Wireless fingerprint intelligent positioning method based on adaptive WKNN

Abstract

The application discloses a wireless fingerprint intelligent positioning method based on adaptive WKNN. The method comprises the following steps: firstly, a positioning area division method based on greedy clustering is adopted, the number of sub-areas and the area center are determined by using the position distance and feature distance relationship between reference points, and the sub-area division in the offline stage is completed; secondly, the number and set of the nearest neighbor reference points are adaptively determined according to the relationship between the position distance and the feature distance; finally, a positioning method based on statistical law correction is adopted to correct the weight of each nearest neighbor reference point, and then the positioning information of a terminal to be positioned is obtained. By using the technical scheme of the application, the positioning accuracy in the online stage is effectively improved, and the complexity of the position matching in the online positioning stage is reduced.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of intelligent positioning, in particular to a wireless fingerprint intelligent positioning method based on adaptive WKNN. BACKGROUND

[0002] With the continuous development of information technology, the intelligent technology industry is also developing, gradually moving towards the door of Internet of Things. Position is an important information, and the positioning service based on positioning technology is also increasingly concerned. It has penetrated into human daily life and is an indispensable application for human beings. It can help us better manage our behavior and improve the quality of life, and also provide us with more convenience. Due to its potential great practical value, countries around the world are continuously investing in it, and many mature positioning technologies have also begun to be widely used. Positioning technology has a wide range of applications in fields such as navigation, aviation, military, and natural disaster prevention, such as maritime search and rescue, positioning and tracking of unmanned aerial vehicles, precise position calibration of aerial vehicles, and automatic navigation of marine vessels. In daily life, positioning technology is also widely used, such as personnel search, positioning search, route planning, vehicle navigation, and the like, making it more convenient for people to travel.

[0003] Positioning is achieved by periodically sending signals by mobile terminal devices or tag wristbands carried by personnel in a network-covered place. After the base station receives the signal, the signal is transmitted to the designated positioning server, and the positioning server runs the program algorithm to determine the position of the personnel. The current positioning methods include centroid positioning method, multilateral positioning method, fingerprint positioning method, etc., but these methods have more or less shortcomings, such as the centroid positioning method, which requires dense base station deployment to achieve high-precision positioning, resulting in huge economic costs. Therefore, the fingerprint positioning method is generally used. The fingerprint positioning method includes two stages: one is the offline stage, in which a number of reference points are set in the positioning area. The reference point is a specific crowd in the fingerprint database in the fingerprint matching algorithm, and the signal characteristic value such as RSSI, LQI, CQI, etc. is collected at each reference point. The signal characteristics are the fingerprint samples in the fingerprint database of the fingerprint matching algorithm, and all the fingerprint samples collected at the reference points constitute the high-quality fingerprint database required by the fingerprint matching positioning algorithm. The second stage is the online stage, in which the terminal to be positioned collects the characteristic information at the current position to form real-time fingerprints, and then matches the fingerprints in the offline stage fingerprint database through the designed fingerprint matching algorithm to obtain the final position estimation result.

[0004] So far, a lot of progress has been made in the field of positioning by applying different signal features, and the use of sub-area division to improve the efficiency of fingerprint matching in the online stage has become a hot research topic. The key technology is to divide the positioning area into multiple sub-areas according to the collected training samples, and to obtain the final positioning coordinates by taking the signal received by the terminal as the input, so as to solve the problem of online positioning and offline data fingerprint library matching redundancy in different scenarios. At the same time, how to improve the matching accuracy in the online stage is also a difficult problem to be solved. SUMMARY

[0005] The purpose of the application is to solve the problems existing in the prior art, and provide a wireless fingerprint intelligent positioning method based on adaptive WKNN.

[0006] Technical scheme: The application provides a wireless fingerprint intelligent positioning method based on adaptive WKNN. The method comprises the following steps:

[0007] dividing the to-be-positioned area into a plurality of sub-areas;

[0008] determining the number and set of the nearest neighbor reference points of the to-be-positioned point;

[0009] using a positioning method based on statistical rule correction to correct the weight of each nearest neighbor reference point, and then obtaining the positioning information of the to-be-positioned point.

[0010] Further, a positioning area division method based on greedy clustering is used to divide the to-be-positioned area, and the specific steps comprise:

[0011] Step 1: determining the initial area center C1 as the reference point with the maximum average position distance among all reference points, wherein the position distance of the i-th reference point is dis_i i ={dis_i i,1 ,dis_i i,2 ,...,dis_i i,N}, dis_i i,j is the position distance between the i-th reference point and the j-th reference point, and N is the total number of reference points in the fingerprint library in the positioning area;

[0012] Step 2: determining the coverage radius p1 of the sub-area 1 where the initial area center C1 is located:

[0013]

[0014] wherein, is the set of feature distances between C1 and the remaining reference points in the fingerprint library, is the feature distance between C1 and the a-th reference point except itself in the fingerprint library, RSSC1,m RSSm represents the signal eigenvalue of the mth base station received by C1 a,m RSSma represents the signal eigenvalue of the mth base station received by the ath reference point, M is the total number of base stations participating in positioning;

[0015] Step 3: Perform the next sub-area division:

[0016] For sub-area c, if the number of sub-areas currently divided is 1, find the point farthest from the position of C1 among the remaining reference points belonging to the sub-area as the regional center C of sub-area c c ; if the number of sub-areas currently divided is greater than 1, find the point farthest from the average position distance of the regional centers of all divided sub-areas among the remaining reference points as the regional center C of sub-area c c ; calculate the characteristic distance between C c and the remaining reference points, and determine the coverage radius p c of sub-area c;

[0017] Step 4: Repeat step 3 until all reference points are divided into their respective sub-areas;

[0018] Step 5: Update the regional center of each sub-area to the average of the fingerprint information of all reference points in each sub-area.

[0019] Further, before step 5, it is also necessary to determine whether there is a reference point that is repeatedly divided into its respective sub-area, if so, further calculate the characteristic distance between the reference point and the regional center of each sub-area, and select the sub-area corresponding to the regional center with the closest characteristic distance as the sub-area to which the reference point belongs.

[0020] Further, the specific steps for determining the number and set of nearest neighbor reference points include:

[0021] Step 1: Calculate the similarity of the signal values collected at the to-be-positioned point and the signal values collected at the regional centers of each sub-area, and select the sub-area corresponding to the maximum similarity as the belonging area of the to-be-positioned point;

[0022] Step 2: Calculate the base station similarity and Sorenson distance between the signal values collected at the to-be-positioned point and the signal values collected at each reference point in the belonging area, multiply the two to obtain the weighted characteristic distance between each reference point in the belonging area and the to-be-positioned point, and record the reference point corresponding to the minimum weighted characteristic distance as F;

[0023] Step 3: Sort all reference points in the belonging area according to the position coordinate distance from F from small to large to form a set B_Loc;

[0024] Step 4: Sort the reference points in the home area according to the corresponding weighted feature distance from small to large, and take the first n WFD reference points with the weighted feature distance lower than the set threshold value to construct a new feature distance set NA_Fea;

[0025] Step 5: Take the first n WFD reference points in the set B_Loc to construct a new coordinate set NB_Loc;

[0026] Step 6: The intersection of NA_Fea and NB_Loc is the nearest neighbor reference point set, and the number of common reference points of NA_Fea and NB_Loc is the number of nearest neighbor reference points.

[0027] Further, the set threshold value is (the maximum weighted feature distance + the minimum weighted feature distance) / 2.

[0028] Further, the calculation method of the similarity of the signal value collected at the to-be-positioned point and the signal value collected at the center of each sub-area is:

[0029]

[0030] wherein, RSS on,m denotes the signal value of the mth base station collected at the center of the sub-area c, and M denotes the number of base stations.

[0031] Further, in step 2, the calculation method of the base station similarity of the signal value collected at the to-be-positioned point and the signal value collected at each reference point in the home area is:

[0032]

[0033] wherein, SIM AP (u, s i-on ) denotes the base station similarity between the signal value collected at the to-be-positioned point and the ith-on reference point in the home area, U on denotes the set of base stations corresponding to the signal value collected at the to-be-positioned point, and s i-on denotes the set of base stations corresponding to the signal value collected at the ith-on reference point in the home area.

[0034] The calculation method of the Sorensen distance of the signal value collected at the to-be-positioned point and the signal value collected at each reference point in the home area is:

[0035]

[0036] wherein, RSS i-on,mrepresents the signal value of the mth base station collected at the ith reference point in the home area, and M represents the number of base stations;

[0037] The calculation method of the weighted feature distance between each reference point in the home area and the terminal to be positioned is as follows:

[0038]

[0039] Further, the positioning method based on the correction of the statistical law is used to correct the weight of each adjacent reference point, and then the positioning information of the terminal to be positioned is obtained, and the specific process includes:

[0040] Step S1, the first reference point rp1 in the nearest neighbor reference point set RPc={rp1, rp2,..., rp k} is recorded as the reference point, the weighted feature distance and the position distance between the reference point rp1 and other nearest neighbor reference points are calculated, the weighted feature distance and the position distance are sorted from small to large, the weighted feature distance set Sor-on=[Sor1, Sor2,..., Sor k-1 ] and the position distance set Loc-on=[Loc1, Loc2,..., Loc k-1 ] are obtained;

[0041] Step S2, the offset degree between the weighted feature distance set Sor-on and the position distance set Loc-on is calculated, the offset degrees are sorted from small to large, and the offset degree set ζ=[ξ1, ξ2,..., ξ k-1 ] is obtained;

[0042] Step S3, according to the odd-even characteristics of the element number in the set , the set is divided into the offset odd set and the offset even set χ+κ=k-1;

[0043] Step S4, the offset odd set and the offset even set are normalized respectively, and the deviation value of the nearest neighbor reference point outside the reference point rp1 is obtained;

[0044] Step S5, if , the deviation value in the offset odd set is positive, and the deviation value in the offset even set is negative; otherwise, the deviation value in the offset odd set is negative, and the deviation value in the offset even set is positive;

[0045] Step S6, the deviation value of the reference point rp1 is set as 0, and the deviation value set B=[β1, β2,..., β k ] is obtained, and satisfies

[0046] Step S7, final coordinate calculation for the to-be-positioned point:

[0047]

[0048] wherein (x h ,y h ) represents the position information of the hth nearest neighbor reference point, (x,y) is the position coordinate of the to-be-positioned point, ω h represents the weight of the hth nearest neighbor reference point.

[0049] Further, the calculation formula of the weight ω h of the hth nearest neighbor reference point in step S7 is as follows:

[0050]

[0051] wherein WFD h represents the weighted feature distance of the hth nearest neighbor reference point.

[0052] Further, the calculation formula of the offset degree set ζ = [ξ1, ξ2,..., ξ k-1 ] in step S2 is as follows:

[0053]

[0054] wherein i-sk = 1, 2,..., k-1, b = 1, 2,..., k-1, and η is a parameter.

[0055] Beneficial effects: The method for region division of the present application divides the fingerprint library into multiple sub-regions, and the division of the sub-regions makes it possible to complete the sub-region matching of the received signal first, and only the reference positions in the region need to be matched, thereby effectively reducing the complexity of the position matching in the online positioning stage. In addition, the improved adaptive weighted K nearest neighbor algorithm is used to correct the weight value according to the statistical law and to calculate the position of the to-be-positioned point, thereby effectively improving the accuracy of the wireless intelligent positioning. BRIEF DESCRIPTION OF DRAWINGS

[0056] Figure 1 is the flowchart of the present application;

[0057] Figure 2 is the flowchart of the sub-region division method in the offline stage of the present application. DETAILED DESCRIPTION

[0058] The present application will be further described in detail below with reference to the accompanying drawings

[0059] The application provides a wireless fingerprint intelligent positioning method based on adaptive WKNN, first, a positioning area division method based on greedy clustering is used for area division, the number of sub-areas and area centers are determined according to the position distance and feature distance relationship between reference points, and the sub-area division in the offline stage is completed; second, the number and set of the nearest neighbor reference points are adaptively determined, the number K and set of the adjacent reference points of real-time data are adaptively determined according to the relationship between the position distance and the feature distance; finally, a positioning method based on statistical law correction is used to correct the weight of each adjacent reference point, and then the positioning information of the terminal to be positioned is obtained. The method combines the sub-area division method based on hierarchical greedy and the weighted nearest neighbor matching algorithm in the positioning technology, reduces the fingerprint matching time for online real-time positioning, and improves the positioning accuracy.

[0060] The application mainly includes three contents:

[0061] First, a positioning area division method based on greedy clustering is used for area division, the number of sub-areas and area centers are determined according to the position distance and feature distance relationship between reference points, and the sub-area division in the offline stage is completed;

[0062] Second, the number and set of the nearest neighbor reference points are adaptively determined, the number K and set of the adjacent reference points of real-time data are adaptively determined according to the relationship between the position distance and the feature distance;

[0063] Third, a positioning method based on statistical law correction is used to correct the weight of each adjacent reference point, and then the positioning information of the terminal to be positioned is obtained.

[0064] I. Sub-area division

[0065] Suppose that the total number of reference points collected in the positioning area is N, the total number of base stations participating in positioning is M, and the collected RSS feature data is R i-clu ={RSS i-clu,1 ,RSS i-clu,2 ,...,RSS i-clu,M}, wherein RSS i-clu,m indicates the RSS data collected by the i-clu reference point from the m base station.

[0066] First, the initial area center is determined.

[0067] The distance between each area center should be as far as possible, the position distance between all reference points in the fingerprint database is calculated, that is, the position distance of the i reference point is dis_l i ={dis_l i,1 ,dis_l i,2 ,...,dis_l i,NThe initial region center C1 is the reference point with the largest average position distance among all the reference points in the fingerprint database. The position distance between the ith reference point and the jth reference point is calculated as shown in the following formula:

[0068]

[0069] where (x i ,y i ) and (x j ,y j ) represent the position information of the ith reference point and the jth reference point in the fingerprint database, respectively.

[0070] Secondly, the coverage radius of the initial region center C1 is determined.

[0071] The feature distance between the initial region center C1 and the remaining reference points in the fingerprint database is calculated where,

[0072]

[0073] where, RSS a,m represents the signal feature value of the mth base station received by the initial region center C1.

[0074] The coverage radius p1 of the initial region center C1 is determined based on the maximum and minimum values of the above feature distance;

[0075]

[0076] At this time, when the feature distance between a reference point in the fingerprint database and the initial region center C1 is within the coverage radius, the reference point is in the sub-region 1 (initial region center C1).

[0077] Then, the next sub-region division is continued for the reference points that have not been divided into sub-regions.

[0078] For the sub-region c, if the number of divided regions is 1, the point with the largest position distance from the initial region center C1 is found among the remaining reference points as the region center C c ; if the number of divided regions is greater than 1, the point with the largest average position distance from all the region centers is found among the remaining reference points as the region center C c ; the feature distance between C c and the remaining reference points is calculated, and the coverage radius p c of C c is determined in the same way as C1, and the reference points within the radius range are the reference points in the sub-region.

[0079] Repeat the above steps until all reference points have completed the sub-region division.

[0080] After all sub-region division is completed, it is also necessary to determine whether there is a reference point being repeatedly divided into regions. If so, further calculate the feature distance between the reference point and each region center, and the sub-region corresponding to the nearest region center is the region to which the reference point belongs.

[0081] Finally, update the center of each sub-region: the mean of the fingerprint information of all reference points in each sub-region is the center of the sub-region.

[0082] II. Adaptive determination of the number and set of nearest neighbor reference points

[0083] Suppose the signal value collected by the user carrying the mobile device at the to-be-located point is R on ={RSS on,1 ,RSS on,2 ,...,RSS on,M}. First, compare the similarity with each sub-region center to determine the sub-region attribution C on of the to-be-located point, and the number of reference point samples in the sub-region C on is n Con ; calculate the base station similarity of R on with each reference point sample in C on , which is calculated as follows:

[0084]

[0085] where SIM AP (u,s i-on ) represents the base station similarity between real-time data and each reference point in the region, U on represents the set of signal sources received by the to-be-located point in the online stage, and s i-on represents the set of signal sources received by the i-onth reference point in the sub-region C on .

[0086] Calculate the Sorensen distance (feature distance) between R on and each reference point in C on to obtain where,

[0087]

[0088] where RSS i-on,m represents the signal feature value received by the i-onth reference point from the mth base station.

[0089] Then, the weighted feature distance WFD between each reference point in C on and the to-be-located point is:

[0090]

[0091] The reference points in the set A_Fea are sorted in ascending order according to the corresponding WFD.

[0092] Secondly, the reference point F with the minimum WFD is determined, and the position coordinate distance between F and C on and the rest of the reference points is calculated. The reference points in the set B_Loc are sorted in ascending order according to the coordinate distance Edis, and the reference point F is placed in the first position in the set B_Loc. The position coordinate distance between F and the jc-th reference point other than F in C on is calculated as shown in the formula.

[0093]

[0094] Then, the set A_Fea and the set B_Loc are compared: a threshold value ε is set, and if the distance of the first n WFD reference points in the set A_Fea is lower than the threshold value, the first n WFD reference points are retained to construct a new feature distance set NA_Fea; at the same time, the first n WFD reference points in the coordinate set B_Loc are selected to construct a new coordinate set NB_Loc; at this time, the final determined nearest neighbor reference point set is constructed by the reference points common to the two sets NA_Fea and NB_Loc, and the number of the nearest neighbor reference points k is the number of the common reference points k. The calculation of the threshold value ε is shown in the following formula:

[0095]

[0096] Finally, the nearest neighbor reference point set RPc = {rp1, rp2,..., rp k} is obtained, wherein rp1 is the reference point F with the minimum WFD.

[0097] III. Weight correction based on statistical law

[0098] In the online phase, the final coordinate calculation is performed by the above-mentioned nearest neighbor reference point set RPc = {rp1, rp2,..., rp k} of the to-be-positioned point. In the coordinate calculation process, the weighting process is modified based on the relationship between the position distance and the feature distance between the neighboring reference points.

[0099] Firstly, the first reference point in the set of nearest neighbor reference points RPc, i.e. the reference point with the smallest feature distance, is recorded as the reference point, and the weighted feature distance and position distance between the reference point rp1 and the other (k-1) reference points are calculated to obtain the weighted feature distance set Sor-on = [Sor1, Sor2,..., Sor k-1 ] and the position distance set Loc-on = [Loc1, Loc2,..., Loc k-1 ], and the data in them are sorted from small to large.

[0100] Secondly, the deviation degree between the weighted feature distance set Sor-on and the position distance set Loc-on is calculated, and the deviation degree of the i-th reference point is calculated as follows:

[0101]

[0102] wherein η is a model parameter, which is set to 1.8 here.

[0103] Therefore, the deviation degree set can be obtained for the (k-1) points. The deviation degrees are sorted from small to large, and they are divided into the odd deviation set and the even deviation set χ+κ=k-1 according to the subscript, and the deviation values of the reference points are obtained by normalizing the two sets, as follows:

[0104]

[0105] The deviation value β F of the reference point is set to 0, the sizes of the odd deviation set and the even deviation set are compared to determine the sign of the deviation value, if min(ζ odd ) < min(ζ even ), the deviation value in the odd deviation set is positive, and the deviation value in the even deviation set is negative; otherwise, the deviation value in the odd deviation set is negative, and the deviation value in the even deviation set is positive; finally, a deviation value set B = [β1, β2,..., β k ] can be obtained, and it satisfies

[0106] Finally, the final coordinate calculation of the to-be-positioned point is performed:

[0107]

[0108] wherein (x h , y h ) represents the position information of the h-th nearest neighbor reference point, (x, y) is the position information of the to-be-positioned point obtained by calculation, ω hThe weight of the hth nearest neighbor reference point is calculated as follows:

[0109]

[0110] where WFD h represents the weighted feature distance of the hth nearest neighbor reference point.

[0111] The technical features of the above embodiments can be combined in any manner. In order to make the description simple, all possible combinations of the technical features in the above embodiments are not described, however, as long as the combinations of the technical features do not contradict, they should be considered as falling within the scope of the present disclosure.

[0112] The above embodiments only express several implementation manners of the present application, and the description is relatively specific and detailed, but it should not be understood as a limitation on the scope of the patent. It should be pointed out that, for ordinary skilled persons in the art, several modifications and improvements can be made without departing from the concept of the present application, and these all fall within the protection scope of the present application. Therefore, the protection scope of the patent of the present application should be subject to the appended claims.

Claims

1. A wireless fingerprint intelligent positioning method based on adaptive WKNN, characterized in that, The method comprises: Dividing the region to be positioned into several sub-regions; Determining the number and set of the nearest neighbor reference points of the point to be positioned; Using a positioning method based on statistical rule correction to correct the weight of each nearest neighbor reference point, and then obtaining the positioning information of the point to be positioned, the specific process comprising: Step S1, taking the first reference point rp1 in the nearest neighbor reference point set RPc={rp1, rp2,..., rp k} as a reference point, calculating the weighted feature distance and position distance between the reference point rp1 and other nearest neighbor reference points, sorting the weighted feature distance and position distance from small to large to obtain the weighted feature distance set Sor-on=[Sor1, Sor2,..., Sor k-1 ] and the position distance set Loc-on=[Loc1, Loc2,..., Loc k-1 ]. Step S2, calculate the offset degree between the weighted feature distance set Sor-on and the location distance set Loc-on, and sort the offset degrees from small to large to obtain an offset degree set Step S3, grouping the elements numbered according to the parity of the offset into the offset odd group and the offset even group χ+κ=k-1; Step S4, normalizing the offset odd set and the offset even set respectively to obtain the offset value of the nearest neighbor reference point outside the reference point rp1; Step S5, if then the offset value in the odd offset set is positive and the offset value in the even offset set is negative; otherwise, the offset value in the odd offset set is negative and the offset value in the even offset set is positive; Step S6, set the deviation value of the reference point rp1 as 0, and obtain the deviation value set B = [β1, β2,..., βn] and satisfy k ] and satisfy Step S7, performing final coordinate calculation on the point to be positioned: wherein (x h ,y h ) represents the position information of the hth nearest neighbor reference point, (x,y) is the position coordinate of the point to be positioned, ω h h represents the weight of the hth nearest neighbor reference point; The weight ω of the hth nearest neighbor reference point in the step S7 h The calculation formula is: where WFD h denotes the weighted feature distance of the h-th nearest reference point.

2. The method of claim 1, wherein, Using a positioning region division method based on greedy clustering to divide the region to be positioned, the specific steps comprising: Step 1: Determine the initial region center C1 as the reference point with the largest mean positional distance among all reference points, where the positional distance of the i-th reference point is dis_l. i ={dis_l i,1 ,dis_l i,2 ,...,dis_l i,N }, dis_l i,j Let N be the distance between the i-th reference point and the j-th reference point, and let N be the total number of reference points in the fingerprint database within the positioning area. Step 2: determining the coverage radius p1 of the sub-region 1 where the initial region center C1 is located: wherein, is a set of feature distances between C1 and the rest of the reference points in the fingerprint library, is a feature distance between C1 and the a-th reference point in the fingerprint library, excluding C1, denotes the m-th signal feature value of C1 received from the m-th base station, RSS a,m Rm(a) represents the signal characteristic value of the mth base station received by the ath reference point, and M is the total number of base stations participating in positioning. Step 3: dividing the next sub-region: For sub-region c, if the number of sub-regions currently divided is 1, find the point farthest from the position of C1 among the remaining reference points of the sub-region to which the sub-region c belongs as the region center C of the sub-region c c ; if the number of sub-regions currently divided is > 1, find the point farthest from the position mean of the region centers of all sub-regions currently divided among the remaining reference points as the region center C of the sub-region c c ; calculate the feature distance between C c and the remaining reference points, determine the coverage radius p of the sub-region c c ; Step 4: repeating step 3 until all reference points are divided into the sub-regions they belong to; Step 5: updating the region center of each sub-region to the mean value of the fingerprint information of all reference points in each sub-region.

3. The method of claim 2, wherein, Before step 5, it is also necessary to determine whether there is a reference point that is repeatedly divided into a sub-region, if so, further calculate the feature distance between the reference point and the region center of each sub-region, and select the sub-region corresponding to the region center with the closest feature distance as the sub-region where the reference point belongs to.

4. The method of claim 1, wherein, The specific steps for determining the number and set of the nearest neighbor reference points comprise: Step ①: calculating the similarity between the signal value collected at the point to be positioned and the signal value collected at the region center of each sub-region, and selecting the sub-region corresponding to the maximum similarity as the belonging region of the point to be positioned; Step ②: calculating the base station similarity and the Sorensen distance between the signal value collected at the point to be positioned and the signal value collected at each reference point in the belonging region, multiplying the two to obtain the weighted feature distance between each reference point in the belonging region and the point to be positioned, and recording the reference point corresponding to the minimum weighted feature distance as F; Step ③: sorting all reference points in the belonging region according to the position coordinate distance between them and F from small to large to form a set B_Loc; Step 4: Sort the reference points in the home area according to the corresponding weighted feature distance from small to large, and take the first n reference points with the weighted feature distance lower than the set threshold to construct a new feature distance set NA_Fea. WFD Step 4: Sort the reference points in the home area according to the corresponding weighted feature distance from small to large, and take the first n reference points with the weighted feature distance lower than the set threshold to construct a new feature distance set NA_Fea. Step ⑤: Take the first n WFD reference points in the set B_Loc to construct a new coordinate set NB_Loc; Step ⑥: the intersection of NA_Fea and NB_Loc is the set of the nearest neighbor reference points, and the number of common reference points of NA_Fea and NB_Loc is the number of the nearest neighbor reference points.

5. The method of claim 1, wherein, The threshold value is set as (maximum weighted feature distance + minimum weighted feature distance) / 2.

6. The method of claim 4, wherein, The calculation method of the similarity between the signal value collected at the point to be positioned and the signal value collected at the region center of each sub-region is: wherein, denotes the signal value of the mth base station collected at the center of the sub-area c, RSS on,m denotes the signal value of the mth base station collected at the point to be positioned, M denotes the number of base stations.

7. The method of claim 4, wherein, In step ②, the calculation method of the base station similarity between the signal value collected at the point to be positioned and the signal value collected at each reference point in the belonging region is: wherein SIM AP (u,s i-on ) represents the similarity of the base station between the signal value collected at the point to be located and the i-onth reference point in the home area, U on represents the set of base stations corresponding to the signal value collected at the point to be located, s i-on represents the set of base stations corresponding to the signal value collected at the i-onth reference point in the home area; The calculation method of the Sorensen distance between the signal value collected at the point to be positioned and the signal value collected at each reference point in the belonging region is: wherein RSS i-on,m denotes the signal value of the mth base station collected at the ith reference point in the home area, and M denotes the number of base stations; The calculation method of the weighted feature distance between each reference point in the belonging region and the point to be positioned is:

8. The method of claim 1, wherein, The set of offset degrees in the step S2 The calculation formula is: Wherein, i-sk=1,2,…k-1, b=1,2,…k-1, and η is a constant.

Images