Device and method for automatically marking high-precision indoor positioning and determining position information

Abstract

The present invention provides a device (particularly a server device) for obtaining channel state information (CSI) of one or more links between another device and at least one access point (AP) during a training phase. The device is further configured to estimate location information of another device (particularly a mobile device) based on at least one geometric positioning technique during the training phase; generate a database including CSI of one or more links, each CSI being associated with the estimated location information. The present invention also proposes a device (particularly a mobile device) for obtaining, during a testing phase, a database from another device, wherein the database includes CSI of one or more links, each CSI being associated with the estimated location information. The device is further configured to estimate CSI of one or more links between the device and at least one AP during the testing phase; determine location information based on the database and the estimated CSI of the one or more links. Therefore, accurate and cost-effective indoor positioning can be provided.

Description

Technical Field

[0001] The present invention generally relates to the field of determining location information, and more particularly to automatically marking high-precision indoor positioning in a wireless local area network (WLAN). To this end, the present invention proposes a device, such as a server device, which is used to generate a database including channel state information (CSI) of a link between another device (e.g., a wireless device) and an access point (AP) during a training phase, and associate each CSI with location information. The present invention also proposes a device, such as a mobile device or an infrastructure device, which determines the location of the mobile device based on the estimated CSI of the link between the mobile device and one or more APs during a testing phase. Background Art

[0002] Generally, indoor positioning of mobile phones provides a new location-aware service paradigm for various scenarios and environments (e.g., offices, buildings open to the public, intelligent manufacturing, logistics, etc.). However, this new function faces two major difficulties. The first difficulty is the sub-meter accuracy required by the service. The second difficulty is the unavailability of traditional satellite-based positioning signals.

[0003] In addition, using dedicated infrastructure will reduce cost-effectiveness and may not be compatible with a large number of mobile phones. This is why relying on existing wireless communication infrastructure and standards is an ideal solution. For this solution, two major categories of indoor positioning schemes have been studied: geometric techniques and fingerprint strategies. Geometric methods use one or more techniques, such as time of arrival, time difference of arrival, direction of arrival. Geometric techniques are very sensitive to the presence and variation of the multipath distribution of the channel. This is why geometric techniques provide the best performance in line-of-sight configurations. However, this situation is rarely encountered in indoor environments. On the other hand, fingerprint techniques rely on a learned and marked wireless map, which can include different levels of detail about the channels experienced by the mobile phone: signal strength at multiple stations, signal levels on different channels, coarse or fine frequency responses in one or more channels.

[0004] Although these techniques are more adaptable to the presence of multipath, the implementation of these techniques needs to address the following two main issues:

[0005] 1. Data labeling using another ground truth positioning system: Precise locations obtained through manual measurements or another precise positioning system are required to label the training set. Both of these options imply excessive costs due to labor expenses or the existence of alternative infrastructure. These additional expenditures will reduce the cost-effectiveness of the fingerprint solution.

[0006] 2. Environmental changes and the need to update the labeled dataset: In an indoor wireless environment, the channels experienced at each location are prone to unpredictable changes. Due to the obsolescence of the labeled dataset, these changes will degrade the performance of fingerprint technology. Therefore, the set needs to be updated regularly to keep the performance of the fingerprint solution above a predefined threshold.

[0007] Although data labeling in fingerprint solutions can affect the accuracy and cost-effectiveness of the final solution, it has received little attention. Traditional fingerprint solutions inadequately consider the preparation and update processes for labeled training tests. Generally, three main strategies can be considered to handle this preparation:

[0008] 1. Manual or automatic training in a predefined set of locations: This is the basic scenario, which involves manually measuring channel metrics in a predefined set of locations and labeling these channel metrics with the associated locations. Additionally, both the initial preparation and updates require high labor costs. In recent years, with the decline in the cost of unmanned ground vehicles (UGVs), manual training can be accomplished by automatically scanning predefined locations with UGVs. The measurement data stored during the training phase can be deterministic or probabilistic. In the deterministic method, the acquired data is compressed into a single representation (e.g., the best measurement, the average of several measurements). When using a probabilistic strategy, statistical information containing the distribution of the collected metrics is stored during the training process.

[0009] 2. Manual training using reference points and enhancement schemes: In this scenario, the channels are measured at a few locations, and interpolation methods are used to label the unmeasured locations. Finally, the labeled radio map consists of two sets of points, one set where actual measurements were performed and another set that obtained its information through an interpolation scheme based on the previous points.

[0010] 3. Establishing a virtual radio map from a calibration model: This method is based on an empirical model of path loss. The parameters of this model are obtained by fitting a set of manual or automatic measurements. The model using these parameters is used to generate virtual radio maps at any resolution. Then, the nearest point method or one of its variants is used to perform positioning.

[0011] However, all these traditional scenarios require prior manual or automatic work to obtain the labeled dataset. This limits the cost-effectiveness of the related solutions. Summary of the Invention

[0012] In view of the above drawbacks, embodiments of the present invention aim to improve traditional devices and methods. The aim is to provide a device and method for precise and cost-effective indoor positioning in a WLAN. The positioning is based on a dataset that is automatically created and highly accurately marked.

[0013] The above aim is achieved by the embodiments provided in the appended independent claims. Advantageous embodiments of the embodiments are further defined in the dependent claims.

[0014] A first aspect of the present invention provides a device, particularly a server device, for use in a training phase: obtaining channel state information (CSI) of one or more links between another device and at least one access point (AP); estimating position information of the another device (particularly a mobile device) based on at least one geometric positioning technique; generating a database including the CSI of one or more links, each CSI being associated with the estimated position information.

[0015] The device of the first aspect may be a server device. The device may include circuitry. The circuitry may include hardware and software. The hardware may include analog circuitry and / or digital circuitry. In some embodiments, the circuitry includes one or more processors and a non-volatile memory connected to the one or more processors. The non-volatile memory may carry executable program code which, when executed by the one or more processors, causes the device to perform the operations or methods described herein.

[0016] The above device (e.g., a server device) obtains CSI and estimates the position information of another device (e.g., a wireless device). For example, the device may opportunistically use geometric positioning techniques that rely on time of arrival, direction of arrival, or a combination of both. For example, when the accuracy is evaluated to be below a predetermined threshold. The purpose of this evaluation is to ensure precise positioning, which will be used as a label for the collected channel state information.

[0017] In addition, the device may collect CSI and label the CSI using position information obtained by an alternative method. In addition, as long as the above evaluation is affirmative, the device may continue to update the labeled wireless map. In addition, when these conditions are not met, the device may use fingerprinting techniques (such as linear, non-linear regression, nearest neighbor, etc.) to locate the another device (e.g., a wireless device, a mobile device) based on the CSI of the another device. In some embodiments, the labeled CSI is fed into a neural network that operates in a training mode and follows an information bottleneck cost function.

[0018] The device of the first aspect supports precise and cost-effective indoor positioning in a WLAN based on a labeled dataset.

[0019] In an embodiment of the first aspect, the device is further configured to determine, during a training phase, a precision parameter for each estimated location information based on predefined parameters.

[0020] In other embodiments of the first aspect, the predefined parameters are one or more of the following:

[0021] - A predefined number of available channels;

[0022] - A high signal-to-noise ratio on a specific link between another device and the AP;

[0023] - Another device includes an alternative positioning sensor operating under optimal conditions;

[0024] - A statistical confidence metric.

[0025] In other embodiments of the first aspect, the device is further configured to: during the training phase, if the precision parameter is higher than a threshold, update the generated database, where the database is updated at a specific time or at a predetermined time interval.

[0026] In other embodiments of the first aspect, the device is further configured to train a fingerprint technique based on the generated database during the training phase.

[0027] In some embodiments, the device (e.g., a server device) may provide a geometric positioning service, which can be used to provide location information, for example, according to a request from another device. In addition, the device may also provide a labeled data set for training a neural network. The labeled data may also be fed together with a CSI set and location information, which is obtained by an existing service in a mobile device and obtained by using alternative methods such as, for example, a global positioning system (GPS), light detection and ranging (LIDAR), radio detection and ranging (RADAR), etc.

[0028] In other embodiments of the first aspect, the above fingerprint technique is based on a deep learning method, particularly based on a neural network; the device is further configured to train the neural network based on feeding CSI of one or more links to the neural network, and the CSI of the one or more links is labeled with associated location information according to the database.

[0029] In other embodiments of the first aspect, the at least one geometric positioning technique is based on one or more of the following:

[0030] - Direction of Arrival (DOA) positioning technology;

[0031] - Time Difference of Arrival (TDOA) positioning technology;

[0032] - Time of Arrival (TOA) positioning technology.

[0033] In particular, geometric positioning services for determining location information based on CSI can be provided. The service can use one or more geometric positioning technologies (such as DOA, TDOA, TOA) to obtain location information based on a CSI set.

[0034] In other embodiments of the first aspect, the above alternative positioning sensors are based on:

[0035] - GPS sensors;

[0036] - Indoor or outdoor visibility sensors.

[0037] In other embodiments of the first aspect, the above deep learning method is based on:

[0038] - Linear regression algorithm; or

[0039] - Nonlinear regression algorithm; or

[0040] - Nearest neighbor algorithm; or

[0041] - Variational autoencoder using the information bottleneck principle.

[0042] In other embodiments of the first aspect, the device is further configured to: in the test phase, obtain the CSI of one or more links related to another device; determine the quality parameter of at least one CSI; when receiving a positioning request, determine the corresponding location information according to the request based on at least one CSI and the quality parameter.

[0043] In other embodiments of the first aspect, if the quality parameter is higher than the threshold, the location information is determined based on using at least one geometric positioning technology; or, if the quality parameter is less than the threshold, the location information is determined based on the trained fingerprint technology.

[0044] In other embodiments of the first aspect, the CSI of one or more links is determined based on: estimating the channel for consecutive data packets during a predefined time interval; or determining a series of vectors corresponding to the frequency response experienced by a set of consecutive data packets of the used waveform.

[0045] In other embodiments of the first aspect, the quality parameter of at least one CSI is determined based on one or more of the following:

[0046] - Received signal strength;

[0047] - Average signal to interference plus noise ratio (SINR) of all subcarriers;

[0048] - Channel capacity;

[0049] - Effective exponential SNR mapping (EESM) with multiple input multiple output (MIMO) extension;

[0050] - Statistical confidence interval.

[0051] In other embodiments of the first aspect, the device is further configured to: in the training phase, estimate the location information of the other device based on at least one geometric positioning technique in parallel and estimate the location information of the other device based on the trained fingerprint technique, and update the generated database if the accuracy parameter of the location information estimated based on the trained fingerprint technique indicates better accuracy than the accuracy parameter of the location information estimated based on at least one geometric positioning technique.

[0052] A second aspect of the present invention provides a device, particularly a mobile device, for use in a testing phase: obtaining a database from another device, where the database includes CSIs of one or more links, each CSI being associated with estimated location information; estimating CSIs of one or more links between the device and at least one access point (AP); determining location information based on the estimated CSIs of the one or more links and the database.

[0053] The device may be a mobile device. Additionally, the device of the second aspect may include circuitry. The circuitry may include hardware and software. The hardware may include analog circuitry and / or digital circuitry. In some embodiments, the circuitry includes one or more processors and a non-volatile memory connected to the one or more processors. The non-volatile memory may carry executable program code that, when executed by the one or more processors, causes the device to perform the operations or methods described herein.

[0054] In an embodiment of the second aspect, the device is further configured to: in the testing phase, obtain a training model, particularly a trained fingerprint technique, from the other device; determine location information based on the trained fingerprint technique.

[0055] The device of the second aspect supports accurate and cost-effective indoor positioning in a WLAN based on a labeled data set.

[0056] A third aspect of the present invention provides a method for a device (especially a server device), the method comprising: in a training phase, determining the CSI of one or more links between another device and at least one access point (AP); estimating the location information of the another device (especially a mobile device) based on at least one geometric positioning technique; generating a database including the CSI of one or more links, each CSI being associated with the estimated location information.

[0057] In an embodiment of the third aspect, the method further comprises: in the training phase, determining an accuracy parameter for each estimated location information based on predefined parameters.

[0058] In other embodiments of the third aspect, the predefined parameters are one or more of the following:

[0059] - a predefined number of available channels;

[0060] - a high signal-to-noise ratio on a specific link between the another device and the AP;

[0061] - the another device includes an alternative positioning sensor operating under optimal conditions;

[0062] - a statistical confidence measure.

[0063] In other embodiments of the third aspect, the method further comprises: in the training phase, if the accuracy parameter is higher than a threshold, updating the generated database, wherein the database is updated at a specific time or at a predetermined time interval.

[0064] In other embodiments of the third aspect, the method further comprises: in the training phase, training a fingerprint technique based on the generated database.

[0065] In other embodiments of the third aspect, the above fingerprint technique is based on a deep learning method, especially a neural network; the method further comprises training the neural network based on feeding the CSI of one or more links to the neural network, and the CSI of one or more links is labeled with associated location information according to the database.

[0066] In other embodiments of the third aspect, the above at least one geometric positioning technique is based on one or more of the following:

[0067] - direction-of-arrival positioning technique;

[0068] - time-difference-of-arrival positioning technique;

[0069] - time-of-arrival positioning technique.

[0070] In other embodiments of the third aspect, the above-mentioned alternative positioning sensor is based on:

[0071] - GPS sensor;

[0072] - Indoor or outdoor visibility sensor.

[0073] In other embodiments of the third aspect, the above-mentioned deep learning method is based on:

[0074] - Linear regression algorithm; or

[0075] - Nonlinear regression algorithm; or

[0076] - Nearest neighbor algorithm; or

[0077] - Variational autoencoder using the information bottleneck principle.

[0078] In other embodiments of the third aspect, the method further includes: in the test phase, obtaining the CSI of one or more links related to another device; determining the quality parameter of at least one CSI; when receiving a positioning request, based on at least one CSI and the quality parameter, determining the corresponding location information according to the request.

[0079] In other embodiments of the third aspect, if the quality parameter is higher than the threshold, the location information is determined based on using at least one geometric positioning technique; or, if the quality parameter is less than the threshold, the location information is determined based on the trained fingerprint technique.

[0080] In other embodiments of the third aspect, the CSI of one or more links is determined based on: estimating the channel for consecutive data packets during a predefined time interval; or determining a series of vectors corresponding to the frequency response experienced by a set of consecutive data packets of the used waveform.

[0081] In other embodiments of the third aspect, the quality parameter of at least one CSI is determined based on one or more of the following:

[0082] - Received signal strength;

[0083] - Average signal-to-interference-plus-noise ratio of all subcarriers;

[0084] - Channel capacity;

[0085] - Effective exponential SNR mapping with multiple-input multiple-output extension;

[0086] - Statistical confidence interval.

[0087] In other embodiments of the third aspect, the method further includes: in the training phase, estimating the location information of the other device in parallel based on at least one geometric positioning technique and estimating the location information of the other device based on the trained fingerprint technique, and updating the generated database if the accuracy parameter of the location information estimated based on the trained fingerprint technique indicates better accuracy than the accuracy parameter of the location information estimated based on at least one geometric positioning technique.

[0088] The method of the third aspect achieves the same advantages as the device of the first aspect.

[0089] A fourth aspect of the present invention provides a method for a device (particularly a mobile device), the method including: in the testing phase, obtaining a database from another device, where the database includes CSIs of one or more links, each CSI being associated with estimated location information; estimating CSIs of one or more links between the device and at least one AP; determining location information based on the estimated CSIs of the one or more links and the database.

[0090] In an embodiment of the fourth aspect, the method further includes: in the testing phase, obtaining a training model, particularly a trained fingerprint technique, from the other device; determining location information based on the trained fingerprint technique.

[0091] The method of the fourth aspect achieves the same advantages as the device of the second aspect.

[0092] In summary, a two-part positioning technique is proposed, namely, CSI-based geometric positioning and fingerprint positioning performed by a neural network. The fingerprint technique can operate in two possible modes. In the first mode (i.e., the training mode), the fingerprint technique is fed a labeled data set including CSIs associated with corresponding location information. Additionally, in the second mode (i.e., the positioning mode), the fingerprint technique is fed a set of CSIs and location information is determined.

[0093] It should be noted that all devices, elements, units, and apparatuses described in this application can be implemented in software or hardware elements or any kind of combination thereof. All steps performed by various entities described in this application and the functions described as being performed by various entities are intended to mean that the corresponding entities are adapted or used to perform the corresponding steps and functions. Even in the following description of specific embodiments, where a specific function or step to be performed by an external entity is not reflected in the description of the specific detailed elements of the entity performing the specific step or function, it should be clear to those skilled in the art that these methods and functions can be implemented in the corresponding software or hardware elements or any kind of combination thereof. BRIEF DESCRIPTION OF THE DRAWINGS

[0094] The above aspects and embodiments of the present invention will be elaborated in the following description of specific embodiments with reference to the accompanying drawings, in which:

[0095] Figure 1 Schematically shown is a device, particularly a server device, for generating a database during a training phase according to an embodiment of the present invention.

[0096] Figure 2 Schematically shown is a device, particularly a mobile device, for determining location information during a testing phase according to an embodiment of the present invention.

[0097] Figure 3 Is a schematic diagram of a mobile device including a positioning service in a radio access network according to an embodiment of the present invention.

[0098] Figure 4a and Figure 4b Is a schematic diagram of an example of a processing method in an offline phase ( Figure 4a ) and an online phase ( Figure 4b ) according to an embodiment of the present invention.

[0099] Figure 5 Is a schematic diagram of channel state information processing.

[0100] Figure 6 Is a schematic diagram of channel state information processing in a training mode.

[0101] Figure 7 Is a flowchart of a method for a device (particularly a server device) according to an embodiment of the present invention.

[0102] Figure 8 Is a flowchart of a method for a device (particularly a mobile device) according to an embodiment of the present invention. Detailed Description of the Invention

[0103] Figure 1 Schematically shown is a device 100 (particularly a server device) for generating a database 101 during a training phase according to an embodiment of the present invention.

[0104] The device 100 can be, for example, a server device for obtaining CSI 102, 103 of one or more links 121, 131 between another device 110 and at least one AP 120, 130 during a training phase.

[0105] The device 100 is further configured to estimate location information 112, 113 of another device 110 (particularly a mobile device) based on at least one geometric positioning technique during the training phase.

[0106] Device 100 is also used to generate a database 101 during the training phase. The database 101 includes the CSI 102, 103 of one or more links 121, 131, and each CSI 102, 103 is associated with the estimated location information 112, 113.

[0107] Device 100 may include circuitry ( Figure 1 not shown in the figure). The circuitry may include hardware and software. The hardware may include analog circuitry and / or digital circuitry. In some embodiments, the circuitry includes one or more processors and a non-volatile memory connected to the one or more processors. The non-volatile memory may carry executable program code that, when executed by the one or more processors, causes the device to perform the operations or methods described herein.

[0108] Figure 2 Schematically shown is a device 110 (particularly a mobile device) for determining the location information 112, 113 during the testing phase according to an embodiment of the present invention.

[0109] Device 110 may be, for example, a mobile device for obtaining the database 101 from another device 100 during the testing phase, where the database 101 includes the CSI 102, 103 of one or more links 121, 131, and each CSI 102, 103 is associated with the estimated location information 112, 113.

[0110] Device 110 is also used to estimate the CSI 102, 103 of one or more links 121, 131 between the device 110 and at least one access point (AP) 120, 130 during the testing phase.

[0111] Device 110 is also used to determine the location information 112, 113 based on the estimated CSI 102, 103 of one or more links 121, 131 and the database 101 during the testing phase.

[0112] Device 110 may include circuitry ( Figure 2 not shown in the figure). The circuitry may include hardware and software. The hardware may include analog circuitry and / or digital circuitry. In some embodiments, the circuitry includes one or more processors and a non-volatile memory connected to the one or more processors. The non-volatile memory may carry executable program code that, when executed by the one or more processors, causes the device to perform the operations or methods described herein.

[0113] Figure 3 Schematically shown is a mobile device 110 including a positioning service in a radio access network according to an embodiment of the present invention. In Figure 3In an embodiment, the device 110 is connected to the server device 100 in a wireless network, and the server device 100 is capable of providing wireless data services.

[0114] The server device 100 also has a positioning service 301 and an application 302 that can be accessed by the mobile device 110. The server device 100 is connected to the radio access network and can obtain the channel state information 102, 103 of the links 121, 131 between the mobile device 110 and each access point (AP) 120, 130, respectively.

[0115] The channel state information 102, 103 of the APs 120, 130 (e.g., in Figure 3 the i-th channel state information is denoted as CSI i ) may include channel estimates collected for consecutive packets within a predefined duration T. In addition, using the OFDM(A) waveform, the CSI can be obtained based on a series of vectors corresponding to the frequency responses experienced by a set of consecutive packets.

[0116] Figure 4a and Figure 4b respectively show schematic diagrams of examples of the processing method 400A and the processing method 400B in the offline phase ( Figure 4a ) and the online phase ( Figure 4b ) according to an embodiment of the present invention.

[0117] According to Figure 4a and Figure 4b , the architecture is organized around two phases. In Figure 4a , during the offline phase, the labeled CSI is collected, and in Figure 4b , during the online phase, the fingerprint technique (e.g., a neural network in this article) is trained using the aforementioned labeled data to infer the location information based on the CSI.

[0118] In addition, when the traffic load is low, the network can operate in a learning mode. In this case, the CSI is collected from one or more mobile devices through several channels, regardless of whether these mobile devices request the positioning service. The network can use a resource allocation strategy that is beneficial to wide channels and trigger a channel hopping scheme if possible.

[0119] Referring to Figure 4a , the following steps can be performed, for example, by the device 100 that can be a server device and / or by the device 110 that can be a mobile device. Without limiting the present disclosure, hereinafter, the method 400A is discussed based on being performed by the device 100, and the method 400B is discussed based on being performed by the device 110.

[0120] At 401a, the device 100 obtains the CSI of different channels 121, 131.

[0121] At 402a, device 100 performs CSI post - processing, such as determining whether the quality of the CSI is good. Additionally, when the determination is "no", the device proceeds to step 403a. However, when the determination is "yes", the device proceeds to step 404a.

[0122] At step 403a, device 100 determines that the frequency hopping has ended.

[0123] At step 404a, device 100 performs bandwidth concatenation, and device 100 can provide the result to the training unit (e.g., the device can proceed to step 407a).

[0124] At step 405a, device 100 runs the time - of - arrival technique to determine location information.

[0125] At step 406a, device 100 determines the location information.

[0126] At step 407a, device 100 trains the fingerprint technique.

[0127] At step 408a, device 100 obtains the trained model.

[0128] Additionally, the quality of each set of CSI associated with each device 110 can be evaluated. Examples of the channel quality assessment process can be obtained based on various metrics that can be used individually or jointly to evaluate the quality of the CSI. For example, the following metrics can be used:

[0129] · Received signal strength (obtained from the WiFi card),

[0130] · Average power over all sub - carriers, where N is the number of sub - carriers,

[0131] · Channel capacity, e.g., according to where, σ 2 is the variance of the noise,

[0132] · Effective - exponent signal - to - noise ratio mapping (EESM) with multiple - input multiple - output extensions.

[0133] Additionally, good - quality CSI can be used to perform geometric positioning.

[0134] Additionally, method 400B can be performed by device 110.

[0135] At 401b, device 110 obtains CSI for different communication channels 121, 131.

[0136] At 402b, device 110 performs CSI post - processing.

[0137] In 403b, device 110 uses the trained model.

[0138] In 404b, device 110 determines the location information.

[0139] According to Figure 6 the flowchart shown, the obtained location information together with the CSI set is sent as training labeled data to, for example, a neural network. In addition, when the network operates in the normal mode, the location request can be processed according to Figure 5 the flowchart shown.

[0140] Figure 5 is a schematic diagram of CSI processing.

[0141] In Figure 5 process 500, for example, the following steps can be performed by device 100 which is a server device and / or by device 110 which is a mobile device. Without limiting the present disclosure, hereinafter, process 500 is discussed based on being executed by device 100.

[0142] In step 501, device 100 performs CSI quality assessment (determines the accuracy parameter). In addition, when it is determined that the quality of the CSI is medium or poor, device 100 goes to step 502, where the location is inferred using a pre-trained neural network. However, when it is determined that the quality of the CSI is good, the device goes to step 503, where geometric location techniques are used.

[0143] In addition, in step 504, the CSI is fed into the training set for subsequent update of the neural network.

[0144] For example, when the quality of the collected CSI set is good, in step 503, the location service uses one or more geometric techniques to locate mobile device 100. When a location request is received, the location service uses an interpolation scheme that uses the trained neural network to locate mobile device 110 based on the CSI set of mobile device 110.

[0145] Figure 6 shows a schematic diagram of CSI processing in the training mode.

[0146] In Figure 6 process 600, for example, the following steps can be performed by device 100 which is a server device and / or by device 110 which is a mobile device. Without limiting the present disclosure, hereinafter, process 600 is discussed based on being executed by device 100.

[0147] In step 601, device 100 determines whether an alternative method or a geometric method is available. In addition, when the determination is "yes", the device goes to step 602, however, when the determination is "no", device 100 goes to step 603.

[0148] In step 603, the device 100 trains a model (e.g., fingerprint technology). In addition, the trained model can also be used to determine location information.

[0149] In step 603, the device 100 uses the labeled data set in the database 101. In addition, the device 100 can provide the labeled data set to the model to be trained.

[0150] Figure 7 The method 700 for the device 100 (especially the server device) according to an embodiment of the present invention is shown. The method 700 includes performing the following steps in the training phase. As described above, the method 700 can be executed by the device 100.

[0151] The method 700 includes step 701: determining the CSI 102, 103 of one or more links 121, 131 between another device 110 and at least one AP 120, 130.

[0152] The method 700 further includes step 702: estimating the location information 112, 113 of another device 110 (especially the mobile device) based on at least one geometric positioning technique.

[0153] The method 700 further includes step 703: generating a database 101 including the CSI 102, 103 of one or more links 121, 131, and each CSI 102, 103 is associated with the estimated location information 112, 113.

[0154] Figure 8 The method 800 for the device 110 (especially the mobile device) according to an embodiment of the present invention is shown. The method 800 includes performing the following steps in the testing phase.

[0155] As described above, the method 800 can be executed by the device 110.

[0156] The method 800 includes step 801: obtaining the database 101 from another device 100, where the database 101 includes the CSI 102, 103 of one or more links 121, 131, and each CSI 102, 103 is associated with the estimated location information 112, 113.

[0157] The method 800 further includes step 802: estimating the CSI 102, 103 of one or more links 121, 131 between the device 110 and at least one AP 120, 130.

[0158] Method 800 further includes step 803: determining location information 112, 113 based on the estimated CSI 102, 103 of one or more links 121, 131 and the database 101.

[0159] The present invention has been described in connection with various embodiments and implementations taken as examples. However, those skilled in the art and those practicing the claimed invention can understand and make other variations by studying the drawings, this disclosure, and the independent claims. In the claims and the description, the word "comprising" does not exclude other elements or steps, and the indefinite article "a" or "an" does not exclude a plurality. A single element or other unit can implement the functions of several entities or items described in the claims. Stating certain measures in mutually different dependent claims does not indicate that a combination of these measures cannot be used in an advantageous implementation.

Claims

1. A device (100), which is a server device and is used in the training phase: Obtain the channel state information CSI (102, 103) of one or more links (121, 131) between another device (110) and at least one access point AP (120, 130); Estimate the location information (112, 113) of the another device (110) based on at least one geometric positioning technique, where the another device (110) is a mobile device; and Generate a database (101) including the CSI (102, 103) of the one or more links (121, 131), with each CSI (102, 103) associated with the estimated location information (112, 113); Among them, Before estimating the location information (112, 113) of the another device (110) based on at least one geometric positioning technique in the training phase: Determine the accuracy parameter of each estimated location information (112, 113) based on predefined parameters.

2. The device (100) according to claim 1, wherein, The predefined parameters are one or more of the following: - A predefined number of available channels; - A high signal-to-noise ratio on a specific link (121, 131) between the another device (110) and the AP (120, 130); - The another device (110) includes an alternative positioning sensor operating under optimal conditions.

3. The device (100) according to claim 1 or 2 is further used in the training phase: If the precision parameter is higher than the threshold, update the generated database (101), where The database (101) is updated at a specific time or at a predetermined time interval.

4. The device (100) according to any one of claims 1 to 3 is further used in the training phase: Train a fingerprint technique based on the generated database (101).

5. The device (100) according to claim 4, wherein: The fingerprint technique is based on a deep learning method and is based on a neural network; And The device (100) is further used to train the neural network by feeding the CSI (102, 103) of the one or more links (121, 131) to the neural network, and the CSI (102, 103) of the one or more links (121, 131) is labeled with the associated location information (112, 113) according to the database (101).

6. The device (100) according to any one of claims 1 to 5, wherein, The at least one geometric positioning technique is based on one or more of the following: - Direction of Arrival DOA positioning technique; - Time Difference of Arrival TDOA positioning technique; - Time of Arrival TOA positioning technique.

7. The device (100) according to any one of claims 2 to 6, wherein, The alternative positioning sensor is based on: - Global Positioning System GPS sensor; - Indoor or outdoor visibility sensor.

8. The device (100) according to any one of claims 5 to 7, wherein, The deep learning method is based on: - Linear regression algorithm; or - Nonlinear regression algorithm; or - Nearest neighbor algorithm; or - Variational autoencoder using the information bottleneck principle.

9. The device (100) according to any one of claims 1 to 8 is further used in the testing phase: Obtain the CSI (102, 103) of one or more links (121, 131) related to the another device (110); Determine quality parameters of the at least one CSI (102, 103); When a positioning request is received, based on the at least one CSI (102, 103) and the quality parameters, determine corresponding location information (112, 113) according to the request.

10. The apparatus (100) according to claim 9, wherein: If the quality parameter is higher than a threshold, determine the location information (112, 113) based on using the at least one geometric positioning technique; or If the quality parameter is less than the threshold, determine the location information (112, 113) based on the trained fingerprint technique.

11. The apparatus (100) according to any one of claims 1 to 10, wherein, The CSI (102, 103) of the one or more links (121, 131) is determined based on: Estimate the channel for consecutive data packets during a predefined time interval; or Determine a series of vectors corresponding to the frequency response experienced by a set of consecutive data packets of the used waveform.

12. The apparatus (100) according to any one of claims 9 to 11, wherein: The quality parameter of the at least one CSI (102, 103) is determined based on one or more of the following: - Received signal strength; - Average signal-to-interference-plus-noise ratio SINR of all subcarriers; - Channel capacity; - Effective exponential signal-to-noise ratio mapping EESM with multiple-input multiple-output MIMO extension; - Statistical confidence interval.

13. The apparatus (100) according to any one of claims 1 to 12, further configured to, during the training phase: Parallelly estimate the location information (112, 113) of the other device (110) based on the at least one geometric positioning technique and estimate the location information (112, 113) of the other device (110) based on the trained fingerprint technique, and If the accuracy parameter of the location information (112, 113) estimated based on the trained fingerprint technique indicates better accuracy than the accuracy parameter of the location information (112, 113) estimated based on the at least one geometric positioning technique, update the generated database (101).

14. A method (700) for an apparatus (100), the apparatus (100) being a server device, the method (700) comprising, during a training phase: Determine (701) the channel state information CSI (102, 103) of one or more links (121, 131) between another device (110) and at least one access point AP (120, 130); Estimate (702) the location information (112, 113) of the other device (110) based on at least one geometric positioning technique, the other device (110) being a mobile device; and Generate (703) a database (101) including the CSI (102, 103) of the one or more links (121, 131), each CSI (102, 103) being associated with the estimated location information (112, 113); Among them, Before estimating the location information (112, 113) of the other device (110) based on at least one geometric positioning technique during the training phase: Determine the accuracy parameter of each estimated location information (112, 113) based on predefined parameters.

15. A device (110), the device (110) being a mobile device, for use in a testing phase: Obtain a database (101) from another device (100), wherein, The database (101) includes CSIs (102, 103) of one or more links (121, 131), each CSI (102, 103) being associated with estimated location information (112, 113); Estimate the CSIs (102, 103) of one or more links (121, 131) between the device (110) and at least one access point AP (120, 130); and Determine the location information (112, 113) based on the estimated CSIs (102, 103) of the one or more links (121, 131) and the database (101); wherein, before the testing phase: Determine the accuracy parameter of each estimated location information (112, 113) based on predefined parameters.

16. The device (110) according to claim 15, further for use in a testing phase: Obtain a trained model from the other device, the trained model being a trained fingerprint technique; and Determine the location information (112, 113) based on the trained fingerprint technique.

17. A method (800) for a device (110), the device (110) being a mobile device, the method (800) comprising, in a testing phase: Obtain (801) a database (101) from another device (100), wherein, The database (101) includes CSIs (102, 103) of one or more links (121, 131), each CSI (102, 103) being associated with estimated location information (112, 113); Estimate (802) the CSIs (102, 103) of one or more links (121, 131) between the device (110) and at least one access point AP (120, 130); and Determine (803) the location information (112, 113) based on the estimated CSIs (102, 103) of the one or more links (121, 131) and the database (101); wherein, before the testing phase: Determine the accuracy parameter of each estimated location information (112, 113) based on predefined parameters.

Images