A GPS positioning and 4G communication service system for IoT devices

Abstract

The present invention belongs to the field of communication positioning technology, and specifically relates to a GPS positioning and 4G communication service system for an Internet of Things device, including: a data transceiver module for data communication between a device and a server, and bidirectional data transmission through a 4G communication base station; a channel environment classification module for classifying and identifying the LOS or NLOS channel environment in which the device is located; and a hybrid positioning module for dynamically selecting a base station ranging method or GPS positioning technology for positioning according to the classification and identification results of the channel environment classification module. The random forest algorithm is used to classify and identify the channel environment, so that the system can dynamically select a base station ranging method or GPS positioning technology to adapt to the LOS or NLOS channel environment in which the device is located, which solves the limitations of traditional positioning methods in complex environments, improves the accuracy of device positioning, and reduces the energy consumption of the device.

Description

Technical Field

[0001] The present invention belongs to the technical field of communication positioning, and in particular, relates to a GPS positioning and 4G communication service system for an Internet of Things device. Background Art

[0002] In the field of positioning and communication of IoT devices, accurate and efficient positioning technology has always been a key challenge. Traditional positioning methods mostly rely on a single positioning technology, such as UWB (ultra-wideband) positioning or GPS positioning, but in complex environments, such as industrial production, logistics and warehousing management, agriculture and animal husbandry, and other IoT intelligent operation scenarios, especially in complex scenarios where GPS satellite channels are interfered or line-of-sight channels are blocked, the accuracy of traditional positioning is also greatly reduced; traditional methods with a single positioning technology often cannot meet the multiple requirements of high precision and low energy consumption when facing these challenges.

[0003] In addition, devices that rely on GPS or UWB positioning technology need to be equipped with more signal receivers, which significantly increases the complexity of the equipment and significantly increases energy consumption. Especially for IoT devices that run for a long time, such as material transportation AGV vehicles or sports equipment such as drones, higher energy consumption is often an issue that cannot be ignored. Summary of the invention

[0004] In order to solve the technical problem that a single positioning technology in traditional positioning methods cannot meet the multiple requirements of high precision and low energy consumption, the present invention provides a GPS positioning and 4G communication service system for Internet of Things devices, which combines base station ranging and GPS positioning technology, improves positioning accuracy, reduces energy consumption, and enhances the overall performance of the system through real-time channel environment identification and optimization of base station ranging algorithm.

[0005] The purpose of the present invention can be achieved through the following technical solutions:

[0006] A GPS positioning and 4G communication service system for IoT devices, comprising a data transceiver module, a channel environment classification module and a hybrid positioning module;

[0007] The data transceiver module is used for data communication between the device and the server, and realizes two-way data transmission through the 4G communication base station;

[0008] The channel environment classification module is used to classify and identify the LOS or NLOS channel environment in which the device is located;

[0009] The hybrid positioning module is used to dynamically select the base station ranging method or GPS positioning technology for positioning according to the classification and identification results of the channel environment classification module.

[0010] Preferably, the channel environment classification module includes a positioning data acquisition unit for receiving a channel impulse response (CIR) signal of a 4G communication network;

[0011] A feature extraction unit, used to extract multiple statistical features from a channel impulse response (CIR) signal to form feature vector data;

[0012] Preferably, the statistical characteristics include at least three combinations of total energy (ε), kurtosis (κ), skewness (s), average delay spread (τ) and rise time (t) of the received signal.

[0013] An algorithm classification unit, for training a random forest (RF) algorithm based on feature vector data of a known LOS / NLOS channel environment by constructing a random forest (RF) algorithm; outputting a classification result to identify a LOS or NLOS channel environment;

[0014] Preferably, the specific implementation process of the algorithm classification unit includes:

[0015] Build and initialize the random forest (RF) model, set the model hyperparameters; select the classification decision tree as the base classifier; the classification decision tree uses the Gini coefficient to select the best feature split point; select the best features for training by minimizing the Gini coefficient;

[0016] For each set of feature vector data of a known LOS / NLOS channel environment, the corresponding channel environment is marked, i.e., LOS (marked as +1) or NLOS (marked as -1), thus forming a labeled vector data set;

[0017] Divide the labeled vector dataset into training set and test set; use the training set data to train the random forest model, and implement integrated learning of multiple decision tree classification processes through a voting mechanism;

[0018] Input the new feature vector data into the trained random forest model; output the classification result.

[0019] Preferably, the method further comprises: using the test set data to classify and identify the trained random forest model;

[0020] Evaluate model performance by classification accuracy; optimize model performance by adjusting feature selection or hyperparameters.

[0021] Preferably, the hybrid positioning module includes a base station ranging unit and a GPS positioning unit; the base station ranging unit is used in a LOS (line of sight) channel environment, through the linear propagation of signals between the device and the base station, and the angle of arrival (AOA) and time delay (TOA) of the ESPRIT algorithm are used to jointly calculate the position information;

[0022] The GPS positioning unit is used to parse and process the received GPS data in a NLOS (non-line-of-sight) channel environment to obtain location information;

[0023] Preferably, the specific implementation process of the base station ranging unit includes:

[0024] The received signal is sampled to obtain a discrete time series channel impulse response (CIR) signal;

[0025] Perform analog-to-digital conversion on the channel impulse response (CIR) signal and construct a data matrix for the ESPRIT algorithm;

[0026] Perform eigenvalue decomposition or singular value decomposition on the data matrix to obtain the signal subspace and noise subspace;

[0027] According to the signal subspace, two rotationally invariant subspaces are constructed;

[0028] The rotation invariance property is used to estimate the rotation factor between the two subspaces by the least square method; the rotation factor contains the information of the angle of arrival (AOA) and the time delay (TOA).

[0029] The angle of arrival (AOA) and time delay (TOA) of the LOS path are calculated from the rotation factors; the angle of arrival is extracted from the rotation factor using an inverse sine function, and the time delay is related to the phase of the rotation factor;

[0030] Based on the calculated arrival angle and delay of the LOS path, combined with the known position of the base station, the location of the device is calculated using geometric relationships.

[0031] Preferably, the specific implementation process of the GPS positioning unit includes:

[0032] According to receiving channel signals from multiple satellites, obtaining satellite timestamps and NMEA data frames;

[0033] Parse NMEA data frames to extract latitude and longitude information, speed, and azimuth information;

[0034] The parsed NMEA data frames are verified by timestamp to ensure that the timestamps of all data sources are consistent.

[0035] Preferably, it also includes a motion trajectory generation module, which is used to update the real-time motion trajectory of the device in real time by combining the location information obtained by the base station distance measurement module and the GPS positioning module.

[0036] Preferably, the specific implementation process of the motion trajectory generation module includes:

[0037] According to the dynamically selected positioning results, the position of the device is updated moment by moment to construct a set of motion trajectory points;

[0038] When switching between LOS and NLOS channel environments, the interpolation method is used to smooth the transition data;

[0039] In the LOS channel environment, the speed and direction are calculated based on the current and previous positions;

[0040] In the NLOS channel environment, the speed and direction analyzed from GPS data are directly used;

[0041] Generate real-time motion trajectory by graphically displaying the trajectory points, speed and direction information at each moment;

[0042] Preferably, the interpolation method uses linear interpolation, which calculates the value of the middle point in the linear relationship between two known data points to smoothly transition the data; the calculation formula is:

[0043] ;

[0044] In the formula, , are the coordinates of the interpolation point; (x1, y1) are the trajectory points before the transition; (x2, y2) are the trajectory points after the transition; express The ratio in the interval [x1,x2].

[0045] Beneficial effects of the present invention:

[0046] 1. The random forest algorithm is used to classify and identify the channel environment, so that the system can dynamically select the base station ranging method or GPS positioning technology to adapt to the LOS (line-of-sight) or NLOS (non-line-of-sight) channel environment where the device is located, solving the limitations of traditional positioning methods in complex environments, improving the accuracy of device positioning and reducing the energy consumption of the device.

[0047] 2. The application of ESPRIT algorithm in base station ranging enhances the accuracy of signal processing. Combined with rotational invariance and least squares optimization methods, it makes device positioning more accurate, especially in mobile scenarios with high-frequency changes.

[0048] 3. The motion trajectory generation module can not only dynamically update the device position, but also automatically calculate and graphically display the real-time motion speed and direction tracking of the device based on the positioning results in different channel environments, and can provide comprehensive motion status monitoring for IoT devices. BRIEF DESCRIPTION OF THE DRAWINGS

[0049] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings required for describing the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other accompanying drawings can be obtained based on these accompanying drawings without paying creative work.

[0050] Figure 1 This is a schematic diagram of the structural framework of a GPS positioning and 4G communication service system for an Internet of Things device of the present invention.

[0051] Figure 2 This is a flow chart of the steps of a base station ranging unit in a GPS positioning and 4G communication service system of an Internet of Things device of the present invention. DETAILED DESCRIPTION

[0052] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0053] See also Figure 1-Figure 2 As shown, a GPS positioning and 4G communication service system for an Internet of Things device includes a data transceiver module, a channel environment classification module and a hybrid positioning module;

[0054] The data transceiver module is used for data communication between the device and the server, and realizes two-way data transmission through the 4G communication base station;

[0055] The channel environment classification module is used to classify and identify the LOS or NLOS channel environment in which the device is located;

[0056] The hybrid positioning module is used to dynamically select the base station ranging method or GPS positioning technology for positioning according to the classification and identification results of the channel environment classification module.

[0057] It should be noted that the present invention is based on the intelligent operation scenarios of the Internet of Things such as industrial production, logistics and warehousing management, agriculture and animal husbandry, especially in complex scenarios where the GPS satellite channel is interfered or the line-of-sight channel is blocked. Through the above-mentioned modular architecture, the system can dynamically adapt to the selection of the base station ranging method in the LOS environment or the GPS positioning technology in the NLOS environment to achieve high-precision hybrid positioning.

[0058] On the other hand, GPS positioning needs to receive satellite signals, requires a high-sensitivity receiver, and parses location data, and the power consumption required to maintain real-time positioning is relatively high. The base station ranging method relies on signal exchange between the device and the base station, and the power consumption is relatively low. In actual scenarios, the LOS and NLOS environments are dynamically adapted, and the dynamic selection method can avoid activating the base station ranging and GPS positioning modules at the same time. According to the specific power consumption indicators of the device, the energy saving effect can reach 10%-25%. This energy saving effect improves the endurance of the device; it is of great significance for battery-powered devices (such as material transportation AGV vehicles or sports equipment such as drones).

[0059] Preferably, the channel environment classification module includes a positioning data acquisition unit for receiving a channel impulse response (CIR) signal of a 4G communication network;

[0060] A feature extraction unit, used to extract multiple statistical features from a channel impulse response (CIR) signal to form feature vector data;

[0061] Preferably, the statistical characteristics include at least three combinations of total energy (ε), kurtosis (κ), skewness (s), average delay spread (τ) and rise time (t) of the received signal.

[0062] The algorithm classification unit is used to train the random forest (RF) algorithm based on the feature vector data of the known LOS / NLOS channel environment by constructing the random forest (RF) algorithm; output the classification result to identify the LOS or NLOS channel environment.

[0063] Specifically, the channel environment classification module is one of the core modules of this system. Its main function is to classify and identify the channel environment (LOS or NLOS) in which the device is located, so as to dynamically select the appropriate positioning technology (base station ranging or GPS positioning). Through the collaborative work of units such as positioning data acquisition, feature extraction and algorithm classification, the feature vector data is classified by constructing a random forest (RF) algorithm to identify the LOS or NLOS channel environment. Among them, the positioning data acquisition unit receives the channel impulse response (CIR) signal through the device's 4G communication network. The CIR signal contains the physical characteristics of the LOS and NLOS channel environments. It can be known from general knowledge that when a mobile device enters a LOS channel S environment, there is usually an obvious LOS path peak in the CIR signal, while in an NLOS environment, the multipath component of the signal is strong, mainly caused by reflection and scattering, and these physical characteristics are implicit in the signal frequency band. Then, a statistical method is used to calculate and extract characteristic indicators that can distinguish between LOS and NLOS channel environments. Statistical features include the following indicators:

[0064] Total received signal energy (ε): represents the power integral of the signal received in the channel, reflecting the signal strength characteristics of the channel. In an NLOS environment, the signal attenuation is large due to reflection or refraction, so the total received signal energy is small. The calculation formula is:

[0065] ; In the formula, is the amplitude of the i-th sampling point in the CIR signal.

[0066] Kurtosis (κ): It indicates the sharpness of the signal distribution. Since the signal amplitude in the NLOS environment is small, the kurtosis value in the NLOS environment is lower than that in the LOS environment. The calculation formula is:

[0067] ; In the formula, is the mean; is the standard deviation.

[0068] Skewness (s): Characterizes the asymmetry of signal distribution;

[0069] Average delay spread (τ): measures the average delay of the signal propagation path and describes the time spread characteristics of the channel. In an NLOS environment, since the signal undergoes multiple reflections, its average delay spread is usually greater than that in a LOS environment.

[0070] Rise time (t): A measure of the time interval between the strongest component and the first component in a signal. Under LOS conditions, the strongest component of a signal usually corresponds to the first component. Under NLOS conditions, the first component may be attenuated due to obstruction or strong diffraction, so that it usually precedes the strongest component. Therefore, the rise time in an NLOS environment is usually greater than the rise time in a LOS environment.

[0071] The above statistical features are calculated and used as input to the machine learning algorithm to distinguish different channel environments.

[0072] Preferably, the specific implementation process of the algorithm classification unit includes:

[0073] S1) Build and initialize the random forest (RF) model, set the model hyperparameters; select the classification decision tree as the base classifier; the classification decision tree uses the Gini coefficient to select the best feature split point; select the best features for training by minimizing the Gini coefficient;

[0074] In the specific implementation process, set the hyperparameters:

[0075] Number of decision trees (N): affects the stability and accuracy of classification and is usually set to dozens to hundreds.

[0076] Maximum depth: Limit the depth of the tree to prevent overfitting.

[0077] Minimum number of sample splits: The minimum number of samples required for each node.

[0078] Sample selection ratio: When training each tree, a certain proportion of samples are randomly selected to build subtrees.

[0079] Calculate the Gini coefficient of different split points for each feature and select the split point with the smallest Gini coefficient. Gini coefficient calculation formula: ; Where pi is the proportion of samples of the i-th category in the data set D.

[0080] S2) for each set of feature vector data of known LOS / NLOS channel environment, mark its corresponding channel environment, i.e., LOS (marked as +1) or NLOS (marked as -1), thus forming a labeled vector data set;

[0081] S3) Divide the labeled vector dataset into a training set and a test set; use the training set data to train the random forest model, and implement integrated learning of multiple decision tree classification processes through a voting mechanism;

[0082] In the specific implementation process, random sampling (repeatable) is performed from the training set to generate a sample subset for training a single decision tree. Each tree uses the Gini coefficient to split nodes until the stopping condition is met (such as the tree depth limit or the number of leaf node samples is less than the minimum value). The decision tree output is integrated using a voting mechanism: the classification results of all trees are voted on, the minority obeys the majority, and the final classification result is output.

[0083] S4) Input the new feature vector data into the trained random forest model; output the classification result.

[0084] It should be noted that random forest (RF) is a nonlinear classification algorithm based on decision tree integration, and the channel impulse response (CIR) signal characteristics (such as total received signal energy, kurtosis, skewness, etc.) show nonlinear distribution characteristics, and RF can capture the classification mode under different feature combinations through the splitting rules of multiple decision trees; it enables it to efficiently process the complex nonlinear relationship between LOS and NLOS features in the channel environment. In addition, the random forest algorithm in the present invention adopts a subsampling mechanism, which effectively avoids the excessive influence of noise on a single decision tree by randomly selecting data samples and features during the training process, and has a strong anti-interference ability. Finally, the decision mechanism of random forest can explain the decision basis of the model through feature importance scoring; in the LOS / NLOS environment classification task, feature importance analysis can help identify which statistical features (such as total received signal energy or average delay spread) have the greatest impact on the classification results, providing a basis for subsequent algorithm optimization.

[0085] Preferably, the method further comprises: using the test set data to classify and identify the trained random forest model;

[0086] Evaluate model performance by classification accuracy; optimize model performance by adjusting feature selection or hyperparameters.

[0087] In the specific implementation process, a portion of the data from the labeled vector data set is divided as the test set (usually accounting for 20%-30%), and the test set data does not participate in the model training. The classification accuracy of the statistical output results (Accuracy) is used to measure the overall prediction ability of the model. The calculation formula for classification accuracy is:

[0088] ;Wherein, TP: number of true positives (actual is LOS, predicted is LOS); TN: number of true negatives (actual is NLOS, predicted is NLOS); FP: number of false positives (actual is NLOS, predicted is LOS); FN: number of false negatives (actual is LOS, predicted is NLOS).

[0089] Preferably, the hybrid positioning module includes a base station ranging unit and a GPS positioning unit; the base station ranging unit is used in a LOS (line of sight) channel environment, and the angle of arrival (AOA) and time delay (TOA) of the ESPRIT algorithm are used to jointly calculate the position information through the linear propagation of the signal between the device and the base station; the GPS positioning unit is used in a NLOS (non-line of sight) channel environment, and the received GPS data is analyzed and processed to obtain the position information;

[0090] Specifically, the hybrid positioning module can adaptively adjust the positioning algorithm in different wireless channel environments to improve positioning accuracy. For example, in an environment with good line of sight, the base station ranging unit is used first, while in an environment with poor line of sight, GPS data is used for positioning.

[0091] In a LOS environment, the signal propagation path between the device and the base station can be approximated as a straight line, so the angle of arrival (AOA) and time delay (TOA) can be used as important bases for positioning calculations. The angle of arrival reflects the direction from which the signal arrives at the base station, while the time delay reflects the time difference in signal propagation. Through the joint calculation of the two, the distance and direction between the device and the base station can be calculated, thereby obtaining the spatial position of the device. The ESPRIT algorithm is an efficient parameter estimation method that is commonly used to locate signal sources. In this application, the ESPRIT algorithm can calculate the angle of arrival of the signal through the phase difference of the signal received by a single base station, thereby assisting in positioning. The advantage of this algorithm is that it can maintain high accuracy even when the signal is weak or there is interference.

[0092] In the NLOS (non-line-of-sight) channel environment, there are usually problems such as obstruction or reflection in the NLOS environment, which affect the propagation path of the base station signal; therefore, GPS data is used for positioning, and the received GPS signals are analyzed and processed to maximize the positioning accuracy. The GPS positioning unit can provide reliable location information in complex environments.

[0093] Preferably, the specific implementation process of the base station ranging unit includes:

[0094] The received signal is sampled to obtain a discrete time series channel impulse response (CIR) signal;

[0095] In the specific implementation process, 4G communication base stations communicate with mobile devices in two directions through antenna arrays; the received signals are usually processed by the RF front end and then enter the receiving module for time domain sampling. The sampling frequency needs to be selected according to the bandwidth requirements of the system, and the sampling frequency is usually selected to be greater than twice the signal bandwidth. This process involves converting the timing data of the wireless signal into discrete signals for processing; it is used to construct the channel impulse response (CIR), which is the time domain response of the signal propagation between the device and the base station.

[0096] The channel impulse response (CIR) signal is converted into digital form and a data matrix for the ESPRIT algorithm is constructed. The signals of multiple receiving antenna arrays are combined to form a matrix that reflects the spatial characteristics of the signal. The columns of the matrix represent the signals of different receiving antenna arrays, while the rows represent time or different time sampling points.

[0097] Perform eigenvalue decomposition or singular value decomposition on the data matrix to obtain the signal subspace and the noise subspace; the eigenvalue decomposition decomposes the data matrix into the signal subspace and the noise subspace by calculating the eigenvalue and eigenvector of the matrix. The part with a larger eigenvalue usually represents the main information of the signal, and the part with a smaller eigenvalue usually represents the noise.

[0098] According to the signal subspace, two rotationally invariant subspaces are constructed; specifically, these two subspaces represent the propagation characteristics of the signal in different directions or with different time delays. Rotational invariance is one of the core ideas of the ESPRIT algorithm. It is based on the fact that the rotation of the spatial array has the same effect on the signal, so the signal source can be located through the rotationally invariant subspace.

[0099] The rotation invariance property is utilized to estimate the rotation factor between the two subspaces by the least square method; the rotation factor contains the information of the angle of arrival (AOA) and the time delay (TOA); the least square method is an optimization method that can minimize the estimation error to obtain the most accurate rotation factor estimation.

[0100] The angle of arrival (AOA) and time delay (TOA) of the LOS path are calculated from the rotation factors; Angle of arrival (AOA): The angle of arrival is extracted from the rotation factors through the arcsine function. This step is based on the phase information of the received signal and uses the signal reception angle of the multi-antenna array to calculate the AOA.

[0101] Time delay (TOA): Time delay is the reflection of signal propagation time, which is related to the phase change of the rotation factor. The signal propagation delay is calculated by analyzing the phase of the signal and combining the signal propagation path.

[0102] Based on the calculated arrival angle and delay of the LOS path, combined with the known location of the base station, the location of the device is calculated using geometric relationships. A common method is to use triangulation positioning technology to use the arrival angle and delay of the signal to infer the precise location of the device relative to the base station.

[0103] In TOA positioning, the known position of the base station and the signal propagation delay are usually used, combined with the speed of light (i.e. the propagation speed of electromagnetic waves), to calculate the distance between the device and the base station through a geometric algorithm.

[0104] In AOA positioning, the coordinates of the device are calculated using triangulation based on the known location of the base station and the angle of arrival of the device signal.

[0105] The base station ranging unit uses the above steps to accurately calculate the position of the device in the LOS (line-of-sight) channel environment based on the ESPRIT algorithm. This process involves a variety of signal processing and algorithms such as signal sampling, analog-to-digital conversion, eigenvalue decomposition, rotational invariance analysis, and least squares optimization. Through the inference calculation of these algorithm steps, direct and high-precision device positioning is achieved.

[0106] Preferably, the specific implementation process of the GPS positioning unit includes:

[0107] According to the channel signals received from multiple satellites, the satellite timestamp and NMEA data frame are obtained; wherein the position of each satellite (xi, yi, zi) and the received device position (x, y, z) satisfy the pseudo-range formula:

[0108] ; In the formula, p i is the pseudorange of the ith satellite; c is the speed of light; Δt is the signal propagation time.

[0109] Parse NMEA data frames to extract latitude and longitude information, speed, and azimuth information;

[0110] The parsed NMEA data frames are verified by timestamp to ensure that the timestamps of all data sources are consistent.

[0111] Specifically, the data frame in NMEA format includes GPRMC (minimum positioning information) and GPGGA (global positioning system correction data). Extract GPRMC data to parse the speed and azimuth fields; extract GPGGA data to parse the longitude, latitude and altitude information for position calibration and comparison.

[0112] Preferably, it also includes a motion trajectory generation module, which is used to update the real-time motion trajectory of the device in real time by combining the location information obtained by the base station distance measurement module and the GPS positioning module.

[0113] Preferably, the specific implementation process of the motion trajectory generation module includes:

[0114] According to the dynamically selected positioning results, the position of the device is updated moment by moment to construct a set of motion trajectory points;

[0115] When switching between LOS and NLOS channel environments, the interpolation method is used to smooth the transition data;

[0116] In the LOS channel environment, the speed and direction are calculated based on the current and previous positions;

[0117] In the NLOS channel environment, the speed and direction analyzed from GPS data are directly used;

[0118] Generate real-time motion trajectory by graphically displaying the trajectory points, speed and direction information at each moment.

[0119] Specifically, the motion trajectory generation module first obtains the current location of the device in each time period based on the location information obtained from the base station ranging module and the GPS positioning module. The location information at each time point is recorded and added to the trajectory point set, gradually constructing the complete motion trajectory data.

[0120] In actual applications, the motion trajectory of the device may experience a switch from the LOS channel environment to the NLOS channel environment, or vice versa. At this time, the positioning information provided by the base station ranging module and the GPS positioning module may produce discontinuous or sudden jumps. In order to avoid the uneven motion trajectory, the motion trajectory generation module needs to use an interpolation method to smoothly transition. For example, linear interpolation is performed between two positioning information or other more complex interpolation methods (such as spline interpolation) are used to smoothly connect the trajectory data. The choice of interpolation method can be based on actual needs. For example, in a high dynamic environment, using a more accurate interpolation method can effectively avoid error accumulation.

[0121] The updated trajectory points (latitude and longitude or 3D coordinates), speed and direction information are displayed through a graphical interface or map display module. Different graphical tools, such as Web map API or custom geographic information system (GIS) modules, can be used to visualize the real-time motion trajectory.

[0122] Preferably, the interpolation method uses linear interpolation, which calculates the value of the middle point in the linear relationship between two known data points to smoothly transition the data; the calculation formula is:

[0123] ;

[0124] In the formula, , are the coordinates of the interpolation point; (x1, y1) are the trajectory points before the transition; (x2, y2) are the trajectory points after the transition; express The ratio in the interval [x1,x2].

[0125] The motion trajectory generation module provides an accurate and continuous device motion trajectory update method by combining the data of the base station ranging module and the GPS positioning module. The module can use the interpolation method to smoothly transition the data when the channel environment switches, ensuring seamless trajectory updates in different environments. At the same time, different positioning methods in LOS and NLOS environments are used to calculate the speed and direction of the device and generate a graphical motion trajectory in real time. Ultimately, a high-precision, smoothly transitioned device motion trajectory diagram can be provided to achieve real-time position tracking and dynamic monitoring.

[0126] The GPS positioning and 4G communication service system for IoT devices proposed in this invention combines the channel environment recognition of machine learning algorithms, dynamic positioning technology selection, accurate base station ranging and GPS positioning technology methods, and provides a high-precision and high-robust IoT device positioning and communication service system. The system can work flexibly in different channel environments, ensure the accuracy and real-time performance of positioning, and is suitable for a variety of complex environments, providing reliable positioning support for IoT applications.

[0127] The above contents are merely examples and explanations of the structure of the present invention. The technicians in this technical field may make various modifications or additions to the specific embodiments described or replace them in a similar manner. As long as they do not deviate from the structure of the invention or exceed the scope defined by the claims, they should all fall within the protection scope of the present invention.

Claims

1. A GPS positioning and 4G communication service system for IoT devices, characterized in that: It includes a data transceiver module, a channel environment classification module and a hybrid positioning module; The data transceiver module is used for data communication between the device and the server, and realizes two-way data transmission through the 4G communication base station; The channel environment classification module is used to classify and identify the LOS or NLOS channel environment in which the device is located; The channel environment classification module includes a positioning data acquisition unit for receiving a channel impulse response signal of a 4G communication network; A feature extraction unit, used to extract multiple statistical features from the channel impulse response signal to form feature vector data; An algorithm classification unit, for training the random forest algorithm based on feature vector data of known LOS and NLOS channel environments by constructing the random forest algorithm; and outputting classification results; The hybrid positioning module is used to dynamically select a base station ranging method or a GPS positioning technology for positioning according to the classification and identification results of the channel environment classification module; the hybrid positioning module includes a base station ranging unit and a GPS positioning unit; The base station ranging unit is used in a LOS channel environment to calculate the location information by using the arrival angle and delay of the ESPRIT algorithm through the linear propagation of the signal between the device and the base station. The specific implementation process of the base station ranging unit includes: The received signal is sampled to obtain a discrete time series channel impulse response signal; Perform analog-to-digital conversion on the channel impulse response signal and construct a data matrix for the ESPRIT algorithm; Perform eigenvalue decomposition or singular value decomposition on the data matrix to obtain the signal subspace and noise subspace; According to the signal subspace, two rotationally invariant subspaces are constructed; The rotation invariance property is used to estimate the rotation factor between the two subspaces by the least square method; the rotation factor contains information about the arrival angle and the delay. Calculate the arrival angle and delay of the LOS path from the rotation factors; Based on the calculated arrival angle and delay of the LOS path, combined with the known location of the base station, the location of the device is calculated using geometric relationships; The GPS positioning unit is used to parse and process the received GPS data in the NLOS channel environment to obtain the position information.

2. The GPS positioning and 4G communication service system for an Internet of Things device according to claim 1, characterized in that: In the feature extraction unit, the statistical features include at least three combinations of total energy, kurtosis, skewness, average delay spread and rise time of the received signal.

3. The GPS positioning and 4G communication service system for an Internet of Things device according to claim 1, characterized in that: The specific implementation process of the algorithm classification unit includes: Build and initialize the random forest model, set the model hyperparameters; select the classification decision tree as the base classifier; the classification decision tree uses the Gini coefficient to select the best feature split point; select the best features for training by minimizing the Gini coefficient; For each set of feature vector data of known LOS and NLOS channel environments, the corresponding channel environment is marked to form a labeled vector data set; Divide the labeled vector dataset into training set and test set; use the training set data to train the random forest model, and implement integrated learning of multiple decision tree classification processes through a voting mechanism; Input the new feature vector data into the trained random forest model; output the classification result.

4. The GPS positioning and 4G communication service system for an Internet of Things device according to claim 3, characterized in that: Also includes: Use the test set data to classify and identify the trained random forest model; Evaluate model performance by classification accuracy; optimize model performance by adjusting feature selection or hyperparameters.

5. The GPS positioning and 4G communication service system for an Internet of Things device according to claim 1, characterized in that: The specific implementation process of the GPS positioning unit includes: According to receiving channel signals from multiple satellites, obtaining satellite timestamps and NMEA data frames; Parse NMEA data frames to extract latitude and longitude information, speed, and azimuth information; The parsed NMEA data frames are verified by timestamp to ensure that the timestamps of all data sources are consistent.

6. The GPS positioning and 4G communication service system for an Internet of Things device according to claim 1, characterized in that: It also includes a motion trajectory generation module, which is used to update the real-time motion trajectory of the device in real time by combining the location information obtained by the base station distance measurement module and the GPS positioning module.

7. The GPS positioning and 4G communication service system for an Internet of Things device according to claim 6, characterized in that: The specific implementation process of the motion trajectory generation module includes: According to the dynamically selected positioning results, the position of the device is updated moment by moment to construct a set of motion trajectory points; When switching between LOS and NLOS channel environments, the interpolation method is used to smooth the transition data; In the LOS channel environment, the speed and direction are calculated based on the current and previous positions; In the NLOS channel environment, the speed and direction analyzed from GPS data are directly used; The trajectory points, speed and direction information at each moment are graphically generated into a real-time motion trajectory.

8. The GPS positioning and 4G communication service system for an Internet of Things device according to claim 7, characterized in that: The interpolation method uses linear interpolation, which calculates the value of the middle point in the linear relationship between two known data points to smooth the transition data; the calculation formula is: ； In the formula, , are the coordinates of the interpolation point; (x1, y1) are the trajectory points before the transition; (x2, y2) are the trajectory points after the transition; express The ratio in the interval [x1,x2].

Images