CN114611388A - Wireless channel characteristic screening method based on artificial intelligence - Google Patents
Wireless channel characteristic screening method based on artificial intelligence Download PDFInfo
- Publication number
- CN114611388A CN114611388A CN202210206436.0A CN202210206436A CN114611388A CN 114611388 A CN114611388 A CN 114611388A CN 202210206436 A CN202210206436 A CN 202210206436A CN 114611388 A CN114611388 A CN 114611388A
- Authority
- CN
- China
- Prior art keywords
- cell
- feature
- building
- signal
- height
- 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
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F30/00—Computer-aided design [CAD]
- G06F30/20—Design optimisation, verification or simulation
- G06F30/27—Design optimisation, verification or simulation using machine learning, e.g. artificial intelligence, neural networks, support vector machines [SVM] or training a model
-
- 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
- H04B17/3912—Simulation models, e.g. distribution of spectral power density or received signal strength indicator [RSSI] for a given geographic region
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F2119/00—Details relating to the type or aim of the analysis or the optimisation
- G06F2119/02—Reliability analysis or reliability optimisation; Failure analysis, e.g. worst case scenario performance, failure mode and effects analysis [FMEA]
Abstract
The invention relates to a wireless channel characteristic screening method based on artificial intelligence, and belongs to the technical field of wireless channel modeling. The invention carries out the wireless channel model feature screening based on artificial intelligence by summarizing the feature complete set related to the wireless channel model, and the high-efficiency machine learning model depends on the strong correlation between the input variable and the problem target, thereby reducing the complexity of the calculation time on the premise of ensuring the reliability of the wireless channel model.
Description
Technical Field
The invention belongs to the technical field of wireless channel modeling, and particularly relates to a wireless channel characteristic screening method based on artificial intelligence.
Background
The propagation process of radio waves is extremely complex, and factors such as plains, hills, oceans, forests, lakes, the curvature of the earth, atmospheric attenuation, building density and the like on a propagation path all cause the electromagnetic waves to generate complex conditions such as transmission, diffraction, scattering, reflection, refraction and the like. In the wireless channel modeling process, the essence of feature engineering is to convert the original data into parameters capable of best characterizing the target problem and make the dynamic range of each parameter in a relatively stable range, so that proper feature subsets need to be screened to improve the modeling accuracy.
The data set acquisition places are laboratory and external field environments, and specifically comprise engineering parameter data, map data and wireless channel RSRP label data of a plurality of cells. Wherein the training data set contains a plurality of files, each file representing data within a cell. Each row of the file represents related data of a test area with a fixed size in a cell, the row number is uncertain (the row number of the cell with a larger area is more and vice versa according to the size of the cell), the column number is 18 fixed columns, wherein the first 9 columns are engineering parameter data of a site; the middle 8 columns are map data; the last column is RSRP tag data. Table 1 is one of the rows of data, shown as a sample:
TABLE 1 training data examples
Each section below describes the specific meaning of each piece of data.
3.1 engineering parameter data
The engineering parameter data records the engineering parameter information of a station in a certain cell, and the total number of the engineering parameter information is 9. The corresponding meanings of the fields are shown in the table 2:
TABLE 2 field meanings of engineering parameter data
Name of field | Means of | Unit of |
Cell Index | Cell unique identifier | - |
Cell X | Grid position, X coordinate of site to which cell belongs | - |
Cell Y | Grid position, Y-coordinate of site to which cell belongs | - |
Height | Height of cell transmitter relative to ground | m |
Azimuth | Horizontal direction angle of cell transmitter | Deg |
Electrical Downtilt | Vertical electrical downtilt for cell transmitter | Deg |
Mechanical Downtilt | Vertical mechanical down tilt angle of cell transmitter | Deg |
Frequency Band | Cell transmitter center frequency | MHz |
RS Power | Cell transmitter transmit power | dBm |
To facilitate data processing, maps are rasterized. Each grid represents a 5m X5 m area (as shown in fig. 1 below), where (Cell X, Cell Y) records the coordinates of the upper left corner of the grid where the station is located. Other engineering parameters (Height, Azimuth, Electrical downlink, Mechanical downlink) are shown in fig. 1, where the Mechanical Downtilt is achieved by adjusting the bracket behind the antenna panel, which is a physical signal Downtilt; the Electrical Downtilt is realized by adjusting a coil inside the antenna, and is an Electrical Downtilt. The actual signal line downtilt is the sum of the mechanical downtilt and the electrical downtilt.
3.2 map data description
The map data mainly comprises information such as terrain, height and the like of a test place, and is divided into 8 fields of information. The corresponding meanings of the fields are shown in table 3. In consideration of the diversity and complexity of test sites in a map, actual transmission environments in urban areas, industrial areas, rural areas, business areas, etc. are abstracted into numbers. The actual feature type corresponding to the feature type name number can be seen in table 4.
TABLE 3 field meanings of map data
TABLE 4 number meanings of surface feature type names
Like the engineering parameter data, the map data is rasterized, each grid representing a 5m X5 m area, where (X, Y) records the coordinates of the upper left corner of the grid on which the map is located.
Table 4 gives the number meaning of the feature type name, wherein the feature type implies a large amount of height information. Although a large amount of data exists, the data has contents contradictory to the feature type index description, and therefore data cleaning is required. For example, when the landmark type index is 10, the buildings of the grid point still have a building height of less than 60m, for example, the observation point of the coordinate (411170,3395480) of the cell number 2461901 has a building height of 12m, but the landmark type index is 10, which contradicts that the corresponding building height is higher than 60 m. Similarly, when the ground object type index is 13, the building heights of the grid points need to be smaller than 20m, but the buildings larger than 20m still exist according to the table data, so that data cleaning is needed according to abnormal data generated in the process that the ground object type index corresponds to the actual ground object height.
3.3 RSRP tag data
RSRP (Reference Signal Receiving Power) tag data. RSRP is one of the key parameters in a cellular network that may represent the radio signal strength and the physical layer measurement requirements, the average of the received signal power over all res (resource elements) carried by the reference signal. By comparing the measured received power with the known transmitted power, the attenuation of the radio wave signal by the radio wave transmission path can be obtained.
Reference Signal Received Power (RSRP) tag data is used in supervised learning as an actual measurement result for comparison with the results predicted by the machine learning model. The data has 1 field in total, and the corresponding meanings are shown in table 5.
Table 5 field meanings of RSRP tag data table
Name of field | Means of | Unit |
RSRP | Reference signal received power of grid (X, Y), tag column | dBm |
Since the radio signal is mostly in mW level, it is converted into dBm by polarizing it. dBm is a unit representing an absolute value of power, and the conversion formula is:
0dBm=10lg(1mW) (1)
the above formula can also be understood as 1mW being 0dBm, and dBm being a negative number for a radio signal less than 1 mW. In the actual wireless signal transmission process, the signal receiving side is difficult to achieve the receiving power of 1mW, so that the wireless signals dBm are all negative numbers, and the maximum value is 0. dBm is 0 only in the ideal case, when the receiver receives all the signals transmitted by the transmitter. Generally speaking, the larger the dBm value, the higher the signal strength, the better the receiving effect, but considering the economic cost in practical application, when a region receives a dBm value between 0-50dBm, or between 0-70dBm, the region signal value is considered to be good. When the received wireless signal is less than-70 dBm, the phenomena of unstable transmission and slow speed occur, and at the moment, the wireless network cannot be normally used. In the research, the value of the evaluation index weak coverage judgment threshold P is set as-103 dBm, namely when the dBm value received by an area is between 0 and 103dBm, the signal value of the area is considered to be good.
According to the radio wave propagation process, if a radio channel model is to be designed, it is first necessary to design a feature set for radio wave transmission. The feature corpus mainly comprises: the method comprises three types of wireless channel model characteristics, namely a characteristic based on a spatial position, a characteristic based on a signal deflection angle, a characteristic based on a transmission environment obstacle and the like.
Disclosure of Invention
Technical problem to be solved
The technical problem to be solved by the invention is as follows: how to screen the characteristics of the wireless channel model reduces the complexity of calculation time on the premise of ensuring the reliability of the wireless channel model.
(II) technical scheme
In order to solve the technical problem, the invention provides a wireless channel characteristic screening method based on artificial intelligence, which comprises the following steps:
step one, establishing a feature complete set of a wireless channel model according to the radio wave propagation characteristics;
and step two, screening a characteristic subset which can be used for building a wireless channel model by using an artificial intelligence correlation algorithm, wherein the characteristic subset is a variable in the wireless channel model.
(III) advantageous effects
In the invention, the characteristic complete set related to the wireless channel model is summarized, the characteristic screening of the wireless channel model based on artificial intelligence is carried out, and the high-efficiency machine learning model depends on the strong correlation between the input variable and the problem target, thereby reducing the complexity of the calculation time on the premise of ensuring the reliability of the wireless channel model.
Drawings
FIG. 1 is a coordinate schematic of a rasterized terrain;
FIG. 2 is a schematic representation of engineering parameter data;
FIG. 3 is a flow chart of a feature design of the present invention;
FIG. 4 is a flow chart of feature screening and evaluation according to the present invention;
fig. 5 is a three-dimensional scene representation of signal transmitter and receiver signal transmission.
Detailed Description
In order to make the objects, contents, and advantages of the present invention clearer, the following detailed description of the embodiments of the present invention will be made in conjunction with the accompanying drawings and examples.
The wireless channel modeling for predicting the radio wave propagation characteristics is the basis for constructing a wireless communication system, and with the rapid development and popularization of the fifth generation mobile communication technology, the establishment of a high-precision low-delay wireless channel model is a hot problem in the current research. In the process of establishing a wireless channel model, the model parameters can be regarded as wireless channel characteristic parameters, and the wireless channel model characteristic screening method based on artificial intelligence improves the wireless channel prediction accuracy and reduces the time complexity. Among them, the radio wave propagation path is complex, the wireless channel model parameters are dynamically variable, and determining the model parameters is an important task.
The general scheme design idea of the invention is as follows: on the premise of ensuring the reliability of the wireless channel, a feature complete set of a wireless channel model is established, a better result is extracted from original data by using feature engineering in deep learning, and then optimal prediction is achieved, so that the calculation complexity is increased while the modeling precision of the wireless channel is improved, and reference is provided for the subsequent establishment of the wireless channel model. At present, in the wireless channel modeling process, because the measured data quantity is small, previous researchers all obtain variables in a wireless channel model through experience, and then fit a model formula.
The method is based on a large amount of engineering actual measurement parameter data, utilizes machine learning and deep learning deep mining radio wave propagation characteristics, and screens the characteristics with large correlation with the target label. The calculated amount is reduced on the basis of ensuring the reliability of the wireless channel model, so that theoretical support and technical reference are provided for the base station deployment of a mobile operator.
Referring to fig. 3 and 4, the scheme of the invention comprises the following steps:
step one, establishing a feature complete set of a wireless channel model according to the radio wave propagation characteristics
The method comprises the following steps of:
step 11, data processing
The data used at this time contains 4000 cells, the test points of each cell are different, 12011832 test units are totally arranged, and no missing value exists.
In order to facilitate the subsequent construction of interpretable features, a wireless signal transmission model considering the three-dimensional space position information of a signal base station and a receiving point is designed, and then the features are constructed through a scene. Meanwhile, the validity of the structural characteristics is visually verified by data visualization analysis, and a scene model for signal transmission of a signal transmitter and a receiving point is shown in fig. 5.
Based on the general signal transmission scenario model of the target point and the base station in fig. 5, it is assumed that the sea level is located in a plane H { (x, y, z) | z ═ 0}, which is specifically expressed as follows:
in fig. 5, θ is an actual emission angle of the signal line, and is mathematically expressed as a sum of a Mechanical down tilt (Mechanical down tilt) and an Electrical down tilt (Electrical down tilt), that is, θD=θMD+θED;
The Cell Altitude of the grid (Cell X, Cell Y) where the Cell site is located is represented as hca;
The Building Height of Cell Building in the grid (Cell X, Cell Y) where the Cell site is located is represented as hcb;
The effective antenna height of Cell X, Cell Y grid, i.e. the effective antenna height of mobile station is hlc=hb-hcb;
The Height of the cell transmitter relative to the building, Height, is denoted hc;
Effective height h of cell transmitter base station antennarc=Cell Y+hc;
The height difference between the grid observation point and the base station antenna is expressed as Δ h ═ hcb+hc;
The elevation of the grid observation point is ha;
Current grid cell and transmitter levelThe distance is denoted dh;
The 3D Euclidean distance between the current grid point B and the signal antenna A is represented as D;
the grid (X, Y) Altitude is denoted as hb;
The horizontal directional angle Azimuth of the signal transmitter is denoted as α;
the angle between the signal line and the line AB is marked as the signal deflection angle and is expressed as beta.
The coordinate of A is (x)0,y0,h0) The coordinates of B are (x)1,y1,h1) Wherein the base station horizontal coordinate is (x)0,y0),h0=hc+hcb+hca(ii) a The horizontal coordinate of the receiving point is (x)1,y1),h1=hb+ha。
The above completes the data processing steps, and the following performs the steps of feature creation.
Step 12, designing characteristics based on space positions of radio wave transmission
According to the background knowledge of radio wave transmission, the strength of a radio signal has a direct relation with the transmission distance. The distance d from the combined radio signal to A, B is defined by the relative height difference Δ h and the horizontal distance d between A and BhDetermined jointly, therefore, d, Δ h, dhAs a one-dimensional feature based on spatial location.
A. Horizontal distance d of BhComprises the following steps:
A. the distance d of B is:
step 13, designing characteristics based on signal deflection angles
Considering that the height of the receiving antenna relative to the ground is not given, it is assumed that the average received signal strength is measured on the ground of the grid where the test points are located.
(1) Total base station down tilt angle thetaD:
θD=θMD+θED (4)
In the formula, thetaMDAt a vertical mechanical down-tilt angle, thetaEDIs a vertical electrical downtilt.
In the formula, the coordinate position of the grid test point is (x, y);
the signal transmitted by the base station antenna has a certain concentration, for example, the signal on the back of the antenna is usually weaker than the signal on the front of the antenna at the same angle and distance, i.e. the signal strength is related to the angle between the signal line and the line AB. When other conditions are the same, the smaller the included angle beta between the AB connecting line and the signal line is, the stronger the signal at the point B is. The base station is used as the center of a circle, and the smaller the connecting line between points on the circumference and the base station and the included angle of the signal wire are, the stronger the signal is.
Since the transmitter horizontal direction angle Azimuth is given in the data, i.e., the angle α increases clockwise from the positive Y-axis direction, it is first changed to α' which is commonly used in the two-dimensional plane and increases counterclockwise from the positive X-axis direction:
α′=(2π-α)+π/2 (5)
then the signal line vectorThe vector between the AB lines isSignal lineCorresponding to the connection of points A-BIn a 3D environment, the angle β can be represented by the cosine:
at this time, the process of the present invention,andthe angle in the horizontal plane can be calculated by the same method as described above using the x and y components of the two vectors. And recording an included angle cos alpha and a horizontal plane included angle cos beta in a 3D transmission environment as a signal-based horizontal deflection angle characteristic, wherein the part of characteristics are important characteristics of a training model.
Step 14, designing a feature set based on the radio wave transmission environment barrier
Through the analysis of the Cost231-Hata model, the transmission of the wireless signal is not only related to the spatial position relation of the transmitting and receiving ends, but also related to the signal transmission environment. The radio wave transmission environment is complex and variable, and the radio wave is no longer transmitted in a single path due to various uncertain factors such as the height of a mountain on a transmission path, the density of buildings, the reflection area of a lake and the like. This is also an important factor that the traditional transmission model cannot accurately describe the signal transmission at a finer granularity. Therefore, designing the transmission environment characteristics and realizing reasonable representation of the transmission environment are key points and difficulties for establishing characteristic engineering. The map data given in the data set about the environment are building Height (Height), Altitude (elevation), and ground feature type (Clutter Index). Therefore, robust extraction of transmission environment features should be achieved from three aspects of building distribution and height, cell terrain and ground object types.
(1) Building distribution and height characteristics Mb:
In the formula, Ab、Bb、CbRespectively, are coordinate values representing X, Y, Z (different measurement points building height) axis directions in the building point cloud map. The covariance matrix of the building point cloud is a symmetric matrix according to the covariance matrix property,thus Cov (A)b,Bb)=Cov(Bb,Ab),Cov(Ab,Cb)=Cov(Cb,Ab),Cov(Bb,Cb)=Cov(Cb,Bb). The covariance matrix M of the map point cloudbThere are 6 valid element parameters: cov (A)b,Bb)、Cov(Ab,Cb)、Cov(Bb,Cb) And the respective variances, Cov (A) for the coordinate column vectors A and Bb,Bb) Calculated using the formula:
in the formula, AbiRepresents a column vector AbThe value of the ith element of (1), BbiRepresents a column vector BbThe value of the ith element of (a),is a column vector AbThe average value of (a) of (b),is a column vector BbIs measured.
And obtaining the three-dimensional point cloud of the cell building by using the cell coordinates, the building height data and the RSRP tag data in the transmission environment. The method comprises the steps of extracting features of a cell building, converting the problem of extracting the features of a three-dimensional rigid body target into a problem of extracting the features of the three-dimensional rigid body target, and realizing robust extraction of target point cloud features, wherein the cell building point cloud covariance matrix reflects the distribution and height features of the cell building, so that in the step, the set features are the number of obstacles (obsacle _ num) on a connecting line of an observation point and a base station point, and the transmission loss suffered by the densities of different observation points and the building degree densities on a transmission path is calculated.
Step 15: summarizing all features that have been established
By analyzing traditional wireless channel path loss models such as a Cost231-Hata model, distance information used by the traditional wireless channel path loss models takes a logarithm with a base of 10, and geometric angles can be represented by a trigonometric function. The features and label columns in the original dataset and the new features constructed by the present invention are shown in table 5:
table 5 all features given in the data set with the label column and the new features of the construct
And step two, screening a feature subset which can be used for building a wireless channel model by using an artificial intelligence correlation algorithm, wherein the feature subset is a feature set with a large correlation with a target label in the feature full set, and the feature subset is a variable in the wireless channel model.
Step 21, performing feature screening by using a feature screening method based on machine learning, wherein two methods are adopted for screening, so that feature reliability is improved
The method comprises the following steps: filtering type feature screening method
The filtering type characteristic screening principle is that firstly, according to an agreed rule, grading is carried out according to characteristic divergence or correlation, a threshold value is set, characteristic selection is carried out on a data set, and then a learner is trained. The feature selection process is independent of the subsequent learner, and therefore the computation speed is fast. That is, the features are first "filtered" using a feature selector, and then the filtered features are used to train the neural network model. Filtered feature screening may statistically rank features, with higher ranking features having a higher value. Common filtration methods include T-test (T-test), Chi-square test (X-test)2-test), Pearson Correlation Coefficient (Pearson Correlation Coefficient),PCC), the Maximum Information Coefficient (MIC).
The invention uses two filtering methods of the Pearson correlation coefficient and the maximum information coefficient according to different index evaluation feature scores.
The method 2 comprises the following steps: embedded-based feature screening method
The main idea of embedded feature screening is to select the best features capable of improving accuracy in the trained models and then learn the features. In the process of determining the model, the characteristics meaningful for training the model are selected, and the model training process and the characteristic screening process are integrated.
Two embedded feature evaluation methods are Random Forest (RF) and XGBoost lifting tree models. The random forest is a supervised learning method and is trained based on a bagging method. The so-called bagging method, namely bootstrapping aggregation, adopts a method of randomly selecting and putting back training data, then constructing a classifier, establishing a plurality of unrelated decision tree models, and finally combining the learned models together to obtain a more accurate and stable prediction effect.
The XGboost lifting tree is used for calculating feature scores by utilizing the sum of the splitting times of each tree of all features. For example, if the feature is split once in the first tree, twice in the second tree, and n times in the nth tree … …, the score for the feature is (1+2+. + n). The parameter selection process of the XGBoost is very important, so that the operation of the parameter shuffle can be performed. And finally, performing score average operation on features and socre obtained based on the XGboost with different parameter combinations to screen out high-score features.
The filtering feature screening method has the advantages that the traditional statistical measurement method can be used for rapidly judging the importance of the features, but the correlation of each feature and a target variable needs to be independently examined in the process, and the correlation information and the combination effect between different features are ignored. The wrapped and embedded feature evaluation methods can better evaluate the importance of features in consideration of the effects generated by different feature combinations, but the calculation time overhead is large.
TABLE 6 Scoring of different characteristics under respective evaluation index
Step 22, obtaining a single feature screening result based on machine learning, and performing simulation analysis
For feature-to-target correlations, two filter evaluations and CVs were used for scoring. Regarding the correlation between the features and the targets, a weighted evaluation (the sum of weights is 1) which is scored by 3 evaluation indexes is defined, the features are given a comprehensive target correlation score by integrating feature results obtained by different evaluation modes, and the target correlation score is distributed evenly by the weights, so that the target correlation score formula is as follows:
Score=ω1SPCC+ω2SMIC+ω3SXGBoost (9)
the results of scoring the constructed unique features using the above 3 evaluation indexes are shown in table 6. Where the scores have been appropriately scaled for observation, subject to the conditions under which the data is collected. The absolute value of each evaluation index is divided into 100 points, and the total points are distributed to each feature to complete the scoring. The higher the number of scores obtained for a feature, the greater the correlation.
TABLE 7 relevance scores and rankings of different characteristics under various evaluation indexes
Step 23: feature screening method based on deep learning
The characteristic subset of the wireless channel model is combined by using an artificial intelligence correlation algorithm, 36 single characteristics are constructed based on a radio wave transmission process, but in the process of actually establishing the radio wave prediction model and the wireless channel, if a neural network is required to be trained efficiently, the characteristics need to be subjected to further data mining. By using different feature filtering methods in machine learning, the dimensions of the features constructed in connection with the radio wave transmission process are also different. Therefore, in order to merge multi-scale features into more discriminative power and create new features from the existing features, the method automatically completes the work by Deep Feature Synthesis (DFS).
Depth feature synthesis is a method that can quickly create feature combinations with different depths, and constructs new, deeper features by decomposing complex mass data into digital components. The depth feature synthesis process is a visualization process, and new features constructed by the depth feature synthesis process can be explained in a practical meaning. Featuretools is an open source library for executing automatic feature engineering, and by converting different relational data sets into feature matrixes which can be used for machine learning, the depth feature synthesis can be rapidly promoted, so that more time is provided for focusing on other aspects of machine learning model construction.
Featuretools mainly involves three basic concepts: entity (entity), relationship (relationship) and operator (principal), which in effect is a framework that provides a transformation from single tables and multi-table bridging. After depth feature synthesis using Featuretools, several new features will be generated. The new features are generated by Primary key (Cell Index) aggregation from the child table to the parent table based on easily understood feature primitives.
TABLE 8 Featuretools Generation of partial New features
It can be seen that the engineering data set used by the method is the actually measured communication data set, and the actual meanings corresponding to the labels are introduced one by one. And selecting proper characteristics capable of better describing the transmission loss of the wireless channel according to a traditional wireless channel path loss model and the provided data information, and evaluating the scientificity and the practical engineering significance of the constructed characteristics by calculating scores on the target tags. In the process of actually establishing the radio wave prediction model and the wireless channel, if the neural network is required to be trained efficiently, the features need to be subjected to further data mining. Therefore, in order to merge multi-scale features into more discriminative power, new features are created from the existing features, and the feature selection work in a wireless channel is automatically completed by adopting depth feature synthesis.
The above description is only a preferred embodiment of the present invention, and it should be noted that, for those skilled in the art, several modifications and variations can be made without departing from the technical principle of the present invention, and these modifications and variations should also be regarded as the protection scope of the present invention.
Claims (10)
1. A wireless channel characteristic screening method based on artificial intelligence is characterized by comprising the following steps:
step one, establishing a feature complete set of a wireless channel model according to the radio wave propagation characteristics;
and step two, screening a characteristic subset which can be used for building a wireless channel model by using an artificial intelligence correlation algorithm, wherein the characteristic subset is a variable in the wireless channel model.
2. The method of claim 1, wherein step one comprises the data processing step of: step 11, designing a wireless signal transmission model considering the three-dimensional spatial position information of the signal base station and the receiving point; and a feature establishing step performed based on the data processing step.
3. The method of claim 2, wherein the signal transmission scene models of the target point and the base station site established in the step one are as follows:
assuming that the sea level is H { (x, y, z) | z { (0 }), and θ is the actual emission angle of the signal line, and is mathematically expressed as the sum of the mechanical and electronic downtilt angles, that is, θD=θMD+θED;
The altitude of the grid (Cell X, Cell Y) where the Cell site is located is represented as hca;
The building height of the grid (Cell X, Cell Y) in which the Cell site is located is denoted by hcb;
The effective antenna height of Cell X, Cell Y grid, i.e. the effective antenna height of mobile station is hlc=hb-hcb;
The Height of the cell transmitter relative to the building, Height, is denoted hc;
Effective height h of cell transmitter base station antennarc=Cell Y+hc;
The height difference between the grid observation point and the base station antenna is expressed as Δ h ═ hcb+hc;
The elevation of the grid observation point is ha;
The current grid cell horizontal distance from the transmitter is denoted dh;
The 3D Euclidean distance between the current grid point B and the signal antenna A is represented as D;
the grid (X, Y) Altitude is denoted as hb;
The horizontal directional angle Azimuth of the signal transmitter is denoted as α;
the angle between the signal line and the line AB is marked as the signal deflection angle and is expressed as beta.
The coordinate of A is (x)0,y0,h0) The coordinates of B are (x)1,y1,h1) Wherein the base station horizontal coordinate is (x)0,y0),h0=hc+hcb+hca(ii) a The horizontal coordinate of the receiving point is (x)1,y1),h1=hb+ha。
4. The method of claim 3, wherein the feature establishing step specifically comprises:
step 12, designing characteristics based on the space position of radio wave transmission;
step 13, designing a characteristic based on a signal deflection angle;
step 14, designing a feature set based on the radio wave transmission environment barrier;
step 15: all features that have been established are determined.
5. The method according to claim 4, wherein step 12 is specifically:
since the distance d from the radio signal to A, B is defined by the relative height difference Δ h and the horizontal distance d between A and BhDetermined jointly, therefore, d, Δ h, dhAs a one-dimensional feature based on spatial location;
A. horizontal distance d of BhComprises the following steps:
A. the distance d of B is:
6. the method according to claim 5, wherein step 13 is specifically:
the average received signal strength is supposed to be measured on the ground of the grid where the test point is located;
(1) total base station down tilt angle thetaD:
θD=θMD+θED (3)
In the formula, thetaMDAt a vertical mechanical down-tilt angle, thetaEDIs vertical toElectrical downtilt;
in the formula, the coordinate position of the grid test point is (x, y);
given a transmitter horizontal angle, i.e., the angle α increases clockwise from the positive Y-axis, it is first changed to a' that increases counterclockwise from the positive X-axis, which is commonly used in two-dimensional planes:
α′=(2π-α)+π/2 (4)
then the signal line vectorThe vector between the AB lines isSignal lineCorresponding to the line connecting points A-BThe angle β is represented by the cosine in a 3D environment:
at this time, the process of the present invention,andand calculating the included angle on the horizontal plane through the x and y components of the two vectors, and recording the included angle cos alpha and the included angle cos beta of the horizontal plane in the 3D transmission environment as the horizontal deflection angle characteristic based on the signal.
7. The method according to claim 6, wherein step 14 is specifically:
robust extraction of transmission environment characteristics is achieved from three aspects of building distribution and height, district terrain and ground object types;
(1) building distribution and height characteristics Mb:
In the formula, Ab、Bb、CbRespectively, the coordinate values of the building point cloud map representing the X, Y, Z axis direction, and the covariance matrix of the building point cloud is a symmetric matrix, so that Cov (A)b,Bb)=Cov(Bb,Ab),Cov(Ab,Cb)=Cov(Cb,Ab),Cov(Bb,Cb)=Cov(Cb,Bb) Covariance matrix M of map point cloudsbThere are 6 valid element parameters: cov (A)b,Bb)、Cov(Ab,Cb)、Cov(Bb,Cb) And the respective variances, Cov (A) for the coordinate column vectors A and Bb,Bb) Calculated using the formula:
in the formula, AbiRepresents a column vector AbThe value of the ith element of (1), BbiRepresents a column vector BbThe value of the ith element of (a),is a column vector AbThe average value of (a) of (b),is a column vector BbThe mean value of (a);
the three-dimensional point cloud of the cell building can be obtained by utilizing the cell coordinates, the building height data and the RSRP label data in the transmission environment, the feature extraction is carried out on the cell building, the problem of feature extraction on the three-dimensional rigid body target is solved, the robust extraction of the target point cloud feature is realized, the cell building point cloud covariance matrix reflects the distribution and height features of the cell building, so in the step 14, the set features are the number of barriers on the observation point-base station point connecting line, and the transmission loss of the density of the building degree on different observation points and transmission paths can be calculated.
8. The method of claim 1, wherein step two specifically comprises:
step 21, performing feature screening by using a feature screening method based on machine learning;
and step 22, obtaining a single feature screening result based on machine learning, and performing simulation analysis.
9. The method of claim 8, wherein the step 21 of feature screening is performed using a filtered feature screening method and an embedded feature screening method.
10. The method of claim 8, wherein step two further comprises step 23: and (4) performing feature screening by using a feature screening method based on deep learning.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202210206436.0A CN114611388B (en) | 2022-03-02 | 2022-03-02 | Wireless channel characteristic screening method based on artificial intelligence |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202210206436.0A CN114611388B (en) | 2022-03-02 | 2022-03-02 | Wireless channel characteristic screening method based on artificial intelligence |
Publications (2)
Publication Number | Publication Date |
---|---|
CN114611388A true CN114611388A (en) | 2022-06-10 |
CN114611388B CN114611388B (en) | 2022-10-25 |
Family
ID=81861464
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN202210206436.0A Active CN114611388B (en) | 2022-03-02 | 2022-03-02 | Wireless channel characteristic screening method based on artificial intelligence |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN114611388B (en) |
Cited By (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN115952929A (en) * | 2023-03-13 | 2023-04-11 | 南京信息工程大学 | Neural network-based channel prediction method and system under rainfall weather condition |
CN116017488A (en) * | 2023-01-04 | 2023-04-25 | 中国人民解放军61905部队 | Boundary search-based large-area scattering communication coverage prediction method and system |
Citations (8)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN1269952A (en) * | 1997-01-28 | 2000-10-11 | 艾利森电话股份有限公司 | Frequency packing for dynamic frequency allocation in a radiocommunication system |
CN101896007A (en) * | 2010-07-30 | 2010-11-24 | 华为终端有限公司 | Wireless module, terminal and storage method |
CN102111873A (en) * | 2009-12-23 | 2011-06-29 | 中国移动通信集团公司 | Method and device for selecting visible base station as well as method and device for locating terminal |
CN104105106A (en) * | 2014-07-23 | 2014-10-15 | 武汉飞脉科技有限责任公司 | Wireless communication network intelligent-antenna-covered scene automatic classification and recognition method |
CN106980879A (en) * | 2017-04-28 | 2017-07-25 | 厦门大学 | Based on the sea channel parameter estimation method for improving SVMs and Motar transport |
CN108027836A (en) * | 2015-09-24 | 2018-05-11 | 高通股份有限公司 | With the system and method for the query image access images of capture |
CN108307309A (en) * | 2018-01-31 | 2018-07-20 | 重庆邮电大学 | A kind of location of mobile users computational methods based on neighbor base station RSRP information |
WO2022021602A1 (en) * | 2020-07-27 | 2022-02-03 | 南京邮电大学 | 5g reference signal received power prediction method based on geographic information |
-
2022
- 2022-03-02 CN CN202210206436.0A patent/CN114611388B/en active Active
Patent Citations (8)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN1269952A (en) * | 1997-01-28 | 2000-10-11 | 艾利森电话股份有限公司 | Frequency packing for dynamic frequency allocation in a radiocommunication system |
CN102111873A (en) * | 2009-12-23 | 2011-06-29 | 中国移动通信集团公司 | Method and device for selecting visible base station as well as method and device for locating terminal |
CN101896007A (en) * | 2010-07-30 | 2010-11-24 | 华为终端有限公司 | Wireless module, terminal and storage method |
CN104105106A (en) * | 2014-07-23 | 2014-10-15 | 武汉飞脉科技有限责任公司 | Wireless communication network intelligent-antenna-covered scene automatic classification and recognition method |
CN108027836A (en) * | 2015-09-24 | 2018-05-11 | 高通股份有限公司 | With the system and method for the query image access images of capture |
CN106980879A (en) * | 2017-04-28 | 2017-07-25 | 厦门大学 | Based on the sea channel parameter estimation method for improving SVMs and Motar transport |
CN108307309A (en) * | 2018-01-31 | 2018-07-20 | 重庆邮电大学 | A kind of location of mobile users computational methods based on neighbor base station RSRP information |
WO2022021602A1 (en) * | 2020-07-27 | 2022-02-03 | 南京邮电大学 | 5g reference signal received power prediction method based on geographic information |
Non-Patent Citations (2)
Title |
---|
刘燕都等: "虚拟化测控站体系结构设计与关键技术", 《电讯技术》 * |
武敏等: "一种深度残差网络构建的5G无线智能传播模型", 《电讯技术》 * |
Cited By (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN116017488A (en) * | 2023-01-04 | 2023-04-25 | 中国人民解放军61905部队 | Boundary search-based large-area scattering communication coverage prediction method and system |
CN116017488B (en) * | 2023-01-04 | 2023-09-01 | 中国人民解放军61905部队 | Boundary search-based large-area scattering communication coverage prediction method and system |
CN115952929A (en) * | 2023-03-13 | 2023-04-11 | 南京信息工程大学 | Neural network-based channel prediction method and system under rainfall weather condition |
Also Published As
Publication number | Publication date |
---|---|
CN114611388B (en) | 2022-10-25 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
Ahmadien et al. | Predicting path loss distribution of an area from satellite images using deep learning | |
CN114611388B (en) | Wireless channel characteristic screening method based on artificial intelligence | |
EP2659697B1 (en) | Method for the prediction of coverage areas of a cellular network | |
CN110418354A (en) | It is a kind of that propagation model wireless network planning method is exempted from based on machine learning | |
CN108696932A (en) | It is a kind of using CSI multipaths and the outdoor fingerprint positioning method of machine learning | |
CN109874146B (en) | Method and device for predicting path loss | |
WO2022021602A1 (en) | 5g reference signal received power prediction method based on geographic information | |
CN110263111A (en) | Land use based on priori knowledge/cover information Space monitoring method | |
EP3659276B1 (en) | Method and apparatus for analyzing communication environments and designing networks in consideration of trees | |
CN112305559A (en) | Power transmission line distance measuring method, device and system based on ground fixed-point laser radar scanning and electronic equipment | |
JP2019211342A (en) | Weather analyzer, weather analysis method, and program | |
Wang et al. | Optimizing the ultra-dense 5G base stations in urban outdoor areas: Coupling GIS and heuristic optimization | |
CN112528508A (en) | Electromagnetic visualization method and device | |
Portilla-Yandún et al. | On the statistical analysis of ocean wave directional spectra | |
CN106842191B (en) | A kind of acquisition methods of Ionospheric Parameters | |
CN104614722B (en) | A kind of method blocked based on signal-to-noise ratio Discrimination Radar | |
Masood et al. | Interpretable AI-based large-scale 3D pathloss prediction model for enabling emerging self-driving networks | |
CN111458691A (en) | Building information extraction method and device and computer equipment | |
CN114828026A (en) | Base station planning method, device, equipment, storage medium and program product | |
CN113839725B (en) | Method and device for predicting wireless signal propagation | |
CN107205237A (en) | The optimization method and device of a kind of network structure | |
Nagao et al. | A study on urban structure map extraction for radio propagation prediction using XGBoost | |
CN110366188B (en) | Interference measurement point deployment method, interference measurement path planning method and system | |
Wang et al. | Meta-learning approaches for indoor path loss modeling of 5G communications in smart factories | |
CN109889981A (en) | A kind of localization method and system based on two sorting techniques |
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 |