Radio signal monitoring and positioning method and system based on big data

Abstract

The present invention relates to the field of radio signal monitoring and positioning technology, specifically to a radio signal monitoring and positioning method and system based on big data. The method comprises: capturing a target radio signal and performing entanglement, preliminary encoding, and compression processing to obtain initial data; performing data preprocessing on the initial data and extracting features from the preprocessed data; enhancing the extracted signal features; fusing the enhanced signal features with geographic information data and environmental data in a spatiotemporal topology model; and training a deep learning model using historical signal data and corresponding location tags, with the model outputting a predicted value of the real-time position coordinates of the radio signal. The present invention solves the problems of incomplete signal collection, inaccurate feature extraction, and significant environmental interference in positioning, such as traditional methods, and achieves autonomous monitoring of radio signals and real-time high-precision positioning.

Description

Technical Field

[0001] The present invention relates to the technical field of radio signal monitoring and positioning, and in particular to a radio signal monitoring and positioning method and system based on big data. Background Art

[0002] Radio signal monitoring and positioning is the process of monitoring and analyzing the characteristics and propagation status of radio signals through technical means, and determining the location of the signal source.

[0003] In the field of radio signal monitoring and positioning, traditional technologies rely on conventional collection equipment, and have problems such as limited coverage and susceptibility to environmental interference; feature extraction mostly uses ordinary algorithms, which make it difficult to exploit the complex characteristics of signals; positioning has poor adaptability to dynamic environmental changes, low positioning accuracy, and insufficient real-time performance; as wireless communication scenarios become increasingly complex, higher requirements are placed on the accuracy, real-timeness, and comprehensiveness of signal monitoring and positioning, and new technical solutions are urgently needed to solve these problems. Summary of the Invention

[0004] The purpose of the present invention is to address the problems existing in the background technology and propose a radio signal monitoring and positioning method and system based on big data.

[0005] The technical solution of the present invention: a radio signal monitoring and positioning method based on big data, comprising the following steps:

[0006] S1. Capture the target radio signal, entangle it with the target radio signal to obtain an entangled signal, convert the entangled signal into a digital signal, and perform preliminary encoding and compression processing on the digital signal to obtain initial data;

[0007] S2. Preprocess the initial data, including data cleaning, denoising and normalization, to remove noise and outliers in the data and unify the data format and range;

[0008] S3, input the pre-processed data into the chaotic neural network, and extract the features of the signal data through the multi-layer structure and chaotic mapping mechanism of the network;

[0009] S4, fusing the extracted signal features with relevant data in the spatiotemporal topological model, and enhancing the features by combining geographic information and environmental factors;

[0010] S5, fusing the enhanced signal features with the geographic information data and environmental data in the spatiotemporal topological model to form a comprehensive feature vector for subsequent positioning analysis;

[0011] S6. Use historical signal data and corresponding location labels to train the deep learning model, input the feature data obtained from the spatiotemporal topology analysis into the trained deep learning model, and the model outputs the real-time location coordinate prediction value of the radio signal.

[0012] Preferably, S31, constructing a neural network with chaotic neurons, the dynamic equation of each chaotic neuron is Where, is the neuron state at the nth moment; μ2 is the chaos parameter.

[0013] Preferably, each chaotic neuron receives a weighted input signal from a neuron in the previous layer and outputs the signal after neuron calculation; the input signal received by the i-th chaotic neuron and the output are calculated based on the following formula:

[0014] ;

[0015] ;

[0016] Where, is the connection weight from the jth previous layer neuron to the ith current neuron; is the output of the jth neuron in the previous layer; is the bias of the current neuron; M is the number of neurons in the previous layer; is the output signal after the current neuron is activated; f() is the activation function.

[0017] Preferably, S32, using the collected pre-processed data to train the chaotic neural network, using the back propagation algorithm to update the network weights, and using the mean square error as the loss function, the calculation formula is as follows:

[0018] ;

[0019] Where N is the number of samples; yk is the true value; pyk is the predicted value; k is the sample number, i is a positive integer, k∈[1, N], and N is the total number of samples.

[0020] Preferably, S3 further includes:

[0021] S33, the trained chaotic neural network maps the input signal data to the feature space and outputs a feature vector containing the key features of the signal;

[0022] S34. When the input signal data enters the network, each chaotic neuron follows its dynamic equation Perform iterative updates;

[0023] S35. When signal features propagate from low layers to high layers in the network, neurons in each layer will generate new feature representations based on the input features and their own weight parameters through weighted summation and activation function operations.

[0024] Preferably, S4 includes:

[0025] S41, associating the extracted signal features with the geographic information and environmental data in the spatiotemporal topology model to establish a corresponding relationship between the features and the spatiotemporal nodes;

[0026] S42. Based on the spatiotemporal topology model, an efficient spatiotemporal index structure is established based on the time and space dimensions;

[0027] S43, associating the signal characteristics with the geographic information based on the location information of the signal collection node and the signal propagation path model;

[0028] S44, establishing a correspondence between the signal characteristics and the environmental data based on the monitoring time and location of the environmental data and the time and location information of the signal collection;

[0029] S45. Establish an association accuracy verification mechanism to check the rationality of data association through logical verification;

[0030] S46. Use the attention mechanism to enhance the features to obtain enhanced signal features, and assign different weights to different enhanced signal features based on geographic information and environmental factors.

[0031] Preferably, S5 includes:

[0032] S51, combining the enhanced signal features, geographic information data, and environmental data into a new feature vector;

[0033] S52, using the principal component analysis algorithm, extracting features from different sources respectively, and then fusing the extracted features;

[0034] S53. Based on the fused feature data, analyze the signal propagation path and feature changes in the spatiotemporal topology model;

[0035] S54. Compare the characteristic vectors of the signal at different time and space nodes, calculate the characteristic difference using the Euclidean distance formula, and obtain the characteristic difference analysis result.

[0036] The present invention also discloses a radio signal monitoring and positioning system based on big data, which applies the above-mentioned radio signal monitoring and positioning method based on big data, specifically including:

[0037] a data acquisition module, configured to capture a target radio signal, entangle the target radio signal to obtain an entangled signal, convert the entangled signal into a digital signal, and perform preliminary encoding and compression processing on the digital signal to obtain initial data;

[0038] The data preprocessing module is used to preprocess the initial data, including data cleaning, denoising and normalization, to remove noise and outliers in the data and unify the data format and range;

[0039] The feature extraction module is used to input the pre-processed data into the chaotic neural network and extract the features of the signal data through the multi-layer structure and chaotic mapping mechanism of the network;

[0040] The feature enhancement module is used to fuse the extracted signal features with relevant data in the spatiotemporal topology model and enhance the features by combining geographic information and environmental factors;

[0041] The data training and output module is used to train the deep learning model using historical signal data and corresponding location labels. The feature data obtained by spatiotemporal topology analysis is input into the trained deep learning model, and the model outputs the real-time location coordinate prediction value of the radio signal.

[0042] Compared with the prior art, the above technical solution of the present invention has the following beneficial technical effects:

[0043] The use of quantum entanglement-assisted signal acquisition, chaotic neural network feature extraction and spatiotemporal topological positioning algorithm breaks through the spatiotemporal limitations of traditional acquisition equipment, extracts signal features more accurately, and fully considers the spatiotemporal dynamic changes of signal propagation. It solves the problems of incomplete signal acquisition, inaccurate feature extraction, and positioning affected by large environmental interference in traditional methods, and realizes autonomous monitoring of radio signals and real-time high-precision positioning. BRIEF DESCRIPTION OF THE DRAWINGS

[0044] Figure 1 This is a method block diagram of embodiment 1 proposed by the present invention. DETAILED DESCRIPTION

[0045] Example 1, as Figure 1 As shown, the radio signal monitoring and positioning method based on big data proposed in the present invention includes the following steps:

[0046] S1. Capture the target radio signal, entangle it with the target radio signal to obtain an entangled signal, convert the entangled signal into a digital signal, and perform preliminary encoding and compression processing on the digital signal to obtain initial data. It should be noted that entanglement with the target radio signal to obtain an entangled signal can be obtained by deploying quantum entanglement acquisition nodes and equipment and performing operations on the captured radio signal. This is a prior art and will not be elaborated on here.

[0047] S2. Preprocess the initial data, including data cleaning, denoising and normalization, to remove noise and outliers in the data and unify the data format and range;

[0048] Statistical analysis methods are used to identify and remove noise and outliers in the data. For time series data, the 3σ principle is adopted, that is, a data point is considered an outlier when the difference between the data point and the mean exceeds 3 times the standard deviation. The calculation formula is μ1±3σ, where μ1 is the mean and σ is the standard deviation.

[0049] Use the wavelet denoising algorithm to smooth the data, decompose the data into different frequency domains through wavelet transform, remove the high-frequency noise part, and then perform inverse transform to reconstruct the signal;

[0050] The data is normalized to the interval [0, 1] using the min-max normalization method;

[0051] S3. Input the pre-processed data into the chaotic neural network, and extract features from the signal data through the network's multi-layer structure and chaotic mapping mechanism. The method includes:

[0052] S31. Construct a neural network with chaotic neurons. The dynamic equation of each chaotic neuron is: Where, is the neuron state at the nth moment; μ2 is the chaotic parameter; it should be noted that when μ2 is in a specific interval, the system exhibits chaotic characteristics;

[0053] Each chaotic neuron receives a weighted input signal from the neurons in the previous layer and calculates the output through the neuron. The input signal received by the i-th chaotic neuron and the output are calculated based on the following formula:

[0054] ;

[0055] ;

[0056] Where, is the connection weight from the jth previous layer neuron to the ith current neuron; is the output of the jth neuron in the previous layer; is the bias of the current neuron; M is the number of neurons in the previous layer; is the output signal after the current neuron is activated; f() is the activation function; the activation function can be implemented by the Sigmoid function;

[0057] S32. Use the collected pre-processed data to train the chaotic neural network, use the back propagation algorithm to update the network weights, and use the mean square error as the loss function. The calculation formula is as follows:

[0058] ;

[0059] Where N is the number of samples; yk is the true value; pyk is the predicted value; k is the sample number, i is a positive integer, k∈[1, N], N is the total number of samples;

[0060] S33. The trained chaotic neural network maps the input signal data to a feature space and outputs a feature vector containing key features of the signal, the method comprising:

[0061] Before inputting the signal data into the chaotic neural network, the data is further formatted and normalized according to the network input requirements; different types of radio signal data, such as time domain waveform data and frequency domain spectrum data, are uniformly converted into a tensor form suitable for network input; for example, for time domain signal sequences , reshape it into a three-dimensional tensor , where the first dimension represents the number of samples, which is 1 here, the second dimension is the time series length T, and the third dimension is the number of channels, which is 1 here; at the same time, the Z-score normalization method is used to standardize the data;

[0062] S34. When the input signal data enters the network, each chaotic neuron follows its dynamic equation Perform iterative updates; in this process, due to the extreme sensitivity of the chaotic system to the initial conditions, even if there is a slight difference in the input signal, the state evolution trajectory of the chaotic neuron will produce huge differences; for example, when μ is 3.9, the two initial states only differ by After dozens of iterations, the difference in state values ​​of chaotic neurons will grow exponentially. By cascading multiple layers of chaotic neurons, the network can extract signal features at different levels and scales. For example, the bottom-layer chaotic neurons can capture local details of the signal, such as instantaneous frequency changes; while the upper-layer chaotic neurons can integrate these local features to extract the global structural features of the signal, such as the modulation pattern and periodicity of the signal.

[0063] S35. When the signal features propagate from the lower layer to the higher layer in the network, the neurons in each layer will generate new feature representations based on the input features and their own weight parameters through weighted summation and activation function operation; among them, the sin-cos activation function is used. ; Where g1, g2, g3 and g4 are adjustable parameters;

[0064] After being processed by multiple layers of chaotic neurons and network layers, the chaotic neural network maps the input signal data to a high-dimensional feature space. In order to make the extracted features more distinguishable and interpretable, a feature transformation layer can be added after the output layer; wherein, the principal component analysis (PCA) method is used to reduce the dimensionality of the feature vector, mapping the high-dimensional features to a low-dimensional space while retaining the most important feature information; for example, let the original feature vector be v, and the feature vector after PCA transformation be , where W is the PCA transformation matrix, which consists of the eigenvectors of the covariance matrix of the original eigenvectors;

[0065] S4. Fusing the extracted signal features with relevant data in the spatiotemporal topology model, and enhancing the features by combining geographic information and environmental factors, the method includes:

[0066] S41, associating the extracted signal features with the geographic information and environmental data in the spatiotemporal topology model to establish a corresponding relationship between the features and the spatiotemporal nodes;

[0067] Specifically, unique identification information is added to the different types of data extracted, such as signal feature data, geographic information data, and environmental data. For example, for signal feature data, identifications such as acquisition time and acquisition node number are added; geographic information data is added with geographic coordinate code and geographic entity type identification; and environmental data is added with monitoring time and monitoring location identification.

[0068] S42. Based on the spatiotemporal topology model, an efficient spatiotemporal index structure is established based on the time and space dimensions. Specifically, a quadtree spatial data structure is used, combined with segmented storage of time series, to divide and store different types of data according to their corresponding spatiotemporal nodes.

[0069] S43. Correlating signal characteristics with geographic information based on the location information of the signal collection node and the signal propagation path model. Specifically, utilizing the spatial analysis function of a geographic information system (GIS) to determine the geographic area or geographic entity corresponding to the signal characteristics. For example, if the signal characteristics indicate strong signal obstruction, analyzing the distribution of buildings around the location through GIS can correlate the signal obstruction characteristics with specific building entities. If the signal exhibits abnormal attenuation in a specific terrain area, the attenuation characteristics can be correlated with the geographic attributes of the terrain area.

[0070] S44. Establish a correspondence between signal characteristics and environmental data based on the monitoring time and location of the environmental data, as well as the time and location information of signal collection. For example, when the meteorological data monitored at a certain moment indicates heavy rainfall, and the radio signal collected at that time period and location fluctuates in strength, the signal strength fluctuation characteristics are associated with the meteorological environmental data of heavy rainfall. For electromagnetic environment monitoring data, if electromagnetic interference in a specific frequency band is detected in a certain area, and the signals collected in the same area have anomalies in the same frequency band, the signal anomaly characteristics are associated with the electromagnetic interference data.

[0071] S45. Establish an association accuracy verification mechanism to check the rationality of data association through logical verification. For example, historical data and known signal propagation patterns are used to verify the logical consistency of the association between signal characteristics and geographic information and environmental data. For contradictory or unreasonable data pairs in the association results, manual intervention or automatic correction is performed.

[0072] S46. Use the attention mechanism to enhance the features and obtain enhanced signal features. Different weights are assigned to different enhanced signal features according to geographic information and environmental factors. The calculation formula for assigning weights is as follows:

[0073] ;

[0074] Where, Representation characteristics and features , sim() is the similarity function; H is the number of features; a, b and h are the feature numbers respectively;

[0075] S5. Fusing the enhanced signal features with the geographic information data and environmental data in the spatiotemporal topology model to form a comprehensive feature vector for subsequent positioning analysis. The method includes:

[0076] S51, combining the enhanced signal features, geographic information data and environmental data into a new feature vector; illustratively, the enhanced signal feature vector is , the geographic information feature vector is , the environmental data feature vector is , then it is spliced ​​into a new feature vector ;

[0077] S52, using the principal component analysis algorithm, extract features from different sources respectively, and then fuse the extracted features; for example, construct a feature matrix of different feature vectors to obtain an enhanced signal feature matrix , geographic information feature matrix and environmental data feature matrix , respectively calculate the covariance matrix of the feature matrix 、 and By solving the eigenvalues ​​and eigenvectors of the covariance matrix and selecting the first d principal components, we can get the reduced dimension feature matrix 、 and , and finally concatenate them into the fused feature matrix ;

[0078] S53. Analyze the signal propagation path and characteristic changes in the spatiotemporal topology model based on the fused characteristic data. The method includes:

[0079] Based on the spatiotemporal topology model and signal characteristics, a graph theory method is used to construct a signal propagation path graph. The nodes in the graph represent spatiotemporal nodes, and the edges represent signal propagation paths. The edge weights are determined based on the signal attenuation and propagation time. The spatiotemporal nodes are used as the nodes of the graph, and the nodes are connected to form edges based on the possible propagation paths of the signal. The edge weight setting comprehensively considers multiple factors, such as the attenuation of the signal propagating between two nodes, the propagation time, and the path loss. These factors are then weighted summed to determine the edge weight.

[0080] Compare the feature vectors of the signal at different time and space nodes, use the Euclidean distance formula to calculate the feature difference, and obtain the feature difference analysis results. The Euclidean distance formula is as follows:

[0081] ; Where TY is the feature difference; 、 are the c-th components of the two eigenvectors respectively, where c is a positive integer, c∈[1,m], and m is the total number of components;

[0082] S6. Based on the characteristic difference analysis results and the signal propagation path diagram, determine the approximate propagation direction and range of the signal; specifically, for a series of and the corresponding eigenvectors , calculate the gradient direction of the feature vector between adjacent nodes; for example, taking the two-dimensional feature space as an example, let the adjacent nodes and The eigenvectors are and , then the feature gradient direction The calculation formula is: By calculating the feature gradient direction between multiple adjacent nodes, the direction with the highest frequency is taken as the preliminary direction of signal propagation. This method uses the trend of feature changes to quickly determine the approximate direction of the signal.

[0083] The deep learning model is trained using historical signal data and corresponding location labels. The loss function uses mean squared error, and the model parameters are updated using a stochastic gradient descent algorithm. It should be noted that the deep learning model can be a convolutional neural network (CNN) or a recurrent neural network (RNN).

[0084] The feature data obtained from spatiotemporal topology analysis is input into the trained deep learning model, and the model outputs the real-time location coordinate prediction value of the radio signal.

[0085] In the second embodiment, the radio signal monitoring and positioning system based on big data proposed by the present invention is applied to the radio signal monitoring and positioning method based on big data proposed in the first embodiment, and specifically includes:

[0086] a data acquisition module, configured to capture a target radio signal, entangle the target radio signal to obtain an entangled signal, convert the entangled signal into a digital signal, and perform preliminary encoding and compression processing on the digital signal to obtain initial data;

[0087] The data preprocessing module is used to preprocess the initial data, including data cleaning, denoising and normalization, to remove noise and outliers in the data and unify the data format and range;

[0088] The feature extraction module is used to input the pre-processed data into the chaotic neural network and extract the features of the signal data through the multi-layer structure and chaotic mapping mechanism of the network;

[0089] The feature enhancement module is used to fuse the extracted signal features with relevant data in the spatiotemporal topology model and enhance the features by combining geographic information and environmental factors;

[0090] The data training and output module is used to train the deep learning model using historical signal data and corresponding location labels. The feature data obtained by spatiotemporal topology analysis is input into the trained deep learning model, and the model outputs the real-time location coordinate prediction value of the radio signal.

[0091] The embodiments of the present invention are described in detail above with reference to the accompanying drawings, but the present invention is not limited thereto. Various changes can be made within the scope of knowledge possessed by those skilled in the art without departing from the spirit of the present invention.

Claims

1. A radio signal monitoring and positioning method based on big data, characterized in that: The following steps are involved: S1. Capture the target radio signal through the quantum entanglement acquisition device, perform quantum entanglement with the target radio signal to obtain an entangled signal, convert the entangled signal into a digital signal, and perform preliminary encoding and compression processing on the digital signal to obtain initial data; S2. Preprocess the initial data, including data cleaning, denoising and normalization, to remove noise and outliers in the data and unify the data format and range; S3, input the pre-processed data into the chaotic neural network, and extract the features of the signal data through the multi-layer structure and chaotic mapping mechanism of the network; S4, fusing the extracted signal features with relevant data in the spatiotemporal topological model, and enhancing the features by combining geographic information and environmental factors; S5, fusing the enhanced signal features with the geographic information data and environmental data in the spatiotemporal topological model to form a comprehensive feature vector for subsequent positioning analysis; S6. Use historical signal data and its corresponding location label data to train the deep learning model, input the feature data obtained from the spatiotemporal topology analysis into the trained deep learning model, and the model outputs the real-time location coordinate prediction value of the radio signal.

2. The radio signal monitoring and positioning method based on big data according to claim 1, characterized in that: S31. Construct a neural network with chaotic neurons. The dynamic equation of each chaotic neuron is: Where, is the neuron state at the nth moment; μ2 is the chaos parameter.

3. The radio signal monitoring and positioning method based on big data according to claim 2, characterized in that: Each chaotic neuron receives a weighted input signal from the neurons in the previous layer and calculates the output through the neuron. The input signal received by the i-th chaotic neuron and the output are calculated based on the following formula: ； ； Where, is the connection weight from the jth previous layer neuron to the ith current neuron; is the output of the jth neuron in the previous layer; is the bias of the current neuron; M is the number of neurons in the previous layer; is the output signal after the current neuron is activated; f() is the activation function.

4. The radio signal monitoring and positioning method based on big data according to claim 3, characterized in that: The chaotic neural network is trained using the collected pre-processed data, and the network weights are updated using the back propagation algorithm. The loss function uses the mean square error, and the calculation formula is as follows: ； Where N is the number of samples; yk is the true value; pyk is the predicted value; k is the sample number, i is a positive integer, k∈[1, N], and N is the total number of samples.

5. The radio signal monitoring and positioning method based on big data according to claim 4, characterized in that: S3 also includes: S34, the trained chaotic neural network maps the input signal data to the feature space and outputs a feature vector containing the key features of the signal; S35. When the input signal data enters the network, each chaotic neuron follows its dynamic equation Perform iterative updates; S36. When signal features propagate from lower layers to higher layers in the network, neurons in each layer will generate new feature representations based on the input features and their own weight parameters through weighted summation and activation function operations.

6. The radio signal monitoring and positioning method based on big data according to claim 5, characterized in that S4 include: S41, associating the extracted signal features with the geographic information and environmental data in the spatiotemporal topology model to establish a corresponding relationship between the features and the spatiotemporal nodes; S42. Based on the spatiotemporal topology model, an efficient spatiotemporal index structure is established based on the time and space dimensions; S43, associating the signal characteristics with the geographic information based on the location information of the signal collection node and the signal propagation path model; S44. Establish a corresponding relationship between signal characteristics and environmental data based on the monitoring time and location of environmental data and the time and location information of signal acquisition. S45. Establish an association accuracy verification mechanism to check the rationality of data association through logical verification; S46. Use the attention mechanism to enhance the features to obtain enhanced signal features, and assign different weights to different enhanced signal features based on geographic information and environmental factors.

7. The radio signal monitoring and positioning method based on big data according to claim 6, characterized in that S5 include: S51, combining the enhanced signal features, geographic information data, and environmental data into a new feature vector; S52, using the principal component analysis algorithm, extracting features from different sources respectively, and then fusing the extracted features; S53. Based on the fused feature data, analyze the signal propagation path and feature changes in the spatiotemporal topology model; S54. Compare the characteristic vectors of the signal at different time and space nodes, calculate the characteristic difference using the Euclidean distance formula, and obtain the characteristic difference analysis result.

8. A radio signal monitoring and positioning system based on big data, applied to the radio signal monitoring and positioning method based on big data according to any one of claims 1 to 7, characterized in that: Specifically include: a data acquisition module, configured to capture a target radio signal, entangle the target radio signal to obtain an entangled signal, convert the entangled signal into a digital signal, and perform preliminary encoding and compression processing on the digital signal to obtain initial data; The data preprocessing module is used to preprocess the initial data, including data cleaning, denoising and normalization, to remove noise and outliers in the data and unify the data format and range; The feature extraction module is used to input the pre-processed data into the chaotic neural network and extract the features of the signal data through the multi-layer structure and chaotic mapping mechanism of the network; The feature enhancement module is used to fuse the extracted signal features with the relevant data in the spatiotemporal topology model and enhance the features by combining geographic information and environmental factors; The data training and output module is used to train the deep learning model using historical signal data and corresponding location labels. The feature data obtained by spatiotemporal topology analysis is input into the trained deep learning model, and the model outputs the real-time location coordinate prediction value of the radio signal.

Images