CN104217231A - RFID positioning system and positioning method based on non-accurate anchor nodes - Google Patents

Abstract

The invention discloses a RFID positioning system and a RFID positioning method based on non-accurate anchor nodes. The system mainly comprises goods shelves, goods, RFID tags, a reader-writer, an antenna and the like. In an environment to be positioned, the goods shelves with known coordinate positions are deployed, the goods with RFID tags are placed on the goods shelves, accurate positions of the goods are unknown and only start and stop positions of each kind of goods on the goods shelves can be obtained. Through using the goods tags without accurate position information as the anchor nodes, based on distribution of goods categories, a RFID positioning system is constructed, and an effective positioning method is proposed. According to the RFID system, the positioning operation can be completed by only using scanned goods classification information without obtaining accurate position information of the goods.

Description

A kind of RFID location system based on non-precision anchor node and localization method

Technical field

The present invention relates to a kind of RFID localization method, specifically a kind of RFID location system based on non-precision anchor node and localization method.

Background technology

Along with the widespread use of RFID technique, increasing RFID label tag is used, make in some environment (such as supermarket, library, kinds of goods warehouse etc.) that the existence of a large amount of RFID label tag can be had, although can not obtain the exact position of these anchor nodes, its classification position scope is known.Also the demand that personnel are positioned is had in above-mentioned environment, to provide humanized service, as: navigation, neighbouring kinds of goods inquiry etc.But only rely on RFID label tag in existing environment at present, and cannot accurate locating personnel, must by means of reference label.

And the label on kinds of goods, originally be that information for identifying kinds of goods is convenient to know its relevant information, but due to the basic radio characteristics (directivity etc. that signal attenuation, signal are propagated) comprised in RFID technique, actually can expand other functions, but prior art does not fully excavate potentiality wherein.The present invention, by utilize these to seem the goods label that there is no other use for it, can position work to target object.

Summary of the invention

Technical matters to be solved by this invention is to provide a kind of RFID location system based on non-precision anchor node and method thereof, this system need not arrange reference label node in addition, the RFID label tag on existing kinds of goods is only utilized to position operation, can accurate consumer positioning position.

A kind of RFID location system based on non-precision anchor node of the present invention, it comprises following assembly:

Shelf: for putting the kinds of goods posting RFID label tag, the position of shelf is fixed, and the start-stop position knowing kinds of goods classification on each shelf;

RFID label tag: be attached to the common RFID label tag comprising inventory information on kinds of goods;

Mobile device: it carries RFID reader and antenna, is promoted by user, by the location to mobile device, the positional information of user can be obtained;

RFID reader and antenna: for collecting the label information of kinds of goods, and by Internet Transmission to the message processing module in calculation server;

Message processing module: by the label information of the kinds of goods collected, calculates the position of mobile device.

Present invention also offers the localization method of the RFID location system based on non-precision anchor node, it comprises the following steps:

1) mobile device of user's propelling loading RFID reader and antenna, enter shelf article deploy RFID label tag can in localizing environment, in this environment, the position of shelf is fixed, and the y coordinate position namely respectively arranging shelf is known;

2) FRID label information on periphery article collected by RFID reader and antenna, and carries out position arithmetic operation by network afferent message processing module;

3) message processing module obtains the information of classification belonging to each RFID label tag, because target place to be positioned shelf y coordinate position is known, only need calculate the x coordinate position at target place to be positioned, get final product localizing objects.

For step 3) concrete position fixing process have following several paralleling method:

One: message processing module obtains the information of classification belonging to each RFID label tag, because the reference position x of each kind _i,sand end position x _i,eknown, use various types of in the number of tags n that is identified _ifor weighting weights, namely the method be averaging by weighting can calculate the x coordinate position of target, and the following formula of this computation process is described:

$\hat{x}={\mathrm{\Σ}}_{i=1}^k{w}_i\·\frac{x_{i,s}+x_{i,e}}{2},$

$w_i=\frac{n_i}{{\mathrm{\Σ}}_{i=1}^k{n}_i}$

Here, k is the species number be identified, w _ibe the weights of corresponding classification, it obtains divided by all number of tags be identified by using the number of tags be identified in this kind; In this approach, the centre position of kinds of goods kind is used as the position of kind.

Its two:

Message processing module obtains the information of classification belonging to each RFID label tag, because the reference position x of each kind _i,sand end position x _i,eknown, and the projector distance of each RFID label tag distance antenna on identified region is nearer, its RSSI value is higher, uses total RSSI value of each kinds of goods kind as weights here, namely the method be averaging by weighting can calculate the x coordinate position of target, and computing method are described below:

$\hat{x}={\mathrm{\Σ}}_{i=1}^k{w}_i\·\frac{x_{i,s}+x_{i,e}}{2},$

$w_i=\frac{s_i}{{\mathrm{\Σ}}_{i=1}^k{s}_i}$

Wherein, k represents that k the kinds of goods with the strongest RSSI mean value chosen are classified, s _ifor the mean value of the RSSI value of all labels in the kinds of goods classification of acquisition, w _ifor according to s _ithe each classification weights calculated, because the start-stop position of each classification is known, can calculate the positional information of mobile device according to this algorithm.

Its three:

Message processing module obtains the information of classification belonging to each RFID label tag, because the reference position x of each kind _i,sand end position x _i,eknown, meanwhile, in scanning process, the number of tags of each classification the inside can be known, using these each tag along sort numbers obtained as a vector, the scope that all labelings be identified cover is as possible position scope, from the left side of possible position scope, every △ d distance, just centered by possible position, a circle is outwards drawn with the radius between main cog region, a series of number of tags vector table can be obtained, the label vector calculated by the number of tags vector sum of match cognization, by a k nearest neighbor algorithm, find k most similar vector, again according to their possible position, estimate the x coordinate position of target.

Its four:

Under obtaining different RF ID label densities situation by experiment, the relation training set T between RSSI intensity and label to antenna projector distance _d, by calculating each tag distances antenna projector distance, provide the anticipation position of each label for projected position; By calculating the center of these anticipation positions, be calculative target location.

The present invention directly uses the label be attached on kinds of goods, positions user, to provide more personalized service, as: the location Based service such as navigation, the inquiry of neighbouring kinds of goods, Instant Ads propelling movement.When user push shopping cart advance shopping time, the reader that shopping cart is equipped with by antenna scanning and the label information obtained in surrounding enviroment, uses the respective algorithms of follow-up proposition herein can calculate the Position Approximate of client.The present invention does not have the goods label of precise position information for anchor node by using these, based on the distribution situation of kinds of goods kind, has built a kind of RFID location system, and has proposed four kinds of effective localization methods.This system does not need the precise position information obtaining kinds of goods, without the need to setting up reference label network, only needing can complete positioning action by the kinds of goods classified information scanned, simplifying operating performance, reduce use expense.

Accompanying drawing explanation

Fig. 1 is system architecture diagram of the present invention,

Fig. 2 is the prototype figure that in the present invention, shelf and kinds of goods are disposed,

Fig. 3 is integral deployment figure of the present invention,

Fig. 4 is the workflow diagram based on RSSI method,

Fig. 5 is classification and matching algorithm principle figure in the present invention,

Fig. 6 is middle distance Voting Algorithm schematic diagram of the present invention,

Fig. 7 is middle distance Voting Algorithm process flow diagram of the present invention.

Specific implementation method

Below in conjunction with accompanying drawing, the invention will be further described.

As Fig. 1, RFID location system of the present invention, its composition comprises shelf, passive RFID tags, mobile device, RFID reader+antenna, the calculation server with message processing module and wireless network.Fig. 2 is the prototype figure that in the present invention, shelf and kinds of goods are disposed.Wherein RFID label tag is attached on kinds of goods, and kinds of goods are placed on shelf by its classification, and the position in length and breadth of shelf is known, and the start-stop position of putting kinds of goods classification is above known.Mobile device is laid RFID reader and antenna, for obtaining the label information of periphery.Message processing module is ARM9 development board, and the statistics of the label information obtained by process and RSSI information, return the positional information of mobile device, i.e. the positional information of user.

Position fixing process of the present invention is:

1) mobile device of user's propelling loading RFID reader and antenna, enter shelf article deploy RFID label tag can in localizing environment, in this environment, the position of shelf is fixed, and the y coordinate position namely respectively arranging shelf is known;

2) FRID label information on periphery article collected by RFID reader and antenna, and carries out position arithmetic operation by network afferent message processing module;

3) message processing module obtains the information of classification belonging to each RFID label tag, because target place to be positioned shelf y coordinate position is known, only need calculate the x coordinate position at target place to be positioned, get final product localizing objects.

Above-mentioned steps 3) several location algorithms of using are described below:

As shown in entire system deployment diagram in Fig. 3, because target object place to be positioned shelf y coordinate position is known, in this case, only need calculate x coordinate position, the position of target object to be positioned can be obtained.Therefore this problem is converted into and asks one-dimensional coordinate problem.

Algorithm one: based on sorting technique

This be one compared with based on solution.By scanning, reader can obtain a series of label information, and can obtain the information of classification belonging to label.Because the reference position x of each kind _i,sand end position x _i,eknown, use various types of in the number of tags n that is identified _ifor weighting weights, namely the method be averaging by weighting can calculate the position of target object.This computation process can be described with following formula:

$\hat{x}={\mathrm{\Σ}}_{i=1}^k{w}_i\·\frac{x_{i,s}+x_{i,e}}{2},$

$w_i=\frac{n_i}{{\mathrm{\Σ}}_{i=1}^k{n}_i}$

Here, k is the species number be identified, w _ibe the weights of corresponding classification, it obtains divided by all number of tags be identified by using the number of tags be identified in this kind.In this approach, the centre position of kinds of goods kind is used as the position of kind.And required by being.

Algorithm two: based on RSSI method

In a upper algorithm, only used the label number of the kinds of goods kind that each recognizes, but RSSI comes for the distribution of measurement label, is a very important factor first.Mention in the preceding article, the projector distance of distance antenna on identified region is nearer, and its RSSI value is higher.Here use total RSSI value of each kinds of goods kind as weights, computing method can be described below:

$\hat{x}={\mathrm{\Σ}}_{i=1}^k{w}_i\·\frac{x_{i,s}+x_{i,e}}{2},$

$w_i=\frac{s_i}{{\mathrm{\Σ}}_{i=1}^k{s}_i}$

Wherein, k represents that k the kinds of goods with the strongest RSSI mean value chosen are classified, s _ifor the mean value of the RSSI value of all labels in the kinds of goods classification of acquisition, w _ifor according to s _ithe each classification weights calculated, because the start-stop position of each classification is known, can calculate the positional information (Fig. 4) of mobile device according to this algorithm.

Algorithm three classification and matching algorithm

In scanning process, the number of tags of each classification the inside can be known, using these each tag along sort numbers obtained as a vector.The scope that all labelings be identified are covered is as possible position scope, from the left side of possible position scope, every △ d distance, just centered by possible position, outwards draw a circle with the radius between main cog region, we can obtain a series of number of tags vector table.The label vector calculated by the number of tags vector sum of match cognization, with crossing a k nearest neighbor algorithm, finds k most similar vector, then according to their possible position, estimates final position.

As shown in Figure 5, region shown in light circle is between main cog region, and its radius is r.According to scanning result, we can obtain a tag along sort number vector V ₀={ n _0,1, n _0,2..., n _{0, s}.Based on the relation training set T identifying radius and label densities and reader energy _a, we can obtain the radius r between the main cog region under present case, and in figure, region shown in dashed circle is between possible main cog region.Based on the relation training set T of identification label number and label densities and reader energy _b, we can calculate label densities ρ.With this dotted line ring box live part, each classification also can generate a tag along sort number vector V _i=(n _{i, 1}, n _{i, 2}..., n _i,s).By more vectorial V ₀with vectorial V _ibetween similarity, we can obtain k most similar vector, and obtain their possible position x ₁..., x _k, then passing through the weighting algorithm of a vector distance, we can calculate final position location.

Algorithm four is apart from Voting Algorithm

For solving based on the restriction in the average weighted algorithm of RSSI, we have done great many of experiments, under obtaining different label densities situation, and the relation training set T between RSSI intensity and label to antenna projector distance _d.By calculating each tag distances antenna projector distance, provide the anticipation position of each label for projected position.By calculating the center of these anticipation positions, be calculative target location.

Based on training set T _d, we can calculate the distance of each tag distance sky line projection by method of interpolation.We are with the centre position of labeling as the position d of label _i,j.Therefore each classification can calculate an anticipation position: wherein d _i,jless, its anticipation position is more accurate, and we are to the d calculated _i,jcarry out positive sequence sequence, k anticipation position before obtaining, calculates its mean value, can calculate final position location.Fig. 6 is this localization method schematic diagram, and Fig. 7 is this localization method process flow diagram.

Claims (6)

1., based on a RFID location system for non-precision anchor node, it is characterized in that comprising following assembly:

Shelf: for putting the kinds of goods posting RFID label tag, the position of shelf is fixed, and the start-stop position knowing kinds of goods classification on each shelf;

RFID label tag: be attached to the common RFID label tag comprising inventory information on kinds of goods;

Mobile device: it carries RFID reader and antenna, is promoted by user, by the location to mobile device, the positional information of user can be obtained;

RFID reader and antenna: for collecting the label information of kinds of goods, and by Internet Transmission to the message processing module in calculation server;

Message processing module: by the label information of the kinds of goods collected, calculates the position of mobile device.

2. described in claim 1 based on the localization method of the RFID location system of non-precision anchor node, it is characterized in that comprising the following steps:

1) mobile device of user's propelling loading RFID reader and antenna, enter shelf article deploy RFID label tag can in localizing environment, in this environment, the position of shelf is fixed, and the y coordinate position namely respectively arranging shelf is known;

2) FRID label information on periphery article collected by RFID reader and antenna, and carries out position arithmetic operation by network afferent message processing module;

3) message processing module obtains the information of classification belonging to each RFID label tag, because target place to be positioned shelf y coordinate position is known, only need calculate the x coordinate position at target place to be positioned, get final product localizing objects.

3. the localization method of the RFID location system based on non-precision anchor node according to claim 2, is characterized in that step 3) detailed process be:

Message processing module obtains the information of classification belonging to each RFID label tag, because the reference position x of each kind _i,sand end position x _i,eknown, use various types of in the number of tags n that is identified _ifor weighting weights, namely the method be averaging by weighting can calculate the x coordinate position of target, and the following formula of this computation process is described:

$\hat{x}={\mathrm{\Σ}}_{i=1}^k{w}_i\·\frac{x_{i,s}+x_{i,e}}{2},$

$w_i=\frac{s_i}{{\mathrm{\Σ}}_{i=1}^k{s}_i}$

Here, k is the species number be identified, w _ibe the weights of corresponding classification, it obtains divided by all number of tags be identified by using the number of tags be identified in this kind; In this approach, the centre position of kinds of goods kind is used as the position of kind.

4. the localization method of the RFID location system based on non-precision anchor node according to claim 2, is characterized in that step 3) detailed process be:

Message processing module obtains the information of classification belonging to each RFID label tag, because the reference position x of each kind _i,sand end position x _i,eknown, and the projector distance of each RFID label tag distance antenna on identified region is nearer, its RSSI value is higher, uses total RSSI value of each kinds of goods kind as weights here, namely the method be averaging by weighting can calculate the x coordinate position of target, and computing method are described below:

$\hat{x}={\mathrm{\Σ}}_{i=1}^k{w}_i\·\frac{x_{i,s}+x_{i,e}}{2},$

$w_i=\frac{s_i}{{\mathrm{\Σ}}_{i=1}^k{s}_i}$

Wherein, k represents that k the kinds of goods with the strongest RSSI mean value chosen are classified, s _ifor the mean value of the RSSI value of all labels in the kinds of goods classification of acquisition, w _ifor according to s _ithe each classification weights calculated, because the start-stop position of each classification is known, can calculate the positional information of mobile device according to this algorithm.

5. the localization method of the RFID location system based on non-precision anchor node according to claim 2, is characterized in that step 3) detailed process be:

Message processing module obtains the information of classification belonging to each RFID label tag, because the reference position x of each kind _i,sand end position x _i,eknown, meanwhile, in scanning process, the number of tags of each classification the inside can be known, using these each tag along sort numbers obtained as a vector, the scope that all labelings be identified cover is as possible position scope, from the left side of possible position scope, every △ d distance, just centered by possible position, a circle is outwards drawn with the radius between main cog region, a series of number of tags vector table can be obtained, the label vector calculated by the number of tags vector sum of match cognization, by a k nearest neighbor algorithm, find k most similar vector, again according to their possible position, estimate the x coordinate position of target.

6. the localization method of the RFID location system based on non-precision anchor node according to claim 2, is characterized in that step 3) detailed process be:

Under obtaining different RF ID label densities situation by experiment, the relation training set T between RSSI intensity and label to antenna projector distance _d, by calculating each tag distances antenna projector distance, provide the anticipation position of each label for projected position; By calculating the center of these anticipation positions, be calculative target location.