Positioning method and related device

Abstract

The invention provides a positioning method and a related device, and relates to the technical field of mobile communication, and the method comprises the steps: obtaining an intermediate positioning result, the intermediate positioning result comprises a first positioning result and a second positioning result, the first positioning result is obtained based on a first positioning mode, and the second positioning result is obtained based on a second positioning mode, the first positioning mode is different from the second positioning mode, the positioning result of the terminal is obtained based on the intermediate positioning result and a dynamic environment parameter, the dynamic environment parameter is used for configuring the weight of the intermediate positioning result, and the dynamic environment parameter represents at least one of the number of base stations associated with the terminal, the number of satellites associated with the terminal and the state of a sensor associated with the terminal; therefore, the positioning result of the terminal not only combines the results of multiple positioning modes, but also considers the number of base stations, the number of satellites and the state of the sensor associated with the terminal, thereby being beneficial to further improving the precision of the positioning result of the terminal, and improving the positioning technology.

Description

Technical Field

[0001] This application relates to the field of mobile communication technologies, and in particular, to a positioning method and related devices. Background Art

[0002] With the continuous innovation and wide popularization of communication technologies, positioning technologies based on mobile communication technologies have been further developed.

[0003] Currently, there is still room for improvement in positioning technologies based on mobile communication technologies. For example, the positioning accuracy in some scenarios needs to be improved. Summary of the Invention

[0004] This application provides a positioning method and related devices, aiming to improve the existing positioning method, such as improving the accuracy of positioning results. The disclosed technical solutions are as follows:

[0005] In a first aspect of this application, a positioning method is provided. In some implementation manners, this method can be executed by a network device, or can also be executed by components (such as circuits, chips, or chip systems, etc.) configured in the network device, and can also be implemented by a logic module or software that can implement all or part of the functions of the network device. In other implementation manners, this method can be executed by a terminal, or can also be executed by components (such as circuits, chips, or chip systems, etc.) configured in the terminal, and can also be implemented by a logic module or software that can implement all or part of the functions of the terminal. This application does not make any limitations in this regard.

[0006] The method includes: obtaining an intermediate positioning result, where the intermediate positioning result includes: a first positioning result and a second positioning result. The first positioning result is obtained based on a first positioning method, and the second positioning result is obtained based on a second positioning method. The first positioning method is different from the second positioning method. Based on the intermediate positioning result and dynamic environment parameters, a positioning result of the terminal is obtained. The dynamic environment parameters are used to configure the weight of the intermediate positioning result, and the dynamic environment parameters characterize at least one of the number of base stations associated with the terminal, the number of satellites associated with the terminal, and the status of sensors associated with the terminal.

[0007] Exemplarily, the first positioning result is related to at least one of the dynamic environment parameters. For example, the parameters used to obtain the first positioning result are obtained by means of at least one of a base station, a satellite, and a sensor. The second positioning result is related to at least one of the dynamic environment parameters. For example, the parameters used to obtain the second positioning result are obtained by means of at least one of a base station, a satellite, and a sensor. Therefore, the dynamic environment parameters can reflect the accuracy of the intermediate positioning result.

[0008] Since the first positioning method is different from the second positioning method, intermediate positioning results can be obtained from multiple dimensions. Also, since the positioning result of the terminal is obtained based on the intermediate positioning results, it is beneficial to improve the accuracy of the terminal's positioning result. Additionally, since the dynamic environment parameters are used to configure the weights of the intermediate positioning results, the positioning result of the terminal can also consider the number of base stations, satellites, and sensor status associated with the terminal, which is beneficial to further improve the accuracy of the terminal's positioning result, thus improving the positioning technology.

[0009] In some implementation manners, the process of obtaining the second positioning result based on the second positioning method includes: by invoking the first deep learning model, obtaining the position information output by the first deep learning model based on a parameter combination (also known as a comprehensive parameter combination), where the first deep learning model has an adaptation relationship with the parameter combination. Obtaining the second positioning result based on the deep learning model is beneficial to improving the accuracy of the second positioning result. Moreover, since the first deep learning model has an adaptation relationship with the parameter combination, the utilization rate of the first deep learning model can be improved. For example, the computing power of the first deep learning model is adapted to the parameter combination, and the accuracy of the second positioning result is further improved.

[0010] In some implementation manners, the first deep learning model having an adaptation relationship with the parameter combination includes: the first deep learning model having an adaptation relationship with at least one of the features (such as features containing time series data or data having a spatial relationship) and dimensions (such as the number of dimensions of the data) of the parameter combination. The adaptation relationship is used to select the first deep learning model based on the parameter combination, which can not only avoid wasting the resources of the first deep learning model but also improve the accuracy of the output result of the first deep learning model.

[0011] In some implementation manners, the first deep learning model having an adaptation relationship with the features of the parameter combination includes at least one of the following: a deep learning model of the long short-term memory network type is adapted to a parameter combination containing time series data; a deep learning model of the convolutional neural network type is adapted to a parameter combination containing a spatial relationship; in addition, the first target model in the first deep learning model is also adapted to the indoor and outdoor scenarios where the terminal is located, and the indoor and outdoor scenarios include: indoor, outdoor, from indoor to outdoor, or from outdoor to indoor. The adaptation relationships of the first deep learning model with different dimensions of the parameter combination can lay a foundation for more flexible model selection and also improve the universality of the positioning method for different scenarios.

[0012] In some implementations, the parameter combination includes the following data: the number of transceiver antennas of the terminal, the frequency band used by the terminal, the number of groups of channel state information (CSI) historical data obtained by the terminal, and the number of groups of received signal strength indication (RSSI) historical data. The method innovatively uses the above data combination for positioning, which is beneficial to improving the accuracy of the positioning result. Alternatively, the parameter combination includes at least one of the following data: satellite positioning data and sensor data. Satellite positioning data is more suitable for outdoor open areas and is vulnerable to influences such as occlusion. Sensor data can compensate for the problems caused by occlusion, which is beneficial to improving the accuracy of the positioning result.

[0013] In some implementations, before obtaining the intermediate positioning result, it further includes: downloading a first deep learning model from a model management platform, where the first deep learning model is selected by the model management platform from multiple models based on an adaptation relationship. The model management platform is beneficial for the configuration and training of various models, etc., so as to be able to efficiently process the models.

[0014] In some implementations, obtaining the positioning result of the terminal based on the intermediate positioning result and dynamic environment parameters includes: by invoking a second deep learning model, obtaining the positioning result output based on the intermediate positioning result and dynamic environment parameters. The second deep learning model includes a hidden layer, and the hidden layer is used to assign weights to the intermediate positioning result. Using a deep learning model to implement the assignment of weights, the deep learning model can fully learn the situation of the intermediate positioning result and weight configuration, which is beneficial to improving the accuracy of the positioning result of the terminal.

[0015] In some implementations, the second deep learning model is adapted to the processing result of the intermediate positioning result. The processing result includes at least one of the following: information on the positioning method for obtaining the intermediate positioning result (such as the number and type of information, etc.), and information on the intermediate positioning result obtained by each positioning method (such as the number of groups and features, etc.), so as to improve the accuracy of the output result of the second deep learning model and make full use of the resources of the second deep learning model.

[0016] In some implementations, before obtaining the positioning result of the terminal based on the intermediate positioning result and dynamic environment parameters, it further includes: downloading a second deep learning model from a model management platform, where the second deep learning model is selected by the model management platform from multiple models based on the adaptation relationship between the second deep learning model and the processing result.

[0017] In some implementations, after obtaining the positioning result of the terminal, it further includes: comparing the positioning result of the terminal with the first positioning result to obtain a first difference; based on the first difference and the first threshold, adjusting the configuration of the signal sent by the base station, where the signal is the basis for obtaining the first positioning result by the first positioning method. Adjusting the configuration of the signal sent by the base station is beneficial for the terminal to obtain a signal with better quality, thereby improving the accuracy of the first positioning result and further improving the accuracy of the positioning result of the terminal.

[0018] In some implementations, after obtaining the positioning result of the terminal, it further includes: comparing the positioning result of the terminal with the second positioning result to obtain a second difference; based on the second difference and the second threshold, retraining at least one of the first deep learning model for obtaining the second positioning result and the second deep learning model for obtaining the positioning result of the terminal, so as to improve the accuracy of the positioning result of the terminal when using the first deep learning model and the second deep learning model for positioning again.

[0019] The second aspect of the present application provides a positioning device, including modules for implementing the method provided by the first aspect of the present application.

[0020] The third aspect of the present application provides a positioning device, including: one or more processors, and a memory; the memory is used to store program codes; the processors are used to run the program codes so that the positioning device implements the method provided by the first aspect of the present application.

[0021] The fourth aspect of the present application provides a computer-readable storage medium, on which instructions are stored, and when the instructions are run on an electronic device, the electronic device is caused to execute the method provided by the first aspect of the present application.

[0022] The fifth aspect of the present application provides a computer program product, on which instructions are stored, and when the computer program product is run on an electronic device, the electronic device is caused to implement the method provided by the first aspect of the present application.

[0023] The sixth aspect of the present application provides a chip system, including: at least one processor and an interface, where the interface is used to receive code instructions and transmit them to at least one processor; the at least one processor runs the code instructions to implement the method provided by the first aspect of the present application. BRIEF DESCRIPTION OF THE DRAWINGS

[0024] In order to more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the following will briefly introduce the drawings required for use in the description of the embodiments or the prior art. Obviously, the following drawings are some embodiments of the present invention. For those of ordinary skill in the art, other drawings can be obtained based on these drawings without creative efforts.

[0025] Figure 1 It is an example diagram of a mobile communication system;

[0026] Figure 2 It is an example diagram of a communication system for implementing a positioning method provided by an embodiment of the present application;

[0027] Figure 3 It is an example diagram for obtaining a positioning result based on various models in a model management platform provided by an embodiment of the present application;

[0028] Figure 4 It is an example diagram of a meta-learner provided by an embodiment of the present application;

[0029] Figure 5 It is a flowchart of a positioning method provided by an embodiment of the present application;

[0030] Figure 6 It is a flowchart of another positioning method provided by an embodiment of the present application;

[0031] Figure 7 It is an example diagram of the composition of a positioning device provided by an embodiment of the present application;

[0032] Figure 8 It is an example diagram of the composition of another positioning device provided by an embodiment of the present application;

[0033] Figure 9 It is an example diagram of the composition of another positioning device provided by an embodiment of the present application. Detailed implementation manners

[0034] The terms "first", "second", "third", etc. in the specification, claims and drawings of the present application are used to distinguish different objects, rather than to limit a specific order.

[0035] In the embodiments of the present application, words such as "in some implementation manners" or "for example" are used to give examples, illustrations or explanations, and should not be construed as being more preferred or having more advantages than other embodiments or design solutions.

[0036] Figure 1 It is an example of a mobile communication system, and the mobile communication system includes a terminal, a base station, and a core network device.

[0037] The mobile communication system can be a second-generation (2G) communication system, a third-generation (3G) communication system, a long term evolution (LTE) system, a fifth-generation (5G) communication system, an LTE and 5G hybrid architecture, a 5G new radio (5G NR) system, or a new communication system emerging in future communication development, etc.

[0038] The terminal device can be in various forms. For example, a mobile phone, a tablet (Pad), a computer with wireless transceiver function, a virtual reality (VR) terminal device, an augmented reality (AR) terminal device, a wireless terminal device in industrial control, a vehicle-mounted terminal device, a wireless terminal device in self-driving, a wireless terminal device in remote medical, a wireless terminal device in smart grid, a wireless terminal device in transportation safety, a wireless terminal device in smart city, a wireless terminal device in smart home, a wearable terminal device, and so on. Sometimes, the terminal device can also be referred to as user equipment (UE), access terminal device, vehicle-mounted terminal device, industrial control terminal device, UE unit, UE station, mobile station, mobile unit, remote station, remote terminal device, mobile device, UE terminal device, wireless communication device, UE agent, or UE device, etc. The terminal device can also be a fixed terminal device or a mobile terminal device.

[0039] The access network device can be a terrestrial base station or a non-terrestrial network (NTN) device. The NTN device can also be referred to as a base station and / or a satellite access node (SAN).

[0040] A base station is any device located on the network side and having wireless transceiver functions, including but not limited to: evolved base stations in LTE (NodeB or eNB or e-NodeB, evolutional Node B), base stations (gNodeB or gNB) or transmission receiving points (TRP) in new radio (NR), base stations evolved by 3GPP subsequently, access nodes in Wi-Fi systems, wireless relay nodes, wireless backhaul nodes, etc. A base station can be: a macro base station, a micro base station, a pico base station, a small station, a relay station, or a balloon station, etc. A base station can include one or more co-located or non-co-located TRPs. A base station can also be a radio controller, a centralized unit (CU), and / or a distributed unit (DU) in the scenario of a cloud radio access network (CRAN). A base station can communicate with a terminal device or communicate with a terminal device through a relay station.

[0041] The core network device and the base station can be independent and different physical devices, or the functions of the core network device can be integrated with the logical functions of the base station on the same physical device, or some functions of the core network device and some functions of the base station can be integrated on one physical device. Figure 1 In this case, taking the core network device including a location management function (LMF) as an example.

[0042] Based on Figure 1 An example process for implementing a positioning function based on the system shown includes: The base station sends the positioning parameters obtained from the terminal device to the LMF, and the LMF obtains a positioning result based on the received data.

[0043] Based on different types of positioning parameters, positioning methods are classified into several types as described below:

[0044] 1. Traditional cellular positioning: Using cellular network signals such as positioning reference signals (PRS) and positioning algorithms to obtain a positioning result.

[0045] Traditional cellular positioning is restricted by the base station layout and signal propagation characteristics. In remote and complex urban environments, the positioning accuracy is as low as dozens of meters or even hundreds of meters, with poor scalability and easy to cause communication bandwidth congestion when the positioning is not good.

[0046] 2. Positioning based on Channel State Information (CSI) and / or Received Signal Strength Indicator (RSSI).

[0047] CSI-based positioning has high hardware requirements and poor environmental adaptability. RSSI-based positioning is affected by multipath effects, hotspot distribution, and signal stability, and it is difficult for the accuracy of positioning results to meet the requirements of high-precision scenarios.

[0048] 3. Satellite-based positioning: Obtain location information based on satellite positioning data such as Global Positioning System (GPS) data acquired by the terminal.

[0049] In complex environments such as urban high-rise areas, mountainous areas, and bad weather, satellite-based positioning has problems of insufficient scalability and low accuracy because satellite signals are interfered by factors such as buildings, terrain, or weather, and the accuracy deteriorates from several meters to more than dozens of meters.

[0050] 4. Sensor data-based positioning: The terminal acquires sensor data collected by sensors carried on the terminal and sends the sensor data to the LMF through the base station. The LMF obtains the location information of the terminal based on the sensor data.

[0051] Sensor data-based positioning has the following problems:

[0052] The accuracy of sensors carried on the terminal is limited. For example, the errors of inertial sensors are prone to accumulation. Also, because the fusion algorithms of various types of sensor data are complex and sensors are greatly affected by the environment, the accuracy of sensor data is low, which reduces the accuracy of positioning results. Moreover, it also faces problems of high resource consumption of sensor devices and difficulty in balancing performance and resources.

[0053] In summary, the current positioning technologies based on wireless communication systems are difficult to meet the high-precision positioning requirements in complex environments and are also difficult to achieve a balance between performance improvement and device resource consumption, which poses a major challenge to improving the accuracy, stability, and universality of positioning.

[0054] To solve the above problems, embodiments of the present application provide a positioning method. The following will describe in detail the positioning method provided by the embodiments of the present application with reference to the accompanying drawings.

[0055] Take Figure 2 as an example. The positioning method provided by the embodiments of the present application is implemented by the system shown in Figure 2 The system includes at least one base station ( Figure 2Take multiple base stations as an example), at least one terminal and core network equipment. Figure 2 Compared with Figure 1 The difference is at least that Figure 2 In addition to the LMF, the core network equipment in Figure 2 also includes a model management platform. Taking the LMF and the model management platform as independent devices as an example, in addition, the model management platform can also be integrated into the LMF.

[0056] The model management platform is used to store and manage various types of artificial intelligence (AI) models and can provide the download of AI models to terminals or network devices. In addition to being set up in the core network in the form of a server or the like, the model management platform can also be a cloud server, a personal computer (PC), or a laptop computer, etc.

[0057] Figure 2 Examples of components for obtaining positioning parameters are also shown in , such as GPS, gyroscopes, accelerometers, and barometric pressure sensors. It can be understood that these components can be independent of the terminal or set in the terminal. In the case where the components are independent of the terminal, the components can send the collected parameters to the terminal.

[0058] In the positioning scenario, the LMF is used to efficiently store and manage various types of positioning-related data, such as base station locations and signal propagation model parameters, etc. At the same time, it accurately manages and maintains the context information of the terminal. The context information covers positioning technology support and mobile status, etc., and can also update its status changes in real time, providing key support for the positioning service and other key functions of the core network.

[0059] First, the various models in the model management platform and the model training strategy will be described below.

[0060] Figure 3 Shows the technical logic for obtaining a positioning result based on various models in the model management platform:

[0061] The model management platform is configured with a base learner and a meta-learner.

[0062] The base learner can be understood as a model that obtains a positioning result based on at least one type of positioning data. Exemplarily, the base learner includes:

[0063] Cellular positioning base learner: Abbreviated as the first type of base learner, used to implement traditional cellular positioning functions. Exemplarily, the cellular positioning base learner obtains a positioning result based on the cellular positioning method without the participation of a model.

[0064] Deep learning model based on CSI and RSSI: Abbreviated as the second type of base learner, which is used to implement the positioning function based on CSI data and / or RSSI data.

[0065] Deep learning model based on GPS and sensors: Abbreviated as the third type of base learner, which is used to implement the positioning function based on GPS data and / or sensor data.

[0066] It can be understood that the deep learning model based on CSI and RSSI, and the deep learning model based on GPS and sensors, are base learners that need to call the deep learning model to obtain the positioning result.

[0067] The meta-learner can be understood as a model that outputs the positioning result based on the output results of the base learners.

[0068] Still taking Figure 3 as an example, in the prediction stage, the cellular positioning base learner outputs the positioning result (X1, Y1), the deep learning model based on CSI and RSSI outputs the positioning result (X2, Y2), and the deep learning model based on GPS and sensors outputs the positioning result (X3, Y3).

[0069] The meta-learner outputs the final positioning result (X, Y) based on (X1, Y1), (X2, Y2), (X3, Y3) and the dynamic environment parameters (number of base stations, number of satellites, sensor status).

[0070] That is, the positioning result output by the meta-learner can be recorded as: (X, Y) = MLP([X1, Y1, X2, Y2, X3, Y3, number of base stations, number of satellites, sensor status]).

[0071] Among them, the number of base stations refers to the number of base stations participating in the positioning, the number of satellites refers to the number of satellites participating in the positioning, the sensor status indicates whether the sensor is normal, such as normal, faulty, etc., and the sensor status can also be the accuracy of the sensor, etc.

[0072] Exemplarily, the meta-learner adopts a Multilayer Perceptron (MLP), and the specific structure is as Figure 4 shown, including an input layer, a hidden layer, and an output layer.

[0073] Exemplarily, the input layer contains 9 neurons corresponding to 9-dimensional feature inputs. In the hidden layer part, the first hidden layer is set with 128 neurons, and the Rectified Linear Unit (ReLU) is used as the activation function to learn the correlation between the dynamic environment parameters and the output of the base learner. Through feature crossing, dynamic weight allocation for the positioning results of the outputs of each base learner based on the dynamic environment parameters is achieved, that is, the first hidden layer can dynamically allocate weights for the positioning results of the outputs of each base learner based on the dynamic environment parameters.

[0074] The second hidden layer has 64 neurons and also uses the ReLU activation function to further extract high-order features. After the two hidden layers, a Dropout layer ( Figure 4 not shown in the figure) is set. The Dropout layer randomly discards 20% of the neurons to prevent the model from overfitting and enhance the generalization ability of the model. The output layer consists of 2 neurons corresponding to the longitude and latitude (X, Y) of the final output respectively.

[0075] The training process of the base learner includes:

[0076] For the deep learning model based on CSI and RSSI (i.e., the second type of base learner), input sample data, where the sample data contains CSI and / or RSSI related parameters (abbreviated as CSI / RSSI sample data), and obtain the positioning result output by the second type of base learner, denoted as (X2, Y2). Through (X2, Y2) and the accurate position label corresponding to the CSI / RSSI sample data, adjust the model parameters of the second type of base learner to obtain the trained second type of base learner.

[0077] For the deep learning model based on GPS and sensors (i.e., the third type of base learner), input sample data, where the sample data contains GPS data and / or sensor data (abbreviated as GPS / sensor sample data), and obtain the positioning result output by the third type of base learner, denoted as (X3, Y3). Through (X3, Y3) and the accurate position label corresponding to the GPS / sensor sample data, adjust the model parameters of the third type of base learner.

[0078] Exemplarily, during the process of training the base learner, LMF may also select an appropriate base learner to be trained based on the features of the sample data, which will be described in combination with the subsequent embodiments. The specific content of the above sample data for training will also be described in the subsequent embodiments.

[0079] After completing the training of the second type of base learner and the third type of base learner, based on the trained second type of base learner and the third type of base learner, train the meta-learner. The process is as follows:

[0080] Input CSI / RSSI sample data (different from the sample data used to train the second - type base learner above) to the second - type base learner, obtain the positioning result output by the second - type base learner, denoted as (X2, Y2), and input GPS / sensor sample data (different from the sample data used to train the third - type base learner above) to the second - type base learner, obtain the positioning result output by the third - type base learner, denoted as (X3, Y3).

[0081] LMF calls the first - type base learner and uses the cellular positioning method to obtain the positioning result, denoted as (X1, Y1).

[0082] Input (X1, Y1), (X2, Y2), (X3, Y3), as well as the base station number sample, satellite number sample, and sensor status sample, into the meta - learner, obtain the predicted coordinates (X, Y) output by the meta - learner, and adjust the model parameters of the meta - learner based on the sample label (i.e., the true position label) and the predicted coordinates (X, Y).

[0083] Exemplarily, during the process of adjusting the model parameters of the meta - learner based on the sample label and the predicted coordinates (X, Y), the mean square error (MSE) is selected as the loss function. By calculating the square of the Euclidean distance between the predicted coordinates and the true position label, the difference degree between the model prediction result and the true position label is measured. The optimizer selects the Adam optimizer, which has the characteristic of adaptive learning rate. The initial learning rate is set to 0.001. This adaptive learning rate adjustment method helps the model converge to the optimal solution faster. During the training process, through the backpropagation algorithm, the chain rule is used to calculate the gradient, and then the weights and biases of each neuron in the MLP are updated, so that the model is continuously optimized.

[0084] The trained base learner and meta - learner are stored in the model management platform for subsequent calls to achieve positioning.

[0085] The following will be combined with Figure 2 each device shown below to describe the positioning method provided by the embodiments of the present application in more detail.

[0086] Figure 5 is the flow of a positioning method disclosed in the embodiments of the present application, including the following steps:

[0087] S101. The base station sends a Positioning Reference Signal (PRS) to the terminal. Correspondingly, the terminal receives the PRS.

[0088] Exemplarily, the base station sends PRS to the terminals within the target area, and the target area can be determined based on relevant protocols.

[0089] The PRS includes base station identification, timestamp, signal configuration information, etc.

[0090] S102. The terminal performs positioning measurements based on the PRS to obtain measurement data.

[0091] Exemplarily, the measurement data includes at least one of signal strength, time of arrival, and angle of arrival, etc.

[0092] S103. The terminal sends the measurement data to the base station. Correspondingly, the base station receives the measurement data.

[0093] S104. The base station obtains a processing result by processing the measurement data.

[0094] Exemplarily, the processing methods include but are not limited to at least one of time calibration, angle calibration, filtering, and outlier removal.

[0095] S105. The base station sends the processing result to the LMF. Correspondingly, the LMF receives the processing result.

[0096] S106. The LMF obtains a positioning result based on the processing result.

[0097] For the purpose of distinguishing from subsequent positioning results, this is called the first type of positioning result here.

[0098] Exemplarily, in this step, the LMF realizes the purpose of using traditional cellular positioning by invoking the first type of base learner.

[0099] S107. The LMF sends the first type of positioning result to the terminal. Correspondingly, the terminal receives the first type of positioning result.

[0100] It can be understood that S101 - S107 is a process of positioning using traditional cellular positioning methods. For more specific implementation methods of each step, reference can be made to relevant protocols, which will not be elaborated here.

[0101] S108. The terminal acquires and preprocesses CSI data and RSSI data.

[0102] Exemplarily, the CSI data includes CSI channel gain data, and the channel gain data may be a channel gain factor, a channel gain vector, or a channel gain matrix. The CSI data may be currently collected data or historical CSI data, that is, previously collected and stored CSI data.

[0103] Exemplarily, the RSSI data includes RSSI strength and RSSI change sequence, and the RSSI data may be currently collected RSSI data or historical RSSI data.

[0104] Exemplarily, the ways of preprocessing CSI data include but are not limited to: adopting the min-max normalization method to map the channel gain data to [0, 1] to eliminate the significant differences in the channel gain range under different scenarios; using mean filtering for noise reduction to effectively weaken the adverse effects of noise interference in the wireless channel on the CSI accuracy.

[0105] The ways of preprocessing historical CSI data include but are not limited to: accurately sorting the historical CSI data from different moments and measurement periods according to the time stamp in chronological order to achieve data alignment; if there is a situation where the time accuracy of the data is inconsistent, using time interpolation means to interpolate the historical CSI data to ensure that the data is evenly and continuously distributed on the time axis.

[0106] Exemplarily, the ways of preprocessing RSSI data include but are not limited to: using Kalman filtering to smooth signal fluctuations and remove anomalies to optimize the quality of RSSI data. For historical RSSI data, similar preprocessing methods to those of historical CSI data can be adopted for processing, which will not be elaborated here.

[0107] S109. The terminal extracts the first comprehensive parameter combination based on the CSI data and the RSSI data.

[0108] For the purpose of distinguishing from the subsequent content, the comprehensive parameter combination in this step is called the first comprehensive parameter combination.

[0109] Exemplarily, the first comprehensive parameter combination includes the following parameters: the number of transceiver antennas, the frequency band used, the number of groups of CSI historical data, and the number of groups of RSSI historical data.

[0110] Examples of the extraction methods for each parameter in the above first comprehensive parameter combination are as follows:

[0111] By analyzing the channel gain data of CSI, such as the dimension of the channel gain matrix, and combining the relationship between multiple matrix dimensions and communication protocols in a complex MIMO system, the number of transceiver antennas can be determined;

[0112] In the currently used communication system, such as the NR system, by analyzing signaling such as the Master Indication Block (MIB) and the System Information Block1 (SIB1) to extract the frequency band identifier, the used frequency band can be clarified.

[0113] The preprocessed CSI historical data is grouped and statistically analyzed at a set time interval. When data loss or anomalies occur, the time window is corrected or redefined to obtain the number of groups of CSI historical data, that is, how many groups of CSI historical data there are.

[0114] First, synchronize the time calibration of the RSSI data and the CSI data, and then count the number of groups of RSSI historical information according to the set time interval by a statistical method similar to that of CSI to obtain the number of groups of RSSI historical data.

[0115] S110. The terminal sends the first comprehensive parameter combination to the model management platform. Correspondingly, the model management platform receives the first comprehensive parameter combination.

[0116] S111. The model management platform selects a suitable model based on the first comprehensive parameter combination. In this step, it is assumed that the model selected by the model management platform and adapted to the first comprehensive parameter combination is the second type of base learner.

[0117] Exemplarily, the model management platform selects a suitable AI model from a predefined model library according to at least one of the characteristics and dimensions of the first comprehensive parameter combination. For example, if the first comprehensive parameter combination contains time series data (such as the number of CSI and / or RSSI historical groups), a base learner of the Long Short-Term Memory (LSTM) type or the Transformer type is selected. If it is necessary to process spatial relationships (i.e., there are spatial relationships in the data) (such as the first comprehensive parameter combination contains Multiple-Input Multiple-Output (MIMO) data), a base learner of the Convolutional Neural Network (CNN) type is selected. In this step, because the first comprehensive parameter combination contains the number of CSI historical data groups and the number of RSSI historical data groups, a base learner based on LSTM or Transformer is selected, which is called the second type of base learner.

[0118] Exemplarily, the above selection rules can be pre-configured in the model management platform, or the adaptation relationships (i.e., corresponding relationships) between each base learner and the characteristics and / or dimensions of the parameter combination can be pre-configured in the model management platform, so as to lay a foundation for selecting a base learner based on the comprehensive parameter combination.

[0119] Selecting a base learner adapted to the first comprehensive parameter combination is beneficial to improving the accuracy of the positioning result output by the base learner based on the first comprehensive parameter combination.

[0120] S112. The terminal downloads the second type of base learner from the model management platform.

[0121] Exemplarily, the model management platform sends information such as the identifier of the second type of base learner to the terminal, and the terminal downloads the second type of base learner from the model management platform based on this information.

[0122] S113. The terminal uses the first comprehensive parameter combination as the input of the second type of base learner, calls the second type of base learner, and obtains the second type of positioning result output by the second type of base learner.

[0123] It can be understood that the second type of positioning result is a positioning result obtained based on the first comprehensive parameter combination, that is, CSI data and RSSI data.

[0124] S114. The terminal acquires and preprocesses GPS data and sensor data.

[0125] Exemplarily, the sensor data includes at least one of the following: gyroscope data, accelerometer data, and barometric pressure sensor data.

[0126] Exemplarily, the preprocessing methods include but are not limited to: cleaning the data, removing obvious error or abnormal data points by setting reasonable data range thresholds, and normalizing the data of different sensors, and aligning the data through timestamps to ensure the temporal synchronization of different sensor data.

[0127] Exemplarily, preprocessing can also be performed on certain types of sensor data. For gyroscope data, abnormal high-frequency fluctuations caused by measurement noise or high-frequency interference are filtered out through a low-pass filter; for another example, for accelerometer data, high-frequency noise generated due to vibrations and the like is removed to make the data smoother and more capable of reflecting the true motion state of the object.

[0128] S115. The terminal extracts the second comprehensive parameter combination based on the GPS data and the sensor data.

[0129] Exemplarily, the second comprehensive parameter combination includes at least one of the following: the average value of the rotational angular velocity, the variance of the rotational angular velocity, the cumulative change in the rotational angle, the peak value of the acceleration, the valley value of the acceleration, the frequency of acceleration change, the barometric pressure change rate, the position dilution of precision, longitude, latitude, and altitude.

[0130] Examples of the extraction methods for each parameter in the above-mentioned second comprehensive parameter combination are as follows:

[0131] The gyroscope data features are extracted as the average value, variance, and cumulative change in the rotational angle over a period of time, etc., to reflect the rotation state and stability of the device. The accelerometer features are extracted as the peak value, valley value, and frequency of acceleration change, etc., and the barometric pressure sensor features are extracted as the barometric pressure change rate. The position information (longitude, latitude, altitude) contained in the GPS data is fused with other sensor data to obtain the second comprehensive parameter combination.

[0132] S116. The terminal sends the second comprehensive parameter combination to the model management platform. Correspondingly, the model management platform receives the second comprehensive parameter combination.

[0133] S117. The model management platform selects a third type of base learner that adapts to the second comprehensive parameter combination based on the second comprehensive parameter combination.

[0134] Exemplarily, in this step, a third type of base learner is selected based on the dimension of the data included in the second comprehensive parameter combination.

[0135] Exemplarily, a third type of base learner can also be selected based on the type of data included in the second comprehensive parameter combination:

[0136] Based on the type of sensor data included in the second comprehensive parameter combination and the accuracy data corresponding to the sensor data, analyze the environment where the terminal is located. For example, judge the current environment type (such as urban canyon, open area, indoor, etc.) based on altitude, etc.;

[0137] Based on the type of sensor data included in the second comprehensive parameter combination, identify the motion state of the terminal, and identify the user's current motion state (walking, driving, stationary, etc.) through rotation and acceleration features;

[0138] Evaluate the current reliability of each sensor according to the variance and change rate of the sensor data.

[0139] Based on the environment where the terminal is located, the motion state, and the current reliability of each sensor, select a third type of base learner. For example, when the position dilution of precision (PDOP) is high, the acceleration changes frequently but moderately, the air pressure changes slowly, and the GPS signal is intermittent, then select a third type of base learner suitable for the driving scenario in a densely built city.

[0140] Correspondingly, data of each scenario can be collected as sample data, and the third type of base learner matching each scenario is trained using the sample data. For example, a large amount of GPS and sensor data are collected in different environmental scenarios (such as city, suburb, indoor, etc.), the data in each scenario is labeled, the real position coordinates are recorded, and the sensor modes in different motion states (stationary, walking, driving, etc.) are collected.

[0141] In the model training stage: Train a dedicated positioning model for different environmental types and motion states, use deep learning architectures (such as CNN, LSTM, etc.) to process spatio-temporal sequence data, and adopt transfer learning techniques to enable the model to adapt to new environments.

[0142] Model classification and storage: The trained models are classified and stored according to the applicable environment and motion characteristics, a model index database is established, and the best applicable conditions of each model (that is, the corresponding relationship between the third type of base learner and the adapted scenario) are recorded.

[0143] S118. The terminal downloads the third type of base learner from the model management platform.

[0144] S119. The terminal takes the second comprehensive parameter combination as the input of the third type of base learner, invokes the third type of base learner, and obtains the third type of positioning result output by the third type of base learner.

[0145] It can be understood that the third type of positioning result is a positioning result obtained based on the second comprehensive parameter combination, that is, GPS data and sensor data.

[0146] S120. The terminal processes the first type of positioning result, the second type of positioning result, and the third type of positioning result to obtain the third comprehensive parameter combination.

[0147] The third comprehensive parameter combination includes the number of positioning modes and the number of groups of positioning results of each positioning mode.

[0148] The positioning mode is the positioning method. In this embodiment, take the cellular positioning mode implemented by the first type of base learner, the positioning mode implemented based on CSI data and RSSI data by the second type of base learner, and the positioning mode implemented based on GPS data and sensor data by the third type of base learner as examples. In this case, in this embodiment, based on the foregoing steps, the number of positioning modes is three. One positioning mode can obtain multiple groups of positioning results. For example, if the terminal obtains GPS data and sensor data at 10 moments, then each moment's data among these 10 moments can obtain a set of longitude and latitude based on the third type of base learner. Therefore, a total of 10 groups of second type of positioning results can be obtained at these 10 moments. The number of arrays of positioning results of each positioning mode refers to the number of positioning results obtained by each positioning mode.

[0149] S121. The terminal sends the third comprehensive parameter combination to the model management platform. Correspondingly, the model management platform receives the third comprehensive parameter combination.

[0150] S122. The model management platform selects a meta-learner adapted to the third comprehensive parameter combination.

[0151] Exemplarily, based on at least one of the information of the positioning modes (the positioning mode is also called the positioning method) included in the third comprehensive parameter combination and / or the information of the positioning results of each positioning mode (the positioning results of each positioning mode are also called the intermediate positioning results obtained by each positioning method), an adapted meta-learner is selected.

[0152] The information on the positioning mode included in the third comprehensive parameter combination includes but is not limited to the type and quantity of the positioning mode. Examples of the type include but are not limited to: types of positioning modes such as cellular positioning, CSI- and RSSI-based positioning, sensor-based positioning, and GPS-based positioning. The quantity indicates how many positioning modes there are.

[0153] The information on the positioning results of each positioning mode includes: the number of groups of the positioning results of each positioning mode and the characteristics of multiple groups of positioning results of the same positioning mode.

[0154] Based on at least one of the information on the positioning mode and / or the number of groups of the positioning results of each positioning mode, the richness of the positioning data (how many groups of valid data each technology provides) and the assessment of the environmental complexity can be obtained.

[0155] Assume that the number of positioning modes is 3, and the positioning modes are cellular positioning, CSI- and RSSI-based positioning, and sensor-based positioning respectively. The number of groups of the positioning results of each positioning mode and the characteristics of multiple groups of positioning results of each positioning mode are as follows:

[0156] • Cellular positioning: 3 groups (with a large dispersion);

[0157] • CSI- and RSSI-based positioning: 2 groups (with a high consistency);

[0158] • Sensor-based positioning: 5 groups (with a small number of outliers).

[0159] From the information on the above positioning modes, it can be known that GPS-based positioning is not available, and based on the characteristics of multiple groups of positioning results of each positioning mode, a meta-learner optimized for the "indoor-outdoor transition area" can be matched.

[0160] It can be understood that meta-learners adapted to various third comprehensive parameter combinations can be pre-trained. Meta-learners adapted to various third comprehensive parameter combinations are meta-learners adapted to multiple scenarios. Multiple scenarios are, for example, urban canyons (areas with dense high-rise buildings), open suburbs, indoor environments, and transportation hubs, etc.

[0161] S123. The terminal downloads the meta-learner from the model management platform.

[0162] Exemplarily, the model management platform sends information such as the identifier of the meta-learner to the terminal, and the terminal downloads the meta-learner from the model management platform based on this information.

[0163] S124. The terminal inputs the first type of positioning results, the second type of positioning results, the third type of positioning results, and the dynamic environment parameters into the meta-learner and calls the meta-learner to output the positioning results.

[0164] Exemplarily, the number of base stations in the dynamic environmental parameters can be obtained based on terminal measurements. The number of satellites in the dynamic environmental parameters can also be obtained by the terminal through measurement or obtained by the terminal from the satellites communicating with the terminal. The sensor status in the dynamic environmental parameters can be obtained by the terminal from each sensor, such as the status of the sensor carried in the sensing data.

[0165] As mentioned above, since the meta-learner learns the correlation between the dynamic environmental parameters and the output of the base learner during training, it can dynamically allocate weights to the positioning results output by each base learner based on the dynamic environmental parameters. Therefore, the accuracy of the output result of the meta-learner can be further improved.

[0166] Figure 5 The shown process has the following beneficial effects:

[0167] 1. Multiple base learners obtain positioning results from different dimensions, and the positioning results of different dimensions are fused to obtain the final positioning result. Compared with a single positioning method, the accuracy of the final positioning result is higher. Moreover, when the accuracy of some positioning methods in different dimensions is limited, such as when the accuracy of cellular positioning and satellite positioning in the basement decreases, the method of fusing the positioning results of different dimensions can also ensure the accuracy of the final positioning result. Therefore, it has strong tolerance and universality for the environment.

[0168] 2. For the base learner and meta-learner of the deep learning model class, they are both selected based on the information of the output data (i.e., the comprehensive parameter combination) (such as at least one of type, feature, and dimension). The model is more adaptable to the input data, which can further improve the accuracy of the positioning result. Moreover, selecting the model based on the parameters (such as each comprehensive parameter combination) is also beneficial to improving the utilization rate of resources. That is to say, it can avoid wasting the computing power resources of the model and fully utilize the computing power resources of the model to obtain a positioning result with accuracy meeting the requirements.

[0169] 3. The meta-learner can output the final positioning result based on the dynamic environmental parameters, which can further improve the accuracy of the positioning result.

[0170] Figure 6 It is another positioning method provided by the embodiments of the present application. Different from Figure 5 the method shown, after obtaining the first type of positioning result, LMF does not send the first type of positioning result to the terminal. After the terminal obtains the second type of positioning result, it sends the second type of positioning result to LMF. After the terminal obtains the third type of positioning result, it sends the third type of positioning result to LMF.

[0171] Different from S123, such as S224, the LMF downloads the meta-learner from the model management platform. Different from S124, such as S225, the LMF inputs the first type of positioning result, the second type of positioning result, the third type of positioning result, and the dynamic environment parameters into the meta-learner, obtains the positioning result output by the meta-learner, and after obtaining the positioning result, sends the positioning result to the terminal. Correspondingly, the terminal receives the positioning result. Except for the above differences, Figure 6 other steps in Figure 5 can be referred to

[0172] Hereinafter, the positioning method provided in the above embodiments will be illustrated by way of two specific scenarios.

[0173] High-precision positioning in a dynamic urban canyon environment:

[0174] In this example, the devices participating in positioning include: multiple densely distributed base stations in the city, a terminal equipped with multiple sensors, and an LMF, and a model management platform is set in the LMF. The above first type of base learner, second type of base learner, third type of base learner, and MLP meta-learner are all trained and stored in the model management platform.

[0175] In order to improve the positioning response speed and reduce the latency, common and lightweight base learners are downloaded and cached by the terminal from the model management platform.

[0176] The terminal, the base station, and the LMF cooperate to execute the positioning process provided in the above embodiments to obtain the positioning result.

[0177] In this example, after obtaining the positioning result, the algorithms and / or models used in the positioning process can also be optimized and adjusted based on the positioning result:

[0178] An example of optimization and adjustment is: comparing the final positioning result with the positioning result output by the first type of base learner. If the gap is greater than the first threshold, the corresponding network device is triggered to reconfigure the base station, such as adjusting the transmission power and time-frequency resource allocation of the base station PRS signal, or optimizing the signal coverage in the indoor-outdoor edge area, etc., to increase the strength of the PRS signal.

[0179] Another example of optimization and adjustment is: real-time monitoring of the gap between the first type of positioning result, the second type of positioning result, and the third type of positioning result and the final positioning result. If the gap is greater than the second threshold (such as 5 meters), trigger the re-training of the base learner and / or the meta-learner to ensure the long-term stability of the model in complex scenarios.

[0180] It can be understood that the terminal or LMF can upload the final positioning result and the data used to obtain the final positioning result (such as PRS, CSI data, RSSI data, GPS data, and sensor data) to the model management platform, so that the model management platform stores them as new training samples for subsequent retraining of the model. The training samples obtained in the dynamic urban canyon environment can enhance the characteristics of the model to resist the multipath interference in the urban canyon.

[0181] Seamless positioning enhancement in indoor-outdoor handover scenarios:

[0182] In this example, the devices participating in positioning include: base stations in an indoor shopping mall and an outdoor square, terminals configured with multi-band antennas and various high-precision sensors, and the aforementioned LMF and model management platform.

[0183] Different from the above embodiments, the models in the model management platform are adapted not only to the characteristics and / or dimensions of various types of data, but also to indoor and outdoor scenarios. For example, indoor and outdoor scenarios are divided into being indoors, outdoors, from indoors to outdoors, and from outdoors to indoors, and at least one model is adapted to each scenario. For example, the main model adapted to the indoor scenario is the aforementioned second type of base learner, and the auxiliary model is a model for positioning based on inertial sensor data. Another example is that the model adapted to the outdoor scenario is the aforementioned first type of base learner and a model for positioning based on GPS.

[0184] Similarly, the meta-learner can also be adapted to indoor and outdoor scenarios. For example, the first meta-learner trained with outdoor data is adapted to the outdoor scenario.

[0185] A lightweight scene classification model is also deployed in the model management platform. The scene classification model is used to analyze and identify the current scenario where the terminal is located, that is, whether the terminal is currently indoors, outdoors, from indoors to outdoors, or from outdoors to indoors.

[0186] The terminal can download the lightweight scene classification model and some base learners from the model management platform.

[0187] In this example, the difference from the foregoing embodiments is further that the LMF deploys a backup positioning mechanism. The backup positioning mechanism may be one of the foregoing positioning methods, such as the method of positioning based on GPS data and sensor data, or may be the positioning method provided by the embodiments of the present application. The purpose of deploying the backup is to obtain a positioning result using another positioning process in the case where one positioning process cannot give a positioning result. For example, in the LMF, a first positioning network element and a second positioning network element are deployed. The first positioning network element uses the positioning method provided by this embodiment for positioning, and the second positioning network element positions based on GPS data and sensor data. By default, the first positioning network element is used for positioning. In the case where the first positioning network element cannot perform positioning (such as equipment failure), the second positioning network element is enabled for positioning to ensure the stability of the positioning service.

[0188] The process of the positioning method based on the above configuration includes:

[0189] The terminal calls a lightweight scene classification model, combines the obtained data such as the current CSI data, and identifies the current scene. If the current scene is indoor, it calls the second type of base learner and the model for positioning based on inertial sensor data to obtain a positioning result. Exemplarily, the positioning result output by the second type of base learner and the positioning result output by the model for positioning based on inertial sensor data are input into the meta-learner adapted to the indoor scene to obtain the final positioning result output by the meta-learner. The inertial sensor data can compensate for the error caused by the occlusion of the CSI signal, which is beneficial to improving the accuracy of the final positioning result.

[0190] In the indoor scene, if the inertial sensor is disturbed by vibration and the inertial sensor data is abnormal, the meta-learner can, based on the dynamic environment parameter of the sensor state, reduce the weight of the positioning result output by the model for positioning based on inertial sensor data, and preferentially rely on the positioning result output by the second type of base learner (that is, preferentially rely on CSI data), so the accuracy of the final positioning result can be improved.

[0191] If the current scene is outdoor, it calls the first type of base learner adapted to the outdoor scene and the model for positioning based on GPS respectively to obtain positioning results, and then obtains the final positioning result based on the meta-learner adapted to the outdoor scene. The meta-learner can suppress the noise caused by the GPS signal jump, which is beneficial to improving the accuracy of the final positioning result.

[0192] It can be understood that because the called model is adapted to the outdoor scene, it is necessary to accurately obtain the input data of the model. Therefore, if the terminal determines that the GPS signal is lost, it enables a backup model (base learner), such as a model for positioning based on sensors.

[0193] Combined with the redundant backup mechanism, the above process can be implemented by the first positioning network element. In the case where the first positioning network element cannot perform positioning (such as equipment failure), the second positioning network element is enabled for positioning.

[0194] In this example, the optimization adjustment measurement described in the above example can also be performed to further improve the positioning accuracy.

[0195] In this example, for the indoor-outdoor handover scenario, the positioning performance of different scenarios can also be regularly evaluated. For example, the positioning result is compared with the actual measurement result to obtain the performance evaluation result. For areas with poor performance, such as areas with high-frequency indoor-outdoor scenario handovers like shopping mall entrances and exits, an MLP meta-learner dedicated to this area is trained to further reduce the handover latency and positioning jitter.

[0196] In summary, the positioning method provided by the above embodiments has the following beneficial effects:

[0197] 1. Multi-technology fusion improves positioning accuracy:

[0198] Existing technologies (such as traditional cellular networks, GPS, sensor positioning, etc.) rely on single or a few technologies, and static weight fusion cannot dynamically perceive environmental changes, resulting in insufficient accuracy in complex scenarios.

[0199] In the embodiments of the present application, through an integrated learning framework, the preliminary results of traditional positioning, CSI / RSSI positioning based on deep learning, and GPS / sensor positioning are combined with dynamic environmental parameters (number of base stations, number of satellites, sensor status), and input into the MLP meta-learner. The weights of each technology are adjusted in real time using the dynamic environmental parameters (such as reducing the weight of traditional positioning when the number of base stations is insufficient). By feature cross-suppressing the deviations caused by signal interference and environmental mutations, multi-dimensional data complementarity and scene adaptive fusion are achieved, significantly improving the positioning accuracy in complex environments (such as urban canyons, indoor-outdoor handovers).

[0200] 2. Integrated learning enhances environmental adaptability:

[0201] Existing solutions only support the switching between traditional and AI positioning, lack the deep fusion of multi-technology data, have limited environmental adaptability, and are difficult to handle complex scenarios such as multipath interference and non-line-of-sight propagation.

[0202] In the embodiments of the present application, the MLP meta-learner is trained based on a large amount of actual scenario data. Through the high-order feature interaction of the MLP hidden layer, the potential relationship between different positioning technologies is mined (such as using sensors to compensate for GPS signal occlusion). Combining the feature interaction and scene recognition parameters of multi-source positioning technologies, it dynamically adapts to different environments (such as indoor, outdoor, mountainous areas), achieving high-robustness positioning across scenarios.

[0203] 3. Coordinated optimization of resource efficiency and reliability:

[0204] The prior art fails to achieve dynamic optimization of resources and accuracy, relies on fixed model configurations, resulting in resource waste or insufficient accuracy, and lacks a fault tolerance mechanism, with a high risk of common mode failure.

[0205] In the embodiments of the present application, the technology combination is flexibly adjusted according to environmental conditions (such as selecting an adapted model based on indoor and outdoor scenarios), combined with selecting a model based on a comprehensive parameter combination, and redundant design is adopted to achieve a dynamic balance between resource consumption and positioning accuracy, which is applicable to safety-critical fields such as rail transit.

[0206] Figure 7 It is an example of the composition of a positioning device provided by the embodiments of the present application. The positioning device may be a terminal, including but not limited to electronic devices such as mobile phones and smart wearable devices (such as smart watches). Taking a mobile phone as an example, the positioning device may include a processor 110, an internal memory 120, a display screen 130, an antenna 1, an antenna 2, a mobile communication module 140, and a wireless communication module 150, etc.

[0207] It can be understood that the structure illustrated in this embodiment does not constitute a specific limitation on the positioning device. In other embodiments, the positioning device may include more or fewer components than those shown in the figure, or combine certain components, or split certain components, or have different component arrangements. The illustrated components may be implemented in hardware, software, or a combination of software and hardware.

[0208] The processor 110 may include one or more processing units. For example, the processor 110 may include an application processor (AP), a modulation and demodulation processor, a digital signal processor (DSP), and / or a baseband processor, etc.

[0209] The internal memory 120 may be used to store computer-executable program code, and the executable program code includes instructions. The processor 110 executes various functions of the electronic device by running the instructions stored in the internal memory 120.

[0210] The wireless communication function of the electronic device may be implemented by the antenna 1, the antenna 2, the mobile communication module 140, the wireless communication module 150, the modulation and demodulation processor, and the baseband processor, etc.

[0211] The antenna 1 and the antenna 2 are used to transmit and receive electromagnetic wave signals.

[0212] The mobile communication module 140 may provide solutions for wireless communications including 2G / 3G / 4G / 5G, etc. applied to the electronic device.

[0213] In some embodiments, the mobile communication module 140 includes a communication interface that is coupled to the processor 110. The communication interface may be a transceiver or an input / output interface. In some embodiments, when the positioning device is a chip configured in a terminal, the communication interface may be an input / output interface.

[0214] The wireless communication module 150 can provide solutions for wireless communications applied to an electronic device, including wireless local area networks (WLANs) (such as wireless fidelity (Wi-Fi) networks), Bluetooth (BT), global navigation satellite system (GNSS), frequency modulation (FM), near field communication (NFC), infrared technology (IR), etc.

[0215] In addition, an operating system runs on the above components. For example, iOS operating system, Android operating system, Windows operating system, etc. Application programs can be installed and run on the operating system.

[0216] Figure 8 This is a composition example of another positioning device provided by the embodiments of the present application. The positioning device may be a network device, such as an LMF. Figure 8 A simplified structural schematic diagram of a network device is shown. The network device includes: at least one processor 210, at least one memory 220, at least one transceiver 230, at least one network interface 240, and one or more antennas 250. The processor 210, the memory 220, the transceiver 230, and the network interface 240 are connected, for example, through a bus. In the embodiments of the present application, the connection may include various interfaces, transmission lines, or buses, etc. This embodiment does not limit this. The antenna 250 is connected to the transceiver 230. The network interface 240 is used to enable the network element to be connected to other communication devices through a communication link. For example, the network interface 240 may include a network interface between the network element and the network element in the core network, such as an S1 interface. The network interface may include a network interface between the network element and other network elements, such as an X2 or Xn interface.

[0217] Figure 8The processor 210 shown in [Figure] can specifically perform the functions of each step executed by the LMF in the above positioning method. The memory 220 can perform the storage actions in the above positioning method. The transceiver 230 and the antenna 250 can execute the transceiver actions in the above positioning method. The network interface 240 can perform the actions of interacting between the LMF and the terminal and / or the base station in the above positioning method.

[0218] The processor 210 may include, but is not limited to, at least one of the following: a central processing unit (CPU), a microprocessor, a digital signal processor (DSP), a microcontroller unit (MCU), or various computing devices that run software such as an artificial intelligence processor. Each computing device may include one or more cores for executing software instructions to perform operations or processing. The processor may be a separate semiconductor chip or may be integrated with other circuits into a semiconductor chip. For example, it may form a system-on-chip (SoC) with other circuits (such as codec circuits, hardware acceleration circuits, or various bus and interface circuits), or it may be integrated as an internal processor of an ASIC. The ASIC integrated with the processor may be separately packaged or may be packaged together with other circuits. In addition to the cores for executing software instructions to perform operations or processing, the processor may further include necessary hardware accelerators, such as a field programmable gate array (FPGA), a programmable logic device (PLD), or a logic circuit for implementing dedicated logical operations.

[0219] The memory 220 may include at least one of the following types but is not limited thereto: a read-only memory (ROM) or other types of static storage devices that can store static information and instructions, a random access memory (RAM) or other types of dynamic storage devices that can store information and instructions, or it may also be an electrically erasable programmable-only memory (EEPROM).

[0220] The transceiver 230 can be used to support the reception or transmission of radio frequency signals between network elements and other devices, and the transceiver 230 can be connected to the antenna 250. The transceiver 230 includes a transmitter Tx and a receiver Rx. Specifically, one or more antennas 250 can receive radio frequency signals. The receiver Rx of the transceiver 230 is used to receive radio frequency signals from the antenna, convert the radio frequency signals into digital baseband signals or digital intermediate frequency signals, and provide the digital baseband signals or digital intermediate frequency signals to the processor 210 so that the processor 210 can further process the digital baseband signals or digital intermediate frequency signals, such as demodulation processing and decoding processing. In addition, the transmitter Tx in the transceiver 230 is also used to receive the modulated digital baseband signals or digital intermediate frequency signals from the processor 210, convert the modulated digital baseband signals or digital intermediate frequency signals into radio frequency signals, and transmit the radio frequency signals through one or more antennas 250. Specifically, the receiver Rx can selectively perform one-stage or multi-stage down-conversion processing and analog-to-digital conversion processing on the radio frequency signals to obtain digital baseband signals or digital intermediate frequency signals, and the order of the down-conversion processing and the analog-to-digital conversion processing can be adjusted. The transmitter Tx can selectively perform one-stage or multi-stage up-conversion processing and digital-to-analog conversion processing on the modulated digital baseband signals or digital intermediate frequency signals to obtain radio frequency signals, and the order of the up-conversion processing and the digital-to-analog conversion processing can be adjusted. The digital baseband signals and the digital intermediate frequency signals can be collectively referred to as digital signals.

[0221] The transceiver 230 can also be referred to as an input / output interface, a communication interface, etc. In some embodiments, when the above positioning device is a chip configured in a satellite, the transceiver 230 can be an input / output interface.

[0222] It should be understood that Figure 8 by way of example only and not limitation, the above network device including a processor, a memory, and a transceiver may not depend on Figure 8 the structure shown.

[0223] The embodiments of the present application also provide a positioning device.

[0224] As Figure 9 shown, the positioning device 300 includes a processing module 301 and a transceiver module 302. In some embodiments, the positioning device may further include a storage module 303, and the storage module can be used to store instructions (codes or programs) and / or data. The processing module 301 and the transceiver module 302 can be coupled to the storage module 303. For example, the processing module 301 can read the instructions (codes or programs) and / or data in the storage module to implement the corresponding methods. The above-mentioned various modules can be set independently, or partially or fully integrated.

[0225] The processing module 301, the transceiver module 302, and the storage module 303 in the embodiments of the present application are used to enable the positioning device 300 to implement the functions of the network device (such as a core network device or a base station) in the above method embodiments, or to enable the positioning device 300 to implement the functions of the terminal in the above method embodiments.

[0226] The following takes the positioning device 300 for implementing the functions of the terminal in the above method embodiments as an example to describe each module in the positioning device.

[0227] In some embodiments, the positioning device 300 may correspondingly implement the functions or steps implemented by the terminal in the above method embodiments. Specifically, the processing module 301 is used to obtain an intermediate positioning result, where the intermediate positioning result includes: a first positioning result and a second positioning result. The first positioning result is obtained based on a first positioning method, the second positioning result is obtained based on a second positioning method, and the first positioning method is different from the second positioning method; based on the intermediate positioning result and dynamic environment parameters, obtain the positioning result of the terminal, where the dynamic environment parameters are used to configure the weight of the intermediate positioning result, and the dynamic environment parameters characterize at least one of the number of base stations associated with the terminal, the number of satellites associated with the terminal, and the sensor state associated with the terminal.

[0228] During the execution of the above process by the processing module 301, the interaction function of the required information or data is executed by the transceiver module 302. For the specific interaction of the required information or data, reference may be made to the above method embodiments.

[0229] It should be noted that the information interaction, execution process, etc. between the above device modules, due to the same concept as the method embodiments of the present application, bring the same technical effects as the method embodiments of the present application. For the specific content, reference may be made to the description in the method embodiments shown above in the present application, and details are not described herein again.

[0230] In other embodiments, the positioning device 300 may correspondingly implement the functions or steps implemented by the LMF in the above method embodiments. Specifically, the processing module 301 is used to obtain an intermediate positioning result, where the intermediate positioning result includes: a first positioning result and a second positioning result. The first positioning result is obtained based on a first positioning method, the second positioning result is obtained based on a second positioning method, and the first positioning method is different from the second positioning method; based on the intermediate positioning result and dynamic environment parameters, obtain the positioning result of the terminal, where the dynamic environment parameters are used to configure the weight of the intermediate positioning result, and the dynamic environment parameters characterize at least one of the number of base stations associated with the terminal, the number of satellites associated with the terminal, and the sensor state associated with the terminal.

[0231] During the process of the processing module 301 executing the above process, the interaction function of the required information or data is executed by the transceiver module 302. For the specific interaction of the required information or data, reference can be made to the above method embodiments.

[0232] It should be noted that for the information interaction, execution process, etc. between the above-mentioned device modules, since they are based on the same concept as the method embodiments of the present application, the technical effects brought by them are the same as those of the method embodiments of the present application. For the specific content, reference can be made to the description in the method embodiments shown above in the present application, and details will not be repeated here.

[0233] Those skilled in the art can clearly understand that for the explanation and beneficial effects of the relevant content in any of the above-provided positioning devices, reference can be made to the corresponding method embodiments provided above, and details will not be repeated here.

[0234] The embodiment of the present application also provides a processor, including: an input circuit, an output circuit, and a processing circuit. Among them: the processing circuit is used to receive a signal through the input circuit and transmit the signal through the output circuit, so that the processor executes the positioning method described in the above embodiment.

[0235] In the specific implementation process, the processor can be one or more chips. The input circuit can be an input pin, the output circuit can be an output pin, and the processing circuit can be transistors, gate circuits, flip-flops, and various logic circuits, etc. The input signal received by the input circuit can be input by, for example, but not limited to, a receiver. The signal output by the output circuit can be output to, for example, but not limited to, a transmitter and transmitted by the transmitter. Moreover, the input circuit and the output circuit can be the same circuit, and this circuit is used as the input circuit and the output circuit at different times respectively. The embodiment of the present application does not limit the specific implementation manners of the processor and various circuits.

[0236] The embodiment of the present application also provides a chip system, which includes one or more processors and is used to call and run the instructions stored in the memory from the memory, so that the positioning method described in the above embodiment is executed. The chip system can be composed of chips or can include chips and other discrete devices. Among them, the chip system can include an input circuit or interface for sending information or data, and an output circuit or interface for receiving information or data.

[0237] The embodiment of the present application also provides a computer-readable storage medium, in which instructions are stored. When the instructions are run on one or more computing devices, the one or more computing devices are enabled to execute the positioning method described in the above embodiment.

[0238] A computer-readable storage medium may be a non-transitory computer-readable storage medium. For example, the non-transitory computer-readable storage medium may be a read-only memory (ROM), a random access memory (RAM), a CD-ROM, a magnetic tape, a floppy disk, and an optical data storage device, etc.

[0239] The embodiments of the present application further provide a computer program product. When the computer program product is executed by one or more computing devices, the one or more computing devices execute any one of the foregoing positioning methods. The computer program product may be a software installation package. In the case where any one of the foregoing positioning methods needs to be used, the computer program product may be downloaded and executed on a computer.

[0240] As described above, the above embodiments are only used to illustrate the technical solutions of the present application, rather than to limit it; although the present application has been described in detail with reference to the foregoing embodiments, those of ordinary skill in the art should understand that: they can still modify the technical solutions recorded in the foregoing embodiments, or perform equivalent replacements on some of the technical features; and these modifications or replacements do not make the essence of the corresponding technical solutions deviate from the scope of the technical solutions of the embodiments of the present application.

Claims

1. A positioning method, characterized in that, Including: Obtain an intermediate positioning result, where the intermediate positioning result includes: a first positioning result and a second positioning result. The first positioning result is obtained based on a first positioning method, and the second positioning result is obtained based on a second positioning method. The first positioning method is different from the second positioning method. Based on the intermediate positioning result and dynamic environment parameters, obtain the positioning result of the terminal. The dynamic environment parameters are used to configure the weights of the intermediate positioning result, and the dynamic environment parameters characterize at least one of the number of base stations associated with the terminal, the number of satellites associated with the terminal, and the state of the sensors associated with the terminal.

2. The method according to claim 1, wherein The process of obtaining the second positioning result based on the second positioning method includes: By invoking a first deep learning model, obtain the position information output by the first deep learning model based on a parameter combination. The first deep learning model has an adaptation relationship with the parameter combination.

3. The method according to claim 2, wherein The first deep learning model having an adaptation relationship with the parameter combination includes: The first deep learning model has an adaptation relationship with at least one of the features and dimensions of the parameter combination.

4. The method according to claim 3, characterized in that, The first deep learning model having an adaptation relationship with the features of the parameter combination includes at least one of the following: A deep learning model of the long short-term memory network type is adapted to a parameter combination including time series data; A deep learning model of the convolutional neural network type is adapted to a parameter combination including spatial relationships; The first target model in the first deep learning model is also adapted to the indoor / outdoor scenario where the terminal is located. The indoor / outdoor scenario includes: indoor, outdoor, from indoor to outdoor, or from outdoor to indoor.

5. The method according to any one of claims 2-4, characterized in that, The parameter combination includes the following data: The number of transceiver antennas of the terminal, the frequency band used by the terminal, the number of groups of channel state information (CSI) historical data obtained by the terminal, and the number of groups of received signal strength indication (RSSI) historical data; Alternatively, the parameter combination includes at least one of the following data: satellite positioning data and sensor data.

6. The method according to any one of claims 2-4, characterized in that Before obtaining the intermediate positioning result, it further includes: Download the first deep learning model from the model management platform. The first deep learning model is selected by the model management platform from multiple models based on the adaptation relationship.

7. The method according to claim 1, wherein The obtaining the positioning result of the terminal based on the intermediate positioning result and dynamic environment parameters includes: By invoking a second deep learning model, obtain the positioning result output based on the intermediate positioning result and dynamic environment parameters. The second deep learning model includes a hidden layer, and the hidden layer is used to assign weights to the intermediate positioning result.

8. The method according to claim 7, wherein The second deep learning model is adapted to the processing result of the intermediate positioning result. The processing result includes at least one of the following: information on the positioning method for obtaining the intermediate positioning result, and information on the intermediate positioning result obtained by each positioning method.

9. The method according to claim 7 or 8, characterized in that, Before obtaining the positioning result of the terminal based on the intermediate positioning result and dynamic environment parameters, it further includes: Download the second deep learning model from the model management platform, where the second deep learning model is selected by the model management platform from multiple models based on the adaptation relationship between the second deep learning model and the processing result, and the processing result includes at least one of the following: information on the positioning method for obtaining the intermediate positioning result and information on the intermediate positioning result obtained by each positioning method.

10. The method according to claim 1 or 7, characterized in that, After obtaining the positioning result of the terminal, it further includes: Compare the positioning result of the terminal with the first positioning result to obtain a first difference; Based on the first difference and the first threshold, adjust the configuration of the signal sent by the base station, where the signal is the basis for obtaining the first positioning result by the first positioning method.

11. The method according to claim 1 or 7, characterized in that, After obtaining the positioning result of the terminal, it further includes: Compare the positioning result of the terminal with the second positioning result to obtain a second difference; Based on the second difference and the second threshold, retrain at least one of the first deep learning model for obtaining the second positioning result and the second deep learning model for obtaining the positioning result of the terminal.

12. A positioning device, characterized in that, It includes: A module for implementing the positioning method according to any one of claims 1-11.

13. A positioning device, characterized in that, It includes: One or more processors and a memory; the memory is used to store program code; The processor is used to run the program code so that the positioning device implements the positioning method according to any one of claims 1 to 11.

14. A computer-readable storage medium, characterized in that, Instructions are stored thereon, and when the instructions are run on an electronic device, the electronic device is caused to execute the positioning method according to any one of claims 1 to 11.

15. A chip system, characterized in that, It includes: At least one processor and an interface, where the interface is used to receive code instructions and transmit them to the at least one processor; The at least one processor runs the code instructions to implement the positioning method according to any one of claims 1 to 11.

Images