Positioning method, neural network training method, device, equipment and medium

Abstract

The application provides a positioning method, a neural network training method, a device, equipment and a medium. The method comprises the following steps: acquiring a first distance difference, a first angle and first information; wherein the first distance difference is used to represent the measured distance difference between a terminal device and two network devices, the first angle is used to represent the measured angle of the terminal device relative to the two network devices, and the first information is used to represent the channel state between the terminal device and the two network devices; and the terminal device is positioned based on the first distance difference, the first angle and the first information.

Description

Localization methods, neural network training methods, devices, equipment and media Technical Field

[0001] This application relates to the field of artificial intelligence technology, specifically to a positioning method, a neural network training method, a device, equipment, and a medium. Background Technology

[0002] Positioning is a technology that measures the location of terminal devices using systems such as satellites and cellular data networks. It can be applied to scenarios such as mobile phone positioning, maritime navigation, car navigation, logistics tracking, environmental monitoring, vehicle networking, and virtual reality. However, in positioning scenarios (such as non-line-of-sight (NLoS) propagation positioning scenarios), how to accurately locate terminal devices is a problem that urgently needs to be solved.

[0003] Summary of the Invention

[0004] This application provides a localization method, a neural network training method, an apparatus, a device, and a medium.

[0005] Firstly, the positioning method provided in this application includes:

[0006] Acquire a first distance difference, a first angle, and first information; wherein, the first distance difference is used to characterize the measured distance difference between the terminal device and the two network devices, the first angle is used to characterize the measured angle of the terminal device relative to the two network devices, and the first information is used to characterize the channel state between the terminal device and the two network devices;

[0007] The terminal device is located based on the first distance difference, the first angle, and the first information.

[0008] Secondly, the neural network training method provided in this application includes:

[0009] Acquire the third distance difference, the third angle, and the second information; wherein, the third distance difference is used to characterize the measured distance difference between the terminal device and the two network devices, the third angle is used to characterize the measured angle of the terminal device relative to the two network devices, and the second information is used to characterize the channel state between the terminal device and the two network devices.

[0010] The third distance difference, the third angle, and the second information are input into the neural network to be trained to obtain the fourth distance difference and the fourth angle.

[0011] The first loss is determined based on the fifth distance difference and the fourth distance difference; and the second loss is determined based on the fifth angle and the fourth angle; wherein the fifth distance difference is used to characterize the real distance difference between the terminal device and the two network devices, and the fifth angle is used to characterize the real angle of the terminal device relative to the two network devices.

[0012] Based on the first loss and the second loss, the network parameters of the neural network to be trained are adjusted so that the loss obtained based on the adjusted neural network satisfies the convergence condition to obtain the target neural network; the target neural network is used to locate the terminal device.

[0013] Thirdly, the positioning device provided in this application includes:

[0014] The first acquisition unit is configured to acquire a first distance difference, a first angle, and first information; wherein the first distance difference is used to characterize the measured distance difference between the terminal device and the two network devices, the first angle is used to characterize the measured angle of the terminal device relative to the two network devices, and the first information is used to characterize the channel state between the terminal device and the two network devices.

[0015] The positioning unit is configured to locate the terminal device based on a first distance difference, a first angle, and first information.

[0016] Fourthly, the network-connected training apparatus provided in this application includes:

[0017] The second acquisition unit is configured to acquire a third distance difference, a third angle, and second information; wherein the third distance difference is used to characterize the measured distance difference between the terminal device and the two network devices, the third angle is used to characterize the measured angle of the terminal device relative to the two network devices, and the second information is used to characterize the channel state between the terminal device and the two network devices.

[0018] The second processing unit is configured to input the third distance difference, the third angle, and the second information into the neural network to be trained to obtain the fourth distance difference and the fourth angle.

[0019] The second processing unit is further configured to determine a first loss based on a fifth distance difference and a fourth distance difference; and to determine a second loss based on a fifth angle and a fourth angle; wherein the fifth distance difference is used to characterize the actual distance difference between the terminal device and the two network devices, and the fifth angle is used to characterize the actual angle of the terminal device relative to the two network devices.

[0020] The second processing unit is further configured to adjust the network parameters of the neural network to be trained based on the first loss and the second loss, so that the loss obtained based on the adjusted neural network satisfies the convergence condition to obtain the target neural network; the target neural network is used to locate the terminal device.

[0021] Fifthly, the computer device provided in this application includes a memory, a processor, and a transceiver. The memory is used to store computer programs; the processor is connected to the memory and is used to retrieve and run the computer programs from the memory to implement the methods of the first or second aspect described above; the transceiver is used for receiving and sending information during the process of sending and receiving information with other external devices.

[0022] Sixthly, the chip provided in this application includes a memory, a processor, and a transceiver. The memory is used to store a computer program; the processor is connected to the memory and is used to retrieve and run the computer program from the memory, causing a device with the chip installed to perform the methods described in the first or second aspect; the transceiver is used for receiving and sending information during the exchange of information with the device or the chip.

[0023] Seventhly, the present application provides a computer-readable storage medium storing a computer program that, when executed by at least one processor, implements the methods of the first or second aspect described above.

[0024] Eighthly, the computer program product provided in this application includes a computer program or instructions that, when executed by a processor, implement the methods of the first or second aspect described above.

[0025] Ninthly, the computer program provided in this application causes a computer to perform the methods described in the first or second aspect above.

[0026] This application provides a positioning method that acquires a first distance difference, a first angle, and first information. The first distance difference characterizes the measured distance difference between a terminal device and two network devices, the first angle characterizes the measured angle of the terminal device relative to the two network devices, and the first information characterizes the channel state between the terminal device and the two network devices. Based on the first distance difference, the first angle, and the first information, the terminal device is positioned. Thus, by introducing the first information into the positioning process of the terminal device and combining the first distance difference, the first angle, and the first information, the positioning accuracy of the terminal device can be improved. Attached Figure Description

[0027] The accompanying drawings, which are included to provide a further understanding of this application and form part of this application, illustrate exemplary embodiments and are used to explain this application, but do not constitute an undue limitation of this application. In the drawings:

[0028] Figure 1 is a schematic diagram of a communication architecture;

[0029] Figure 2 is a flowchart illustrating a positioning method provided in an embodiment of this application;

[0030] Figure 3 is a flowchart illustrating a neural network training method provided in an embodiment of this application;

[0031] Figure 4 is a schematic diagram showing the location of a terminal device and a network device according to an embodiment of this application;

[0032] Figure 5 is a schematic diagram of a hierarchical latent variable model provided in an embodiment of this application;

[0033] Figure 6 is a schematic diagram of a neural network structure provided in an embodiment of this application;

[0034] Figure 7 is a schematic diagram of the structural composition of a positioning device 700 provided in an embodiment of this application;

[0035] Figure 8 is a schematic diagram of the structural composition of a neural network training device 800 provided in an embodiment of this application;

[0036] Figure 9 is a schematic diagram of the structural composition of a computer device provided in an embodiment of this application;

[0037] Figure 10 is a schematic structural diagram of a chip provided in an embodiment of this application. Detailed Implementation

[0038] The technical solutions of the embodiments of this application will now be described with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of this application, and not all of them. All other embodiments obtained by those skilled in the art based on the embodiments of this application without creative effort are within the scope of protection of this application.

[0039] Figure 1 is a schematic diagram of an application scenario of an embodiment of this application.

[0040] As shown in Figure 1, the communication system 100 may include a terminal device 110 and a network device 120. The network device 120 can communicate with the terminal device 110 via an air interface. Multi-service transmission is supported between the terminal device 110 and the network device 120.

[0041] It should be understood that the embodiments of this application are only illustrated by way of example with communication system 100, but the embodiments of this application are not limited thereto. That is to say, the technical solutions of the embodiments of this application can be applied to various communication systems, such as: Long Term Evolution (LTE) system, LTE Time Division Duplex (TDD), Universal Mobile Telecommunication System (UMTS), Internet of Things (IoT) system, Narrow Band Internet of Things (NB-IoT) system, enhanced Machine-Type Communications (eMTC) system, 5G communication system (also known as New Radio (NR) communication system), or future communication systems, etc.

[0042] In the communication system 100 shown in Figure 1, the network device 120 can be an access network device that communicates with the terminal device 110. The access network device can provide communication coverage for a specific geographical area and can communicate with the terminal device 110 located within that coverage area.

[0043] Network device 120 may be an evolved Node B (eNB or eNodeB) in an LTE system, or a Next Generation Radio Access Network (NG RAN) device, or a base station (gNB) in an NR system, or a radio controller in a Cloud Radio Access Network (CRAN), or network device 120 may be a relay station, access point, vehicle-mounted equipment, wearable device, hub, switch, bridge, router, or network equipment in a future evolved Public Land Mobile Network (PLMN), etc.

[0044] Terminal device 110 can be any terminal device, including but not limited to terminal devices that are connected to network device 120 or other terminal devices via wired or wireless connections.

[0045] For example, terminal equipment 110 can refer to an access terminal, user equipment (UE), user unit, user station, mobile station, mobile station, remote station, remote terminal, mobile device, user terminal, terminal, wireless communication equipment, user agent, or user device. An access terminal can be a cellular phone, cordless phone, Session Initiation Protocol (SIP) phone, IoT device, satellite handheld terminal, Wireless Local Loop (WLL) station, Personal Digital Assistant (PDA), handheld device with wireless communication capabilities, computing device or other processing device connected to a wireless modem, in-vehicle device, wearable device, terminal equipment in a 5G network, or terminal equipment in a future evolved network, etc.

[0046] Terminal device 110 can be used for device-to-device (D2D) communication.

[0047] Figure 1 illustrates an exemplary network device and two terminal devices. Optionally, the communication system 100 may include multiple network devices, and each network device may include other numbers of terminal devices within its coverage area. This application embodiment does not limit this.

[0048] It should be noted that Figure 1 is only an example illustrating the system to which this application applies. Of course, the method shown in the embodiments of this application can also be applied to other systems.

[0049] In the following description, references are made to “some embodiments,” which describe a subset of all possible embodiments. However, it is understood that “some embodiments” may be the same subset or different subsets of all possible embodiments and may be combined with each other without conflict.

[0050] It should be understood that the terms "first, second, third" used in the embodiments of this application are only used to distinguish similar objects and do not represent a specific order of objects. It is understood that "first, second, third" can be interchanged in a specific order or sequence where permitted, so that the embodiments of this application described herein can be implemented in an order other than that illustrated or described herein.

[0051] It should also be understood that the term "and / or" in this article is merely a description of the relationship between related objects, indicating that three relationships can exist. For example, A and / or B can represent: A existing alone, A and B existing simultaneously, or B existing alone. Additionally, the character " / " in this article generally indicates that the preceding and following related objects have an "or" relationship.

[0052] It should also be understood that the "predefined" or "predefined rules" mentioned in the embodiments of this application can be implemented by pre-storing corresponding codes, tables, or other means that can be used to indicate relevant information in the device (e.g., including terminal devices and network devices), and this application does not limit the specific implementation method. For example, predefined can refer to those defined in a protocol. It should also be understood that in the embodiments of this application, the "protocol" can refer to standard protocols in the field of communication, such as LTE protocol, NR protocol, and related protocols applied to future communication systems, and this application does not limit it.

[0053] To facilitate understanding of the technical solutions of the embodiments of this application, the relevant technologies of the embodiments of this application are described below. The following relevant technologies are optional solutions and can be combined with the technical solutions of the embodiments of this application in any way, and they all fall within the protection scope of the embodiments of this application.

[0054] The technical terms used in the embodiments of this application are briefly introduced below.

[0055] Non-line-of-sight (NFS): This term describes the signal propagation environment between a terminal device and a network device. When the signal propagation environment between the terminal device and the network device is non-line-of-sight, it indicates that there is an obstruction to the signal propagation between them.

[0056] Distance difference, also known as time difference of arrival (TDOA), is used to characterize the time difference between when a signal transmitted by the terminal device arrives at the two network devices, taking one terminal device and two network devices as an example. Distance difference can be divided into measured distance difference and true distance difference. Generally, there is an error between the measured distance difference and the true distance difference. The smaller the error, the higher the accuracy of the measured distance difference; the larger the error, the lower the accuracy of the measured distance difference.

[0057] Angle: Also known as azimuth (AoA). Taking a terminal device and a network device as an example, the angle can be used to characterize the angle between the terminal device and the network device; or, the angle can be used to characterize the angle between the network device and the terminal device. Angles can be divided into measured angles and true angles. Generally, there is an error between the measured angle and the true angle. The smaller the error, the higher the accuracy of the measured angle; the larger the error, the lower the accuracy of the measured angle.

[0058] Positioning is a technology that measures the location of terminal devices through systems such as satellites and cellular data networks. It can be applied to scenarios such as mobile phone positioning, maritime navigation, car navigation, logistics tracking, environmental monitoring, vehicle networking, and virtual reality.

[0059] In related technologies, there are various methods for locating terminal devices. For example, in 5G machine learning, terminal devices can be located by directly collecting data (such as fingerprint positioning); another example is by measuring the distance difference between the terminal device and two network devices; yet another example is by measuring the angle between the terminal device and two network devices; still another example is by using both the distance difference between the terminal device and two network devices and the angle between the terminal device and the two network devices to locate the terminal device.

[0060] However, in some positioning scenarios, the accuracy of positioning terminal devices is not high. Taking non-line-of-sight (NLS) propagation positioning scenarios as an example, the impact of NLS propagation on measurement distance difference and measurement angle may not be considered. For instance, in NLS propagation positioning scenarios, on the one hand, there may be no angle of arrival, thus reducing the accuracy of the measurement angle; on the other hand, the deviation between the measured distance difference and the actual distance difference may be large. Based on these two aspects, if network devices directly use the measured distance difference and / or measurement angle to locate terminal devices, lacking other supporting information, the positioning results will be inaccurate. Therefore, how to accurately locate terminal devices is an urgent problem to be solved.

[0061] Based on this, embodiments of this application provide a positioning method that obtains a first distance difference, a first angle, and first information. The first distance difference characterizes the measured distance difference between the terminal device and two network devices, the first angle characterizes the measured angle of the terminal device relative to the two network devices, and the first information characterizes the channel state between the terminal device and the two network devices. The terminal device is positioned based on the first distance difference, the first angle, and the first information. Thus, by introducing the first information into the positioning process of the terminal device and combining the first distance difference, the first angle, and the first information, the positioning accuracy of the terminal device can be improved.

[0062] To facilitate understanding of the technical solutions of the embodiments of this application, the technical solutions of this application are described in detail below through specific embodiments. The above-mentioned related technologies are optional solutions and can be arbitrarily combined with the technical solutions of the embodiments of this application, all of which fall within the protection scope of the embodiments of this application. The embodiments of this application include at least some of the following contents.

[0063] Figure 2 is a flowchart illustrating a positioning method provided in an embodiment of this application. As shown in Figure 2, the method may include the following steps.

[0064] S210. Obtain a first distance difference, a first angle, and first information; wherein, the first distance difference is used to characterize the measured distance difference between the terminal device and the two network devices, the first angle is used to characterize the measured angle of the terminal device relative to the two network devices, and the first information is used to characterize the channel state between the terminal device and the two network devices.

[0065] S220. Based on the first distance difference, the first angle, and the first information, the terminal device is located.

[0066] It should be noted that, in the embodiments of this application, the positioning method can be applied to a positioning device or a computer device integrated with such a device. The computer device can be either one of the two network devices, or other devices; this embodiment of the application does not limit the specific type of device.

[0067] In some embodiments, obtaining the first distance difference and the first angle may include: acquiring the first distance difference and the first angle by building a 5G signal acquisition platform and using an array antenna.

[0068] In some embodiments, obtaining the first information may include: building a 5G signal acquisition platform and using an array antenna to receive a sounding reference signal (SRS) sent by a terminal device; and obtaining the first information based on the SRS.

[0069] It should be noted that one or more of the first distance difference, the first angle, and the first information can be obtained directly by the computer device or obtained by the computer device through other computer devices. This application embodiment does not limit this.

[0070] For example, other computer devices can directly obtain the first distance difference and the first angle, and send the first distance difference and the first angle to the computer device.

[0071] For example, other computer devices can receive the SRS sent by the terminal device, obtain the first information based on the SRS, and then send the first information to the computer device.

[0072] For example, other computer devices can receive the SRS sent by the terminal device and send the SRS to the computer device; the computer device can obtain the first information based on the SRS.

[0073] In some embodiments, the first information may include one or more of the following:

[0074] The received signal energy of Channel State Information (CSI) signals;

[0075] The maximum amplitude of the CSI signal;

[0076] The rise time of the CSI signal;

[0077] Average delay of CSI signal;

[0078] Root mean square spread of CSI signals; and

[0079] Kubularity of CSI signal.

[0080] For example, the received signal energy of a CSI signal can be denoted as ε. r , ε r The formula for calculating ε is shown below: r =∫ T |r(t)| 2 dt (1)

[0081] Where t represents a subcarrier; T represents the total number of subcarriers.

[0082] For example, the maximum amplitude of the CSI signal can be denoted as r. max r max The calculation formula is as follows:

[0083] Where t represents a subcarrier.

[0084] For example, the rise time of the CSI signal can be denoted as t. rise , t rise The formula for calculating t is shown below: rise =min{t:|r(t)|≥βr max}-min{t:|r(t)|≥ασ n} (3)

[0085] Where t represents the subcarrier; βr max For the first threshold, ασ n The second threshold is used to constrain t. rise So that t rise Located within a certain range.

[0086] It should be understood that the first threshold and the second threshold may be specified by the protocol or set in other ways, and the embodiments of this application do not limit this.

[0087] For example, the average delay of the CSI signal can also be called the average excess delay of the CSI signal, and the average delay of the CSI signal can be denoted as T. MED T MED The calculation formula is as follows:

[0088] Where t represents a subcarrier; T represents the total number of subcarriers; ε r This refers to the received signal energy of the CSI signal.

[0089] For example, the root mean square spread of the CSI signal can be denoted as T RMS T RMS The calculation formula is as follows:

[0090] Where t represents a subcarrier; T represents the total number of subcarriers; ε r T represents the received signal energy of the CSI signal. MED The average time delay of the CSI signal.

[0091] For example, the kurtosis of a CSI signal can be denoted as K, and the formula for calculating K is as follows:

[0092] Where t represents a subcarrier; T represents the total number of subcarriers; μ |r| The third threshold is used to constrain K so that K is within a certain range.

[0093] It should be understood that the third threshold may be specified by the protocol or set in other ways, and the embodiments of this application do not limit it.

[0094] This application provides a positioning method that acquires a first distance difference, a first angle, and first information. The first distance difference characterizes the measured distance difference between a terminal device and two network devices, the first angle characterizes the measured angle of the terminal device relative to the two network devices, and the first information characterizes the channel state between the terminal device and the two network devices. Based on the first distance difference, the first angle, and the first information, the terminal device is positioned. Thus, by introducing the first information into the positioning process of the terminal device and combining the first distance difference, the first angle, and the first information, the positioning accuracy of the terminal device can be improved.

[0095] In some embodiments, locating a terminal device based on a first distance difference, a first angle, and first information may include: correcting the first distance difference based on the first information to obtain a second distance difference; correcting the first angle based on the first information to obtain a second angle; and locating the terminal device based on the second distance difference and the second angle.

[0096] It should be noted that by using the second distance difference to locate the terminal device, the location range of the terminal device can be obtained; by using the second angle to locate the terminal device, the location range of the terminal device can be further narrowed.

[0097] Compared to methods that locate terminal devices based on a first distance difference and a first angle, this method, by correcting the first distance difference and the first angle using first information, and then locating the terminal device based on the obtained second distance difference and second angle, can achieve more accurate positioning.

[0098] There are two ways to correct the first distance difference and the first angle based on the first information:

[0099] Method #A can determine a first mapping relationship between the distance difference before correction and the distance difference after correction, and a second mapping relationship between the angle before correction and the angle after correction, based on the first information; the second distance difference can be obtained based on the first mapping relationship and the first distance difference; the second angle can be obtained based on the second mapping relationship and the first angle.

[0100] This method allows for the determination of a first mapping relationship and a second mapping relationship based on the first information. This, in turn, enables the acquisition of a second distance difference and a second angle. This simplifies the acquisition of the second distance difference and the second angle; subsequent positioning of the terminal device based on the obtained second distance difference and second angle allows for more accurate positioning.

[0101] Method #B involves inputting the first distance difference, the first angle, and the first information into the target neural network. In the target neural network, the first distance difference and the first angle are corrected using the first information to obtain the second distance difference and the second angle.

[0102] It should be noted that the target neural network can be a pre-trained neural network.

[0103] In related technologies, the calibration of distance difference and angle is based on different theories, frameworks and algorithms, without taking into account the correlation between distance difference and angle, and lacks an integrated joint design.

[0104] Based on this, in some embodiments, the first distance difference, the first angle, and the first information are input into the target neural network to obtain the second distance difference and the second angle.

[0105] This method achieves two objectives: firstly, considering the correlation between the first distance difference and the first angle, the target neural network can be used to jointly calibrate the first distance difference and the first angle to obtain the second distance difference and the second angle; secondly, the subsequent positioning of the terminal device based on the obtained second distance difference and the second angle can make the positioning more accurate.

[0106] In some embodiments, the target neural network includes a target variational encoder; inputting a first distance difference, a first angle, and first information into the target neural network to obtain a second distance difference and a second angle may include: inputting the first distance difference, the first angle, and the first information into the target variational encoder to obtain a first latent variable and a second latent variable; determining the second distance difference based on the first latent variable; and determining the second angle based on the second latent variable.

[0107] Furthermore, in some embodiments, the first latent variable is obtained by fusing the first information with the first distance difference in the target variational encoder, and the second latent variable is obtained by fusing the first information with the first angle in the target variational encoder.

[0108] For example, the first latent variable can be a distance difference latent variable (also known as a distance difference codeword).

[0109] For example, the second latent variable can be an angle latent variable (also known as an angle codeword).

[0110] This method achieves two objectives: firstly, by employing variational inference, the representational capability of the encoder can be further enhanced, making the encoder's output more accurate; secondly, by fully utilizing the effective information in the first information, the first information can be fused with the first distance difference to obtain the first latent variable, and the first information can be fused with the first angle to obtain the second latent variable, thereby making the subsequent second distance difference and second angle obtained based on the first and second latent variables more accurate.

[0111] There are two ways to obtain the second distance difference based on the first latent variable:

[0112] One possible implementation is to determine the third mapping relationship between the latent variables and the distance difference; based on the third mapping relationship and the first latent variable, the second distance difference can be obtained.

[0113] This method obtains the second distance difference based on the third mapping relationship and the first latent variable, which simplifies the acquisition of the second distance difference. Subsequent use of the second distance difference to locate the terminal device can make the positioning more accurate.

[0114] Another possible implementation is that the target neural network may also include a target first regressor; determining the second distance difference based on the first latent variable may include: inputting the first latent variable into the target first regressor to obtain the second distance difference.

[0115] For example, the target first regressor can be the target distance difference regressor (also known as the target distance difference calibrator).

[0116] By using this method, the first latent variable can be input into the target first regressor, which can make the obtained second distance difference more accurate. Subsequently, the second distance difference can be used to locate the terminal device, making the positioning more precise.

[0117] There are two ways to obtain the second angle based on the second latent variable:

[0118] One possible implementation is to determine the fourth mapping relationship between the latent variables and the angle; based on the fourth mapping relationship and the second latent variable, the second angle can be obtained.

[0119] This method obtains the second angle based on the fourth mapping relationship and the second latent variable, which simplifies the acquisition of the second angle. Subsequent use of the second angle to locate the terminal device can make the positioning more accurate.

[0120] Another possible implementation is that the target neural network may also include a target second regressor; determining the second angle based on the second latent variable may include: inputting the second latent variable into the target second regressor to obtain the second angle.

[0121] For example, the target second regressor can be a target angle regressor (also known as a target angle calibrator).

[0122] By using this method, the second latent variable can be input into the target second regressor, making the obtained second angle more accurate. Subsequently, the second angle can be used to locate the terminal device, making the positioning more precise.

[0123] It should be understood that the target neural network is a pre-trained neural network. The following sections, with reference to Figures 3 and 4, illustrate the process of obtaining the target neural network by training the neural network to be trained.

[0124] Figure 3 is a flowchart illustrating a neural network training method provided in an embodiment of this application. As shown in Figure 3, the method may include the following steps.

[0125] S310. Obtain the third distance difference, the third angle, and the second information; wherein, the third distance difference is used to characterize the measured distance difference between the terminal device and the two network devices, the third angle is used to characterize the measured angle of the terminal device relative to the two network devices, and the second information is used to characterize the channel state between the terminal device and the two network devices.

[0126] S320. Input the third distance difference, the third angle, and the second information into the neural network to be trained to obtain the fourth distance difference and the fourth angle.

[0127] S330. Based on the fifth distance difference and the fourth distance difference, determine the first loss; and based on the fifth angle and the fourth angle, determine the second loss; wherein the fifth distance difference is used to characterize the real distance difference between the terminal device and the two network devices, and the fifth angle is used to characterize the real angle of the terminal device relative to the two network devices.

[0128] S340. Based on the first loss and the second loss, the network parameters of the neural network to be trained are adjusted so that the loss obtained based on the adjusted neural network satisfies the convergence condition to obtain the target neural network; the target neural network is used to locate the terminal device.

[0129] It should be noted that, in the embodiments of this application, the training method of the neural network can be applied to a neural network training device or a computer device integrated with such a device. The computer device can be a network device or other devices, and this application embodiment does not limit the specific type of device.

[0130] In some embodiments, obtaining the third distance difference and the third angle may include: acquiring the third distance difference and the third angle by building a 5G signal acquisition platform and using an array antenna.

[0131] In some embodiments, obtaining the second information may include: building a 5G signal acquisition platform and using an array antenna to receive the SRS sent by the terminal device; and obtaining the second information based on the SRS.

[0132] It should be noted that one or more of the third distance difference, the third angle, and the second information can be obtained directly by the computer device or obtained by the computer device through other computer devices. This application embodiment does not limit this.

[0133] In some embodiments, the second information may include one or more of the following:

[0134] The received signal energy of the CSI signal;

[0135] The maximum amplitude of the CSI signal;

[0136] The rise time of the CSI signal;

[0137] Average delay of CSI signal;

[0138] Root mean square spread of CSI signals; and

[0139] Kubularity of CSI signal.

[0140] It should be noted that the calculation formulas for each parameter in the second information can refer to formulas (1) to (6) in the aforementioned embodiments, and will not be repeated in this application embodiment.

[0141] It should also be noted that during the training of the neural network, since the position of the terminal device is not fixed, the obtained third distance difference may include one or more distance differences, the third angle may include one or more angles, and the second information may include one or more pieces of information; wherein, at a certain position of the terminal device, a distance difference, two angles, and two pieces of information between the terminal device and the two network devices can be obtained.

[0142] For example, Figure 4 is a schematic diagram of the location of a terminal device and a network device provided in an embodiment of this application. As shown in Figure 4, a third distance difference, a third angle, and second information can be obtained at 77 locations where the terminal device is located. That is, at location 1 where the terminal device is located, distance difference 1, angle 1 and angle 2, and information 1 and information 2 are obtained; at location 2 where the terminal device is located, distance difference 2, angle 3 and angle 4, and information 3 and information 4 are obtained; ...; at location 77 where the terminal device is located, distance difference 77, angle 153 and angle 154, and information 153 and information 154 are obtained. The third distance difference may include distance difference 1, distance difference 2, ..., distance difference 77; the third angle may include angle 1, angle 2, ..., angle 154; and the second information may include information 1, information 2, ..., information 154.

[0143] In some embodiments, determining the first loss based on the fifth distance difference and the fourth distance difference may include: performing a subtraction operation on the fifth distance difference and the fourth distance difference to obtain the first loss.

[0144] In some embodiments, determining the second loss based on the fifth angle and the fourth angle may include: subtracting the fifth angle and the fourth angle to obtain the second loss.

[0145] It should be noted that when the neural network to be trained includes one or more of the following: a variational encoder to be trained, a first regressor to be trained, a second regressor to be trained, and a variational decoder to be trained, adjusting the network parameters of the neural network to be trained based on the first loss and the second loss may include: adjusting one or more of the following based on the first loss and the second loss:

[0146] The network parameters of the variational encoder to be trained;

[0147] The network parameters of the first regressor to be trained;

[0148] The network parameters of the second regressor to be trained;

[0149] Network parameters of the variational decoder to be trained.

[0150] It should be understood that in the embodiments of this application, network parameters may also be referred to as weights, weights, etc., and the embodiments of this application do not limit them to this.

[0151] It should also be understood that before inputting the third distance difference, the third angle, and the second information into the neural network to be trained, one or more of the network parameters of the variational encoder, the first regressor, the second regressor, and the variational decoder to be trained can be initialized.

[0152] In some embodiments, inputting the third distance difference, the third angle, and the second information into the neural network to be trained may include: inputting the third distance difference, the third angle, the second information, the learning rate, and the batch processing length into the neural network to be trained.

[0153] The learning rate can be used to control the learning efficiency of the neural network to be trained.

[0154] The batch length can be used to control the processing efficiency of the neural network to be trained.

[0155] In some embodiments, adjusting the network parameters of the neural network to be trained based on the first loss and the second loss may include: adjusting the network parameters of the neural network to be trained based on the first loss and the second loss and using a gradient descent algorithm.

[0156] Based on the first loss and the second loss, the network parameters of the neural network to be trained can be adjusted in the following two possible ways:

[0157] One possible implementation involves determining a fifth loss based on the first and second losses, and then adjusting the network parameters of the neural network to be trained based on the fifth loss.

[0158] The determination of the fifth loss based on the first and second losses may include:

[0159] The first and second losses are summed to obtain the fifth loss; or,

[0160] The fifth loss is obtained by multiplying the first and second losses.

[0161] Another possible implementation involves using a first loss and a second loss to adjust the network parameters of the neural network to be trained, respectively.

[0162] This method adjusts the network parameters of the neural network to be trained based on the first loss and the second loss, thereby obtaining the target neural network. Subsequently, the positioning accuracy of the terminal device can be improved by using the target neural network to locate the terminal device.

[0163] In some embodiments, the loss obtained by the adjusted neural network satisfies the convergence condition, which may include: the loss obtained by the adjusted neural network is less than or equal to a fourth threshold.

[0164] The fourth threshold can be specified by the protocol, set by the network device, or set in other ways. This application embodiment does not limit this.

[0165] It should be understood that the target neural network is used to locate the terminal device. This can be understood as follows: after obtaining the target neural network, the terminal device can be located based on the output parameters of the target neural network (such as the second distance difference and the second angle mentioned in the previous embodiments).

[0166] This application provides a method for training a neural network. The method involves acquiring a third distance difference, a third angle, and second information; inputting these three information into a neural network to be trained to obtain a fourth distance difference and a fourth angle; determining a first loss based on the fifth and fourth distance differences; determining a second loss based on the fifth and fourth angles; adjusting the network parameters of the neural network to be trained based on the first and second losses to ensure that the loss obtained from the adjusted neural network satisfies the convergence condition, thus obtaining a target neural network; and using the target neural network to locate a terminal device. In this way, by training the neural network to be trained, a target neural network can be obtained, and subsequent positioning of the terminal device based on the target neural network can improve the positioning accuracy of the terminal device.

[0167] In some embodiments, the second information may be used to correct the third distance difference to obtain the fourth distance difference in the neural network to be trained; and to correct the third angle to obtain the fourth angle.

[0168] Using this method, in the neural network to be trained, the second information can be used to correct the third distance difference and the third angle. Based on the obtained fourth distance difference and the fourth angle, the neural network to be trained can be trained to obtain the target neural network. Subsequently, the positioning of the terminal device can be improved based on the target neural network.

[0169] In some embodiments, the neural network to be trained includes a variational encoder to be trained; inputting the third distance difference, the third angle, and the second information into the neural network to be trained to obtain the fourth distance difference and the fourth angle may include:

[0170] The third distance difference, the third angle, and the second information are input into the variational encoder to be trained to obtain the third and fourth latent variables.

[0171] The fourth distance difference is determined based on the third latent variable;

[0172] The fourth angle is determined based on the fourth latent variable.

[0173] For example, the third latent variable can be a distance difference latent variable (also known as a distance difference codeword).

[0174] For example, the fourth latent variable can be an angle latent variable (also known as an angle codeword).

[0175] This method, employing the idea of ​​variational inference, can further enhance the encoder's representational ability, making the encoder's output more accurate. The fourth distance difference and fourth angle obtained based on the third and fourth latent variables will also be more accurate.

[0176] In some embodiments, adjusting the network parameters of the neural network to be trained based on the first loss and the second loss may include: inputting a third distance difference, a third angle, and second information into the neural network to be trained to obtain a fourth distance difference, a fourth angle, and reconstructed second information; determining a third loss based on the second information and the reconstructed second information; and adjusting the network parameters of the neural network to be trained based on the first loss, the second loss, and the third loss.

[0177] It should be noted that determining the third loss based on the second information and the reconstructed second information may include: performing a subtraction operation on the second information and the reconstructed second information to obtain the third loss.

[0178] Based on the first loss, second loss, and third loss, the network parameters of the neural network to be trained can be adjusted in the following two possible ways:

[0179] One possible implementation involves determining a fourth loss based on the first, second, and third losses; and then adjusting the network parameters of the neural network to be trained based on the fourth loss.

[0180] The determination of the fourth loss based on the first loss, the second loss, and the third loss may include:

[0181] The first, second, and third losses are summed to obtain the fourth loss; or,

[0182] The fourth loss is obtained by multiplying the first, second, and third losses.

[0183] Another possible implementation involves using a first loss, a second loss, and a third loss to adjust the network parameters of the neural network to be trained, respectively.

[0184] This method adjusts the network parameters of the neural network to be trained based on the first loss, the second loss, and the third loss, thereby obtaining the target neural network. Subsequently, the positioning accuracy of the terminal device can be improved by using the target neural network to locate the terminal device.

[0185] In some embodiments, the neural network to be trained includes a variational encoder to be trained; inputting the third distance difference, the third angle, and the second information into the neural network to be trained to obtain the fourth distance difference, the fourth angle, and the reconstructed second information may include:

[0186] The third distance difference, the third angle, and the second information are input into the variational encoder to be trained to obtain the third and fourth latent variables.

[0187] The fourth distance difference is determined based on the third latent variable;

[0188] Determine the fourth angle based on the fourth latent variable;

[0189] Based on the third and fourth latent variables, the reconstructed second information is determined.

[0190] This method, employing the idea of ​​variational inference, can further enhance the representational ability of the encoder, making the encoder's output more accurate. The fourth distance difference, fourth angle, and reconstructed second information obtained based on the third and fourth latent variables will also be more accurate.

[0191] In some embodiments, the third latent variable is obtained by fusing the second information and the third distance difference in the variational encoder, and the fourth latent variable is obtained by fusing the second information and the third angle in the variational encoder.

[0192] This method can fully utilize the effective information in the second information, fuse the second information with the third distance difference to obtain the third latent variable, and fuse the second information with the third angle to obtain the fourth latent variable. This makes the fourth distance difference, the fourth angle, and the reconstructed second information obtained based on the third and fourth latent variables more accurate.

[0193] There are two ways to obtain the fourth distance difference based on the third latent variable:

[0194] One possible implementation is to determine the third mapping relationship between the latent variables and the distance difference; based on the third mapping relationship and the third latent variable, the fourth distance difference can be obtained.

[0195] This method simplifies the acquisition of the fourth distance difference by obtaining the fourth distance difference based on the third mapping relationship and the third latent variable. Subsequently, using the fourth distance difference to train the neural network to be trained can result in a higher accuracy of the target neural network.

[0196] Another possible implementation is that the neural network to be trained may also include a first regressor to be trained; determining the fourth distance difference based on the third latent variable may include: inputting the third latent variable into the first regressor to be trained to obtain the fourth distance difference.

[0197] For example, the first regressor to be trained can be a distance difference regressor to be trained, or it can be called a distance difference calibrator to be trained.

[0198] By using this method, the third latent variable can be input into the first regressor to be trained, which can make the obtained fourth distance difference more accurate. Subsequently, the neural network to be trained can be trained using the fourth distance difference, which can make the obtained target neural network more accurate.

[0199] There are two ways to obtain the fourth angle based on the fourth latent variable:

[0200] One possible implementation is to determine the fourth mapping relationship between the latent variables and the angle; based on the fourth mapping relationship and the fourth latent variable, the fourth angle can be obtained.

[0201] This method simplifies the acquisition of the fourth angle by obtaining the fourth mapping relationship and the fourth latent variable. Subsequently, using the fourth angle to train the neural network can result in higher accuracy of the target neural network.

[0202] Another possible implementation is that the neural network to be trained may also include a second regressor to be trained; determining the fourth angle based on the fourth latent variable may include: inputting the fourth latent variable into the second regressor to be trained to obtain the fourth angle.

[0203] For example, the second regressor to be trained can be an angle regressor to be trained, or it can be called an angle calibrator to be trained.

[0204] By using this method, the fourth latent variable can be input into the second regressor to be trained, which can make the obtained fourth angle more accurate. Subsequently, using the fourth angle to train the neural network can result in a higher accuracy of the target neural network.

[0205] There are two ways to obtain the reconstructed second information based on the third and fourth latent variables:

[0206] One possible implementation is to determine the fifth mapping relationship between the two latent variables and the information (which is used to characterize the channel state between the terminal device and the two network devices); based on the third latent variable, the fourth latent variable and the fifth mapping relationship, the reconstructed second information can be obtained.

[0207] This method obtains the reconstructed second information based on the third, fourth, and fifth latent variables and the mapping relationship. This simplifies the acquisition of the reconstructed second information. Subsequently, using the reconstructed second information to train the neural network can result in a higher accuracy of the target neural network.

[0208] Another possible implementation is that the network to be trained also includes a variational decoder to be trained; the reconstructed second information is determined based on the third and fourth hidden variables, which may include: inputting the third and fourth hidden variables into the variational decoder to be trained to obtain the reconstructed second information.

[0209] This method, employing the idea of ​​variational inference, can further enhance the representational ability of the decoder, making the decoder's output more accurate. Subsequently, using the reconstructed second information to train the neural network to be trained can result in a higher accuracy of the target neural network.

[0210] It should be understood that this application embodiment uses two network devices as an example. It can be understood that more than two network devices can also be used in the training of neural networks and the positioning of terminal devices. This application embodiment does not limit this.

[0211] It should also be understood that using more than two network devices to locate a terminal device can achieve higher positioning accuracy compared to using two network devices to locate the terminal device; similarly, using more than two network devices to train a neural network can achieve higher accuracy in the target neural network compared to using two network devices to train the neural network.

[0212] For example, taking the positioning of a terminal device as an example, the terminal device can be positioned based on the measurement distance difference between the terminal device and two or more network devices, the measurement angle of the terminal device relative to the two or more network devices, and the channel status between the terminal device and the two or more network devices.

[0213] The measurement distance difference between the terminal device and two or more network devices may include: dividing two or more network devices into pairs and the measurement distance difference between the terminal device and each group of network devices.

[0214] In this embodiment, a target neural network can be obtained by training a neural network to be trained, and the second distance difference and second angle obtained by the target neural network can be used to locate the terminal device. The training method and localization method of the neural network provided in this embodiment are described in detail below with reference to specific application scenarios.

[0215] Example 1: Training a neural network.

[0216] The training of the neural network in this application embodiment mainly includes the following steps:

[0217] The first step involves building a 5G signal acquisition platform based on commercial 5G equipment. This platform includes an array antenna. Using the array antenna, the third distance difference and third angle between the terminal device and two network devices are obtained. Furthermore, the SRS transmitted by the terminal device can be acquired using the array antenna. Based on the SRS, second information (i.e., the CSI data before reconstruction, which is the measured CSI data) can be obtained. This second information may include the received signal energy of the CSI signal, the maximum amplitude of the CSI signal, the rise time of the CSI signal, the average delay of the CSI signal, the root mean square spread of the CSI signal, the kurtosis of the CSI signal, etc.

[0218] For example, as shown in Figure 4, the third distance difference, the third angle, and the second information between the terminal device and the two network devices can be recorded based on the different positions of the terminal device until no new identifiable features are added (that is, when the position of the terminal device is changed again, the distance difference, angle, and the second information are almost the same as the existing position).

[0219] The second step involves inputting the third distance difference, the third angle, and the second information into the neural network to be trained. This neural network can include four sub-network modules: a variational encoder to be trained, a distance difference regressor to be trained (or a distance difference calibrator to be trained, i.e., the first regressor to be trained in the aforementioned embodiments), an angle regressor to be trained (or an angle calibrator to be trained, i.e., the second regressor to be trained in the aforementioned embodiments), and a variational decoder to be trained.

[0220] The variational encoder and variational decoder can be combined to form a variational autoencoder (VAE).

[0221] It should be noted that, in the embodiments of this application, a hierarchical latent variable model based on distance difference latent variables and angle latent variables is built to design the neural network to be trained, so that the joint distance difference estimation (i.e., obtaining the second distance difference by correcting the first distance difference) and angle estimation (i.e., obtaining the second angle by correcting the first angle) can be performed based on the trained neural network (i.e., the target neural network).

[0222] Let the fourth distance difference be denoted as d, the fourth angle as θ, and the third latent variable as z. d The fourth hidden variable is denoted as z. θ Let the second piece of information be denoted as x. Then, the construction of the hierarchical latent variable model can include:

[0223] (1) In general (e.g., without SRS), the distributions of d and θ are independent. That is, d and θ satisfy the following distribution formula: p(d,θ)=p(d)p(θ) (7)

[0224] (2)z d Given d, independent of θ and z θ ;and z θ Given θ, independent of d and z d In other words, d and z d ,θ,z θ Distributions that satisfy the following formula: p(z) d ,z θ ,d,θ)=p(z d |d)p(z θ |θ) (8)

[0225] (3) Given z d and z θ Under the condition that x is independent of d and θ: p(x|z) d ,z θ ,d,θ)=p(x|Vz d ,z θ (9)

[0226] (4) The above assumptions describe the generation process of x in the hierarchical latent variable model. This generation process can derive the following decomposition relationship between latent variables: p(x,z) d ,z θ ,d,θ)=p(d)p(θ)p(z d |d)p(z θ |θ)p(x|z d ,z θ (10)

[0227] Based on formula (10), the hierarchical latent variable model shown in Figure 5 can be designed. As shown in Figure 5, the third latent variable z can be obtained based on the fourth distance difference d. d The fourth hidden variable z can be obtained based on the fourth angle θ. θ Based on the third hidden variable z d and the fourth hidden variable z θ We can obtain the second piece of information, x.

[0228] Based on this, the estimation of the fourth distance difference d and the fourth angle θ can be performed through the following steps: Based on the second information x, estimate the third latent variable z. d and the fourth hidden variable z θ According to the third hidden variable zd Estimate the fourth distance difference d; based on the fourth hidden variable z θ Estimate the fourth angle θ. Therefore, a neural network to be trained can be designed such that the neural network takes the second information x as input (and can also take the third distance difference and the third angle as input), and outputs the fourth distance difference d and the fourth angle θ.

[0229] It should be noted that, in the embodiments of this application, the four sub-network modules involved in the neural network to be trained—the variational encoder to be trained, the distance difference regressor to be trained, the angle regressor to be trained, and the variational decoder to be trained—can be designed based on the Evidence Lower Bound (ELBO).

[0230] For example, the formula for calculating the lower bound of evidence is as follows:

[0231] It should be understood that L ELBO Log-likelihood with the true log (i.e. The difference between the approximate distribution and the true posterior distribution is the Kullback-Leibler divergence (KL Divergence), which can also be called relative entropy. That is, in formula (11), D KL (q(z d ,z θ |x;Φ)||p(z d ,z θ When )) equals 0, the inequality shown in formula (11) can be considered equal. This can be achieved by considering the unknown parameters Φ, Ω, Maximize L ELBO To approximate

[0232] Dataset It includes N independent and identically distributed data samples, each of which includes reconstructed second information. (i.e., the reconstructed CSI data), the fourth distance difference d, and the fourth angle θ, are used as unknown parameters Φ, Ω, The training. The trained Φ is denoted as Φ. * The trained Ω is denoted as Ω. * After training Recorded as Post-training Recorded as Then we have the following formula:

[0233] Considering variational distribution Conditional distribution Conditional distribution Conditional distribution This is not easy to implement, so we can consider designing an approximate variational distribution q(z) of the variational encoder to be trained. d ,z θ |x;Φ), where Φ is the network parameter of the variational encoder to be trained; design the approximate conditional distribution of the distance difference regressor to be trained. Define the network parameters for the distance difference regressor to be trained; design the approximate conditional distribution for the angle regressor to be trained. Define the network parameters for the angle regressor to be trained; design the approximate conditional distribution of the variational decoder to be trained. Ω represents the network parameters of the variational decoder to be trained.

[0234] Based on this, a neural network as shown in Figure 6 can be constructed. This neural network may include a variational encoder 610, a distance difference regressor 620, an angle regressor 630, and a variational decoder 640. Based on the neural network shown in Figure 6, the reconstructed second information can be obtained. The fourth distance difference d and the fourth angle θ, based on d, θ, for network parameters (i.e., unknown parameters) Φ, Ω, Training is conducted to obtain the trained Φ * Ω * ,

[0235] The third step, as shown in Figure 6, involves fusing the second information and the third distance difference in the variational encoder 610 to obtain the third latent variable, fusing the second information and the third angle to obtain the fourth latent variable, inputting the third latent variable into the distance difference regressor 620 to obtain the fourth distance difference, inputting the fourth latent variable into the angle regressor 630 to obtain the fourth angle, and inputting the third and fourth latent variables into the variational decoder 640 to obtain the reconstructed second information (i.e., the reconstructed CSI data).

[0236] The fourth step involves obtaining the first loss based on the fourth distance difference and the true distance difference (i.e., the fifth distance difference); obtaining the second loss based on the fourth angle and the true angle (i.e., the fifth angle); obtaining the third loss based on the second information and the reconstructed second information; accumulating the first, second, and third losses to obtain the fourth loss (i.e., the loss function); and using the fourth loss to adjust the network parameters of the neural network to be trained so that the loss obtained based on the adjusted neural network satisfies the convergence condition, thus obtaining the target neural network.

[0237] For example, for network parameters Φ, Ω, The training process may include:

[0238] (1) Input the third distance difference, the third angle, and the second information into the neural network to be trained; in addition, the learning rate and batch processing length can also be input into the neural network to be trained.

[0239] (2) Based on the third distance difference, the third angle, and the second information, the fourth loss is obtained;

[0240] (3) Based on the fourth loss, the gradient descent algorithm is used to adjust the network parameters Φ, Ω, and Training continues until the loss obtained by the trained neural network satisfies the convergence condition. At this point, the trained Φ can be obtained. * Ω * ,

[0241] Example 2: Localization of terminal devices based on target neural networks.

[0242] This application's embodiments are applicable to the field of 5G signal positioning. Based on a target neural network, it can perform integrated calibration of a first distance difference and a first angle to obtain a second distance difference and a second angle. This allows for the positioning of terminal devices based on the combined second distance difference and second angle. This helps reduce distance difference and angle errors, thereby improving positioning accuracy.

[0243] This application embodiment establishes a target neural network for estimating multiple parameters (such as the second distance difference and the second angle), which improves the positioning capability of the current 5G base station positioning system in scenarios such as non-line-of-sight propagation.

[0244] This application's embodiments employ the idea of ​​variational inference (such as a target variational encoder or a target variational decoder) to further enhance the representation capability of the target neural network. By extracting position-related latent variables (such as a first latent variable or a second latent variable) from the first information (i.e., CSI data), the effective information in the first information is fully utilized. By building a target neural network, the joint correction of the first distance difference and the first angle is achieved, thereby enhancing the positioning system's ability to resist non-line-of-sight propagation.

[0245] This application proposes a 5G positioning method based on joint calibration of distance difference and angle (such as joint calibration of first distance difference and first angle), which avoids the problems of separate calibration of distance difference and angle in related technologies, as well as large positioning errors of terminal devices under non-line-of-sight propagation, and has good adaptability and practicality.

[0246] In this embodiment of the application, the localization of the terminal device based on the target neural network mainly includes the following steps:

[0247] (1) Input the first information, the first distance difference and the first angle into the trained variational encoder (i.e. the target variational encoder) to obtain the first latent variable and the second latent variable;

[0248] (2) Input the first latent variable into the trained distance difference regressor (i.e., the target first regressor) and input the second latent variable into the trained angle regressor (i.e., the target second regressor);

[0249] (3) Output the second distance difference through the trained distance difference regressor and output the second angle through the trained angle regressor;

[0250] (3) Position the terminal device based on the second distance difference and the second angle.

[0251] The positioning method provided in this application improves the positioning capability of terminal devices in non-line-of-sight scenarios. On one hand, it enables the construction of a 5G signal acquisition platform based on commercial 5G equipment, without requiring the rebuilding of a complete system and platform, making it more versatile in commercial applications. On the other hand, this application can learn from CSI data, integrating the correlation between distance difference and angle, and mining latent variables of distance difference and angle, thereby improving positioning accuracy in non-line-of-sight scenarios. Furthermore, this application constructs a VAE-based neural network based on a lower bound of evidence, and by using a gradient descent algorithm, the neural network exhibits good convergence, resulting in robust positioning capabilities.

[0252] In this application embodiment, due to the wide applicability and low cost of 5G, obtaining CSI data from 5G uplink SRS signals has good application and practice. Furthermore, the positioning method provided in this application embodiment can also be applied to scenarios such as mobile hotspots (e.g., Wi-Fi), ultra-wideband (UWB) signals, and Bluetooth signals.

[0253] The preferred embodiments of this application have been described in detail above with reference to the accompanying drawings. However, this application is not limited to the specific details of the above embodiments. Within the scope of the technical concept of this application, various simple modifications can be made to the technical solutions of this application, and these simple modifications all fall within the protection scope of this application. For example, the various specific technical features described in the above specific embodiments can be combined in any suitable manner without contradiction. To avoid unnecessary repetition, this application will not describe the various possible combinations separately. Furthermore, various different embodiments of this application can also be arbitrarily combined, as long as they do not violate the spirit of this application, they should also be considered as the content disclosed in this application. Moreover, without conflict, the various embodiments and / or the technical features in the various embodiments described in this application can be arbitrarily combined with the prior art, and the resulting technical solutions should also fall within the protection scope of this application.

[0254] It should also be understood that, in the various method embodiments of this application, the order of the above-mentioned processes does not imply the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of this application.

[0255] Figure 7 is a schematic diagram of the structural composition of a positioning device 700 provided in an embodiment of this application, which is applied to a computer device. As shown in Figure 7, the positioning device 700 may include:

[0256] The first acquisition unit 710 is configured to acquire a first distance difference, a first angle, and first information; wherein the first distance difference is used to characterize the measured distance difference between the terminal device and the two network devices, the first angle is used to characterize the measured angle of the terminal device relative to the two network devices, and the first information is used to characterize the channel state between the terminal device and the two network devices.

[0257] The positioning unit 720 is configured to locate the terminal device based on a first distance difference, a first angle, and first information.

[0258] In some embodiments, the positioning device 700 may further include:

[0259] The first processing unit 730 is configured to perform correction processing on the first distance difference based on the first information to obtain a second distance difference; and to perform correction processing on the first angle based on the first information to obtain a second angle.

[0260] The positioning unit 720 is also configured to locate the terminal device based on a second distance difference and a second angle.

[0261] In some embodiments, the first processing unit 730 is further configured to input the first distance difference, the first angle, and the first information into the target neural network to obtain the second distance difference and the second angle; the positioning unit 720 is further configured to locate the terminal device based on the second distance difference and the second angle.

[0262] In some embodiments, the target neural network includes a target variational encoder; the first processing unit 730 is further configured to input a first distance difference, a first angle and first information into the target variational encoder to obtain a first latent variable and a second latent variable; determine a second distance difference based on the first latent variable; and determine a second angle based on the second latent variable.

[0263] In some embodiments, the first latent variable is obtained by fusing the first information with the first distance difference in the target variational encoder, and the second latent variable is obtained by fusing the first information with the first angle in the target variational encoder.

[0264] In some embodiments, the target neural network further includes a target first regressor; the first processing unit 730 is further configured to input a first latent variable into the target first regressor to obtain a second distance difference.

[0265] In some embodiments, the target neural network further includes a target second regressor; the first processing unit 730 is further configured to input a second latent variable into the target second regressor to obtain a second angle.

[0266] In some embodiments, the first information includes one or more of the following:

[0267] The received signal energy of the Channel State Information (CSI) signal;

[0268] Maximum amplitude of CSI signal;

[0269] The rise time of the CSI signal;

[0270] Average delay of CSI signal;

[0271] Root mean square spread of CSI signals; and

[0272] Kubularity of CSI signal.

[0273] This application provides a positioning device that can acquire a first distance difference, a first angle, and first information. The first distance difference characterizes the measured distance difference between a terminal device and two network devices, the first angle characterizes the measured angle of the terminal device relative to the two network devices, and the first information characterizes the channel state between the terminal device and the two network devices. The terminal device is positioned based on the first distance difference, the first angle, and the first information. Thus, by introducing the first information into the positioning process of the terminal device and combining the first distance difference, the first angle, and the first information, the positioning accuracy of the terminal device can be improved.

[0274] Those skilled in the art should understand that the description of the positioning device in the embodiments of this application can be understood with reference to the description of the positioning method in the embodiments of this application.

[0275] Figure 8 is a schematic diagram of the structure of a neural network training device 800 provided in an embodiment of this application, applied to a computer device. As shown in Figure 8, the neural network training device 800 may include:

[0276] The second acquisition unit 810 is configured to acquire a third distance difference, a third angle, and second information; wherein the third distance difference is used to characterize the measured distance difference between the terminal device and the two network devices, the third angle is used to characterize the measured angle of the terminal device relative to the two network devices, and the second information is used to characterize the channel state between the terminal device and the two network devices.

[0277] The second processing unit 820 is configured to input the third distance difference, the third angle, and the second information into the neural network to be trained to obtain the fourth distance difference and the fourth angle.

[0278] The second processing unit 820 is further configured to determine a first loss based on a fifth distance difference and a fourth distance difference; and to determine a second loss based on a fifth angle and a fourth angle; wherein the fifth distance difference is used to characterize the real distance difference between the terminal device and the two network devices, and the fifth angle is used to characterize the real angle of the terminal device relative to the two network devices.

[0279] The second processing unit 820 is further configured to adjust the network parameters of the neural network to be trained based on the first loss and the second loss, so that the loss obtained based on the adjusted neural network satisfies the convergence condition to obtain the target neural network; the target neural network is used to locate the terminal device.

[0280] In some embodiments, the second information is used to correct the third distance difference in the neural network to be trained to obtain a fourth distance difference; and to correct the third angle to obtain a fourth angle.

[0281] In some embodiments, the neural network to be trained includes a variational encoder to be trained; the second processing unit 820 is further configured to input a third distance difference, a third angle, and second information into the variational encoder to obtain a third latent variable and a fourth latent variable; determine a fourth distance difference based on the third latent variable; and determine a fourth angle based on the fourth latent variable.

[0282] In some embodiments, the second processing unit 820 is further configured to input the third distance difference, the third angle, and the second information into the neural network to be trained to obtain the fourth distance difference, the fourth angle, and the reconstructed second information; determine the third loss based on the second information and the reconstructed second information; and adjust the network parameters of the neural network to be trained based on the first loss, the second loss, and the third loss.

[0283] In some embodiments, the second processing unit 820 is further configured to determine a fourth loss based on the first loss, the second loss, and the third loss; and to adjust the network parameters of the neural network to be trained based on the fourth loss.

[0284] In some embodiments, the neural network to be trained includes a variational encoder to be trained; the second processing unit 820 is further configured to input a third distance difference, a third angle, and second information into the variational encoder to obtain a third latent variable and a fourth latent variable; determine a fourth distance difference based on the third latent variable; determine a fourth angle based on the fourth latent variable; and determine the reconstructed second information based on the third latent variable and the fourth latent variable.

[0285] In some embodiments, the third latent variable is obtained by fusing the second information and the third distance difference in the variational encoder, and the fourth latent variable is obtained by fusing the second information and the third angle in the variational encoder.

[0286] In some embodiments, the neural network to be trained further includes a first regressor to be trained; the second processing unit 820 is further configured to input a third latent variable into the first regressor to obtain a fourth distance difference.

[0287] In some embodiments, the neural network to be trained further includes a second regressor to be trained; the second processing unit 820 is further configured to input a fourth latent variable into the second regressor to obtain a fourth angle.

[0288] In some embodiments, the network to be trained further includes a variational decoder second processing unit 820, which is further configured to input the third and fourth hidden variables into the variational decoder to obtain the reconstructed second information.

[0289] In some embodiments, the second information includes one or more of the following:

[0290] The received signal energy of the Channel State Information (CSI) signal;

[0291] The maximum amplitude of the CSI signal;

[0292] The rise time of the CSI signal;

[0293] Average delay of CSI signal;

[0294] Root mean square spread of CSI signals; and

[0295] Kubularity of CSI signal.

[0296] This application provides a neural network training device that can acquire a third distance difference, a third angle, and second information; input the third distance difference, third angle, and second information into the neural network to be trained to obtain a fourth distance difference and a fourth angle; determine a first loss based on the fifth distance difference and the fourth distance difference; and determine a second loss based on the fifth angle and the fourth angle; adjust the network parameters of the neural network to be trained based on the first loss and the second loss so that the loss obtained based on the adjusted neural network satisfies the convergence condition to obtain a target neural network; the target neural network is used to locate a terminal device. Thus, by training the neural network to be trained, a target neural network can be obtained, and subsequent positioning of the terminal device based on the target neural network can improve the positioning accuracy of the terminal device.

[0297] Those skilled in the art should understand that the description of the neural network training apparatus in the embodiments of this application can be understood with reference to the description of the neural network training method in the embodiments of this application.

[0298] Figure 9 is a schematic diagram of the structural composition of a computer device according to an embodiment of this application. The computer device 900 shown in Figure 9 may include a processor 910, a memory 920, and a transceiver 930, wherein:

[0299] Memory 920 is used to store computer programs;

[0300] The processor 910, connected to the memory 920, is used to call and run computer programs from the memory 920 to implement the methods in the embodiments of this application;

[0301] The transceiver 930, also known as a communication interface, is used for receiving and sending information when exchanging information with other external devices.

[0302] In some embodiments, the memory 920 may be a separate device independent of the processor 910, or it may be integrated into the processor 910.

[0303] In some embodiments, the transceiver 930 may include an input interface. The processor 910 can control this input interface to communicate with other external devices; specifically, it can receive information or data sent by other external devices.

[0304] In some embodiments, the transceiver 930 may include an output interface. The processor 910 can control this output interface to communicate with other external devices; specifically, it can send information or data to other external devices.

[0305] In some embodiments, transceiver 930 may include a transmitter and a receiver. Transceiver 930 may further include antennas, and the number of antennas may be one or more.

[0306] Figure 10 is a schematic structural diagram of a chip provided in an embodiment of this application. The chip 1000 shown in Figure 10 includes a processor 1010, which can call and run computer programs from memory to implement the methods in the embodiments of this application.

[0307] In some embodiments, as shown in FIG10, chip 1000 may further include memory 1020. Processor 1010 may retrieve and run computer programs from memory 1020 to implement the methods described in the embodiments of this application.

[0308] The memory 1020 can be a separate device independent of the processor 1010, or it can be integrated into the processor 1010.

[0309] In some embodiments, the chip 1000 may further include an input interface 1030. The processor 1010 can control the input interface 1030 to communicate with other devices or chips; specifically, it can acquire information or data sent by other devices or chips.

[0310] In some embodiments, the chip 1000 may further include an output interface 1040. The processor 1010 may control the output interface 1040 to communicate with other devices or chips, specifically, to output information or data to other devices or chips.

[0311] In some embodiments, the chip can be applied to the computer device in the embodiments of this application, and the chip can implement the corresponding processes implemented by the computer device in the various methods of the embodiments of this application. For the sake of brevity, it will not be described in detail here.

[0312] It should be understood that the chip mentioned in the embodiments of this application may also be referred to as a system-on-a-chip, system chip, chip system, or system-on-a-chip, etc.

[0313] It should be understood that the processor in the embodiments of this application may be an integrated circuit chip with signal processing capabilities. In implementation, the steps of the above method embodiments can be completed by integrated logic circuits in the processor's hardware or by instructions in software form. The processor described above can be a general-purpose processor, a digital signal processor (DSP), an application-specific integrated circuit (ASIC), a field-programmable gate array (FPGA), or other programmable logic devices, discrete gate or transistor logic devices, or discrete hardware components. It can implement or execute the methods, steps, and logic block diagrams disclosed in the embodiments of this application. The general-purpose processor can be a microprocessor or any conventional processor. The steps of the methods disclosed in the embodiments of this application can be directly embodied in the execution of a hardware decoding processor, or executed by a combination of hardware and software modules in the decoding processor. The software modules can be located in random access memory, flash memory, read-only memory, programmable read-only memory, electrically erasable programmable memory, registers, or other mature storage media in the art. The storage medium is located in memory, and the processor reads information from the memory and, in conjunction with its hardware, completes the steps of the above method.

[0314] It is understood that the memory in the embodiments of this application can be volatile memory or non-volatile memory, or may include both volatile and non-volatile memory. The non-volatile memory can be read-only memory (ROM), programmable read-only memory (PROM), erasable programmable read-only memory (EPROM), electrically erasable programmable read-only memory (EEPROM), or flash memory. The volatile memory can be random access memory (RAM), which is used as an external cache. By way of example, but not limitation, many forms of RAM are available, such as Static Random Access Memory (SRAM), Dynamic Random Access Memory (DRAM), Synchronous DRAM (SDRAM), Double Data Rate SDRAM (DDR SDRAM), Enhanced Synchronous DRAM (ESDRAM), Synchlink DRAM (SLDRAM), and Direct Rambus RAM (DR RAM). It should be noted that the memory used in the systems and methods described herein is intended to include, but is not limited to, these and any other suitable types of memory.

[0315] This application also provides a computer-readable storage medium storing a computer program that, when executed by at least one processor, implements the method described in this application.

[0316] In some embodiments, the computer-readable storage medium may be applied to the computer device in the embodiments of this application, and the computer program causes the computer to perform the corresponding processes implemented by the computer device in the various methods of the embodiments of this application. For the sake of brevity, these will not be described in detail here.

[0317] This application also provides a computer program product, which includes a computer storage medium storing a computer program. The computer program includes instructions executable by at least one processor, which, when executed by at least one processor, implement the methods described in this application.

[0318] In some embodiments, the computer program product can be applied to the computer device in the embodiments of this application, and the computer program instructions cause the computer to execute the corresponding processes implemented by the computer device in the various methods of the embodiments of this application. For the sake of brevity, they will not be described in detail here.

[0319] This application also provides a computer program that causes a computer to perform the methods described in this application.

[0320] In some embodiments, the computer program can be applied to the computer device in the embodiments of this application. When the computer program is run on the computer, it causes the computer to execute the corresponding processes implemented by the computer device in the various methods of the embodiments of this application. For the sake of brevity, it will not be described in detail here.

[0321] Those skilled in the art will recognize that the units and algorithm steps of the various examples described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of this application.

[0322] Those skilled in the art will understand that, for the sake of convenience and brevity, the specific working processes of the systems, devices, and units described above can be referred to the corresponding processes in the foregoing method embodiments, and will not be repeated here.

[0323] In the embodiments provided in this application, it should be understood that the disclosed systems, apparatuses, and methods can be implemented in other ways. For example, the apparatus embodiments described above are merely illustrative; for instance, the division of units is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the mutual coupling or direct coupling or communication connection shown or discussed may be through some interfaces; the indirect coupling or communication connection of apparatuses or units may be electrical, mechanical, or other forms.

[0324] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of this embodiment according to actual needs.

[0325] In addition, the functional units in the various embodiments of this application can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit.

[0326] If the aforementioned functions are implemented as software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, or a portion of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of this application. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, ROM, RAM, magnetic disks, or optical disks.

[0327] The above description is merely a specific embodiment of this application, but the scope of protection of this application is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in this application should be included within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.

Claims

1. A positioning method, the method comprising: Acquire a first distance difference, a first angle, and first information; wherein, the first distance difference is used to characterize the measured distance difference between the terminal device and the two network devices, the first angle is used to characterize the measured angle of the terminal device relative to the two network devices, and the first information is used to characterize the channel state between the terminal device and the two network devices; The terminal device is located based on the first distance difference, the first angle, and the first information.

2. The method according to claim 1, wherein, The step of locating the terminal device based on the first distance difference, the first angle, and the first information includes: Based on the first information, the first distance difference is corrected to obtain the second distance difference; Based on the first information, the first angle is corrected to obtain the second angle; The terminal device is located based on the second distance difference and the second angle.

3. The method of claim 1 or 2, wherein, The step of locating the terminal device based on the first distance difference, the first angle, and the first information includes: The first distance difference, the first angle, and the first information are input into the target neural network to obtain the second distance difference and the second angle. The terminal device is located based on the second distance difference and the second angle.

4. The method according to claim 3, wherein, The target neural network includes a target variational encoder; The step of inputting the first distance difference, the first angle, and the first information into the target neural network to obtain the second distance difference and the second angle includes: The first distance difference, the first angle, and the first information are input into the target variational encoder to obtain the first latent variable and the second latent variable. Based on the first latent variable, determine the second distance difference; The second angle is determined based on the second latent variable.

5. The method according to claim 4, wherein, The first latent variable is obtained by fusing the first information with the first distance difference in the target variational encoder, and the second latent variable is obtained by fusing the first information with the first angle in the target variational encoder.

6. The method according to claim 4 or 5, wherein, The target neural network further includes a target first regressor; determining the second distance difference based on the first latent variable includes: The first latent variable is input into the target first regressor to obtain the second distance difference.

7. The method according to any one of claims 4 to 6, wherein, The target neural network further includes a target second regressor; determining the second angle based on the second latent variable includes: The second latent variable is input into the target second regressor to obtain the second angle.

8. The method according to any one of claims 1 to 7, wherein, The first information includes one or more of the following: The received signal energy of the Channel State Information (CSI) signal; The maximum amplitude of the CSI signal; The rise time of the CSI signal; Average delay of CSI signal; Root mean square spread of CSI signals; and Kubularity of CSI signal.

9. A method for training a neural network, the method comprising: Acquire a third distance difference, a third angle, and second information; wherein the third distance difference is used to characterize the measured distance difference between the terminal device and the two network devices, the third angle is used to characterize the measured angle of the terminal device relative to the two network devices, and the second information is used to characterize the channel state between the terminal device and the two network devices; The third distance difference, the third angle, and the second information are input into the neural network to be trained to obtain the fourth distance difference and the fourth angle. A first loss is determined based on the fifth distance difference and the fourth distance difference; and a second loss is determined based on the fifth angle and the fourth angle; wherein the fifth distance difference is used to characterize the true distance difference between the terminal device and the two network devices, and the fifth angle is used to characterize the true angle of the terminal device relative to the two network devices; Based on the first loss and the second loss, the network parameters of the neural network to be trained are adjusted so that the loss obtained based on the adjusted neural network satisfies the convergence condition to obtain the target neural network; the target neural network is used to locate the terminal device.

10. The method according to claim 9, wherein, The second information is used to correct the third distance difference in the neural network to be trained to obtain the fourth distance difference; and to correct the third angle to obtain the fourth angle.

11. The method according to claim 9 or 10, wherein, The neural network to be trained includes a variational encoder to be trained; The step of inputting the third distance difference, the third angle, and the second information into the neural network to be trained to obtain the fourth distance difference and the fourth angle includes: The third distance difference, the third angle, and the second information are input into the variational encoder to obtain the third latent variable and the fourth latent variable; Based on the third latent variable, the fourth distance difference is determined; The fourth angle is determined based on the fourth hidden variable.

12. The method according to any one of claims 9 to 11, wherein, The step of adjusting the network parameters of the neural network to be trained based on the first loss and the second loss includes: The third distance difference, the third angle, and the second information are input into the neural network to be trained to obtain the fourth distance difference, the fourth angle, and the reconstructed second information; Based on the second information and the reconstructed second information, a third loss is determined; Based on the first loss, the second loss, and the third loss, the network parameters of the neural network to be trained are adjusted.

13. The method according to claim 12, wherein, Also includes: Based on the first loss, the second loss, and the third loss, a fourth loss is determined; Based on the fourth loss, the network parameters of the neural network to be trained are adjusted.

14. The method according to claim 12 or 13, wherein, The neural network to be trained includes a variational encoder to be trained; the step of inputting the third distance difference, the third angle, and the second information into the neural network to be trained to obtain the fourth distance difference, the fourth angle, and the reconstructed second information includes: The third distance difference, the third angle, and the second information are input into the variational encoder to obtain the third latent variable and the fourth latent variable; Based on the third latent variable, the fourth distance difference is determined; The fourth angle is determined based on the fourth hidden variable; Based on the third and fourth hidden variables, the reconstructed second information is determined.

15. The method according to any one of claims 11 to 14, wherein, The third latent variable is obtained by fusing the second information and the third distance difference in the variational encoder, and the fourth latent variable is obtained by fusing the second information and the third angle in the variational encoder.

16. The method according to any one of claims 11 to 15, wherein, The neural network to be trained further includes a first regressor to be trained; determining the fourth distance difference based on the third latent variable includes: The third latent variable is input into the first regressor to obtain the fourth distance difference.

17. The method according to any one of claims 11 to 16, wherein, The neural network to be trained also includes a second regressor to be trained; determining the fourth angle based on the fourth latent variable includes: The fourth latent variable is input into the second regressor to obtain the fourth angle.

18. The method according to any one of claims 14 to 17, wherein, The network to be trained also includes a variational decoder to be trained; The step of determining the reconstructed second information based on the third latent variable and the fourth latent variable includes: The third and fourth hidden variables are input into the variational decoder to obtain the reconstructed second information.

19. The method according to any one of claims 9 to 18, wherein, The second information includes one or more of the following: The received signal energy of the Channel State Information (CSI) signal; Maximum amplitude of CSI signal; The rise time of the CSI signal; Average delay of CSI signal; Root mean square spread of CSI signals; and Kubularity of CSI signal.

20. A positioning device, the device comprising: The first acquisition unit is configured to acquire a first distance difference, a first angle, and first information; wherein the first distance difference is used to characterize the measured distance difference between the terminal device and the two network devices, the first angle is used to characterize the measured angle of the terminal device relative to the two network devices, and the first information is used to characterize the channel state between the terminal device and the two network devices. The positioning unit is configured to locate the terminal device based on the first distance difference, the first angle, and the first information.

21. A training apparatus for a neural network, the apparatus comprising: The second acquisition unit is configured to acquire a third distance difference, a third angle, and second information; wherein the third distance difference is used to characterize the measured distance difference between the terminal device and the two network devices, the third angle is used to characterize the measured angle of the terminal device relative to the two network devices, and the second information is used to characterize the channel state between the terminal device and the two network devices. The second processing unit is configured to input the third distance difference, the third angle, and the second information into the neural network to be trained to obtain the fourth distance difference and the fourth angle. The second processing unit is further configured to determine a first loss based on a fifth distance difference and a fourth distance difference; and to determine a second loss based on a fifth angle and a fourth angle; wherein the fifth distance difference is used to characterize the actual distance difference between the terminal device and the two network devices, and the fifth angle is used to characterize the actual angle of the terminal device relative to the two network devices; The second processing unit is further configured to adjust the network parameters of the neural network to be trained based on the first loss and the second loss, so that the loss obtained based on the adjusted neural network satisfies the convergence condition to obtain the target neural network; the target neural network is used to locate the terminal device.

22. A computer device, the computer device comprising: Memory, used to store computer programs; A processor, connected to the memory, is configured to retrieve and run the computer program from the memory to implement the method of any one of claims 1 to 8, or to implement the method of any one of claims 9 to 19; A transceiver is used to receive and send information when exchanging information with other external devices.

23. A chip, the chip comprising: Memory, used to store computer programs; A processor, connected to the memory, is configured to retrieve and run a computer program from the memory, causing a device on which the chip is mounted to perform the method as described in any one of claims 1 to 8, or to perform the method as described in any one of claims 9 to 19; A transceiver is used to receive and send information during the exchange of information with a device or chip.

24. A computer-readable storage medium storing a computer program that, when executed by at least one processor, implements the method as claimed in any one of claims 1 to 8, or implements the method as claimed in any one of claims 9 to 19.

25. A computer program product comprising a computer program or instructions which, when executed by a processor, implement the steps of the method as claimed in any one of claims 1 to 8; or implement the steps of the method as claimed in any one of claims 9 to 19.

26. A computer program that causes a computer to perform the method as claimed in any one of claims 1 to 8, or to implement the method as claimed in any one of claims 9 to 19.

Images