Wireless radio frequency fingerprint data enhancement and positioning method and device and storage medium

Abstract

The invention relates to the technical field of wireless communication and positioning, in particular to a wireless radio frequency fingerprint data enhancement and positioning method, device and equipment and a computer storage medium. The wireless radio frequency fingerprint data enhancement and positioning method comprises the following steps: firstly, constructing a signal acquisition system, acquiring fingerprint data, preprocessing the fingerprint data, and constructing a radio map after rasterizing a physical map; secondly, modeling and supplementing fingerprints at missing positions in the original radio map to generate a fine-grained radio map; thirdly, using the generative model to obtain more sample data based on the complete radio map; and 4, carrying out position inference by utilizing the expanded radio map and combining machine learning and deep learning algorithms. The efficient positioning method is provided for a moving target positioning system in a GPS failure scene, manual intervention and acquisition time are reduced through automatic data enhancement and signal supplementation, and rapid and accurate moving target positioning is achieved.

Description

Technical Field

[0001] The present invention relates to the field of wireless communication and positioning technology, and in particular to a method, device, equipment and computer storage medium for wireless radio frequency fingerprint data enhancement and positioning. Background Art

[0002] Positioning mobile targets (such as drones, unmanned vehicles, and objects or people carrying mobile devices on cellular networks) can promptly detect and eliminate security threats, thus safeguarding public safety. However, in complex urban environments, forests, canyons, and other environments where GPS signals are weak or ineffective, positioning mobile targets cannot rely solely on GPS or satellite positioning systems like Beidou. Therefore, positioning technologies such as radio frequency and ultra-wideband are often used to achieve high-precision positioning with fast update rates.

[0003] The interaction between the radio frequency signals generated by a mobile target and the positioning system allows the signal strength or arrival time of the radio signal emitted by the mobile target to be calculated, thereby determining the distance between the target and the positioning system. To determine the location of a mobile target, at least three distributed radio receiving devices / base stations are required. The distances between the target and different base stations are calculated using a trilateration algorithm to construct a positioning equation. An optimization algorithm is then used to calculate the optimal solution, thereby determining the mobile target's position. Although trilateration is simple in principle, it is highly dependent on ranging accuracy. Alternatively, the time difference of arrival (TDOA) of wireless signals at different base stations is widely used to determine the target's location, but time synchronization between base stations is a complex and challenging issue. In multi-antenna arrays, time angle of arrival positioning can be used. In addition to the aforementioned methods, wireless fingerprint positioning is currently the predominant positioning method for GPS failure scenarios. This method essentially matches perceived signal features with features in a database. Rapid advances in pattern recognition have enabled the use of cutting-edge machine learning techniques to improve positioning accuracy.

[0004] The accuracy of mobile target positioning using wireless fingerprinting is primarily determined by the sophistication of the radio map / RF fingerprint library and the accuracy of the matching algorithm. In GPS-degraded scenarios, mobile target positioning using wireless fingerprinting presents two challenges. First, wireless signal characteristics can change due to environmental fluctuations and multipath effects, affecting positioning accuracy. Second, developing a high-quality fingerprint library requires extensive field data collection, consuming significant human and material resources. Furthermore, updating and maintaining the library presents an ongoing challenge. Summary of the Invention

[0005] Therefore, the technical problem to be solved by the present invention is to overcome the problems of low accuracy and high cost of wireless fingerprint positioning in the prior art.

[0006] To solve the above technical problems, the present invention provides a wireless radio frequency fingerprint data enhancement and positioning method, comprising:

[0007] Construct an original coarse-grained radio map based on the physical environment map and the original radio signals of the mobile targets;

[0008] Based on a deep learning model, the original coarse-grained radio map is completed to generate a fine-grained radio map;

[0009] Based on the data enhancement model, fingerprint data enhancement is performed on the fine-grained radio map to generate an enhanced fine-grained radio map;

[0010] Collect radio signals of mobile targets in real time, obtain target fingerprint data, perform data preprocessing on the target fingerprint data, and construct target radio frequency fingerprint features;

[0011] Based on the positioning model, the target radio frequency fingerprint feature is matched with the enhanced fine-grained radio map to predict the target location information.

[0012] Preferably, constructing the original coarse-grained radio map based on the physical environment map and the original radio signal of the mobile target includes:

[0013] Rasterize the physical environment map;

[0014] Use the radio map acquisition system to obtain the original radio signals of mobile targets;

[0015] extracting original fingerprint data based on the original radio signal of the mobile target;

[0016] An original coarse-grained radio map is constructed based on the rasterized physical environment map and the original fingerprint data.

[0017] Preferably, the completing the original coarse-grained radio map based on the deep learning model to generate a fine-grained radio map includes:

[0018] Perform data preprocessing on the original coarse-grained radio map;

[0019] Constructing a deep learning model that takes random noise and the original coarse-grained radio map after data preprocessing as input, using the deep learning model to learn the propagation pattern and local characteristics of radio signals in space, and generating a co-modulation generative model through training;

[0020] The co-modulation generation model is used to perform model inference and completion on fingerprint data of unobserved or missing areas in the original coarse-grained radio map to generate a fine-grained radio map.

[0021] Preferably, the constructing of a deep learning model that takes random noise and the original coarse-grained radio map after data preprocessing as input, using the deep learning model to learn the propagation pattern and local characteristics of radio signals in space, and generating a co-modulation generation model through training includes:

[0022] A deep learning model based on conditional co-modulation generative adversarial networks is constructed. The original coarse-grained radio map after data preprocessing is used as the spatial prior input. A multi-scale channel state information feature encoder is introduced to construct the modulation vector. Random noise is added to simulate the diversity of unobserved or missing areas.

[0023] The deep learning model is used to learn the nonlinear mapping relationship from coarse-grained fingerprint features to fine-grained fingerprint features, as well as the propagation pattern and local characteristics of radio signals in space. The generator network is used to infer the fingerprint data distribution of unobserved or missing areas based on the existing radio data characteristics, while the discriminator network evaluates the authenticity of the generated data and continuously optimizes the model generation capability to obtain a co-modulation generation model.

[0024] Preferably, performing fingerprint data enhancement on the fine-grained radio map based on a data enhancement model to generate an enhanced fine-grained radio map includes:

[0025] Build a data augmentation model and train it based on a fine-grained radio map to learn the distribution characteristics of radio signals and generate a set of radio frequency fingerprint samples;

[0026] The radio frequency fingerprint sample set is fused with the fine-grained radio map to generate an enhanced fine-grained radio map.

[0027] Preferably, the step of constructing a data enhancement model and performing training based on a fine-grained radio map to learn the distribution characteristics of radio signals and generate a radio frequency fingerprint sample set includes:

[0028] Build a data enhancement model based on generative adversarial networks and train it based on fine-grained radio maps to learn the distribution characteristics of radio signals.

[0029] During the training process, the generator network is used to simulate the radio signal characteristics of different locations, and the discriminator network is used to evaluate the authenticity of the generated data, so as to continuously optimize the model generation capability and generate a radio frequency fingerprint sample set.

[0030] Preferably, the matching of the target radio frequency fingerprint feature with the enhanced fine-grained radio map based on the positioning model to predict the target location information includes:

[0031] Build positioning models based on machine learning or deep learning;

[0032] Based on the positioning model, the target radio frequency fingerprint feature is matched with the enhanced fine-grained radio map to predict target location information.

[0033] The present invention also provides a wireless radio frequency fingerprint data enhancement and positioning device, comprising:

[0034] A coarse-grained radio map construction module is used to construct an original coarse-grained radio map based on a physical environment map and original radio signals of mobile targets;

[0035] A fine-grained radio map construction module, configured to complete the original coarse-grained radio map based on a deep learning model to generate a fine-grained radio map;

[0036] an enhanced fine-grained radio map construction module, configured to perform fingerprint data enhancement on the fine-grained radio map based on a data enhancement model to generate an enhanced fine-grained radio map;

[0037] A target radio frequency fingerprint feature construction module is used to collect radio signals of mobile targets in real time, obtain target fingerprint data, perform data preprocessing on the target fingerprint data, and construct target radio frequency fingerprint features;

[0038] The target location information prediction module is used to match the target radio frequency fingerprint feature with the enhanced fine-grained radio map based on a positioning model to predict the target location information.

[0039] The present invention also provides a wireless radio frequency fingerprint data enhancement and positioning device, comprising:

[0040] memory for storing computer programs;

[0041] The processor is configured to implement the above-mentioned steps of the wireless radio frequency fingerprint data enhancement and positioning method when executing the computer program.

[0042] The present invention also provides a computer-readable storage medium, on which a computer program is stored. When the computer program is executed by a processor, the steps of the above-mentioned wireless radio frequency fingerprint data enhancement and positioning method are implemented.

[0043] The above technical solution of the present invention has the following advantages over the prior art:

[0044] The wireless radio frequency fingerprint data enhancement and positioning method described in this invention first constructs a signal acquisition system to acquire and preprocess fingerprint data, then rasterizes the physical map to construct a radio map. Second, fingerprints at locations missing from the original radio map are modeled and supplemented to generate a fine-grained radio map. Third, a generative model is used to obtain more sample data based on the complete radio map. Fourth, the expanded radio map is used to perform position inference in combination with machine learning and deep learning algorithms. This invention provides an efficient positioning method for mobile target positioning systems in GPS-invalid scenarios. Through automated data enhancement and signal supplementation, it reduces manual intervention and acquisition time, achieving rapid and accurate mobile target positioning. BRIEF DESCRIPTION OF THE DRAWINGS

[0045] In order to make the content of the present invention more clearly understood, the present invention is further described in detail below based on specific embodiments of the present invention in conjunction with the accompanying drawings, wherein:

[0046] Figure 1 This is a flow chart of the implementation of a wireless radio frequency fingerprint data enhancement and positioning method provided by the present invention;

[0047] Figure 2 It is a visualization diagram of part of the fingerprint in the original coarse-grained radio fingerprint map provided by an embodiment of the present invention;

[0048] Figure 3 This is a comparison diagram of the coarse-grained radio map provided by an embodiment of the present invention before and after completion. DETAILED DESCRIPTION

[0049] The core of the present invention is to provide a wireless radio frequency fingerprint data enhancement and positioning method, device, equipment and computer storage medium, which effectively improves the accuracy of positioning and reduces costs.

[0050] In order to enable those skilled in the art to better understand the present invention, the present invention is further described in detail below with reference to the accompanying drawings and specific embodiments. Obviously, the embodiments described are only some embodiments of the present invention, not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative work are within the scope of protection of the present invention.

[0051] Please refer to Figure 1 , Figure 1 This is a flowchart of the implementation of a wireless radio frequency fingerprint data enhancement and positioning method provided by the present invention; the specific operation steps are as follows:

[0052] S101: constructing an original coarse-grained radio map based on the physical environment map and the original radio signals of the mobile target;

[0053] S102: Based on the deep learning model, the original coarse-grained radio map is completed to generate a fine-grained radio map;

[0054] S103: Based on the data enhancement model, perform fingerprint data enhancement on the fine-grained radio map to generate an enhanced fine-grained radio map;

[0055] S104: collecting radio signals of a mobile target in real time, obtaining target fingerprint data, performing data preprocessing on the target fingerprint data, and constructing target radio frequency fingerprint features;

[0056] S105: Based on the positioning model, match the target radio frequency fingerprint feature with the enhanced fine-grained radio map to predict target location information.

[0057] Based on the above embodiment, this embodiment describes step S101 in detail:

[0058] Based on the physical environment map and the original radio signals of the mobile targets, the original coarse-grained radio map is constructed, which includes:

[0059] Rasterize the physical environment map;

[0060] Use the radio map acquisition system to obtain the original radio signals of mobile targets;

[0061] extracting original fingerprint data based on the original radio signal of the mobile target;

[0062] An original coarse-grained radio map is constructed based on the rasterized physical environment map and the original fingerprint data.

[0063] Among them, the rasterization operation is to divide a continuous area or space into multiple fixed-size grid units (rectangles or squares) according to certain rules to facilitate processing, analysis and visualization. In the radio fingerprint positioning method, the rasterization operation can create a unified coordinate reference system for the map, so that the wireless signal data at different locations can correspond to specific spatial areas. For example, in a 6m×6m area, rasterization is performed at intervals of 1m, which can be divided into 49 areas. In this embodiment, the physical environment map is rasterized and the monitoring area is divided into a regular grid for subsequent positioning and analysis of signal data.

[0064] The radio map acquisition system consists of a software-defined radio (SDR) and multiple receiving antennas. Based on the mobile target's different communication signal types, frequency bands, modulation schemes, and other communication information, an appropriate acquisition system can be selected and configured with an acquisition program. For example, the USRP series uses a USRP B210 as an SDR device, equipped with two VERT2450 omnidirectional antennas, to support wireless signal acquisition in the 2.4 GHz and 5 GHz bands. These bands are commonly used for drone communications, Wi-Fi signals, Bluetooth signals, and other applications, making them widely applicable. The acquisition program can be written in software platforms such as GNU Radio or LabVIEW to control the acquisition system to accurately capture specific wireless signals. For example, in a mobile target localization experiment in an IEEE 802.11p vehicle-to-vehicle communication environment, the acquisition system can be configured to receive OFDM signals in the 5.89 GHz band. The received signal is amplified and frequency-converted by the acquisition system's RF front-end, converting it to a baseband signal. This signal is then converted to a digital signal by an analog-to-digital converter and demodulated by a digital signal processor to obtain the mobile target's original radio signal.

[0065] The raw radio signals of the mobile target are stored on the data storage card using a data processing system (including a data storage card and a processing unit), and the processing unit performs data processing and analysis. Different fingerprint construction methods can be selected during the data processing process: received signal strength (RSS) and channel state estimation (CSI); these are not limited in the present invention.

[0066] As a single scalar value, RSS has the advantages of being easy to obtain and having low computational overhead. However, past experiments have shown that RSS is easily affected by factors such as obstacles, interference, and device orientation, and fluctuates greatly. In addition, RSS only represents the overall signal strength and cannot reflect the multipath characteristics of the signal. Nowadays, RSS is gradually being used less as a fingerprint in radio fingerprint positioning methods. CSI can provide fine-grained subcarrier-level information, including amplitude and phase, which can better characterize wireless channel characteristics. It captures multipath effects and fading characteristics, is more sensitive to short-term channel changes, and is suitable for dynamic scenarios. Currently, most radio fingerprint positioning methods choose CSI as the fingerprint for fingerprint library construction.

[0067] Therefore, in this embodiment, after obtaining the original radio signal of the mobile target, it is subjected to frequency domain analysis, the frequency domain channel response is calculated, and the noise is removed by filtering to extract the CSI information; and the gridded physical environment map and the CSI fingerprint data of the signal collection are combined to complete the construction and precise modeling of the original coarse-grained radio map; in a specific embodiment, the desired UAV communication signal is transformed into the frequency domain using a discrete Fourier transform, and the time domain signal is transformed into the frequency domain. The signal is decomposed using technologies such as orthogonal frequency division multiplexing (OFDM) to recover the signals of different subcarriers to obtain the CSI fingerprint data.

[0068] Among them, during the acquisition process of the original radio signal of the mobile target, the target needs to move at some grids to obtain CSI fingerprint data at different positions. For example, at a certain grid, the acquisition system is used to collect data with a sampling rate and bandwidth of 10MHz and a duration of 3s. After the above data transformation, the baseband signal can be restored to a CSI fingerprint containing 64 subcarriers. The fingerprint has corresponding position coordinate information. The fingerprint and coordinate information are saved together in the original coarse-grained radio map. The visualization of some fingerprints in the original coarse-grained radio fingerprint map is as follows Figure 2 shown.

[0069] Based on the above embodiment, this embodiment describes step S102 in detail:

[0070] Completing the original coarse-grained radio map based on the deep learning model to generate a fine-grained radio map includes:

[0071] Perform data preprocessing on the original coarse-grained radio map;

[0072] A deep learning model is constructed that takes a random noise vector and the original coarse-grained radio map after data preprocessing as joint input. The deep learning model is used to learn the propagation pattern and local characteristics of radio signals in space, and a co-modulation generation model is generated after training. Specifically:

[0073] A deep learning model is constructed that takes a random noise vector and the original coarse-grained radio map after data preprocessing as joint input. This deep learning model is used to learn the nonlinear propagation laws and local multipath characteristic distribution of radio signals in three-dimensional space. Through end-to-end training, a generative network with spatial generalization capabilities is obtained, namely the co-modulation generative model.

[0074] The co-modulation generation model is used to perform model inference and completion on the fingerprint data of unobserved or missing areas in the original coarse-grained radio map to generate a fine-grained radio map. That is, by inputting the target coordinates, the generator of the co-modulation generation model is driven to output high-precision fingerprint data of the location.

[0075] Data preprocessing of the original coarse-grained radio map includes checking data integrity, adjusting the data structure to suit the subsequent model input format, and normalizing fingerprint data such as signal strength, RSS, or CSI at different locations. Specifically, fingerprint data for unobserved locations in the original coarse-grained radio map is set to 0. This data consists of three dimensions: location coordinates, antenna, and subcarrier. This data is then transformed, converting the antenna dimension to the third dimension. This means that the coarse-grained radio fingerprint map is treated as a two-channel image and normalized.

[0076] Among them, the design and construction of a deep learning model that takes random noise and processed original coarse-grained radio maps as input to learn the propagation pattern and local characteristics of radio signals in space, and improve the signal completion capability through generative adversarial training or other learning mechanisms include:

[0077] A deep learning model based on conditional co-modulation generative adversarial networks is constructed. The original coarse-grained radio map after data preprocessing is used as the spatial prior input (i.e., a type of coordination). A multi-scale channel state information feature encoder is introduced to construct the modulation vector. Random noise is added and mapped to a style vector (i.e., another type of coordination) through a feedforward neural network to simulate the diversity of unobserved or missing regions. This introduces noise into the generation condition. Finally, the concept of co-modulation is implemented through weighted fusion of feature space.

[0078] The deep learning model is used to learn the nonlinear mapping relationship from coarse-grained fingerprint features to fine-grained fingerprint features, as well as the propagation mode and local characteristics of radio signals in space (i.e., the multi-scale propagation characteristics and local field strength distribution laws in complex spatial environments). The generator network is used to infer the fingerprint data distribution of unobserved or missing areas based on the existing radio data characteristics, while the discriminator network evaluates the authenticity of the generated data and continuously optimizes the model generation capability. At the same time, conditional information is dynamically injected into each level of the generator through radial transformation to solve the problem of mutual interference between low-frequency and high-frequency features, thereby obtaining a co-modulation generation model.

[0079] The trained co-modulation generation model has successfully learned the radio signal propagation pattern and spatial characteristics, and can infer and complete the CSI or RSS fingerprints of the missing areas, thereby generating a fine-grained radio map. For example, during the construction of the original coarse-grained radio map, 9 locations were selected from 49 locations without manual measurement to reduce manpower and time consumption. The co-modulation generation model can be used to complete the coarse-grained radio map containing the missing location fingerprints; the effect of the original coarse-grained radio map before and after completion is shown as follows: Figure 3 shown.

[0080] Based on the above embodiment, this embodiment describes step S103 in detail:

[0081] Performing fingerprint data enhancement on the fine-grained radio map based on a data enhancement model to generate an enhanced fine-grained radio map includes:

[0082] Build a data augmentation model and train it based on fine-grained radio maps to learn the distribution characteristics of radio signals, generate a set of RF fingerprint samples, and enhance the diversity of the dataset;

[0083] The radio frequency fingerprint sample set is fused with the fine-grained radio map to enrich the sample quantity and spatial distribution characteristics of the radio map, thereby generating an enhanced fine-grained radio map.

[0084] Among them, the data enhancement model is constructed and trained based on the fine-grained radio map to learn the distribution characteristics of radio signals and generate a radio frequency fingerprint sample set including:

[0085] Build a data enhancement model based on generative adversarial networks and train it based on fine-grained radio maps to learn the distribution characteristics of radio signals.

[0086] During training, the generator network simulates radio signal characteristics at different locations, such as signal fading patterns, multipath effects, and noise interference. The discriminator network evaluates the authenticity of the generated data to continuously optimize the model's generation capabilities. Through this adversarial training mechanism, the model is able to synthesize a set of generated RF fingerprint samples that are highly similar to real data.

[0087] The generated RF fingerprint data is fused with the fine-grained radio map to obtain an enhanced fine-grained radio map, which makes up for the deficiency of measurement data and improves the generalization ability of the positioning model, making it more robust and adaptable in actual deployment.

[0088] Based on the above embodiment, this embodiment describes step S104 in detail:

[0089] The steps described above are all offline processing. Mobile target positioning still requires online data collection and matching. The acquisition equipment remains in its original location. The mobile target moves within the monitoring area and transmits radio signals. The radio map acquisition system, controlled by the acquisition code, receives the data and uses the data processing system to preprocess the collected fingerprint data, including denoising, filtering, and feature extraction, to construct the target's corresponding RF fingerprint.

[0090] Based on the above embodiment, this embodiment describes step S105 in detail:

[0091] Design and train a positioning model based on machine learning or deep learning, such as K-nearest neighbor (KNN), random forest (RF), deep neural network (DNN), or convolutional neural network (CNN). Utilize RF fingerprint features to match them with an enhanced fine-grained radio map to infer the target's most likely location. By comparing it with an existing enhanced fine-grained radio map, the model can predict the target's spatial location based on RF signal characteristics, achieving precise positioning.

[0092] In summary, the present invention, based on the wireless fingerprint positioning method, can achieve high-precision positioning of mobile targets in GPS failure scenarios according to actual needs, significantly reducing the material and time costs in the radio map construction process, improving the degree of automation of the construction process, saving a large amount of manual intervention, and achieving a significant improvement in the accuracy and completeness of the radio map. In addition, by enhancing the diversity and robustness of the data set, the model demonstrates strong generalization capabilities in different environments and situations, thereby improving the accuracy and stability of the positioning system in practical applications. This provides a reliable fingerprint library enhancement method based on co-modulation generation for the theoretical method of radio frequency fingerprint data enhancement and mobile target positioning, which has strong applicability.

[0093] The embodiment of the present invention further provides a wireless radio frequency fingerprint data enhancement and positioning device; the specific device may include:

[0094] A coarse-grained radio map construction module is used to construct an original coarse-grained radio map based on a physical environment map and original radio signals of mobile targets;

[0095] A fine-grained radio map construction module, configured to complete the original coarse-grained radio map based on a deep learning model to generate a fine-grained radio map;

[0096] an enhanced fine-grained radio map construction module, configured to perform fingerprint data enhancement on the fine-grained radio map based on a data enhancement model to generate an enhanced fine-grained radio map;

[0097] A target radio frequency fingerprint feature construction module is used to collect radio signals of mobile targets in real time, obtain target fingerprint data, perform data preprocessing on the target fingerprint data, and construct target radio frequency fingerprint features;

[0098] The target location information prediction module is used to match the target radio frequency fingerprint feature with the enhanced fine-grained radio map based on a positioning model to predict the target location information.

[0099] The wireless radio frequency fingerprint data enhancement and positioning device of this embodiment is used to implement the aforementioned wireless radio frequency fingerprint data enhancement and positioning method. Therefore, the specific implementation methods of the wireless radio frequency fingerprint data enhancement and positioning device can be seen in the embodiment part of the wireless radio frequency fingerprint data enhancement and positioning method above. For example, the coarse-grained radio map construction module, the fine-grained radio map construction module, the enhanced fine-grained radio map construction module, the target radio frequency fingerprint feature construction module, and the target position information prediction module are respectively used to implement steps S101, S102, S103, S104 and S105 in the above-mentioned wireless radio frequency fingerprint data enhancement and positioning method. Therefore, its specific implementation methods can refer to the description of the corresponding embodiments of each part and will not be repeated here.

[0100] A specific embodiment of the present invention further provides a wireless radio frequency fingerprint data enhancement and positioning device, comprising: a memory for storing a computer program; and a processor for implementing the steps of the above-mentioned wireless radio frequency fingerprint data enhancement and positioning method when executing the computer program.

[0101] A specific embodiment of the present invention further provides a computer-readable storage medium, on which a computer program is stored. When the computer program is executed by a processor, the steps of the above-mentioned wireless radio frequency fingerprint data enhancement and positioning method are implemented.

[0102] Those skilled in the art will appreciate that the embodiments of the present application can be provided as methods, systems, or computer program products. Therefore, the present application can adopt the form of a complete hardware embodiment, a complete software embodiment, or an embodiment in combination with software and hardware. Moreover, the present application can adopt the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to magnetic disk storage, CD-ROM, optical storage, etc.) that contain computer-usable program code.

[0103] The present application is described with reference to the flowcharts and / or block diagrams of the methods, devices (systems), and computer program products according to the embodiments of the present application. It should be understood that each process and / or box in the flowchart and / or block diagram, as well as the combination of the processes and / or boxes in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to produce a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the steps in the process. Figure 1 a process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0104] These computer program instructions may also be stored in a computer readable memory that can direct a computer or other programmable data processing device to work in a specific manner, so that the instructions stored in the computer readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1 a process or multiple processes and / or boxes Figure 1 The function specified in one or more boxes.

[0105] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operational steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing the instructions executed on the computer or other programmable device for implementing the process. Figure 1 a process or multiple processes and / or boxes Figure 1 A step that specifies a function in one or more boxes.

[0106] Obviously, the above embodiments are merely examples for clarity of explanation and are not intended to limit the implementation methods. Those skilled in the art will appreciate that other variations or modifications can be made based on the above description. It is not necessary and impossible to enumerate all implementation methods here. Obvious variations or modifications arising therefrom remain within the scope of protection of the present invention.

Claims

1. A wireless radio frequency fingerprint data enhancement and positioning method, characterized in that: include: Construct an original coarse-grained radio map based on the physical environment map and the original radio signals of the mobile targets; Based on a deep learning model, the original coarse-grained radio map is completed to generate a fine-grained radio map; Based on the data enhancement model, fingerprint data enhancement is performed on the fine-grained radio map to generate an enhanced fine-grained radio map; Collect radio signals of mobile targets in real time, obtain target fingerprint data, perform data preprocessing on the target fingerprint data, and construct target radio frequency fingerprint features; Based on the positioning model, the target radio frequency fingerprint feature is matched with the enhanced fine-grained radio map to predict the target location information.

2. The method for enhancing and positioning wireless radio frequency fingerprint data according to claim 1, characterized in that: The constructing of the original coarse-grained radio map based on the physical environment map and the original radio signal of the mobile target includes: Rasterize the physical environment map; Use the radio map acquisition system to obtain the original radio signals of mobile targets; extracting original fingerprint data based on the original radio signal of the mobile target; An original coarse-grained radio map is constructed based on the rasterized physical environment map and the original fingerprint data.

3. The method for enhancing and positioning wireless radio frequency fingerprint data according to claim 1, wherein: The method of completing the original coarse-grained radio map based on the deep learning model to generate a fine-grained radio map includes: Perform data preprocessing on the original coarse-grained radio map; Constructing a deep learning model that takes random noise and the original coarse-grained radio map after data preprocessing as input, using the deep learning model to learn the propagation pattern and local characteristics of radio signals in space, and generating a co-modulation generative model through training; The co-modulation generation model is used to perform model inference and completion on fingerprint data of unobserved or missing areas in the original coarse-grained radio map to generate a fine-grained radio map.

4. The method for enhancing and positioning wireless radio frequency fingerprint data according to claim 3, wherein: The method of constructing a deep learning model that takes random noise and the original coarse-grained radio map after data preprocessing as input, using the deep learning model to learn the propagation pattern and local characteristics of radio signals in space, and generating a co-modulation generation model after training includes: A deep learning model based on conditional co-modulation generative adversarial networks is constructed. The original coarse-grained radio map after data preprocessing is used as the spatial prior input. A multi-scale channel state information feature encoder is introduced to construct the modulation vector. Random noise is added to simulate the diversity of unobserved or missing areas. The deep learning model is used to learn the nonlinear mapping relationship from coarse-grained fingerprint features to fine-grained fingerprint features, as well as the propagation pattern and local characteristics of radio signals in space. The generator network is used to infer the fingerprint data distribution of unobserved or missing areas based on the existing radio data characteristics, while the discriminator network evaluates the authenticity of the generated data and continuously optimizes the model generation capability to obtain a co-modulation generation model.

5. The method for enhancing and positioning wireless radio frequency fingerprint data according to claim 1, wherein: The performing fingerprint data enhancement on the fine-grained radio map based on the data enhancement model to generate an enhanced fine-grained radio map includes: Build a data augmentation model and train it based on a fine-grained radio map to learn the distribution characteristics of radio signals and generate a set of radio frequency fingerprint samples; The radio frequency fingerprint sample set is fused with the fine-grained radio map to generate an enhanced fine-grained radio map.

6. The method for enhancing and positioning wireless radio frequency fingerprint data according to claim 5, characterized in that: The data enhancement model is constructed and trained based on a fine-grained radio map to learn the distribution characteristics of radio signals and generate a radio frequency fingerprint sample set, including: Build a data enhancement model based on generative adversarial networks and train it based on fine-grained radio maps to learn the distribution characteristics of radio signals. During the training process, the generator network is used to simulate the radio signal characteristics of different locations, and the discriminator network is used to evaluate the authenticity of the generated data, so as to continuously optimize the model generation capability and generate a radio frequency fingerprint sample set.

7. The method for enhancing and positioning wireless radio frequency fingerprint data according to claim 1, wherein: The matching of the target radio frequency fingerprint feature with the enhanced fine-grained radio map based on the positioning model to predict the target location information includes: Build positioning models based on machine learning or deep learning; Based on the positioning model, the target radio frequency fingerprint feature is matched with the enhanced fine-grained radio map to predict target location information.

8. A wireless radio frequency fingerprint data enhancement and positioning device, characterized in that: include: A coarse-grained radio map construction module is used to construct an original coarse-grained radio map based on a physical environment map and original radio signals of mobile targets; A fine-grained radio map construction module, configured to complete the original coarse-grained radio map based on a deep learning model to generate a fine-grained radio map; an enhanced fine-grained radio map construction module, configured to perform fingerprint data enhancement on the fine-grained radio map based on a data enhancement model to generate an enhanced fine-grained radio map; A target radio frequency fingerprint feature construction module is used to collect radio signals of mobile targets in real time, obtain target fingerprint data, perform data preprocessing on the target fingerprint data, and construct target radio frequency fingerprint features; The target location information prediction module is used to match the target radio frequency fingerprint feature with the enhanced fine-grained radio map based on a positioning model to predict the target location information.

9. A wireless radio frequency fingerprint data enhancement and positioning device, characterized in that: include: memory for storing computer programs; A processor, configured to implement the steps of a wireless radio frequency fingerprint data enhancement and positioning method as claimed in any one of claims 1 to 7 when executing the computer program.

10. A computer-readable storage medium, characterized in that The computer-readable storage medium stores a computer program, and when the computer program is executed by the processor, the steps of the wireless radio frequency fingerprint data enhancement and positioning method according to any one of claims 1 to 7 are implemented.

Images