sequence, if so, determining that the abnormality occurs in the current time period, otherwise, determining that the abnormality does not occur.
In the embodiment of the present invention, the specific method in step S1 is: acquiring frequency band scanning data in a historical time period, taking out frequency spectrum data with the channel center frequency of XHz, and generating a first CSI sequence, a first C by a threshold methodChannel state information, CSI, in an SI sequence1The calculation formula of (2) is as follows:
wherein 0 represents that the channel is in an idle state, 1 represents that the channel is in an occupied state, e1Representing measured level values in a historical period of time, E1The noise threshold value in the historical time period is represented, X represents the set frequency, and X can be set according to actual conditions or human needs.
The method is mainly characterized in that the detection problem is solved by using a prediction thought, and a channel occupancy rate sequence of frequency sweep data is selected as a prediction characteristic of a signal. In order to analyze the data sequence subsequently, whether the channel is occupied or not is visually expressed according to the level value, and the level data in the channel is converted into binary forms of 0 and 1 through a threshold value method.
As shown in FIG. 3, the maximum, minimum and difference energy level diagrams of the partial noise channel are obtained, and the analysis of the actual data collected by the present invention shows that the minimum energy level of the sweep data channel at 935MHz-960MHz is about-3 dBuV, i.e. the noise threshold E is set to be 5 dBuV.
In the embodiment of the present invention, the specific method in step S2 is: the first CSI sequence of each day is divided into M1 segments, and the calculation formula is as follows:
M1=W1/m1
wherein, W1 represents the total number of time slots of the frequency band scanning data in the historical time period, and m1 represents the number of time slots contained in each sequence in the historical time period;
the calculation formula of the first channel occupancy sequence S is:
wherein,representing the working time of frequency points in historical time periods,Representing the observed time within the historical time period.
The data of the occupied signal channel represents the percentage of the time of a specific frequency band or frequency point in which a signal works within a certain observation time, and the signal using condition of the frequency band or frequency point can be effectively reflected. Spectrum occupancy plays an important role in spectrum management. If the allocated frequency band is found to be frequently in an idle state in the spectrum management record, or the allocated frequency band is frequently in a use state, the spectrum occupancy rate can be reflected to remind a spectrum manager to check, and the use condition of the frequency band can be evaluated according to the value of the spectrum occupancy rate. If sudden abnormality of the spectrum occupancy rate is found, the abnormal condition of the signal operation can be judged according to the abnormality. The frequency spectrum occupancy can be divided into a frequency channel occupancy and a frequency band occupancy. Channel occupancy. The term "time occupancy" refers to the analysis of the spectrum occupancy over a period of time. For a specific detection frequency band, when the signal frequency spectrum is greater than a set level threshold, the signal is judged to be occupied, and the time occupied by the signal in the observation time is called as the working time. If the observation time of a given frequency point is TsAnd then the time occupancy is expressed as the working time T of the frequency pointuWith total observation time TsThe ratio of. The observation time can be defined by monitoring personnel, the sampling interval is smaller, the more the acquired frequency spectrum data is, the larger the occupied storage space is, the closer the measurement result is to a real value, and the more accurate the monitoring result is. Particularly, according to the calculation scheme in the method, the proportion of the state "1" in each section of CSI sequence is counted, and the sequence formed by the statistics is the channel occupancy rate sequence.
The invention mainly observes the change of the channel occupancy rate and can find that the channel occupancy rate is unstable in time. But it can be seen that the occupancy of the partial channels has periodicity, as shown in fig. 2, which also satisfies the inference of the frequency usage law of the users. The frequency spectrum occupancy rate of a period of time in the future is predicted by learning the change rule of the frequency channel occupancy rate of the history period of the regular channel, namely the frequency channel occupancy rate has predictability. Therefore, the channel occupancy rate of the next time period of a certain frequency point can be predicted by using the long-term and short-term memory neural network, and the purpose of detecting the signal abnormality is further realized.
In the embodiment of the present invention, step S4 includes the following sub-steps:
s41: equally dividing the first channel occupancy rate sequence S into Y subsequences with the length of time _ step1, and predicting frequency band scanning data in a historical time period by using a stored neural network model to obtain a second channel occupancy rate sequence Q;
s42: and taking out the last time-step 1-length subsequence in the first channel occupancy rate sequence S, and predicting the time-step 2-length data in the current time period by using the stored neural network model to obtain a third channel occupancy rate sequence T.
In the embodiment of the present invention, the specific method in step S5 is: acquiring frequency band scanning data in the current time period, taking out frequency spectrum data with the channel center frequency of XHz, and generating a second CSI sequence through a threshold method, wherein the CSI sequence comprises Channel State Information (CSI)2The calculation formula of (2) is as follows:
wherein 0 represents that the channel is in an idle state, 1 represents that the channel is in an occupied state, e2Representing measured level values in the current time period, E2Representing the noise threshold for the current time period and X representing the set frequency.
In the embodiment of the present invention, the specific method in step S6 is: the second CSI sequences of each day are equally divided into M2 segments, and the calculation formula is as follows:
M2=W2/m2
wherein, W2 represents the total number of time slots of the frequency band scanning data in the current time period, and m2 represents the number of time slots contained in each sequence of the current time period;
the calculation formula of the fourth channel occupancy rate sequence R is:
wherein,the working time of the frequency point in the current time period is shown,representing the observed time within the current time period.
In the embodiment of the present invention, in step S7, the deviation percentage P of the first channel occupancy rate sequence and the second channel occupancy rate sequence1The calculation formula of (2) is as follows:
wherein s isiElement representing a first sequence S of occupancy rates of channels, qiElements, n, representing a second channel occupancy sequence Q1Representing the total number of corresponding channel occupancy sequence elements in the first channel occupancy sequence and the second channel occupancy sequence;
in step S8, the percentage of deviation P between the third channel occupancy sequence and the fourth channel occupancy sequence2The calculation formula of (2) is as follows:
wherein, tiElements, r, representing a third frequency channel occupancy sequence TiElement, n, representing a fourth channel occupancy sequence R2Representing the total number of corresponding channel occupancy sequence elements in the third channel occupancy sequence and the fourth channel occupancy sequence.
As shown in fig. 4, centered on the channelFrequency sweep data at 943.8MHz is an example. The point of one channel occupancy is calculated by data of 400 (the number m of time slots included in each sequence is 400) time slots, according to the characteristics of actually acquired data, the data of each time slot lasts for about 0.9S, experimental data of three days is preprocessed by data to obtain a channel occupancy sequence length 568 (namely, the channel occupancy sequence S in a historical time slot is 568), 90% of the experimental data is selected as training data, namely, the training set length of the channel occupancy is 511, the experiment uses the data of the last three steps to predict the data of the last two steps, specifically, time _ step1 is 3, time _ step2 is 2, namely, the time sequence S needing training is {1, 2, 3, 4, 5, 6, 7, 8}, and the time sequence is processed as trainX ═ 1, 2, 3, 7, 8}, and the data is processed into a sequence of [1, 2, 3 ] or 3],[2,3,4],[3,4,5],[4,5,6],trainY={[4,5],[5,6],[6,7],[7,8]Then one single sample will get a trainX dimension of [1, 3, 1 ]]The first 1 represents the number of samples, 3 represents the number of time step1, the second 1 represents the dimension, the rainy dimension is [1, 2, 1%]The first 1 represents the number of samples, 2 represents the number of time step2, and the second 1 represents the dimension. The dimensionality of all signal sample data of the training set after being processed by the channel occupancy (batch is the number of samples of single training in training) [ batch, 3, 1]And [ batch, 2, 1]The dimension obtained by converting the signal is then tranix ═ batch, 3, 1]And train y ═ batch, 2, 1]The channel occupancy rate information of each data sample is input into a neural network for training, and [ batch, 3, 1 [ ]]Conversion to [ batch 3, 1 ]]To [ batch/10, 30 ]]Finally to [ batch, 2, 1]And (4) as a prediction result, calculating loss by comparing with train Y, continuing training until the network is saturated (the training times of the experiment are about 1000 times), obtaining a network model and storing the network model. Through training a stored good network model, prediction fitting is carried out on a part of known channel occupancy rate sequences to obtain a predicted relative error P of the channel occupancy rate sequences1Finally, through the data of the known sequence, the occupancy rate of the channel in the next time period (data of two points, the duration is about 720s) is predicted, and the deviation percentage P between the predicted value and the actual value is calculated2By comparison of P1And P2The size of the sample to achieve the detection purpose.
The working principle and the process of the invention are as follows: the invention selects the channel occupancy rate as a prediction quantity, and judges whether the frequency spectrum is abnormal or not by comparing the prediction result with the actual measurement result. The invention aims to solve the problems of selection of prediction characteristics and judgment of abnormity in a scheme of carrying out abnormity detection by using a long-short term memory network (LSTM), does not need to use data in abnormity when a detection model is established, and solves the limitations that most of the existing algorithms are classified based on a supervised method, and the type of an abnormal signal adopts artificial standard definition. The invention can update the established network model in real time. Since spectrum usage or reception conditions have a possibility of changing, the anomaly detection system should have an update capability to adapt to the new changes. If the actually measured current frequency band scanning data is not abnormal, the channel occupancy rate sequence can be added into the channel occupancy rate sequence in the historical time period to form a new sequence, and a new network model is trained and stored.
The invention has the beneficial effects that:
(1) the spectrum anomaly detection method of the invention is an obvious unsupervised detection scheme. In the process of establishing the detection model, data with abnormal frequency spectrum are not used, and the method mainly uses a long-term and short-term memory network to learn the rule of the occupancy rate of the channel under the normal working condition and classifies the data which do not conform to the rule as abnormal data.
(2) The invention simplifies the acquired complex frequency band scanning data into a frequency channel occupancy rate sequence by utilizing the concept of frequency channel occupancy rate, can effectively represent the use state of a frequency spectrum, and trains a network model only by learning the rule of the frequency channel occupancy rate sequence.
It will be appreciated by those of ordinary skill in the art that the embodiments described herein are intended to assist the reader in understanding the principles of the invention and are to be construed as being without limitation to such specifically recited embodiments and examples. Those skilled in the art can make various other specific changes and combinations based on the teachings of the present invention without departing from the spirit of the invention, and these changes and combinations are within the scope of the invention.
Claims (7)
1. A frequency spectrum abnormity detection method based on channel occupancy prediction is characterized by comprising the following steps:
s1: acquiring frequency band scanning data in a historical time period and generating a first CSI sequence;
s2: averaging the first CSI sequences of each day to obtain a first channel occupancy rate sequence;
s3: training a neural network model by using a long-term and short-term memory network until the neural network model is saturated and stored;
s4: inputting the first channel occupancy rate sequence into a stored neural network model for prediction to obtain a second channel occupancy rate sequence and a third channel occupancy rate sequence;
s5: acquiring frequency band scanning data in the current time period and generating a second CSI sequence;
s6: averaging the second CSI sequences of each day to obtain a fourth channel occupancy rate sequence;
s7: calculating the deviation percentage of the first channel occupancy rate sequence and the second channel occupancy rate sequence;
s8: calculating the deviation percentage of the occupancy rate sequence of the third channel and the occupancy rate sequence of the fourth channel;
s9: and comparing whether the deviation percentage of the first channel occupancy rate sequence and the second channel occupancy rate sequence is smaller than the deviation percentage of the third channel occupancy rate sequence and the fourth channel occupancy rate sequence, if so, determining that the abnormality occurs in the current time period, otherwise, determining that the abnormality does not occur.
2. The method for detecting spectrum anomaly based on channel occupancy prediction as claimed in claim 1, wherein the specific method of step S1 is: acquiring frequency band scanning data in a historical time period, taking out frequency spectrum data with the channel center frequency of XHz, and generating a first CSI sequence through a threshold method, wherein CSI in the first CSI sequence is Channel State Information (CSI)1The calculation formula of (2) is as follows:
wherein 0 represents that the channel is in an idle state, 1 represents that the channel is in an occupied state, e1Representing measured level values in a historical period of time, E1Representing the noise threshold over the historical period of time and X representing the set frequency.
3. The method for detecting spectrum anomaly based on channel occupancy prediction as claimed in claim 1, wherein the specific method of step S2 is: the first CSI sequence of each day is divided into M1 segments, and the calculation formula is as follows:
M1=W1/m1
wherein, W1 represents the total number of time slots of the frequency band scanning data in the historical time period, and m1 represents the number of time slots contained in each sequence in the historical time period;
the calculation formula of the first channel occupancy sequence S is:
4. The method for detecting spectrum abnormality based on channel occupancy prediction as claimed in claim 1, wherein said step S4 includes the following sub-steps:
s41: equally dividing the first channel occupancy rate sequence Q into Y subsequences with the length of time _ step1, and predicting frequency band scanning data in a historical time period by using a stored neural network model to obtain a second channel occupancy rate sequence Q;
s42: and taking out the last subsequence with the length of time _ step1 in the first channel occupancy rate sequence S, and predicting data with the length of time _ step2 in the current time period by using the stored neural network model to obtain a third channel occupancy rate sequence T.
5. The method for detecting spectrum anomaly based on channel occupancy prediction as claimed in claim 1, wherein the specific method of step S5 is: acquiring frequency band scanning data in the current time period, taking out frequency spectrum data with the channel center frequency of XHz, and generating a second CSI sequence through a threshold method, wherein the CSI sequence comprises Channel State Information (CSI)2The calculation formula of (2) is as follows:
wherein 0 represents that the channel is in an idle state, 1 represents that the channel is in an occupied state, e2Representing measured level values in the current time period, E2Representing the noise threshold for the current time period and X representing the set frequency.
6. The method for detecting spectrum anomaly based on channel occupancy prediction as claimed in claim 1, wherein the specific method of step S6 is: the second CSI sequences of each day are equally divided into M2 segments, and the calculation formula is as follows:
M2=W2/m2
wherein, W2 represents the total number of time slots of the frequency band scanning data in the current time period, and m2 represents the number of time slots contained in each sequence of the current time period;
the calculation formula of the fourth channel occupancy rate sequence R is:
7. The method for detecting spectrum abnormality based on channel occupancy prediction as claimed in claim 1, wherein in said step S7, the deviation percentage P of the first channel occupancy sequence and the second channel occupancy sequence1The calculation formula of (2) is as follows:
wherein s isiElement representing a first sequence S of occupancy rates of channels, qiElements, n, representing a second channel occupancy sequence Q1Representing the total number of corresponding channel occupancy sequence elements in the first channel occupancy sequence and the second channel occupancy sequence;
in the step S8, the deviation percentage P between the third channel occupancy rate sequence and the fourth channel occupancy rate sequence2The calculation formula of (2) is as follows:
wherein, tiElements, r, representing a third frequency channel occupancy sequence TiElement, n, representing a fourth channel occupancy sequence R2Representing the total number of corresponding channel occupancy sequence elements in the third channel occupancy sequence and the fourth channel occupancy sequence.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202110323351.6A CN113098640B (en) | 2021-03-26 | 2021-03-26 | Frequency spectrum anomaly detection method based on channel occupancy prediction |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202110323351.6A CN113098640B (en) | 2021-03-26 | 2021-03-26 | Frequency spectrum anomaly detection method based on channel occupancy prediction |
Publications (2)
Publication Number | Publication Date |
---|---|
CN113098640A true CN113098640A (en) | 2021-07-09 |
CN113098640B CN113098640B (en) | 2022-03-08 |
Family
ID=76669767
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN202110323351.6A Active CN113098640B (en) | 2021-03-26 | 2021-03-26 | Frequency spectrum anomaly detection method based on channel occupancy prediction |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN113098640B (en) |
Cited By (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN117421644A (en) * | 2023-12-19 | 2024-01-19 | 商飞智能技术有限公司 | Wireless equipment intrusion detection method and device |
Citations (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
WO2018005210A1 (en) * | 2016-06-29 | 2018-01-04 | Microsoft Technology Licensing, Llc | Predictive anomaly detection in communication systems |
CN108089071A (en) * | 2016-11-17 | 2018-05-29 | 上海创远仪器技术股份有限公司 | A kind of ambient threshold interference detection method |
CN110071913A (en) * | 2019-03-26 | 2019-07-30 | 同济大学 | A kind of time series method for detecting abnormality based on unsupervised learning |
CN110138475A (en) * | 2019-05-08 | 2019-08-16 | 南京邮电大学 | A kind of adaptive threshold channel occupation status prediction technique based on LSTM neural network |
-
2021
- 2021-03-26 CN CN202110323351.6A patent/CN113098640B/en active Active
Patent Citations (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
WO2018005210A1 (en) * | 2016-06-29 | 2018-01-04 | Microsoft Technology Licensing, Llc | Predictive anomaly detection in communication systems |
CN108089071A (en) * | 2016-11-17 | 2018-05-29 | 上海创远仪器技术股份有限公司 | A kind of ambient threshold interference detection method |
CN110071913A (en) * | 2019-03-26 | 2019-07-30 | 同济大学 | A kind of time series method for detecting abnormality based on unsupervised learning |
CN110138475A (en) * | 2019-05-08 | 2019-08-16 | 南京邮电大学 | A kind of adaptive threshold channel occupation status prediction technique based on LSTM neural network |
Non-Patent Citations (3)
Title |
---|
SREERAJ RAJENDRAN等: "Unsupervised Wireless Spectrum Anomaly Detection With Interpretable Features", 《IEEE TRANSACTIONS ON COGNITIVE COMMUNICATIONS AND NETWORKING》 * |
WEI HONGHAO等: "Spectrum anomalies autonomous detection in cognitive radio using Hidden Markov Models", 《2015 IEEE ADVANCED INFORMATION TECHNOLOGY, ELECTRONIC AND AUTOMATION CONTROL CONFERENCE (IAEAC)》 * |
王磊等: "基于时间序列分析的频谱异常自主检测和稳健估计方法", 《电子学报》 * |
Cited By (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN117421644A (en) * | 2023-12-19 | 2024-01-19 | 商飞智能技术有限公司 | Wireless equipment intrusion detection method and device |
CN117421644B (en) * | 2023-12-19 | 2024-05-31 | 商飞智能技术有限公司 | Wireless equipment intrusion detection method and device |
Also Published As
Publication number | Publication date |
---|---|
CN113098640B (en) | 2022-03-08 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN110263846B (en) | Fault diagnosis method based on fault data deep mining and learning | |
CN112232447B (en) | Construction method of complete sample set of power equipment state monitoring data | |
US20210048807A1 (en) | Method of modeling multi-mode degradation process and predicting remaining useful life | |
CN105205288B (en) | Satellite based on the Evolution Modes prediction technique of state in orbit for a long time | |
CN112232366B (en) | Electrical equipment fault early warning method and system based on RFID monitoring | |
US20100329320A1 (en) | Noise detection method, noise detection apparatus, simulation method, simulation apparatus, and communication system | |
CN110059845B (en) | Metering device clock error trend prediction method based on time sequence evolution gene model | |
CN116306320B (en) | Method and system for managing production quality of non-vascular lumen guide wire | |
CN114167838B (en) | Multi-scale health assessment and fault prediction method for servo system | |
CN111401553A (en) | Missing data filling method and system based on neural network | |
CN116552306B (en) | Monitoring system and method for direct current pile | |
CN110991668A (en) | Electric vehicle power battery monitoring data analysis method based on association rule | |
Honghao et al. | Spectrum anomalies autonomous detection in cognitive radio using hidden markov models | |
CN116684878B (en) | 5G information transmission data safety monitoring system | |
CN116691418B (en) | Charging method capable of automatically distributing control power | |
CN113098640B (en) | Frequency spectrum anomaly detection method based on channel occupancy prediction | |
CN116559667A (en) | Model training method and device, battery detection method and device, equipment and medium | |
CN117154263A (en) | Lithium battery cascade utilization charging and discharging system and control method | |
CN113610188A (en) | Bow net contact force non-section abnormity identification method and device | |
CN115983477A (en) | Load prediction method based on K-means clustering and convolutional neural network model | |
CN116915582A (en) | Diagnosis and analysis method and device for fault root cause of communication terminal | |
CN117473351B (en) | Power supply information remote transmission system based on Internet of things | |
CN113918433A (en) | Adaptive intelligent network equipment performance index abnormity detection device and method | |
CN117172601A (en) | Non-invasive load monitoring method based on residual total convolution neural network | |
CN116070140B (en) | Power distribution substation safe operation state monitoring system and method |
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 |