CN102638888A - Indoor positioning fingerprint grouping method based on signal statistics - Google Patents

Abstract

The invention provides an indoor positioning fingerprint grouping method based on signal statistics. The method includes selecting N reference points from a positioning scene and acquiring the number ( hereinafter referred to as m) of access points in the positioning scene; performing T signal intensity samplings in the reference points and recording signal intensity information of the access points in each sampling; dividing the signal intensity information of the access points which have same strongest signal intensities in the signal intensity information of the T samplings into a group and recording the number of the signal intensity information of each group; judging whether the number is larger than or equal to a frequency-time threshold value alpha or not, if so, averaging the signal intensity information of the group as a group of fingerprints of the nth reference point, setting m coarse groupings for the fingerprints, selecting the strongest access point in the fingerprints corresponding to the reference points in each coarse grouping, and finally differing the fingerprints. The indoor positioning fingerprint grouping method based on the signal statistics is simple in operation, and the number of fingerprints needed to be compared during positioning is effectively reduced, so that cost is lowered.

Description

Indoor positioning fingerprint group technology based on signal statistics

Technical field

The present invention relates to communication and radio network technique field, be specifically related to a kind of indoor positioning fingerprint group technology based on signal statistics.

Background technology

Location based services (Location-based service is called for short LBS) is a kind of very important application in mobile computing and radio communication.LBS not only has huge using value in commerce, and also leaves no choice but important for public safety, communications and transportation and urgent rescue.More well-known navigation system has GPS of USA, the Big Dipper of China, the Galileo and the Muscovite lattice Lip river Paasche of European Union.These navigation systems are all very good at outdoor positioning result, still all lost their ample scope for abilities in indoor these navigation systems, because faint framing signal can't be passed through wall.Under this background, people develop a lot of indoor locating systems, mainly contain based on infrared, based on hyperacoustic and based on wireless lan (wlan).Because the navigation system based on WLAN is to utilize the wlan network that has existed, therefore this system cost is cheap, and in the last few years, the laying of WLAN is more and more wider, more and more closeer, and is increasing based on the navigation system application region of WLAN.

The WLAN navigation system roughly can be divided into two types, based on propagation model and position-based fingerprint.Because indoor environment more complicated, and radio signal meeting generation diffraction, reflection, scattering and multipath transmission in communication process, some difficult parameters that cause propagation model cause based on the Positioning System of propagation model generally relatively poor to confirm.The navigation system of position-based fingerprint mainly is divided into two stages: training stage and positioning stage.Training stage, choose some reference points and repeatedly sample, note the corresponding signal strength information of each access point, store in the fingerprint base.At positioning stage; Common localization method has NNSS (signal space arest neighbors method); Portable terminal obtains each access point signals strength information, through with fingerprint base in the fingerprint of each reference point compare, find out the position of the highest reference point of similarity as portable terminal.This method is simple; Be prone to realize, but for bigger location scene, the quantity of reference point can be quite huge; Therefore in the single location, need the fingerprint amount of comparison also quite huge; To accomplish the exigent location of real-time this moment and be suitable difficulty, be difficult to especially realize under the very big situation of particularly parallel Location Request amount, and the comparison amount too greatly also can increase load of server with can energy consumption.Fingerprint comparison amount when therefore reducing real-time positioning is necessary.

Have a lot of scholars to invent multiple group technology abroad, it is the grouping index with q maximum in reference point access point that group technology is arranged, and will have q access point of identical maximum to be divided into one group.This group technology is not considered the fluctuation of signal, and site undetermined is easy to take place the misclassification group during location.Also have group technology to utilize KNN (K Mean Method) method will have the reference point of similarity signal strength information to be divided into a group; The fingerprint of all reference points is averaged; The signal strength values of average as whole group; Earlier with the signal strength values of the signal strength information in site undetermined and each group relatively, it is maximum to find out similarity during the location, and then in group, looks for the reference point of similarity maximum.More than two kinds of methods all do not consider the situation that the training stage is different with the antenna gain at positioning stage terminal, thereby positioning accuracy is very possible descends because antenna gain different.

Summary of the invention

The object of the present invention is to provide a kind of indoor positioning fingerprint group technology based on signal statistics, its work is simple, each group of back that divides into groups the fingerprint number reduce greatly; And be not prone to the situation that mistake is divided into groups; In the time of effectively reducing to locate the fingerprint quantity that will compare, improve positioning accuracy, reduce the work load of location-server; Cut down the consumption of energy, improve locating speed.

The present invention realizes through following technical scheme:

A kind of indoor positioning fingerprint group technology based on signal statistics may further comprise the steps:

(1) in the scene of location, choose N reference point, the quantity of obtaining access point in the scene of location is m;

(2) carry out T signal strength signal intensity sampling in reference point, note the signal strength information of each sampling access point, the t time sample information of n reference point is designated as $S_n^t=(s_{n1}^t,s_{n2}^t,...s_{\mathrm{Nm}}^t),$ T=1 wherein, 2 ..., T;

(3) there is the signal strength information of the access point of identical strongest signal strength to be divided into a group in the signal strength information of each reference point with T sampling; And writing down the number of the signal strength information of each group, the number of the signal strength information of i group of n reference point is designated as

each reference point has m such group at most;

(4) judge number Whether more than or equal to frequency threshold alpha; If then the signal strength information of this group is averaged as one group of fingerprint of n reference point, otherwise abandons the signal strength information of this group; Each reference point can have m group fingerprint at the most like this, and i fingerprint of n reference point is designated as ${\stackrel{\‾}{S}}_n\left(i\right)=({\stackrel{\‾}{s}}_{n1}\left(i\right),{\stackrel{\‾}{s}}_{n2}\left(i\right),...,{\stackrel{\‾}{s}}_{\mathrm{Nm}}\left(i\right);f_n^i);$

(5) for fingerprint is provided with m rough segmentation group, be designated as C ₁, C ₂..., C _m, for all fingerprints of n reference point, i the strongest corresponding reference point of access point will be assigned to rough segmentation group C with this fingerprint in the fingerprint _iIn, the access point of signal strength signal intensity strongest signal strength is in i the fingerprint of n reference point

Therefore reference point n can assign to rough segmentation group C _iIn, fingerprint

Also will be stored in rough segmentation group C _iIn;

(6) choose access point the strongest in the fingerprint of the reference point correspondence in each rough segmentation group for reference to access point, fingerprint is carried out difference, the access point of signal strength signal intensity strongest signal strength in i the fingerprint of n reference point The signal strength signal intensity of other access points with

Differ from, the difference fingerprint that obtains is designated as ${\stackrel{\‾}{S}}_n^d\left(i\right)=({\stackrel{\‾}{s}}_{n1}^d(i,1),{\stackrel{\‾}{s}}_{n2}^d(i,2),...,{\stackrel{\‾}{s}}_{\mathrm{Nm}}^d(i,m),f_n^i);$

(7) the individual segmentation group of m* (m-1) is set, is designated as C ₁₂, C ₁₃..., C _{(m-1) m}

(8) find out the strongest, inferior strong differential signal in the difference fingerprint, suppose the difference fingerprint of n reference point ${\stackrel{\‾}{S}}_n^d\left(i\right)=({\stackrel{\‾}{s}}_{n1}^d(i,1),{\stackrel{\‾}{s}}_{n2}^d(i,2),...,{\stackrel{\‾}{s}}_{\mathrm{Nm}}^d(i,m),f_n^i)$ In the strongest, inferior two strong differential signals do

Then reference point n will be assigned to segmentation group C _{I, j}In;

(9) judge

Reference point n whether is less than or equal to similarity threshold β, if then also can be assigned to segmentation group C _{I, k}In, otherwise reference point n can not assigned to segmentation group C _{I, k}In;

(10), repeat above-mentioned steps (5) to (9) to all fingerprints of all reference points.

The present invention has following advantage and technique effect:

1, a reference point; With maximum access point is the grouping foundation, and the sample information that repeatedly collects is divided into a plurality of groups, and the frequency of occurrence threshold value is set; Consider that not only a reference point different fingerprints can occur in different positioning times; And rejected the fingerprint of accidental appearance, not only reduce the possibility that the location mistiming is divided into groups, and reduced fingerprint quantity to a certain extent.

A similarity threshold is set during 2, segmentation group, has reduced because the possibility that the mistake that signal fluctuation causes is divided into groups.

3, fingerprint quantity has reducing on the magnitude with respect to the fingerprint amount of whole locating area in the segmentation group, thereby locating speed accelerates computation burden decline, energy consumption reduction.

4, divide into groups in last fingerprint be the signal differential fingerprint, than the group technology of general direct employing signal strength signal intensity, effectively suppressed because the difference of terminal gain makes the positioning accuracy variation as fingerprint.

Description of drawings

Fig. 1 is the indoor positioning fingerprint group technology flow chart that the present invention is based on signal statistics.

Fig. 2 is T sampling back of a reference point signal strength information storage node composition.

Rough segmentation group sketch map in the scene of a location of Fig. 3.

Reference point received signals fingerprint figure after Fig. 4 rough segmentation group.

The fingerprint segmentation group sketch map of a rough segmentation group of Fig. 5 confidential reference items examination point.

Embodiment

As shown in Figure 1, the indoor positioning fingerprint group technology based on signal statistics of the present invention may further comprise the steps:

(1) choose N reference point in the location in the scene, reference point tries not to be chosen in the seldom position of process of people, such as corner or near the position of metope, definition is located in the scene has m access point;

(2) carry out T signal strength signal intensity sampling at each with reference to examination point; Note the signal strength information of each sampling; The t time sample information of n reference point is designated as the signal strength information that

Fig. 2 has shown T sampling in n the reference point; If can't detect certain access point in the sampled signal of reference point; Then this time in signal strength information, with-100dBm signal strength signal intensity as this access point;

(3) there is the sample information of the access point of identical strongest signal strength to be divided into a group in the signal strength information with T sampling; And writing down the number of the signal strength information of each group, the number of i panel signal strength information of n reference point is designated as the such group of

each reference point has m at most;

(4) frequency threshold alpha is set, for reference point, if

Then the signal strength information of this group is averaged as one group of fingerprint of this reference point, on the contrary then abandon this group signal strength information, each reference point can have m fingerprint at the most like this, i fingerprint of n reference point is designated as ${\stackrel{\‾}{S}}_n\left(i\right)=({\stackrel{\‾}{s}}_{n1}\left(i\right),{\stackrel{\‾}{s}}_{n2}\left(i\right),...{\stackrel{\‾}{s}}_{\mathrm{Nm}}\left(i\right);f_n^i);$

(5) m rough segmentation group is set, is designated as C ₁, C ₂..., C _m,, the reference point of the access point that identical strongest signal strength is arranged in the fingerprint is assigned in the same rough segmentation group fingerprint of n reference point for all reference points

The access point of middle signal strength signal intensity strongest signal strength is

Therefore this fingerprint of reference point n can be assigned to rough segmentation group C _iIn, fingerprint

Also will be stored in rough segmentation group C _iIn.

Fig. 3 has provided the sketch map of a rough segmentation group; Fig. 3 has shown has the location of 5 access points scene, sampling number T=200, and the access point the strongest according to each reference point is divided into 5 rough segmentation groups with all reference points; The reference point that has among the figure can be assigned in a plurality of rough segmentation groups; Because in T the sampling, maybe be because the fluctuation of signal, access point the strongest on reference point different sampling stages is inequality; And the frequency that occurs is greater than the threshold alpha of the frequency, and the corresponding fingerprint of identical reference point is different in the different here rough segmentation groups.

Fig. 4 has provided the instance of the fingerprint of a reference point correspondence, if stipulate here, and T=200, α=20, then article one fingerprint of this reference point can be assigned to rough segmentation group C ₁, the second fingerprint can be assigned to rough segmentation group C ₂In;

(6) choose access point the strongest in the fingerprint of the reference point correspondence in each rough segmentation group for fingerprint is carried out difference, a fingerprint of n fingerprint point with reference to access point ${\stackrel{\‾}{S}}_n\left(i\right)=({\stackrel{\‾}{s}}_{n1}\left(i\right),{\stackrel{\‾}{s}}_{n2}\left(i\right),...,{\stackrel{\‾}{s}}_{\mathrm{Nm}}\left(i\right);f_n^i),$ Signal strength signal intensity with reference to access point does

The signal strength signal intensity of other access points with

Differ from, the difference fingerprint that obtains is designated as ${\stackrel{\‾}{S}}_n^d\left(i\right)=({\stackrel{\‾}{s}}_{n1}^d(i,1),{\stackrel{\‾}{s}}_{n2}^d(i,2),...,{\stackrel{\‾}{s}}_{\mathrm{Nm}}^d(i,m),f_n^i)\·$

The signal strength signal intensity that the access point of strongest signal strength is corresponding in article one fingerprint among Fig. 4 is-40.3, and fingerprint is (0 ,-2.2 ,-7.9 ,-8 ,-19.9 after the difference; 140);

(7) the individual segmentation group of m* (m-1) is set, is designated as C ₁₂, C ₁₃..., C _{(m-1) m}

(8) find out the strongest, inferior strong differential signal in the said difference fingerprint, suppose a difference fingerprint of n reference point ${\stackrel{\‾}{S}}_n^d\left(i\right)=({\stackrel{\‾}{s}}_{n1}^d(i,1),{\stackrel{\‾}{s}}_{n2}^d(i,2),...,{\stackrel{\‾}{s}}_{\mathrm{Nm}}^d(i,m),f_n^i)$ In two the strongest differential signals do Reference point n will be assigned to segmentation group C _{I, j}In, similarity threshold β＞0 is set, relatively

With the difference of remaining differential signal, if satisfy

Then reference point n not only can be assigned to segmentation group C _{I, j}, and can assign to segmentation group C _{I, k}In.

Among Fig. 4, the differentiated signal strength information of article one fingerprint is (0 ,-2.2 ,-9.9 ,-8 ,-19.9; 140), if β=6 are set here, this fingerprint at first can be assigned to segmentation group C ₁₂In, here-2.2-(8)=5.8＜6, so this fingerprint also can be assigned to segmentation group C ₁₄In.

Fig. 5 has provided a rough segmentation group C ₄The situation of the fingerprint segmentation group that the confidential reference items examination point is corresponding, reference point RP _kThe difference of second differentiated signal strength information and the 3rd differentiated signal strength information is less than similarity threshold β in the corresponding fingerprint, so RP _kWill be assigned to segmentation group C ₄₂, C ₄₃In;

(9) all fingerprints to all reference points all divide into groups according to the step of (1)～(8).

Claims (2)

1. the indoor positioning fingerprint group technology based on signal statistics is characterized in that, comprises following steps:

(1) in the scene of location, choose N reference point, the quantity of obtaining access point in the scene of said location is m;

(2) carry out T signal strength signal intensity sampling in said reference point, note the signal strength information of each sampling access point, the t time sample information of n reference point is designated as $S_n^t=(s_{n1}^t,s_{n2}^t,...s_{\mathrm{Nm}}^t),$ T=1 wherein, 2 ..., T;

(3) there is the signal strength information of the access point of identical strongest signal strength to be divided into a group in the signal strength information of each reference point with T sampling; And writing down the number of the signal strength information of each group, the number of the signal strength information of i group of n reference point is designated as

each reference point has m such group at most;

(4) judge number Whether more than or equal to frequency threshold alpha; If; Then the signal strength information of this group is averaged as one group of fingerprint of said n reference point; Otherwise abandon the signal strength information of this group, each reference point can have m group fingerprint at the most like this, and i fingerprint of n reference point is designated as ${\stackrel{\‾}{S}}_n\left(i\right)=({\stackrel{\‾}{s}}_{n1}\left(i\right),{\stackrel{\‾}{s}}_{n2}\left(i\right),...,{\stackrel{\‾}{s}}_{\mathrm{Nm}}\left(i\right);f_n^i);$

(5) for said fingerprint m rough segmentation group is set, is designated as C ₁, C ₂..., C _m, for all fingerprints of n reference point, i the strongest corresponding reference point of access point will be assigned to rough segmentation group C with this fingerprint in the fingerprint _iIn, the access point of signal strength signal intensity strongest signal strength is in i the fingerprint of n reference point

Therefore reference point n can assign to rough segmentation group C _iIn, fingerprint

Also will be stored in rough segmentation group C _iIn;

(6) choose access point the strongest in the fingerprint of the reference point correspondence in each rough segmentation group for reference to access point, fingerprint is carried out difference, the access point of signal strength signal intensity strongest signal strength in i the fingerprint of n reference point

The signal strength signal intensity of other access points with

Differ from, the difference fingerprint that obtains is designated as ${\stackrel{\‾}{S}}_n^d\left(i\right)=({\stackrel{\‾}{s}}_{n1}^d(i,1),{\stackrel{\‾}{s}}_{n2}^d(i,2),...,{\stackrel{\‾}{s}}_{\mathrm{Nm}}^d(i,m),f_n^i);$

(7) the individual segmentation group of m* (m-1) is set, is designated as C ₁₂, C ₁₃..., C _{(m-1) m}

(8) find out the strongest, inferior strong differential signal in the said difference fingerprint, suppose the difference fingerprint of n reference point ${\stackrel{\‾}{S}}_n^d\left(i\right)=({\stackrel{\‾}{s}}_{n1}^d(i,1),{\stackrel{\‾}{s}}_{n2}^d(i,2),...,{\stackrel{\‾}{s}}_{\mathrm{Nm}}^d(i,m),f_n^i)$ In the strongest, inferior two strong differential signals do

Then reference point n will be assigned to segmentation group C _{I, j}In;

(9) judge

Reference point n whether is less than or equal to similarity threshold β, if then also can be assigned to segmentation group C _{I, k}In, otherwise reference point n can not assigned to segmentation group C _{I, k}In;

(10), repeat above-mentioned steps (5) to (9) to all fingerprints of all reference points.

2. indoor positioning fingerprint group technology according to claim 1 is characterized in that, the value of said frequency threshold value be 0.1T between the 0.25T, said similarity threshold β value is the truth of a matter e of natural logrithm.

Images