Improved tracking loop based on FNN and operation method of anti-interference 5G opportunity signal positioning receiver

Abstract

The application belongs to the technical field of 5G NAVSOP receiving, and particularly relates to a 5G NAVSOP receiving device, which comprises a synchronous adjustment module, a signal preprocessing module, a tracking unit, a pseudo-range acquisition module and a linear transformer.

Description

Technical Field

[0001] This invention belongs to the field of 5G NAVSOP receiving technology, specifically relating to an operation method for an anti-interference 5G opportunistic signal positioning receiver based on an improved tracking loop using FNN. Background Technology

[0002] With the development of the information age, location services have become increasingly important. Accurate location information is indispensable in emergency rescue, urban planning, logistics and transportation, and personal navigation. However, in urban environments, high-rise buildings, heavy traffic, and interference from electronic signals make precise positioning extremely difficult. To find an alternative to global navigation satellite systems, researchers have begun to study an innovative navigation technology called opportunistic signal navigation, which uses commercial communication signals as opportunistic signals for positioning instead of traditional navigation signals. Although research on opportunistic signal positioning has made progress, achieving widespread and accurate positioning in urban environments remains challenging. Currently used signals such as Wi-Fi, Bluetooth, and Ultra-Wide Band (UWB) all have limitations; their signals are unevenly distributed and have low coverage, hindering the widespread application of opportunistic signals. However, the development of fifth-generation wireless technology has brought new possibilities to opportunistic signal navigation. The high bandwidth and ultra-dense network deployment characteristics of 5G signals can provide extensive and continuous signal coverage in specific areas, thus providing high-quality opportunistic positioning signal sources.

[0003] With the completion of the 5G NAVSOP research framework, specific challenges such as mitigating multipath effects are receiving increasing attention. Various GNSS multipath mitigation strategies, such as antenna technology and high-resolution correlators (HRC), have been considered for use in cellular signal-based navigation receivers. However, filtering low-altitude signals is difficult because multipath mitigation methods using antenna technology are not entirely feasible, as signals from terrestrial 5G stations typically exhibit low elevation angles. To address this challenge, Ali A. Abdallah and his team employed a reinforcement learning-based ground vehicle localization method that demonstrated significant effectiveness in reducing multipath effects, thus affirming the enormous potential of machine learning in this field. However, the challenge of multipath propagation becomes increasingly critical in complex environments, especially under conditions of large signal-to-noise ratio variations and fluctuating multipath conditions. A significant gap exists in lightweight machine learning algorithms that can effectively track pseudoranges without pre-training. Developing such algorithms is crucial for maintaining signal integrity and accurately extracting localization data in multipath scenarios, which is essential for high-precision applications such as autonomous driving. Summary of the Invention

[0004] This invention provides an operation method for an FNN-based improved tracking loop and anti-interference 5G opportunistic signal positioning receiver, which solves the problems of low positioning accuracy and unstable pseudorange acquisition caused by traditional EML technology.

[0005] This invention is achieved through the following technical solution:

[0006] An improved tracking loop based on FNN, the improved tracking loop includes a synchronization adjustment module, a signal preprocessing module, a tracking unit, a pseudo-distance acquisition module, and a linear transformer;

[0007] Both the synchronization adjustment module and the signal preprocessing module receive 5G NR signals. The synchronization adjustment module transmits the signal to the tracking unit, the signal preprocessing module transmits the signal to the pseudo-range acquisition module, the pseudo-range acquisition module transmits the signal to the linear converter, and the linear converter and the tracking unit transmit signals to each other.

[0008] The tracking unit includes a phase rotation module, an EML discriminator, and a pre-trained network. The phase rotation module receives signals from the synchronization adjustment module and transmits them to the EML discriminator. The EML discriminator transmits the signals to the pre-trained network, and the pre-trained network transmits the signals to the linear converter. The phase rotation module and the linear converter transmit signals to each other.

[0009] Furthermore, the improved tracking loop includes a discrimination function based on morning and evening power. The TOA estimate of the discrimination function based on morning and evening power is normalized by sampling time T and divided into integer and fractional parts. The zero point and TOA estimate data of the discriminator function are obtained using the TDL channel loop. The zero point and TOA estimate data of the discriminator function are used as the reference information of the phase detector in the tracking loop and are transformed into empirical parameters after fitting.

[0010] The integer part is used to adjust the position of the FFT window, while the fractional part controls the magnitude of the phase rotation;

[0011] The pre-trained network processes the original EML output, generates new time delay tracking results, and uses them as input for the next tracking loop.

[0012] Furthermore, the EML discriminator evaluates the time alignment of the signal by comparing the power difference between the early and late signals; this is generated in the frequency domain through phase rotation, and its mathematical expression is shown in Equation (1):

[0013]

[0014] Among them, S k ε represents the DMRS value on subcarrier k in the frequency domain, ε represents the preset time offset, and N represents the total number of subcarriers;

[0015] The output of the EML discriminator is the power difference between the early and late signals, used to indicate the time deviation of the signals. The measure of the deviation can be expressed by formula (2):

[0016]

[0017] Where N represents the total number of subcarriers, and C represents the received signal strength. The normalized S-curve is represented by formula (3).

[0018]

[0019] in, This represents the normalized timing error.

[0020] Normalization factor k dll The value is obtained from the derivative of the discriminator's S-curve at 0, as shown in formula (4).

[0021]

[0022] The further normalized discriminator can be expressed as formula (5).

[0023]

[0024] The delay estimate η is adjusted based on the discriminator's output, and the update rule is defined as shown in formula (6).

[0025]

[0026] Furthermore, a linear transformation is used to correct the output of the EML discriminator. The relevant transformation formula is detailed in formula (7):

[0027]

[0028] Where η represents the estimated time delay.

[0029] Furthermore, the pre-trained network uses the feature data of the discriminator function as the training set, and the training set preparation process specifically includes the following steps:

[0030] Step S1: Based on the 5G protocol, the basic transmission signal s(k) is generated;

[0031] Step S2: Delay τ to s(k) by applying different normalization symbols;

[0032] Step S3: Adjust Nb from -50 to 50 and set γ = 0.01 to simulate signals with different fractional delays, where (Nb max -Nbmin γ = 1, used to ensure that the trained latency tracking network can identify fractional delays within a sampling point.

[0033] Furthermore, in step S2, noise with different signal-to-noise ratios is added to each delayed signal to generate signals s(τ) with different SNR levels, wherein the SNR values ​​are set to an arithmetic sequence from 5dB to 50dB with a step size of 5dB, covering {5,10,15,,50}dB.

[0034] Furthermore, after generating the 5G analog signal, the signal is sent to a preset SDR for data preprocessing and pseudorange acquisition; then these data are sent to the tracking loop; the output value of the tracking loop is used as the feature of the training set, and the relative relationship between the estimated TOA value and the true TOA value is used as the label.

[0035] Each data point (x i ,y i All of them conform to the following formula (8),

[0036]

[0037] in, The feature value of the i-th sample obtained from the normalized discriminator function, label y i It is a binary value that represents the state of the current TOA estimate relative to the true TOA; if the current estimated TOA lags behind the true TOA, the label is 0; if the current estimated TOA leads the true TOA, the label is 1.

[0038] Furthermore, a feedforward neural network is used to construct an improved tracking loop, specifically,

[0039] The network structure includes an input layer, two fully connected layers, and a ReLU activation layer, and finally a softmax layer to achieve binary classification.

[0040] Each fully connected layer contains 50 neurons and is designed to capture complex data features and patterns;

[0041] The Adam optimizer was used, with 100 training epochs, a mini-batch size of 64 as described in this invention, and an initial learning rate of 0.01.

[0042] An operation method for an anti-interference 5G opportunity signal positioning receiver based on an improved tracking loop using an FNN, the operation method comprising the following steps:

[0043] Step 1: Receive 5G signals using USRP and perform data preprocessing; during data preprocessing, location information elements are expanded.

[0044] Step 2: Based on the data preprocessing in Step 1, perform preliminary delay estimation;

[0045] Step 3: Based on the preliminary estimated delay from Step 2, perform pseudorange measurement for a single base station;

[0046] Step 4: Based on the single base station pseudorange measurement in Step 3, construct a base station pseudorange dataset, which includes the pseudorange results obtained in Step 3 and the cell ID set obtained in Step 1.

[0047] Step 5: Based on the base station pseudorange dataset from Step 4, the navigation result is obtained using the improved tracking loop based on FNN as described in claim 1.

[0048] A computer device includes a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein when the processor executes the computer program, it implements the circuit described above.

[0049] A computer-readable storage medium storing a computer program that, when executed by a processor, implements the circuit described above.

[0050] The beneficial effects of this invention are:

[0051] The stability and accuracy of this invention are significantly improved.

[0052] This invention ensures that the data used for network training is highly authentic and diverse.

[0053] This invention not only demonstrates the potential of ML technology in improving the accuracy and robustness of 5G positioning technology, but also provides a practical method to address the challenges of positioning in complex environments.

[0054] This invention can effectively balance computational efficiency and performance.

[0055] This invention proposes a linear transformation strategy that optimizes the tracking loop using parameters determined through simulation experiments, thereby reducing the inconsistency between the discriminator function zeros and the TOA estimate in theory and practice.

[0056] This invention enhances the performance of the tracking loop. Experiments have shown that using a feedforward neural network to improve the tracking loop achieves the most accurate improvement.

[0057] This invention enables the acquisition of 5G opportunity signals by utilizing the high-performance USRP-B210 device.

[0058] Through field testing, this invention demonstrates that the designed 5G NAVSOP receiver can achieve meter-level ranging accuracy in urban environments, showcasing the feasibility and reliability of the ranging algorithm in real-world conditions.

[0059] The ability of this invention to maintain stable positioning performance in a highly interfering environment demonstrates its significant advantages in anti-interference. Attached Figure Description

[0060] Figure 1 This is a schematic diagram of the structure of the present invention.

[0061] Figure 2 This is a schematic diagram of the data preparation process of the present invention.

[0062] Figure 3 These are schematic diagrams illustrating the algorithm performance under different SNR values. (a) shows the algorithm performance when RMSE is 0.66m, (b) shows the algorithm performance when RMSE is 0.64m, and (c) shows the algorithm performance when RMSE is 0.62m.

[0063] Figure 4 This invention provides a histogram analyzing the root mean square error accuracy of each algorithm under different signal-to-noise ratios with a 95% probability.

[0064] Figure 5 This is a schematic diagram of the simulation scene construction of the present invention.

[0065] Figure 6 This is a comparison diagram of the tracking loop improved by ML, the traditional EML tracking loop, the trajectory using only the tracking loop, and the actual trajectory in Scenario 2 of the present invention.

[0066] Figure 7 This is a bar chart comparing the average positioning errors of various algorithms in scenario two of this invention.

[0067] Figure 8 This is a comparison diagram of the tracking loop improved by ML, the traditional EML tracking loop, the trajectory using only the tracking loop, and the real trajectory in scenario three of this invention.

[0068] Figure 9 This is a bar chart comparing the average positioning errors of various algorithms in scenario three of this invention.

[0069] Figure 10 This is a schematic diagram of the experimental path of the present invention.

[0070] Figure 11 This is the experimental path diagram of the present invention.

[0071] Figure 12This is a line graph of the path-distance measurement results of the present invention.

[0072] Figure 13 This is a line graph showing the path-two distance measurement results of the present invention.

[0073] Figure 14 This is a line graph showing the measured positioning coverage of the present invention.

[0074] Figure 15 These are pseudorange measurement maps, where (a) is the pseudorange measurement map of the SIC algorithm and (b) is the pseudorange measurement map of the ICSP algorithm. Detailed Implementation

[0075] In the following description, specific details such as particular system architectures and techniques are set forth for illustrative purposes and not for limitation, in order to provide a thorough understanding of the embodiments of this application. However, those skilled in the art will understand that this application may also be implemented in other embodiments without these specific details. In other instances, detailed descriptions of well-known systems, apparatuses, circuits, and methods are omitted so as not to obscure the description of this application with unnecessary detail.

[0076] It should be understood that, when used in this specification and the appended claims, the term "comprising" indicates the presence of the described features, integrals, steps, operations, elements and / or components, but does not exclude the presence or addition of one or more other features, integrals, steps, operations, elements, components and / or collections thereof.

[0077] It should also be understood that the terminology used in this application specification is for the purpose of describing particular embodiments only and is not intended to limit the application. As used in this application specification and the appended claims, the singular forms “a,” “an,” and “the” are intended to include the plural forms unless the context clearly indicates otherwise.

[0078] The following is in conjunction with the appendix to this application specification. Figure 1-9 The technical solutions in the embodiments of this application are clearly and completely described. Obviously, the described embodiments are only a part of the embodiments of this application, and not all of the embodiments. Based on the embodiments in this application, all other embodiments obtained by those of ordinary skill in the art without creative effort are within the scope of protection of this application.

[0079] Many specific details are set forth in the following description in order to provide a full understanding of this application. However, this application may also be implemented in other ways different from those described herein. Those skilled in the art can make similar extensions without departing from the spirit of this application. Therefore, this application is not limited to the specific embodiments disclosed below.

[0080] Implementation Method 1

[0081] This embodiment proposes an improved tracking loop based on FNN, which includes a synchronization adjustment module, a signal preprocessing module, a tracking unit, a pseudo-distance acquisition module, and a linear transformer.

[0082] Both the synchronization adjustment module and the signal preprocessing module receive 5G NR signals. The synchronization adjustment module transmits the signal to the tracking unit, the signal preprocessing module transmits the signal to the pseudo-range acquisition module, the pseudo-range acquisition module transmits the signal to the linear converter, and the linear converter and the tracking unit transmit signals to each other.

[0083] The tracking unit includes a phase rotation module, an EML discriminator, and a pre-trained network. The phase rotation module receives signals from the synchronization adjustment module and transmits them to the EML discriminator. The EML discriminator transmits the signals to the pre-trained network, and the pre-trained network transmits the signals to the linear converter. The phase rotation module and the linear converter transmit signals to each other.

[0084] Furthermore, the improved tracking loop includes a discrimination function based on morning and evening power. The TOA estimate of the discrimination function based on morning and evening power is normalized by sampling time T and divided into integer and fractional parts. The zero point and TOA estimate data of the discriminator function are obtained using the TDL channel loop. The zero point and TOA estimate data of the discriminator function are used as the reference information of the phase detector in the tracking loop and are transformed into empirical parameters after fitting.

[0085] The integer part is used to adjust the position of the FFT window, while the fractional part controls the magnitude of the phase rotation;

[0086] The pre-trained network obtained through offline training can process the original EML output, generate new time delay tracking results, and use them as input for the next tracking loop.

[0087] Furthermore, the EML discriminator evaluates the time alignment of the signals by comparing the power difference between the early and late signals; the early and late signals are generated by applying a slight time offset to the reference signal to simulate signal advance or delay that may occur in a multipath environment; these signals are generated in the frequency domain by phase rotation, and their mathematical expression is shown in Equation (1):

[0088]

[0089] Among them, S k ε represents the DMRS value on subcarrier k in the frequency domain, ε represents the preset time offset, and N represents the total number of subcarriers; these calculations help the EML discriminator determine whether the signal matches the estimated time of arrival.

[0090] The output of the EML discriminator is the power difference between the early and late signals, used to indicate the time deviation of the signals. The measure of the deviation can be expressed by formula (2):

[0091]

[0092] Where N represents the total number of subcarriers, and C represents the received signal strength. The normalized S-curve can be expressed as formula (3).

[0093]

[0094] in, ε represents the normalized timing error, and ε represents the preset time offset, where ε is 0.5 to achieve the best effect.

[0095] Normalization factor k dll The value is obtained from the derivative of the discriminator's S-curve at 0, as shown in formula (4).

[0096]

[0097] The further normalized discriminator can be expressed as formula (5).

[0098]

[0099] The delay estimate η is adjusted based on the discriminator's output, and the update rule is defined as shown in formula (6).

[0100]

[0101] The working principle of the tracking loop has been analyzed above. However, in multipath and noisy environments, the zeros of the discriminator function often do not match the actual TOA (Time of Arrival). To address the discriminator function zeros and TOA estimates in theoretical and practical applications, this section uses simulation experiments conducted in a TDL (Transmission of Delay) channel environment to initially obtain a series of zeros and TOA estimates for the discriminator function. These data are used as reference information for the phase detector in the tracking loop and are transformed into empirical parameters through fitting. These empirical parameters are applied to adjust the normalized delay ratio in simulation and field test data, and are applied to the design of tracking loops enhanced by linear transform technology.

[0102] By iteratively updating the tracking loop, the signal delay can be effectively estimated and tracked, achieving time alignment correction. Although this method is quite accurate in multipath delay estimation, it faces numerous challenges in complex urban environments with high dynamics and significant multipath effects. In such environments, the discriminator's output may be affected by noise interference or path variations in the actual delay value; therefore, further optimization of the delay estimation correction strategy is needed. This invention employs a linear transformation strategy, applying a linear correction factor to adjust the discriminator output, thereby reducing the deviation between the discriminator's zero point and the actual arrival time.

[0103] Furthermore, a linear transformation is used to correct the output of the EML discriminator. The relevant transformation formula is detailed in formula (7):

[0104]

[0105] Here, η represents the estimated time delay; the linear transformation is defined by parameters α and β and optimized through extensive simulations under different signal-to-noise ratios and multipath delay conditions. Using these simulation data for linear regression analysis, the values ​​of α and β can be accurately determined, making the time delay estimate provided by the EML discriminator closer to the actual value. Applying this method enables more accurate time alignment correction in complex signal environments. Through this improvement, the tracking loop can more effectively handle multipath propagation and noise interference, thereby enhancing the stability and accuracy of the entire system.

[0106] Furthermore, the pre-trained network uses the feature data of the discriminator function as the training set, that is, the output of the discriminator function is used as the input of the pre-trained network, and the output of the pre-trained network is used as the output of the tracking loop. The goal of the tracking loop is to fine-tune the current TOA estimate in order to better align with the true TOA value. Therefore, this invention constructs this problem as a classification task, where the network's goal is to predict whether the current TOA is lagging or ahead of the true TOA; Figure 2 As shown,

[0107] The training set preparation process specifically includes the following steps.

[0108] Step S1: Based on the 5G protocol, the basic transmission signal s(k) is generated;

[0109] Step S2: Delay τ to s(k) by applying different normalization symbols;

[0110] Step S3: Adjust Nb from -50 to 50 and set γ = 0.01 to simulate signals with different fractional delays, where (Nb max -Nb minγ = 1, used to ensure that the trained latency tracking network can identify fractional delays within a sampling point.

[0111] To make the simulated signal closer to the real environment, this paper incorporates simulated multipath effects into the tapped delay (TDL) channel model to enhance the realism of the simulation. The signal processed by the TDL channel more accurately reflects the propagation characteristics of the real signal.

[0112] Furthermore, in order to further improve network stability and interference resistance, in step S2, noise with different signal-to-noise ratios is added to each delayed signal to generate signals s(τ) with different SNR levels, wherein the SNR values ​​are set to an arithmetic sequence from 5dB to 50dB with a step size of 5dB, covering {5,10,15,,50}dB.

[0113] This setup provides a rich and realistic data foundation for network training by simulating various signal-to-noise environments, ensuring the recognition accuracy and robustness of the latency tracking network in different environments.

[0114] Furthermore, after generating the 5G analog signal, the signal is sent to a preset SDR for data preprocessing and pseudorange acquisition; then these data are sent to the tracking loop; the output value of the tracking loop is used as the feature of the training set, and the relative relationship between the estimated TOA value and the true TOA value is used as the label.

[0115] Each data point (x i ,y i All of them conform to the following formula (8),

[0116]

[0117] in, The feature value of the i-th sample obtained from the normalized discriminator function, label y i It is a binary value that represents the state of the current TOA estimate relative to the true TOA; if the current estimated TOA lags behind the true TOA, the label is 0; if the current estimated TOA leads the true TOA, the label is 1.

[0118] The adjustment magnitude Δt for the tracking loop will be determined based on the output of the pre-trained network to minimize classification error and optimize TOA estimation. The dataset will be split into training and test sets to evaluate the model's performance during training and to validate its generalization ability after training. In this way, the pre-trained network will learn to recognize the complex relationship between input features and TOA states, thereby improving the accuracy of TOA estimation.

[0119] Furthermore, to ensure that the improved tracking loop developed based on the ML algorithm can meet the high real-time requirements of positioning and navigation, this invention places particular emphasis on the lightweight characteristics of the ML algorithm when selecting it. Lightweight algorithms can reduce the model's demand for computational resources while maintaining computational efficiency, thus adapting to the system's requirements for rapid decision-making and processing capabilities. Based on this consideration, this invention selects three representative lightweight ML algorithms to construct the improved tracking loop: Support Vector Machine (SVM).

[0120] Support Vector Machine (SVM), Random Average Decision Trees (RADT), and Free Neural Networks (FNN) are considered ideal choices for achieving fast and accurate binary classification tasks due to their efficient learning capabilities and small model size. Through in-depth analysis and comparison of these lightweight algorithms, this study aims to find the improved tracking loop that best meets the real-time requirements of the system.

[0121] Support Vector Machine

[0122] A linear kernel function is used to take advantage of the linear distribution characteristics of the data and automatically adjust the kernel function size to adapt to different data characteristics and optimize classification results.

[0123] The box constraint parameter is set to 1 by default, but can be adjusted according to experimental needs to balance the complexity of the model and the training effect.

[0124] We chose not to perform data standardization in order to preserve the original distribution characteristics of the data.

[0125] Stochastic average decision tree

[0126] Build 100 decision trees and use a majority voting mechanism to improve the accuracy and stability of classification.

[0127] This method fully leverages the advantages of ensemble learning to improve the model's ability to generalize to new data.

[0128] An improved tracking loop is constructed using a feedforward neural network, specifically,

[0129] The network structure includes an input layer, two fully connected layers, and a ReLU activation layer, and finally a softmax layer to achieve binary classification.

[0130] Each fully connected layer contains 50 neurons and is designed to capture complex data features and patterns;

[0131] The Adam optimizer is used, with 100 training epochs, a mini-batch size of 64 as described in this invention, and an initial learning rate of 0.01. This method optimizes the training process and improves model performance.

[0132] Table 1 Accuracy Assessment

[0133]

[0134] The three pre-trained tracking networks with different time delays were validated using independently generated validation set data. Classification accuracy was used as the evaluation criterion, and the experimental results are shown in Table 1. The results show that the tracking loop improved using FNN achieved the highest accuracy.

[0135] Scene 1

[0136] Building upon the theory of improved tracking loops that combine machine learning (ML) techniques with linear transducers, this section evaluates in detail the performance of ML-enhanced tracking loops in multipath propagation environments, particularly within the context of TDL channel models. Signals were fed into several pre-trained networks, including SVM, RADT, and FNN, using EML and linear transducers as standard inputs. This experiment aimed to evaluate the tracking loop performance under varying signal-to-noise ratios and diverse time delays.

[0137] The simulation parameters are configured as shown in Table 2 (the simulation parameters should be set according to the actual receiving environment).

[0138] By simulating 5G signal generation, multipath channel processing, reception, and resolution, and using the tracking loop structure designed in previous chapters, the TOA value is estimated. After adjusting the SNR conditions and conducting 150 independent randomized repeated experiments for each network, the following results can be obtained: Figure 3 The experimental results.

[0139] Table 2 Simulation parameter configuration

[0140]

[0141] The experimental results show that, in this scenario, improving the tracking loop using the ML algorithm can achieve a low-error ranging effect. Specifically, improving the tracking loop using SVM achieves a ranging accuracy of 0.66m RMS error, while improving it using RADT achieves 0.64m RMS error. In comparison, the FNN network achieves the highest accuracy and possesses the strongest noise and multipath resistance, reaching a ranging accuracy of 0.62m RMS error.

[0142] Scene 2

[0143] To further verify the effectiveness of FNN in improving the tracking loop, this paper selects a second scenario for simulation in a complex urban environment. This environment has many obstructions, a dense distribution of 5G base stations, and high signal power. Multiple 5G opportunistic signal data from various base stations can be received simultaneously at the same point, thus enabling positioning via 5G opportunistic signals under these conditions. For the realism of the experiment, this paper selects a 3D building and road model composed of some buildings in a certain urban area of ​​Hong Kong. The map boundary range is [114.14596, 114.14815; 22.28350, 22.28614].

[0144] By reading the OSM data, a model was created for the key building complex, and the building heights were set to match the actual situation. A total of 20 5G base stations were deployed on some of the rooftops, and their latitude (°), longitude (°), and height (m) coordinates are shown in Table 3.

[0145] Table 3 Base Station Location Distribution Configuration

[0146]

[0147] They were labeled on the map with 20 numbers. Based on the road information on the selected area map, two suitable points were chosen as the initial point and the endpoint. During the trajectory from the initial point to the endpoint, five intermediate points were selected as connection points to form the movement trajectory. These points were connected sequentially along the road from the initial point, and all connecting lines were straight lines along the road. The coordinates of the initial point, the endpoint, and the five connection points are shown in Table 4.

[0148] Connecting these seven points yields the actual trajectory simulated in this environment.

[0149] Table 4 Path Planning

[0150]

[0151] Experimental scenario setup, such as Figure 5 As shown in Section 3.4, a 5G opportunity signal source and its propagation are simulated using the FSPL model. The actual trajectory speed in the simulation is 1 m / s, the total path length from the starting point to the ending point is 272 m, the running time is 272 s, and the direction is always along the road in the forward direction. A point is sampled every second during the movement, and this point is sequentially compared with 20 5G base stations to determine signal reception. The base station number and coordinates that can receive the 5G opportunity signal are recorded for subsequent calculations. When calculating the 5G opportunity signal, the coordinates of each point are calculated using a triangulation method based on least squares estimation.

[0152] Table 5 Channel Parameter Configuration

[0153]

[0154] The following comparisons will be made between the tracking loop improved using ML, the traditional EML tracking loop, and the trajectory using only the tracking loop and the real trajectory. Figure 6 As shown.

[0155] Depend on Figure 7 Experimental results show that the improved tracking loop using FNN achieves a 13% improvement in positioning accuracy compared to the traditional EML tracking loop. Simulation results demonstrate that the enhancement method proposed in this invention significantly enhances positioning capabilities.

[0156] Scene 3

[0157] To further verify the effectiveness of FNN in improving the tracking loop, this invention simulates another environment in a complex city, in which the propagation channel of all signals emitted by base stations is adjusted to the TDL-A channel, and the channel parameter configuration is shown in Table 6.

[0158] The following comparisons will be made between the tracking loop improved using ML, the traditional EML tracking loop, and the trajectory using only the tracking loop and the real trajectory. Figure 8 As shown.

[0159] Depend on Figure 9 Experimental results show that the performance of the OMP algorithm degrades in the TDL-A scenario, while the performance of the FNN-improved EML algorithm remains the best, with an 18% improvement in accuracy compared to the EML algorithm.

[0160] Table 6 Channel Parameter Configuration

[0161]

[0162] The present invention aims to improve the accuracy and robustness of 5G positioning technology in complex environments.

[0163] Implementation Method 2

[0164] This embodiment provides a transfer method for an anti-interference 5G opportunistic signal positioning receiver based on an improved FNN tracking loop. A series of field experiments were completed at the South Stadium of Harbin Engineering University using commercial 5G signals.

[0165] (1) High-gain omnidirectional custom antenna: The device operates in the frequency range of 3.4 GHz to 3.6 GHz and has a gain of 10 dBi.

[0166] (2) Portable inverter power supply: This device supplies power to the experimental equipment.

[0167] (3) Host computer terminal: This device is used for real-time processing of test data.

[0168] (4) USRP-B210: This device is mainly used for signal acquisition. It has a built-in GPS Disciplined Oscillator (GPSDO) module to achieve precise time synchronization. Its hardware clock is controlled by GPSDO and can downconvert and sample 5G signals at a rate of 15.36 MSps.

[0169] (5) GPS receiver and antenna: The DOVE482 high-precision navigation terminal is used to accurately estimate the location of pedestrians on the ground. This receiver achieves centimeter-level positioning accuracy through differential positioning technology. This chapter will use the measurement results of this device as the absolute reference standard for the actual distance.

[0170] Table 7 Experimental Parameter Configuration

[0171]

[0172] The USRP-B210 is a software-defined radio developed by Ettus Research. Based on Analog Devices' AD9361 RF front-end and Xilinx's Zynq-7020 FPGA, it supports a wide frequency range from 70MHz to 6GHz, covering the current 5G cellular signal bands. The device achieves high-precision time synchronization through its built-in GPSDO module.

[0173] The software receiving platform's application is developed in C++ and interacts with the USRP device using the UHD library to receive wireless signals. Key parts of the program include initializing and configuring the USRP device, main function design, receive data flow control, and time synchronization and compensation. The following sections will elaborate on each part.

[0174] Initialization and Configuration: This program first includes the necessary header files, including the UHD library interface (for communication with USRP devices), the standard input / output library (for console interaction), string processing, and file operation libraries, providing support for the program's basic operations. The program also defines several namespaces (e.g., namespace po = boost::program_options;) to simplify code writing and enhance code readability and maintainability.

[0175] Main Function Design: The main function uses the macro definition UHD_SAFE_MAIN(int argc, char* argv[]) to enhance cross-platform compatibility, which is the recommended practice of the UHD library. Inside this function, the program parses command-line arguments, allowing users to customize the configuration of the USRP device, such as sampling rate, frequency, gain, and antenna selection. This provides users with flexibility, enabling them to configure the USRP device according to specific application requirements.

[0176] Receive Data Flow Control: The program defines the `recv_to_file` template function, which is the core of data reception and processing. This function configures the USRP device to initiate continuous or limited-number sample reception and writes the received samples to a file. This step is crucial in the data acquisition process, ensuring effective data storage and subsequent processing. The `recv_to_file` function implements the key logic for GPSDO time synchronization.

[0177] Time Synchronization and Compensation Design: First, the program obtains the current GPS time by calling the sensor interface of the USRP device management board. Then, it sets the internal clock of the USRP device to align with the GPS time when the next Pulse Per Second (PPS) signal arrives. This alignment process is accomplished by setting the USRP time precisely before the arrival of the next PPS, ensuring the accuracy of data acquisition time, which is crucial for this time-synchronization-dependent application scenario. To further improve the accuracy of time synchronization, the program implements a time compensation mechanism. This mechanism first calculates the time difference (time_difference) between the last PPS signal (time_last_pps) of the USRP device and the current GPS time (gps_current_time). Based on this time difference, the program determines whether time compensation is needed and the specific compensation method. If the calculated time difference is greater than 0.5 seconds, this indicates a significant error in time synchronization, requiring compensation. In this case, the program adjusts the time difference by subtracting 1 second to reflect the actual time offset. If the time difference is less than -1 second, the program will enter a loop compensation process. In each loop, the program pauses for a period of 100 milliseconds, recalculates the time difference, and attempts to reduce the time error by increasing it by 1 second. This loop will execute a maximum of 10 times to avoid the risk of an infinite loop. Through the above compensation logic, the program can dynamically adjust the time setting of the USRP device to ensure that it remains highly consistent with the actual GPS time. The compensated time (compensated_time) will be used to set the start time of the receiving task, thereby ensuring the accuracy of the data acquisition time stamp.

[0178] like Figure 10As shown, an operation method for an anti-interference 5G opportunity signal positioning receiver based on an improved tracking loop using an FNN is disclosed. The operation method includes the following steps:

[0179] Step 1: Receive 5G signals using USRP and perform data preprocessing; during data preprocessing, location information elements are expanded.

[0180] Step 2: Based on the data preprocessing in Step 1, perform preliminary delay estimation;

[0181] Step 3: Based on the preliminary estimated delay from Step 2, perform pseudorange measurement for a single base station;

[0182] Step 4: Based on the single base station pseudorange measurement in Step 3, construct a base station pseudorange dataset, which includes the pseudorange results obtained in Step 3 and the cell ID set obtained in Step 1.

[0183] Step 5: Based on the base station pseudorange dataset from Step 4, the navigation result is obtained using the improved tracking loop based on FNN as described in claim 1.

[0184] 5G NAVSOP ranging experiment and results analysis

[0185] To preliminarily verify the effectiveness and ranging performance of the designed 5G NAVSOP receiver, this paper conducted a series of field tests using commercial 5G signals at the South Sports Field of Harbin Engineering University.

[0186] The carrier frequency of the 5G base station has been predicted and used as prior information in the ranging phase. The selected base station parameters are shown in Table 8. The experimental path is as follows: Figure 11 As shown,

[0187] Table 8 Experimental Base Station Parameters

[0188]

[0189] In the experiment, a 5G NAVSOP receiver was used to collect signals from a commercial receiver. A centimeter-level GNSS receiver was also used in this experiment to obtain the precise location information of pedestrians and 5G base stations.

[0190] Two pedestrians walked along two predetermined paths at a fixed speed for 70 seconds. To enhance the comparability of the data, participants paused for approximately 5 seconds before each experiment and remained at pre-marked absolute locations using high-precision positioning equipment for 8 seconds. This was designed to ensure the accuracy and contrast of the data collection, thereby facilitating better analysis of the experimental results.

[0191] By performing super-resolution pseudorange calculations on the offline data acquired by the receiver using the designed software receiver, the following results can be obtained: Figure 12 and Figure 13 The experimental results.

[0192] Data analysis shows that the average distance measurement error for route 2 is 3.17m, while the average distance measurement error for route 1 is 2.68m.

[0193] Experimental results show that, in a simple urban environment with a single macro base station, the 5G NAVSOP receiver designed in this invention can achieve meter-level ranging accuracy. Field test results also confirm the feasibility and reliability of the ranging algorithm proposed in this study in practical applications.

[0194] ICSP Anti-interference Experiment and Result Analysis

[0195] Commercial 5G signals were used as the signal source in the test. This was to comprehensively evaluate the algorithm's resilience.

[0196] Interference robustness was assessed using a virtual base station-based verification method. The study generated a virtual base station signal source through software simulation, allowing for customization of parameters such as transmit power, frequency, and location to simulate the transmission signal characteristics of a real base station, thus precisely controlling the interference intensity. Experiments simulated the attenuation of 5G signals in space, which not only simulated the propagation characteristics of base station signals in natural environments but also facilitated the evaluation of the algorithm's robustness under different interference intensities.

[0197] The basic parameters of the base stations used in the experiment are listed in Table 9.

[0198] Table 9 Experimental Base Station Parameters

[0199]

[0200] Location coverage test

[0201] The purpose of this experiment was to evaluate the performance of various interference cancellation algorithms when faced with the influence of an increased virtual interference source. The experimental setup included three real base stations and one virtual interference source. The signal strength of the virtual interference source was manually adjusted to test and compare the robustness of each algorithm. The entire test process was independently repeated 100 times. The test results show... Figure 14 The figure illustrates the relationship between interference signal strength and effective base station recovery rate. Effective base station recovery rate refers to the proportion of base stations that successfully extract positioning observation data from all test base stations. As interference strength increases, the ICSP algorithm significantly outperforms traditional algorithms such as cross-correlation, SIC, and SCP methods.

[0202] Especially under high interference conditions, the ICSP algorithm can maintain a near-constant recovery rate, demonstrating its excellent ability to cope with strong interference in positioning systems.

[0203] Positioning accuracy test

[0204] The specific experimental setup was as follows: positioning samples were collected every 4 seconds at a fixed point, for a total duration of 120 seconds. The interference intensity was set at 28 dBm. Figure 14 It can be seen that under this level of interference, both the SCP and cross-correlation algorithms cannot achieve an effective base station recovery rate of 75% (i.e., recovering the effective positioning information of three base stations), and therefore cannot perform effective positioning. Therefore, the experiment mainly compared the performance of the SIC and ICSP algorithms.

[0205] The pseudorange measurement results show Figure 15 In the comparison, the ICSP algorithm can calculate the pseudorange relative to four base stations, while the SIC algorithm can only calculate the pseudorange relative to three base stations. After transforming the base station and receiver coordinates from the LLH coordinate system to the ECEF coordinate system, the least squares (LS) positioning algorithm was used for solving the problem. The positioning results showed that the average positioning error of the SIC algorithm was 13.82 meters, while the average positioning error of the ICSP algorithm was 11.66 meters. These results demonstrate the significant advantages of the ICSP algorithm in terms of both the number of base station information recovered and the positioning accuracy.

[0206] Implementation Method 3

[0207] This invention provides an electronic device including a memory, a processor, and a computer program stored in the memory and executable on the processor. The memory stores software programs and modules, and the processor executes various functional applications and data processing by running the software programs and modules stored in the memory. The memory and processor are connected via a bus. Specifically, the processor implements any step in Embodiment 1 by running the computer program stored in the memory.

[0208] It should be understood that, in the embodiments of the present invention, the processor may be a central processing unit (CPU), or it may be other general-purpose processors, digital signal processors (DSPs), application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. A general-purpose processor may be a microprocessor or any conventional processor.

[0209] Memory may include read-only memory, flash memory, and random access memory, and provides instructions and data to the processor. Some or all of the memory may also include non-volatile random access memory.

[0210] It should be understood that if the integrated modules / units described above are implemented as software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, all or part of the processes in the methods described above can also be implemented by a computer program instructing related hardware. This computer program can be stored in a computer-readable storage medium, and when executed by a processor, it can implement the steps of the various method embodiments described above. The computer program includes computer program code, which can be in the form of source code, object code, executable files, or certain intermediate forms. The computer-readable medium can include: any entity or device capable of carrying the computer program code, recording media, USB flash drives, portable hard drives, magnetic disks, optical disks, computer memory, read-only memory (ROM), random access memory (RAM), electrical carrier signals, telecommunication signals, and software distribution media, etc. It should be noted that the content contained in the computer-readable storage medium can be appropriately increased or decreased according to the requirements of legislation and patent practice in the jurisdiction.

[0211] The above description of the disclosed embodiments enables those skilled in the art to make or use the invention. Various modifications to these embodiments will be readily apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the invention. Therefore, the invention is not to be limited to the embodiments shown herein, but is to be accorded the widest scope consistent with the principles and novel features disclosed herein.

[0212] Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the above-described division of functional units and modules is merely an example. In practical applications, the above functions can be assigned to different functional units and modules as needed, that is, the internal structure of the above device can be divided into different functional units or modules to complete all or part of the functions described above. The functional units and modules in the embodiments can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The integrated unit can be implemented in hardware or as a software functional unit. Furthermore, the specific names of the functional units and modules are only for easy differentiation and are not intended to limit the scope of protection of this invention. The specific working process of the units and modules in the above system can be referred to the corresponding process in the foregoing method embodiments, and will not be repeated here.

[0213] It should be noted that the methods and detailed examples provided in the above embodiments can be incorporated into the apparatus and devices provided in the embodiments for mutual reference, and will not be repeated here.

[0214] Those skilled in the art will recognize that the units and algorithm steps of the various examples described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementations should not be considered beyond the scope of this invention.

[0215] The above embodiments are only used to illustrate the technical solutions of the present invention, and are not intended to limit them. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention, and should all be included within the protection scope of the present invention.

Claims

1. An improved tracking loop based on FNN, characterized in that, The improved tracking loop includes a synchronization adjustment module, a signal preprocessing module, a tracking unit, a pseudo-distance acquisition module, and a linear converter; Both the synchronization adjustment module and the signal preprocessing module receive 5G NR signals. The synchronization adjustment module transmits the signal to the tracking unit, the signal preprocessing module transmits the signal to the pseudo-range acquisition module, the pseudo-range acquisition module transmits the signal to the linear converter, and the linear converter and the tracking unit transmit signals to each other. The tracking unit includes a phase rotation module, an EML discriminator, and a pre-trained network. The phase rotation module receives signals from the synchronization adjustment module and transmits them to the EML discriminator. The EML discriminator transmits the signals to the pre-trained network, and the pre-trained network transmits the signals to the linear converter. The phase rotation module and the linear converter transmit signals to each other. The improved tracking loop includes a discrimination function based on morning and evening power. The TOA estimate of the discrimination function based on morning and evening power is normalized by sampling time T and divided into integer and fractional parts. The zero point and TOA estimate data of the discriminator function are obtained by using the TDL channel loop. The zero point and TOA estimate data of the discriminator function are used as the reference information of the phase detector in the tracking loop and are transformed into empirical parameters after fitting. The integer part is used to adjust the position of the FFT window, while the fractional part controls the magnitude of the phase rotation; The pre-trained network processes the original EML output, generates new time delay tracking results, and uses them as input for the next tracking loop; The EML discriminator evaluates the time alignment of signals by comparing the power difference between early and late signals; it is generated in the frequency domain through phase rotation, and its mathematical expression is shown in formula (1): （1） in, Represents frequency domain subcarriers DMRS value on This represents the preset time offset. Represents the total number of subcarriers; The output of the EML discriminator is the power difference between the early and late signals, which indicates the time deviation of the signal. The measure of the deviation can be expressed as formula (2): （2） in, Represents the total number of subcarriers. Represents the received signal strength. The normalized S-curve is represented by formula (3). （3） in, This represents the normalized timing error; Normalization factor The value is obtained from the derivative of the discriminator's S-curve at 0, as shown in formula (4). （4） The further normalized discriminator can be expressed as formula (5). （5） Delay estimation Adjustments are made based on the output of the discriminator, and the update rule definition is shown in formula (6). （6） The output of the EML discriminator is corrected by a linear transformation. The relevant transformation formula is detailed in formula (7): （7） in, This represents an estimate of the time delay.

2. The improved tracking loop according to claim 1, characterized in that, The pre-trained network uses the feature data of the discriminator function as the training set. The training set preparation process specifically includes the following steps. Step S1: Based on the 5G protocol, the basic transmission signal was generated. ; Step S2: By applying different normalized symbol delays arrive ; Step S3: Adjustment From -50 to 50 and set =0.01 to simulate signals with different fractional delays, where This is used to ensure that the trained latency tracking network can identify fractional delays within a single sampling point.

3. The improved tracking loop according to claim 2, characterized in that, In step S2, noise with different signal-to-noise ratios was added to each delayed signal, resulting in signals with different SNR levels. The SNR value is set to an arithmetic sequence from 5dB to 50dB with a step size of 5dB, covering {5,10,15,50}dB.

4. The improved tracking loop according to claim 3, characterized in that, After generating a 5G analog signal, the signal is sent to a preset SDR for data preprocessing and pseudorange acquisition; then these data are sent to the tracking loop; the output value of the tracking loop is used as the feature of the training set, and the relative relationship between the estimated TOA value and the true TOA value is used as the label. Each data point All conform to the following formula (8). （8） in, The first one is obtained from the normalized discriminator function. i Feature values ​​and labels of each sample It is a binary value that represents the state of the current TOA estimate relative to the true TOA; if the current estimated TOA lags behind the true TOA, the label is 0; if the current estimated TOA leads the true TOA, the label is 1.

5. The improved tracking loop according to claim 4, characterized in that, An improved tracking loop is constructed using a feedforward neural network, specifically, The network structure includes an input layer, two fully connected layers, and a ReLU activation layer, and finally a softmax layer to achieve binary classification. Each fully connected layer contains 50 neurons and is designed to capture complex data features and patterns; The Adam optimizer was used, with 100 training epochs, a mini-batch size of 64 as described in this invention, and an initial learning rate of 0.

01.

6. A method for operating an anti-interference 5G opportunistic signal positioning receiver based on an improved tracking loop using an FNN, characterized in that, The operation method of the anti-interference 5G opportunity signal positioning receiver uses any of the FNN-based improved tracking loops as described in claims 1-5, and the operation method of the anti-interference 5G opportunity signal positioning receiver includes the following steps: Step 1: Receive 5G signals using USRP and perform data preprocessing; during data preprocessing, location information elements are expanded. Step 2: Based on the data preprocessing in Step 1, perform preliminary delay estimation; Step 3: Based on the preliminary estimated delay from Step 2, perform pseudorange measurement for a single base station; Step 4: Based on the single base station pseudorange measurement in Step 3, construct a base station pseudorange dataset, which includes the pseudorange results obtained in Step 3 and the cell ID set obtained in Step 1. Step 5: Based on the base station pseudorange dataset from Step 4, the navigation result is obtained using the improved tracking loop based on FNN as described in claim 1.

7. A computer device, characterized in that, It includes a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the improved tracking loop as described in any one of claims 1 to 5, or the processor executes the computer program to implement the steps of the method as described in claim 6.

8. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores a computer program that, when executed by a processor, implements the improved tracking loop as described in any one of claims 1 to 5, or, when executed by a processor, implements the steps of the method as described in claim 6.

Images