Method for improving positioning
By applying data enhancement technology in 5G mobile communication, including jitter conversion, time offset conversion and CVAE methods, a new training data set is generated, which solves the problem of insufficient data sets, improves the performance and accuracy of the positioning model, and reduces the The amount of work done by dataset collection.
Patent Information
- Application Number
- CN202410517011.0
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Priority Date
- 2023-05-12
- Filing Date
- 2024-04-26
- Publication Date
- 2024-11-12
AI Technical Summary
In new 5G radio mobile communications, it is difficult to obtain sufficiently large or balanced tag datasets, resulting in limited positioning performance of artificial intelligence or machine learning models, especially when using incomplete or unbalanced datasets, which are difficult to improve positioning accuracy.
Through data augmentation techniques, including jitter transformation and time offset transformation, a new training data set is generated, and the probability distribution of the original data is learned using the conditional variational autocoding (CVAE) method to generate new samples that follow the distribution to improve the positioning model performance.
In the absence of large amounts of additional data samples, improve positioning performance, reduce the workload and time of data set collection, and enhance the accuracy and generalization capabilities of the positioning model.
Smart Images

Figure CN118945809A_ABST
Abstract
Description
Technical Field
[0001] The present invention relates generally to mobile communications and, more particularly, to improving positioning through data enhancement for artificial intelligence or machine learning (AI / ML) positioning of devices in mobile communications. Background Art
[0002] Unless otherwise indicated, the approaches described in this section are not prior art to the claims listed below and are not admitted to be prior art by inclusion in this section.
[0003] In fifth generation (5G) new radio (NR) mobile communications, artificial intelligence (AI) or machine learning (ML) solutions are introduced to facilitate the positioning of devices (e.g., user equipment (UE)). The performance of AI / ML models depends on the sample density of the training dataset (e.g., # samples / square meter). The larger the size of the training dataset (i.e., the smaller the average distance between samples), the smaller the positioning error (e.g., in meters) until a saturation point is reached. However, it is not easy to obtain or acquire datasets, especially labeled datasets, and it is difficult to obtain sufficiently large or balanced datasets (e.g., uniformly distributed UEs). This often leads to a major problem when trying to train these AI / ML models with incomplete, unbalanced (i.e., non-uniformly distributed), or privacy-challenging datasets.
[0004] Typically, data augmentation techniques can be used to address these issues and thus improve AI / ML positioning performance. In this way, the overhead of training dataset collection can also be reduced. Therefore, in newly developed wireless communication networks, how to improve positioning performance with limited data samples becomes an important issue. Therefore, it is necessary to provide appropriate solutions to improve positioning by using data augmentation techniques in mobile communications. Summary of the invention
[0005] The following summary of the invention is illustrative only and is not intended to be limiting in any way. That is, the following summary of the invention is provided to introduce the concepts, key points, benefits and beneficial effects of the novel and non-obvious technologies described herein. Selected embodiments are further described in the detailed description below. Therefore, the following summary of the invention is not intended to identify the essential features of the claimed subject matter, nor is it intended to be used to determine the scope of the claimed subject matter.
[0006] An object of the present invention is to propose a solution or scheme wherein positioning is improved by data enhancement using artificial intelligence or machine learning (AI / ML) for devices in mobile communications.
[0007] In one aspect, a method may include a training entity sending a request for auxiliary data to a data collection entity. The method may also include the training entity receiving the auxiliary data from the data collection entity. The method may also include the training entity performing model training on a positioning model based on the auxiliary data. The method may also include the training entity determining location information through the positioning model.
[0008] In one aspect, a method may include a data collection entity receiving an auxiliary data request from a training entity. The method may also include the data collection entity generating auxiliary data according to the request. The method may also include the data collection entity sending the auxiliary data to the training entity. The auxiliary data may include at least one of the following: a set of delay profiles of a channel, a set of labels marking each delay profile, a set of quality indicators of the delay profiles and labels, a set of enhanced training data, a set of data enhancement indicators for enhanced training data, a set of parameters or time offset values associated with enhanced training data, an actual data set size, and statistical information related to the data enhancement method.
[0009] In one aspect, a method may include a device acquiring a data input sample. The method may also include the device performing data enhancement to generate enhanced training data based on the data input sample. The method may also include the device performing model training on a positioning model based on the enhanced training data. The method may also include the device determining location information of the device through the positioning model.
[0010] It is worth noting that although the description provided herein may be in the context of certain radio access technologies, networks and network topologies (e.g., Long Term Evolution (LTE), LTE-Advanced, LTE-Advanced Pro, fifth generation systems (5G), new radio (NR), Internet of Things (IoT) and narrowband Internet of Things (NB-IoT), Industrial Internet of Things (IIoT), Beyond 5G (B5G) and Sixth Generation Systems (6G)), the proposed concepts, schemes and any variations or derivatives thereof can be implemented on, for and through any other type of radio access technologies, networks and network topologies. Therefore, the scope of the present invention is not limited to the examples described herein. BRIEF DESCRIPTION OF THE DRAWINGS
[0011] The accompanying drawings are included to provide a further understanding of the invention and are incorporated into and constitute a part of the present invention. The accompanying drawings illustrate embodiments of the invention and together with the description are used to explain the principles of the invention. It is understood that in order to clearly illustrate the concepts of the present invention, the accompanying drawings are not necessarily drawn to scale, and some components shown may be shown in a scale that exceeds the size in the actual embodiment.
[0012] Figure 1 is a schematic diagram of an example scenario described according to an embodiment of the present invention.
[0013] Figure 2 is a schematic diagram of an exemplary jitter enhancement method described according to an embodiment of the present invention.
[0014] Figure 3 is a schematic diagram of an exemplary time offset enhancement method described according to an embodiment of the present invention.
[0015] Figure 4 is a block diagram of an exemplary positioning system described according to an embodiment of the present invention.
[0016] Figure 5 is a flowchart of an example process described according to an embodiment of the present invention.
[0017] Figure 6 is a flowchart of an example process described according to an embodiment of the present invention.
[0018] Figure 7 is a flowchart of an example process described according to an embodiment of the present invention. DETAILED DESCRIPTION
[0019] Detailed embodiments and implementations of the claimed subject matter are disclosed herein. However, it should be understood that the disclosed embodiments and implementations are merely illustrations of the claimed subject matter, which can be implemented in various forms. However, the present invention can be implemented in many different forms and should not be construed as being limited to the exemplary embodiments and implementations set forth herein. On the contrary, these multiple exemplary embodiments and implementations are provided so that the description of the present invention is comprehensive and complete, and will fully convey the scope of the present invention to those skilled in the art. In the following description, the details of known features and technologies may be omitted to avoid unnecessarily obscuring the presented embodiments and implementations.
[0020] Overview
[0021] Embodiments of the present invention relate to various technologies, methods, schemes and / or solutions for improving positioning of devices in mobile communications with AI / ML positioning data enhancement. According to the present invention, multiple possible solutions are implemented individually or in combination. That is, although these possible solutions are described separately below, two or more of these possible solutions may also be implemented in combination or in another manner.
[0022] AI / ML positioning includes direct AI / ML positioning and AI / ML assisted positioning. The performance of these two AI / ML positioning models can be improved by collecting more training data sets. The present invention provides a solution for acquiring / generating enhanced data for AI / ML model training. Data enhancement has always been a critical task when available data is unbalanced or insufficient. Traditionally, in fields such as image recognition, different transformations have been applied to the data, such as cropping, scaling, mirroring, color enhancement, or translation. However, these algorithms cannot be directly applied to positioning. Therefore, it is necessary to provide an appropriate data enhancement solution suitable for positioning in mobile communications.
[0023] In view of the above, the present invention proposes some schemes related to data enhancement technology, which can be directly applied to the positioning of devices (e.g., user equipment or network equipment) in mobile communications. According to the scheme of the present invention, the proposed data enhancement method includes at least jitter conversion and time offset conversion, which add noise and time offset to the original input samples, respectively, to synthesize new data sets. For jitter conversion, it is assumed that the added noise conforms to a Gaussian distribution. For time offset conversion, it is assumed that the added time offset conforms to a truncated Gaussian distribution with μ=0 as the mean. Both methods modify the data input samples directly. In addition, conditional variational autoencoding (CVAE) is an AI data enhancement method that learns the probability distribution of the original data and generates new samples that follow the distribution. By applying these data enhancement methods, the size of the data set used for positioning can be increased and expanded based on the original data samples. Therefore, the positioning performance can be improved without the need to collect a large number of additional data samples.
[0024] Figure 1It is an example scenario 100 described according to an embodiment of the present invention. Scenario 100 involves at least one training entity and one data collection entity, which may be part of a wireless communication network (e.g., an LTE network, a 5G / NR network, an IoT / NB-IoT / IIoT network, a 6G network, and / or an NTN network). For example, each of the training entity and the data collection entity may include at least one of a positioning reference unit (PRU) (e.g., a device with a known location), a user equipment (UE), a base station, a location management function (LMF), a server, and a network node (e.g., a gNB). Scenario 100 shows a framework for positioning using a data enhancement method. The training entity may send a request for auxiliary data to the data collection entity. The request may include at least one of a data enhancement indicator indicating the type of data enhancement required, a required or preferred data set size, and statistical information related to the data enhancement method. For example, the data enhancement indicator may indicate which data enhancement method is preferred or adopted. If a time offset enhancement method is selected, the request may include a reference standard deviation of a time error distribution (e.g., a synchronization error and / or a UE transmit or receive (TX / RX) time error). The data collection entity may receive a request from the training entity.
[0025] The data collection entity may perform data processing upon request to generate auxiliary data. The data collection entity may send the auxiliary data to the training entity. The auxiliary data may include, but is not limited to, a set of channel delay profiles, a set of labels marking each delay profile, a set of quality indicators of delay profiles and labels, a set of enhanced training data, a set of data enhancement indicators for enhanced training data, a set of parameters or time offset values associated with enhanced training data, an actual data set size, and at least one of statistical information associated with the data enhancement method. The channel delay profile may include at least one of a channel impulse response (CIR), a power delay profile (PDP), and a reference signal received power (RSRP). The label may include location information.
[0026] The training entity may receive auxiliary data from the data collection entity. After acquiring the auxiliary data, the training entity may perform model training on the positioning model based on the auxiliary data. The positioning model may include an AI / ML positioning model. The training entity may then determine / infer location information through the positioning model. The location information may include an estimated location of the training entity and / or soft information related to the estimated location. For example, the estimated location may include coordinate information of the training entity. The soft information may include a variance or standard deviation of the estimated location.
[0027] In some embodiments, the training entity may perform data enhancement to generate more training data sets. Enhanced training data can be used to train a positioning model to improve positioning accuracy. Specifically, the training entity may have an original data input sample (i.e., an original data set). The original data input sample may include at least one CIR, PDP, and RSRP. These data input samples may be obtained by performing some measurements, or may be received from other devices. The training entity may send the data input sample to the data collection entity. The data input sample may be used by the data collection entity to generate auxiliary data. The auxiliary data may be associated with the data input sample. For example, the training entity may send the data input sample to the data collection entity and request at least one label for each data input sample (e.g., CIR, PDP, or RSRP) from the data collection entity. The auxiliary data generated by the data collection entity may be associated with the data input sample and at least one label. In another example, the training entity may obtain the label of the data input sample by itself during the training data collection phase.
[0028] After receiving the auxiliary data from the data collection entity, the training entity can perform data enhancement to generate enhanced training data based on the auxiliary data (e.g., data input samples and labels). For example, the training entity can perform data enhancement by at least one of the following methods: jitter enhancement method, time offset enhancement method, artificial intelligence enhancement method, and machine learning enhancement method. Then, the training entity can use the enhanced training data to perform model training on the positioning model. After the model training is completed, the positioning accuracy can be improved. Therefore, by using data enhancement, the training entity is able to improve the accuracy of the positioning model based on the original data input samples. The training entity does not need to collect more data input samples for model training.
[0029] In some embodiments, the data collection entity may perform data enhancement to generate more training data sets. Specifically, the data collection entity may receive raw data input samples from the training entity. The data input samples may include at least one of the following: channel impulse response, power delay spectrum, and reference signal received power. The data collection entity may also receive auxiliary data requests from the training entity. The request may include a data enhancement indicator, which may indicate a required or preferred data enhancement method, and a required or preferred data set size (e.g., the amount of data requested). The request may also include statistical information related to the enhancement method. For example, if a time offset enhancement method is selected, the request may include a reference standard deviation of a time error distribution (e.g., synchronization error and / or user equipment transmission or reception (TX / RX) time error). The data collection entity may then perform data processing to generate auxiliary data. For example, the data collection entity may add at least one label to the data input sample. The auxiliary data may be associated with the data input sample and at least one label. The data processing may include performing data enhancement and adding a data enhancement indicator to each sample to indicate whether data enhancement was performed and which data enhancement method was used. For example, the data enhancement indicator may indicate that the new data is generated by data enhancement method #N. The data enhancement method may include at least one of the following: a jitter enhancement method, a time offset enhancement method, an artificial intelligence enhancement method, and a machine learning enhancement method.
[0030] In some cases, the data collection entity may collect raw data input samples (e.g., channel impulse response CIR, power delay profile PDP, or reference signal received power RSRP) on its own. The data collection entity may perform some measurements to obtain the raw data input samples, or obtain the raw data input samples from other databases. After receiving a request from the training entity, the data collection entity may perform data processing based on the raw data input samples to generate auxiliary data. The data processing may include adding a data enhancement indicator to each sample to indicate which data enhancement method was used or whether data enhancement was performed, and / or performing data enhancement.
[0031] After the data processing is completed, the data collection entity may send at least one of the auxiliary data, the actual data set size, and statistical information related to the data enhancement method to the training entity. The training entity may decide whether to perform data enhancement and which samples to select for training based on at least one of the following: data enhancement indicators, actual data set size, and statistical information related to the data enhancement method. Therefore, by using data enhancement, the data collection entity is able to generate and provide more training data sets to the training entity based on the original data input samples. The data collection entity does not need to collect additional measurement data, such as more channel impulse responses, power delay spectra, and reference signal received powers.
[0032] In some embodiments, the training entity and the data collection entity may be implemented in the same device. The device may include at least one of the following: a positioning reference unit, a user equipment, a base station, a location management function, a server, and a network node. The communication between the training entity and the data collection entity is located inside the device. In this embodiment, the device may obtain data input samples. For example, the device may perform some measurements to obtain at least one of the following: a channel impulse response, a power delay spectrum, and a reference signal received power. The device may perform data enhancement to generate enhanced training data based on the data input samples. Data enhancement may be performed by at least one of the following: a jitter enhancement method, a time offset enhancement method, an artificial intelligence enhancement method, and a machine learning enhancement method. The device may then perform model training on the positioning model based on the enhanced training data. The device may determine the location information of the device through the positioning model. The location information may include an estimated location of the device or soft information related to the estimated location.
[0033] In some embodiments, data enhancement and model training can be performed iteratively. Specifically, the training entity or data collection entity can perform a first data enhancement based on the original data set to generate a first enhanced data set. The first enhanced data set can be used to train the positioning model. Then, the training entity or data collection entity can perform a first data enhancement based on the first enhanced data set to generate a second enhanced data set. The second enhanced data set can be used to train the positioning model again. By iterative data enhancement and model training, the performance of the positioning model can become more accurate.
[0034] In some embodiments, after model training is complete, the device can use the positioning model to infer its location. For example, the device can receive a positioning reference signal (PRS) and use the positioning model to infer the location of the device based on the positioning reference signal. In some embodiments, when the scene or position of the device changes, the device can further determine whether to request new assistance data and / or retrain the positioning model.
[0035] In the present invention, several data enhancement methods are proposed, including transformation methods and artificial intelligence methods. The transformation data enhancement techniques in the present invention take data input samples and synthesize new samples by modifying these data and applying different transformations. These transformations are applied directly to the data. Unlike transformation methods, artificial intelligence methods aim to learn the probability distribution of data to generate completely new samples in an attempt to mimic the data distribution. Artificial intelligence methods can include, for example but not limited to, conditional variational autoencoder (CVAE) methods.
[0036] Figure 2is an example jitter enhancement method 200 described in accordance with an embodiment of the present invention. The jitter enhancement method 200 may be performed by at least one of: a training entity and a data collection entity, which may be part of a wireless communication network (e.g., an LTE network, a 5G / NR network, an IoT / NB-IoT / IIoT network, a 6G network, and / or an NTN network). The jitter enhancement method includes adding noise to a channel impulse response or a power delay spectrum to perform data enhancement. In the present invention, it is assumed that the data is noisy, which is true in many cases, i.e., when processing received positioning reference signals. The jitter enhancement method attempts to exploit the noise of the data and simulate it to generate new samples. Typically, Gaussian noise is added to each sample. The mean μ and standard deviation σ of this noise define the size and shape of the deformation, so it is different in each application. The jitter process may be defined as follows,
[0037] X′(∈)={x 1 +∈ 1 , ..., x t +∈ t , ..., x T +∈ t}
[0038] Wherein, ∈ represents the noise added vector of each sample, which can be used as time domain CIR / PDP / RSRP, etc. x represents a sample consisting of CIR / PDP or RSRP of N-point fast Fourier transform (N-FFT point). According to an embodiment of the present invention, x can be measurement data (e.g., raw data input sample), and X′(∈) can be enhanced training data.
[0039] Figure 3 is an example time offset enhancement method 300 described in accordance with an embodiment of the present invention. The time offset enhancement method 300 may be performed by at least one of: a training entity and a data collection entity, which may be part of a wireless communication network (e.g., an LTE network, a 5G / NR network, an IoT / NB-IoT / IIoT network, a 6G network, and / or an NTN network). It is very reasonable that network synchronization errors or transmit / receive (TX / RX) timing errors are included in the actual received data. The time offset enhancement method attempts to utilize the timing error of the data and simulate it to generate new samples. The time offset is obtained by truncated Gaussian modeling with a mean of μ=0, a standard deviation of T1 nanoseconds (ns), and a distribution truncated range of [-2*T1, 2*T1]. The time offset process may be defined as follows,
[0040] X′(Δt)={x 1+Δt , ..., x t+Δt , ..., x T+Δt}
[0041] Wherein, Δt represents the time addition of each sample, and Δt∈[-2*T1, 2*T1]. According to an embodiment of the present invention, x may be measurement data (eg, initial data input samples), and X′(Δt) may be enhanced training data.
[0042] In some embodiments, the jitter or time offset enhancement method may need to be adjusted on a case-by-case basis, since in some cases the effect of jitter or time offset may lead to negative learning. That is, the conversion method for improved positioning performance strongly depends on the quality of the original training data set. Quality indicators of the training data set can be reported by the data generation or collection entity, and / or requested by different entities. Taking time offset as an example, in order to improve performance, the standard deviation T1 used for data enhancement should not be selected to have a value that is very different from the timing error of the original input data. Therefore, it should be selected with caution.
[0043] In some embodiments, positioning performance can be improved by data enhancement. Based on the quality indicators of the original data set, the conversion method can synthesize new samples close to the input data set by adding noise / timing errors that conform to the assumed distribution. For example, when a sampled signal is received, there will be a timing error of ±0.5 sampling intervals due to the limitation of sampling accuracy. The time offset enhancement method can add a small time offset to each original sample to expand the data set size. Due to the increase in the size of the training data set, the feature extraction of the samples during the model training process is more sufficient, which ultimately leads to improved positioning performance.
[0044] In certain embodiments, data augmentation may be used to improve the generalization performance of the positioning model. For example, when the positioning model is trained by a dataset with UE / gNB RX and TX timing error t1 (nanoseconds) and tested in a deployment scenario with UE / gNB RX and TX timing error t2 (nanoseconds), for a given t1, the positioning accuracy is better when t2 is less than t1 than when t2 is equal to t1. The positioning accuracy is worse when t2 is greater than t1 than when t2 is equal to t1. The greater the difference between t1 and t2, the greater the performance degradation. To address the problem of decreased positioning accuracy when testing the model in a deployment scenario with t2 greater than t1, a timing error of a truncated Gaussian distribution may be added to the training dataset with timing error t1 so that the timing error of the generated new data may be increased to t2. With data augmentation, in the absence of training data with timing error t2, a target dataset may still be generated to retrain the positioning model to improve performance in the deployment scenario.
[0045] The improvement in the generalization performance of the positioning model based on data augmentation is due to the fact that the signal-to-noise ratio (SNR) of the test dataset is lower than that of the training dataset, or the timing error of the test dataset is larger than that of the training dataset. In other words, the quality of the test dataset is lower than that of the training dataset. When the quality of the test dataset is higher than that of the training dataset, the generalization performance of the model can be guaranteed.
[0046] In addition to the above-mentioned transformation enhancement methods, artificial intelligence enhancement methods such as variational autoencoders (VAE) can also be used for data enhancement. VAE is a generative model that learns to generate new data by mapping the data to a low-dimensional space (encoding) and then mapping it back to the original space (decoding), while learning the distribution of latent variables. It is a neural network that uses a probabilistic approach to simulate data distribution. VAE learns to generate new data by sampling from the learned latent space. Conditional variational autoencoder (CVAE) extends VAE to a conditional generative model. This means that the generated output is conditioned on additional information such as category labels or input data. In CVAE, an additional set of conditional variables are incorporated into the encoder and decoder networks. In the present invention, the CVAE data enhancement method can be used to improve positioning performance.
[0047] In some embodiments, the training entity may estimate the position through a direct artificial intelligence or machine learning (AI / ML) positioning model. The training entity may use measurement data or raw data (e.g., channel impulse response (CIR), power delay profile (PDP), or reference signal received power (RSRP)) as input to the direct AI / ML positioning model. The direct AI / ML positioning model may directly output the estimated position. Alternatively, the training entity may estimate the position through an AI / ML assisted positioning model. The AI / ML assisted positioning model may be used to estimate time information, line-of-sight (LOS) / non-line-of-sight (NLOS) indication, time of arrival (TOA) parameter, or time difference of arrival (TDOA) parameter. The LOS indicator may indicate that the corresponding path between the training entity and the network node is LOS. The NLOS indicator may indicate that the corresponding path between the training entity and the network node is NLOS. The TOA parameter is the TOA of the path between the network node and the training entity. The training entity may use the output of the AI / ML assisted positioning model to determine the estimated position. For example, the training entity may determine the estimated position based on the LOS indicator, the location of the network node (e.g., transmittable from the network node), and the TOA parameter.
[0048] According to the solution proposed by the present invention, the training data set of the positioning model can be expanded or generated based on the original data set through a data enhancement method. Therefore, only a portion of the original data set (e.g., 5%-10%) is needed to train the positioning model and improve the accuracy of the positioning model. The workload and time of collecting the original data set can be significantly reduced. The performance of the positioning model can be improved.
[0049] Illustrative Embodiments
[0050] Figure 4 The exemplary positioning system 400 described in accordance with an embodiment of the present invention has an exemplary training entity 410 and an exemplary data collection entity 420. Each of the training entity 410 and the data collection entity 420 can perform various functions to implement the schemes, technologies, processes and methods described herein related to improving the positioning of user equipment and network devices in mobile communications through data enhancement, including the scenarios or schemes described above and the processes 500, 600 and 700 described below.
[0051] The training entity 410 or the data collection entity 420 may be part of an electronic device, which may be a UE such as a portable or mobile device, a wearable device, an in-vehicle device or vehicle, a wireless communication device, or a computing device. For example, the training entity 410 or the data collection entity 420 may be implemented in a smartphone, a smart watch, a personal digital assistant, an electronic control unit in a vehicle, a digital camera, or a computing device such as a tablet, a laptop, or a notebook computer. The training entity 410 or the data collection entity 420 may also be part of a machine-type device, which may be an IoT, NB-IoT, IIoT device such as a fixed or static device, a home device, a wired communication device, or a computing device. For example, the communication device 310 may be implemented in a smart thermostat, a smart refrigerator, a smart door lock, a wireless speaker, or a home control center. Alternatively, the training entity 410 or the data collection entity 420 may be part of a network device, which may be a network node, such as a satellite, a base station, a small base station, a router, a gateway, a positioning reference unit, a location management function, or a server. For example, the training entity 410 or the data collection entity 420 may be implemented in an eNodeB in an LTE network, a gNB in a 5G / NR, IoT, NB-IoT or IIoT network, or in a satellite or base station in a 6G network.
[0052] In some embodiments, the training entity 410 or the data collection entity 420 may be implemented in the form of one or more integrated circuit (IC) chips, such as, but not limited to, one or more single-core processors, one or more multi-core processors, one or more reduced instruction set computing (RISC) processors, or one or more complex instruction set computing (CISC) processors. The training entity 410 or the data collection entity 420 may include Figure 44, such as processor 412 or 422 and transceiver or interface 416 or 426. Transceiver or interface 416 or 426 can be used to communicate with other devices or entities. Training entity 410 or data collection entity 420 may also include one or more other components that are not related to the solution proposed by the present invention (e.g., internal power supply, display device and / or user interface device), so for the sake of brevity and simplicity, these components of training entity 410 or data collection entity 420 are not shown in FIG. Figure 4 It is not shown in the figure and is not described below.
[0053] In one aspect, each of processor 412 and processor 422 may be implemented in the form of one or more single-core processors, one or more multi-core processors, or one or more CISC or RISC processors. That is, even if the singular term "processor" is used herein to refer to processor 412 and processor 422, each of processor 412 and processor 422 may include multiple processors in some embodiments and may include a single processor in other embodiments according to the present invention. In another aspect, each of processor 412 and processor 422 may be implemented in the form of hardware (and, optionally, firmware) having electronic components, which may include, for example but not limited to, one or more transistors, one or more diodes, one or more capacitors, one or more resistors, one or more inductors, one or more memristors, and / or one or more varactors configured and arranged to achieve the specific purpose of the present invention. In other words, in accordance with various embodiments of the present invention, at least in some embodiments, each of processor 412 and processor 422 may be used as a dedicated machine specially designed, configured, and arranged to perform specific tasks including data enhancement in an apparatus (e.g., represented by training entity 410 or data collection entity 420) according to various embodiments of the present invention.
[0054] In some embodiments, the training entity 410 may also include a transceiver or interface 416 coupled to the processor 412 and capable of transmitting and receiving data. In some embodiments, the training entity 410 may also include a memory 414 connected to the processor 412 and capable of being accessed by the processor 412 and storing data therein. In some embodiments, the data collection entity 420 may also include a transceiver or interface 426 connected to the processor 422 and capable of transmitting and receiving data. In some embodiments, the data collection entity 420 may also include a memory 424 coupled to the processor 422 and capable of being accessed by the processor 422 and storing data therein. Therefore, the training entity 410 and the data collection entity 420 may communicate with each other through the transceiver or interface 416 and the transceiver or interface 426. For better understanding, the following describes the operation, functions and capabilities of the training entity 410 and the data collection entity 420, which are provided in a mobile communication environment, wherein the training entity 410 and the data collection entity 420 are implemented as a communication device, a user equipment or a network node of a communication network or as a communication device.
[0055] In some embodiments, the processor 412 may send a request for auxiliary data to the data collection entity 420 via the transceiver or interface 416. The processor 412 may receive the auxiliary data from the data collection entity 420 via the transceiver or interface 416. The processor 412 may perform model training on the positioning model based on the auxiliary data. The processor 412 may determine the location information through the positioning model.
[0056] In some embodiments, the auxiliary data includes at least one of: a set of delay spectrum of the channel, a set of labels for marking each delay spectrum, a set of quality indicators of the delay spectrum and the labels, a set of enhanced training data, a set of data enhancement indicators for the enhanced training data, a set of parameters or time offset values associated with the enhanced training data, an actual data set size, and statistical information related to the data enhancement method.
[0057] In some embodiments, the training entity 410 may include at least one of the following: a positioning reference unit (PRU), a user equipment (UE), a base station, a location management function (LMF), a server, and a network node.
[0058] In some embodiments, the request may include at least one of: a data augmentation indicator indicating the type of data augmentation required, a required or preferred data set size, and statistical information related to the data augmentation method.
[0059] In some embodiments, the processor 412 may perform data augmentation to generate augmented training data based on the auxiliary data. The processor 412 may perform model training based on the augmented training data.
[0060] In some embodiments, the processor 412 may perform data enhancement by at least one of: a jitter enhancement method, a time offset enhancement method, an artificial intelligence (AI) enhancement method, and a machine learning (ML) enhancement method. The AI enhancement method may include a conditional variational autoencoder (CVAE) method.
[0061] In some embodiments, the processor 412 may obtain the data input samples via the transceiver / interface 416. The processor 412 may send the data input samples to the data collection entity 420 via the transceiver or interface 416. Auxiliary data may be associated with the data input samples.
[0062] In some embodiments, processor 412 may request at least one label for the data input sample through transceiver or interface 416. Processor 412 may perform data augmentation to generate augmented training data based on the at least one label and the data input sample.
[0063] In some embodiments, the location information may include an estimated location of the training entity or soft information related to the estimated location.
[0064] In some embodiments, the processor 422 may receive a request for assistance data from the training entity 410 via the transceiver or interface 426. The processor 422 may generate the assistance data in accordance with the request. The processor 422 may send the assistance data to the training entity 410 via the transceiver or interface 426.
[0065] In some embodiments, the data collection entity 420 may include at least one of the following: a positioning reference unit (PRU), a user equipment (UE), a base station, a location management function (LMF), a server, and a network node.
[0066] In certain embodiments, processor 422 may perform data augmentation to generate augmented training data.
[0067] In some embodiments, the processor 422 may perform data enhancement by at least one of: a jitter enhancement method, a time offset enhancement method, an artificial intelligence (AI) enhancement method, and a machine learning (ML) enhancement method. The AI enhancement method may include a conditional variational autoencoder (CVAE) method.
[0068] In some embodiments, the processor 422 may obtain the data input sample through the transceiver or the interface 426. The processor 422 may add at least one tag to the data input sample. The auxiliary data may be associated with the data input sample and the at least one tag.
[0069] In some embodiments, the training entity 410 and the data collection entity 420 may be implemented in the same device. In some embodiments, the processor of the device may obtain data input samples. The processor may perform data enhancement to generate enhanced training data based on the data input samples. The processor may also perform model training on the positioning model based on the enhanced training data. The processor may determine the location information of the device through the positioning model.
[0070] In some embodiments, the apparatus may include at least one of: a positioning reference unit (PRU), a user equipment (UE), a base station, a location management function (LMF), a server, and a network node.
[0071] In some embodiments, the data input samples acquired by the apparatus may include at least one of the following: a channel impulse response (CIR), a power delay profile (PDP), and a reference signal received power (RSRP).
[0072] In some embodiments, the device may perform data enhancement by at least one of: a jitter enhancement method, a time offset enhancement method, an artificial intelligence (AI) enhancement method, and a machine learning (ML) enhancement method. The AI enhancement method may include a conditional variational autoencoder (CVAE) method.
[0073] In some embodiments, the location information determined by the device may include an estimated location of the device or soft information related to the estimated location.
[0074] Illustrative Process
[0075] Figure 5 It is an example process 500 shown according to an embodiment of the present invention. Whether in part or in whole, process 500 may represent an aspect of implementing the various proposed designs, concepts, schemes, systems and methods described above. More specifically, process 500 may represent an aspect of implementing the features of training entity 410. Process 500 may include one or more operations, actions or functions shown by one or more of blocks 510-540. Although the blocks shown are discrete, the blocks in process 500 may be split into more blocks, combined into fewer blocks or some blocks may be deleted depending on the desired implementation. In addition, the blocks of process 500 may be arranged according to Figure 5 The process 500 may be performed in the order shown, or, alternatively, may be performed in a different order. Process 500 may also be repeated in part or in its entirety. Training entity 410, any suitable UE, or a device of its type may implement process 500. For illustrative purposes only and not limiting the scope, process 500 is described below in the context of training entity 410. Process 500 may start at block 510.
[0076] At block 510 , flow 500 may include the training entity 410 sending a request for auxiliary data to the data collection entity. Flow 500 may proceed from block 510 to block 520 .
[0077] At block 520 , process 500 may include training entity 410 receiving auxiliary data from a data collection entity. Process 500 may proceed from block 520 to block 530 .
[0078] At block 530 , the process 500 may include the training entity 410 performing model training on the positioning model based on the auxiliary data. The process 500 may proceed from block 530 to block 540 .
[0079] At block 540 , the process 500 may include the training entity 410 determining position information via the positioning model.
[0080] In some embodiments, the process 500 may also include the training entity 410 performing data augmentation to generate augmented training data based on the auxiliary data. The process 500 may also include the training entity 410 performing model training based on the augmented training data.
[0081] In some embodiments, the process 500 may further include the training entity 410 obtaining a data input sample. The process 500 may further include the training entity 410 sending the data input sample to the data collection entity.
[0082] In some embodiments, the process 500 may further include the training entity 410 requesting at least one label for the data input sample. The process 500 may further include the training entity 410 performing data augmentation to generate augmented training data based on the at least one label and the data input sample.
[0083] Figure 6 It is an example process 600 shown according to an embodiment of the present invention. Whether in part or in whole, process 600 can represent an aspect of implementing the various proposed designs, concepts, schemes, systems and methods mentioned above. More specifically, process 600 can represent an aspect of the implementation of the features of data collection entity 420. Process 600 can include one or more operations, actions or functions shown by one or more of blocks 610-630. Although the blocks shown are discrete, depending on the desired implementation method, the blocks in process 600 can be split into more blocks, combined into fewer blocks, or some blocks can be deleted. In addition, the blocks of process 600 can be arranged according to Figure 6The process 600 may be performed in the order shown, or, alternatively, may be performed in a different order. The process 600 may also be repeated in part or in its entirety. The data collection entity 420, any suitable UE, or a device of its type may implement the process 600. For illustrative purposes only and not limiting the scope, the process 600 is described below in the context of the data collection entity 420. The process 600 may start at block 610.
[0084] At block 610 , process 600 may include data collection entity 420 receiving a request for auxiliary data from a training entity. Process 600 may proceed from block 610 to block 620 .
[0085] At block 620 , the process 600 may include the data collection entity 420 generating auxiliary data in accordance with the request. From block 620 , the process 600 may proceed to block 630 .
[0086] At block 630, the process 600 may include the data collection entity 420 sending auxiliary data to the training entity. The auxiliary data may include at least one of: a set of delay profiles of the channel, a set of labels marking each delay profile, a set of quality indicators of the delay profiles and the labels, a set of enhanced training data, a set of data enhancement indicators for the enhanced training data, a set of parameters or time offset values associated with the enhanced training data, an actual data set size, and statistical information related to the data enhancement method.
[0087] In some embodiments, the process 600 may further include the data collection entity 420 performing data augmentation to generate augmented training data.
[0088] In some embodiments, the process 600 may further include the data collection entity 420 obtaining the data input sample. The process 600 may further include the data collection entity 420 adding at least one tag to the data input sample. The auxiliary data may be associated with the data input sample and the at least one tag.
[0089] Figure 7 It is an example process 700 shown according to an embodiment of the present invention. Whether in part or in whole, process 700 may represent an aspect of implementing the various proposed designs, concepts, schemes, systems and methods described above. More specifically, process 700 may represent an aspect of the implementation of device features including a training entity 410 and a data collection entity 420. Process 700 may include one or more operations, actions or functions shown by one or more of blocks 710-740. Although the blocks shown are discrete, the blocks in process 700 may be split into more blocks, combined into fewer blocks or some blocks may be deleted depending on the desired implementation. In addition, the blocks of process 700 may be arranged according to Figure 7The process 700 may be performed in the order shown, or, alternatively, may be performed in a different order. Process 700 may also be repeated in part or in its entirety. A communication device, any suitable UE, or a device of its type may implement process 700. For illustrative purposes only and not limiting the scope, process 700 is described below in the context of a communication device. Process 700 may start at block 710.
[0090] At block 710 , process 700 may include a processor of the device acquiring a data input sample. Process 700 may proceed from block 710 to block 720 .
[0091] At block 720 , process 700 may include the processor performing data augmentation to generate augmented training data based on the data input samples. Process 700 may proceed from block 720 to block 730 .
[0092] At block 730 , the process 700 may include the processor performing model training on the positioning model based on the enhanced training data. The process 700 may proceed from block 730 to block 740 .
[0093] At block 740 , process 700 may include the processor determining location information of the device using the positioning model.
[0094] Additional Notes
[0095] The subject matter described herein sometimes shows different components contained in or connected to different other components. However, it should be understood that the multiple depicted architectures are only examples, and in fact many other architectures that implement the same function can be implemented. In a conceptual sense, any arrangement of components that implement the same function is effectively "associated", so that the desired function is realized. Therefore, regardless of the architecture or intermediate components, any two components combined to implement a specific function herein can be regarded as "associated" with each other, so that the desired function is realized. Similarly, any two components so associated can also be regarded as "operationally connected" or "operationally coupled" to each other to achieve the desired function, and any two components that can be so associated can also be regarded as "operationally connected" to each other to achieve the desired function. The specific examples of coupling in operation include but are not limited to components that can be matched and / or physically interact and / or components that can interact wirelessly and / or wirelessly interact and / or components that can interact logically and / or logically interact.
[0096] Further, with respect to any plural and / or singular terms used herein, those skilled in the art may convert from plural to singular and / or from singular to plural when appropriate for context and / or application. For the sake of clarity, various singular and plural interchanges may be explicitly set forth herein.
[0097] Furthermore, those skilled in the art will understand that, in general, the terms used herein and especially in the appended claims (e.g., the bodies of the appended claims) are generally intended to be "open" terms, e.g., the term "comprising" should be interpreted as "including but not limited to", the term "having" should be interpreted as "having at least", the term "including" should be interpreted as "including but not limited to", etc. Those skilled in the art will also understand that if a specific number of an introduced claim recitation is intended, such intent will be explicitly recited in the claim, and in the absence of such recitation, such intent is absent. For example, to aid understanding, the appended claims may contain the use of the introductory phrases "at least one" and "one or more". However, the use of such phrases should not be interpreted as implying that any particular claim included in a claim recitation through the introduction of the indefinite article "a" or "an" will be limited to embodiments containing only one such recitation, even when the same claim contains the introductory phrases "one or more" or "at least one" and an indefinite article such as "a" or "an", for example, "a and / or an" should be interpreted as meaning "at least one" or "one or more", and the same applies to the use of definite articles used to introduce claim recitations. In addition, even if a specific number of introduced claim recitations is explicitly recited, those skilled in the art will recognize that such recitation should be interpreted as meaning at least the recited number, for example, the unmasked recitation of "two recitations" means at least two recitations or two or more recitations in the absence of other modifiers. In addition, where a convention similar to "at least one of A, B, and C, etc." is used, those skilled in the art will understand that the meaning of this convention is generally intended to be interpreted as such (e.g., "a system having at least one of A, B, and C" will include but not be limited to systems having A alone, B alone, C alone, A and B together, A and C together, B and C together, and / or A, B, and C together, etc.). Where a convention similar to "at least one of A, B, or C, etc." is used, those skilled in the art will understand that the meaning of this convention is generally intended to be interpreted as such (e.g., "a system having at least one of A, B, or C" will include but not be limited to systems having A alone, B alone, C alone, A and B together, A and C together, B and C together, and / or A, B, and C together, etc.). Those skilled in the art will also understand that any transitional words and / or phrases that actually indicate two or more optional items, whether in the specification, claims, or drawings, should be understood to contemplate the possibility of including one of such items, any of such items, or both of such items. For example, the phrase "A or B" will be understood to include the possibilities of "A" or "B" or "A and B."
[0098] As can be seen from the above, it can be understood that various embodiments of the present invention have been described herein for illustrative purposes, and various modifications can be made without departing from the scope and spirit of the present invention. Therefore, the various embodiments disclosed herein are not meant to be restrictive, and the true scope and spirit are determined by the appended claims.
Claims
1. A method for improving positioning, comprising: Sending, by the training entity, a request for auxiliary data to the data collection entity; Receiving, by the training entity, the auxiliary data from the data collection entity; By means of the training entity, model training is performed on the positioning model based on the auxiliary data; as well as By means of the training entity, position information is determined via the positioning model.
2. The method for improving positioning according to claim 1, characterized in that: The auxiliary data includes at least one of the following: The delay profile set of the channel; A set of labels for marking each delay profile; The delay profile and a set of quality indicators of the label; Enhance the training data set; A set of data augmentation indicators for the augmented training data; a set of parameters or time offset values associated with the augmented training data; Actual dataset size; as well as Statistics related to data augmentation methods.
3. The method for improving positioning according to claim 1, characterized in that: The training entity includes at least one of the following: a positioning reference unit, a user equipment, a base station, a location management function, a server and a network node.
4. The method for improving positioning according to claim 1, characterized in that: The request includes at least one of the following: a data augmentation indicator indicating the type of data augmentation required, a preferred dataset size, and statistical information related to the data augmentation method.
5. The method for improving positioning according to claim 1, characterized in that: Further including: By the training entity, data augmentation is performed to generate enhanced training data based on the auxiliary data, wherein the model training is performed based on the enhanced training data.
6. The method for improving positioning according to claim 5, characterized in that: The data enhancement is performed by at least one of: a jitter enhancement method, a time offset enhancement method, an artificial intelligence enhancement method, and a machine learning enhancement method.
7. The method for improving positioning according to claim 6, characterized in that: The artificial intelligence enhancement method includes a conditional variational autoencoding method.
8. The method for improving positioning according to claim 1, characterized in that: Further including: Obtain data input samples through the training entity; as well as The data input sample is sent to the data collection entity through the training entity, wherein the auxiliary data is associated with the data input sample.
9. The method for improving positioning according to claim 8, characterized in that: Further including: requesting, by the training entity, at least one label for the data input sample; as well as Data augmentation is performed by the training entity to generate augmented training data based on the at least one label and the data input sample.
10. The method for improving positioning according to claim 1, characterized in that: The position information includes an estimated position of the training entity or soft information related to the estimated position.
11. A method for improving positioning, comprising: Receiving, by the data collection entity, a request for auxiliary data from the training entity; generating, by the data collection entity, the auxiliary data according to the request; as well as The auxiliary data is sent to the training entity by the data collection entity, wherein the auxiliary data includes at least one of the following: The delay profile set of the channel; A set of labels for marking each delay profile; The delay profile and a set of quality indicators of the label; Enhance the training data set; A set of data augmentation indicators for the augmented training data; a set of parameters or time offset values associated with the augmented training data; Actual dataset size; and Statistics related to data augmentation methods.
12. The method for improving positioning according to claim 11, characterized in that: The data collection entity includes at least one of the following: a positioning reference unit, a user equipment, a base station, a location management function, a server and a network node.
13. The method for improving positioning according to claim 11, characterized in that: Further including: Data augmentation is performed by the data collection entity to generate the augmented training data.
14. The method for improving positioning according to claim 13, characterized in that: The data enhancement is performed by at least one of: a jitter enhancement method, a time offset enhancement method, an artificial intelligence enhancement method, and a machine learning enhancement method.
15. The method for improving positioning according to claim 11, characterized in that: Further including: obtaining a sample of data input through the data collection entity; as well as At least one label is added to the data input sample by the data collection entity, wherein the auxiliary data is associated with the data input sample and the at least one label.
16. A method for improving positioning, comprising: obtaining, by a processor of the device, a data input sample; By the processor, performing data augmentation to generate augmented training data based on the data input sample; By means of the processor, model training is performed on the positioning model based on the enhanced training data; as well as The processor determines location information of the device via the positioning model.
17. The method for improving positioning according to claim 16, characterized in that: The device includes at least one of the following: a positioning reference unit, a user equipment, a base station, a location management function, a server and a network node.
18. The method for improving positioning according to claim 16, characterized in that: The data input sample includes at least one of the following: a channel impulse response, a power delay spectrum, and a reference signal received power.
19. The method for improving positioning according to claim 16, characterized in that: The data enhancement is performed by at least one of: a jitter enhancement method, a time offset enhancement method, an artificial intelligence enhancement method, and a machine learning enhancement method.
20. The method for improving positioning according to claim 16, characterized in that: The location information includes an estimated location of the device or soft information related to the estimated location.
