CN110166930A - A kind of indoor orientation method and system based on WiFi signal - Google Patents

Abstract

The invention discloses an indoor positioning method and system based on WiFi signals. The method includes: selecting a reference point in the area to be positioned to collect data information, after data preprocessing and access point screening, using the signal strength and standard deviation of the reference point corresponding to the access point as the location feature, and constructing a location fingerprint database; The position signal vector is used to calculate the characteristic Euclidean distance, and the characteristic distance and the actual physical position distance are used as weights to estimate the final position coordinates when calculating the coordinates. The system includes an offline sampling module for obtaining a location fingerprint database including reference point location coordinates and location features; an online positioning module for obtaining the location coordinates of the location to be measured by querying the location fingerprint database by collecting the location signal to be measured. The present invention fully considers the fluctuation information of the indoor WiFi signal, takes AP stability as one of the features to measure its positioning contribution, and improves the positioning accuracy and robustness of the positioning system while ensuring a small calculation amount as much as possible.

Description

A method and system for indoor positioning based on WiFi signals

technical field

The invention belongs to the technical field of communication, and more specifically, relates to an indoor positioning method and system based on WiFi signals.

Background technique

With the development of smart phones and the Internet, the use of mobile terminals to achieve location positioning has greatly facilitated people's daily life. Satellite-based GPS outdoor positioning technology has been very mature. In contrast, due to the obstruction of GPS signals by buildings and the usually complex indoor environment, GPS positioning technology is difficult to apply indoors. Usually, the existing indoor WiFi, Bluetooth and other wireless signals are selected as the signal source, and the position is estimated according to the geometric measurement method or the fingerprint positioning method.

The location fingerprint positioning algorithm based on WiFi signals is mainly divided into two stages, the offline sampling stage and the online positioning stage. The off-line sampling stage mainly completes the construction of the fingerprint database, that is, establishes the association between the physical location in the positioning area and the characteristics of the wireless signal. Using a smart phone pair, first select a series of reference points for the area division grid, and use the smart phone to collect the WiFi signal strength of each access point (AP, Access Point) that can be received at the reference point position. After processing, the signal strength (RSS, Received Signal Strength) corresponding to each reference point is obtained, and stored in the database according to a certain format as a fingerprint map. In the online positioning stage, use the mobile phone to collect the fingerprint data of the point at the location to be tested, and match it with the data in the fingerprint map through a certain matching algorithm to estimate the current location coordinates. The schematic diagram of the algorithm principle is shown in Figure 1.

In the indoor environment, wireless signals are affected by reflection, refraction, and diffraction, and the signal attenuation is irregular. In addition, the environment is complex and changeable. Human body movements and obstacles will cause fluctuations in wireless signals, making AP signals unstable. Therefore, in the offline stage, it is necessary to perform certain denoising and screening preprocessing on the signal. At present, matching algorithms mainly include deterministic algorithms such as nearest neighbor algorithm and probabilistic algorithms such as Bayesian algorithm. In recent years, algorithms such as neural network and support vector machine have also been applied in the process of indoor positioning fingerprint matching. However, the different APs of these existing WiFi fingerprint-based indoor positioning methods fluctuate at various reference points due to various factors, and the signals that are too unstable are not suitable for recording into the database as reference APs, and because the received signals at each location Different APs have different distances and positions, resulting in different strength and stability performances, so they make different contributions to the positioning algorithm.

The WiFi signal is affected by the multipath effect and the environment, so that the deterministic data in the fingerprint database cannot reflect its real distribution. To solve this problem, many methods based on probability statistics have been proposed. The Nibble system proposed by the University of California in the United States will The signal-to-noise ratio is used as a signal characteristic parameter to construct a fingerprint database; some scholars have proposed the Horus system, which uses a probability statistical model to store RSS Gaussian distribution fitting in the fingerprint map. Although these methods record the volatility of WiFi into the fingerprint library as a feature through the probability model, which more comprehensively reflects the real situation of the signal in the environment, but also increase the storage cost of the database, and also increase the complexity of the subsequent online matching algorithm. Spend.

Contents of the invention

In view of the defects of the prior art, the purpose of the present invention is to provide an indoor positioning method and system based on WiFi signals, aiming to solve the problems of unstable AP signals and complex matching algorithms in the existing positioning methods.

In order to achieve the above object, according to one aspect of the present invention, an indoor positioning method based on WiFi signals is provided, which is characterized in that it includes an offline sampling stage and an online positioning stage, wherein the offline sampling stage includes:

The area to be positioned is divided into grids, and the grid points are used as reference points, and the original data information of each AP received at each reference point is collected;

Preprocessing the raw data information to obtain the filtered data information;

The signal strength and standard deviation of each AP received at each reference point in the filtered data information are used as location features;

Establish a location fingerprint database by using the location coordinates and location features of the reference point;

The online orientation phase includes:

Collect the WiFi signals of each AP received at the location to be tested, and obtain the signal vector of the location to be tested;

Calculate the characteristic Euclidean distance between the location signal vector to be measured and the location characteristics of each reference point in the location fingerprint database;

Select the four reference points with the smallest characteristic Euclidean distance as candidate reference points;

According to the position coordinates of the candidate reference points, the position coordinates to be measured are calculated by combining the characteristic Euclidean distance and the actual physical position weighted.

Preferably, the original data information includes the coordinates of the reference point, the received name, MAC address, and signal strength of each AP.

Preferably, the preprocessing of the raw data information includes filtering out the moment when the RSS value jumps to 0, filtering out the fixed APs installed in the environment, and then sorting according to the AP signal strength from high to low, filtering out the signal strength lower than the preset Set the AP value to filter out the AP whose stability is lower than the preset value.

Preferably, the stability is the standard deviation of AP signal strength values.

Preferably, establishing the location fingerprint database by using the location coordinates of the reference point and the location characteristics includes establishing a mapping relationship between the location coordinates of the reference point and the location characteristics at the location, generating records and storing them in the database.

Preferably, the location fingerprint database includes the coordinates of the reference point, the name and MAC address of each AP, the mean value and standard deviation of the signal strength of each AP.

Preferably, assuming that the location to be tested receives signals from n APs, the signal strength at the jth AP is RSS _j . For the position of the i-th sampling point, the signal strength at the j-th AP is rss _ij . Since the performance capabilities of APs received at each location are different, each AP is given different weights according to the signal strength and stability of the AP, and then the characteristic Euclidean distance is calculated. The calculation formula is:

Among them, d _i is the weighted characteristic Euclidean distance between the location to be tested and the i-th sampling point in the location fingerprint database, w _ij is the weight of the j-th AP at the i-th sampling point, calculated according to the standard deviation of the signal strength of the AP After the contribution is made, the weight is calculated by the following formula:

After the candidate reference point is selected, calculate the sum of the physical position coordinate distances between each point of the four candidate reference points and other points. For the ith candidate reference point is dl _i , use the following formula to calculate:

Comprehensively consider the characteristic Euclidean distance d _i and the reference point physical distance dl _i as the weighting factor w _i of the i-th candidate reference point:

The final calculation of the position coordinates of the point to be measured is:

Based on the principle of a deterministic algorithm, the present invention extracts the standard deviation of WiFi signal fluctuations as a characteristic value representing its stability, and improves them in the offline sampling fingerprint library construction stage and the online positioning fingerprint matching stage respectively, and filters out fuzzy, Redundant unstable access points, and make full use of the performance information of APs at each location in the positioning stage, and calculate the characteristic Euclidean distance by weighting, so as to improve the accuracy of the positioning algorithm while saving the cost of the database as much as possible.

According to another aspect of the present invention, an indoor positioning system based on WiFi signals is provided, including:

An off-line sampling module for obtaining a location fingerprint database comprising reference point location coordinates and location features;

The online positioning module is used to obtain the coordinates of the location to be measured by collecting the signal of the location to be measured and querying the location fingerprint database.

Preferably, the off-line sampling module includes:

The collection unit is used to collect raw data information;

a screening unit, configured to screen the raw data information collected by the collection unit;

The construction unit is used for constructing a location fingerprint database by using the filtered data information.

Preferably, the online positioning module includes:

Calculation unit, used to calculate the characteristic Euclidean distance of the position to be measured;

The positioning unit is used for locating the position coordinates to be measured according to the position coordinates of the reference point.

Through the above technical solutions conceived in the present invention, compared with the prior art, the following can be obtained

Beneficial effect:

1. The present invention improves the data preprocessing algorithm in the construction phase of the offline fingerprint library, filters out AP abnormal points to obtain reliable WiFi signal strengths and stores them in the database as fingerprints, and proposes an AP selection algorithm to select reliable ones that can effectively provide positioning Refer to the access point, filter out fuzzy and unstable APs, improve positioning accuracy, and reduce the size of the location fingerprint database;

2. The present invention introduces the standard deviation as the characteristic basis for measuring the fluctuation characteristics of AP without increasing the database pressure and online matching calculation complexity as much as possible, and the later online matching algorithm comprehensively considers the strength and stability of the AP at this position as Its contribution to positioning is given different weights to calculate the feature distance according to different expressiveness;

3. When the present invention selects the reference points to estimate the coordinates, it takes into account the ambiguity of the features, and each reference position contributes differently to the coordinates of the points to be measured. Simple calculation of the coordinate mean will introduce a large error. Considering the actual physical distance of the reference points and Feature distance, a weighted algorithm is proposed to estimate the coordinates to be measured.

Description of drawings

FIG. 1 is a schematic diagram of the principle of an indoor positioning method based on WiFi signals in the prior art;

Fig. 2 is a schematic flow chart of an indoor positioning method based on WiFi signals provided by the present invention;

Fig. 3 is a schematic diagram of distribution of regional reference points in an indoor positioning method provided by an embodiment of the present invention.

Detailed ways

In order to make the object, technical solution and advantages of the present invention clearer, the present invention will be further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described here are only used to explain the present invention, not to limit the present invention. In addition, the technical features involved in the various embodiments of the present invention described below can be combined with each other as long as they do not constitute conflicts with each other.

The main solution of the present invention is to filter out APs that are not suitable for reference through the AP selection algorithm in the offline sampling stage, and use the fluctuation standard deviation of the AP as a feature to measure its stability, and store it together with the processed RSS intensity value into the corresponding The location is used as a fingerprint; in the online positioning stage, the strength and stability of different APs at each location are considered comprehensively as its contribution factor, and the weighted AP feature distance is calculated. By selecting an appropriate number of reference points for the location of the measurement point, the final analysis of the reference points Contribution degree realizes weighted coordinate estimation, as shown in Figure 2, and the specific steps are as follows:

Step S1: Divide the area to be positioned into a grid, establish a coordinate system, divide the spatial coordinates according to the grid, use the grid point as a reference point, and collect the signal strength received from each AP at each reference point multiple times value, the collected information includes reference point coordinates, AP name, MAC address, and signal strength value.

1.1 Selection of reference points for the positioning area: select the area shown in Figure 3 as the positioning area to establish a two-dimensional coordinate system, and divide it into grids according to a certain interval distance. The division density depends on the size of the scene and the capacity of the database. This example selects the reference point The point interval is 0.8m, the grid center is selected to assign position coordinates according to the coordinate system, and the grid point is used as the reference point sampling position. The reference point position information is represented by (id, x, y), where id indicates that the point is in the fingerprint database The serial number, x, y represent the position coordinates of the point in the coordinate system respectively.

1.2 Reference point raw data information collection: Personnel hold the smart device terminal, sequentially select the reference points in the previous section, use the collection software, and sample all the AP signals that can be received at each position for a period of time. , The collection frequency is related to the terminal device. In this example, the sampling frequency is set to 1 Hz, and each sampling point is continuously sampled for 10 minutes. The information obtained by signal collection includes: AP name, MAC address, and signal strength.

Step S2: preprocessing the raw data information collected in step S1. Because WiFi signals are affected by multiple factors such as environmental multipath effects, people walking, and building occlusions, the strength of collected WiFi signals fluctuates over time. It is necessary to preprocess the original data to remove the influence of noise, and calculate the location features that need to be stored in the location fingerprint database.

The specific process is as follows:

2.1 At certain moments, the signal of an AP is unstable and jumps to 0, indicating that the signal is too weak at that moment and the device cannot detect the signal, and the value at this moment is eliminated.

2.2 At a certain reference point position, for the received i-th AP, calculate the mean value of the signal strength within a period of time after filtering and record it as

2.3 Calculate the signal standard deviation σ _i of each AP at each reference point as the basis for measuring its stability. The standard deviation of the i-th AP is calculated by the following formula:

in, Indicates the jth signal strength value from the ith AP, and N _i is the number of APs.

Step S3: Use the AP selection algorithm to filter out access points that are not suitable for reference, such as personal hotspots with large fluctuations, weak signals, and instability. The specific operation steps are as follows:

3.1 First filter out some personal hotspots through the SSID name, select a fixed access network in the environment as the AP to be screened, and then sort them from strong to weak according to the signal strength value.

3.2 Set the initial AP number K value and the initial RSS intensity threshold Th _rss , that is, take the first K APs as available APs, compare the Kth AP signal with the Th _rss size, if it is greater, enter 3.3, and if it is less, enter 3.4.

3.3 Continue to compare the RSS of the k+1th AP with Th _rss , if it is greater than that, continue to compare until it is less than that and enter 3.4.

3.4 Set the fluctuation threshold Th _sd , calculate the standard deviation of all the APs selected in the previous steps, and when it is less than Th _sd , it will be the finally selected AP.

Step S4: Build a location fingerprint database: establish a mapping relationship between the location coordinates of the reference point of the positioning area and the location features at the location, generate a record, and store it in the database. Each record includes: reference position coordinates x, y, AP name, MAC address, average value and standard deviation of the signal strength of each AP after processing. The specific record format is shown in the figure. For the i-th record, the coordinates of the reference point are (X _i , Y _i ), and the fingerprints corresponding to the received n APs are:

For the same location fingerprint database, the APs that can be received by two points that are far away are not the same. If they are collected in two areas separately, it may lead to problems such as inconsistent AP collection sequence. Match the corresponding AP bands for the later real-time positioning stage. Come to the question, in order to simplify the calculation, when the acquisition software is in the offline experiment, for the newly emerging AP signal, only add information to the end of the AP file. When preprocessing the data, for the undetected data, a small value -100dBm is uniformly given. as default.

Step S5: In the online matching stage, collect the WiFi signal strength at the location to be tested, hold the smart terminal device, detect all the received AP signals in real time, compare them with the AP MAC address records saved in the location fingerprint database, and filter out Corresponding to the signal strength of the AP, it is assumed that the received signal strength at the jth AP is RSS _j . For the undetected signals of existing APs in the fingerprint library, the default signal strength value is -100dBm.

Step S6: The fingerprint of the position to be measured is compared with the records in the position fingerprint database, and the characteristic Euclidean distance between the fingerprint of the current position to be measured and each fingerprint in the position fingerprint database is calculated.

For the i-th reference point, assuming that the location fingerprint database finally selects n available APs, then for the j-th AP received, the signal strength is rss _ij . Since the performance capabilities of APs received at each location are different, each AP is given different weights according to the signal strength and stability of the AP, and then the characteristic distance is calculated. The specific calculation method is:

Among them, d _i is the weighted feature distance between the WiFi fingerprint of the location to be tested and the ith sampling point in the location fingerprint database, w _ij is the weight of the jth AP at the i location, and the contribution is calculated according to the signal strength of the AP, that is, the standard deviation After that, the weight is calculated by the following formula:

Step S7: After calculating the characteristic Euclidean distance between the location to be measured and each reference point in the location fingerprint database, sort the distances from small to large, and select the first 4 reference points as candidate reference points.

Step S8: According to the position coordinates of the selected candidate reference points, the position to be measured is estimated by weighted coordinate calculation method. First calculate the sum of physical position coordinate distances between each point in the candidate reference point and other points. For the i-th reference point is dl _i , use the following formula to calculate:

Comprehensively consider the characteristic Euclidean distance d _i and the reference point physical distance dl _i as the weighting factor w _{i of the reference point i} :

The final calculation of the position coordinates of the point to be measured is:

Those skilled in the art can easily understand that the above descriptions are only preferred embodiments of the present invention, and are not intended to limit the present invention. Any modifications, equivalent replacements and improvements made within the spirit and principles of the present invention, All should be included within the protection scope of the present invention.

Claims (10)

1. a kind of indoor location method based on WiFi signal, is characterized in that, comprises the following steps: The area to be positioned is divided into grids, and the grid points are used as reference points to collect raw data information; Preprocessing the raw data information to obtain filtered data information; Using the signal strength and standard deviation of each AP received at each reference point in the filtered data information as a location feature; establishing a location fingerprint database by using the location coordinates of the reference point and the location features; Collect the WiFi signals of each AP received at the location to be tested, and obtain the signal vector of the location to be tested; Calculating the characteristic Euclidean distance between the position signal vector to be measured and the position features of each reference point in the position fingerprint database; Selecting the four minimum reference points in the characteristic Euclidean distance as candidate reference points; The position coordinates to be measured are calculated according to the position coordinates of the candidate reference points. 2. The method according to claim 1, wherein the original data information includes the coordinates of the reference point, the name, MAC address, and signal strength of each AP received. 3. The method according to claim 1, wherein the preprocessing of the raw data information includes filtering out the moment when the RSS value jumps to 0, filtering out APs whose names are not fixed, and filtering out signal strength APs lower than the preset value, filter out APs whose stability is lower than the preset value. 4. The method according to claim 3, wherein the stability is the standard deviation of AP signal strength. 5. The method according to claim 1, wherein said establishing a location fingerprint database using said reference point location coordinates and said location features comprises establishing a mapping between said reference point location coordinates and location features at the location relation. 6. The method according to claim 5, wherein the location fingerprint database includes the location coordinates of the reference point, the name and MAC address of each AP, the mean value and standard deviation of the signal strength of each AP. 7. The method according to claim 1, wherein the calculating the position coordinates to be measured according to the position coordinates of the candidate reference points comprises calculating the candidate The weight of the reference point, weighted to estimate the coordinates of the position to be measured. 8. A system based on the indoor positioning method of the WiFi signal according to any one of claims 1 to 7, characterized in that it comprises: An off-line sampling module for obtaining a location fingerprint database comprising reference point location coordinates and location features; The online positioning module is used to obtain the coordinates of the location to be measured by collecting the signal of the location to be measured and querying the location fingerprint database. 9. system as claimed in claim 8, is characterized in that, described off-line sampling module comprises: The collection unit is used to collect raw data information; a screening unit, configured to screen the raw data information collected by the collection unit; The construction unit is used for constructing a location fingerprint database by using the filtered data information. 10. The system according to claim 8, wherein the online positioning module comprises: Calculation unit, used to calculate the characteristic Euclidean distance of the position to be measured; The positioning unit is used for locating the position coordinates to be measured according to the position coordinates of the reference point.