Positioning methods, systems, terminal devices, and media based on time difference of arrival.

Abstract

This application relates to the field of base station positioning technology, and discloses a positioning method, system, terminal device, and medium based on time difference of arrival (TDOA). The TDOA-based positioning method includes: calculating the online differential signal strength and online TDOA of multiple non-reference base stations relative to a reference base station; inputting the online differential signal strength into a pre-trained TDOA deviation prediction model to obtain the predicted TDOA deviation output by the model; correcting the online TDOA based on the predicted TDOA deviation to obtain a target corrected TDOA; and determining the positioning location of the target node based on the target corrected TDOA. This TDOA-based positioning method can significantly reduce the dimensionality of the input data while correcting the online TDOA to obtain an accurate target corrected TDOA, enabling more efficient and accurate positioning.

Description

Technical Field

[0001] This application relates to the field of base station positioning technology, and in particular to a positioning method, system, terminal device and medium based on time difference of arrival. Background Technology

[0002] Ultra-wideband (UWB) positioning technology, based on Time Difference of Arrival (TDoA), has become the mainstream solution for indoor centimeter-level accurate positioning due to the nanosecond-level pulse width and high temporal resolution advantages of UWB signals. However, it still faces severe challenges in actual deployment: factors such as multipath reflection, non-line-of-sight propagation, RF link hardware delay, and clock drift cause strong nonlinearity and non-stationary bias in TDoA observations, affecting positioning accuracy; traditional filtering methods rely on the assumption of error stationarity and are prone to failure in dynamic occlusion environments; while deep learning models based on high-dimensional channel impulse response (CIR) have fitting capabilities, they are difficult to meet the real-time correction requirements of low-power terminals due to their high data dimensionality and computational overhead. Summary of the Invention

[0003] In view of this, embodiments of this application provide a positioning method, system, terminal device, and medium based on time difference of arrival (TDoA), aiming to effectively solve the problem of inaccurate positioning caused by deviations in TDoA observations.

[0004] In a first aspect, embodiments of this application provide a positioning method based on time difference of arrival, comprising: Calculate the online differential signal strength and online time-of-arrival difference (TOD) of multiple non-reference base stations relative to the reference base station; the calculation of the online differential signal strength and online TOD of multiple non-reference base stations relative to the reference base station includes: The system acquires the online signal strength of the reference base station and each of the non-reference base stations when receiving signals in real time, and acquires the online time difference of arrival of the multiple non-reference base stations relative to the reference base station when receiving the signals in real time. The online signal strength and the online time difference of arrival are statically calibrated respectively to obtain the online calibration signal strength and the online calibration time difference of arrival. The difference between the online calibration signal strength of each non-reference base station and the online calibration signal strength of the reference base station is calculated to obtain the online initial differential signal strength. After smoothing and filtering the online initial differential signal strength and the online calibration arrival time difference, the online differential signal strength and the online arrival time difference are obtained respectively. The online differential signal strength is input into a pre-trained time difference deviation prediction model to obtain the predicted time difference deviation output by the time difference deviation prediction model; wherein, the time difference deviation prediction model is trained based on the offline differential signal strength and offline arrival time difference deviation of the non-reference base station relative to the reference base station; The online arrival time difference is corrected based on the predicted time difference deviation to obtain the target corrected time difference, and the location of the target node is determined based on the target corrected time difference.

[0005] In a first possible embodiment of the first aspect, the time difference deviation prediction model is trained in the following manner: The offline differential signal strength and offline time difference of arrival deviation of multiple non-reference base stations relative to the reference base station are obtained to obtain the model training dataset; The regression model is trained based on the training dataset of the model to obtain the time difference deviation prediction model.

[0006] In a second possible embodiment of the first aspect, the step of statically calibrating the online signal strength and the online time difference of arrival to obtain the online calibration signal strength and the online calibration time difference of arrival includes: The average signal strength of each of the non-reference base stations and the reference base station during the static calibration phase is obtained, and the average time difference of arrival of the multiple non-reference base stations relative to the reference base station when receiving the signal is obtained; The difference between the online signal strength and the average signal strength is calculated to obtain the online calibration signal strength; The difference between the online arrival time difference and the average arrival time difference is calculated to obtain the online calibration arrival time difference.

[0007] In a third possible embodiment of the first aspect, obtaining the offline differential signal strength and time-of-arrival deviation of the plurality of non-reference base stations relative to the reference base station includes: The offline signal strength of the reference base station and each of the non-reference base stations when receiving signals is obtained, and the offline time of arrival difference of each of the non-reference base stations relative to the reference base station when receiving the signal is obtained; The offline signal strength and the offline time difference of arrival are statically calibrated respectively to obtain the offline calibration signal strength and the offline calibration time difference of arrival. Calculate the difference between the offline calibration signal strength of each non-reference base station and the offline calibration signal strength of the reference base station to obtain the initial offline differential signal strength; Calculate the difference between the offline calibration arrival time difference and the theoretical arrival time difference for each of the above calculations to obtain the initial arrival time difference deviation; After smoothing and filtering the offline initial differential signal strength and the initial arrival time difference deviation, the offline differential signal strength and the arrival time difference deviation are obtained.

[0008] In a fourth possible embodiment of the first aspect, the step of statically calibrating the offline signal strength and the offline time difference of arrival to obtain the offline calibration signal strength and the offline calibration time difference of arrival includes: The average signal strength of each non-reference base station and the reference base station during the static calibration phase is obtained, and the average time difference of arrival of each non-reference base station relative to the reference base station during the static calibration phase is obtained. The difference between the offline signal strength and the average signal strength is calculated to obtain the offline calibration signal strength; The difference between the offline arrival time difference and the average arrival time difference is calculated to obtain the offline calibration arrival time difference.

[0009] In a fifth possible embodiment of the first aspect, correcting the online arrival time difference based on the predicted time difference deviation includes: The difference between the online arrival time difference and the predicted time difference is calculated to obtain the target corrected time difference.

[0010] In a sixth possible embodiment of the first aspect, determining the location of the target node based on the target correction time difference includes: Based on the target correction time difference, the reference base station, and the base station coordinates of each non-reference base station, multiple hyperbolic positioning constraint equations are established. Each hyperbolic positioning constraint equation is used to characterize the difference between the distance from the positioning location to the non-reference base station and the distance to the reference base station, which is equal to the product of the speed of light and the target correction time difference. The hyperbolic positioning constraint equation is converted into a system of linear equations with the positioning position as the unknown. The initial positioning position of the target node is obtained by analyzing the system of linear equations. Based on the initial positioning position, the distance difference function between the positioning position and each of the non-reference base stations and the reference base station is iteratively solved to obtain the position correction amount of the positioning position; The location is corrected based on the location correction amount to obtain a location estimate, until the distance between the location correction amounts obtained in two adjacent iterations is less than a preset convergence threshold, and the location estimate is taken as the location.

[0011] Secondly, embodiments of this application provide a positioning system based on time difference of arrival, comprising: The online data acquisition module is used to calculate the online differential signal strength and online time difference of arrival of multiple non-reference base stations relative to the reference base station. The online data acquisition module is further configured to acquire in real time the online signal strength of the reference base station and each of the non-reference base stations when receiving signals, and to acquire in real time the online time difference of arrival of the multiple non-reference base stations relative to the reference base station when receiving the signals; perform static calibration on the online signal strength and the online time difference of arrival to obtain the online calibration signal strength and the online calibration time difference of arrival; calculate the difference between the online calibration signal strength of each non-reference base station and the online calibration signal strength of the reference base station to obtain the online initial differential signal strength; and perform smoothing filtering on the online initial differential signal strength and the online calibration time difference of arrival to obtain the online differential signal strength and the online time difference of arrival. The deviation prediction module is used to input the online differential signal strength into a pre-trained time difference deviation prediction model to obtain the predicted time difference deviation output by the time difference deviation prediction model; wherein, the time difference deviation prediction model is trained based on the offline differential signal strength and offline arrival time difference deviation of the non-reference base station relative to the reference base station; The time difference correction module is used to correct the online arrival time difference based on the predicted time difference deviation to obtain the target corrected time difference; The positioning module is used to determine the location of the target node based on the target correction time difference.

[0012] Thirdly, embodiments of this application provide a terminal device, the terminal device including a processor and a memory, the memory storing a computer program, and the processor executing the computer program to implement the above-described positioning method based on time difference of arrival.

[0013] Fourthly, embodiments of this application provide a computer-readable storage medium storing a computer program, which, when executed by a processor, implements the aforementioned positioning method based on time difference of arrival.

[0014] The embodiments of this application have the following beneficial effects: This embodiment of a positioning method based on time difference of arrival includes: calculating the online differential signal strength and online time difference of arrival of multiple non-reference base stations relative to a reference base station; inputting the online differential signal strength into a pre-trained time difference deviation prediction model to obtain the predicted time difference deviation output by the time difference deviation prediction model; wherein, the time difference deviation prediction model is trained based on the offline differential signal strength and time difference deviation of the non-reference base stations relative to the reference base station; correcting the online time difference of arrival based on the predicted time difference deviation to obtain the target corrected time difference; and determining the positioning location of the target node based on the target corrected time difference. Based on the above scheme, this positioning method based on time difference of arrival (TDOA) constructs differential signal strengths between a reference base station and each non-reference base station. While significantly reducing the dimensionality of the input data, it can obtain the predicted time difference deviation output by the time difference deviation prediction model based on the differential signal strength. This time difference deviation prediction model can learn the mapping relationship between differential signal strength and time difference of arrival. Based on the predicted time difference deviation, the arrival time difference can be corrected to obtain an accurate target corrected time difference. Based on the target corrected time difference, more efficient and accurate UWB-TDoA positioning can be achieved, meeting the real-time correction requirements of low-power terminals. Attached Figure Description

[0015] To more clearly illustrate the technical solutions of the embodiments of this application, the accompanying drawings used in the embodiments will be briefly introduced below. It should be understood that the following drawings only show some embodiments of this application and should not be regarded as a limitation of the scope. For those skilled in the art, other related drawings can be obtained based on these drawings without creative effort.

[0016] Figure 1 A schematic diagram of a first embodiment of the positioning method based on time difference of arrival in this application is shown; Figure 2 A second flowchart illustrating the positioning method based on time difference of arrival according to an embodiment of this application is shown; Figure 3 A schematic diagram of a third type of positioning method based on time difference of arrival, according to an embodiment of this application, is shown. Figure 4 A schematic diagram of the offline training process of the time difference deviation prediction model according to an embodiment of this application is shown; Figure 5 This paper presents a comparison chart of the online arrival time difference error before and after correction in an embodiment of this application. Figure 6 A flowchart illustrating the online positioning stage of an embodiment of this application is shown; Figure 7 A schematic diagram of a fourth type of positioning method based on time difference of arrival, according to an embodiment of this application, is shown. Figure 8 This paper presents a comparison chart of the positioning results before and after online arrival time difference correction according to an embodiment of this application. Figure 9 A schematic diagram of the structure of a positioning system based on time difference of arrival according to an embodiment of this application is shown.

[0017] Explanation of key component symbols: 200 - Positioning system based on time difference of arrival; 210 - Online data acquisition module; 220 - Deviation prediction module; 230 - Time difference correction module; 240 - Positioning module. Detailed Implementation

[0018] The technical solutions in the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, and not all embodiments.

[0019] The components of the embodiments of this application described and illustrated in the accompanying drawings can be arranged and designed in a variety of different configurations. Therefore, the following detailed description of the embodiments of this application provided in the drawings is not intended to limit the scope of the claimed application, but merely to illustrate selected embodiments of the application. All other embodiments obtained by those skilled in the art based on the embodiments of this application without inventive effort are within the scope of protection of this application.

[0020] In the following text, the terms "comprising," "having," and their cognates, which may be used in various embodiments of this application, are intended only to indicate a particular feature, number, step, operation, element, component, or combination thereof, and should not be construed as primarily excluding the presence of one or more other features, numbers, steps, operations, elements, components, or combinations thereof, or adding the possibility of one or more combinations thereof. Furthermore, the terms "offline," "online," "third," etc., are used only for distinguishing descriptions and should not be construed as indicating or implying relative importance.

[0021] Unless otherwise specified, all terms used herein (including technical and scientific terms) shall have the same meaning as commonly understood by one of ordinary skill in the art to which the various embodiments of this application pertain. Terms (such as those defined in commonly used dictionaries) shall be interpreted as having the same meaning as in their contextual meaning in the relevant technical field and shall not be construed as having an idealized or overly formal meaning, unless clearly defined in the various embodiments of this application.

[0022] The following detailed description of some embodiments of this application is provided in conjunction with the accompanying drawings. Unless otherwise specified, the following embodiments and features can be combined with each other.

[0023] Ultra-wideband (UWB) positioning technology, based on time difference of arrival (TDOA), is a method that fully utilizes the high temporal resolution of UWB signals to achieve precise positioning. This technology measures the time difference of flight of UWB signals received by multiple base stations from the same tag, and then solves a hyperbolic equation to determine the tag's location with high accuracy. The extremely short pulse width of UWB signals makes them more effective than Bluetooth and WiFi in resisting the effects of multipath interference in complex indoor environments, ensuring positioning accuracy down to the decimeter or even centimeter level.

[0024] Currently, among mainstream high-precision wireless positioning technologies, time-of-arrival (TOA) positioning methods require strict clock synchronization between the tag and the base station, often implemented using two-way ranging (TWR). However, this method increases communication complexity and energy consumption, reducing the number of tags available. In contrast, TDoA-based solutions only require synchronization between base stations; the tag only needs to transmit signals to achieve positioning, significantly reducing positioning latency and energy consumption, thereby greatly simplifying the communication process and enhancing tag scalability.

[0025] Thanks to the strong penetration, low power consumption, and anti-interference performance of UWB signals, TDoA-based UWB positioning systems have demonstrated broad application prospects in practical applications. This technology is not only suitable for densely populated and complex indoor environments such as large shopping malls, stadiums, and underground parking lots, but also plays an important role in warehousing and logistics, industrial automation, and smart cities. With the rapid development of the Internet of Things and intelligent systems, TDoA-based UWB positioning solutions will provide low-cost, high-precision positioning services for various scenarios, meeting the ever-increasing positioning demands.

[0026] Although ultra-wideband (UWB) signals possess good anti-interference capabilities to some extent due to their nanosecond-level pulse width and high time resolution, they are still inevitably affected by various non-ideal factors in practical applications. On the one hand, co-channel interference, transient pulse noise, and equipment circuit noise in complex electromagnetic environments reduce the signal-to-noise ratio (SNR) of the received signal, directly affecting the accuracy of time of arrival (TDOA) extraction. On the other hand, antenna group delay, hardware link delay, and frequency drift and phase jitter of the base station's local clock introduce systematic deviations into the measurements. Furthermore, reflection, obstruction, and scattering effects in indoor environments cause significant multipath propagation of the signal. Even though UWB has a certain multipath resolution capability, it may still misjudge the arrival time of the direct path in non-line-of-sight (NLOS) or strong reflection scenarios. When positioning accuracy requirements reach the centimeter level, the impact of the above-mentioned errors will be further amplified, leading to significant deviations and instability between TDoA measurements and positioning results.

[0027] To correct errors in TDoA measurements, existing research primarily utilizes the statistical correlation between Channel Impulse Response (CIR) and ranging bias to establish estimation models for measurement bias. Common methods include traditional filtering algorithms such as Kalman filtering and particle filtering, or nonlinear mapping models based on neural networks. However, filtering methods typically rely on the stationarity of the error and high accuracy assumptions. In real-world environments, TDoA measurement bias is easily affected by personnel movement, occlusion changes, and reference base station switching, exhibiting strong nonlinear and non-stationary characteristics. This makes it difficult for filtering models to converge stably, resulting in limited correction effectiveness. On the other hand, while neural network-based methods can fit complex nonlinear relationships, CIR data itself has high dimensionality and a large sampling rate. To fully explore the fine-grained correspondence between CIR and measurement bias, it is often necessary to construct large-scale network models and perform extensive training. This not only increases the model storage and computational burden but also limits the system's real-time performance and online deployment capabilities, making it difficult to meet the requirements of low-power, low-latency UWB positioning applications.

[0028] To address the interference issues of multipath interference and clutter on UWB-TDoA positioning, this application provides a positioning method, system, terminal device, and medium based on time difference of arrival. The aim is to utilize a time difference deviation prediction model to output a predicted time difference deviation, thereby determining the target correction time difference based on the predicted time difference deviation. This eliminates the impact of interference on UWB-TDoA positioning and simultaneously solves the problem of large CIR data dimensionality, achieving more efficient and accurate UWB-TDoA positioning.

[0029] The following describes the positioning method based on time difference of arrival using some specific embodiments.

[0030] Figure 1 A flowchart illustrating a positioning method based on time difference of arrival (TDOA) according to an embodiment of this application is shown. Exemplarily, the positioning method based on TDOA includes the following steps: S110, calculate the online differential signal strength and online time difference of arrival of each non-reference base station relative to the reference base station.

[0031] In this embodiment, the base station is a fixed UWB transceiver device pre-deployed at a known three-dimensional coordinate location to receive UWB signals. Multiple base stations are deployed at the known three-dimensional coordinate location, with one base station serving as a reference base station and the others serving as non-reference base stations.

[0032] For example, the online differential signal strength is the difference between the signal strength of each non-reference base station and the signal strength of the reference base station during the online positioning phase, and the online time difference of arrival is the difference between the signal arrival time of each non-reference base station and the signal arrival time of the reference base station during the online positioning phase.

[0033] In one embodiment, such as Figure 2 As shown, obtaining the online differential signal strength and online time difference of arrival includes the following steps: S111, real-time acquisition of the online signal strength of the reference base station and each non-reference base station when receiving signals, and real-time acquisition of the online time difference of arrival of multiple non-reference base stations relative to the reference base station when receiving signals.

[0034] In an exemplary manner, during the online positioning phase, multiple base stations receive pulse signals transmitted by the target node in real time, acquire the online signal strength and online timestamp of the received signal at each base station, and calculate the difference between the online timestamp of each non-reference base station and the online timestamp of the reference base station to obtain the online time difference of arrival. Since the differences in arrival times and signal strengths among the various base stations are significant, these differences can be reduced through preliminary static calibration.

[0035] S112, statically calibrate the online signal strength and online time difference of arrival respectively to obtain the online calibration signal strength and online calibration time difference of arrival.

[0036] In one embodiment, the average signal strength of each non-reference base station and reference base station during the static calibration phase is obtained, and the difference between the online signal strength and the average signal strength is calculated to obtain the online calibration signal strength.

[0037] Exemplary, setting the first The online signal strength sequences collected by each base station during the calibration phase are as follows: ; In the formula, Indicates the first The online signal strength sequence of each base station, Indicates the first step in the calibration phase The signal strength of the first data collected by each base station, Indicates the first The signal strength collected by the base station a second time Indicates the first The first base station The signal strength acquired in the second acquisition Indicates the number of samples. Here is a positive integer representing the base station number. The average signal strength of each base station can be obtained by averaging the received signal strength. The formula for calculating the average signal strength is: ; In the formula, Indicates the first The average signal strength collected by each positioning base station Indicates the first The first positioning base station The strength of the positioning signal received this time.

[0038] As an example, the average signal strength is subtracted from the online signal strength acquired at any given time to obtain the calibrated online calibration signal strength. The formula for calculating the online calibration signal strength is: ; In the formula, Indicates the first A positioning base station in The online calibration signal strength is collected in real time. Indicates the online positioning stage A positioning base station in The online signal strength collected at all times.

[0039] In another embodiment, the average time difference of arrival (TDOA) when multiple non-reference base stations receive signals relative to the reference base station is obtained; the difference between the online TDOA and the average TDOA is calculated to obtain the online calibration TDOA.

[0040] Exemplary, and similarly, static calibration is performed on the signal arrival time difference between the reference base station and the non-reference base station. Let's assume that during the calibration phase... The arrival time difference between the non-reference base stations and the reference base station is: The average time difference of arrival (MTA) can be obtained by averaging the values ​​during the calibration phase. The formula for calculating the MTA is: ; In the formula, Indicates the first The average time difference of arrival of each non-reference base station relative to the reference base station Indicates the first step in the calibration phase The first non-reference base station The arrival time difference relative to the reference base station is used to obtain the online calibration arrival time difference after removing system bias. The formula for calculating the online calibration arrival time difference is: ; Indicates in Time of the first The online calibration arrival time difference of each non-reference base station relative to the reference base station. Indicates the online positioning phase Time of the first The online arrival time difference between a non-reference base station and the reference base station.

[0041] In this embodiment, the average signal strength and average time difference of arrival of each base station are used as a benchmark. The average value is subtracted from the online signal strength and online time difference of arrival, which can effectively eliminate systematic deviations such as base station hardware delay and clock synchronization error, and reduce the fluctuation between each online time difference of arrival and online signal strength.

[0042] S113, calculate the difference between the online calibration signal strength of each non-reference base station and the online calibration signal strength of the reference base station to obtain the online initial differential signal strength.

[0043] As an example, since signal strength has spatial correlation between different base stations, in order to enhance the sensitivity of features to changes in target location, this application further constructs differential signal strength. The differential operation suppresses common interference components while preserving spatial difference information, thereby improving the robustness of input features to environmental changes.

[0044] In one embodiment, the formula for calculating the initial differential signal strength online is: ; In the formula, Indicates in Time of the first The online initial differential signal strength of each non-reference base station relative to the reference base station. Indicates in Time of the first Online calibration signal strength of non-reference base stations, Indicates in The online calibration signal strength of the base station is constantly referenced.

[0045] This constitutes the online initial differential signal intensity sequence: ; In the formula, Indicates in The online initial differential signal intensity sequence at time 10:00. Indicates in The initial online differential signal strength of the first non-reference base station relative to the reference base station at time step [time value]. Indicates in The initial online differential signal strength of the second non-reference base station relative to the reference base station at time [time]. Indicates in Time of the first The online initial differential signal strength of each non-reference base station relative to the reference base station.

[0046] S114: After smoothing and filtering the online initial differential signal strength and the online calibration arrival time difference, the online differential signal strength and the online arrival time difference are obtained.

[0047] In one embodiment, to suppress high-frequency random noise and transient pulse interference while highlighting the main characteristics of signal changes, smoothing filters are applied to the online initial differential signal strength and the online calibration arrival time difference, respectively: ; ; In the formula, Indicates in At any given time, the online differential signal strength after smoothing and filtering. This indicates the time range for the convolution kernel to traverse. It signifies a historical moment. These are the kernel function weights, used to weight the time difference. It is mapped to a weight value to control the contribution ratio of each sampling point in the neighborhood to the current output. Indicates in The online initial differential signal intensity sequence at time 1. Indicates in At any given moment, the arrival time difference deviation after smoothing filtering. Indicates in The online calibration arrival time difference sequence includes... The time difference of arrival of all online calibrations at any given moment.

[0048] This application uses a Gaussian kernel function for filtering, and its weights are defined as follows: ; In the formula, Indicates the time difference, i.e. , The kernel width parameter controls the width of the Gaussian bell curve, thus determining the degree of smoothness. This is achieved through a Gaussian kernel function. For the current time t, around Within range and Perform a weighted average to obtain the smoothed output. and It can effectively suppress high-frequency noise and transient pulse interference while preserving the main variation characteristics of the data.

[0049] In this embodiment, a sliding window smoothing filter is introduced to preprocess the online initial differential signal strength and the online calibration arrival time difference. On the one hand, high-frequency random noise and transient pulse interference are effectively suppressed, avoiding noise spikes from interfering with model training. On the other hand, since the target node moves continuously during actual positioning, the time difference exhibits a slow change characteristic over time. The smoothing filter can fully utilize this temporal continuity, making the filtered data more realistically reflect the propagation path and environmental change trends, rather than instantaneous noise disturbances. Therefore, the jitter in the model input features is weakened, and the trend structure is highlighted, providing a more physically meaningful input expression for the subsequent time difference deviation prediction model to extract stable temporal correlation features.

[0050] S120, the online differential signal strength is input into the pre-trained time difference deviation prediction model to obtain the predicted time difference deviation output by the time difference deviation prediction model; wherein, the time difference deviation prediction model is trained based on the offline differential signal strength and arrival time difference deviation of the non-reference base station relative to the reference base station.

[0051] In one embodiment, the offline differential signal strength and time difference of arrival deviation of multiple non-reference base stations relative to the reference base station are obtained to obtain a model training dataset; a regression model is trained based on the model training dataset to obtain a time difference deviation prediction model.

[0052] For example, the time difference of arrival (TDOA) is the deviation of the difference between the signal arrival time of each non-reference base station and the signal arrival time of the reference base station from the theoretical time difference of arrival. The model training dataset includes all offline differential signal strengths and TDOAs at different times during the offline phase.

[0053] In one implementation, such as Figure 3 As shown, the process of constructing the model training dataset includes the following steps: S121, obtain the offline signal strength of the reference base station and each non-reference base station when receiving signals, and obtain the offline time difference of arrival of each non-reference base station relative to the reference base station when receiving signals.

[0054] As an example, during the offline training phase, the target node remains stationary at a known fixed location for a period of time, and the offline signal strength and offline timestamp of the signals received by each base station are collected. The difference between the offline timestamp of each non-reference base station and the offline timestamp of the reference base station is calculated to obtain the offline arrival time difference.

[0055] S122, perform static calibration on the offline signal strength and offline time difference of arrival respectively to obtain the offline calibration signal strength and offline calibration time difference of arrival.

[0056] In one embodiment, the average signal strength of each non-reference base station and the reference base station during the static calibration phase is obtained, and the average time difference of arrival of each non-reference base station relative to the reference base station during the static calibration phase is obtained; the difference between the offline signal strength and the average signal strength is calculated to obtain the offline calibration signal strength; the difference between the offline time difference of arrival and the average time difference of arrival is calculated to obtain the offline calibration time difference of arrival.

[0057] It is understandable that the calculation methods for offline calibration signal strength and offline calibration arrival time difference are the same as those for online calibration signal strength and online calibration arrival time difference, and will not be elaborated here.

[0058] S123, calculate the difference between the offline calibration signal strength of each non-reference base station and the offline calibration signal strength of the reference base station to obtain the initial offline differential signal strength.

[0059] In this embodiment, the offline initial differential signal strength is the difference in signal strength between the non-reference base station and the reference base station during the offline training phase. The calculation method for the offline initial differential signal strength is the same as that for the online initial differential signal strength.

[0060] S124, calculate the difference between the arrival time difference of each offline calibration and the theoretical arrival time difference to obtain the initial arrival time difference deviation.

[0061] For example, the theoretical time difference of arrival (TDOA) refers to the time difference of arrival of each non-reference base station relative to the reference base station under ideal error-free conditions. The initial TDOA deviation is obtained by calculating the difference between the offline calibration TDOA and the theoretical TDOA. By using the theoretical TDOA as a benchmark, the TDOA deviation is used as a modelable initial deviation, providing a supervisory label for subsequent learning of its nonlinear mapping relationship using differential signal strength-driven models.

[0062] S125: After smoothing and filtering the offline initial differential signal strength and the initial arrival time difference deviation, the offline differential signal strength and arrival time difference deviation are obtained.

[0063] In this embodiment, the online positioning stage and offline training have similar preprocessing operations. The filtering process for the offline initial differential signal strength is the same as the filtering process for the online initial differential signal strength described above. The filtering process for the initial arrival time difference deviation is the same as the filtering process for the online calibration arrival time difference described above, and will not be repeated here.

[0064] Exemplary, the regression model is a multi-input temporal convolutional network (MTCN), and the time difference deviation prediction model is the trained regression model used to establish a nonlinear mapping relationship between differential signal strength and arrival time difference deviation. To better capture this mapping relationship, this application also integrates a convolutional block attention module (CBAM) to construct the time difference deviation prediction model to capture the data correlation between base stations.

[0065] In one embodiment, during model training, the non-reference base station and the reference base station constitute an offline differential signal strength matrix, and a sliding time window of length L is used to extract joint temporal features. Therefore, the single-step model input tensor is defined as: ; In the formula, The core input data represents a three-dimensional tensor, which is a set of offline differential signal intensities that are constructed and used as input to a model (such as a neural network regression model). Tensor The dimension, R represents the real number field, offline dimension This indicates the number of differential base station channels. It represents the total number of base stations participating in the positioning system, with one base station selected as the reference base station, and the remaining... Each non-reference base station forms a differential pair with the reference base station; therefore... The dimension represents the number of independent base station pairs involved in constructing the input data; each base station pair corresponds to one differential base station channel. The online dimension L corresponds to the time dimension, indicating the length of a continuous time segment on the time axis used to construct the input offline differential signal strength sample. The third dimension... This represents the offline differential signal strength value for each channel, indicating the signal strength of each differential base station channel at each time point. This structure simultaneously preserves the joint feature representation of both the inter-base station difference dimension and the temporal evolution dimension.

[0066] The corresponding supervision labels are the arrival time difference deviation scalar matrix under the same window: ; In the formula, This represents the supervised label data, which is the correct answer that the model needs to learn during training, i.e., the arrival time difference bias. Represents the dimension of label y. R represents the real number field with shape [formula missing]. Offline maintenance Indicates the number of differential base station channels, online maintenance This indicates that for each differential base station channel, the corresponding label is the time difference of arrival deviation.

[0067] In this embodiment, the input and tags exist They are one-to-one correspondences in terms of dimensions. Provided Each channel is in Historical differential signal strength values ​​at each time point, Then it provides The model's task is to learn the arrival time difference deviation of each channel within the same time window. The offline differential signal strength, to The mapping relationship between arrival time difference deviation.

[0068] In one embodiment, the main body of the time difference deviation prediction model is composed of cascaded multi-layer two-dimensional causal dilated TCN residual blocks. Assuming a total of H stacked two-dimensional TCN residual blocks, the model's receptive field expands exponentially with the number of layers, allowing the network to cover the entire time window range with a relatively small number of layers, fully exploring the slow variation of arrival time difference deviation over time and cross-base station correlation patterns. Each residual block includes: (a) A two-dimensional causal dilated convolutional layer employs a causal convolutional structure in the time dimension to ensure that the output at any given time depends only on the current and historical inputs, thus avoiding the leakage of future information; a dilation coefficient is set in the time dimension. The kernel size is ,in, Indicates the first In convolutional operations, the dilation rate applied over time is... Indicates the kernel width of the time dimension. This represents the kernel width of the base station dimension, thereby enabling the joint extraction of cross-base station correlation features and long-term time-series dependency features.

[0069] (b) Batch normalization layer; (c) ReLU activation layer; (d) Dropout regularization layer; (e) Residual skip connection structure, used to add input features to convolutional output to alleviate the gradient vanishing problem in deep network training.

[0070] In one implementation, the convolutional attention module is embedded as follows: a convolutional attention module is introduced after the output of each two-dimensional TCN residual block. This module consists of a cascaded Channel Attention Mechanism (CAM) and Spatial Attention Mechanism (SAM). Channel Attention: Global average pooling and max pooling are performed on the two-dimensional features respectively, and differential channel weight coefficients for each base station are generated via a shared multilayer perceptron, enabling the model to adaptively highlight base station combinations that have a more significant impact on error. Spatial Attention: Average pooling and max pooling are performed on the channel-weighted features along the channel dimension, and time-base station joint attention weights are generated via two-dimensional convolution to highlight key time segments and key base station interaction areas.

[0071] In this embodiment, through the convolutional attention module, the time difference deviation prediction model can adaptively allocate attention resources in the joint time-base station feature space, suppress noise interference areas, and enhance feature responses that are strongly correlated with arrival time difference deviation.

[0072] In another embodiment, the regression output layer of the time difference deviation prediction model is used to output the final predicted time difference deviation. After feature extraction by multiple TCN-CBAM layers, the output features of the regression output layer are compressed into a spatially dimension-independent global feature vector by a global average pooling layer, and then the predicted time difference deviation is obtained through a fully connected layer. The output formula for the predicted time difference deviation is: ; In the formula, This indicates the prediction time difference bias in the model output. This represents the time difference deviation prediction model.

[0073] In one implementation, the model training process uses Mean Absolute Error (MAE) as the training loss function. This loss function is more robust to non-Gaussian noise and anomalous perturbations in the data, making it suitable for regression tasks. An early stopping mechanism is introduced during training. Training is automatically terminated when the validation set loss in the model training data does not decrease further within a preset number of consecutive rounds, in order to prevent overfitting and improve training efficiency.

[0074] In this embodiment, a multi-input temporal convolutional neural network combined with an attention mechanism is proposed to model the temporal nonlinear relationship between differential signal strength and time difference of arrival (TDOA). This enables the model to adaptively highlight base station channels and key time segments that have a more significant impact on the error, thereby enhancing its ability to express dynamic environmental disturbances such as multipath abrupt changes and occlusion variations. Simultaneously, the use of a mean absolute error loss function and an early stopping training strategy makes the model more robust to anomalous samples and improves its generalization ability, thus enhancing the reliability of online TDOA prediction.

[0075] In one implementation, such as Figure 4 The diagram illustrates the offline training process for the time difference deviation prediction model. In the data preprocessing stage, the offline signal strength is statically calibrated to obtain the offline calibrated signal strength, and the offline time difference of arrival (TDOA) is statically calibrated to obtain the offline calibrated TDOA. After differential construction (calculating the difference between the offline calibrated signal strength of the non-reference base station and the offline calibrated signal strength of the reference base station), the initial offline differential signal strength is obtained. Then, after deviation construction (calculating the deviation between the offline calibrated TDOA and the theoretical TDOA), the initial TDOA deviation is obtained. Finally, after smoothing and filtering, the obtained offline differential signal strength and TDOA deviation are used for model training to obtain the final time difference deviation prediction model.

[0076] In one implementation, the smoothed and filtered online differential signal strength is input into a pre-trained time difference deviation prediction model for online time difference of arrival (TDOA) correction. The model mapping relationship is as follows: ; In the formula, exist Prediction time difference deviation at time, Indicates in The online differential signal strength sequence at time t, including all smoothed and filtered online differential signal strengths. The predicted time difference deviation output by the time difference deviation prediction model represents the arrival time difference error after environmental adaptive correction.

[0077] In this embodiment, online differential signal strength features are constructed based on a reference base station, transforming the input from absolute signal quantity into a relative difference between base stations. This reduces common interference and common-mode noise while retaining the sensitivity of measurement deviations to changes in the propagation environment and significantly reducing the dimensionality of the input data. Therefore, compared to modeling methods based on high-dimensional CIR data, this application maintains a high sensitivity to environmental disturbances with lower computational complexity, meeting the requirements of real-time positioning systems for low latency and low power consumption.

[0078] S130, based on the predicted time difference deviation, the online arrival time difference is corrected to obtain the target corrected time difference, and the positioning position of the target node is determined according to the target corrected time difference.

[0079] In one embodiment, the difference between the online arrival time difference and the predicted time difference is calculated to obtain the target corrected time difference. In this embodiment, the smoothed and filtered online arrival time difference is corrected online, and the target corrected time difference after online correction is: ; In the formula, Indicates the time difference for target correction. This indicates the time difference of arrival online.

[0080] Exemplary, such as Figure 5 As shown, this is a comparison chart of the online arrival time difference error before and after correction. Before the online arrival time difference correction, the online arrival time difference error fluctuated significantly. After the real-time time difference correction, the online arrival time difference error fluctuated less, indicating that the target corrected time difference is more accurate after correction.

[0081] In this embodiment, as Figure 6 The diagram illustrates the process of online positioning. After static calibration, the online signal strength is obtained as the online calibration signal strength. This is then used for differential construction, specifically calculating the difference between the online calibration signal strength of the non-reference base station and the reference base station, to obtain the initial online differential signal strength. The online time difference of arrival (TDoA) is obtained through static calibration to obtain the online calibration arrival time difference. During model correction, the initial online differential signal strength, after smoothing and filtering, is input into the model, and the predicted time difference deviation is output. During TDoA correction, the difference between the smoothed online arrival time difference and the predicted time difference deviation is calculated to obtain the target correction time difference. Finally, in hyperbolic positioning, the target node's location is determined based on the target correction time difference.

[0082] In one embodiment, such as Figure 7 As shown, determining the location of the target node based on the target correction time difference includes the following steps: S131. Based on the target correction time difference, the reference base station and the base station coordinates of each non-reference base station, multiple hyperbolic positioning constraint equations are established. Each hyperbolic positioning constraint equation is used to characterize the difference between the distance from the positioning location to the non-reference base station and the distance to the reference base station, which is equal to the product of the speed of light and the target correction time difference.

[0083] As an example, after obtaining the corrected target time difference, a hyperbolic positioning constraint equation is constructed to solve for the target position. This case uses the Chan-Taylor joint positioning algorithm to solve for the target position. This method combines the closed-form fast initial value solution capability of the Chan algorithm with the high-precision optimization capability of Taylor iteration, which can significantly improve the convergence speed and stability of hyperbolic positioning.

[0084] Let the coordinates of the reference base station be... , No. The coordinates of the non-reference base stations are The target node's location is Then we have the hyperbolic positioning constraint equation: ; In the formula, c represents the speed of light. Indicates the distance from the target node to the non-reference base station. geometric distance, This represents the geometric distance from the target node to the reference base station.

[0085] S132 transforms the hyperbolic positioning constraint equation into a system of linear equations with the positioning position as the unknown, and obtains the initial positioning position of the target node by analyzing the system of linear equations.

[0086] In this embodiment, squaring the hyperbolic positioning constraint equation above, substituting it into the base station coordinate values, and rearranging the equations yields a linear system of equations: ; In the formula, A is the observation matrix, and b is the residual, which can be expressed by the base station coordinates and the target correction time difference, respectively: , Unknown quantity Represented as , This represents the straight-line distance from the target node to the reference base station.

[0087] In one embodiment, the least squares method is used to solve the linear equation system to obtain the closed-form initial value solution of the target node's location, i.e., the initial location. The expression for calculating the initial location is: ; In the formula, It is a four-dimensional column vector. The first three dimensions represent the estimated coordinates of the target node's initial location in three-dimensional space.

[0088] S133, based on the initial positioning position, iteratively solve the distance difference function between the positioning position and each non-reference base station and reference base station to obtain the position correction amount of the positioning position.

[0089] In one implementation, using the initial positioning position as the initial value for iteration, an analytical first-order Taylor expansion of the distance difference function is performed to obtain the Jacobian linearization formula. Exemplarily, the estimated coordinates of the initial positioning position are... That is, the initial value of Taylor's iteration, for the distance difference function at the estimated point. Performing a first-order Taylor expansion yields the Jacobian linearization formula. The expression for the distance difference function is: ; In the formula, Indicates the distance from the target node to the non-reference base station. Distance difference from the reference base station.

[0090] The expression for the Jacobi linearization formula is: ; In the formula, J is the Jacobian matrix. Initial positioning position Relative to the actual location Positional error. Indicates based on initial positioning position The calculated theoretical distance difference The distance difference function represents the distance difference function in The gradient matrix at that point.

[0091] In another implementation, the position correction is obtained by solving the Jacobian linearization formula, and the position is corrected based on the position correction to obtain a position estimate. In each iteration, the Jacobian linearization formula is solved again based on the position estimate, and the position estimate is updated; until the distance between the position corrections obtained in two adjacent iterations is less than a preset convergence threshold, the position is output.

[0092] In this embodiment, the Jacobian linearization formula is solved using the least squares method, and the position estimate is iteratively updated: ; In the formula, This represents the position correction amount, i.e., the initial positioning position. Relative to the actual location Positional error.

[0093] S134, the positioning position is corrected based on the position correction amount to obtain the estimated positioning position value until the distance between the position correction amounts obtained in two adjacent iterations is less than the preset convergence threshold, and the estimated positioning position value is used as the positioning position.

[0094] In one embodiment, the first The expression for the estimated location value obtained in the next iteration is: ; Indicates the first The estimated location after the next iteration. Indicates the first The estimated location after the next iteration.

[0095] In this embodiment, a preset convergence threshold is used as a control parameter to determine the termination of the iteration, representing the minimum distance difference between the position corrections obtained from two adjacent iterations. The iterations are repeated until the distance between the position corrections obtained from two adjacent iterations is less than the preset convergence threshold; at this point, the estimated location calculated in the current iteration is taken as the final determined location.

[0096] Exemplary, such as Figure 8As shown, the images compare the positioning results before and after online time-of-arrival (TOA) correction. Gray dots represent the positioning locations, and dark blue lines represent the actual movement trajectory of the target node. Image A shows the positioning results before TOA correction, with a calculated positioning error of 6.9 cm. Image B shows the positioning results after TOA correction, with a calculated positioning error of 1.8 cm. Therefore, TOA correction makes the positioning location more closely match the actual movement trajectory, thus improving positioning accuracy.

[0097] Figure 9 A schematic diagram of a positioning system 200 based on time difference of arrival according to an embodiment of this application is shown. Exemplarily, the positioning system 200 based on time difference of arrival includes: The online data acquisition module 210 is used to calculate the online differential signal strength and online time difference of arrival of multiple non-reference base stations relative to the reference base station; The online data acquisition module 210 is also used to acquire in real time the online signal strength of the reference base station and each non-reference base station when receiving signals, and to acquire in real time the online time difference of arrival of multiple non-reference base stations relative to the reference base station when receiving signals; to perform static calibration on the online signal strength and online calibration time difference to obtain the online calibration signal strength and online calibration time difference; to calculate the difference between the online calibration signal strength of each non-reference base station and the online calibration signal strength of the reference base station to obtain the online initial differential signal strength; and to perform smoothing filtering on the online initial differential signal strength and online calibration time difference to obtain the online differential signal strength and online time difference of arrival.

[0098] The deviation prediction module 220 is used to input the online differential signal strength into the pre-trained time difference deviation prediction model to obtain the predicted time difference deviation output by the time difference deviation prediction model; wherein, the time difference deviation prediction model is trained based on the offline differential signal strength and offline arrival time difference deviation of the non-reference base station relative to the reference base station. The time difference correction module 230 is used to correct the online arrival time difference based on the predicted time difference deviation to obtain the target corrected time difference; The positioning module 240 is used to determine the positioning position of the target node based on the target correction time difference.

[0099] It is understood that the system in this embodiment corresponds to the positioning method based on time difference of arrival in the above embodiments, and the options in the above embodiments are also applicable to this embodiment, so they will not be described again here.

[0100] This application also provides a terminal device, exemplary of which includes a processor and a memory, wherein the memory stores a computer program, and the processor executes the computer program to enable the terminal device to perform the functions of the various modules in the above-described positioning method based on time difference of arrival or the above-described positioning system based on time difference of arrival.

[0101] The processor can be an integrated circuit chip with signal processing capabilities. The processor can be a general-purpose processor, including at least one of the following: Central Processing Unit (CPU), Graphics Processing Unit (GPU), Network Processor (NP), Digital Signal Processor (DSP), Application Specific Integrated Circuit (ASIC), Field Programmable Gate Array (FPGA), or other programmable logic devices, discrete gate or transistor logic devices, or discrete hardware components. The general-purpose processor can be a microprocessor or any conventional processor, capable of implementing or executing the methods, steps, and logic block diagrams disclosed in the embodiments of this application.

[0102] Memory can be, but is not limited to, Random Access Memory (RAM), Read Only Memory (ROM), Programmable Read-Only Memory (PROM), Erasable Programmable Read-Only Memory (EPROM), and Electrically Erasable Programmable Read-Only Memory (EEPROM). Memory is used to store computer programs, and the processor can execute these programs upon receiving execution instructions.

[0103] This application also provides a computer-readable storage medium for storing computer programs used in the aforementioned terminal devices. For example, the computer-readable storage medium may include, but is not limited to, various media capable of storing program code, such as a USB flash drive, a portable hard drive, a read-only memory (ROM), a random access memory (RAM), a magnetic disk, or an optical disk.

[0104] In the several embodiments provided in this application, it should be understood that the disclosed systems and methods can also be implemented in other ways. The system embodiments described above are merely illustrative. For example, the flowcharts and block diagrams in the accompanying drawings show the architecture, functionality, and operation of possible implementations of systems, methods, and computer program products according to various embodiments of this application. In this regard, each block in a flowchart or block diagram may represent a module, segment, or portion of code, which contains one or more executable instructions for implementing a specified logical function. It should also be noted that, in alternative implementations, the functions marked in the blocks may occur in a different order than those marked in the drawings. For example, two consecutive blocks may actually be executed substantially in parallel, and they may sometimes be executed in reverse order, depending on the functions involved. It should also be noted that each block in the block diagram and / or flowchart, and combinations of blocks in the block diagram and / or flowchart, can be implemented using a dedicated hardware-based system that performs the specified function or action, or using a combination of dedicated hardware and computer instructions.

[0105] In addition, the functional modules or units in the various embodiments of this application can be integrated together to form an independent part, or each module can exist independently, or two or more modules can be integrated to form an independent part.

[0106] If a function is implemented as a software module and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, or part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a smartphone, personal computer, server, or network device, etc.) to execute all or part of the steps of the methods of the various embodiments of this application.

[0107] The above are merely specific embodiments of this application, but the scope of protection of this application is not limited thereto. Any changes or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in this application should be included within the scope of protection of this application.

Claims

1. A positioning method based on time difference of arrival, characterized in that, include: Calculate the online differential signal strength and online time difference of arrival of multiple non-reference base stations relative to the reference base station; The calculation of the online differential signal strength and online time-of-arrival difference of multiple non-reference base stations relative to the reference base station includes: The system acquires the online signal strength of the reference base station and each of the non-reference base stations when receiving signals in real time, and acquires the online time difference of arrival of the multiple non-reference base stations relative to the reference base station when receiving the signals in real time. The online signal strength and the online time difference of arrival are statically calibrated respectively to obtain the online calibration signal strength and the online calibration time difference of arrival. The difference between the online calibration signal strength of each non-reference base station and the online calibration signal strength of the reference base station is calculated to obtain the online initial differential signal strength. After smoothing and filtering the online initial differential signal strength and the online calibration arrival time difference, the online differential signal strength and the online arrival time difference are obtained respectively. The online differential signal strength is input into a pre-trained time difference deviation prediction model to obtain the predicted time difference deviation output by the time difference deviation prediction model; wherein, the time difference deviation prediction model is trained based on the offline differential signal strength and arrival time difference deviation of the non-reference base station relative to the reference base station; The online arrival time difference is corrected based on the predicted time difference deviation to obtain the target corrected time difference, and the location of the target node is determined based on the target corrected time difference.

2. The positioning method based on time difference of arrival according to claim 1, characterized in that, The time difference deviation prediction model is trained in the following manner: The offline differential signal strength and time difference of arrival deviation of multiple non-reference base stations relative to the reference base station are obtained to obtain the model training dataset; The regression model is trained based on the training dataset of the model to obtain the time difference deviation prediction model.

3. The positioning method based on time difference of arrival according to claim 1, characterized in that, The step of statically calibrating the online signal strength and the online time difference of arrival to obtain the online calibration signal strength and the online calibration time difference of arrival includes: The average signal strength of each of the non-reference base stations and the reference base station during the static calibration phase is obtained, and the average time difference of arrival of the multiple non-reference base stations relative to the reference base station when receiving the signal is obtained; The difference between the online signal strength and the average signal strength is calculated to obtain the online calibration signal strength; The difference between the online arrival time difference and the average arrival time difference is calculated to obtain the online calibration arrival time difference.

4. The positioning method based on time difference of arrival according to claim 2, characterized in that, The step of obtaining the offline differential signal strength and time-of-arrival deviation of the multiple non-reference base stations relative to the reference base station includes: The offline signal strength of the reference base station and each of the non-reference base stations when receiving signals is obtained, and the offline time of arrival difference of each of the non-reference base stations relative to the reference base station when receiving the signal is obtained; The offline signal strength and the offline time difference of arrival are statically calibrated respectively to obtain the offline calibration signal strength and the offline calibration time difference of arrival. Calculate the difference between the offline calibration signal strength of each non-reference base station and the offline calibration signal strength of the reference base station to obtain the initial offline differential signal strength; Calculate the difference between the offline calibration arrival time difference and the theoretical arrival time difference for each of the above calculations to obtain the initial arrival time difference deviation; After smoothing and filtering the offline initial differential signal strength and the initial arrival time difference deviation, the offline differential signal strength and the arrival time difference deviation are obtained.

5. The positioning method based on time difference of arrival according to claim 4, characterized in that, The step of statically calibrating the offline signal strength and the offline time difference of arrival to obtain the offline calibrated signal strength and the offline calibrated time difference of arrival includes: The average signal strength of each non-reference base station and the reference base station during the static calibration phase is obtained, and the average time difference of arrival of each non-reference base station relative to the reference base station during the static calibration phase is obtained. The difference between the offline signal strength and the average signal strength is calculated to obtain the offline calibration signal strength; The difference between the offline arrival time difference and the average arrival time difference is calculated to obtain the offline calibration arrival time difference.

6. The positioning method based on time difference of arrival according to claim 1, characterized in that, The step of correcting the online arrival time difference based on the predicted time difference deviation includes: The difference between the online arrival time difference and the predicted time difference is calculated to obtain the target corrected time difference.

7. The positioning method based on time difference of arrival according to claim 1, characterized in that, The step of determining the location of the target node based on the target correction time difference includes: Based on the target correction time difference, the reference base station, and the base station coordinates of each non-reference base station, multiple hyperbolic positioning constraint equations are established. Each hyperbolic positioning constraint equation is used to characterize the difference between the distance from the positioning location to the non-reference base station and the distance to the reference base station, which is equal to the product of the speed of light and the target correction time difference. The hyperbolic positioning constraint equation is converted into a system of linear equations with the positioning position as the unknown. The initial positioning position of the target node is obtained by analyzing the system of linear equations. Based on the initial positioning position, the distance difference function between the positioning position and each of the non-reference base stations and the reference base station is iteratively solved to obtain the position correction amount of the positioning position; The location is corrected based on the location correction amount to obtain a location estimate, until the distance between the location correction amounts obtained in two adjacent iterations is less than a preset convergence threshold, and the location estimate is taken as the location.

8. A positioning system based on time difference of arrival, characterized in that, include: The online data acquisition module is used to calculate the online differential signal strength and online time difference of arrival of multiple non-reference base stations relative to the reference base station. The online data acquisition module is also used to acquire in real time the online signal strength of the reference base station and each of the non-reference base stations when receiving the signal, and to acquire in real time the online time difference of arrival of the multiple non-reference base stations relative to the reference base station when receiving the signal; The online signal strength and the online time difference of arrival are statically calibrated to obtain the online calibration signal strength and the online calibration time difference of arrival. The difference between the online calibration signal strength of each non-reference base station and the online calibration signal strength of the reference base station is calculated to obtain the online initial differential signal strength. The online initial differential signal strength and the online calibration time difference of arrival are then smoothed and filtered to obtain the online differential signal strength and the online time difference of arrival. The deviation prediction module is used to input the online differential signal strength into a pre-trained time difference deviation prediction model to obtain the predicted time difference deviation output by the time difference deviation prediction model; wherein, the time difference deviation prediction model is trained based on the offline differential signal strength and arrival time difference deviation of the non-reference base station relative to the reference base station; The time difference correction module is used to correct the online arrival time difference based on the predicted time difference deviation to obtain the target corrected time difference; The positioning module is used to determine the location of the target node based on the target correction time difference.

9. A terminal device, characterized in that, The terminal device includes a processor and a memory, the memory storing a computer program, and the processor executing the computer program to implement the positioning method based on time difference of arrival as described in any one of claims 1-7.

10. A computer-readable storage medium, characterized in that, The system contains a computer program that, when executed by a processor, implements the positioning method based on time difference of arrival as described in any one of claims 1-7.

Images