Positioning method and device based on AI model

Abstract

The application provides an AI model-based positioning method and device, and relates to the technical field of communication. The method comprises the following steps: a terminal receives a PRS signal sent by an access network device; the terminal measures the PRS signal to obtain PRS data; the PRS data at least comprises a line-of-sight (LOS) confidence, a reference signal time difference (RSTD), and a terminal receiving and sending time difference (RXTX); the terminal performs quantization processing on the PRS data to obtain first data associated with the PRS data; the first data comprises a first parameter, a second parameter, and a third parameter; wherein the first parameter is obtained by quantizing the LOS confidence, the second parameter is obtained by quantizing the RSTD, and the third parameter is obtained by quantizing the RXTX; the first parameter, the second parameter, and the third parameter are parameters of the same dimension; and the terminal inputs the first data into an AI model to determine the position information of the terminal. The method can solve the problem of excessively high storage space and computing resources of the AI model, and improve the positioning accuracy.

Description

Technical Field

[0001] This application relates to the field of communication technology, and in particular to a positioning method and apparatus based on an AI model. Background Technology

[0002] In the field of wireless communication, terminals can deploy artificial intelligence / machine learning (AI / ML) models to achieve terminal positioning, that is, to determine the location information of the terminal.

[0003] AI / ML models can achieve terminal positioning through various methods, such as triangulation and fingerprint positioning. However, existing positioning methods consume excessive computational resources for AI / ML models, resulting in weak terminal computing power. Furthermore, excessive computational resources lead to large data storage volumes for AI / ML models, resulting in large models that occupy significant terminal storage space. Additionally, this results in longer data processing times for the terminal, making it difficult to ensure high-precision positioning and thus limiting positioning accuracy. Summary of the Invention

[0004] This application provides a positioning method and apparatus based on an AI model, which can solve the problem of excessive storage space and computing resources of the model and improve positioning accuracy.

[0005] Firstly, an AI model-based positioning method is provided, applied to a terminal. The terminal is equipped with an AI model used to locate the terminal's position. The method includes: the terminal receiving a PRS signal sent from an access network device; the terminal measuring the PRS signal to obtain PRS data; the PRS data including at least the line-of-sight (LOS) confidence score, reference signal time difference (RSTD), and terminal receive-transmit time difference (RXTX); the terminal quantizing the PRS data to obtain first data associated with the PRS data; the first data including a first parameter, a second parameter, and a third parameter; wherein the first parameter is obtained by quantizing the LOS confidence score, the second parameter is obtained by quantizing the RSTD, and the third parameter is obtained by quantizing the RXTX; the first, second, and third parameters are parameters of the same dimension; and the terminal inputting the first data into the AI ​​model to determine the terminal's location information.

[0006] Based on the method provided in this application, the terminal obtains PRS data by measuring the PRS signal and performs quantization processing on the PRS data, unifying the various parameters in the PRS data to the same dimension, reducing the computing resources and storage space of the AI ​​model on the terminal, thereby improving positioning accuracy.

[0007] In one possible implementation of the first aspect, the first data is RGB data, the first parameter is the red pixel information R in the RGB data, the second parameter is the green pixel information G in the RGB data, and the third parameter is the blue pixel information B in the RGB data.

[0008] In one possible implementation of the first aspect, the terminal is equipped with multiple antennas, each of which is used to receive a PRS signal, one PRS signal corresponds to one RGB data, the multiple antennas correspond to multiple RGB data, and the multiple RGB data constitute an RGB image.

[0009] In one possible implementation of the first aspect, the terminal inputs first data into the AI ​​model to determine the terminal's location information, including: the terminal inputs an RGB image into the AI ​​model to determine the terminal's location information. In another possible implementation of the first aspect, the AI ​​model includes a first model; the terminal inputting an RGB image into the AI ​​model to determine the terminal's location information includes: the terminal determining whether the size of the RGB image is less than a threshold; if the size of the RGB image is less than the threshold, the RGB image is input into the first model to determine the terminal's location information.

[0010] In one possible implementation of the first aspect, the AI ​​model includes a second model; the method further includes: if the size of the RGB image is greater than or equal to a threshold, the terminal inputs the RGB image into the second model for dimensionality reduction processing; the terminal inputs the dimensionality-reduced RGB image into the first model to determine the terminal's location information.

[0011] Based on the method provided in this application, a second model is added for dimensionality reduction, in addition to the first model for localization. When the size of the RGB image is larger than a threshold, the RGB image is input into the second model for dimensionality reduction, and then the dimensionality-reduced result is input into the first model for localization. This reduces the amount of data processing and improves computational efficiency.

[0012] In one possible implementation of the first aspect, the first parameter satisfies the following quantization formula:

[0013]

[0014] Among them, P LOS For LOS confidence, P LOS ∈[0,1], β is the distribution coefficient, used to indicate the complexity of the PRS signal propagation environment, and round() means rounding the value in ().

[0015] The second parameter satisfies the following quantization formula:

[0016]

[0017] Where RTSD is the reference signal time difference, RTSD max The maximum value of the time difference of the reference signal, RTSD min The minimum time difference of the reference signal is indicated by round(), which rounds the value within the parentheses to the nearest integer.

[0018] The third parameter satisfies the following quantization formula:

[0019]

[0020] Where RXTX is the time difference between terminal reception and transmission. min RXTX represents the minimum time difference between the terminal's receive and transmit signals. max This represents the maximum value of the time difference between the terminal's reception and transmission. `round()` indicates that the value within the parentheses is rounded to the nearest integer.

[0021] In one possible implementation of the first aspect, the method further includes: the terminal sending indication information to the access network device, the indication information being used to instruct the location management function (LMF) to send the terminal's location information, the indication information including the terminal's location information.

[0022] In a second aspect, a communication device is provided, the communication device comprising one or more processors; the one or more processors are configured to execute computer programs or instructions, such that when the one or more processors execute the computer programs or instructions, a communication method as described in the first aspect and any possible implementation thereof is performed.

[0023] In one possible implementation of the second aspect, the communication device further includes one or more memories coupled to one or more processors, the memories used to store the aforementioned computer program or instructions. In one possible implementation, the memories are located outside the communication device. In another possible implementation, the memories are located inside the communication device. In this application, the processor and memory can also be integrated into a single device, i.e., the processor and memory can be integrated together. In one possible implementation, the communication device further includes a transceiver for receiving and / or transmitting information.

[0024] In one possible design, the communication device further includes one or more communication interfaces coupled to one or more processors, and the communication interfaces are used to communicate with other modules outside the device.

[0025] Thirdly, this application provides a communication device, which includes an interface circuit and a logic circuit; the interface circuit is used for inputting and / or outputting information; the logic circuit is used to execute a communication method as described in the first aspect and any possible implementation thereof, and to process and / or generate information based on the information.

[0026] Fourthly, this application provides a computer-readable storage medium storing computer instructions or programs that, when executed on a computer, cause the method of the first aspect and any possible implementation thereof to be performed.

[0027] Fifthly, this application provides a chip comprising: a processor coupled to a memory for storing programs or instructions, wherein when the program or instructions are executed by the processor, a method as described in the first aspect and any possible implementation thereof is performed.

[0028] In a sixth aspect, this application provides a computer program product containing computer instructions that, when run on a computer, causes the method of the first aspect and any possible implementation thereof to be executed.

[0029] In a seventh aspect, this application provides a communication system including communication means for implementing the method in the first aspect.

[0030] The technical effects of any of the design methods in aspects two through seven can be seen in the technical effects of any of the possible implementation methods in aspect one above, and will not be elaborated further. Attached Figure Description

[0031] Figure 1 This application provides a schematic diagram of the architecture of a communication system.

[0032] Figure 2 This is a schematic diagram illustrating the deployment of an AI model on a terminal for localization, as provided in an embodiment of this application.

[0033] Figure 3 This is a schematic diagram of a trilateration method provided in an embodiment of this application;

[0034] Figure 4a A schematic diagram of the interaction process of a localization method based on an AI model provided in an embodiment of this application;

[0035] Figure 4b A schematic diagram of the interaction process of another AI model-based localization method provided in the embodiments of this application;

[0036] Figure 4c A schematic diagram of the interaction process of another AI model-based localization method provided in the embodiments of this application;

[0037] Figure 4d A schematic diagram of the interaction process of another AI model-based localization method provided in the embodiments of this application;

[0038] Figure 5 A schematic diagram illustrating the training process of a first model provided in an embodiment of this application;

[0039] Figure 6 A schematic diagram illustrating the training process of a second model provided in an embodiment of this application;

[0040] Figure 7 This is a schematic diagram of the structure of a communication device provided in an embodiment of this application;

[0041] Figure 8 This is a schematic diagram of another communication device provided in an embodiment of this application. Detailed Implementation

[0042] The technical solutions of the embodiments of this application will be described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, and not all embodiments. The terms "first," "second," etc., are used for descriptive purposes only, and the features of "first," "second," etc., may explicitly or implicitly include one or more of the features. In the description of this embodiment, unless otherwise stated, "multiple" means two or more.

[0043] In embodiments of this application, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Without further limitation, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes that element.

[0044] In the embodiments of this application, "at least one (item)" refers to one or more. "More than one" refers to two or more. "At least two (items)" refers to two or three or more. "And / or" is merely a description of the relationship between related objects, indicating that three relationships can exist. For example, A and / or B can represent: A alone, A and B simultaneously, and B alone, where A and B can be singular or plural. In addition, the character " / " in this document generally indicates that the related objects before and after are in an "or" relationship. "At least one (item) of the following" or similar expressions refer to any combination of these items, including any combination of single or plural items. For example, at least one (item) of a, b, or c can represent: a, b, c, "a and b", "a and c", "b and c", or "a and b and c", where a, b, and c can be single or multiple. Both "...when" and "if" indicate that a corresponding action will be taken under certain objective circumstances. They are not time limits, nor do they require a judgment action to be taken when the action is taken, nor do they imply any other limitations.

[0045] In the embodiments of this application, the terms "exemplary" or "for example" are used to indicate that something is an example, illustration, or description. Any embodiment or design that is described as "exemplary" or "for example" in the embodiments of this application should not be construed as being more preferred or advantageous than other embodiments or design. Specifically, the use of terms such as "exemplary" or "for example" is intended to present the relevant concepts in a specific manner to facilitate understanding.

[0046] In this application, "sending information to...(terminal)" can be understood as the destination of the information being the terminal. This can include sending information to the terminal directly or indirectly. "Receiving information from...(terminal)" can be understood as the source of the information being the terminal, and can include receiving information from the terminal directly or indirectly. The information may undergo necessary processing between the source and destination, such as format changes, but the destination can understand the valid information from the source.

[0047] In the description of this application, it should be noted that, unless otherwise expressly specified and limited, the terms "set" and "connection" should be interpreted broadly, for example, they can refer to a fixed connection, a detachable connection, or an integral connection; those skilled in the art can understand the specific meaning of the above terms in this application according to the specific circumstances.

[0048] To ensure clarity and conciseness in the description of the following embodiments, a brief introduction to the relevant concepts or technologies is given first:

[0049] 1. Artificial Intelligence

[0050] Artificial intelligence (AI) is the ability of computers to perform tasks that originally required human intelligence. Its core is to endow computers with abilities such as perception, learning, reasoning, and decision-making, thereby simulating human thought and behavior. AI is the simulation of human intelligent behavior through computer programs; it can be understood as the intelligence exhibited by machines created by humans. The goals of AI include understanding intelligence by constructing computer programs that demonstrate symbolic reasoning or reasoning.

[0051] 2. Machine Learning

[0052] Machine learning (ML) is an implementation of artificial intelligence. It's a method that endows machines with the ability to perform functions that are impossible through direct programming. In practical terms, machine learning trains models with data, enabling computers to learn patterns from data and make accurate predictions or decisions on new data. There are many methods of machine learning, such as neural networks (NNs), decision trees, and support vector machines. Machine learning theory primarily involves designing and analyzing algorithms that allow computers to learn automatically. Machine learning algorithms are a class of algorithms that automatically analyze data to obtain patterns and use these patterns to predict unknown data.

[0053] 3. AI Model

[0054] An AI model is an algorithm or computer program that enables AI functionality. An AI model represents the mapping relationship between the model's input and output; in other words, an AI model is a function model that maps an input of a certain dimension to an output of a certain dimension. The parameters of this function model can be obtained through machine learning training. For example, f(x) = ax 2 +b (a≠0) is a quadratic function model, which can be viewed as an AI model, where a and b are the parameters of this AI model, and a and b can be obtained through machine learning training. In this application, the AI ​​model can also be called an AI / ML model; the two are equivalent and interchangeable.

[0055] This application provides a positioning method based on an AI model. The positioning method can be applied to communication systems, which may include, but are not limited to, wireless communication systems, such as narrowband internet of things (NB-IoT), global system for mobile communications (GSM), enhanced data rate for GSM evolution (EDGE), wideband code division multiple access (WCDMA), code division multiple access 2000 (CDMA2000), time division-synchronization code division multiple access (TD-SCDMA), long term evolution (LTE), the 5th generation (5G), and future systems.

[0056] The scenarios for which this communication system is applicable may include, but are not limited to: terrestrial cellular communication, non-terrestrial network (NTN), satellite communication, high altitude platform station (HAPS) communication, vehicle-to-everything (V2X) communication, integrated access and backhaul (IAB) communication, and reconfigurable intelligent surface (RIS) communication.

[0057] Figure 1 This is a schematic diagram of the architecture of a communication system provided in an embodiment of this application. Figure 1 As shown, the communication system of this application may include: a terminal 100, an access network device 101, and a core network device 102. The terminal 100 can communicate with the access network device 101, and the access network device 101 can communicate with the core network device 102.

[0058] Terminal 100 may include one or more devices. Terminal 100 is a device with wireless transceiver capabilities. Terminal 100 can be a wireless terminal or a wired terminal. A wireless terminal can be a device providing voice and / or other service data connectivity to a user, a handheld device with wireless connectivity, or other processing devices connected to a wireless modem. The wireless terminal can communicate with one or more core networks via the RAN. The wireless terminal can be a mobile terminal, such as a mobile phone (or "cellular" phone), and a computer with a mobile terminal, for example, a portable, pocket-sized, handheld, computer-embedded, or vehicle-mounted mobile device, which exchanges voice and / or data with the wireless access network. A wireless terminal can also be referred to as a system, subscriber unit, subscriber station, mobile station (MS), mobile station, remote station, remote terminal, access terminal, user terminal, user agent, user device or user equipment, user equipment (UE), terminal unit, terminal station, remote station, mobile device, terminal, wireless communication equipment, terminal agent, or terminal device, etc., without limitation.

[0059] Access network device 101 may include one or more devices. Access network device 101 may also be referred to as radio access network device, radio access network (RAN), radio access network function, or radio access network unit. Access network device 101 may include base stations, WLAN access points, or WiFi nodes, etc. A base station may be referred to as Node B, evolved Node B (eNB), access point, base transceiver station (BTS), radio base station, radio transceiver, basic service set (BSS), extended service set (ESS), home B node, home evolved B node, transmitting receiving point (TRP), or any other suitable term in the field, provided that the same technical effect is achieved. The base station is not limited to specific technical terms.

[0060] In one network architecture, the access network device 101 may include a centralized unit (CU) node, or a distributed unit (DU) node, or a RAN device including CU nodes and DU nodes, or a RAN device including a control plane CU node (CU-CP node), a user plane CU node (CU-UP node), and a DU node.

[0061] Core network equipment 102 refers to equipment in the core network (CN) that provides service support for terminal equipment. Core network equipment can be mobility management equipment. Core network equipment may include, but is not limited to, at least one of the following: core network node, core network function, mobility management entity (MME), access and mobility management function (AMF), session management function (SMF), location management function (LMF), user plane function (UPF), policy control function (PCF), policy and charging rules function (PCRF), edge application server discovery function (EASDF), unified data management (UDM), unified data repository (UDR), home subscriber server (HSS), centralized network configuration (CNC), network repository function (NRF), network exposure function (NEF), local NEF (or L-NEF), binding support function (BSF), application function (AF), etc.

[0062] The embodiments of this application can be applied to terminal positioning scenarios. Currently, terminal positioning scenarios are very widespread, mainly including the following scenarios:

[0063] 1. Vehicle navigation scenario: It can track the vehicle's location in real time, helping drivers plan the best route and avoid congestion and getting lost.

[0064] 2. Vehicle logistics scenario: Real-time monitoring of vehicle transportation trajectory, driving route and other information.

[0065] 3. Logistics and warehousing scenarios: It can monitor the location and storage status of goods in real time, so as to dispatch them in a timely manner and ensure the safety of goods.

[0066] 4. Outdoor exploration scenarios: It can be applied to outdoor exploration activities such as mountaineering and hiking to ensure personal safety.

[0067] The above scenarios require the terminal to have positioning capabilities, enabling real-time determination of the terminal's location and acquisition of its location information. Therefore, research on terminal positioning is becoming increasingly widespread.

[0068] In the Rel-19 AI / ML localization project, the terminal can use artificial intelligence / machine learning (AI / ML) to obtain its location information. AI / ML localization can include two methods: direct AI / ML (D-AI / ML) localization and AI / ML assisted positioning (A-AI / ML) localization. D-AI / ML localization focuses more on extracting useful information from raw data and making direct predictions; while A-AI / ML localization focuses more on using machine learning models to optimize and adjust traditional methods (such as preliminary localization results based on signal strength, geometric calculations, or rule matching) to make them closer to the true location.

[0069] When a terminal uses AI / ML for positioning, it's crucial to ensure consistent configuration between the terminal and the access network equipment to maintain overall coordination and accuracy of the positioning service. Simultaneously, the terminal needs to understand how to use AI / ML positioning methods for inference. This includes mastering key steps such as data preprocessing, model selection, and parameter adjustment to ensure the effectiveness and reliability of the positioning process. Furthermore, the terminal must explicitly indicate to the LMF whether its positioning information was generated using an AI / ML model. This step is essential for the LMF's subsequent processing of positioning data, evaluation of positioning quality, and collaboration with other network functions. During the 118th meeting of Working Group 1 (WG1 Meeting #118) of the 3rd Generation Partnership Project (3GPP), Protocol 15 proposed that the LMF provide the terminal with location request information and report configuration. Specifically, the LMF generates and sends a location request message to the terminal, explicitly informing the terminal of the required positioning method to support AI / ML positioning inference.

[0070] For example, deploying an AI model in the terminal allows for location tracking, such as... Figure 2 The diagram illustrates the deployment of an AI model for localization on a terminal device. Figure 2 As shown, taking a terminal, base station, and LMF as examples, the terminal deploys an AI model. The AI ​​model can be positioned using D-AI / ML or A-AI / ML. For instance, the base station sends a positioning reference signal (PRS) to the terminal. The terminal receives and parses the PRS information, obtains the specific configuration parameters of the PRS resources (such as the frequency domain location, time domain period, subcarrier spacing, etc.), and synchronizes them to the corresponding time and frequency domain locations, receiving the PRS signal. After processing the PRS signal, the terminal obtains PRS data, inputs the PRS data into the AI ​​model, and the AI ​​model outputs the terminal's current location information. Further, the terminal sends location information to the base station. After receiving the terminal's location information, the base station forwards it to the LMF. Upon receiving the location information from the base station, the LMF performs a series of processes based on this location information to support location services and related applications in the network. For example, as a core positioning function entity in the 5G network, the LMF performs various processing on the received location information, including verification, optimization, storage, distribution, location service support, network optimization, security processing, and coordination with other network functions. Through these processes, LMF is able to provide reliable and efficient support for a variety of location-based services and applications.

[0071] Typically, when an AI model is deployed on a terminal, the two most common methods for locating the model are as follows.

[0072] The first method: triangular positioning.

[0073] Among them, the trilateration method refers to the method by which the terminal determines its location information through geometric calculation based on the distances from multiple known reference points (such as base stations and Wi-Fi access points) to the terminal.

[0074] Among them, the triangulation method refers to the terminal determining its position information through geometric calculation based on angle measurements (such as angle of arrival (AOA)) and the position of reference points.

[0075] Taking the trilateration method as an example, for instance, such as Figure 3 The diagram illustrates the trilateration method. This method uses three known base stations (A, B, and C) and measurements such as Received Signal Strength Indicator (RSSI), Time of Arrival (TOA), and Time Difference of Arrival (TDOA) to determine the distance d from base station A to terminal D. a Size, distance d from base station B to terminal D b Size, distance d from base station C to terminal D c The size of the value is used to determine the position information of terminal D by using geometric intersections (such as the intersections of circles).

[0076] Triangulation and triangulation are two common positioning methods, but they have significant drawbacks in terms of positioning accuracy, robustness, and computational complexity. For example, triangulation methods are typically based on distance measurements (such as estimating distance using RSSI or TOA), using only distance information to determine location; triangulation methods are typically based on angle measurements (such as using AOA), using only angle information to determine location. Therefore, triangulation and triangulation methods cannot combine multiple different types of data; they can only rely on the accuracy of a single type of data (such as distance or angle), placing high demands on the precision of the hardware.

[0077] The second method: fingerprint positioning.

[0078] For example, fingerprint localization methods are divided into an offline acquisition stage and an online localization stage.

[0079] The offline data acquisition phase includes: the terminal selecting several location points within the target positioning area and acquiring signal features corresponding to each location point. These signal features may include RSSI, Channel State Information (CSI), and AOA, etc. The terminal then converts these signal features into numerical vectors, forming fingerprints. Each location point's fingerprint (i.e., numerical vector) is mapped one-to-one with its location coordinates, generating a fingerprint database. Essentially, the fingerprint database stores the fingerprints (i.e., numerical vectors) of several location points and their corresponding location coordinates.

[0080] The online positioning phase includes: the terminal acquiring signal features of the current location in real time and converting them into numerical vectors to form a real-time fingerprint; then, matching the real-time fingerprint with a fingerprint database, and outputting the current location coordinates based on the matching result. Methods such as the k-order nearest neighbor algorithm can be used to match the real-time fingerprint with the fingerprint database, thereby obtaining the location coordinates corresponding to the real-time fingerprint.

[0081] However, when the target location area is large, the number of location points increases, leading to a larger data volume in the fingerprint database and consequently, a larger AI model. Consequently, the data processing time during fingerprint matching in the online positioning phase becomes longer. Therefore, fingerprint positioning methods are typically used for indoor positioning.

[0082] For example, when the target positioning area is large, the terminal can combine a deep convolutional neural network to divide the target positioning area into several sub-regions before positioning.

[0083] For example, during the offline data acquisition phase, the terminal divides the target location area in the spatial coordinate system into several sub-regions at equal intervals along the x and y directions, and assigns a different number to each sub-region. Then, the terminal selects several location points within each sub-region and measures and collects the fingerprint matrix of each location point. Finally, using the fingerprint matrix of the location points as input and the number of the sub-region where the location point is located as output, a convolutional neural network model is trained.

[0084] For example, the terminal can generate a fingerprint database by mapping the location coordinates of several location points to their sub-region numbers. This fingerprint database is used for subsequent positioning.

[0085] During the online positioning phase, the terminal measures and collects the fingerprint matrix of the current location in real time. Then, the fingerprint matrix of the current location is input into a trained convolutional neural network model, which outputs the predicted sub-region number. Next, the weighted k-nearest neighbors (WKNN) algorithm is used to match the fingerprint matrix of the current location with the aforementioned fingerprint database to obtain the coordinates of the current location.

[0086] However, the fingerprint localization method's use of convolutional neural networks for sub-region segmentation may introduce boundary issues. Furthermore, the method involves multiple processing stages (such as model training and terminal localization using the model), the additional use of WKNN matching, and its dependence on a fingerprint database, which may lead to deficiencies in overall localization accuracy, real-time performance, and system maintainability.

[0087] Based on this, the embodiments of this application provide a positioning method and apparatus based on an AI model. The terminal obtains PRS data by measuring the PRS signal and performs quantization processing on the PRS data, unifying the various parameters in the PRS data to the same dimension, thereby reducing the computing resources and storage space of the AI ​​model on the terminal and thus improving positioning accuracy.

[0088] The following embodiments of this application will be used to illustrate the concept of having Figure 1 Taking the terminal 100, access network device 101, and core network device 102 of the structure shown as examples, the positioning method provided in this application will be described in detail with reference to the accompanying drawings and application scenarios.

[0089] Please see Figure 4a , Figure 4a This is a schematic diagram illustrating the interaction process of an AI model-based localization method provided in an embodiment of this application. The method is applied to a terminal, access network equipment, and core network equipment, wherein the terminal can be... Figure 1 The terminal or device within the terminal, network equipment can be Figure 1 The access network equipment or devices within the access network equipment, the core network equipment may be Figure 1 The core network equipment or devices within the core network equipment. Taking LMF as an example, such as... Figure 4a As shown, the specific steps of this application embodiment are as follows.

[0090] S401. The access network device sends a PRS signal to the terminal. Correspondingly, the terminal receives the PRS signal from the access network device.

[0091] Among them, the PRS signal is used by the terminal to achieve positioning.

[0092] For example, a terminal receives a PRS signal from an access network device and processes it via an antenna and a receiving link. For instance, considering that the PRS signal received by the terminal may be affected by factors such as noise and attenuation, the terminal needs to preprocess the PRS signal. For example, preprocessing includes denoising and synchronization. Denoising refers to removing noise and interference from the signal to improve the purity of the PRS signal. For example, the terminal can use a filter to remove noise from the PRS signal. Synchronization refers to aligning the PRS signal in the time or frequency domain.

[0093] In one implementation, such as Figure 4b As shown, S401 above, namely, the access network device sending a PRS signal to the terminal, or the terminal receiving a PRS signal from the access network device, specifically includes:

[0094] S401a: The terminal sends a location request to the access network device. Correspondingly, the access network device receives the location request from the terminal.

[0095] For example, a location request can be used to request access network devices to allocate PRS resources to support the terminal's location. For example, a location request can also be a message or an instruction; the type of location request is not limited. For instance, a location request can also be described as a location message, a location request message, or a location request instruction, etc., which will not be elaborated further.

[0096] For example, a location request may include any one of the following 1 to 4:

[0097] 1. Locate the type of request.

[0098] For example, the type of location request may include an identifier for the location request that explicitly indicates that the location request is a location request message. For instance, the location request may be represented as: LTE positioning protocol ancillary user equipment positioning information request (LPPa UE Positioning Information Request).

[0099] 2. Terminal capability information.

[0100] For example, the terminal's capability information may include the PRS frequency bands supported by the terminal, bandwidth capabilities, antenna capabilities, positioning capabilities, etc.

[0101] The PRS frequency band supported by the terminal can be understood as the operating frequency band of the PRS supported by the terminal. For example, the PRS frequency band supported by the terminal may include: Sub-6GHz band (frequency range 1, FR1), millimeter wave band (frequency range 2, FR2), etc., without limitation.

[0102] Among them, bandwidth capability is used to indicate the maximum PRS bandwidth that the terminal can receive.

[0103] Antenna capability is used to indicate the antenna configuration of the terminal, which may affect beamforming and reception capabilities. For example, the antenna configuration of the terminal may include: a single transmit dual receive antenna configuration (1 transmit 2 receive, 1T2R), a dual transmit quad receive antenna configuration (2 transmit 4 receive, 2T4R), etc., without limitation.

[0104] The positioning capability is used to indicate the positioning technologies supported by the terminal. For example, the positioning technologies supported by the terminal may include: observed time difference of arrival (OTDOA), uplink time difference of arrival (UL-TDOA), multi-round trip time (Multi-RTT), etc., without limitation.

[0105] 3. Measurement requirements.

[0106] For example, measurement requirements may include positioning accuracy requirements, measurement type requirements, measurement cycle requirements, etc.

[0107] The positioning accuracy requirement indicates the desired positioning accuracy of the terminal. For example, the desired positioning accuracy may include meter-level, sub-meter-level, etc., without limitation.

[0108] The measurement type requirement indicates the type of measurement requested by the terminal. For example, the measurement type requested by the terminal may include: measuring the reference signal time difference (RSTD), AOA, and TOA, etc., without limitation.

[0109] The measurement period requirement is used to indicate the measurement period desired by the terminal. For example, the desired measurement period may include periodic measurement, single measurement, etc., and is not limited thereto.

[0110] 4. PRS configuration.

[0111] For example, PRS configuration may include PRS resource set ID, PRS resource ID, time-frequency resource requirements, repetition factor, etc.

[0112] The PRS resource set ID is used to indicate the PRS resource set identifier requested by the terminal.

[0113] Among them, the PRS resource ID is used to indicate the PRS resource identifier requested by the terminal.

[0114] The time-frequency resource requirement is used to indicate the terminal's desired PRS time-frequency resource allocation. For example, the terminal's desired PRS time-frequency resource allocation may include: subcarrier spacing, number of symbols, etc., and is not limited.

[0115] The repeatability factor is used to indicate the desired number of PRS repetitions at the terminal, and it is used to improve measurement accuracy.

[0116] It should be noted that the above are merely some examples of the content of location requests and do not constitute a limitation on this application. In practical applications, location requests may also include other content, which will not be elaborated here.

[0117] S401b: The access network device sends PRS information to the terminal based on the positioning request. Correspondingly, the terminal receives the PRS information from the access network device.

[0118] Among them, PRS information is used to provide the configuration parameters required for the terminal to receive PRS signals.

[0119] For example, PRS information includes PRS resource set ID, PRS resource ID, time-frequency location, period, repetition factor, power control parameters, etc., and is not limited to these.

[0120] S401c: The terminal parses the PRS information to obtain the PRS resources.

[0121] The PRS resource is used to indicate the specific allocation of the PRS signal in the time domain. For example, the PRS resource may include resources such as time, frequency, and code domain.

[0122] For example, a PRS resource consists of a PRS resource set and a PRS resource. The PRS resource set indicates a group of PRS resources with the same attributes (such as period and bandwidth); the PRS resource indicates a specific time-frequency resource unit used to carry the PRS signal.

[0123] S401d: The access network device sends a PRS signal to the terminal based on the PRS resource. Correspondingly, the terminal receives the PRS signal from the access network device based on the PRS resource.

[0124] After the terminal receives the PRS signal, such as Figure 4a As shown, the method also includes:

[0125] S402. The terminal measures the PRS signal and obtains PRS data.

[0126] The PRS data includes at least the following: line-of-sight (LOS) confidence, RSTD, and UErxtx time difference (RXTX), without limitation.

[0127] It should be noted that the PRS data in S402 is only an example. In actual applications, the PRS data may also include other data, such as reference signal received power (RSRP) and Doppler shift, without limitation.

[0128] In addition, PRS data may also include the size of the positioning area and the signal-to-noise ratio obtained from PRS information. Optionally, PRS data may also include the target location, etc., without limitation.

[0129] The LOS confidence level indicates the degree of confidence that the PRS signal propagates through a line-of-sight path. Alternatively, the LOS confidence level indicates the probability that the communication link between the terminal and the access network equipment propagates through a line-of-sight path.

[0130] LOS confidence level can be represented as P LOS P LOS The value range is usually between 0 and 1. P LOS The closer the value is to 1, the greater the likelihood that the communication link between the terminal and the access network equipment is a line-of-sight propagation path. Correspondingly, P... LOS The closer the value is to 0, the greater the likelihood that the PRS signal propagates along a non-line-of-sight (NLOS) path between the terminal and the access network equipment. Both LOS and NLOS are important aspects of channel characteristics. LOS indicates that there are no obstructions in the PRS signal propagation path, allowing the PRS signal to propagate in a straight line. NLOS indicates that there are obstructions in the PRS signal propagation path, requiring the PRS signal to propagate through reflection, diffraction, or by penetrating obstacles.

[0131] Optional, P LOS It can be a pre-set value (i.e., a preset value), such as a value pre-defined by the terminal, a value pre-defined by the access network device, or a value pre-defined by the protocol, etc., without limitation.

[0132] Optional, PLOS This can be calculated using information such as RSSI and channel characteristics. For example, when the RSSI value is greater than -50dBm, the probability that the communication link between the terminal and the access network device is in a line-of-sight state is 80%. Therefore, when the actually measured RSSI value is greater than -50dBm, P can be... LOS Set it to 0.8.

[0133] RSTD is used to indicate the time difference between the arrival of different PRS signals at the terminal.

[0134] RXTX is used to indicate the time difference between the terminal receiving the PRS signal and sending the uplink reference signal.

[0135] S403. The terminal performs quantization processing on the PRS data to obtain the first data associated with the PRS data.

[0136] The first data includes the first parameter, the second parameter, and the third parameter.

[0137] The first parameter can be obtained by the terminal quantifying the LOS confidence level in the PRS data, the second parameter can be obtained by the terminal quantifying the RSTD in the PRS data, and the third parameter can be obtained by the terminal quantifying the RXTX in the PRS data; the first, second, and third parameters are parameters of the same dimension.

[0138] It should be noted that the correspondence between the first, second, and third parameters and the various data in the PRS data is only one example. In practical applications, the first parameter can also be obtained by the terminal quantizing the RSTD in the PRS data; the second parameter can also be obtained by the terminal quantizing the RXTX in the PRS data; the third parameter can also be obtained by the terminal quantizing the LOS confidence level in the PRS data, or there can be other correspondences, which will not be listed here.

[0139] For example, in this embodiment, quantization processing refers to the standardization of PRS data by the terminal. Standardization processing can also be understood as scaling the data proportionally to make it fall into a specific range or transforming it to a fixed range. That is, in this embodiment, standardization processing can be scaling the PRS data proportionally to transform it to a fixed range (e.g., between 0 and 255). In other words, in this embodiment, quantization processing (or standardization processing) unifies the various data included in the PRS data to the same dimension, eliminating the dimensional differences between the various parameters included in the PRS data.

[0140] In the embodiments of this application, quantization processing may include various processing methods, such as text data quantization, categorical data quantization, normalization processing, etc., and is not limited thereto. Optionally, quantization processing may also include RGB data quantization. For example, RGB data quantization refers to converting PRS data into RGB numerical form to facilitate AI model training and improve the accuracy of the AI ​​model.

[0141] For example, the terminal can use RGB data quantization to quantize PRS data into RGB data. Based on this, the terminal can quantize the LOS confidence measure in the PRS data into red pixel information R in the RGB data; the terminal can quantize the RSTD in the PRS data into green pixel information G in the RGB data; and the terminal can quantize the RXTX in the PRS data into blue pixel information B in the RGB data.

[0142] In other words, the first parameter can be the red pixel information R in the RGB data, the second parameter can be the green pixel information G in the RGB data, and the third parameter can be the blue pixel information B in the RGB data. For ease of description, the following embodiments will all use RGB data as the first data as an example for illustration, and will not be repeated here.

[0143] For example, the terminal can use the following quantization formula (1) to quantize the LOS confidence level in the PRS data into the red pixel information R in the RGB data.

[0144]

[0145] Among them, P LOS For LOS confidence, P LOS ∈[0,1],P LOS The relevant descriptions can be found in the above embodiments, and will not be repeated here.

[0146] Here, β is the distribution coefficient, used to indicate the complexity of the PRS signal propagation environment. Different β values ​​can be selected based on different channel characteristics.

[0147] For example, in open areas, there are fewer obstructions along the PRS signal propagation path, increasing the likelihood that the communication link between the terminal and the access network equipment propagates along the LOS path. LOS The larger the value of β, the lower the value can be. Correspondingly, in densely populated urban areas, there are more obstructions along the PRS signal propagation path, resulting in a more significant multipath effect during PRS signal propagation. This increases the likelihood that the communication link between the terminal and the access network equipment will propagate along an NLOS path, i.e., P... LOS The smaller the value, the higher the β can be. β is generally an empirical value.

[0148] The `round()` function rounds the value within parentheses to the nearest integer. For example, `round(10.6)` rounds 10.6 to the nearest integer, such as 11. `round(10.2)` rounds 10.6 to the nearest integer, such as 10. `round(2.5)` might return 2 in some implementations because the `round()` function typically rounds a number to an even value when the value is exactly the midpoint between two integers.

[0149] For example, the terminal can use the following quantization formula (2) to quantize the RTSD in the PRS data into the green pixel information G in the RGB data.

[0150]

[0151] RTSD stands for Reference Signal Time Difference, used to indicate the time difference between the arrival times of different PRS signals at the terminal. max The maximum value of the time difference of the reference signal, RTSD min The minimum time difference of the reference signal is indicated by round(), which rounds the value within the parentheses to the nearest integer.

[0152] For example, the terminal can use the following quantization formula (3) to quantize RXTX in PRS data into blue pixel information B in RGB data.

[0153]

[0154] Wherein, RXTX is the terminal receive-transmit time difference, used to indicate the time difference between the terminal receiving the PRS signal and transmitting the uplink reference signal. min RXTX represents the minimum time difference between the terminal's receive and transmit signals. max This represents the maximum value of the time difference between the terminal's reception and transmission. `round()` indicates that the value within the parentheses is rounded to the nearest integer.

[0155] In summary, based on this solution, quantization processing can transform PRS data from different dimensions to the same dimension. For example, quantization processing can map LOS confidence, RSTD, and RXTX to the R, G, and B pixel information channels respectively, thus transforming data from different dimensions into a single dimension. Quantization processing allows LOS confidence, RSTD, and RXTX to be used more fairly and effectively for subsequent AI model training.

[0156] Compared to fingerprint positioning methods, the AI ​​model-based positioning method (also known as the RGB positioning method) provided in this application does not require manual feature selection or dimensionality reduction processing. Moreover, it is not affected by subjective factors and will not miss features. It can automatically learn features from training data through AI models and obtain the correlation between training data, thereby reducing costs and errors.

[0157] Furthermore, when using the RGB positioning method to predict the terminal's location information, it is not necessary to match real-time fingerprints with a fingerprint database as in fingerprint positioning methods. Moreover, the time complexity of the RGB positioning method does not increase linearly with the size of the fingerprint database. The RGB positioning method only needs to input the RGB image into the AI ​​model to determine the terminal's location information. Whether a single RGB image or multiple RGB images are input into the AI ​​model, the output of the AI ​​model (i.e., the terminal's location information) can be obtained quickly. This results in fast positioning speed with minimal computational and storage requirements, fulfilling the need for high-precision positioning.

[0158] It should be noted that the terminal can be equipped with multiple antennas, each of which is used to receive one PRS signal. After receiving a PRS signal through one antenna, the terminal measures the PRS signal to obtain PRS data, and then quantizes the PRS data into RGB data. In other words, one antenna corresponds to one RGB data.

[0159] For example, the multiple antennas set in the terminal can form one or more antenna arrays. Taking one antenna array as an example, since one antenna corresponds to one RGB data, one antenna array corresponds to multiple RGB data, and multiple RGB data can form an RGB image. That is, one antenna array corresponds to one RGB image.

[0160] For example, by mapping RGB data to RGB images, a two-dimensional arrangement structure can be designed so that LOS confidence, RSTD, and RXTX exhibit local correlation on the RGB image, which is helpful for the training of subsequent AI models.

[0161] S404. The terminal inputs the first data into the AI ​​model to determine the terminal's location information.

[0162] It is understandable that the first data is obtained by the terminal after quantizing the PRS data. For example, the first data can be RGB data. That is, the terminal inputs RGB data into the AI ​​model to obtain the terminal's location information.

[0163] Based on this scheme, the LOS confidence, RSTD, and RXTX in PRS data are mapped to the three pixel information channels of red pixel information R, green pixel information G, and blue pixel information B through quantization processing. This transforms data from different dimensions to the same dimension, which helps the AI ​​model learn the complementary relationships between various data in RGB data, enabling the AI ​​model to learn the features of RGB data more fully.

[0164] Understandably, since one antenna corresponds to one RGB data point, and one antenna array corresponds to multiple RGB data points, these multiple RGB data points can form an RGB image. Based on this, the terminal can input an RGB image into an AI model to determine the terminal's location information. For example, by inputting an RGB image into an AI model to locate the terminal, its movement trajectory can be determined.

[0165] In this embodiment, by mapping RGB data to an RGB image, a two-dimensional arrangement structure can be designed, allowing LOS confidence, RSTD, and RXTX to exhibit local correlation on the RGB image. This local correlation makes it easier to extract local features using AI models, thus enabling AI models to represent the relationship between input and output without requiring overly deep or complex structures. Understandably, the AI ​​model structure obtained in this way is relatively simple.

[0166] Considering that different scenarios have different positioning accuracy requirements, higher positioning accuracy requires more PRS data to achieve the desired precision. However, more PRS data also means more RGB data, resulting in a larger RGB image size. This leads to increased data processing load for the AI ​​model, longer training time, and lower computational efficiency. Therefore, when inputting RGB images into the AI ​​model, the size of the RGB image can be determined to avoid excessively large RGB image sizes that would cause excessive data processing load, longer training time, and lower computational efficiency.

[0167] For example, an AI model may include a first model and a second model. (e.g.) Figure 4c As shown, the terminal inputs the RGB image into the AI ​​model to determine the terminal's location information, which may include the following steps, i.e., S404 may include the following steps:

[0168] S4011. The terminal determines whether the size of the RGB image is less than the threshold.

[0169] The threshold can be an empirical value preset by the terminal. In this application embodiment, the specific value of the threshold is not limited, and the actual setting shall prevail.

[0170] For example, if the size of the RGB image is less than the threshold, step S4012 is executed; if the size of the RGB image is greater than or equal to the threshold, steps S4013 and S4014 are executed.

[0171] S4012. The terminal inputs the RGB image into the first model to determine the terminal's location information.

[0172] The first model is used to predict the location information of the terminal, that is, the output of the first model is the location information of the terminal.

[0173] Understandably, the first model takes an RGB image as input and outputs the terminal's location information.

[0174] S4013. The terminal inputs the RGB image into the second model for dimensionality reduction processing to obtain the dimensionality-reduced RGB image.

[0175] The second model is used to perform dimensionality reduction on RGB images.

[0176] Understandably, the second model takes an RGB image as input and outputs a dimensionality-reduced RGB image.

[0177] Dimensionality reduction is used to reduce the dimensionality (number of features) of data, mapping high-dimensional data to a low-dimensional space. For example, dimensionality reduction can map an input RGB image to a low-dimensional RGB image (the dimensionality-reduced RGB image).

[0178] For example, dimensionality reduction can reduce computational complexity and compress storage space while retaining important information.

[0179] It is understood that dimensionality reduction can include various methods, such as feature extraction, feature selection, matrix factorization, and deep learning methods, and the embodiments of this application do not limit this.

[0180] S4014. The terminal inputs the dimensionality-reduced RGB image into the first model to determine the terminal's location information.

[0181] Understandably, the input to the first model is a dimensionality-reduced RGB image, and the output is the terminal's location information.

[0182] This application embodiment, based on the use of the first model for localization, adds a second model for dimensionality reduction. When the size of the RGB image is larger than a threshold, the RGB image is input into the second model for dimensionality reduction, and then the dimensionality-reduced result is input into the first model for localization, which can reduce the amount of data processing.

[0183] Furthermore, different scenarios have different positioning accuracy requirements, and the corresponding number of antenna arrays also varies. According to the embodiments of this application, the terminal can pre-set multiple second models based on different numbers of antenna arrays during the training phase. Then, during the usage phase, the corresponding second model can be input based on the known number of antenna arrays to achieve dimensionality reduction processing of the RGB image, improving computational efficiency.

[0184] Optional, such as Figure 4d As shown, the method also includes:

[0185] S405. The terminal sends its location information to the access network device, and the access network device receives the location information from the terminal accordingly.

[0186] For example, location information can be carried in radio resource control (RRC) messages such as RRC setup complete message and RRC reconfiguration complete message, without limitation.

[0187] Optionally, after performing step S405, steps S406 and S407 can also be performed.

[0188] S406. The terminal sends an instruction message to the access network device. Correspondingly, the access network device receives the instruction message from the terminal.

[0189] The indication information is used to instruct the access network equipment to send the terminal's location information to the LMF.

[0190] For example, the indication information may include the terminal's location information, terminal identifier, priority, etc.

[0191] The terminal identifier is used to uniquely identify the terminal. For example, the terminal identifier may include: subscription permanent identifier (SUPI), globally unique temporary identifier (5G-GUTI), international mobile subscriber identity (IMSI), or other unique identifiers, without limitation.

[0192] The priority is used to indicate the priority at which access network devices send terminal location information to the LMF. For example, the priority can include high, medium, low, etc., and is not limited.

[0193] S407. The access network device sends the terminal's location information to the LMF. Correspondingly, the LMF receives the terminal's location information from the access network device.

[0194] Based on the above embodiments, the terminal can deploy an AI model, inputting RGB data into the AI ​​model to achieve terminal positioning. Furthermore, since the terminal quantizes PRS data into RGB data, the various parameters in the PRS data can be unified to the same dimension, allowing the AI ​​model to learn the features of the RGB data more fully, reducing the computational resources and storage space required for the AI ​​model on the terminal.

[0195] In other words, in this embodiment, the AI ​​model has the function of converting RGB data into the location information of the terminal. Referring to the accompanying drawings, this embodiment also provides a training process for the AI ​​model, enabling it to output the terminal's location information based on RGB data.

[0196] It is understood that the AI ​​model in this application embodiment includes a first model and a second model. The training process of the first model and the second model will be described below.

[0197] For example, the training process of the first model includes:

[0198] 1.1 The access network device sends a PRS signal to the terminal. Correspondingly, the terminal receives the PRS signal from the access network device.

[0199] 1.2. The terminal measures the PRS signal to obtain PRS training data.

[0200] 1.3 The terminal performs quantization processing on the PRS training data to obtain RGB training data.

[0201] 1.4 The terminal generates RGB images from multiple RGB training data.

[0202] It should be noted that the specific implementation process of steps 1.1 to 1.4 above is similar to that of S401 to S404 in the above embodiments. For details, please refer to the relevant descriptions in the above embodiments, which will not be repeated here.

[0203] For example, such as Figure 5 As shown, the training process of the first model also includes:

[0204] 1.5. The terminal inputs the RGB image into the convolutional layer.

[0205] For example, such as Figure 5 As shown, the terminal can input the quantized RGB image or the dimensionality-reduced RGB image into the convolutional layer through the input layer.

[0206] Convolutional layers are used to extract features from RGB images. For example, a terminal can use convolutional layers to extract features from RGB images, such as edges and textures.

[0207] 1.6 The terminal inputs the output of the convolutional layer into the pooling layer.

[0208] It is understandable that the output of the convolutional layer is the features of the RGB image, such as edges and textures.

[0209] The pooling layer is used to reduce the dimensionality of the extracted features.

[0210] 1.7 The terminal inputs the output of the pooling layer into the fully connected layer.

[0211] It is understandable that the output of the pooling layer is the feature of the dimensionality-reduced RGB image, such as the dimensionality-reduced edges and textures.

[0212] The fully connected layer is used to map features to location coordinates.

[0213] 1.8 The terminal inputs the output of the fully connected layer into the next convolutional layer, pooling layer, and fully connected layer, repeating the operation multiple times.

[0214] Understandably, repeatedly performing operations on convolutional layers, pooling layers, and fully connected layers can enhance the feature learning ability of AI models, improve their generalization ability, accelerate the training process, and avoid overfitting.

[0215] 1.9 The terminal computing AI model uses the normalized mean squared error (NMSE) between the predicted location coordinates (terminal location information) output by the output layer and the target location coordinates to optimize the parameters in the AI ​​model using an adaptive moment estimation (Adam) optimizer or other optimizers, ultimately achieving the set NMSE requirement.

[0216] The optimizer is used to adjust the parameters of the AI ​​model (such as convolutional kernel weights, fully connected layer weights, etc.) during training to minimize NMSE.

[0217] For example, Adam is an adaptive learning rate optimization algorithm that combines the advantages of the adaptive gradient algorithm (Adagrad) and the root mean square propagation algorithm (RMSProp), and can adaptively adjust the learning rate of each parameter.

[0218] For example, the training process of the second model includes:

[0219] 2.1 The access network device sends a PRS signal to the terminal. Correspondingly, the terminal receives the PRS signal from the access network device.

[0220] 2.2 The terminal measures the PRS signal to obtain PRS training data.

[0221] 2.3 The terminal performs quantization processing on the PRS training data to obtain RGB training data.

[0222] 2.4 The terminal generates RGB images from multiple RGB training data.

[0223] It should be noted that the specific implementation process of steps 2.1 to 2.4 above can be referred to the specific implementation process of steps 1.1 to 1.4 above, and will not be repeated here.

[0224] For example, such as Figure 6 As shown, the training process of the second model also includes:

[0225] 2.5. The terminal inputs the RGB image into the convolutional layer.

[0226] For example, such as Figure 6 As shown, the terminal can input the quantized RGB image into the convolutional layer through the input layer.

[0227] Convolutional layers are used to extract features from RGB images. For example, a terminal can use convolutional layers to extract features from RGB images, such as edges and textures.

[0228] 2.6 The terminal inputs the output of the convolutional layer into the average pooling layer.

[0229] It is understandable that the output of the convolutional layer is the features of the RGB image, such as edges and textures.

[0230] The average pooling layer is used to reduce the dimensionality of the extracted features.

[0231] 2.7 The terminal inputs the output of the average pooling layer into the batch normalization layer.

[0232] It is understandable that the output of the average pooling layer is the feature of the dimensionality-reduced RGB image, such as the dimensionality-reduced edges and textures.

[0233] The batch normalization layer is used to normalize the mean and variance of each small batch of data, making the data distribution more stable, avoiding the problems of gradient vanishing or gradient explosion, and thus improving the convergence speed and stability of the model.

[0234] 2.8 The terminal will input the output of the batch normalization layer into the linear rectifier function to activate the layer.

[0235] The linear rectifier activation layer is used to maintain the linear relationship of the data.

[0236] 2.9 The terminal inputs the output of the linear rectified function activation layer into the next convolutional layer, average pooling layer, batch normalization layer, and linear rectified function activation layer, repeating the operation multiple times.

[0237] It is understandable that repeatedly using convolutional layers, average pooling layers, batch normalization layers, and linear rectified function activation layers can enable AI models to gradually extract more complex features, improve their expressive power, accelerate their training process, and enhance their generalization ability.

[0238] 2.10. The terminal inputs the output of the linear rectifier activation layer into the fully connected layer.

[0239] The fully connected layer is used to map features to location coordinates.

[0240] 2.11. The terminal will input the output of the fully connected layer into the discard layer.

[0241] Dropout layers are used to improve the model's generalization ability by reducing the risk of relying on specific neurons, forcing the network to learn stronger features, and reducing the risk of overfitting.

[0242] 2.12. The terminal will discard the output of the layer and input the batch normalization layer.

[0243] The batch normalization layer is used to prevent gradient vanishing or gradient explosion problems, allowing the network to use a larger learning rate, thereby accelerating convergence.

[0244] 2.13. The terminal will input the output of the batch normalization layer into the next fully connected layer, the discard layer, and the batch normalization layer, and repeat this operation.

[0245] Understandably, repeatedly performing operations on fully connected layers, dropout layers, and batch normalization layers can enhance the feature learning ability of AI models, improve their generalization ability, accelerate the training process, and avoid overfitting.

[0246] 2.14. The terminal outputs the dimensionality-reduced RGB image through the output layer.

[0247] It is understandable that the size of the output dimensionality-reduced RGB image meets the requirement of the input RGB image of the first model, that is, the size of the output dimensionality-reduced RGB image is less than the threshold.

[0248] Based on the above training process, the resulting AI models (including the first and second models) do not require manual feature selection. The AI ​​models automatically learn features from the training data, reducing costs and errors. During the AI ​​model usage phase, if the size of the RGB image is less than a threshold, the terminal can input the RGB image into the first model to determine its location. If the size of the RGB image is less than or equal to the threshold, the terminal can input the RGB image into the second model to obtain a dimensionality-reduced RGB image, and then input the dimensionality-reduced RGB image into the first model to determine the terminal's location, achieving rapid location of the terminal and improving computational efficiency.

[0249] It is understood that in the embodiments of this application, the executing entity may perform some or all of the steps in the embodiments of this application. These steps or operations are merely examples, and the embodiments of this application may also perform other operations or variations thereof. Furthermore, the steps may be executed in different orders as presented in the embodiments of this application, and it is not necessary to execute all the operations in the embodiments of this application.

[0250] The foregoing primarily describes the solutions provided in this application from the perspective of device-to-device interaction. It is understood that each device, in order to achieve the aforementioned functions, includes corresponding hardware structures and / or software modules for executing each function. Those skilled in the art should readily recognize that, based on the algorithm steps of the examples described in conjunction with the embodiments disclosed herein, this application can be implemented in hardware or a combination of hardware and computer software. Whether a function is executed in hardware or by computer software driving hardware depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of this application.

[0251] This application embodiment can divide each device into functional modules according to the above method example. For example, each function can be divided into a separate functional module, or two or more functions can be integrated into one processing module. The integrated module can be implemented in hardware or as a software functional module. It should be noted that the module division in this application embodiment is illustrative and only represents one logical functional division. In actual implementation, there may be other division methods.

[0252] When dividing each function into modules according to its corresponding function. Figure 7 A communication device 1700 is shown, which can perform the above-described... Figures 4a to 6The actions performed by the terminal and network device in the method shown, and all related content of each step involved in the above method embodiments, can be referenced from the functional description of the corresponding functional module. The technical effects that can be obtained can be referred to the above method embodiments, and will not be repeated here.

[0253] The communication device 1700 may include a transceiver module 1701 and a processing module 1702. Exemplarily, the communication device 1700 may be a communication equipment, or a chip or other combination device or component having the aforementioned communication device functions applied in a communication equipment. When the communication device 1700 is a communication equipment, the transceiver module 1701 may be a transceiver, which may include an antenna and radio frequency circuits, etc.; the processing module 1702 may be a processor (or processing circuit), such as a baseband processor, which may include one or more central processing units (CPUs). When the communication device 1700 is a component having the aforementioned communication device functions, the transceiver module 1701 may be a radio frequency unit; the processing module 1702 may be a processor (or processing circuit), such as a baseband processor. When the communication device 1700 is a chip system, the transceiver module 1701 may be an input / output interface of a chip (e.g., a baseband chip); the processing module 1702 may be a processor (or processing circuit) of the chip system, and may include one or more central processing units. It should be understood that the transceiver module 1701 in the embodiments of this application can be implemented by a transceiver or transceiver-related circuit components; the processing module 1702 can be implemented by a processor or processor-related circuit components (or, referred to as processing circuit).

[0254] For example, transceiver module 1701 can be used to perform... Figures 4a to 6 In the illustrated embodiment, all transmit and receive operations performed by the communication device, and / or other processes used to support the technology described herein; the processing module 1702 can be used to perform Figures 4a to 6 The embodiments shown include all operations performed by the communication device other than the transmit and receive operations, and / or other processes used to support the techniques described herein.

[0255] As another feasible approach Figure 7 The transceiver module 1701 can be replaced by a transceiver unit, which can integrate the functions of the transceiver module 1701; the processing module 1702 can be replaced by a processor, which can integrate the functions of the processing module 1702. Furthermore, Figure 7 The communication device 1700 shown may also include a memory.

[0256] This application embodiment also provides a method such as Figure 8The communication device 1800 shown can be a terminal or a chip or system-on-a-chip within a terminal; it can also be a network device or a chip or system-on-a-chip within a network device. For example... Figure 8 As shown, the communication device 1800 includes a processor 1801, a transceiver 1802, and a communication line 1803.

[0257] Furthermore, the communication device 1800 may also include a memory 1804. The processor 1801, the memory 1804, and the transceiver 1802 can be connected via a communication line 1803.

[0258] The processor 1801 can be a central processing unit (CPU), a general-purpose processor, a network processor (NP), a digital signal processor (DSP), a microprocessor, a microcontroller, a programmable logic device (PLD), or any combination thereof. The processor 1801 can also be other devices with processing capabilities, such as circuits, devices, or software modules, without limitation.

[0259] Transceiver 1802 is used to communicate with other devices or other communication networks. Other communication networks can be Ethernet, RAN, wireless local area networks (WLAN), etc. Transceiver 1802 can be a module, circuit, transceiver, or any device capable of enabling communication.

[0260] Communication line 1803 is used to transmit information between the components included in communication device 1800.

[0261] Memory 1804 is used to store instructions. These instructions can be computer programs.

[0262] The memory 1804 can be a read-only memory (ROM) or other type of static storage device that can store static information and / or instructions; it can also be a random access memory (RAM) or other type of dynamic storage device that can store information and / or instructions; it can also be an electrically erasable programmable read-only memory (EEPROM), a compact disc read-only memory (CD-ROM) or other optical disc storage, optical disc storage (including compressed optical discs, laser discs, optical discs, digital universal optical discs, Blu-ray discs, etc.), magnetic disk storage media or other magnetic storage devices, etc., without limitation.

[0263] It should be noted that the memory 1804 can exist independently of the processor 1801 or can be integrated with the processor 1801. The memory 1804 can be used to store instructions, program code, or some data, etc. The memory 1804 can be located inside or outside the communication device 1800, without limitation. The processor 1801 is used to execute the instructions stored in the memory 1804 to implement the communication method provided in the following embodiments of this application.

[0264] In one example, processor 1801 may include one or more CPUs, for example Figure 8 CPU0 and CPU1 in the CPU.

[0265] As an optional implementation, the communication device 1800 includes multiple processors, for example, besides Figure 8 In addition to processor 1801, it may also include processor 1807.

[0266] As an optional implementation, the communication device 1800 also includes an output device 1805 and an input device 1806. For example, the input device 1806 is a device such as a keyboard, mouse, microphone, or joystick, and the output device 1805 is a device such as a display screen or speaker.

[0267] It should be noted that the communication device 1800 can be a desktop computer, laptop computer, network server, mobile phone, tablet computer, wireless terminal, embedded device, chip system, or something else. Figure 8 Equipment with a similar structure. Furthermore... Figure 8 The structural composition shown does not constitute a limitation on the communication device, except... Figure 8In addition to the components shown, the communication device may include more or fewer components than illustrated, or combine certain components, or have different component arrangements.

[0268] In this embodiment of the application, the chip system may be composed of chips or may include chips and other discrete devices.

[0269] Furthermore, the actions, terms, etc., involved in the various embodiments of this application can be referenced interchangeably without limitation. The message names or parameter names in the messages exchanged between the various devices in the embodiments of this application are merely examples, and other names may be used in specific implementations without limitation.

[0270] This application also provides a computer program product that, when executed by a computer, can implement the functions of any of the above method embodiments.

[0271] This application also provides a computer program that, when executed by a computer, can implement the functions of any of the above method embodiments.

[0272] This application also provides a computer-readable storage medium. All or part of the processes in the above method embodiments can be implemented by a computer program instructing related hardware. This program can be stored in the computer-readable storage medium, and when executed, it can include the processes of the above method embodiments. The computer-readable storage medium can be an internal storage unit of the terminal (including a data sending end and / or a data receiving end) of any of the foregoing embodiments, such as the terminal's hard disk or memory. The computer-readable storage medium can also be an external storage device of the terminal, such as a plug-in hard disk, smart media card (SMC), secure digital (SD) card, flash card, etc., equipped on the terminal. Further, the computer-readable storage medium can include both the terminal's internal storage unit and external storage devices. The computer-readable storage medium is used to store the computer program and other programs and data required by the terminal. The computer-readable storage medium can also be used to temporarily store data that has been output or will be output.

[0273] In the several embodiments provided in this application, it should be understood that the disclosed apparatus and methods can be implemented in other ways. For example, the apparatus embodiments described above are merely illustrative; for instance, the division of modules or units is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another device, or some features may be ignored or not executed. Furthermore, the mutual coupling or direct coupling or communication connection shown or discussed may be through some interfaces; the indirect coupling or communication connection between devices or units may be electrical, mechanical, or other forms.

[0274] The units described as separate components may or may not be physically separate. A component shown as a unit can be one or more physical units; that is, it can be located in one place or distributed in multiple different locations. Some or all of the units can be selected to achieve the purpose of this embodiment according to actual needs.

[0275] Furthermore, the functional units in the various embodiments of this application can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The integrated unit can be implemented in hardware or as a software functional unit.

[0276] If the integrated unit is implemented as a software functional unit and sold or used as an independent product, it can be stored in a readable storage medium. Based on this understanding, the technical solution of this application embodiment, or all or part of the technical solution, can be embodied in the form of a software product. This software product is stored in a storage medium and includes several instructions to cause a device (which may be a microcontroller, chip, etc.) or processor to execute all or part of the steps of the methods described in the various embodiments of this application. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, ROM, RAM, magnetic disks, or optical disks.

[0277] The above embodiments are only used to illustrate the technical solutions of this application, and are not intended to limit them. Although this application has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the embodiments of this application.

Claims

1. A localization method based on an AI model, characterized in that, The method is applied to a terminal, wherein the terminal is equipped with an AI model, and the AI ​​model is used to locate the position of the terminal; the method includes: Receive PRS signals sent from the access network device; The PRS signal is measured to obtain PRS data; the PRS data includes at least the line-of-sight transmission LOS confidence, reference signal time difference RSTD, and terminal receive-transmit time difference RXTX. The PRS data is quantized to obtain first data associated with the PRS data; the first data includes a first parameter, a second parameter, and a third parameter; wherein, the first parameter is obtained by the terminal quantizing the LOS confidence, the second parameter is obtained by the terminal quantizing the RSTD, and the third parameter is obtained by the terminal quantizing the RXTX; the first parameter, the second parameter, and the third parameter are parameters of the same dimension; The first data is input into the AI ​​model to determine the location information of the terminal.

2. The method according to claim 1, characterized in that, The first data is RGB data, the first parameter is the red pixel information R in the RGB data, the second parameter is the green pixel information G in the RGB data, and the third parameter is the blue pixel information B in the RGB data.

3. The method according to claim 2, characterized in that, The terminal is equipped with multiple antennas, each of which is used to receive a PRS signal. One PRS signal corresponds to one RGB data, and the multiple antennas correspond to multiple RGB data, which together form an RGB image.

4. The method according to claim 3, characterized in that, The step of inputting the first data into the AI ​​model to determine the location information of the terminal includes: The RGB image is input into the AI ​​model to determine the location information of the terminal.

5. The method according to claim 4, characterized in that, The AI ​​model includes a first model; the step of inputting the RGB image into the AI ​​model to determine the location information of the terminal includes: Determine whether the size of the RGB image is less than a threshold; If the size of the RGB image is less than a threshold, the RGB image is input into the first model to determine the location information of the terminal.

6. The method according to claim 5, characterized in that, The AI ​​model includes a second model; the method further includes: If the size of the RGB image is greater than or equal to the threshold, the RGB image is input into the second model for dimensionality reduction processing; The dimensionality-reduced RGB image is input into the first model to determine the location information of the terminal.

7. The method according to any one of claims 1-6, characterized in that, The first parameter satisfies the following quantization formula: Among them, P LOS For the LOS confidence level, P LOS ∈[0,1], β is the distribution coefficient, used to indicate the complexity of the PRS signal propagation environment, and round() means rounding the value in (). The second parameter satisfies the following quantization formula: Wherein, RTSD is the time difference of the reference signal, RTSD max The maximum value of the time difference of the reference signal, RTSD min The minimum value of the time difference of the reference signal is , and round() means rounding the value in (). The third parameter satisfies the following quantization formula: Wherein, RXTX is the time difference between the terminal's reception and transmission. min RXTX is the minimum value of the time difference between the terminal's reception and transmission. max This represents the maximum value of the time difference between the terminal's reception and transmission. round() indicates that the value within () is rounded to the nearest integer.

8. The method according to any one of claims 1-7, characterized in that, The method further includes: Send indication information to the access network device, the indication information being used to instruct the location management function (LMF) to send the location information of the terminal, the indication information including the location information of the terminal.

9. A communication device, characterized in that, The communication device includes a processor; the processor is configured to run a computer program or instructions that cause the method as described in any one of claims 1-8 to be performed.

10. A communication device, characterized in that, The communication device includes an interface circuit and a logic circuit; the interface circuit is used for inputting and / or outputting information; the logic circuit is used for performing the method as described in any one of claims 1-8.

11. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores computer instructions or programs that, when executed on a computer, cause the method as described in any one of claims 1-8 to be performed.

12. A computer program product, characterized in that, The computer program product includes a computer program or computer instructions; when some or all of the computer program or computer instructions are run on a computer, the method as described in any one of claims 1-8 is performed.

Images