Indoor and outdoor seamless positioning methods, devices, and media that integrate GNSS signal characteristics and Bluetooth fingerprints

Abstract

This invention discloses a seamless indoor and outdoor positioning method, device, and medium that integrates GNSS signal features and Bluetooth fingerprints. The method includes: constructing a GNSS temporal fingerprint database; constructing a Bluetooth fingerprint database; during the positioning phase, interpolating and extracting multi-dimensional signal features matching the current timestamp from the GNSS temporal fingerprint database, and combining these features with the original GNSS temporal fingerprint database to form an interpolated GNSS temporal fingerprint database; selecting multi-dimensional signal features of satellites from the interpolated GNSS temporal fingerprint database and matching them with real-time observation features, outputting a GNSS candidate position set and a matching score; performing similarity matching between the current received signal strength value and the Bluetooth fingerprint database, outputting a Bluetooth reference position set and its matching score; and outputting the final position estimate based on the relationship between the number of observed satellites and a threshold, using a visibility weighting strategy. This invention enables seamless switching, high precision, and continuous stable indoor and outdoor positioning in various complex environments.

Description

Technical Field

[0001] This invention relates to the field of positioning and navigation technology, specifically to a seamless indoor and outdoor positioning method, device, and medium that integrates GNSS signal characteristics and Bluetooth fingerprints. Background Technology

[0002] Currently, the mainstream positioning method in outdoor environments is the Global Navigation Satellite System (GNSS), including BeiDou, GPS, and GLONASS. GNSS has the advantages of global coverage and high accuracy, but in environments such as urban canyons, tunnel entrances, and densely populated areas with tall buildings, satellite signals are easily blocked or reflected, resulting in problems such as large positioning errors, discontinuities, or even complete failure.

[0003] For indoor environments, commonly used positioning technologies include Wi-Fi fingerprinting, Bluetooth Low Energy (BLE) fingerprinting, and UWB. Among these, Bluetooth fingerprinting has gained widespread application in spaces such as shopping malls, train stations, and office buildings due to its low power consumption and ease of deployment. However, because Bluetooth fingerprinting inherently relies on manual deployment within the environment, it cannot be effectively used in open or semi-enclosed environments, and it lacks a unified integration mechanism with GNSS systems, making it difficult to achieve seamless switching and continuous tracking between indoor and outdoor environments.

[0004] Furthermore, existing GNSS and BLE positioning systems mostly employ their own independent algorithms and database management methods, lacking structured modeling and historical utilization of GNSS signal characteristics, and lacking the ability to match and predict in the time dimension; in terms of multi-source fusion, there is still a lack of a mechanism that can intelligently adjust weights according to dynamic changes in the environment, and it is also difficult to balance real-time performance and positioning stability.

[0005] Existing Bluetooth fingerprint positioning systems are difficult to integrate seamlessly with GNSS positioning systems and lack multi-source signal fusion mechanisms.

[0006] Existing Bluetooth fingerprint positioning methods are mainly used in indoor scenarios and usually operate independently of GNSS positioning systems. The lack of a unified fusion strategy makes it impossible to achieve seamless switching and continuous tracking in practical applications. In particular, when dynamically entering or exiting buildings or semi-enclosed spaces, significant positioning jumps or drifts are likely to occur.

[0007] Currently, there is a lack of effective solutions for "fingerprinting" GNSS signals themselves. Traditional GNSS positioning focuses on real-time coordinate calculation rather than structured storage and temporal modeling of signal characteristics, making it difficult to achieve time interpolation and dynamic matching based on historical features.

[0008] Existing multi-source fusion methods typically employ fixed or manually configured weighting methods, which cannot intelligently adjust the trust level of different signal sources according to changes in the actual environment, thus limiting the robustness and adaptability of fusion positioning algorithms in complex environments. Summary of the Invention

[0009] This invention addresses the shortcomings of existing technologies by providing a seamless indoor and outdoor positioning method, device, and medium that integrates GNSS signal characteristics and Bluetooth fingerprints. It can achieve deep integration of GNSS and BLE without adding extra complex hardware, and has adaptive weight adjustment and timing modeling capabilities, enabling seamless switching, high precision, and continuous and stable indoor and outdoor positioning in various complex environments.

[0010] To achieve the above objectives, the present invention adopts the following technical solution:

[0011] A seamless indoor / outdoor positioning method integrating GNSS signal characteristics and Bluetooth fingerprints includes the following steps:

[0012] Within the target area, multidimensional signal characteristics of each visible satellite are collected, and the corresponding timestamps and the actual locations of the collection points on the map are recorded to construct a GNSS time-series fingerprint database with a time structure.

[0013] Deploy Bluetooth beacons in an indoor environment, collect the received signal strength and beacon ID from each Bluetooth beacon at different reference points, and associate them with timestamps and reference point locations to build a Bluetooth fingerprint database;

[0014] During the positioning phase, multi-dimensional signal features that best match the current timestamp are interpolated from the GNSS time-series fingerprint database and extracted based on the current timestamp. These features are then combined with the original GNSS time-series fingerprint database to form an interpolated GNSS time-series fingerprint database.

[0015] For satellites in a public satellite cluster, the corresponding multidimensional signal features are selected from the interpolated GNSS time-series fingerprint database. Multidimensional similarity matching is performed between the multidimensional signal features and real-time observation features to output a set of GNSS candidate locations and a matching score.

[0016] The received signal strength value from the Bluetooth beacon is matched with the Bluetooth fingerprint database to output a set of Bluetooth reference locations and their matching scores.

[0017] If the number of observed satellites is greater than or equal to the threshold, then the top K matching scores with the highest scores are selected from the GNSS candidate location set. G The location result is obtained by taking a position-weighted average of the candidate points;

[0018] If the number of observed satellites is below the threshold but not zero, then the GNSS candidate location set and the Bluetooth reference location set are fused to output the optimal location estimate;

[0019] If the number of observed satellites is 0, select K from the Bluetooth reference position set. B Given the nearest reference point K, find this K. BThe average coordinates of the reference points are used as the positioning result.

[0020] To optimize the above technical solution, the specific measures also include:

[0021] Furthermore, the acquisition of multidimensional signal features from each visible satellite specifically involves: acquiring and recording multidimensional signal features at fixed time intervals;

[0022] The specific steps for constructing the GNSS time-series fingerprint database with a time structure are as follows:

[0023] The GNSS time-series fingerprint database is modeled in the following form:

[0024]

[0025] In the formula, For GNSS time-series fingerprint database, For each fingerprint record, the actual coordinates of the fingerprint collection point on the map are provided. For the first m A timestamp, It is the pseudo-random noise code of the m-th satellite. For the nth satellite in the nth... m A multidimensional feature vector of timestamps; , For the nth satellite in the nth... m Signal-to-noise ratio of each timestamp For the nth satellite in the nth... m The elevation angle of a timestamp For the nth satellite in the nth... m The azimuth of a timestamp, For the nth satellite in the nth... m The pseudo-distance residual of each timestamp This indicates the total number of time points recorded in the GNSS time-series fingerprint database. This indicates the number of visible satellites at each point in time.

[0026] Within each time window, mean filtering, principal component analysis, and feature stability judgment are performed on the multidimensional signal features corresponding to the same pseudo-random noise code. The feature stability judgment is as follows: if the feature standard deviation is lower than the threshold, the multidimensional signal features within that time window are considered stable fingerprints, reducing the precision of the stored data; otherwise, they are considered high dynamic regions, maintaining the precision of the original record.

[0027] The target area for feature acquisition is divided into partitions according to a preset grid size, and a spatial partition index is constructed. Within each spatial unit, the GNSS time-series fingerprint is linearly compressed over time.

[0028] The GNSS time-series fingerprint database is managed using a multi-index structure, including a first-level index, a second-level index, and a three-dimensional composite index. The first-level index is a time index, the second-level index is a satellite index, and the three-dimensional composite index is a combination of the time index, the spatial partition index, and the satellite index into a joint query key.

[0029] Furthermore, the data format of the Bluetooth fingerprint database is as follows: reference point number, reference point location coordinates, and the received signal strength from each Bluetooth beacon received by the reference point, with each Bluetooth beacon distinguished by its MAC address.

[0030] Furthermore, the step of interpolating and extracting the multidimensional signal features that best match the current timestamp from the GNSS time-series fingerprint database based on the current timestamp specifically involves:

[0031] Obtain the current GNSS observation timestamp t, and find the two records in the GNSS time series fingerprint database that are closest in time to the current observation timestamp, and denote them as timestamps. and ,satisfy For all in For common satellites observed at both timestamps, extract their values ​​at each time. The satellite's observation timestamp is estimated by taking the multidimensional feature vectors observed at two timestamps and performing linear interpolation on these two multidimensional feature vectors. Multidimensional feature vectors The complete interpolation vector is obtained by combining the interpolated GNSS multidimensional feature vectors of all interpolable satellites. Introducing interpolation confidence index This is used to quantify the impact of the time span of interpolation on accuracy.

[0032] Furthermore, the step of selecting corresponding multi-dimensional signal features from the interpolated GNSS time-series fingerprint database, performing multi-dimensional similarity matching between the multi-dimensional signal features and real-time observation features, and outputting a GNSS candidate location set and matching score specifically involves:

[0033] It has The data comes from the interpolated GNSS time series fingerprint database. GNSS fingerprint records:

[0034]

[0035] Each GNSS candidate point In addition to GNSS features, it also includes the corresponding location coordinates and confidence level, which are collectively denoted as:

[0036]

[0037] in, This indicates that the interpolated GNSS time-series fingerprint database contains data at time [time]. The A fingerprint, which contains a set of multidimensional feature vectors of all satellites at time t; This represents the actual location coordinates of the i-th fingerprint collection point on the map. This coordinate information was recorded during the fingerprint database construction phase. This represents the interpolation confidence level; the interpolation confidence level of the interpolated fingerprint is... The interpolation confidence level in the original GNSS time-series fingerprint database is considered to be 0.

[0038] Based on public satellite sets Feature alignment and matching are performed, and the common satellite set is the intersection of the currently observed set of visible satellites and the set of satellites contained in a certain GNSS fingerprint;

[0039] For any one of the public satellites The feature vector of the current observations of this satellite From the first At any moment fingerprints Extracting the multidimensional feature vector of the satellite , In the formula, Indicates the signal-to-noise ratio. Indicates the elevation angle. Indicates azimuth. Represents pseudorange residuals;

[0040] Calculate the weighted Euclidean distance of the satellite between the two eigenvectors. The formula is as follows:

[0041]

[0042] In the formula, This represents the normalized weight of the k-th dimension feature. Eigenvectors representing observations The k-th dimension feature, This represents the k-th dimension of the feature vector in a fingerprint record;

[0043] Public satellite collection It contains N satellites, and the overall feature matching distance is the average of the distances between all satellites:

[0044]

[0045] In the formula, The GNSS matching distance between the feature vector of the current observation and the i-th fingerprint sample;

[0046] A confidence penalty term is introduced to adjust the matching distance; the adjusted matching distance... The definition is as follows:

[0047]

[0048] in, As a penalty weighting coefficient, For interpolation confidence level;

[0049] The adjusted matching distance is obtained using a normalization function. Convert to candidate point GNSS matching score :

[0050]

[0051] For all GNSS candidate points Calculate its GNSS matching score Sort by GNSS matching score from highest to lowest, denoted as:

[0052]

[0053] In the formula, This represents the sorted GNSS candidate locations and their corresponding matching score sets; Indicates the first Spatial coordinates of GNSS candidate points This indicates its corresponding GNSS matching score. It represents the number of GNSS candidate points.

[0054] Furthermore, the step of performing a similarity match between the currently received signal strength value from the Bluetooth beacon and the Bluetooth fingerprint database, and outputting the Bluetooth reference location set and its matching score, specifically involves:

[0055] The received signal strengths of all Bluetooth beacons received by the receiving device at the current moment are used to form the RSSI observation vector. :

[0056]

[0057] in, This represents the received signal strength of the b-th visible beacon. The beacons are arranged in ascending order of ID and filled with missing values.

[0058] Reference samples relevant to the current scenario are selected from the Bluetooth fingerprint database to form a matching candidate set. Each Bluetooth fingerprint record in the matching candidate set includes: a reference RSSI vector. Reference point number and reference point coordinates ;

[0059] Calculate the weighted Euclidean distance between the RSSI observation vector and the reference RSSI vector:

[0060]

[0061] In the formula, This represents the b-th dimension feature of the RSSI observation vector, i.e., the... The received signal strength of a visible beacon. This represents the b-th dimension of the reference RSSI vector. Indicates the first The weighting coefficients of each visible beacon. This represents the weighted Euclidean distance between the RSSI observation vector and the j-th reference RSSI vector;

[0062] Convert distance values ​​into Bluetooth pairing scores :

[0063]

[0064] Sort Bluetooth pairing scores in descending order and filter the top K. B From several reference points, we obtain the Bluetooth reference location set and its matching confidence:

[0065]

[0066] In the formula, Sort by Bluetooth reference location and corresponding match confidence. Indicates the first The coordinates of the reference points This indicates its corresponding Bluetooth pairing score.

[0067] Furthermore, the linear interpolation of the two multidimensional feature vectors specifically involves:

[0068] Retrieve the two timestamp records that are closest in time to the current timestamp t from the GNSS fingerprint database. and , so that:

[0069]

[0070] in, The system presets a maximum interpolation window; if If the fingerprint is too sparse, the interpolation fails; otherwise, it is considered that the fingerprint is too dense. and Linear interpolation is performed on each dimension of the timestamp's multidimensional feature vector:

[0071]

[0072] In the formula, express The i-th dimension of the multidimensional feature vector of a timestamp. express The i-th dimension of the multidimensional feature vector of a timestamp. Indicates the current GNSS observation timestamp The interpolation of the i-th dimension feature of the multidimensional feature vector; the interpolation of all dimensions of features constitutes the visible satellite. Current GNSS observation timestamp Interpolation of multidimensional feature vectors , Indicates visible satellites Signal-to-noise ratio at timestamp t Indicates visible satellites At the elevation angle of timestamp t Indicates visible satellites At the azimuth angle of timestamp t, Indicates visible satellites The pseudo-range residual at timestamp t.

[0073] Furthermore, the fusion of the GNSS candidate location set and the Bluetooth reference location set to output the optimal location estimate specifically involves:

[0074] The GNSS candidate location set is represented as: Bluetooth reference location set is represented as ;

[0075] In the formula, This represents the sorted GNSS candidate locations and their corresponding matching score sets; Indicates the first Spatial coordinates of GNSS candidate points This indicates its corresponding GNSS matching score. It is the number of GNSS candidate points. Sort by Bluetooth reference location and corresponding match confidence. Indicates the first The coordinates of a Bluetooth reference point This indicates its corresponding Bluetooth pairing score. This refers to the number of Bluetooth reference points;

[0076] GNSS candidate points and Bluetooth reference points are merged into a unified candidate pool. :

[0077]

[0078] The scores of GNSS candidate points and Bluetooth reference points were normalized separately:

[0079]

[0080] In the formula, This represents the normalized GNSS matching score of the i-th GNSS candidate point. This represents the normalized Bluetooth pairing score for the j-th Bluetooth reference point;

[0081] Calculate the unified candidate pool Fusion score of candidate points for:

[0082]

[0083] in, It is the weighting factor for the score. It is the normalized GNSS matching score of the k-th candidate point in the unified candidate pool. It is the normalized Bluetooth matching score of the k-th candidate point in the unified candidate pool. If the candidate point comes from a GNSS candidate point, then... Only the GNSS matching score portion after normalization of the GNSS candidate points is retained; if the candidate point comes from a Bluetooth reference point, then... Only the normalized Bluetooth pairing score of the Bluetooth reference point is retained;

[0084] For the fusion candidate pool All candidate points are based on the fusion score Sort and select the top-rated ones. The positions are denoted as:

[0085]

[0086] In the formula, Indicates the highest-rated top A set of candidate points; This represents the position coordinates of the k-th candidate point after fusion. This represents the fusion score of the k-th candidate point after fusion.

[0087] Output the weighted average result as the final fusion position:

[0088]

[0089] In the formula, It is the optimal location estimate.

[0090] The present invention also proposes an electronic device, comprising: a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein when the processor executes the computer program, it implements the indoor and outdoor seamless positioning method described above, which integrates GNSS signal features and Bluetooth fingerprints.

[0091] The present invention also proposes a computer-readable storage medium storing a computer program that enables a computer to execute the indoor and outdoor seamless positioning method described above, which integrates GNSS signal features and Bluetooth fingerprints.

[0092] The beneficial effects of this invention are:

[0093] This invention is the first to realize fingerprint management and time-series modeling of GNSS signals. Based on the traditional GNSS system which is mainly based on real-time calculation, it expands a structured, searchable, and interpolable feature database, which greatly improves the reusability and scene adaptability of GNSS positioning in dynamic environments.

[0094] This invention proposes a time interpolation synchronization mechanism and an interpolation confidence modeling method, which effectively solves the problems of rapid changes in GNSS fingerprint features over time and severe feature mismatch, and significantly improves positioning accuracy and continuity.

[0095] This invention designs a score-driven candidate extraction and weighted fusion mechanism, which supports dynamic adjustment of the trust level of GNSS and BLE based on signal availability in different environments, achieving truly adaptive indoor and outdoor seamless positioning. Attached Figure Description

[0096] Figure 1 This is a flowchart of the indoor and outdoor seamless positioning method that integrates GNSS signal characteristics and Bluetooth fingerprints proposed in this invention. Detailed Implementation

[0097] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, and not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those of ordinary skill in the art without creative effort are within the scope of protection of this application.

[0098] Example 1

[0099] This invention proposes a seamless indoor and outdoor positioning method that integrates GNSS signal characteristics and Bluetooth fingerprints. The overall process of this method is as follows: Figure 1 As shown, it includes the following steps:

[0100] Deploy GNSS receivers (high-precision RTK receivers, low-cost GNSS modules, or GNSS chips built into smart terminals) within the target area. Data acquisition equipment is recommended to support multiple constellations (such as GPS, BDS, GLONASS, Galileo) to increase the number of visible satellites and improve anti-obstruction capabilities. Periodically collect multi-dimensional signal characteristics of each visible satellite. The sampling period is recommended to be set between 1 and 10 seconds, depending on the dynamic nature of the scene. Simultaneously record the corresponding timestamp and the actual location of the collection point on the map to construct a GNSS time-series fingerprint database with a time structure. The GNSS time-series fingerprint database is modeled in the following form:

[0101]

[0102] In the formula, For GNSS time-series fingerprint database, For each fingerprint record, the actual coordinates of the fingerprint collection point on the map are provided. For the first m A timestamp, It is the pseudo-random noise code of the m-th satellite. For the nth satellite in the nth... m A multidimensional feature vector of timestamps; , For the nth satellite in the nth... m Signal-to-noise ratio of each timestamp For the nth satellite in the nth... m The elevation angle of a timestamp For the nth satellite in the nth... m The azimuth of a timestamp, For the nth satellite in the nth... m The pseudo-distance residual of each timestamp This indicates the total number of time points recorded in the GNSS time-series fingerprint database. This represents the number of visible satellites at each time point; since the number of visible satellites may differ at different times, therefore... It is actually a variable that changes over time.

[0103] Table 1 shows an example of the data structure for the GNSS time-series fingerprint database:

[0104] Table 1

[0105]

[0106] To reduce the size of fingerprint data and avoid duplicate recording, this invention adopts the following optimization strategy:

[0107] Within each time window, mean filtering, principal component analysis, and feature stability judgment are performed on the multidimensional signal features corresponding to the same pseudo-random noise code. The feature stability judgment is as follows: if the feature standard deviation is lower than the threshold, the multidimensional signal features within that time window are considered stable fingerprints, reducing the precision of the stored data; otherwise, they are considered high dynamic regions, maintaining the precision of the original record.

[0108] The target area for feature acquisition is divided into partitions according to a preset grid size (e.g., 5m×5m), and a spatial partition index is constructed. Within each spatial unit, the GNSS time-series fingerprint is linearly compressed over time. The specific process of linear compression over time is as follows: In each spatial unit, the system arranges GNSS feature records in ascending order of time and uses a sliding window to analyze the feature change trend within a time period. If the feature change is within a preset threshold range (Euclidean distance change is less than one standard deviation) within a specified time period, the intermediate point is replaced by the linear interpolation coefficient or the average of the first and last features, thereby achieving compressed storage in the time dimension.

[0109] The GNSS time-series fingerprint database employs a multi-index structure for management, including a primary index, secondary indexes, and a three-dimensional combined index. The primary index is a time index, supporting rapid retrieval of fingerprint data within a specific time period or near the current time, achieving time proximity matching. The secondary index is a satellite index, supporting rapid querying of fingerprint features for specific satellite numbers within a given time period. The three-dimensional combined index combines the time index, spatial partition index, and satellite index into a joint query key, improving the matching speed and accuracy during positioning. Fingerprint data can also be stored as an offline build version on the terminal or downloaded and updated on demand through a background service, adapting to real-time positioning or cached positioning needs.

[0110] In environments where GNSS signals are unavailable or severely obstructed (such as large indoor venues, semi-open spaces, underground areas, etc.), this invention employs Bluetooth Low Energy (BLE) beacons as auxiliary positioning reference sources. Bluetooth beacons are deployed in the indoor environment, and the received signal strength and beacon ID from each Bluetooth beacon at different reference points are collected and associated with timestamps and reference point locations to construct a Bluetooth fingerprint database. The specific process is as follows:

[0111] Prioritize the deployment of BLE beacons in areas where GNSS positioning is unavailable or has significant errors (such as underground entrances, elevator lobbies, building mezzanines, etc.) to establish stable reference points;

[0112] To avoid RSSI feature confusion caused by uniform and symmetrical beacon arrangement, and to improve fingerprint distinguishability;

[0113] In locations with a simple spatial structure, such as corridors or large rooms, increase the number of beacons to create a distinguishable RSSI gradient;

[0114] Generally, each localization unit area (e.g., 20m×20m) receives at least 3 different beacons to form an RSSI feature vector with sufficient dimensions;

[0115] Broadcast configuration parameters:

[0116] Broadcast interval: It is recommended to set it between 100ms and 500ms;

[0117] Transmit power: can be set from -12dBm to +4dBm, depending on the openness of the scene;

[0118] Operating frequency: Based on the 2.4GHz band standard BLE protocol, using a polling broadcast mechanism;

[0119] The user terminal or positioning device performs traversal sampling within the deployment area, collecting the following RSSI feature data at each reference point:

[0120] Beacon ID: A unique identifier for the receiving BLE broadcast device;

[0121] RSSI (Received Signal Strength Indicator): Received signal strength, measured in dBm;

[0122] Sampling point location information: can be manually marked or automatically recorded in combination with preliminary GNSS positioning results;

[0123] Timestamp (Time Tag): Used to establish the correlation between RSSI and changes over time.

[0124] Because RSSI signals fluctuate significantly due to multipath interference, blockages, and pedestrian traffic, the following preprocessing operations are required:

[0125] Multiple sampling average: For each sampling point, multiple RSSI data points are continuously collected within a short time window (e.g., 5 seconds), and the average value is taken;

[0126] Outlier removal: The z-score method is used to remove mutation outliers in RSSI.

[0127] The data format of the Bluetooth fingerprint database is: reference point number, reference point location coordinates, and the received signal strength from each Bluetooth beacon received by the reference point. Each Bluetooth beacon is distinguished by its MAC address.

[0128] The data format example for the Bluetooth fingerprint database is shown in Table 2:

[0129] Table 2

[0130]

[0131] Reference: Reference point number, for example, B1, B2... represent sampling points;

[0132] X, Y: The planar coordinates of the reference point on the site map or within the area;

[0133] Each column for MAC addresses represents the MAC address of a Bluetooth beacon.

[0134] Each cell value represents the RSSI (dBm) value received from the beacon at that point (the larger the negative value, the weaker the signal).

[0135] A null value indicates that the beacon signal was not received here.

[0136] During the positioning phase, multi-dimensional signal features that best match the current timestamp are interpolated from the GNSS time-series fingerprint database and extracted based on the current timestamp. These features are then combined with the original GNSS time-series fingerprint database to form an interpolated GNSS time-series fingerprint database.

[0137] In satellite navigation scenarios, satellites are in a state of high-speed motion relative to ground receivers: orbital angular velocity of about 15° / h and linear velocity of about 3 km / s. Due to orbital motion and Earth's rotation, the elevation angle, azimuth angle, and signal-to-noise ratio of any satellite at the same station are constantly changing every minute. Multipath blockage also causes periodic fluctuations as the satellite elevation angle changes.

[0138] Therefore, GNSS signal characteristics are inherently highly time-sensitive: satellite SNR and elevation characteristics recorded at the same location within 5 minutes of each other are no longer consistent. If the "currently observed characteristics" are directly matched with the "fingerprints saved at any past time" during real-time positioning, the geometric mismatch between the stars and the sky will lead to large differences in the matched characteristics, positioning drift, and severe jumps.

[0139] To address the feature mismatch problem caused by the rapid changes in GNSS signal characteristics over time, and to avoid excessively large fingerprint database data, this invention proposes a time interpolation synchronization mechanism. This mechanism generates virtual GNSS features aligned with the current time during real-time positioning, ensuring the effectiveness of fingerprint matching. The specific implementation is as follows:

[0140] Obtain the current GNSS observation timestamp t, and find the two records in the GNSS time series fingerprint database that are closest in time to the current observation timestamp, and denote them as timestamps. and ,satisfy ;and ;in, The system presets a maximum interpolation window; if If the fingerprints are too sparse, the interpolation fails; otherwise, for all fingerprints in... For common satellites observed at both timestamps, extract their values ​​at each time. The satellite's observation timestamp is estimated by taking the multidimensional feature vectors observed at two timestamps and performing linear interpolation on these two multidimensional feature vectors. Multidimensional feature vectors The details are as follows:

[0141] right and Linear interpolation is performed on each dimension of the timestamp's multidimensional feature vector:

[0142]

[0143] In the formula, express The i-th dimension of the multidimensional feature vector of a timestamp. express The i-th dimension of the multidimensional feature vector of a timestamp. Indicates the current GNSS observation timestamp The interpolation of the i-th dimension feature of the multidimensional feature vector; the interpolation of all dimensions of features constitutes the visible satellite. Current GNSS observation timestamp Interpolation of multidimensional feature vectors , Indicates visible satellites Signal-to-noise ratio at timestamp t Indicates visible satellites At the elevation angle of timestamp t Indicates visible satellites At the azimuth angle of timestamp t, Indicates visible satellites The pseudo-range residual at timestamp t.

[0144] The complete interpolation vector is obtained by combining the interpolated GNSS multidimensional feature vectors of all interpolable satellites. :

[0145]

[0146] Indicates at time Interpolated GNSS feature vectors of all interpolable satellites, Indicates at two interpolation time points , The set of satellites that exist in the middle (i.e., the satellites that can be interpolated);

[0147] Introducing interpolation confidence index The formula used to quantify the impact of the interpolation time span on accuracy is as follows:

[0148]

[0149] The value range is controlled within [0, 0.5].

[0150] If the interpolation is successful, return [value]. , If the GNSS match fails, the subsequent positioning system will automatically ignore the GNSS match and only use BLE positioning or other modules for support.

[0151] For satellites in a public satellite constellation, corresponding multidimensional signal features are selected from the interpolated GNSS time-series fingerprint database. Multidimensional similarity matching is then performed between these multidimensional signal features and real-time observation features to output a set of GNSS candidate locations and a matching score. The specific process is as follows:

[0152] It has The data comes from the interpolated GNSS time series fingerprint database. GNSS fingerprint records:

[0153]

[0154] Each GNSS candidate point In addition to GNSS features, it also includes the corresponding location coordinates and confidence level, which are collectively denoted as:

[0155]

[0156] in, This indicates that the interpolated GNSS time-series fingerprint database contains data at time [time]. The A fingerprint, which contains a set of multidimensional feature vectors of all satellites at time t; This represents the actual location coordinates of the i-th fingerprint collection point on the map. This coordinate information was recorded during the fingerprint database construction phase. This represents the interpolation confidence level; the interpolation confidence level of the interpolated fingerprint is... The interpolation confidence level in the original GNSS time-series fingerprint database is considered to be 0.

[0157] Based on public satellite sets Feature alignment and matching are performed, and the common satellite set is the intersection of the currently observed set of visible satellites and the set of satellites contained in a certain GNSS fingerprint;

[0158] For any one of the public satellites The feature vector of the current observations of this satellite From the first At any moment fingerprints Extracting the multidimensional feature vector of the satellite , In the formula, Indicates the signal-to-noise ratio. Indicates the elevation angle. Indicates azimuth. Represents pseudorange residuals;

[0159] Calculate the weighted Euclidean distance of the satellite between the two eigenvectors. The formula is as follows:

[0160]

[0161] In the formula, This represents the normalized weight of the k-th dimension feature. Eigenvectors representing observations The k-th dimension feature, This represents the k-th dimension of the feature vector in a fingerprint record;

[0162] Public satellite collection It contains N satellites, and the overall feature matching distance is the average of the distances between all satellites:

[0163]

[0164] In the formula, The GNSS matching distance between the feature vector of the current observation and the i-th fingerprint sample is used to assess their similarity; the smaller the distance, the more similar they are.

[0165] A confidence penalty term is introduced to adjust the matching distance; the adjusted matching distance... The definition is as follows:

[0166]

[0167] in, As a penalty weighting coefficient, For interpolation confidence level;

[0168] The adjusted matching distance is obtained using a normalization function. Convert to candidate point GNSS matching score :

[0169]

[0170] When distance When it approaches 0, the score A value close to 1 indicates a complete match;

[0171] When distance When it increases, the score A value close to 0 indicates a decrease in similarity.

[0172] For all GNSS candidate points Calculate its GNSS matching score Sort by GNSS matching score from highest to lowest, denoted as:

[0173]

[0174] In the formula, This represents the sorted GNSS candidate locations and their corresponding matching score sets; Indicates the first Spatial coordinates of GNSS candidate points This indicates its corresponding GNSS matching score. It represents the number of GNSS candidate points.

[0175] The received signal strength value from the Bluetooth beacon is matched against the Bluetooth fingerprint database to output a set of Bluetooth reference locations and their matching scores; the specific process is as follows:

[0176] The received signal strengths of all Bluetooth beacons received by the receiving device at the current moment are used to form the RSSI observation vector. :

[0177]

[0178] in, This represents the received signal strength of the b-th visible beacon. The beacons are arranged in ascending order of ID and filled with missing values.

[0179] Reference samples relevant to the current scenario are selected from the Bluetooth fingerprint database to form a matching candidate set. Each Bluetooth fingerprint record in the matching candidate set includes: a reference RSSI vector. Reference point number and reference point coordinates ;

[0180] Calculate the weighted Euclidean distance between the RSSI observation vector and the reference RSSI vector:

[0181]

[0182] In the formula, This represents the b-th dimension feature of the RSSI observation vector, i.e., the... The received signal strength of a visible beacon. This represents the b-th dimension of the reference RSSI vector. Indicates the first The weighting coefficients of each visible beacon. This represents the weighted Euclidean distance between the RSSI observation vector and the j-th reference RSSI vector;

[0183] Convert distance values ​​into Bluetooth pairing scores :

[0184]

[0185] Sort Bluetooth pairing scores in descending order and filter the top K. B From several reference points, we obtain the Bluetooth reference location set and its matching confidence:

[0186]

[0187] In the formula, Sort by Bluetooth reference location and corresponding match confidence. Indicates the first The coordinates of the reference points This indicates its corresponding Bluetooth pairing score.

[0188] If the number of observed satellites is greater than or equal to the threshold, it indicates that the current GNSS observation contains a sufficient number of satellites, and the GNSS positioning result has a high confidence level. From the GNSS candidate location set, the top K locations with the highest matching scores are selected. G The location result is obtained by taking a position-weighted average of the candidate points;

[0189] In this embodiment, the first 5 candidate points are selected to form a simplified candidate set. By retaining only the top 5 high-confidence candidate points, the interference of abnormal matching on the fusion results can be effectively avoided.

[0190]

[0191] Use matching score As a weighting factor, a weighted average is calculated for all candidate locations to output the final fused location estimate:

[0192]

[0193] In the formula, This represents the final fusion location estimate.

[0194] If the number of observed satellites is below a threshold but not zero, then the GNSS candidate position set and the Bluetooth reference position set are fused to output the optimal position estimate; specifically:

[0195] The GNSS candidate location set is represented as: Bluetooth reference location set is represented as ;

[0196] In the formula, This represents the sorted GNSS candidate locations and their corresponding matching score sets; Indicates the first Spatial coordinates of GNSS candidate points This indicates its corresponding GNSS matching score. It is the number of GNSS candidate points. Sort by Bluetooth reference location and corresponding match confidence. Indicates the first The coordinates of a Bluetooth reference point This indicates its corresponding Bluetooth pairing score. This refers to the number of Bluetooth reference points;

[0197] GNSS candidate points and Bluetooth reference points are merged into a unified candidate pool. :

[0198]

[0199] The scores of GNSS candidate points and Bluetooth reference points were normalized separately:

[0200]

[0201] In the formula, This represents the normalized GNSS matching score of the i-th GNSS candidate point. This represents the normalized Bluetooth pairing score for the j-th Bluetooth reference point;

[0202] Calculate the unified candidate pool Fusion score of candidate points for:

[0203]

[0204] in, It is the weighting factor for the score. It is the normalized GNSS matching score of the k-th candidate point in the unified candidate pool. It is the normalized Bluetooth matching score of the k-th candidate point in the unified candidate pool. If the candidate point comes from a GNSS candidate point, then... Only the GNSS matching score portion after normalization of the GNSS candidate points is retained; if the candidate point comes from a Bluetooth reference point, then... Only the normalized Bluetooth pairing score of the Bluetooth reference point is retained;

[0205] For the fusion candidate pool All candidate points are based on the fusion score Sort and select the top-rated ones. The positions are denoted as:

[0206]

[0207] In the formula, Indicates the highest-rated top A set of candidate points; This represents the position coordinates of the k-th candidate point after fusion. This represents the fusion score of the k-th candidate point after fusion.

[0208] Output the weighted average result as the final fusion position:

[0209]

[0210] In the formula, It is the optimal location estimate.

[0211] If the number of observed satellites is 0, select K from the Bluetooth reference position set. B Given the nearest reference point K, find this K. B The average coordinates of three reference points are used as the positioning result. In this embodiment, three nearest neighbor reference points are selected, and the average coordinates of these three points are calculated as the positioning result.

[0212]

[0213] This represents the optimal location estimate. , and These are the coordinates of three reference points.

[0214] Example 2

[0215] This invention proposes an electronic device, comprising: a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, it implements the indoor and outdoor seamless positioning method that integrates GNSS signal features and Bluetooth fingerprints as described in Embodiment 1.

[0216] Example 3

[0217] This invention proposes a computer-readable storage medium storing a computer program that enables a computer to execute the indoor and outdoor seamless positioning method that integrates GNSS signal features and Bluetooth fingerprints as described in Embodiment 1.

[0218] In the embodiments disclosed in this application, a computer storage medium may be a tangible medium that may contain or store programs for use by or in conjunction with an instruction execution system, apparatus, or device. The computer storage medium may include, but is not limited to, electronic, magnetic, optical, electromagnetic, infrared, or semiconductor systems, apparatus, or devices, or any suitable combination of the foregoing. More specific examples of computer storage media include electrical connections based on one or more wires, portable computer disks, hard disks, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fibers, portable compact disk read-only memory (CD-ROM), optical storage devices, magnetic storage devices, or any suitable combination of the foregoing.

[0219] Those skilled in the art will recognize that the units and algorithm steps of the various examples described in conjunction with the embodiments disclosed in this application can be implemented in electronic hardware or a combination of computer software and electronic hardware. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of this application.

[0220] The above are merely preferred embodiments of the present invention. The scope of protection of the present invention is not limited to the above embodiments. All technical solutions falling within the scope of the present invention's concept are within the scope of protection of the present invention. It should be noted that for those skilled in the art, any improvements and modifications made without departing from the principles of the present invention should be considered within the scope of protection of the present invention.

Claims

1. A seamless indoor and outdoor positioning method integrating GNSS signal characteristics and Bluetooth fingerprints, characterized in that, Includes the following steps: Within the target area, multidimensional signal characteristics of each visible satellite are collected, and the corresponding timestamps and the actual locations of the collection points on the map are recorded to construct a GNSS time-series fingerprint database with a time structure. Deploy Bluetooth beacons in an indoor environment, collect the received signal strength and beacon ID from each Bluetooth beacon at different reference points, and associate them with timestamps and reference point locations to build a Bluetooth fingerprint database; During the positioning phase, multi-dimensional signal features that best match the current timestamp are interpolated from the GNSS time-series fingerprint database and extracted based on the current timestamp. These features are then combined with the original GNSS time-series fingerprint database to form an interpolated GNSS time-series fingerprint database. For satellites in a public satellite cluster, the corresponding multidimensional signal features are selected from the interpolated GNSS time-series fingerprint database. Multidimensional similarity matching is performed between the multidimensional signal features and real-time observation features to output a set of GNSS candidate locations and a matching score. The received signal strength value from the Bluetooth beacon is matched with the Bluetooth fingerprint database to output a set of Bluetooth reference locations and their matching scores. If the number of observed satellites is greater than or equal to the threshold, then the top K matching scores with the highest scores are selected from the GNSS candidate location set. G The location result is obtained by taking a position-weighted average of the candidate points; If the number of observed satellites is below a threshold but not zero, then the GNSS candidate position set and the Bluetooth reference position set are fused to output the optimal position estimate; specifically, the fusion of the GNSS candidate position set and the Bluetooth reference position set to output the optimal position estimate is as follows: The GNSS candidate location set is represented as: Bluetooth reference location set is represented as ; In the formula, This represents the sorted GNSS candidate locations and their corresponding matching score sets; Indicates the first Spatial coordinates of GNSS candidate points This indicates its corresponding GNSS matching score. It is the number of GNSS candidate points. Sort by Bluetooth reference location and corresponding match confidence. Indicates the first The coordinates of a Bluetooth reference point This indicates its corresponding Bluetooth pairing score. This refers to the number of Bluetooth reference points; GNSS candidate points and Bluetooth reference points are merged into a unified candidate pool. : ； The scores of GNSS candidate points and Bluetooth reference points were normalized separately: ； In the formula, This represents the normalized GNSS matching score of the i-th GNSS candidate point. This represents the normalized Bluetooth pairing score for the j-th Bluetooth reference point; Calculate the unified candidate pool Fusion score of candidate points for: ； in, It is the weighting factor for the score. It is the normalized GNSS matching score of the k-th candidate point in the unified candidate pool. It is the normalized Bluetooth matching score of the k-th candidate point in the unified candidate pool. If the candidate point comes from a GNSS candidate point, then... Only the GNSS matching score portion after normalization of the GNSS candidate points is retained; if the candidate point comes from a Bluetooth reference point, then... Only the normalized Bluetooth pairing score of the Bluetooth reference point is retained; For fusion candidate pool All candidate points are based on the fusion score Sort and select the top-rated ones. The positions are denoted as: ； In the formula, Indicates the highest-rated top A set of candidate points; This represents the position coordinates of the k-th candidate point after fusion. This represents the fusion score of the k-th candidate point after fusion. Output the weighted average result as the final fusion position: ； In the formula, It is the optimal location estimate; If the number of observed satellites is 0, select K from the Bluetooth reference position set. B Find the nearest reference point K. B The average coordinates of the reference points are used as the positioning result.

2. The indoor and outdoor seamless positioning method integrating GNSS signal features and Bluetooth fingerprints as described in claim 1, characterized in that, The specific method for collecting the multidimensional signal features of each visible satellite is to collect and record the multidimensional signal features at fixed time intervals. The specific steps for constructing the GNSS time-series fingerprint database with a time structure are as follows: The GNSS time-series fingerprint database is modeled in the following form: ； In the formula, For GNSS time-series fingerprint database, For each fingerprint record, the actual coordinates of the fingerprint collection point on the map are provided. For the first m A timestamp, It is the pseudo-random noise code of the m-th satellite. For the nth satellite in the nth... m A multidimensional feature vector of timestamps; , For the nth satellite in the nth... m Signal-to-noise ratio of each timestamp For the nth satellite in the nth... m The elevation angle of a timestamp For the nth satellite in the nth... m The azimuth of a timestamp, For the nth satellite in the nth... m The pseudo-distance residual of each timestamp This indicates the total number of time points recorded in the GNSS time-series fingerprint database. This indicates the number of visible satellites at each point in time. Within each time window, mean filtering, principal component analysis, and feature stability assessment are performed on the multidimensional signal features corresponding to the same pseudo-random noise code. The feature stability judgment is as follows: if the feature standard deviation is lower than the threshold, the multidimensional signal feature within the time window is a stable fingerprint, reducing the precision of the stored data; otherwise, it is regarded as a high dynamic area, maintaining the precision of the original record. The target area for feature acquisition is divided into partitions according to a preset grid size, and a spatial partition index is constructed. Within each spatial unit, the GNSS time-series fingerprint is linearly compressed over time. The GNSS time-series fingerprint database is managed using a multi-index structure, including a first-level index, a second-level index, and a three-dimensional composite index. The first-level index is a time index, the second-level index is a satellite index, and the three-dimensional composite index is a combination of the time index, the spatial partition index, and the satellite index into a joint query key.

3. The indoor and outdoor seamless positioning method integrating GNSS signal features and Bluetooth fingerprints as described in claim 1, characterized in that, The data format of the Bluetooth fingerprint database is: reference point number, reference point location coordinates, and the received signal strength from each Bluetooth beacon received by the reference point, with each Bluetooth beacon distinguished by its MAC address.

4. The indoor and outdoor seamless positioning method integrating GNSS signal features and Bluetooth fingerprints as described in claim 1, characterized in that, The specific steps of interpolating and extracting the multidimensional signal features that best match the current timestamp from the GNSS time-series fingerprint database based on the current timestamp are as follows: Obtain the current GNSS observation timestamp t, and find the two records in the GNSS time series fingerprint database that are closest in time to the current observation timestamp, and denote them as timestamps. and ,satisfy For all in For common satellites observed at both timestamps, extract their values ​​at each time. The satellite's observation timestamp is estimated by taking the multidimensional feature vectors observed at two timestamps and performing linear interpolation on these two multidimensional feature vectors. Multidimensional feature vectors The complete interpolation vector is obtained by combining the interpolated GNSS multidimensional feature vectors of all interpolable satellites. Introducing interpolation confidence index This is used to quantify the impact of the time span of interpolation on accuracy.

5. The indoor and outdoor seamless positioning method integrating GNSS signal features and Bluetooth fingerprints as described in claim 1, characterized in that, The process of selecting corresponding multidimensional signal features from the interpolated GNSS time-series fingerprint database, performing multidimensional similarity matching between the multidimensional signal features and real-time observation features, and outputting a set of GNSS candidate locations and a matching score is as follows: It has The data comes from the interpolated GNSS time series fingerprint database. GNSS fingerprint records: ； Each GNSS candidate point In addition to GNSS features, it also includes the corresponding location coordinates and confidence level, which are collectively denoted as: ； in, This indicates that the interpolated GNSS time-series fingerprint database contains data at time [time]. The A fingerprint, which contains a set of multidimensional feature vectors of all satellites at time t; This represents the actual location coordinates of the i-th fingerprint collection point on the map. This coordinate information was recorded during the fingerprint database construction phase. This represents the interpolation confidence level; the interpolation confidence level of the interpolated fingerprint is... The interpolation confidence level in the original GNSS time-series fingerprint database is considered to be 0. Based on public satellite sets Feature alignment and matching are performed, and the common satellite set is the intersection of the currently observed set of visible satellites and the set of satellites contained in a certain GNSS fingerprint; For any one of the public satellites The feature vector of the current observations of this satellite From the first At any moment fingerprints Extracting the multidimensional feature vector of the satellite , In the formula, Indicates the signal-to-noise ratio. Indicates the elevation angle. Indicates azimuth. Represents pseudorange residuals; Calculate the weighted Euclidean distance of the satellite between the two eigenvectors. The formula is as follows: ； In the formula, This represents the normalized weight of the k-th dimension feature. Eigenvectors representing observations The k-th dimension feature, This represents the k-th dimension of the feature vector in a fingerprint record; Public satellite collection It contains N satellites, and the overall feature matching distance is the average of the distances between all satellites: ； In the formula, The GNSS matching distance between the feature vector of the current observation and the i-th fingerprint sample; A confidence penalty term is introduced to adjust the matching distance; the adjusted matching distance... The definition is as follows: ； in, As a penalty weighting coefficient, For interpolation confidence level; The adjusted matching distance is obtained using a normalization function. Convert to candidate point GNSS matching score : ； For all GNSS candidate points Calculate its GNSS matching score Sort by GNSS matching score from highest to lowest, denoted as: ； In the formula, This represents the sorted GNSS candidate locations and their corresponding matching score sets; Indicates the first Spatial coordinates of GNSS candidate points This indicates its corresponding GNSS matching score. It represents the number of GNSS candidate points.

6. The indoor and outdoor seamless positioning method integrating GNSS signal features and Bluetooth fingerprints as described in claim 1, characterized in that, The step of performing a similarity match between the currently received signal strength value from the Bluetooth beacon and the Bluetooth fingerprint database, and outputting a Bluetooth reference location set and its matching score, specifically involves: The received signal strengths of all Bluetooth beacons received by the receiving device at the current moment are used to form the RSSI observation vector. : ； in, This represents the received signal strength of the b-th visible beacon. The beacons are arranged in ascending order of ID and filled with missing values. Reference samples relevant to the current scenario are selected from the Bluetooth fingerprint database to form a matching candidate set. Each Bluetooth fingerprint record in the matching candidate set includes: a reference RSSI vector. Reference point number and reference point coordinates ; Calculate the weighted Euclidean distance between the RSSI observation vector and the reference RSSI vector: ； In the formula, This represents the b-th dimension feature of the RSSI observation vector, i.e., the... The received signal strength of a visible beacon. This represents the b-th dimension of the reference RSSI vector. Indicates the first The weighting coefficients of each visible beacon. This represents the weighted Euclidean distance between the RSSI observation vector and the j-th reference RSSI vector; Convert distance values ​​into Bluetooth pairing scores : ； Sort Bluetooth pairing scores in descending order and filter the top K. B From several reference points, we obtain the Bluetooth reference location set and its matching confidence: ； In the formula, Sort by Bluetooth reference location and corresponding match confidence. Indicates the first The coordinates of the reference points This indicates its corresponding Bluetooth pairing score.

7. The indoor and outdoor seamless positioning method integrating GNSS signal features and Bluetooth fingerprints as described in claim 4, characterized in that, The linear interpolation of the two multidimensional feature vectors specifically involves: Retrieve the two timestamp records that are closest in time to the current timestamp t from the GNSS fingerprint database. and , so that: ； in, The system presets a maximum interpolation window; if If the fingerprint is too sparse, the interpolation fails; otherwise, it is considered that the fingerprint is too dense. and Linear interpolation is performed on each dimension of the timestamp's multidimensional feature vector: ； In the formula, express The i-th dimension of the multidimensional feature vector of a timestamp. express The i-th dimension of the multidimensional feature vector of a timestamp. Indicates the current GNSS observation timestamp The interpolation of the i-th dimension feature of the multidimensional feature vector; the interpolation of all dimensions of features constitutes the visible satellite. Current GNSS observation timestamp Interpolation of multidimensional feature vectors , Indicates visible satellites Signal-to-noise ratio at timestamp t Indicates visible satellites At the altitude angle of timestamp t Indicates visible satellites At the azimuth angle of timestamp t, Indicates visible satellites The pseudo-range residual at timestamp t.

8. An electronic device, characterized in that, include: The memory, the processor, and the computer program stored in the memory and executable on the processor, wherein when the processor executes the computer program, it implements the indoor and outdoor seamless positioning method that integrates GNSS signal features and Bluetooth fingerprints as described in any one of claims 1-7.

9. A computer-readable storage medium storing a computer program, characterized in that, The computer program enables the computer to execute the indoor and outdoor seamless positioning method as described in any one of claims 1-7.

Images