A positioning method, device and storage medium based on AI model

Abstract

A positioning method, device and storage medium based on an AI model relate to the field of communication technology. The terminal device uses a combination of PRS and environmental parameters to determine one or more matching AI model indexes and sends them to the LMF network element to obtain the corresponding AI model. The terminal device uses a combination of environmental parameters to determine the weights of the positioning results output by each AI model, and uses PRS, each AI model and the weights corresponding to each AI model to determine the final positioning result. In addition, the AI ​​model index can also be determined by the LMF network element, and the final positioning result can also be determined by the LMF network element, and the final positioning result can be sent to the terminal device. This solution can use multiple AI models with a high degree of current environmental parameter matching for positioning and fuse the positioning results according to the weights to obtain an accurate final positioning result, which is suitable for areas where the real environment is complex and changeable, or where the AI ​​model training is insufficient.

Description

Technical Field

[0001] The present application relates to the field of wireless communication technology, and in particular to a positioning method, device and storage medium based on an AI model. Background Art

[0002] In new radio (NR) systems or long term evolution (LTE) systems, it is usually necessary to manage the mobility of user equipment (UE) based on base stations, such as positioning the UE. In order to achieve UE mobility management, the accuracy of positioning is particularly important. Artificial Intelligence (AI) has been introduced into wireless communication networks and can be applied to UE positioning scenarios.

[0003] In some related schemes, a model library can be formed by training multiple AI models for different environmental parameter combinations, and the model library can be deployed in the location management function (LMF) network element. When determining the location, the actual environmental parameter combination of the UE that currently needs to be located is obtained, and an AI model with the same environmental parameter combination as the actual environmental parameter combination used in training set generation and model training is matched from the model library, and the output of the AI ​​model is used as the positioning result.

[0004] However, in actual applications, the number of trained AI models is generally limited, and the actual environment is very complex. It may not be possible to have corresponding AI models for all environmental parameter combinations, resulting in the inability to match the applicable AI model according to the actual environmental parameter combination of the UE and the inability to complete positioning. Summary of the invention

[0005] In order to solve the above problems, the present application provides a positioning method, device and storage medium based on an AI model, which can use multiple AI models with a high degree of matching of current environmental parameters for positioning and fuse the positioning results according to weights to obtain accurate final positioning results.

[0006] In a first aspect, the present application provides an AI model-based positioning method that can be applied to a terminal device, the method comprising: determining one or more matching AI model indexes by receiving a positioning reference signal PRS and an environmental parameter combination, wherein the environmental parameter combination includes state parameters of the terminal device and state parameters of the network device; sending first index information, wherein the first index information includes the one or more AI model indexes; receiving an AI model corresponding to each AI model index in the one or more AI model indexes; determining the weight of the positioning result output by each AI model by using the environmental parameter combination; and determining the final positioning result by using the PRS, each AI model and the weight corresponding to each AI model.

[0007] In this implementation, the terminal device can determine the AI ​​model index corresponding to one or more AI models used for positioning, and then send all AI model indexes to the network device, that is, to the LMF network element. The LMF network element matches the corresponding AI model from the AI ​​model library according to each AI model index and sends it to the terminal device. The terminal device can determine the weight corresponding to each AI model, and use each AI model and the corresponding weight to determine the final positioning result. In the prior art, when it is impossible to determine an AI model that completely matches the current environmental parameter combination of the UE, it is directly considered that the positioning cannot be completed. This solution can use multiple AI models with a high degree of matching with the current environmental parameters of the terminal device for positioning, and fuse the positioning results according to the weights, so as to obtain a more accurate final positioning result. It is particularly suitable for service areas with complex and changeable real environments, or areas where the AI ​​models used for positioning are not sufficiently trained, and has high practicality.

[0008] At the same time, after the LMF network element sends the AI ​​model to the terminal device, for scenarios where continuous positioning is required on the terminal device side, the UE side can directly use the received AI model for positioning. If the environmental parameter combination does not change subsequently, there is no need for frequent interaction with the LMF network element, thereby reducing the positioning time delay.

[0009] In a possible implementation manner, the PRS carries one or more of a cell identifier Cell ID, a reference signal received power RSRP, and an angle of arrival AoA relative to a base station.

[0010] In a possible implementation, the PRS carries a cell identifier Cell ID, and the environmental parameter combination includes status parameters of the terminal device and status parameters of the network device; the status parameters of the terminal device include one or more of a spatial area code, a time status parameter, and a scene status parameter, wherein the spatial area code indicates the grid area in which the cell corresponding to each Cell ID is located, the time status parameter is used to indicate the positioning time, and the scene status parameter is an identifier of the scene in which the terminal device is located; the status parameters of the network device indicate the parameter configuration of the network device.

[0011] In a possible implementation, the scene state parameters are determined by the terminal device using detection results of multiple sensors, where the sensors include one or more of the following: an air pressure sensor, an accelerometer, and a gyroscope.

[0012] In a possible implementation, the PRS carries a cell identifier Cell ID, and the state parameters of the terminal device include a spatial area code, a time state parameter, and a scene state parameter. The received positioning reference signal PRS and an environmental parameter combination are used to determine one or more matching AI model indexes, including: determining a first Cell ID list, the first Cell ID list includes all Cell IDs carried in the PRS; respectively determining the intersection of a second Cell ID list corresponding to each spatial area code and the first Cell ID list, the second Cell ID list includes the Cell IDs of all cells in the grid area corresponding to the spatial area code; taking the spatial area code corresponding to the intersection whose number of Cell IDs is greater than or equal to the first threshold as the first screening condition; when the time interval between the positioning time and the reference value corresponding to the time interval is less than a preset interval, taking the time state parameter corresponding to the time interval as the second screening condition, and the reference value corresponding to the time interval is the middle value of the time interval; taking the scene state parameter as the third screening condition; determining the fourth screening condition according to the state parameter of the network device; determining one or more matching AI model indexes according to the first screening condition, the second screening condition, the third screening condition, and the fourth screening condition, as well as a mapping table of the environmental parameter combination and the AI ​​model index.

[0013] In a possible implementation, the PRS carries a cell identifier Cell ID and RSRP, and the state parameters of the terminal device include a spatial area code, a time state parameter, and a scene state parameter. The received positioning reference signal PRS and an environmental parameter combination are used to determine one or more matching AI model indexes, including: determining a first Cell ID list, the first Cell ID list includes all Cell IDs carried in the PRS; determining the intersection of a second Cell ID list corresponding to each spatial area code and the first Cell ID list, respectively, the second Cell ID list includes the Cell IDs of all cells in the grid area corresponding to the spatial area code; using the spatial area code corresponding to the intersection that meets the following conditions as the first screening condition: a Cell whose RSRP is greater than or equal to a preset power threshold; The number of IDs is greater than a first threshold; when the time interval between the positioning time and the reference value corresponding to the time interval is less than a preset interval, the time state parameter corresponding to the time interval is used as the second filtering condition, and the reference value corresponding to the time interval is the middle value of the time interval; the state parameter of the scene is used as the third filtering condition; the fourth filtering condition is determined according to the state parameter of the network device; one or more matching AI model indexes are determined according to the first filtering condition, the second filtering condition, the third filtering condition and the fourth filtering condition, as well as a mapping table of environmental parameter combinations and AI model indexes.

[0014] In a possible implementation, the PRS carries the cell identification Cell ID, RSRP and AoA relative to the base station, and uses the received positioning reference signal PRS and the combination of environmental parameters to determine one or more matching AI model indexes, including: determining a first Cell ID list, the first Cell ID list includes all Cell IDs carried in the PRS; respectively determining the intersection of the second Cell ID list corresponding to each spatial area code and the first Cell ID list, the second Cell ID list includes the Cell IDs of all cells in the grid area corresponding to the spatial area code; using the spatial area code corresponding to the intersection that meets the following conditions as the first screening condition: RSRP is greater than or equal to the preset power threshold, and the AoA relative to the base station meets the Cell ID within a preset angle range. The number of IDs is greater than a first threshold; when the time interval between the positioning time and the reference value corresponding to the time interval is less than a preset interval, the time state parameter corresponding to the time interval is used as the second filtering condition, and the reference value corresponding to the time interval is the middle value of the time interval; the state parameter of the scene is used as the third filtering condition; the fourth filtering condition is determined according to the state parameter of the network device; one or more matching AI model indexes are determined according to the first filtering condition, the second filtering condition, the third filtering condition and the fourth filtering condition, as well as a mapping table of environmental parameter combinations and AI model indexes.

[0015] In one possible implementation, the state parameter of the network device indicates the parameter configuration of the network device, and the weight of the positioning result output by each AI model is determined by using a combination of environmental parameters, specifically including: determining the first weight coefficient of each AI model according to the number of Cell IDs in the intersection corresponding to each spatial area code as the first screening condition, and the first weight coefficient is positively correlated with the number of Cell IDs in the intersection; determining the second weight coefficient of each AI model according to the time interval, and the second weight coefficient is negatively correlated with the time interval; determining the third weight coefficient of each AI model according to the scene state parameter; determining the fourth weight coefficient of each AI model according to the state parameter of the network device; determining the weight coefficient of each AI model according to each first weight coefficient, each second weight coefficient, each third weight coefficient and each fourth weight coefficient; normalizing the weight coefficient of each AI model to obtain the weight of each AI model, and the sum of the weights of each AI model is 1.

[0016] In one possible implementation, the state parameters of the network device indicate the parameter configuration of the network device in different time periods, and the weight coefficient of each AI model is determined based on each first weight coefficient, each second weight coefficient, each third weight coefficient and each fourth weight coefficient, including: for each AI model, performing the following steps: summing the product of the first influence factor and the first weight coefficient, the product of the first influence factor and the third weight coefficient, the product of the second influence factor and the second weight coefficient, and the product of the second influence factor and the fourth weight coefficient to determine the weight coefficient of each AI model, wherein the first influence factor is greater than the second influence factor.

[0017] In one possible implementation, the final positioning result is determined using PRS, each AI model and the weight corresponding to each AI model, including: using PRS as the input of each AI model to obtain the positioning result output by each AI model; multiplying the positioning result output by each AI model with the weight corresponding to each AI model to obtain the final positioning result component corresponding to each AI model; and superimposing the final positioning result components corresponding to each AI model to obtain the final positioning result.

[0018] On the second aspect, the present application also provides an AI model-based positioning method, which can be applied to a terminal device, the method comprising: sending an environmental parameter combination and a positioning reference signal PRS, the environmental parameter combination including state parameters of the terminal device and state parameters of the network device; receiving one or more AI models, each of the one or more AI models corresponding to an AI model index, and each AI model index is determined by the environmental parameter combination and PRS; using the environmental parameter combination to determine the weight of the positioning result output by each AI model; using PRS, each AI model and the weight corresponding to each AI model to determine the final positioning result.

[0019] In this implementation, after the terminal device sends the environmental parameter combination and PRS to the LMF network element, the LMF can determine the AI ​​model index corresponding to one or more AI models used for positioning, and then the LMF network element matches the corresponding AI model from the AI ​​model library according to each AI model index and sends it to the terminal device. After the terminal device side determines the weights corresponding to each AI model, the final positioning result is determined using each AI model and the corresponding weight. In the existing related schemes, when it is impossible to determine an AI model that completely matches the current environmental parameter combination of the terminal device, it will be directly considered that the positioning cannot be completed. The scheme provided in the embodiment of the present application can use multiple AI models with a high degree of matching with the current environmental parameters of the terminal device for positioning, and fuse the positioning results according to the weights, thereby obtaining a more accurate final positioning result. It is particularly suitable for service areas with complex and changeable real environments, or areas where the AI ​​models used for positioning are not sufficiently trained, and has high practicality.

[0020] At the same time, after the LMF network element sends the AI ​​model to the terminal device, for scenarios where continuous positioning is required on the terminal device side, the terminal device side can directly use the received AI model for positioning. If the environmental parameter combination does not change subsequently, there is no need for frequent interaction with the LMF network element, thereby reducing the positioning time delay.

[0021] In a possible implementation manner, the PRS carries one or more of a cell identifier Cell ID, a reference signal received power RSRP, and an angle of arrival AoA relative to a base station.

[0022] In a possible implementation, the PRS carries a cell identifier Cell ID, and the environmental parameter combination includes status parameters of the terminal device and status parameters of the network device; the status parameters of the terminal device include one or more of a spatial area code, a time status parameter, and a scene status parameter, wherein the spatial area code indicates the grid area in which the cell corresponding to each Cell ID is located, the time status parameter is used to indicate the positioning time, and the scene status parameter is an identifier of the scene in which the terminal device is located; the status parameters of the network device indicate the parameter configuration of the network device.

[0023] In a possible implementation, the PRS carries a cell identifier Cell ID, and the state parameters of the terminal device include a spatial area code, a time state parameter, and a scene state parameter. The weight of the positioning result output by each AI model is determined by combining environmental parameters, specifically including: determining a first Cell ID list, the first Cell ID list includes all Cell IDs carried in the PRS; respectively determining the intersection of a second Cell ID list corresponding to each spatial area code and the first Cell ID list, the second Cell ID list includes the Cell IDs of all cells in the grid area corresponding to the spatial area code; taking the spatial area code corresponding to the intersection whose number of Cell IDs is greater than or equal to the first threshold as the first screening condition; determining a first weight coefficient of each AI model according to the number of Cell IDs in the intersection corresponding to each spatial area code as the first screening condition, the first weight coefficient being positively correlated with the number of Cell IDs in the intersection; determining a second weight coefficient of each AI model according to the time interval, the second weight coefficient being negatively correlated with the time interval; determining a third weight coefficient of each AI model according to the scene state parameter; determining a fourth weight coefficient of each AI model according to the state parameter of the network device; The weight coefficient of each AI model is determined according to the first weight coefficient, the second weight coefficient, the third weight coefficient and the fourth weight coefficient; the weight coefficient of each AI model is normalized to obtain the weight of each AI model, and the sum of the weights of each AI model is 1.

[0024] In a possible implementation, the PRS carries a cell identifier Cell ID and RSRP, the state parameters of the terminal device include a spatial area code, a time state parameter and a scene state parameter, and the weight of the positioning result output by each AI model is determined by combining environmental parameters, specifically including: determining a first Cell ID list, the first Cell ID list includes all Cell IDs carried in the PRS; respectively determining the intersection of a second Cell ID list corresponding to each spatial area code and the first Cell ID list, the second Cell ID list includes the Cell IDs of all cells in the grid area corresponding to the spatial area code; using the spatial area code corresponding to the intersection that satisfies the following conditions as the first screening condition: the number of Cell IDs whose RSRP is greater than or equal to a preset power threshold is greater than the first threshold; determining a first weight coefficient of each AI model according to the number of Cell IDs in the intersection corresponding to each spatial area code as the first screening condition, the first weight coefficient being positively correlated with the number of Cell IDs in the intersection; determining a second weight coefficient of each AI model according to the time interval, the second weight coefficient being negatively correlated with the time interval; determining a third weight coefficient of each AI model according to the scene state parameter; determining a fourth weight coefficient of each AI model according to the state parameter of the network device; The weight coefficient of each AI model is determined according to the first weight coefficient, the second weight coefficient, the third weight coefficient and the fourth weight coefficient; the weight coefficient of each AI model is normalized to obtain the weight of each AI model, and the sum of the weights of each AI model is 1.

[0025] In a possible implementation, the PRS carries a cell identifier Cell ID, and the state parameters of the terminal device include a spatial area code, a time state parameter, and a scene state parameter. The weight of the positioning result output by each AI model is determined by combining environmental parameters, specifically including: determining a first Cell ID list, the first Cell ID list includes all Cell IDs carried in the PRS; determining the intersection of a second Cell ID list corresponding to each spatial area code and the first Cell ID list, respectively, the second Cell ID list includes the Cell IDs of all cells in the grid area corresponding to the spatial area code; using the spatial area code corresponding to the intersection that satisfies the following conditions as the first screening condition: RSRP is greater than or equal to a preset power threshold, and the number of Cell IDs whose AoA relative to the base station satisfies a preset angle range is greater than the first threshold; the number of Cell IDs whose RSRP is greater than or equal to the preset power threshold is greater than the first threshold; determining a first weight coefficient for each AI model according to the number of Cell IDs in the intersection corresponding to each spatial area code as the first screening condition, and the first weight coefficient and the Cell ID in the intersection are ... The number of IDs is positively correlated; the second weight coefficient of each AI model is determined according to the time interval, and the second weight coefficient is negatively correlated with the time interval; the third weight coefficient of each AI model is determined according to the scene state parameters; the fourth weight coefficient of each AI model is determined according to the state parameters of the network device; the weight coefficient of each AI model is determined according to the first weight coefficient, the second weight coefficient, the third weight coefficient and the fourth weight coefficient; the weight coefficient of each AI model is normalized to obtain the weight of each AI model, and the sum of the weights of each AI model is 1.

[0026] In one possible implementation, the state parameters of the network device indicate the parameter configuration of the network device in different time periods, and the weight coefficient of each AI model is determined based on each first weight coefficient, each second weight coefficient, each third weight coefficient and each fourth weight coefficient, including: for each AI model, performing the following steps: summing the product of the first influence factor and the first weight coefficient, the product of the first influence factor and the third weight coefficient, the product of the second influence factor and the second weight coefficient, and the product of the second influence factor and the fourth weight coefficient to determine the weight coefficient of each AI model, wherein the first influence factor is greater than the second influence factor.

[0027] In one possible implementation, the final positioning result is determined using PRS, each AI model and the weight corresponding to each AI model, including: using PRS as the input of each AI model to obtain the positioning result output by each AI model; multiplying the positioning result output by each AI model with the weight corresponding to each AI model to obtain the final positioning result component corresponding to each AI model; and superimposing the final positioning result components corresponding to each AI model to obtain the final positioning result.

[0028] On the third aspect, an embodiment of the present application also provides a positioning method for an AI model, which can be applied to a terminal device, including: using the received positioning reference signal PRS and an environmental parameter combination to determine one or more matching AI model indexes, the environmental parameter combination including the state parameters of the terminal device and the state parameters of the network device; sending first index information, PRS and an environmental parameter combination, the first index information including one or more AI model indexes; receiving location information, the location information including the final positioning result, the final positioning result is determined by PRS, one or more AI models, and the weight corresponding to each AI model in the one or more AI models, each AI model corresponds to an AI model index in one or more AI model indexes, and the weight is the weight of the positioning result output by each AI model determined using the environmental parameter combination.

[0029] In this implementation, the terminal device can determine the AI ​​model index corresponding to one or more AI models used for positioning, and then send all AI model indexes to the LMF network element. The LMF network element matches the corresponding AI model from the AI ​​model library according to each AI model index and determines the weight corresponding to each AI model. Then LMF can use each AI model and the corresponding weight to determine the final positioning result, and send the final positioning result to the terminal device. In the existing scheme, when it is impossible to determine an AI model that completely matches the current environmental parameter combination of the terminal device, it will be directly considered that the positioning cannot be completed. The solution of the present application can use multiple AI models with a high degree of matching with the current environmental parameters of the terminal device for positioning, and fuse the positioning results according to the weights, so as to obtain a more accurate final positioning result. It is especially suitable for service areas with complex and changeable real environments, or areas where the AI ​​models used for positioning are not sufficiently trained, and has high practicality.

[0030] At the same time, for scenarios where continuous positioning is not required on the terminal device side, such as single positioning or a smaller number of positioning times, the terminal device side can directly obtain the positioning results sent by the LMF network element. In this way, the terminal device side does not need to receive the AI ​​model sent by the LMF network element, which can reduce the amount of data received and reduce the time delay of single positioning.

[0031] In the fourth aspect, an embodiment of the present application also provides a positioning method for an AI model, which can be applied to a terminal device, including: sending an environmental parameter combination and a positioning reference signal PRS, the environmental parameter combination including state parameters of the terminal device and state parameters of the network device; receiving location information, the location information including the final positioning result, the final positioning result is determined by the PRS, one or more AI models, and the weight corresponding to each AI model in the one or more AI models, the weight being the weight of the positioning result output by each AI model determined using the environmental parameter combination, each AI model corresponds to an AI model index, and each AI model index is determined using the environmental parameter combination and PRS.

[0032] In this implementation, the terminal device side sends the PRS and environmental parameter combination to the LME network element. The LMF network element can use the PRS and environmental parameter combination to determine the AI ​​model index. The LMF network element matches the corresponding AI model from the AI ​​model library according to each AI model index and determines the corresponding weight of each AI model. Then LMF can use each AI model and the corresponding weight to determine the final positioning result, and send the final positioning result to the terminal device. In the existing related schemes, when it is impossible to determine an AI model that completely matches the current environmental parameter combination of the terminal device, it will be directly considered that the positioning cannot be completed. The scheme provided by the embodiment of the present application can use multiple AI models with a high degree of matching with the current environmental parameters of the terminal device for positioning, and fuse the positioning results according to the weights, so as to obtain a more accurate final positioning result. It is particularly suitable for service areas with complex and changeable real environments, or areas where the AI ​​model used for positioning is not sufficiently trained, and has high practicality.

[0033] At the same time, after the LMF network element sends the AI ​​model to the terminal device, for scenarios where continuous positioning is required on the terminal device side, the terminal device side can directly use the received AI model for positioning. If the environmental parameter combination does not change subsequently, there is no need for frequent interaction with the LMF network element, thereby reducing the positioning time delay.

[0034] In the fifth aspect, an embodiment of the present application also provides a positioning method for an AI model, which can be applied to a network device, and the network device can be a LMF network element. The method includes: receiving first index information, the first index information includes one or more AI model indexes, and the one or more AI model indexes are determined by receiving a positioning reference signal PRS and an environmental parameter combination, and the environmental parameter combination includes state parameters of the terminal device and state parameters of the network device; sending an AI model corresponding to each AI model index in the one or more AI model indexes, each AI model is used to determine the final positioning result together with the PRS and the weight corresponding to each AI model, and the weight corresponding to each AI model is the weight of the positioning result output by each AI model determined by the environmental parameter combination.

[0035] In this implementation, the LMF network element matches the corresponding AI model from the AI ​​model library according to each AI model index and determines the corresponding weight of each AI model, and then sends the matched one or more AI models to the terminal device, which performs fusion positioning. It is suitable for service areas with complex and changeable real-world environments, or areas where the AI ​​model used for positioning is not fully trained, and has high practicality.

[0036] In the sixth aspect, an embodiment of the present application also provides a positioning method for an AI model, which can be applied to a network device, and the network device can be a LMF network element. The method includes: receiving an environmental parameter combination and a positioning reference signal PRS, wherein the environmental parameter combination includes state parameters of the terminal device and state parameters of the network device; using the environmental parameter combination and PRS to determine one or more matching AI model indexes; sending an AI model corresponding to each AI model index in one or more AI model indexes, each AI model is used to determine the final positioning result together with the PRS and the weight corresponding to each AI model, and the weight corresponding to each AI model is the weight of the positioning result output by each AI model determined using the environmental parameter combination.

[0037] In this implementation, the LMF network element determines one or more AI model indexes, matches the corresponding AI models from the AI ​​model library according to each AI model index, and determines the weights of each AI model. Then, the matched one or more AI models are sent to the terminal device, and the terminal device performs fusion positioning. It is suitable for service areas with complex and changeable real-world environments, or areas where the AI ​​model used for positioning is not fully trained, and has high practicality.

[0038] In a possible implementation manner, the PRS carries one or more of a cell identifier Cell ID, a reference signal received power RSRP, and an angle of arrival AoA relative to a base station.

[0039] In a possible implementation, the PRS carries a cell identifier Cell ID, and the environmental parameter combination includes status parameters of the terminal device and status parameters of the network device; the status parameters of the terminal device include one or more of a spatial area code, a time status parameter, and a scene status parameter, wherein the spatial area code indicates the grid area in which the cell corresponding to each Cell ID is located, the time status parameter is used to indicate the positioning time, and the scene status parameter is an identifier of the scene in which the terminal device is located; the status parameters of the network device indicate the parameter configuration of the network device.

[0040] In a possible implementation, the PRS carries a cell identifier Cell ID, and the state parameters of the terminal device include a spatial area code, a time state parameter, and a scene state parameter. The environmental parameter combination and the PRS are used to determine one or more matching AI model indexes, specifically including: determining a first Cell ID list, the first Cell ID list includes all Cell IDs carried in the PRS; determining the intersection of a second Cell ID list corresponding to each spatial area code and the first Cell ID list, respectively, the second Cell ID list includes the Cell IDs of all cells in the grid area corresponding to the spatial area code; taking the spatial area code corresponding to the intersection whose number of Cell IDs is greater than or equal to the first threshold as the first screening condition; when the time interval between the positioning time and the reference value corresponding to the time interval is less than the preset interval, taking the time state parameter corresponding to the time interval as the second screening condition, and the reference value corresponding to the time interval is the middle value of the time interval; taking the scene state parameter as the third screening condition; determining the fourth screening condition according to the state parameter of the network device; determining one or more matching AI model indexes according to the first screening condition, the second screening condition, the third screening condition, and the fourth screening condition, as well as the mapping table of the environmental parameter combination and the AI ​​model index.

[0041] In a possible implementation, the PRS carries a cell identifier Cell ID and RSRP, and the state parameters of the terminal device include a spatial area code, a time state parameter, and a scene state parameter. The environmental parameter combination and the PRS are used to determine one or more matching AI model indexes, specifically including: determining a first Cell ID list, the first Cell ID list includes all Cell IDs carried in the PRS; determining the intersection of a second Cell ID list corresponding to each spatial area code and the first Cell ID list, respectively, the second Cell ID list includes the Cell IDs of all cells in the grid area corresponding to the spatial area code; using the spatial area code corresponding to the intersection that meets the following conditions as the first screening condition: a Cell whose RSRP is greater than or equal to a preset power threshold; The number of IDs is greater than a first threshold; when the time interval between the positioning time and the reference value corresponding to the time interval is less than a preset interval, the time state parameter corresponding to the time interval is used as the second filtering condition, and the reference value corresponding to the time interval is the middle value of the time interval; the state parameter of the scene is used as the third filtering condition; the fourth filtering condition is determined according to the state parameter of the network device; one or more matching AI model indexes are determined according to the first filtering condition, the second filtering condition, the third filtering condition and the fourth filtering condition, as well as a mapping table of environmental parameter combinations and AI model indexes.

[0042] In a possible implementation, the PRS carries the cell ID, RSRP, and AoA relative to the base station, and uses the environmental parameter combination and PRS to determine one or more matching AI model indexes, specifically including: determining a first Cell ID list, the first Cell ID list includes all Cell IDs carried in the PRS; determining the intersection of the second Cell ID list corresponding to each spatial area code and the first Cell ID list, respectively, the second Cell ID list includes the Cell IDs of all cells in the grid area corresponding to the spatial area code. ID; the spatial area code corresponding to the intersection that satisfies the following conditions is used as the first filtering condition: RSRP is greater than or equal to the preset power threshold, and the number of CellIDs whose AoA relative to the base station satisfies the preset angle range is greater than the first threshold; when the time interval between the positioning time and the reference value corresponding to the time interval is less than the preset interval, the time state parameter corresponding to the time interval is used as the second filtering condition, and the reference value corresponding to the time interval is the middle value of the time interval; the state parameter of the scene is used as the third filtering condition; the fourth filtering condition is determined according to the state parameter of the network device; according to the first filtering condition, the second filtering condition, the third filtering condition and the fourth filtering condition, and the mapping table of environmental parameter combinations and AI model indexes, one or more matching AI model indexes are determined.

[0043] In the seventh aspect, an embodiment of the present application also provides a positioning method for an AI model, which can be applied to a network device, and the network device can be a LMF network element. The method includes: receiving first index information, a positioning reference signal PRS and an environmental parameter combination, the first index information includes one or more AI model indexes, and the one or more AI model indexes are determined using a combination of PRS and environmental parameters, and the environmental parameter combination includes state parameters of the terminal device and state parameters of the network device; determining the AI ​​models corresponding to each AI model index in the one or more AI model indexes; determining the weight of the positioning result output by each AI model using the environmental parameter combination; determining the final positioning result using PRS, each AI model and the weight corresponding to each AI model; sending location information, and the location information includes the final positioning result.

[0044] In this implementation, the LMF network element can obtain the AI ​​model according to the AI ​​model index matching, determine the final positioning result, and directly send the final positioning result to the terminal device. Among them, the LMF network element determines the weight of the positioning result output by each AI model, and the specific implementation method of determining the final positioning result can refer to the implementation method provided in the first aspect above.

[0045] In the eighth aspect, an embodiment of the present application also provides a positioning method for an AI model, which can be applied to a network device, and the network device can be a LMF network element. The method includes: receiving an environmental parameter combination and a positioning reference signal PRS, wherein the environmental parameter combination includes state parameters of the terminal device and state parameters of the network device; using the environmental parameter combination and PRS to determine one or more matching AI model indexes; determining the AI ​​models corresponding to each AI model index in the one or more AI model indexes; using the environmental parameter combination to determine the weight of the positioning result output by each AI model; using PRS, each AI model and the weight corresponding to each AI model to determine the final positioning result; sending location information, wherein the location information includes the final positioning result.

[0046] In this implementation, the LMF network element can determine the AI ​​model index, obtain the AI ​​model according to the AI ​​model index matching, determine the final positioning result, and directly send the final positioning result to the terminal device. Among them, the LMF network element determines the AI ​​model index, determines the weight of the positioning result output by each AI model, and determines the specific implementation method of the final positioning result. Please refer to the implementation method provided in the first aspect above.

[0047] In the ninth aspect, the present application also provides a terminal device, which includes a processor and a memory; the processor is coupled to the memory; the memory is used to store computer programs and / or instructions; the processor is used to execute the computer programs and / or instructions stored in the memory to implement the AI ​​model-based positioning method provided in the first aspect and any one of the implementation methods of the first aspect, or the second aspect and any one of the implementation methods of the second aspect, or the third aspect, or the fourth aspect.

[0048] In the tenth aspect, the present application also provides a communication device, including a processor, wherein the processor is used to execute a computer program or instruction stored in a memory so that the device executes the AI ​​model-based positioning method provided in the fifth aspect, or the sixth aspect and any one of the implementations of the sixth aspect, or the seventh aspect, or the eighth aspect.

[0049] In the eleventh aspect, the present application also provides a computer program product, which includes a computer program or instruction for executing the AI ​​model-based positioning method provided in the first aspect and any one of the implementations of the first aspect, or the second aspect and any one of the implementations of the second aspect, or the third aspect, or the fourth aspect; or, a computer program or instruction for executing the AI ​​model-based positioning method provided in the fifth aspect, or the sixth aspect and any one of the implementations of the sixth aspect, or the seventh aspect, or the eighth aspect.

[0050] In the twelfth aspect, the present application also provides a computer-readable storage medium, on which a computer program or instructions are stored. When the computer program or instructions are executed, the AI ​​model-based positioning method provided by any one of the above aspects or any one of the implementation methods of any one of the aspects is implemented. BRIEF DESCRIPTION OF THE DRAWINGS

[0051] Figure 1 A schematic diagram of a communication system provided for this application;

[0052] Figure 2 A schematic diagram of a NR system in 5G provided in an embodiment of the present application;

[0053] Figure 3 A flowchart of a positioning method based on an AI model provided in an embodiment of the present application;

[0054] Figure 4 A schematic diagram of a scenario provided for an embodiment of the present application;

[0055] Figure 5 Schematic diagram of the framework for positioning based on the AI ​​model provided in the embodiment of the present application Figure 1 ;

[0056] Figure 6 A flowchart of another positioning method based on an AI model provided in an embodiment of the present application;

[0057] Figure 7 Schematic diagram of the framework for positioning based on the AI ​​model provided in the embodiment of the present application Figure 2 ;

[0058] Figure 8 A flowchart of another positioning method based on an AI model provided in an embodiment of the present application;

[0059] Fig. 9 Schematic diagram of the framework for positioning based on the AI ​​model provided in the embodiment of the present application Figure 3 ;

[0060] Fig.10 A flowchart of another positioning method based on an AI model provided in an embodiment of the present application;

[0061] Fig.11 Schematic diagram of the framework for positioning based on the AI ​​model provided in the embodiment of the present application Figure 4 ;

[0062] Fig.12 A schematic diagram of a terminal device provided in an embodiment of the present application;

[0063] Fig.13 A schematic diagram of a communication device provided in an embodiment of the present application. DETAILED DESCRIPTION

[0064] In the embodiments of the present application, words such as "exemplary" or "for example" are used to indicate examples, illustrations or descriptions. Any embodiment or design described as "exemplary" or "for example" in the embodiments of the present application should not be interpreted as being more preferred or more advantageous than other embodiments or designs. Specifically, the use of words such as "exemplary" or "for example" is intended to present related concepts in a specific way.

[0065] It is understood that the terms "first" and "second" are used for descriptive purposes only and are not to be understood as indicating or implying relative importance or implicitly indicating the number of technical features indicated. Thus, a feature defined as "first" or "second" may explicitly or implicitly include one or more of the features. It is understood that the terms used in the description of the various examples described herein are only for describing specific examples and are not intended to be limiting. As used in the description of the various examples described, the singular forms "a" and "the" are intended to include the plural forms as well, unless the context clearly indicates otherwise.

[0066] The technical solution of the embodiment of the present application can be applied to various communication systems, for example: thgeneration, 5G) or new radio (NR) system, long term evolution (LTE) system, LTE frequency division duplex (FDD) system, LTE time division duplex (TDD) system, wireless local area network (WLAN) system, satellite communication system, future communication system such as the sixth generation mobile communication system, or a fusion system of multiple systems.

[0067] The technical solution provided by this application can also be applied to device-to-device (D2D) communication, vehicle-to-everything (V2X) communication, machine-to-machine (M2M) communication, machine type communication (MTC), and Internet of Things (IoT) communication system or other communication systems. Of course, future communication systems may also have other naming methods, which are still included in the scope of this application, and this application does not make any limitation on this.

[0068] A network element in a communication system can send a signal to another network element or receive a signal from another network element. The signal may include information, signaling or data, etc. The network element may also be replaced by an entity, a network entity, a device, a communication device, a communication module, a node, a communication node, etc., and the network element is used as an example for description in the present disclosure. For example, the communication system may include at least one terminal device and at least one network device. The network device may send a downlink signal to the terminal device, and / or the terminal device may send an uplink signal to the network device. It is understandable that the terminal device in the present disclosure may be replaced by a first network element, and the network device may be replaced by a second network element, and the two perform the corresponding communication method in the present disclosure.

[0069] The terminal devices in the embodiments of the present application include various devices with wireless communication functions, which can be used to connect people, objects, machines, etc. The terminal devices can be widely used in various scenarios, such as: cellular communication, D2D, V2X, peer to peer (P2P), M2M, MTC, IoT, virtual reality (VR), augmented reality (AR), industrial control, automatic driving, telemedicine, smart grid, smart furniture, smart office, smart wear, smart transportation, smart city drones, robots, remote sensing, passive sensing, positioning, navigation and tracking, autonomous delivery and other scenarios.

[0070] The terminal device can be a terminal in any of the above scenarios, such as an MTC terminal, an IoT terminal, etc. The terminal device can be a user equipment (UE), terminal, fixed device, mobile station device or mobile device of the third generation partnership project (3rd generation partnership project, 3GPP) standard, a user unit (UE), a handheld device, a vehicle-mounted device, a wearable device, a cellular phone, a smart phone, a session initialization protocol (SIP) phone, a wireless data card, a personal digital assistant (PDA), a computer, a tablet computer, a notebook computer, a wireless modem, a handheld device (handset), a laptop computer, a computer with wireless transceiver function, a smart book, a vehicle, a satellite, a global positioning system (GPS) device, a target tracking device, an aircraft (such as a drone, a helicopter, a multi-copter, a quadcopter, or an airplane, etc.), a ship, a remote control device, a smart home device, an industrial device, or a device built into the above-mentioned device (for example, a communication module, a modem or a chip in the above-mentioned device, etc.), or other processing devices connected to the wireless modem. For the convenience of description, the terminal device is described below by taking the terminal or UE as an example.

[0071] It should be understood that in some scenarios, the UE can also be used to act as a base station. For example, the UE can act as a scheduling entity that provides sidelink signals between UEs in scenarios such as V2X, D2D or P2P.

[0072] In the embodiment of the present application, the device for realizing the function of the terminal device can be the terminal device, or it can be a device that can support the terminal device to realize the function, such as a chip system or a chip, which can be installed in the terminal device. In the embodiment of the present application, the chip system can be composed of a chip, or it can include a chip and other discrete devices.

[0073] The network device in the embodiment of the present application may be a device for communicating with a terminal device, and the network device may also be referred to as an access network device or a wireless access network device, such as a base station. The network device in the embodiment of the present application may refer to a wireless access network (RAN) node (or device) that connects a terminal device to a wireless network. The base station can broadly cover the following various names, or be replaced with the following names, such as: NodeB, evolved NodeB (eNB), next generation NodeB (gNB), relay station, access point, transmission reception point (TRP), transmitting point (TP), master station, auxiliary station, multi-standard wireless (motor slide retainer, MSR) node, home base station, network controller, access node, wireless node, access point (AP), transmission node, transceiver node, baseband unit (BBU), remote radio unit (RRU), active antenna unit (AAU), remote radio head (RRH), central unit (CU), distributed unit (DU), positioning node, etc.

[0074] The base station may be a macro base station, a micro base station, a relay node, a donor node or the like, or a combination thereof. The base station may also refer to a communication module, a modem or a chip used to be arranged in the aforementioned device or apparatus. The base station may also be a mobile switching center and a device that performs the base station function in D2D, V2X, and M2M communications, a network side device in a 6G network, and a device that performs the base station function in a future communication system. The base station may support networks with the same or different access technologies. The embodiments of the present application do not limit the specific technology and specific device form adopted by the network equipment.

[0075] A base station can be fixed or mobile. For example, a helicopter or drone can be configured to act as a mobile base station, and one or more cells can move according to the location of the mobile base station. In other examples, a helicopter or drone can be configured to act as a device that communicates with another base station.

[0076] In the embodiment of the present application, the device for realizing the function of the network device can be a network device, or a device capable of supporting the network device to realize the function, such as a chip system or a chip, which can be installed in the network device. In the embodiment of the present application, the chip system can be composed of a chip, or can include a chip and other discrete devices.

[0077] Network devices and terminal devices can be deployed on land, including indoors or outdoors, handheld or vehicle-mounted; they can also be deployed on the water; they can also be deployed on aircraft, balloons and satellites in the air. The scenarios in which network devices and terminal devices are located are not limited in the embodiments of the present application. In addition, terminal devices and network devices can be hardware devices, or they can be software functions running on dedicated hardware, software functions running on general-purpose hardware, such as virtualization functions instantiated on a platform (e.g., a cloud platform), or entities including dedicated or general-purpose hardware devices and software functions. The present application does not limit the specific forms of terminal devices and network devices.

[0078] In order to enable persons skilled in the art to more clearly understand the solution of the present application, the application scenarios involved in the technical solution of the present application are first described below.

[0079] See also Figure 1 , which is a schematic diagram of the communication system provided by this application.

[0080] The wireless communication system includes an access network 400 and a core network 200. Optionally, the communication system may further include the Internet 300.

[0081] The wireless access network 100 may be a next generation (e.g., 6G or higher) wireless access network, or a conventional (e.g., 5G, 4G, 3G, or 2G) wireless access network.

[0082] One or more terminal devices 420 (only one of which is shown in the figure) can be connected to each other or to one or more network devices in the access network 400. Figure 1 In the example, the access network 400 includes the network device 410a and the network device 410b.

[0083] Figure 1 The figure is only for illustrative purposes, and the wireless communication system may also include other devices, such as core network devices, wireless relay devices and / or wireless backhaul devices.

[0084] In practical applications, the wireless communication system may include multiple network devices (also called access network devices) at the same time, and may also include multiple terminal devices at the same time, without limitation. A network device may serve one or more terminal devices at the same time. A terminal device may also access one or more network devices at the same time.

[0085] Currently, artificial intelligence (AI) has been introduced into wireless communication networks and has been widely used in many application scenarios of air interface technology. A common application scenario is the positioning of terminal devices based on AI.

[0086] When positioning terminal devices based on AI, AI nodes can be deployed in one or more of the following: network devices, terminal devices, core networks, or positioning devices; or, AI nodes can also be deployed separately, such as in a location other than any of the above devices.

[0087] Optionally, the AI ​​node is used to perform AI-related operations. As an example, the AI-related operations may include: one or more of: model failure test, model performance test, model training test, or data collection.

[0088] For example, the network device may forward the data related to the AI ​​model reported by the terminal device to the AI ​​node, and the AI ​​node may perform the AI-related operations. For another example, the network device or the terminal device may forward the data related to the AI ​​model to the AI ​​node, and the AI ​​node may perform the AI-related operations.

[0089] For another example, the AI ​​node may send the output of AI-related operations, such as one or more of a trained neural network model, model evaluation, or test results, to a network device and / or a terminal device. Optionally, the AI ​​node may directly send the output of AI-related operations to the network device and the terminal device. For another example, the AI ​​node may send the output of AI-related operations to the terminal device via the network device. For another example, the AI ​​node may send the output of AI-related operations to the network device via the terminal device.

[0090] It is understood that the present application does not limit the number of AI nodes. For example, when there are multiple AI nodes, the multiple AI nodes can be divided based on functions, such as different AI nodes are responsible for different functions.

[0091] In the current related solutions, the AI ​​node (including multiple AI models for positioning) can be deployed in the location management function (LMF) network element of the core network 200. The LMF network element can be used to collect positioning-related measurement results, and determine the location information of the terminal device based on the collected measurement results. The LMF network element can also be called a positioning server, a location management device, etc.

[0092] It can be understood that in the present application, the device for implementing the function of the location management device can be a location management device; it can also be a device that can support the location management device to implement the function, such as a chip system, a hardware circuit, a software module, or a hardware circuit plus a software module. The device can be installed in the location management device, or can be used in combination with the location management device.

[0093] For the convenience of description, the following describes the technical solution of the present application by taking the apparatus for implementing the functions of the location management device as a location management device, and the location management device as an LMF as an example.

[0094] See also Figure 2 , which is a schematic diagram of the NR system in 5G provided in an embodiment of the present application.

[0095] The base stations in the next generation radio access network (NG-RAN) can include the next generation NodeB (gNB) and the next generation-evolved NodeB (ng-eNB). Among them, gNB is the base station in 5G, and ng-eNB is the 4G base station accessing the 5G core network.

[0096] The core network includes access and mobility management function (AMF) network elements and location management function (LMF) network elements.

[0097] The user equipment (UE) and gNB communicate through the NR-Uu interface, and the UE and ng-eNB communicate through the LTE-Uu interface. Both the NR-Uu interface and the LTE-Uu interface can be used to transmit positioning-related signaling. The capital letter U represents the user to network interface (UNI), and the lowercase letter u represents universal.

[0098] Among them, LTE-Uu and NR-Uu use non-access stratum (NAS) or radio resource control (RRC) protocol for transmission. The gNB and AMF, as well as the ng-eNB and AMF, communicate through the next generation core network (NG-C) interface, which can be used to transmit positioning-related signaling.

[0099] NG-C uses the NG application protocol (NGAP). The AMF network element and the LMF network element communicate through the NL1 interface, which can be used to transmit positioning-related signaling. The N in NL1 represents the AMF network element, and the L represents the location service device, that is, the LMF network element.

[0100] All or part of the functions implemented by one or more of the UE, base station, or core network can be virtualized, that is, implemented by a proprietary processor, or a general-purpose processor, and corresponding software modules. Since the UE and the base station involve an interface for air interface transmission, the transceiver function of the interface can be implemented by hardware. The core network, such as the aforementioned AMF network element, or LMF network element, can be virtualized. Optionally, one or more functions of the virtualized UE, base station or core network can be implemented by a cloud device, such as a cloud device in an over the top (OTT) system.

[0101] Among them, the UE and LMF can communicate based on the LTE positioning protocol (LPP). The LPP protocol specifies the process of exchanging information between the UE and LMF. In actual deployment, the UE and LMF are not directly connected, but are connected in the form of UE-base station-AMF-LMF. LPP messages can be transparently transmitted across base stations and AMF to achieve interaction between the UE and LMF. Alternatively, the UE and LMF can forward and exchange information through the base station and AMF without restriction.

[0102] The interaction between the base station and LMF can be based on the NR positioning protocol A (NRPPa). The NRPPa protocol specifies the process of exchanging information between the base station and LMF. In actual deployment, the base station and LMF are connected through AMF. The NRPPa protocol is transparent to the AMF, and the NRPPa data unit is transparently transmitted across the AMF to realize the interaction between the LMF and the base station.

[0103] At present, in order to adapt to more generalized scenarios, generally only one AI model is not deployed on the LMF network element side for positioning. Instead, several different AI models need to be generated to form an AI model library to adapt to different environments. Each AI model corresponds to a different combination of environmental parameters.

[0104] The UE can determine the current environmental parameter combination in which it is located, and select the AI ​​model that best matches the current actual environmental parameter combination from the AI ​​model library of the LMF network element. The UE uses the positioning reference signal (PRS) sent by one or more base stations as the input of the AI ​​model, and the output of the AI ​​model is the UE's location information.

[0105] In the relevant scheme, the matching criteria for selecting the most matching AI model from the AI ​​model library of LMF is: the actual environmental parameter combination during location prediction is exactly the same as the environmental parameter combination corresponding to the AI ​​model. The reason for this is that when the training set of the AI ​​model is generated and the model is trained, specific environmental parameter combinations are targeted. Therefore, when using the AI ​​model for location prediction, the same environmental parameter combination should also be applied to ensure that the UE location information output by the AI ​​model is accurate and the location error is reduced.

[0106] Theoretically, if the model training is sufficient and the number of AI models in the AI ​​model library is sufficient, it can be considered that the optimal AI model can be matched for all environmental parameter combinations. However, in practical applications, the number of trained AI models is generally limited, and the real environment is very complex. It may not be possible to have corresponding AI models for all environmental parameter combinations, resulting in the inability to match the applicable AI model according to the actual environmental parameter combination of the UE and the inability to complete positioning.

[0107] In order to solve the above technical problems, this application provides a positioning method, device and storage medium based on an AI model, which can use multiple AI models with a high degree of matching with the current environmental parameters of the UE for positioning, and fuse the positioning results according to the weights to obtain a more accurate final positioning result. It is particularly suitable for service areas with complex and changeable real environments, or areas where the AI ​​model used for positioning is not fully trained, and has high practicality.

[0108] The technical solution provided by this application is described in detail below with reference to the accompanying drawings.

[0109] See also Figure 3 , which is a flowchart of a positioning method based on an AI model provided in an embodiment of the present application.

[0110] In the following description, the terminal device is taken as UE, and the AI ​​model is deployed in the LMF network element as an example. The method includes the following steps:

[0111] S21: The UE receives a positioning reference signal.

[0112] The positioning reference signal (PRS) received by the UE can be sent by multiple base stations. For the convenience of explanation, the embodiment of the present application takes the base station as gNB as an example, but it does not constitute a limitation on the base station type.

[0113] The PRS sent by each base station may carry one or more of a cell ID (Cell ID), a reference signal received power (reference signal received power, RSRP), and an angle of arrival (angle of arrival, AoA) relative to the base station.

[0114] The Cell ID is a unique identifier assigned to each cell. In a possible implementation, the PRS sent by multiple base stations may carry the Cell IDs of the serving cell and the neighboring cell.

[0115] In some embodiments, the UE side may not receive the neighboring cell Cell ID, in which case the neighboring cell's Physical Cell Identifier (PCI) may be used to distinguish the neighboring cells.

[0116] After receiving the PRS sent by each base station, the UE can summarize the Cell IDs of all cells, the RSRP corresponding to each cell, and the AoA relative to each base station.

[0117] The following example illustrates this.

[0118] See also Figure 4 , which is a schematic diagram of a scenario provided in an embodiment of the present application.

[0119] Figure 4 , a schematic diagram of region division and spatial region coding for a service area is illustrated. The service area can be divided according to the regional information and the characteristics of the region. For example, a rectangular area of ​​a certain area can be used as a division basis to divide the service area into multiple grid areas, such as grid area L1, grid area L2, grid area L3 and grid area L4 in the figure. Each grid area has a corresponding spatial region code.

[0120] Each grid area generally includes multiple cells. The number of cells included in each grid area can be the same or different. Figure 4 The same number of cells is taken as an example for explanation. Figure 4 The box labeled 50 represents UE, namely UE50. Figure 4 Each black and white circle in represents a cell, that is, the circle numbered 40 represents a cell, and each cell corresponds to a Cell ID.

[0121] After receiving the PRS sent by each base station, UE50 obtains the Cell IDs of all cells, which are the 9 cells in the dotted box in the figure. It can be seen that since the UE is at the junction of different grid areas, the ranges of the 9 cells are also distributed in different grid areas.

[0122] S22: The UE obtains a combination of environmental parameters of the current environment.

[0123] In order to improve the positioning accuracy of AI model positioning, it is necessary to ensure that the UE environment is highly consistent when the model is trained and the model is applied, that is, the PRS used in model training and the PRS obtained during positioning are obtained in the same environment, thereby ensuring the accuracy of positioning.

[0124] The attributes of the environment in which the UE is located can be characterized by a combination of environmental parameters.

[0125] The environmental parameter combination includes the status parameters of the terminal device and the status parameters of the network device.

[0126] The state parameters of the terminal device can be used to characterize the state parameters of the environment in which the terminal device, that is, the UE, is currently located, which may include but are not limited to spatial state parameters, temporal state parameters and scene state parameters.

[0127] The spatial state parameter may be a spatial region coding, and the spatial region coding may refer to discretization by dividing the continuous space into discrete grid regions and encoding the discretized regions. Figure 4 As shown, grid areas L1 to L4 may have corresponding spatial area codes.

[0128] In practical applications, different regions with different spatial region codes generally use different AI models for positioning. That is, when the AI ​​model is trained, the training data set includes data in a specific region, so the AI ​​model generally also processes the positioning task in that region in a targeted manner.

[0129] In a possible implementation, the service area may be divided into discrete grid areas by using Geohash geographic coding or Google S2 coding, and the spatial area codes corresponding to the grid areas may be determined.

[0130] The spatial state code of the grid area where the UE is located can reflect the approximate physical location range of the current UE. The spatial state code can be determined by the Cell ID carried in the PRS received by the UE, for example Figure 4 The Cell IDs carried in the PRSs received by the UE correspond to nine cells, and it is determined that the given spatial area code corresponds to the grid area L1-L4, which means that the UE is currently in the area covered by the grid area L1-L4.

[0131] In a possible implementation, the UE may store a correspondence between a Cell ID list and a spatial region code, wherein the correspondence includes Cell IDs of all cells included in a grid region corresponding to each spatial region code, for example, see the example shown in Table 1 below.

[0132] Table 1: Correspondence between Cell ID list and spatial region coding

[0133]

[0134] The above correspondence can be stored in the form of a data table on the UE, or it can be obtained in real time through the network when the UE is positioned. It can be understood that the above spatial area coding and Cell ID data structure are only for the convenience of explanation and do not constitute a limitation on the technical solution of this application. In actual applications, other data structures that meet the requirements of relevant standards can be used.

[0135] The time state parameter is parameter information obtained by discretizing continuous time variables, and represents the time information when the UE performs positioning.

[0136] In positioning scenarios, considering the versatility of AI models, the cycle is generally one day, and corresponding time state parameters are set for each time period within a day. For example, a day is discretized into hours and divided into 24 specific time periods, namely [0:00, 1:00), [1:00, 2:00), …, [23:00, 24:00). The signal characteristics received by the UE in each time period may be different, which will affect the positioning accuracy. Therefore, there is a corresponding AI model for each hourly time period.

[0137] Different time periods correspond to different time state parameters. For example, when the time when the UE performs positioning is in [0:00, 1:00), the corresponding time state parameter is T1, or it is directly expressed by the time interval [0:00, 1:00); when the time when the UE performs positioning is in [23:00, 24:00), the corresponding time state parameter is T24, or it is directly expressed by the time interval [23:00, 24:00).

[0138] In a possible implementation, the positioning moment when the UE performs positioning can be determined by the network, and then the time period in which the positioning moment is located is matched and determined, thereby obtaining the time state parameter.

[0139] It is understandable that the above division method is only an example and does not constitute a limitation on the technical solution of the present application. For example, a day can be divided into 12 specific time periods, each time period includes two hours; or a day can be divided into 48 specific time periods, each time period includes 0.5 hours, etc.

[0140] In another possible implementation, the multiple divided time periods may also have intersections. For example, the time periods within a day may be [0:00, 2:00), [1:00, 3:00), [2:00, 4:00), ..., [23:00, 24:00), etc.

[0141] The scene state parameter represents the scene in which the UE is located when positioning. Different scenes have different scene characteristics, which will affect the accuracy of the positioning results of the AI ​​model. Therefore, it is generally necessary to train AI models suitable for different scenes for different scenes. The scenes may include but are not limited to high-speed rail scenes, subway scenes, indoor scenes, underground garage scenes, general outdoor scenes, etc.

[0142] For example, in the high-speed rail scenario, when the UE is on the high-speed rail, it has the characteristics of fast movement speed, large signal penetration loss, and large signal Doppler frequency shift; in the underground garage scenario, when the UE is in the underground garage, the signal strength is low, and signal interruption may occur, resulting in data update delays. Therefore, the model parameters and model types of the AI ​​model used for high-speed rail scenario positioning and the AI ​​model suitable for underground garage scenario positioning may be different.

[0143] Each scene is pre-configured with a corresponding identifier, for example, the high-speed rail scene corresponds to identifier L1, and the underground garage scene corresponds to identifier L2. The identifier can be used as a scene state parameter.

[0144] UE can detect scenes through its own sensors.

[0145] For example, the accelerometer equipped by the UE can detect the magnitude of the acceleration of the UE in different directions, and the gyroscope can be used to determine the movement posture of the UE. The UE can determine, based on the detection results of the accelerometer and the gyroscope, that the UE performs acceleration and deceleration movements multiple times within a short time interval, such as multiple acceleration and deceleration movements within a dozen minutes or a few minutes, and when it is determined that the UE's altitude is lower than the altitude of the grid area corresponding to the spatial area code based on the air pressure detected by the air pressure sensor, it can be determined that the current scene is a subway scene.

[0146] For another example, when the UE determines that the UE's altitude is lower than the altitude of the grid area corresponding to the spatial area coding based on the air pressure detected by the pressure sensor, and the UE determines that the speed change of the UE is small based on the detection results of the accelerometer and the gyroscope, it can be determined that the current scene is an underground garage scene (similar to the basement scene).

[0147] For another example, when the UE determines that the acceleration and speed of the UE meet the characteristics of the high-speed rail scene based on the detection results of the accelerometer and the gyroscope, it can be determined that the current scene is a high-speed rail scene.

[0148] For another example, the UE can determine whether it is in an indoor scene or an outdoor scene according to whether it can currently initiate global navigation satellite system (GNSS) positioning. If GNSS positioning can be initiated, it is considered to be in an outdoor scene, otherwise it is considered to be in an indoor scene.

[0149] In addition, the UE can also determine the scene based on the properties of the currently connected base station device. For example, when the base station currently connected to the UE is identified as a dedicated base station along the high-speed rail line, the current scene can be determined to be a high-speed rail scene. For another example, when the base station currently connected to the UE is identified as a dedicated base station for a large shopping mall, the current scene can be determined to be an indoor scene.

[0150] The network equipment status parameters are generally related to the configuration parameters of the network equipment, which are mainly related to the network optimization measures of the network equipment operator. For the same network equipment, the operator may set different configuration parameters on different dates or in different time periods of the same day. The configuration parameters are specifically related to the load conditions of the network equipment at different times, the maintenance capabilities of the operator, and the business behavior of the operator.

[0151] For example, for the same network device, the configuration parameters at different times such as weekdays (Monday to Friday), ordinary rest days (Saturday and weekends) and other statutory holidays are different, resulting in differences in information processing delays, network quality and other conditions of the network device, which will affect the positioning results.

[0152] Therefore, different AI models are generally trained for different network device status parameters. When positioning, the AI ​​model that matches the current network device status parameters is selected to improve the accuracy of the positioning results.

[0153] The parameters of network devices can be identified in an anonymous and discretized manner.

[0154] For example, for time-related state parameters, continuous time can be discretized into different time periods, such as working day time periods, ordinary rest day time periods, and other statutory holiday time periods. When the UE positioning time is within the discretized working day time period, it can be determined that the current time-related state parameter is the state parameter corresponding to the working day time period.

[0155] S23: The UE uses the received positioning reference signal PRS and the environmental parameter combination to determine one or more matching AI model indexes.

[0156] Compared with existing related solutions, the solution of the present application allows for the determination of multiple available AI models.

[0157] It is understandable that the environmental parameter combinations of the training data used in the training of the multiple AI models determined in this application are not completely consistent with the screening conditions when determining the AI ​​model index. For example, both the time screening conditions and the space screening conditions may obtain multiple matching AI models. This is because the present application scheme takes into account that the model training is not sufficient due to the complex environment in actual application, but in some special cases, the UE cannot match the only optimal model according to the current environmental parameter combination, and can only match several relatively excellent AI models. The environmental parameter combinations corresponding to the training data of these AI models are more or less inconsistent with the actual environmental parameter combinations during positioning. At this time, the present application scheme does not directly determine that the positioning cannot be completed, but believes that even if the most reliable positioning result cannot be obtained at this time, the positioning results of these nearly optimal AI models can be merged to obtain a relatively accurate final positioning result.

[0158] See also Figure 5 , which is a schematic diagram of the framework for positioning based on the AI ​​model provided in the embodiment of the present application Figure 1 .

[0159] The UE aggregates the received PRS information, and determines the AI ​​model indexes corresponding to one or more currently applicable AI models in combination with the acquired environmental parameter combination.

[0160] Among them, in the embodiment of the present application, the use of sensor information to determine some environmental parameters in the environmental parameter combination is taken as an example. In actual applications, the environmental parameters in the environmental parameter combination can also be obtained through user active configuration, or through network equipment, the Internet of Things, the Internet, etc., and the embodiment of the present application does not make specific limitations.

[0161] When selecting the currently applicable AI model, the AI ​​model is mainly matched based on the combination of environmental parameters.

[0162] Take the example that the environment parameter combination includes the state parameters of the terminal device and the state parameters of the network device, and the state parameters of the terminal device specifically include the spatial area code (ie, the spatial state parameters), the time state parameters and the scene state parameters.

[0163] The AI ​​model index is a data structure used for quick query and AI model, which can be understood as the identifier of the AI ​​model.

[0164] Each AI model has a unique corresponding AI model index. The embodiment of the present application does not limit the specific data structure of the AI ​​model index, and the following description is only for example.

[0165] See Table 2 below, which is a table showing the correspondence between various environmental parameter combinations and AI model indexes.

[0166] Table 2: Correspondence between various environmental parameter combinations and AI model indexes

[0167]

[0168] The corresponding relationship in Table 2 can also be called a mapping table. It can be understood that the data structure of each data in Table 2 is only for convenience of explanation and does not constitute a limitation on the technical solution of this application. In actual applications, other data structures that meet the requirements of relevant standards can be used.

[0169] The UE can first summarize all the received PRSs and then obtain all the Cell IDs, see Table 1 and Figure 4 At this time, the determined Cell IDs are Cell11, Cell12, Cell25, Cell17, Cell18, Cell31, Cell41, Cell42 and Cell55 respectively.

[0170] All Cell IDs indicated in the PRS may form a first Cell ID list, that is, the first Cell ID list is the union of the Cell IDs indicated in all PRSs.

[0171] The Cell ID list corresponding to the spatial area code may be referred to as a second Cell ID list.

[0172] Determine the intersection of the first Cell ID list and the second Cell ID list corresponding to each spatial region code.

[0173] There is an intersection between the two lists, that is, there are Cell IDs that appear in both lists. For example, referring to Table 1, the intersection of the first Cell ID list and the second Cell ID list corresponding to the spatial region code L1 is Cell11, Cell12, Cell17, and Cell18; the intersection of the first Cell ID list and the second Cell ID list corresponding to the spatial region code L2 is Cell25 and Cell31; the intersection of the first Cell ID list and the second Cell ID list corresponding to the spatial region code L3 is Cell41 and Cell42; the intersection of the first Cell ID list and the second Cell ID list corresponding to the spatial region code L4 is Cell55.

[0174] In a possible implementation, the UE may narrow the range of the grid area or reduce the number of grid areas participating in the matching according to the service area information (e.g., city information) to reduce the amount of calculation. For example, the UE may roughly determine the location according to the current base station identifier and Cell ID, and reduce the number of grid areas participating in the matching in combination with the regional information. Figure 4 In the calculation, only the grid areas corresponding to the spatial area codes L1-L4 are matched, and other grid areas such as L5 do not need to be matched. In addition, the range of the grid areas involved in the matching can be narrowed. For example, for the grid area L1, when the first Cell ID list of the UE includes Cell12, due to the limited communication capability between the UE and the base station, the 9 cells on the leftmost side of the grid area corresponding to the spatial area code L1 can be excluded from the calculation, because the distance between these 9 cells and Cell12 is too large and there is no intersection with the first Cell ID list.

[0175] When the number of Cell IDs included in the intersection is greater than or equal to the first threshold, the UE considers that the corresponding grid area determines to meet the screening condition of the space state parameter, and the screening condition is also the first screening condition.

[0176] Specifically, when the number of Cell IDs included in the intersection is greater than or equal to the first threshold, the UE uses the spatial area encoding of the second Cell ID list corresponding to the intersection as the spatial screening condition of the AI ​​model index.

[0177] The embodiment of the present application does not specifically limit the first threshold value. In the following description, the first threshold value of 3 is taken as an example.

[0178] Continuing with the above example, for each grid area corresponding to the spatial area codes L1-L4, only the number of Cell IDs in the intersection corresponding to L1 is greater than 3, so the spatial area code L1 is used as the spatial screening condition for the AI ​​model index. At this time, the AI ​​model indexes in Table 2 that meet the spatial screening condition are P1-P8.

[0179] Furthermore, the screening process of the spatial region coding may also refer to the number of Cell ID intersections and other information in the PRS at the same time.

[0180] For example, RSRP is further added for joint judgment. For each Cell ID in the intersection, it is further determined whether the RSRP corresponding to each Cell ID is greater than or equal to the preset power threshold, and whether the number of Cell IDs whose RSRP is greater than or equal to the preset power threshold is greater than or equal to the first threshold. If so, the spatial area encoding of the second Cell ID list corresponding to the intersection is used as the spatial screening condition for the AI ​​model index. Similarly, AoA can also be further added for joint judgment.

[0181] For another example, RSRP and AoA are further added for joint judgment. For each Cell ID in the intersection, it is further determined whether the RSRP corresponding to each Cell ID is greater than or equal to the preset power threshold, and whether the AoA corresponding to each Cell ID meets the preset angle range, and then it is determined whether the number of Cell IDs that meet the above conditions at the same time is greater than or equal to the first threshold. If so, the spatial area encoding of the second Cell ID list corresponding to the intersection is used as the spatial screening condition of the AI ​​model index.

[0182] For the time state parameter, since it has been discretized into a time interval, the UE can compare the positioning time with each time interval to determine the time interval in which the positioning time is located, or the time interval closest to the positioning time. There can be one or more time intervals determined by the time parameter. Please refer to Table 2 above for an example for specific explanation.

[0183] In one example, the time interval corresponding to the time parameter T1 in Table 2 is [7:00, 8:00], and the time interval corresponding to the time parameter T2 is [9:00, 10:00]. When the positioning time of the UE is 7:50, since it is in [7:00, 8:00], the time parameter T1 is used as the time filtering condition. At this time, among P1-P8 that meet the spatial filtering condition, the AI ​​model index that also meets the time filtering condition is P1-P4.

[0184] In another example, the time interval corresponding to the time parameter T1 in Table 2 is [7:00, 8:00], and the time interval corresponding to the time parameter T2 is [9:00, 10:00]. When the positioning time of the UE is 6:50, since it is not in any of the above time periods, the time parameter of the time period closest to the positioning time can be determined as the time screening condition, that is, the time parameter T1 is used as the time screening condition.

[0185] In another example, the time interval corresponding to the time parameter T1 in Table 2 is [7:00, 8:00], and the time interval corresponding to the time parameter T2 is [7:30, 8:30]. When the positioning time of the UE is 7:50, it is within the above two time intervals at the same time, so the time parameters T1 and T2 are both used as time filtering conditions. At this time, P1-P8 meets both the time filtering condition and the space filtering condition.

[0186] In another example, the basis for determining the time screening condition is whether the time interval between the positioning time and the reference value corresponding to the time interval is greater than a preset interval. If less than, the time state parameter corresponding to the time interval is used as the time screening condition, and the time screening condition is also the second screening condition. There can be one or more determined time intervals.

[0187] Among them, the benchmark value is the middle value of the time interval, and the preset interval can be set according to the actual AI model corresponding to the geographical area. The embodiment of the present application is not specifically limited and is illustrated by examples below.

[0188] For example, the time interval corresponding to the time parameter T1 in Table 2 is [8:00,9:00], and the corresponding reference value of [8:00,9:00] is 8:30; the time interval corresponding to the time parameter T2 is [9:00,10:00], and the corresponding reference value of [9:00,10:00] is 9:30. The preset interval can be set to 1 hour. When the positioning time of the UE is 9:20, the time interval between the positioning time and 8:30 is 50 minutes, which is less than the preset interval; the time interval between the positioning time and 9:30 is 10 minutes, which is less than the preset interval, so both time parameter T1 and time parameter T2 can be used as time screening conditions.

[0189] The scene state parameter can identify the scene in which the UE is located during positioning. After the UE determines the scene in which it is located, the scene state parameter corresponding to the scene can be used as a scene filtering condition. The scene filtering condition is the third filtering condition, which is illustrated below with an example.

[0190] Continuing to refer to Table 2, the scene corresponding to the scene state parameter S1 is the high-speed rail scene, and the scene corresponding to the scene state parameter S1 is the subway scene. When the UE determines that the scene it is in is the high-speed rail scene, the scene state parameter S1 is determined as the time screening condition. At this time, among P1-P8 that meet the spatial screening condition, the AI ​​model indexes that also meet the scene screening condition are P1, P2, P5, and P6.

[0191] For the network device state parameters, the network device state parameters can be quantified into discrete intervals for matching. Taking the network device state parameters related to time as an example, the network device screening conditions can be determined in a similar manner to the time state parameters, as illustrated below.

[0192] Continue to refer to Table 2 above. The network device status parameter is Q1, indicating that the configuration parameters of the current network device are the configuration parameters of the working day, and the network device status parameter is Q2, indicating that the configuration parameters of the current network device are the configuration parameters of the holidays. The UE can determine that the current time is on a working day based on the positioning time, and the network device status parameter Q1 is used as the network device filtering condition. At this time, among P1, P2, P5, and P6 that simultaneously meet the spatial filtering conditions (L1), time filtering conditions (T1, T2), and scene filtering conditions (S1), the AI ​​model indexes that also meet the network device filtering conditions are P1 and P5.

[0193] The network device status parameters related to the space can determine the network device screening conditions in a manner similar to the spatial area coding. The network device screening conditions are the fourth screening conditions and will not be described in detail here.

[0194] Through the above method, one or more matching AI model indexes are determined.

[0195] S24: The UE sends the first index information to the LMF network element.

[0196] The first index information includes one or more AI model indexes.

[0197] In a possible implementation, the UE and the LMF network element may communicate based on the LTE positioning protocol (LPP). The first index information may be carried in LPP-related signaling.

[0198] S25: The LMF network element determines an AI model corresponding to each AI model index in one or more AI model indexes.

[0199] Continue to see Figure 5, an AI model library for UE positioning is deployed on the LMF network element side. After the LMF network element obtains one or more AI model indexes, it matches the AI ​​model corresponding to each AI model index from the AI ​​model library.

[0200] S26: The LMF network element sends all determined AI models to the UE.

[0201] In one possible implementation, the UE and LMF can communicate based on LPP, and the relevant data of the AI ​​model can be carried in the LPP-related signaling.

[0202] S27: The UE uses a combination of environmental parameters to determine the weight of the positioning results output by each AI model.

[0203] When fusing the positioning results, the present application solution needs to determine the weight of each AI model, which is also the weight of the positioning result output by the AI ​​model. After the weights corresponding to each AI model are determined, the positioning results of each AI model are multiplied by the corresponding weights, and then the calculated results are accumulated to determine the final positioning result.

[0204] It can be understood that if only one AI model index is determined in S23, the LMF network element sends a corresponding AI model to the UE, then the weight of the AI ​​model can be determined as 1, that is, the positioning result of the AI ​​model can be directly used as the final positioning result. The following introduces the situation when multiple AI model indexes are determined in S23.

[0205] For each AI model, the weight coefficient K of its corresponding AI model can be determined by the following formula:

[0206] (1)

[0207] Where n is the number of environmental parameters in the environmental parameter combination, and n is a positive integer; K i is the weight coefficient corresponding to the i-th environmental parameter; p i is the influencing factor of the i-th environmental parameter. The influencing factor is related to the attribute of the environmental parameter. For example, the first influencing factor related to space is greater than the second influencing factor related to time. The influencing factor is used to calibrate and adjust the weight coefficient corresponding to each environmental parameter. In formula (1), each environmental parameter is weighted and summed with the corresponding influencing factor to obtain the weight coefficient K of each AI model.

[0208] The larger the weight coefficients are, the higher the credibility of the AI ​​model selected by the environmental parameters corresponding to the weight coefficients, and the closer the positioning result of the AI ​​model is to the actual location.

[0209] In the embodiment of the present application, taking the example that various environmental parameters include space state parameters, time state parameters, scene state parameters and network device state parameters, n is 4.

[0210] For the spatial state parameter, that is, the spatial region code, when screening the model, the more the number of Cell IDs in the intersection, the higher the weight coefficient of the corresponding spatial state parameter, and the weight of the spatial state parameter is the first weight coefficient. The first weight coefficient is positively correlated with the number of Cell IDs in the intersection.

[0211] In a possible implementation, the weight coefficient is the ratio of the number of Cell IDs in the intersection corresponding to the spatial region coding to the number of Cell IDs included in the first Cell ID list, which is illustrated below by an example.

[0212] For example, the first Cell ID list formed by all Cell IDs indicated in the PRS includes 10 Cell IDs. The intersection of the second Cell ID list corresponding to the spatial area code L1 and the first Cell ID list includes 5 Cell IDs, and the weight coefficient corresponding to the spatial area code L1 is 0.5; the intersection of the second Cell ID list corresponding to the spatial area code L2 and the first Cell ID list includes 4 Cell IDs, and the weight coefficient corresponding to the spatial area code L1 is 0.4.

[0213] In another possible implementation, the weight coefficient is the ratio between the number of Cell IDs in the intersection corresponding to the spatial region encoding and the number of Cell IDs in the intersection corresponding to all AI model indexes matched in S23, as illustrated below.

[0214] For example, the AI ​​model indexes matched in S23 are P1 and P9 in Table 2, where the spatial region encoding of P1 is L1, and the intersection of the second Cell ID list corresponding to L1 and the first Cell ID list includes 5 Cell IDs. The spatial region encoding of P9 is L2, and the intersection of the second Cell ID list corresponding to L2 and the first Cell ID list includes 3 Cell IDs. Then the weight coefficient corresponding to L1 = 5 / (5+3) = 0.625, and the weight coefficient corresponding to L2 = 3 / (5+3) = 0.375.

[0215] The size of the second weight coefficient of the time state parameter is negatively correlated with the time interval between the reference value of the time interval and the positioning time. The larger the time interval, the smaller the second weight coefficient. The following example illustrates this.

[0216] For example, the time interval corresponding to the time parameter T1 in Table 2 is [8:00, 9:00], and the corresponding reference value of [8:00, 9:00] is 8:30; the time interval corresponding to the time parameter T2 is [9:00, 10:00], and the corresponding reference value of [9:00, 10:00] is 9:30. When the positioning time of the UE is 9:20, the time interval between the positioning time and 8:30 is 50 minutes, and the corresponding second weight coefficient is K 21 ; The time interval between the positioning time and 9:30 is 10 minutes, and the corresponding second weight coefficient is K 22 Then K 22 Greater than K 21 , indicating that the prediction results of the AI ​​model obtained by using T2 as the time screening condition are more reliable.

[0217] In a possible implementation, the weight coefficient K2 of the time state parameter can be determined by the following formula:

[0218] K2=1-Ts / T0 (2)

[0219] Wherein, Ts is the time interval between the reference value of the time interval and the positioning time, and T0 is the preset interval.

[0220] For example, taking the time interval corresponding to the time parameter T1 in Table 2 as [8:00,9:00] as an example, the reference value is 8:30, the positioning time is 9:20, the preset interval is one hour, and the weight coefficient corresponding to T1 is K2=0.167.

[0221] For another example, taking the time interval corresponding to the time parameter T2 in Table 2 as [9:00, 10:00] as an example, the reference value is 9:30, the positioning time is 9:20, the preset interval is one hour, and the weight coefficient corresponding to T1 is K2=0.833.

[0222] The above is only an example. The weight of the time state parameter can be determined in other ways, which will not be described in detail in the embodiments of the present application.

[0223] For scene state parameters, the third weight coefficient corresponding to each scene can be calibrated by big data and AI model analysis. For example, the third weight coefficient corresponding to the high-speed rail scene is 0.2, the weight coefficient corresponding to the indoor scene is 0.5, etc. The embodiment of the present application is not specifically limited. The correspondence between the scene and the weight coefficient can be predetermined and stored in the UE locally, or obtained through the network when the UE is positioned.

[0224] For network device status parameters, a similar method can also be used to determine the weight coefficients corresponding to each network device status parameter. For example, for time-related network device status parameters, a method similar to the time status parameters in the above description can be adopted, and for space-related network device status parameters, a method similar to the space status parameters in the above description can be adopted, and the embodiments of the present application will not be repeated here.

[0225] After the weight coefficient K corresponding to each model is determined by formula (1), the weight coefficients of each AI model are normalized to obtain the weight of the AI ​​model distributed between [0,1]. After normalization, the weight coefficient K of each AI model is converted to weight k, and the weight coefficients of all AI models are added together to equal 1.

[0226] The following is an example. Continuing to refer to Table 2, take AI model indexes P1 and P15 as an example.

[0227] For P1, the weight coefficient K1 corresponding to the spatial region coding is 0.5, the weight coefficient K2 corresponding to the time state parameter is 0.4, the weight K3 corresponding to the scene state parameter is 0.8, and the weight coefficient K4 corresponding to the network device state parameter is 0.5. Among them, K1 and K3 are environmental parameters related to space, and the influence factors are relatively large, for example, the first influence factor is set to 1.1; K2 and K4 are environmental parameters related to time, and the influence factors are relatively small, for example, the second influence factor is set to 0.8. Then, the weight coefficient K corresponding to the AI ​​model index P1 is determined according to formula (1): P1 =1.1*0.5+0.8*0.4+1.1*0.8+0.8*0.5=2.15.

[0228] For P15, the weight coefficient K1 corresponding to the spatial region coding is 0.4, the weight coefficient K2 corresponding to the time state parameter is 0.8, the weight K3 corresponding to the scene state parameter is 0.6, and the weight coefficient K4 corresponding to the network device state parameter is 0.5.

[0229] According to formula (1), the weight coefficient K corresponding to the AI ​​model index P15 is determined P15 =1.1*0.4+0.8*0.4+1.1*0.6+0.8*0.5=1.82.

[0230] K P1 and K P15 Normalize and get the weight k corresponding to P1 P1 =( K P1 ) / ( K P1 +K P15 ), which is about 0.542, and the weight k corresponding to P2 P15 =( K P15 ) / ( KP1 +K P15 ), which is approximately 0.458.

[0231] S28: The UE uses the PRS, each AI model and the weight corresponding to each AI model to determine the final positioning result.

[0232] The UE uses the PRS as the input of each AI model to obtain the positioning results output by each AI model. The positioning results output by the AI ​​model can be the absolute position or relative position of the UE, which is not specifically limited in the embodiments of the present application.

[0233] It is understandable that the types of positioning results output by each AI model are generally the same, for example, both are absolute positions or both are relative positions.

[0234] The absolute position may be longitude and latitude coordinates. The relative position may be the position coordinates of the UE relative to the serving base station and / or the base station corresponding to the neighboring cell. The relative position may also be represented by other methods. The embodiment of the present application does not specifically limit the representation method of the relative position.

[0235] The final positioning result can be determined by the following formula:

[0236] (3)

[0237] Where j is the number of AI models and is a positive integer. j is the weight of the jth AI model. j is the positioning result output by the jth AI model. That is, the positioning results output by each AI model are multiplied by the weights corresponding to each AI model to obtain the final positioning result components corresponding to each AI model, and the final positioning result components corresponding to each AI model are superimposed to obtain the final positioning result.

[0238] The following takes the positioning result as absolute position and the number of AI models as 2 as an example for explanation.

[0239] In one example, the positioning result of AI model P1 is the longitude and latitude coordinates (x1, y1), and the weight k1 of P1 is 0.542; the positioning result of AI model P15 is the longitude and latitude coordinates (x2, y2), and the weight k1 of P15 is 0.458. The final positioning result can be the longitude and latitude coordinates (0.542 x1+0.458x2, 0.542 y1+0.458y2).

[0240] In the embodiment of the present application, the UE can use the weight of the AI ​​model with the highest weight as the credibility of the final positioning result. The higher the weight, the higher the credibility. For example, when only one AI model that best matches is determined, the weight of the AI ​​model is 1, and the corresponding credibility is 1, indicating that the positioning result is predicted by the AI ​​model that fully matches the current environmental parameter combination, and the result is accurate and reliable.

[0241] It can be understood that the division of the above steps is only for the convenience of explanation and does not constitute a limitation on the technical solution of the present application. In actual applications, the order of the above steps may be changed. For example, S27 can be advanced to before S24, or S27 can be performed in parallel with S24.

[0242] In summary, the positioning method provided in the embodiment of the present application is that the UE can determine the AI ​​model index corresponding to one or more AI models used for positioning, and then send all the AI ​​model indexes to the LMF network element. The LMF network element matches the corresponding AI model from the AI ​​model library according to each AI model index and sends it to the UE. The UE can determine the weights corresponding to each AI model, and use each AI model and the corresponding weight to determine the final positioning result. In the prior art, when it is impossible to determine an AI model that completely matches the current environmental parameter combination of the UE, it is directly considered that the positioning cannot be completed. The solution provided in the embodiment of the present application can use multiple AI models with a high degree of matching with the current environmental parameters of the UE for positioning, and fuse the positioning results according to the weights, thereby obtaining a more accurate final positioning result. It is particularly suitable for service areas with complex and changeable real environments, or areas where the AI ​​models used for positioning are not sufficiently trained, and has high practicality. At the same time, after the LMF network element sends the AI ​​model to the UE, for scenarios where continuous positioning is required on the UE side, the UE side can directly use the received AI model for positioning. If the environmental parameter combination does not change subsequently, there is no need for frequent interaction with the LMF network element, thereby reducing the positioning time delay.

[0243] In the above embodiments, the LMF network element sends the determined AI models to the UE, and the UE performs fusion positioning. The following describes an implementation method in which the LMF network element side directly performs fusion positioning based on the determined AI models and sends the final positioning result to the UE side.

[0244] See also Figure 6 , which is a flowchart of another positioning method based on an AI model provided in an embodiment of the present application.

[0245] In the following description, the terminal device is taken as UE, and the AI ​​model is deployed in the LMF network element as an example. The method includes the following steps:

[0246] S41: UE receives PRS.

[0247] The PRS received by the UE can be sent by multiple base stations. For the convenience of explanation, the embodiment of the present application takes the base station as gNB as an example, but it does not constitute a limitation on the base station type.

[0248] The PRS sent by each base station may carry one or more of the Cell ID, RSRP, and AoA relative to the base station. In a possible implementation, the PRS sent by multiple base stations may carry the Cell IDs of the serving cell and the neighboring cell. In some embodiments, the UE side may not receive the neighboring cell Cell ID, in which case the neighboring cell PCI may be used to distinguish the neighboring cells.

[0249] S42: The UE obtains a combination of environmental parameters of the current environment.

[0250] The environmental parameter combination includes the status parameters of the terminal device and the status parameters of the network device.

[0251] The state parameters of the terminal device can be used to characterize the state parameters of the environment in which the terminal device, that is, the UE, is currently located, which may include but are not limited to spatial state parameters, temporal state parameters and scene state parameters.

[0252] For detailed description of the environmental parameter combination, please refer to the introduction in S22 above, which will not be repeated here.

[0253] S43: The UE uses the received positioning reference signal PRS and the environmental parameter combination to determine one or more matching AI model indexes.

[0254] The AI ​​model index is a data structure used for quick query and AI model, which can be understood as the identification of the AI ​​model. Each AI model has a unique corresponding AI model index, and the embodiment of the present application does not limit the specific data structure of the AI ​​model index. The UE can store the correspondence between various environmental parameter combinations and AI model indexes in the form of a mapping table in advance, and see Table 2 above for details.

[0255] The UE may firstly aggregate all received PRSs, and then obtain all Cell IDs. All Cell IDs indicated in the PRSs may form a first Cell ID list.

[0256] The Cell ID list corresponding to the spatial area code may be referred to as a second Cell ID list.

[0257] Determine the intersection of the first Cell ID list and the second Cell ID list corresponding to each spatial region code.

[0258] When the number of Cell IDs included in the intersection is greater than or equal to the first threshold, the UE considers that the corresponding grid area determines to meet the filtering conditions of the spatial state parameters, and encodes the spatial area of ​​the second Cell ID list corresponding to the intersection as the spatial filtering condition of the AI ​​model index.

[0259] Furthermore, the screening process of the spatial area coding can also refer to the number of Cell ID intersections and other information in the PRS at the same time. For example, RSRP and / or AoA are further added for joint judgment, that is, after determining the number of Cell IDs in the intersection, it is further determined whether the RSRP of the base stations corresponding to each Cell ID in the intersection is greater than or equal to the preset power threshold, and / or the intersection, and whether the number of Cell IDs corresponding to the base stations in the set AoA that meet the preset angle range is greater than or equal to the first threshold. If so, the spatial area coding of the second Cell ID list corresponding to the intersection is used as the spatial screening condition for the AI ​​model index.

[0260] For the time state parameter, since it has been discretized into time intervals, the UE can compare the positioning time with each time interval to determine the time interval in which the positioning time is located, or the time interval closest to the positioning time. There can be one or more time intervals determined by the time parameter.

[0261] In one possible implementation, the basis for determining the time screening condition is whether the time interval between the positioning time and the reference value corresponding to the time interval is greater than the preset interval. If less than, the time state parameter corresponding to the time interval is used as the time screening condition. Among them, the reference value is the middle value of the time interval, and the preset interval can be set according to the actual AI model corresponding to the geographical area, which is not specifically limited in the embodiments of the present application.

[0262] The scene state parameters can identify the scene in which the UE is located during positioning. After the UE determines the scene in which it is located, the scene state parameters corresponding to the scene can be used as scene screening conditions.

[0263] For network device status parameters, they can be quantified into discrete intervals for matching. Taking the network device status parameters related to time as an example, the network device screening conditions can be determined in a similar manner to the time status parameters; the network device status parameters related to space can be determined in a similar manner to the spatial region coding.

[0264] Through the above method, one or more matching AI model indexes are determined.

[0265] S44: The UE sends the first index information, PRS and environmental parameter combination to the LMF network element.

[0266] See also Figure 7 , which is a schematic diagram of the framework for positioning based on the AI ​​model provided in the embodiment of the present application Figure 2 .

[0267] Figure 7 The corresponding plan and Figure 5 The difference between the corresponding solutions is that Figure 7 The UE needs to send PRS and environmental parameter combinations to the network device. PRS is used as the input of each AI model, and the environmental parameter combination is used by the LMF side to determine the corresponding weights of each AI model.

[0268] In a possible implementation, the UE and the LMF may communicate based on LPP, and the first index information, PRS, and environmental parameter combination may be carried in LPP-related signaling.

[0269] It is understandable that in another possible implementation, the UE can also determine the weights corresponding to each AI model locally and send the weights directly to the LMF network element, that is, the LPP-related signaling carries the first index information, PRS, and the weights corresponding to each AI model. In this implementation, the LMF network element can directly use the corresponding weights of each AI model sent by the UE for fusion positioning.

[0270] S45: The LMF network element determines an AI model corresponding to each AI model index in one or more AI model indexes.

[0271] The LMF network element side is deployed with an AI model library for UE positioning. The AI ​​model library includes multiple trained AI models, each of which corresponds to an AI model index.

[0272] After the LMF network element obtains one or more AI model indexes, it matches the AI ​​models corresponding to each AI model index from the AI ​​model library.

[0273] S46: The LMF network element uses a combination of environmental parameters to determine the weight of the positioning results output by each AI model.

[0274] In the solution of the embodiment of the present application, the weight of the positioning result output by each AI model is determined by the LMF network element, rather than being performed on the UE. The computing power of the LMF network element can be used to quickly obtain the weight result, thereby improving the data processing speed.

[0275] It can be understood that if only one AI model index is determined in S45, the weight of the AI ​​model can be directly determined as 1, that is, the positioning result of the AI ​​model can be directly used as the final positioning result. The following describes the situation when multiple AI model indexes are determined in S45.

[0276] For each AI model, the weight coefficient K of the corresponding AI model can be determined by the above formula (1).

[0277] The larger the weight coefficients are, the higher the credibility of the AI ​​model selected by the environmental parameters corresponding to the weight coefficients, and the closer the positioning result of the AI ​​model is to the actual location.

[0278] In the embodiment of the present application, taking the example that various environmental parameters include space state parameters, time state parameters, scene state parameters and network device state parameters, n is 4.

[0279] For the spatial state parameters, that is, the spatial region coding, when screening the model, the more the number of Cell IDs in the intersection, the higher the weight coefficient of the corresponding spatial state parameters.

[0280] In a possible implementation manner, the weight coefficient is a ratio of the number of Cell IDs in the intersection corresponding to the spatial region coding to the number of Cell IDs included in the first Cell ID list.

[0281] In another possible implementation, the weight coefficient is the ratio between the number of Cell IDs in the intersection corresponding to the spatial region encoding and the sum of the number of Cell IDs in the intersection corresponding to all AI model indexes matched in S45.

[0282] For the time state parameter, the size of the weight coefficient is negatively correlated with the time interval between the baseline value of the time interval and the positioning time. The larger the time interval, the smaller the weight coefficient.

[0283] In a possible implementation, the weight coefficient K2 of the time state parameter can be determined by the above formula (2).

[0284] For scene state parameters, the weight coefficients corresponding to each scene can be calibrated through big data and AI model analysis, and the correspondence between the scene and the weight can be saved in the LMF network element.

[0285] For network device status parameters, a similar method can also be used to determine the weight coefficients corresponding to each network device status parameter. For example, for time-related network device status parameters, a method similar to the time status parameters in the above description can be adopted, and for space-related network device status parameters, a method similar to the space status parameters in the above description can be adopted, and the embodiments of the present application will not be repeated here.

[0286] After the weight coefficient K corresponding to each model is determined by formula (1), the weight coefficients of each AI model are normalized to obtain the weight of the AI ​​model distributed between [0,1]. After normalization, the weight coefficient K of each AI model is converted to weight k, and the weight coefficients of all AI models are added together to equal 1.

[0287] S47: The LMF network element uses the PRS, each AI model and the weight corresponding to each AI model to determine the final positioning result.

[0288] The LMF network element uses the PRS sent by the UE as the input of each AI model to obtain the positioning results output by each AI model. The positioning result output by a single AI model can be the absolute position or relative position of the UE, which is not specifically limited in the embodiments of the present application.

[0289] It is understandable that the types of positioning results output by each AI model are generally the same, for example, both are absolute positions or both are relative positions.

[0290] The absolute position may be longitude and latitude coordinates. The relative position may be the position coordinates of the UE relative to the serving base station and / or the base station corresponding to the neighboring cell. The relative position may also be represented by other methods. The embodiment of the present application does not specifically limit the representation method of the relative position.

[0291] The final positioning result can be determined by the above formula (3), that is, the positioning result output by each AI model is multiplied by the weight corresponding to each AI model to obtain the final positioning result component corresponding to each AI model, and the final positioning result components corresponding to each AI model are superimposed to obtain the final positioning result corresponding to each AI model.

[0292] S48: The LMF network element sends location information to the UE.

[0293] The location information includes the final positioning result.

[0294] In a possible implementation, the LMF network element may carry location information in the relevant signaling of the LTE positioning protocol.

[0295] It can be understood that the division of the above steps is only for the convenience of explanation and does not constitute a limitation on the technical solution of the present application. In actual applications, the order of the above steps may be changed. For example, S47 can be advanced to before S45, or S47 can be performed in parallel with S45.

[0296] In summary, the positioning method provided in the embodiment of the present application is that the UE can determine the AI ​​model index corresponding to one or more AI models used for positioning, and then send all the AI ​​model indexes to the LMF network element. The LMF network element matches the corresponding AI model from the AI ​​model library according to each AI model index and determines the weight corresponding to each AI model. Then the LMF can use each AI model and the corresponding weight to determine the final positioning result, and send the final positioning result to the UE. In the existing scheme, when it is impossible to determine an AI model that completely matches the current environmental parameter combination of the UE, it will be directly considered that the positioning cannot be completed. The scheme of the present application can use multiple AI models with a high degree of matching with the current environmental parameters of the UE for positioning, and fuse the positioning results according to the weights, so as to obtain a more accurate final positioning result. It is particularly suitable for service areas with complex and changeable real environments, or areas where the AI ​​models used for positioning are not sufficiently trained, and has high practicality.

[0297] At the same time, for scenarios where continuous positioning is not required on the UE side, such as single positioning or fewer positioning times, the UE side can directly obtain the positioning results sent by the LMF network element. In this way, the UE side does not need to receive the AI ​​model sent by the LMF network element, which can significantly reduce the amount of data received and reduce the time delay of single positioning.

[0298] In the above description, taking the UE side storing a mapping table and determining the AI ​​model index as an example, the following describes the implementation method of determining the AI ​​model index by the LMF network element.

[0299] See also Figure 8 , which is a flowchart of another positioning method based on an AI model provided in an embodiment of the present application.

[0300] In the following description, the terminal device is taken as UE, and the AI ​​model is deployed in the LMF network element as an example. The method includes the following steps:

[0301] S61: UE receives PRS.

[0302] The PRS received by the UE may be sent by multiple base stations.

[0303] The PRS sent by each base station may carry one or more of the Cell ID, RSRP, and AoA relative to the base station. In a possible implementation, the PRS sent by multiple base stations may carry the Cell IDs of the serving cell and the neighboring cell. In some embodiments, the UE side may not receive the neighboring cell Cell ID, in which case the neighboring cell PCI may be used to distinguish the neighboring cells.

[0304] S62: The UE obtains a combination of environmental parameters of the current environment.

[0305] The environmental parameter combination includes the status parameters of the terminal device and the status parameters of the network device.

[0306] The state parameters of the terminal device can be used to characterize the state parameters of the environment in which the terminal device, that is, the UE, is currently located, which may include but are not limited to spatial state parameters, temporal state parameters and scene state parameters.

[0307] For detailed description of the environmental parameter combination, please refer to the introduction in S22 above, which will not be repeated here.

[0308] S63: The UE sends an environmental parameter combination and a PRS to the LMF network element.

[0309] See also Fig. 9 , which is a schematic diagram of the framework for positioning based on the AI ​​model provided in the embodiment of the present application Figure 3 .

[0310] Fig. 9 The corresponding plan and Figure 5 and Figure 7 The difference is that: the mapping table is not stored on the UE side, and the mapping table is stored on the LMF network element side. Therefore, the LMF network element needs to determine the AI ​​model index based on the mapping table. Therefore, the UE needs to send the environmental parameter combination and PRS to the LMF network element.

[0311] In a possible implementation, the UE and LMF may communicate based on LPP, and the PRS and environmental parameter combination may be carried in LPP-related signaling.

[0312] S64: The LMF network element uses the environmental parameter combination and PRS to determine one or more matching AI model indexes.

[0313] The UE may send the PRS sent by each base station to the LMF separately in sequence, or aggregate them and send them to the LMF together, which is not specifically limited in the embodiments of the present application.

[0314] The mapping table is a table of correspondence between various environmental parameter combinations and AI model indexes. For details, please refer to Table 2 above.

[0315] After receiving the PRS, the LMF network element aggregates the information of all PRSs to obtain all Cell IDs. All Cell IDs indicated in the PRS can form a first Cell ID list.

[0316] The LMF network element determines the intersection of the first Cell ID list and the second Cell ID list corresponding to each spatial area code.

[0317] When the number of Cell IDs included in the intersection of the LMF network element is greater than or equal to the first threshold, the corresponding grid area is considered to meet the filtering conditions of the spatial state parameters, and the spatial area encoding of the second Cell ID list corresponding to the intersection is used as the spatial filtering condition of the AI ​​model index.

[0318] Furthermore, the screening process of the spatial area coding can also refer to the number of Cell ID intersections and other information in the PRS at the same time. For example, RSRP and / or AoA are further added for joint judgment, that is, after determining the number of Cell IDs in the intersection, it is further determined whether the RSRP of the base stations corresponding to each Cell ID in the intersection is greater than or equal to the preset power threshold, and / or the intersection, and whether the number of Cell IDs corresponding to the base stations in the set AoA that meet the preset angle range is greater than or equal to the first threshold. If so, the spatial area coding of the second Cell ID list corresponding to the intersection is used as the spatial screening condition for the AI ​​model index.

[0319] For the time state parameters, since they have been discretized into time intervals, the LMF network element can compare the positioning time with each time interval to determine the time interval in which the positioning time is located, or the time interval closest to the positioning time. There can be one or more time intervals determined by the time parameters.

[0320] In one possible implementation, the basis for determining the time screening condition is whether the time interval between the positioning time and the reference value corresponding to the time interval is greater than the preset interval. If less than, the time state parameter corresponding to the time interval is used as the time screening condition. Among them, the reference value is the middle value of the time interval, and the preset interval can be set according to the actual AI model corresponding to the geographical area, which is not specifically limited in the embodiments of the present application.

[0321] The scene state parameters can identify the scene in which the UE is located during positioning. After the UE determines the scene in which it is located, the scene state parameters corresponding to the scene can be used as scene screening conditions.

[0322] For network device status parameters, they can be quantified into discrete intervals for matching. Taking the network device status parameters related to time as an example, the network device screening conditions can be determined in a similar manner to the time status parameters; the network device status parameters related to space can be determined in a similar manner to the spatial region coding.

[0323] Through the above method, the LMF network element can determine one or more matching AI model indexes through the mapping table.

[0324] S65: The LMF network element determines an AI model corresponding to each AI model index in the one or more AI model indexes.

[0325] An AI model library for UE positioning is deployed on the LMF network element side. The AI ​​model library includes multiple trained AI models, and each AI model corresponds to an AI model index.

[0326] After the LMF network element obtains one or more AI model indexes, it matches the AI ​​models corresponding to each AI model index from the AI ​​model library.

[0327] S66: LMF sends all determined AI models to UE.

[0328] In one possible implementation, the UE and LMF can communicate based on LPP, and the relevant data of the AI ​​model can be carried in the LPP-related signaling.

[0329] S67: The UE uses a combination of environmental parameters to determine the weight of the positioning results output by each AI model.

[0330] When fusing the positioning results, the present application solution needs to determine the weight of each AI model, which is also the weight of the positioning result output by the AI ​​model. After the weights corresponding to each AI model are determined, the positioning results of each AI model are multiplied by the corresponding weights, and then the calculated results are accumulated to determine the final positioning result.

[0331] It can be understood that if only one AI model index is determined, the weight of the AI ​​model can be directly determined as 1, that is, the positioning result of the AI ​​model can be directly used as the final positioning result. The following describes the situation when multiple AI model indexes are determined.

[0332] For each AI model, the weight coefficient K of the corresponding AI model can be determined by the above formula (1).

[0333] The larger the weight coefficients are, the higher the credibility of the AI ​​model selected by the environmental parameters corresponding to the weight coefficients, and the closer the positioning result of the AI ​​model is to the actual location.

[0334] For the spatial state parameters, that is, the spatial region coding, when screening the model, the more the number of Cell IDs in the intersection, the higher the weight coefficient of the corresponding spatial state parameters.

[0335] In a possible implementation manner, the weight coefficient is a ratio of the number of Cell IDs in the intersection corresponding to the spatial region coding to the number of Cell IDs included in the first Cell ID list.

[0336] In another possible implementation, the weight coefficient is the ratio between the number of Cell IDs in the intersection corresponding to the spatial region encoding and the number of Cell IDs in the intersection corresponding to all AI model indexes.

[0337] For the time state parameter, the size of the weight coefficient is negatively correlated with the time interval between the baseline value of the time interval and the positioning time. The larger the time interval, the smaller the weight coefficient.

[0338] In a possible implementation, the weight coefficient K2 of the time state parameter can be determined by the above formula (2).

[0339] For scene state parameters, the weight coefficients corresponding to each scene can be calibrated through big data and AI model analysis, and the correspondence between the scene and the weight can be saved in the UE.

[0340] After the weight coefficient K corresponding to each model is determined by formula (1), the weight coefficients of each AI model are normalized to obtain the weight of the AI ​​model distributed between [0,1]. After normalization, the weight coefficient K of each AI model is converted to weight k, and the weight coefficients of all AI models are added together to equal 1.

[0341] S68: The UE uses the PRS, each AI model, and the weight corresponding to each AI model to determine the final positioning result.

[0342] The UE uses the acquired PRS as the input of each AI model to obtain the positioning results output by each AI model. The positioning result output by a single AI model can be the absolute position or relative position of the UE, which is not specifically limited in the embodiments of the present application. It is understandable that the type of positioning results output by each AI model is generally the same, for example, both are absolute positions or both are relative positions.

[0343] The absolute position may be longitude and latitude coordinates. The relative position may be the position coordinates of the UE relative to the serving base station and / or the base station corresponding to the neighboring cell. The relative position may also be represented by other methods. The embodiment of the present application does not specifically limit the representation method of the relative position.

[0344] The final positioning result can be determined by the above formula (3), that is, the positioning result output by each AI model is multiplied by the weight corresponding to each AI model to obtain the final positioning result component corresponding to each AI model, and the final positioning result components corresponding to each AI model are superimposed to obtain the final positioning result corresponding to each AI model.

[0345] It can be understood that the division of the above steps is only for the convenience of explanation and does not constitute a limitation on the technical solution of the present application. In actual applications, the order of the above steps may be changed. For example, S67 can be advanced to before S63, or S67 can be performed in parallel with S63.

[0346] In summary, in the positioning method provided by the embodiment of the present application, after the UE sends the environmental parameter combination and PRS to the LMF network element, the LMF can determine the AI ​​model index corresponding to one or more AI models used for positioning, and then the LMF network element matches the corresponding AI model from the AI ​​model library according to each AI model index and sends it to the UE. After the UE side determines the weights corresponding to each AI model, the final positioning result is determined using each AI model and the corresponding weight. In the existing related schemes, when it is impossible to determine an AI model that completely matches the current environmental parameter combination of the UE, it is directly considered that the positioning cannot be completed. The scheme provided by the embodiment of the present application can use multiple AI models with a high degree of matching with the current environmental parameters of the UE for positioning, and fuse the positioning results according to the weights, thereby obtaining a more accurate final positioning result. It is particularly suitable for service areas with complex and changeable real environments, or areas where the AI ​​models used for positioning are not sufficiently trained, and has high practicality.

[0347] At the same time, after the LMF network element sends the AI ​​model to the UE, for scenarios where continuous positioning is required on the UE side, the UE side can directly use the received AI model for positioning. If the environmental parameter combination does not change subsequently, there is no need for frequent interaction with the LMF network element, thereby reducing the positioning time delay.

[0348] The following describes how the LMF network element determines the AI ​​model index and performs fusion positioning.

[0349] See also Fig.10 , which is a flowchart of another positioning method based on an AI model provided in an embodiment of the present application.

[0350] In the following description, the terminal device is taken as UE, and the AI ​​model is deployed in the LMF network element as an example. The method includes the following steps:

[0351] S81: UE receives PRS.

[0352] The PRS received by the UE may be sent by multiple base stations.

[0353] The PRS sent by each base station may carry one or more of the Cell ID, RSRP, and AoA relative to the base station. In a possible implementation, the PRS sent by multiple base stations may carry the Cell IDs of the serving cell and the neighboring cell. In some embodiments, the UE side may not receive the neighboring cell Cell ID, in which case the neighboring cell PCI may be used to distinguish the neighboring cells.

[0354] S82: The UE obtains a combination of environmental parameters of the current environment.

[0355] The environmental parameter combination includes the status parameters of the terminal device and the status parameters of the network device.

[0356] The state parameters of the terminal device can be used to characterize the state parameters of the environment in which the terminal device, that is, the UE, is currently located, which may include but are not limited to spatial state parameters, temporal state parameters and scene state parameters.

[0357] For detailed description of the environmental parameter combination, please refer to the introduction in S22 above, which will not be repeated here.

[0358] S83: The UE sends an environmental parameter combination and a PRS to the LMF network element.

[0359] See also Fig.11 , which is a schematic diagram of the framework for positioning based on the AI ​​model provided in the embodiment of the present application Figure 4 .

[0360] Fig.11 In the scheme shown, the mapping table is not stored on the UE side, and the mapping table is stored on the LMF network element side. The LMF network element needs to determine the AI ​​model index based on the mapping table. Therefore, the UE needs to send the environmental parameter combination and PRS to the LMF network element.

[0361] In a possible implementation, the UE and LMF may communicate based on LPP, and the PRS and environmental parameter combination may be carried in LPP-related signaling.

[0362] S84: The LMF network element uses the environmental parameter combination and PRS to determine one or more matching AI model indexes.

[0363] The UE may send the PRS sent by each base station to the LMF separately in sequence, or aggregate them and send them to the LMF together, which is not specifically limited in the embodiments of the present application.

[0364] The mapping table is a table of correspondence between various environmental parameter combinations and AI model indexes. For details, please refer to Table 2 above.

[0365] After receiving the PRS, the LMF network element aggregates the information of all PRSs to obtain all Cell IDs. All Cell IDs indicated in the PRS can form a first Cell ID list.

[0366] The LMF network element determines the intersection of the first Cell ID list and the second Cell ID list corresponding to each spatial area code. When the number of Cell IDs included in the intersection is greater than or equal to the first threshold, the LMF network element considers that the corresponding grid area determines the screening condition that meets the spatial state parameter, and uses the spatial area code of the second Cell ID list corresponding to the intersection as the spatial screening condition of the AI ​​model index.

[0367] Furthermore, the screening process of the spatial area coding can also refer to the number of Cell ID intersections and other information in the PRS at the same time. For example, RSRP and / or AoA are further added for joint judgment, that is, after determining the number of Cell IDs in the intersection, it is further determined whether the RSRP of the base stations corresponding to each Cell ID in the intersection is greater than or equal to the preset power threshold, and / or the intersection, and whether the number of Cell IDs corresponding to the base stations in the set AoA that meet the preset angle range is greater than or equal to the first threshold. If so, the spatial area coding of the second Cell ID list corresponding to the intersection is used as the spatial screening condition for the AI ​​model index.

[0368] For the time state parameters, since they have been discretized into time intervals, the LMF network element can compare the positioning time with each time interval to determine the time interval in which the positioning time is located, or the time interval closest to the positioning time. There can be one or more time intervals determined by the time parameters.

[0369] In one possible implementation, the basis for determining the time screening condition is whether the time interval between the positioning time and the reference value corresponding to the time interval is greater than the preset interval. If less than, the time state parameter corresponding to the time interval is used as the time screening condition. Among them, the reference value is the middle value of the time interval, and the preset interval can be set according to the actual AI model corresponding to the geographical area, which is not specifically limited in the embodiments of the present application.

[0370] The scene state parameters can identify the scene in which the UE is located during positioning. After the UE determines the scene in which it is located, the scene state parameters corresponding to the scene can be used as scene screening conditions.

[0371] For network device status parameters, they can be quantified into discrete intervals for matching. Taking the network device status parameters related to time as an example, the network device screening conditions can be determined in a similar manner to the time status parameters; the network device status parameters related to space can be determined in a similar manner to the spatial region coding.

[0372] Through the above method, the LMF network element can determine one or more matching AI model indexes through the mapping table.

[0373] S85: The LMF network element determines an AI model corresponding to each AI model index in the one or more AI model indexes.

[0374] The LMF network element side is deployed with an AI model library for UE positioning. The AI ​​model library includes multiple trained AI models, each of which corresponds to an AI model index. After the LMF network element obtains one or more AI model indexes, it matches the AI ​​model corresponding to each AI model index from the AI ​​model library.

[0375] S86: The LMF network element uses a combination of environmental parameters to determine the weight of the positioning results output by each AI model.

[0376] When fusing the positioning results, the present application solution needs to determine the weight of each AI model, which is also the weight of the positioning result output by the AI ​​model. After the weights corresponding to each AI model are determined, the positioning results of each AI model are multiplied by the corresponding weights, and then the calculated results are accumulated to determine the final positioning result.

[0377] It can be understood that if only one AI model index is determined, the weight of the AI ​​model can be directly determined as 1, that is, the positioning result of the AI ​​model can be directly used as the final positioning result. The following describes the situation when multiple AI model indexes are determined.

[0378] For each AI model, the weight coefficient K of the corresponding AI model can be determined by the above formula (1).

[0379] The larger the weight coefficients are, the higher the credibility of the AI ​​model selected by the environmental parameters corresponding to the weight coefficients, and the closer the positioning result of the AI ​​model is to the actual location.

[0380] For the spatial state parameters, that is, the spatial region coding, when screening the model, the more the number of Cell IDs in the intersection, the higher the weight coefficient of the corresponding spatial state parameters.

[0381] In a possible implementation manner, the weight coefficient is a ratio of the number of Cell IDs in the intersection corresponding to the spatial region coding to the number of Cell IDs included in the first Cell ID list.

[0382] In another possible implementation, the weight coefficient is the ratio between the number of Cell IDs in the intersection corresponding to the spatial region encoding and the number of Cell IDs in the intersection corresponding to all AI model indexes.

[0383] For the time state parameter, the size of the weight coefficient is negatively correlated with the time interval between the baseline value of the time interval and the positioning time. The larger the time interval, the smaller the weight coefficient.

[0384] In a possible implementation, the weight coefficient K2 of the time state parameter can be determined by the above formula (2).

[0385] For scene state parameters, the weight coefficients corresponding to each scene can be calibrated through big data and AI model analysis, and the correspondence between the scene and the weight can be saved in the UE.

[0386] After LMF determines the weight coefficient K corresponding to each model through formula (1), the weight coefficient of each AI model is normalized to obtain the weight of the AI ​​model distributed between [0,1]. After normalization, the weight coefficient K of each AI model is converted to weight k, and the weight coefficients of all AI models are added together to equal 1.

[0387] S87: The LMF network element uses the PRS, each AI model, and the weight corresponding to each AI model to determine the final positioning result.

[0388] The LMF network element uses the acquired PRS as the input of each AI model to obtain the positioning results output by each AI model. The positioning result output by a single AI model can be the absolute position or relative position of the UE, which is not specifically limited in the embodiments of the present application. It is understandable that the type of positioning results output by each AI model is generally the same, for example, both are absolute positions or both are relative positions.

[0389] The final positioning result can be determined by the above formula (3), that is, the positioning result output by each AI model is multiplied by the weight corresponding to each AI model to obtain the final positioning result component corresponding to each AI model, and the final positioning result components corresponding to each AI model are superimposed to obtain the final positioning result corresponding to each AI model.

[0390] S88: The LMF network element sends location information to the UE.

[0391] The location information includes the final positioning result. In a possible implementation, the LMF network element may carry the location information in the relevant signaling of the LTE positioning protocol.

[0392] It can be understood that the division of the above steps is only for the convenience of explanation and does not constitute a limitation on the technical solution of the present application. In actual applications, the order of the above steps may be changed. For example, S87 can be advanced to before S84, or S47 can be performed in parallel with S45.

[0393] In summary, the positioning method provided by the embodiment of the present application is that the UE side sends the PRS and environmental parameter combination to the LME network element, and the LMF network element can use the PRS and environmental parameter combination to determine the AI ​​model index. The LMF network element matches the corresponding AI model from the AI ​​model library according to each AI model index and determines the corresponding weight of each AI model. Then LMF can use each AI model and the corresponding weight to determine the final positioning result, and send the final positioning result to the UE. In the existing related schemes, when it is impossible to determine an AI model that completely matches the current environmental parameter combination of the UE, it will be directly considered that the positioning cannot be completed. The scheme provided by the embodiment of the present application can use multiple AI models with a high degree of matching with the current environmental parameters of the UE for positioning, and fuse the positioning results according to the weights, so as to obtain a more accurate final positioning result. It is particularly suitable for service areas with complex and changeable real environments, or areas where the AI ​​model used for positioning is not sufficiently trained, and has high practicality.

[0394] At the same time, after the LMF network element sends the AI ​​model to the UE, for scenarios where continuous positioning is required on the UE side, the UE side can directly use the received AI model for positioning. If the environmental parameter combination does not change subsequently, there is no need for frequent interaction with the LMF network element, thereby reducing the positioning time delay.

[0395] Based on the method provided in the above embodiment, the embodiment of the present application also provides a terminal device, which is described in detail below with reference to the accompanying drawings.

[0396] See also Fig.12 , which is a schematic diagram of a terminal device provided in an embodiment of the present application.

[0397] The terminal device 100 may include a processor 110, an external memory interface 120, an internal memory 121, a universal serial bus (USB) interface 130, a charging management module 140, a power management module 141, a battery 142, an antenna 1, an antenna 2, a mobile communication module 150, a wireless communication module 160, a sensor module 180, a button 190, a motor 191, an indicator 192, a camera 193, a display screen 194, etc.

[0398] The sensor module 180 may include a gyro sensor 180A, an air pressure sensor 180B, an acceleration sensor 180C, and the like.

[0399] It is to be understood that the structure illustrated in the embodiment of the present invention does not constitute a specific limitation on the terminal device 100. In other embodiments of the present application, the terminal device 100 may include more or fewer components than shown in the figure, or combine some components, or split some components, or arrange the components differently. The components shown in the figure may be implemented in hardware, software, or a combination of software and hardware.

[0400] The processor 110 may include one or more processing units, for example, the processor 110 may include an application processor (AP), a modem processor, a graphics processor (GPU), an image signal processor (ISP), a controller, a video codec, a digital signal processor (DSP), a baseband processor, and / or a neural-network processing unit (NPU), etc. Among them, different processing units may be independent devices or integrated into one or more processors. For example, the controller may generate an operation control signal according to an instruction opcode and a timing signal to complete the control of fetching and executing instructions.

[0401] The processor 110 may also be provided with a memory for storing instructions and data. In some embodiments, the memory in the processor 110 is a cache memory. The memory may store instructions or data that the processor 110 has just used or cyclically used. If the processor 110 needs to use the instruction or data again, it may be directly called from the memory. This avoids repeated access, reduces the waiting time of the processor 110, and thus improves the efficiency of the system.

[0402] NPU is a neural-network (NN) computing processor. When the terminal device receives the AI ​​model sent by the LMF network element, it can use the NPU to run the AI ​​model to obtain the final positioning result.

[0403] The air pressure sensor 180B is used to measure air pressure. In some embodiments, the terminal device 100 calculates the altitude through the air pressure value measured by the air pressure sensor 180B to assist in determining the scene in which the terminal device is currently located.

[0404] The gyro sensor 180A can be used to determine the motion posture of the terminal device 100. The acceleration sensor 180C can detect the magnitude of the acceleration of the terminal device 100 in various directions (generally three axes). The terminal device can use the detection results output by the gyro sensor 180A and the acceleration sensor 180C to assist in determining the scene in which the terminal device is currently located.

[0405] The wireless communication function of the terminal device 100 can be implemented through the antenna 1, the antenna 2, the mobile communication module 150, the wireless communication module 160, the modem processor and the baseband processor.

[0406] Antenna 1 and antenna 2 are used to transmit and receive electromagnetic wave signals. Each antenna in terminal device 100 can be used to cover a single or multiple communication frequency bands. Different antennas can also be reused to improve the utilization of antennas. For example, antenna 1 can be reused as a diversity antenna for a wireless local area network. In some other embodiments, the antenna can be used in combination with a tuning switch.

[0407] The mobile communication module 150 can provide solutions for wireless communications including 2G / 3G / 4G / 5G, etc., applied to the terminal device 100. The wireless communication module 160 can provide solutions for wireless communications including wireless local area network (WLAN) (such as Wi-Fi network), Bluetooth (BT), global navigation satellite system (GNSS), frequency modulation (FM), near field communication (NFC), infrared technology (IR), etc., applied to the terminal device 100.

[0408] The processor 110 of the terminal device can be used to execute the computer program and / or instructions stored in the memory to implement the AI ​​model-based positioning method described in the above embodiments.

[0409] The embodiment of the present application also provides a communication device, which can be used as a positioning device. The communication device includes but is not limited to network devices such as base stations and core network units, and the communication device is taken as a core network device for example.

[0410] See also Fig.13 , which is a schematic diagram of a communication device provided in an embodiment of the present application.

[0411] The communication device 1100 shown in the figure includes a processor 1110 , a memory 1120 , and a transceiver 1130 .

[0412] The processor 1110 is mainly used for baseband processing, controlling the communication device 1100, etc. The processor 1110 is usually the control center of the communication device 1100, and is used to control the communication device 1100 to execute the positioning method based on the AI ​​model in the above method embodiment.

[0413] The memory 1120 is mainly used to store computer program codes and data.

[0414] For example, for Figure 5 and Figure 7 In the scheme shown, an AI model library can be stored in the memory, and the AI ​​model library can include multiple AI models.

[0415] For example, for Fig. 9 and Fig.11 In the scheme shown, a mapping table and an AI model library may be stored in the memory, and the AI ​​model library may include multiple AI models.

[0416] The transceiver 1130 is mainly used for receiving and transmitting radio frequency signals and converting radio frequency signals into baseband signals. The transceiver 1130 may also be called a transceiver, a transceiver circuit, etc.

[0417] The transceiver module of the transceiver 1130 may include an antenna 1133 and a radio frequency circuit (not shown in the figure), wherein the radio frequency circuit is mainly used for performing radio frequency processing.

[0418] Optionally, the device for implementing the receiving function in the transceiver 1130 may be regarded as the receiver 1032, and the device for implementing the transmitting function may be regarded as the transmitter 1031. The receiver 1032 may also be referred to as a receiving module, a receiver, or a receiving circuit, etc., and the transmitter 1031 may be referred to as a transmitting module, a transmitter, or a transmitting circuit, etc.

[0419] The processor 1110 and the memory 1120 may include one or more boards, and each board may include one or more processors and one or more memories.

[0420] The processor 1110 is used to read and execute the program in the memory 1120 to control the communication device. If there are multiple boards, each board can be interconnected to enhance the processing capability. As an optional implementation, multiple boards can share one or more processors, or multiple boards can share one or more memories, or multiple boards can share one or more processors at the same time.

[0421] For example, in one implementation, the transceiver module of the transceiver 1130 is used to execute the process of sending and receiving related information executed by the LMF network element in the aforementioned method embodiment. The processor 1110 is used to execute the positioning method executed by the LMF network element in the aforementioned method embodiment.

[0422] It should be understood that Fig.13 This is only an example and not a limitation. The communication device including the processor, the memory and the transceiver may not rely on Fig.13 The structure shown.

[0423] The embodiment of the present application also provides a storage medium, and the computer-readable storage medium can be any available medium that can be stored by a computing device or a data storage device such as a data center containing one or more available media. The available medium can be a magnetic medium (for example, a floppy disk, a hard disk, a tape), an optical medium (for example, a DVD), or a semiconductor medium (for example, a solid-state hard disk), etc. The computer-readable storage medium includes instructions that instruct a terminal device or a communication device to perform the above-mentioned positioning method based on an AI model. The embodiment of the present application also provides another computer-readable storage medium. The computer-readable storage medium includes instructions that instruct a terminal device or a communication device to perform the above-mentioned positioning method based on an AI model.

[0424] The embodiment of the present application also provides a computer program product comprising instructions. The computer program product may be a software or program product comprising instructions that can be run on a terminal device or a communication device, or stored in any available medium. When the computer program product runs on a terminal device or a communication device, the terminal device or the communication device executes the above-mentioned positioning method based on the AI ​​model. The embodiment of the present application also provides a computer program product comprising instructions. When the computer program product runs on a terminal device or a communication device, the terminal device or the communication device executes the above-mentioned positioning method based on the AI ​​model.

[0425] In the embodiments of the present application, "at least one" means one or more, and "more than one" means two or more. "At least one of the following" or similar expressions refers to any combination of these items, including any combination of single items or plural items. For example, at least one of a, b, or c can represent: a, b, c, ab, ac, bc, or abc, where a, b, and c can be single or multiple.

[0426] As described above, the above embodiments are only used to illustrate the technical solutions of the present application, rather than to limit them. Although the present application has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or make equivalent replacements for some of the technical features therein. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present application.

Claims

1. A positioning method based on an AI model, characterized in that: The method comprises: Determine one or more matching AI model indexes by using a received positioning reference signal PRS and an environmental parameter combination, wherein the environmental parameter combination includes a state parameter of a terminal device and a state parameter of a network device; Sending first index information, where the first index information includes the one or more AI model indexes; Receiving an AI model corresponding to each AI model index in the one or more AI model indexes; Determine the weight of the positioning result output by each of the AI ​​models using the combination of environmental parameters; The final positioning result is determined by using the PRS, each of the AI ​​models, and the weight corresponding to each of the AI ​​models.

2. The method according to claim 1, characterized in that: The PRS carries one or more of a cell ID, a reference signal received power RSRP, and an angle of arrival AoA relative to a base station.

3. The method according to claim 2, characterized in that The PRS carries a cell ID, and the environmental parameter combination includes a state parameter of a terminal device and a state parameter of a network device; The state parameter of the terminal device includes one or more of a spatial region code, a time state parameter and a scene state parameter, wherein the spatial region code indicates the grid area where the cell corresponding to each Cell ID is located, the time state parameter is used to indicate the positioning time, and the scene state parameter is an identifier of the scene where the terminal device is located; The status parameter of the network device indicates a parameter configuration of the network device.

4. The method according to claim 3, characterized in that The scene state parameters are determined by the terminal device using detection results of multiple sensors, and the sensors include one or more of the following: an air pressure sensor, an accelerometer, and a gyroscope.

5. The method according to claim 3, characterized in that: The PRS carries a cell ID, the state parameters of the terminal device include a spatial region code, a time state parameter, and a scene state parameter, and the method of using the received positioning reference signal PRS and the combination of environmental parameters to determine one or more matching AI model indexes includes: Determine a first Cell ID list, where the first Cell ID list includes all Cell IDs carried in the PRS; Determine the intersection of a second Cell ID list corresponding to each spatial area code and the first Cell ID list, respectively, where the second Cell ID list includes Cell IDs of all cells in the grid area corresponding to the spatial area code; Using the spatial region code corresponding to the intersection of the number of the Cell IDs being greater than or equal to the first threshold as the first screening condition; When the time interval between the positioning time and the reference value corresponding to the time interval is less than the preset interval, the time state parameter corresponding to the time interval is used as the second screening condition, and the reference value corresponding to the time interval is the middle value of the time interval; The scene state parameter is used as the third screening condition; Determining a fourth screening condition according to the status parameter of the network device; According to the first filtering condition, the second filtering condition, the third filtering condition and the fourth filtering condition, and a mapping table of environmental parameter combinations and AI model indexes, determine one or more matching AI model indexes.

6. The method according to claim 3, characterized in that The PRS carries a cell ID and the RSRP, the state parameters of the terminal device include a spatial region code, a time state parameter, and a scene state parameter, and the method of using a combination of a received positioning reference signal PRS and an environmental parameter to determine one or more matching AI model indexes includes: Determine a first Cell ID list, where the first Cell ID list includes all Cell IDs carried in the PRS; Determine the intersection of a second Cell ID list corresponding to each spatial area code and the first Cell ID list, respectively, where the second Cell ID list includes Cell IDs of all cells in the grid area corresponding to the spatial area code; The spatial region code corresponding to the intersection that meets the following conditions is used as the first screening condition: The number of Cell IDs whose RSRP is greater than or equal to the preset power threshold is greater than a first threshold; When the time interval between the positioning time and the reference value corresponding to the time interval is less than the preset interval, the time state parameter corresponding to the time interval is used as the second screening condition, and the reference value corresponding to the time interval is the middle value of the time interval; The scene state parameter is used as the third screening condition; Determining a fourth screening condition according to the status parameter of the network device; According to the first filtering condition, the second filtering condition, the third filtering condition and the fourth filtering condition, and a mapping table of environmental parameter combinations and AI model indexes, determine one or more matching AI model indexes.

7. The method according to claim 3, characterized in that The PRS carries a cell ID, the RSRP, and the AoA of the relative base station, and the receiving positioning reference signal PRS and the environmental parameter combination are used to determine the matching one or more AI model indexes, including: Determine a first Cell ID list, where the first Cell ID list includes all Cell IDs carried in the PRS; Determine the intersection of a second Cell ID list corresponding to each spatial area code and the first Cell ID list, respectively, where the second Cell ID list includes Cell IDs of all cells in the grid area corresponding to the spatial area code; The spatial region code corresponding to the intersection that meets the following conditions is used as the first screening condition: The RSRP is greater than or equal to a preset power threshold, and the number of Cell IDs whose AoA relative to the base station satisfies a preset angle range is greater than a first threshold; When the time interval between the positioning time and the reference value corresponding to the time interval is less than the preset interval, the time state parameter corresponding to the time interval is used as the second screening condition, and the reference value corresponding to the time interval is the middle value of the time interval; The scene state parameter is used as the third screening condition; Determining a fourth screening condition according to the status parameter of the network device; According to the first filtering condition, the second filtering condition, the third filtering condition and the fourth filtering condition, and a mapping table of environmental parameter combinations and AI model indexes, determine one or more matching AI model indexes.

8. The method according to any one of claims 5 to 7, characterized in that: The state parameter of the network device indicates the parameter configuration of the network device, and the weight of the positioning result output by each AI model is determined by combining the environmental parameters, specifically including: Determining a first weight coefficient of each of the AI ​​models according to the number of Cell IDs in the intersection corresponding to each of the spatial region codes serving as the first screening condition, wherein the first weight coefficient is positively correlated with the number of Cell IDs in the intersection; Determine a second weight coefficient of each of the AI ​​models according to the time interval, wherein the second weight coefficient is negatively correlated with the time interval; Determine a third weight coefficient of each of the AI ​​models according to the scene state parameter; Determining a fourth weight coefficient of each of the AI ​​models according to the state parameters of the network device; Determine a weight coefficient of each of the AI ​​models according to each of the first weight coefficients, each of the second weight coefficients, each of the third weight coefficients, and each of the fourth weight coefficients; The weight coefficients of the AI ​​models are normalized to obtain the weights of the AI ​​models, and the sum of the weights of the AI ​​models is 1.

9. The method according to claim 8, characterized in that The state parameter of the network device indicates the parameter configuration of the network device in different time periods, and the determining, according to each of the first weight coefficients, each of the second weight coefficients, each of the third weight coefficients, and each of the fourth weight coefficients, determines the weight coefficient of each of the AI ​​models, including: For each AI model, perform the following steps: The product of the first influence factor and the first weight coefficient, the product of the first influence factor and the third weight coefficient, the product of the second influence factor and the second weight coefficient, and the product of the second influence factor and the fourth weight coefficient are summed to determine the weight coefficient of each of the AI ​​models, wherein the first influence factor is greater than the second influence factor.

10. The method according to claim 8, characterized in that The using the PRS, each of the AI ​​models, and the weights corresponding to each of the AI ​​models to determine a final positioning result includes: Using the PRS as an input of each of the AI ​​models to obtain a positioning result output by each of the AI ​​models; The positioning result output by each of the AI ​​models is multiplied by the weight corresponding to each of the AI ​​models to obtain the final positioning result component corresponding to each of the AI ​​models; The final positioning result is obtained by superimposing the final positioning result components corresponding to each of the AI ​​models.

11. A positioning method based on an AI model, characterized in that: The method comprises: Sending an environmental parameter combination and a positioning reference signal PRS, wherein the environmental parameter combination includes a state parameter of the terminal device and a state parameter of the network device; Receive one or more AI models, each of the one or more AI models corresponds to an AI model index, and each of the AI ​​model indexes is determined by the combination of the environmental parameters and the PRS; Determine the weight of the positioning result output by each of the AI ​​models using the combination of environmental parameters; The final positioning result is determined by using the PRS, each of the AI ​​models, and the weight corresponding to each of the AI ​​models.

12. The method according to claim 11, characterized in that The PRS carries one or more of a cell ID, a reference signal received power RSRP, and an angle of arrival AoA relative to a base station.

13. The method according to claim 12, characterized in that The PRS carries a cell ID, and the environmental parameter combination includes a state parameter of a terminal device and a state parameter of a network device; The state parameter of the terminal device includes one or more of a spatial region code, a time state parameter and a scene state parameter, wherein the spatial region code indicates the grid area where the cell corresponding to each Cell ID is located, the time state parameter is used to indicate the positioning time, and the scene state parameter is an identifier of the scene where the terminal device is located; The status parameter of the network device indicates a parameter configuration of the network device.

14. The method according to claim 13, characterized in that The PRS carries a cell ID, the state parameters of the terminal device include a spatial region code, a time state parameter, and a scene state parameter, and the weight of the positioning result output by each AI model is determined by combining the environmental parameters, specifically including: Determine a first Cell ID list, where the first Cell ID list includes all Cell IDs carried in the PRS; Determine the intersection of a second Cell ID list corresponding to each spatial area code and the first Cell ID list, respectively, where the second Cell ID list includes Cell IDs of all cells in the grid area corresponding to the spatial area code; Using the spatial region code corresponding to the intersection of the number of the Cell IDs being greater than or equal to the first threshold as the first screening condition; Determining a first weight coefficient of each of the AI ​​models according to the number of Cell IDs in the intersection corresponding to each of the spatial region codes serving as the first screening condition, wherein the first weight coefficient is positively correlated with the number of Cell IDs in the intersection; Determine a second weight coefficient of each of the AI ​​models according to a time interval between the positioning time and a reference value corresponding to the time interval, wherein the second weight coefficient is negatively correlated with the time interval; Determine a third weight coefficient of each of the AI ​​models according to the scene state parameter; Determining a fourth weight coefficient of each of the AI ​​models according to the state parameters of the network device; Determine a weight coefficient of each of the AI ​​models according to each of the first weight coefficients, each of the second weight coefficients, each of the third weight coefficients, and each of the fourth weight coefficients; The weight coefficients of the AI ​​models are normalized to obtain the weights of the AI ​​models, and the sum of the weights of the AI ​​models is 1.

15. The method according to claim 13, characterized in that The PRS carries a cell ID and the RSRP, the state parameters of the terminal device include a spatial region code, a time state parameter and a scene state parameter, and the weight of the positioning result output by each AI model is determined by combining the environmental parameters, specifically including: Determine a first Cell ID list, where the first Cell ID list includes all Cell IDs carried in the PRS; Determine the intersection of a second Cell ID list corresponding to each spatial area code and the first Cell ID list, respectively, where the second Cell ID list includes Cell IDs of all cells in the grid area corresponding to the spatial area code; The spatial region code corresponding to the intersection that meets the following conditions is used as the first screening condition: The number of Cell IDs whose RSRP is greater than or equal to the preset power threshold is greater than a first threshold; Determining a first weight coefficient of each of the AI ​​models according to the number of Cell IDs in the intersection corresponding to each of the spatial region codes serving as the first screening condition, wherein the first weight coefficient is positively correlated with the number of Cell IDs in the intersection; Determine a second weight coefficient of each of the AI ​​models according to a time interval between the positioning time and a reference value corresponding to the time interval, wherein the second weight coefficient is negatively correlated with the time interval; Determine a third weight coefficient of each of the AI ​​models according to the scene state parameter; Determining a fourth weight coefficient of each of the AI ​​models according to the state parameters of the network device; Determine a weight coefficient of each of the AI ​​models according to each of the first weight coefficients, each of the second weight coefficients, each of the third weight coefficients, and each of the fourth weight coefficients; The weight coefficients of the AI ​​models are normalized to obtain the weights of the AI ​​models, and the sum of the weights of the AI ​​models is 1.

16. The method according to claim 13, characterized in that The PRS carries a cell ID, the state parameters of the terminal device include a spatial region code, a time state parameter, and a scene state parameter, and the weight of the positioning result output by each AI model is determined by combining the environmental parameters, specifically including: Determine a first Cell ID list, where the first Cell ID list includes all Cell IDs carried in the PRS; Determine the intersection of a second Cell ID list corresponding to each spatial area code and the first Cell ID list, respectively, where the second Cell ID list includes Cell IDs of all cells in the grid area corresponding to the spatial area code; The spatial region code corresponding to the intersection that meets the following conditions is used as the first screening condition: The RSRP is greater than or equal to a preset power threshold, and the number of Cell IDs whose AoA relative to the base station satisfies a preset angle range is greater than a first threshold; The number of Cell IDs whose RSRP is greater than or equal to the preset power threshold is greater than a first threshold; Determining a first weight coefficient of each of the AI ​​models according to the number of Cell IDs in the intersection corresponding to each of the spatial region codes serving as the first screening condition, wherein the first weight coefficient is positively correlated with the number of Cell IDs in the intersection; Determine a second weight coefficient of each of the AI ​​models according to a time interval between the positioning time and a reference value corresponding to the time interval, wherein the second weight coefficient is negatively correlated with the time interval; Determine a third weight coefficient of each of the AI ​​models according to the scene state parameter; Determining a fourth weight coefficient of each of the AI ​​models according to the state parameters of the network device; Determine a weight coefficient of each of the AI ​​models according to each of the first weight coefficients, each of the second weight coefficients, each of the third weight coefficients, and each of the fourth weight coefficients; The weight coefficients of the AI ​​models are normalized to obtain the weights of the AI ​​models, and the sum of the weights of the AI ​​models is 1.

17. The method according to any one of claims 14 to 16, characterized in that: The state parameter of the network device indicates the parameter configuration of the network device in different time periods, and the determining, according to each of the first weight coefficients, each of the second weight coefficients, each of the third weight coefficients, and each of the fourth weight coefficients, determines the weight coefficient of each of the AI ​​models, including: For each AI model, perform the following steps: The product of the first influence factor and the first weight coefficient, the product of the first influence factor and the third weight coefficient, the product of the second influence factor and the second weight coefficient, and the product of the second influence factor and the fourth weight coefficient are summed to determine the weight coefficient of each of the AI ​​models, wherein the first influence factor is greater than the second influence factor.

18. The method according to any one of claims 14 to 16, characterized in that: The using the PRS, each of the AI ​​models, and the weights corresponding to each of the AI ​​models to determine a final positioning result includes: Using the PRS as an input of each of the AI ​​models to obtain a positioning result output by each of the AI ​​models; The positioning result output by each of the AI ​​models is multiplied by the weight corresponding to each of the AI ​​models to obtain the final positioning result component corresponding to each of the AI ​​models; The final positioning result is obtained by superimposing the final positioning result components corresponding to each of the AI ​​models.

19. A positioning method based on an AI model, characterized in that: The method comprises: Determine one or more matching AI model indexes by using a received positioning reference signal PRS and an environmental parameter combination, wherein the environmental parameter combination includes a state parameter of a terminal device and a state parameter of a network device; Sending first index information, the PRS, and the environmental parameter combination, wherein the first index information includes the one or more AI model indexes; Receive location information, the location information including a final positioning result, the final positioning result being determined by the PRS, one or more AI models, and a weight corresponding to each of the one or more AI models, each of the AI ​​models corresponding to an AI model index in the one or more AI model indexes, and the weight being a weight of a positioning result output by each of the AI ​​models determined using the combination of environmental parameters.

20. A positioning method based on an AI model, characterized in that: The method comprises: Sending an environmental parameter combination and a positioning reference signal PRS, wherein the environmental parameter combination includes a state parameter of the terminal device and a state parameter of the network device; Receive location information, the location information including a final positioning result, the final positioning result is determined by the PRS, one or more AI models, and a weight corresponding to each of the one or more AI models, the weight being the weight of the positioning result output by each AI model determined using the environmental parameter combination, each AI model corresponds to an AI model index, and each AI model index is determined using the environmental parameter combination and the PRS.

21. A positioning method based on an AI model, characterized in that: The method comprises: Receive first index information, where the first index information includes one or more AI model indexes, where the one or more AI model indexes are determined by receiving a positioning reference signal PRS and an environmental parameter combination, where the environmental parameter combination includes a state parameter of a terminal device and a state parameter of a network device; Send an AI model corresponding to each AI model index in the one or more AI model indexes, each AI model is used to determine the final positioning result together with the PRS and the weight corresponding to each AI model, and the weight corresponding to each AI model is the weight of the positioning result output by each AI model determined by the combination of environmental parameters.

22. A positioning method based on an AI model, characterized in that: The method comprises: Receiving an environmental parameter combination and a positioning reference signal PRS, wherein the environmental parameter combination includes a state parameter of a terminal device and a state parameter of a network device; Determine one or more matching AI model indexes using the combination of environmental parameters and the PRS; Send an AI model corresponding to each AI model index in the one or more AI model indexes, each AI model is used to determine the final positioning result together with the PRS and the weight corresponding to each AI model, and the weight corresponding to each AI model is the weight of the positioning result output by each AI model determined by the combination of environmental parameters.

23. The method according to claim 22, characterized in that The PRS carries one or more of a cell ID, a reference signal received power RSRP, and an angle of arrival AoA relative to a base station.

24. The method according to claim 23, characterized in that The PRS carries a cell ID, and the environmental parameter combination includes a state parameter of a terminal device and a state parameter of a network device; The state parameter of the terminal device includes one or more of a spatial region code, a time state parameter and a scene state parameter, wherein the spatial region code indicates the grid area where the cell corresponding to each Cell ID is located, the time state parameter is used to indicate the positioning time, and the scene state parameter is an identifier of the scene where the terminal device is located; The status parameter of the network device indicates a parameter configuration of the network device.

25. The method according to claim 23, characterized in that The PRS carries a cell ID, the state parameters of the terminal device include a spatial region code, a time state parameter, and a scene state parameter, and the use of the environmental parameter combination and the PRS to determine one or more matching AI model indexes specifically includes: Determine a first Cell ID list, where the first Cell ID list includes all Cell IDs carried in the PRS; Determine the intersection of a second Cell ID list corresponding to each spatial area code and the first Cell ID list, respectively, where the second Cell ID list includes Cell IDs of all cells in the grid area corresponding to the spatial area code; Using the spatial region code corresponding to the intersection of the number of the Cell IDs being greater than or equal to the first threshold as the first screening condition; When the time interval between the positioning time and the reference value corresponding to the time interval is less than the preset interval, the time state parameter corresponding to the time interval is used as the second screening condition, and the reference value corresponding to the time interval is the middle value of the time interval; The scene state parameter is used as the third screening condition; Determining a fourth screening condition according to the status parameter of the network device; According to the first filtering condition, the second filtering condition, the third filtering condition and the fourth filtering condition, and a mapping table of environmental parameter combinations and AI model indexes, determine one or more matching AI model indexes.

26. The method according to claim 23, characterized in that The PRS carries a cell ID and the RSRP, the state parameters of the terminal device include a spatial region code, a time state parameter, and a scene state parameter, and the use of the environmental parameter combination and the PRS to determine one or more matching AI model indexes specifically includes: Determine a first Cell ID list, where the first Cell ID list includes all Cell IDs carried in the PRS; Determine the intersection of a second Cell ID list corresponding to each spatial area code and the first Cell ID list, respectively, where the second Cell ID list includes Cell IDs of all cells in the grid area corresponding to the spatial area code; The spatial region code corresponding to the intersection that meets the following conditions is used as the first screening condition: The number of Cell IDs whose RSRP is greater than or equal to the preset power threshold is greater than a first threshold; When the time interval between the positioning time and the reference value corresponding to the time interval is less than the preset interval, the time state parameter corresponding to the time interval is used as the second screening condition, and the reference value corresponding to the time interval is the middle value of the time interval; The scene state parameter is used as the third screening condition; Determining a fourth screening condition according to the status parameter of the network device; According to the first filtering condition, the second filtering condition, the third filtering condition and the fourth filtering condition, and a mapping table of environmental parameter combinations and AI model indexes, determine one or more matching AI model indexes.

27. The method according to claim 23, characterized in that The PRS carries a cell ID, the RSRP, and the AoA of the relative base station, and the use of the environmental parameter combination and the PRS to determine one or more matching AI model indexes specifically includes: Determine a first Cell ID list, where the first Cell ID list includes all Cell IDs carried in the PRS; Determine the intersection of a second Cell ID list corresponding to each spatial area code and the first Cell ID list, respectively, where the second Cell ID list includes Cell IDs of all cells in the grid area corresponding to the spatial area code; The spatial region code corresponding to the intersection that meets the following conditions is used as the first screening condition: The RSRP is greater than or equal to a preset power threshold, and the number of Cell IDs whose AoA relative to the base station satisfies a preset angle range is greater than a first threshold; When the time interval between the positioning time and the reference value corresponding to the time interval is less than the preset interval, the time state parameter corresponding to the time interval is used as the second screening condition, and the reference value corresponding to the time interval is the middle value of the time interval; The scene state parameter is used as the third screening condition; Determining a fourth screening condition according to the status parameter of the network device; According to the first filtering condition, the second filtering condition, the third filtering condition and the fourth filtering condition, and a mapping table of environmental parameter combinations and AI model indexes, determine one or more matching AI model indexes.

28. A positioning method based on an AI model, characterized in that: The method comprises: Receive first index information, a positioning reference signal PRS, and an environmental parameter combination, where the first index information includes one or more AI model indexes, the one or more AI model indexes are determined using the PRS and the environmental parameter combination, and the environmental parameter combination includes a state parameter of a terminal device and a state parameter of a network device; Determine an AI model corresponding to each AI model index in the one or more AI model indexes; Determine the weight of the positioning result output by each of the AI ​​models using the combination of environmental parameters; Determine a final positioning result by using the PRS, each of the AI ​​models, and a weight corresponding to each of the AI ​​models; Sending location information, where the location information includes the final positioning result.

29. A positioning method based on an AI model, characterized in that: The method comprises: Receiving an environmental parameter combination and a positioning reference signal PRS, wherein the environmental parameter combination includes a state parameter of a terminal device and a state parameter of a network device; Determine one or more matching AI model indexes using the combination of environmental parameters and the PRS; Determine an AI model corresponding to each AI model index in the one or more AI model indexes; Determine the weight of the positioning result output by each of the AI ​​models using the combination of environmental parameters; Determine a final positioning result by using the PRS, each of the AI ​​models, and a weight corresponding to each of the AI ​​models; Sending location information, where the location information includes the final positioning result.

30. A terminal device, characterized in that: including a processor and a memory; The processor is coupled to the memory; The memory is used to store computer programs and / or instructions; The processor is used to execute the computer program and / or instructions stored in the memory to implement the AI ​​model-based positioning method as described in any one of claims 1 to 20.

31. A communication device, characterized in that: It includes a processor, which is used to execute a computer program or instruction stored in a memory so that the device performs the AI ​​model-based positioning method described in any one of claims 21 to 29.

32. A computer program product, characterized in that The computer program product includes a computer program or instruction for executing the AI ​​model-based positioning method as described in any one of claims 1 to 20, or includes a computer program or instruction for executing the AI ​​model-based positioning method as described in any one of claims 21 to 29.

33. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores a computer program or instruction, and when the computer program or instruction is executed, the AI ​​model-based positioning method as described in any one of claims 1 to 29 is executed.

Images