CN105072582A - Distance self-adapting wireless sensing network passive positioning method based on RSS distribution - Google Patents

Abstract

The present invention provides a distance self-adapting wireless sensing network passive positioning method based on RSS distribution. The method comprises a first step of deploying wireless sensor nodes in a monitoring region and determining reference distance and migration distance; a second step of dividing grids within a coverage area of communication links corresponding to the reference distance into reference grids and non-reference grids; according to the reference grids and the non-reference grids under the reference distance, determining reference grids and non-reference grids under the migration distance; acquiring an RSS value set of all the grids under the migration distance, obtaining an RSS set of non-sample links, and establishing a grid RSS priori knowledge base under the monitoring area; and a third step of carrying out positioning of targets to be measured according to the established grid RSS priori knowledge base under the monitoring area. The positioning method provided by the present invention is applicable to large-scale scenes, costs on node deployment and priori knowledge base acquisition are greatly lowered, accurate positioning of the targets is realized, interferences of environment noises are reduced by using target moving temporal correlation, and the positioning precision is raised.

Description

Based on the distance adaptive radio sensing network passive type localization method of RSS distribution

Technical field

The invention belongs to the applied technical field of wireless network, be specifically related to one and locate apart from adaptive Sensor Network passive type based on RSS distribution low cost, the method is applied to the passive type target following of the wireless sensor network of wild animal.

Background technology

How effective wild animal has important ecologic niche and ecological functions at occurring in nature, is one of link indispensable in whole ecological chain, carry out monitoring and protect it, seems particularly important.Traditional conservation of wildlife adopts manual type hand-kept, statistics; therefore; there is a lot of drawback in traditional approach; as: lack chronicity, real-time; also certain difficulty and danger is had; in addition, space-time isolates, and is difficult to the comprehensive analysis data obtained being carried out to time, space, phenomenon.And the appearance of current wireless sensor network, provide technical support for solving the problem.

Wireless sensor network is made up of a large amount of distributed sensor nodes being deployed in area to be monitored, it combines the multiple fields technology such as sensor technology, wireless communication technology, embedded technology and computer technology, by various types of transducer, extensive, long-term, real-time acquisition is carried out to information such as the character of material, the state of environment and behavior patterns, and in the mode of self-organizing, perception data is sent to remote data center by 802.15.4 communication protocol.Wherein, the location technology of wireless sensor network be wild animal event trace monitoring provide effective solution.

The challenge of 4 aspects below the movement locus surveying and mapping technology existence of wild animal:

1) sparse deployment.The living environment and habit of not disturbing wild animal are monitored in a basic demand of wild animal monitoring.Therefore the least possible equipment is used to be one of demand towards the conservation of wildlife, i.e. sparse deployment.

2) equipment has nothing to do.Existing most localization method all requires object Portable device to be positioned (as GPS module, RFID label tag), but for wild animal Portable device is not easy to accomplish, and animal protection expert does not advise doing so yet.Therefore need when target not Portable device to realize location be one of demand towards the conservation of wildlife, namely equipment has nothing to do.

3) computing cost.Existing method, to the mapping of target passive type track, is all first locate the estimated position of different time point target have been linked up track mapping again, needs to position calculating at each time point, exist and frequently locate the problem causing computing cost large.Therefore need to reduce that frequently to locate to wild animal the problem causing computing cost large be one of demand towards the conservation of wildlife, i.e. computing cost.

4) track is openness.The position of wild animal movement locus process has openness compared with the position of monitored area.Therefore in order under the condition not reducing mapping precision, reducing mapping track desired data amount and reducing energy consumption is one of demand towards the conservation of wildlife, and namely track is openness.

Up to now, in wireless sensor network, there are many location technologies, have substantially been divided into following 2 classes:

The first kind: active location, i.e. object Portable device.Sensor node evenly or random placement in locating area, signal that the equipment that object carries sends (as electromagnetic wave, infrared, ultrasonic wave etc.) can be detected by wireless sensor network, the signal sent at diverse location place equipment due to object is different, therefore the basic thought of these class methods is the changes being sent signal by checkout equipment, set up the respective function of signal intensity and position, and then object is positioned.If the people such as Kaltiokallio, Liu Yunhao are by the appearance of RSSI (ReceivedSignalStrengthIndicator) signal fluctuation detection target in wireless sensor network, and then position.The advantage of these class methods is positioning precision high (typically as GPS location), and because each object carries differentiable equipment, therefore Multi-target position is simple, is easy to add up destination number.But the shortcoming of the method needs target Portable device, the equipment do not met towards the conservation of wildlife has nothing to do demand.

Equations of The Second Kind: passive type is located.As the passive type based on RSS is located, usual way is that sensor node is evenly deployed in monitored area, and adjacent node communicates, and object is movable in region can cause interference to the node of two communications.Carry out quantifications demarcation by the radio signal RSS that is subject at diverse location place object interference, set up position and RSS value disturb between relation.When disturbed node receives the RSS value of one group of change, the position at object place can be released, as the location of the passive type based on study for representative is waited on a top.The advantage of these class methods is that equipment has nothing to do, and does not need object Portable device also can to target localization.But it is intensive that the shortcoming of the method is node deployment, and cost is high, do not meet the sparse deployment requirements towards the conservation of wildlife; Also have GPS location, camera location or CSI location in addition, but they have and need special installation, high in cost of production shortcoming, these all limit its application in real life, can not meet the demand of conservation of wildlife application.

As can be seen here, based on RSS passive type location because of its do not rely on its equipment carried, do not limit by particular device, the advantage such as with low cost enjoys people to favor, be widely used in reality.These characteristics also meet the needs of conservation of wildlife scene simultaneously.But positioning precision is normally improved by the quantity of the node increasing monitored area in the existing location of the passive type based on RSS, but because the deployment of node can produce dispose cost price and communication cost, relies on this method to there is many problems.Nodes is more, and cost price is larger; Along with the increase disposing nodes, will there is congested and collision in network.And monitoring area is larger, these problems are more obvious.

In addition, existing passive type localization method generally needs to set up in advance the priori storehouse under a special scenes, but the cost building priori storehouse is huge, and therefore when monitored area is very large, existing passive type localization method is just no longer applicable.

Summary of the invention

The defect existed for above-mentioned existing localization method or deficiency, the object of the invention is to, and proposes one and distribute low cost apart from adaptive Sensor Network passive type localization method, to realize positioning wild animal based on RSS.

In order to realize above-mentioned task, the present invention takes following technical solution:

Based on a distance adaptive radio sensing network passive type localization method for RSS distribution, comprise the following steps:

Step 1, disposes wireless sensor node in monitored area, chooses sample link, by sample link coverage area grid division, and according to the length determination reference distance of sample link and migration distance;

Step 2, the grid with reference to the overlay area of the communication link of correspondence is divided into the grid of reference and non-reference grid; According to the grid of reference under reference distance and non-reference grid, determine the grid of reference under migration distance and non-reference grid; Try to achieve the RSS value set of the grid of reference under migration distance and non-reference grid; Obtain the RSS value set of all the other grids under migration distance; Obtain the RSS set of non-sample link according to the RSS value set of all grids under migration distance, build grid RSS priori storehouse under monitored area;

Step 3, utilizes grid RSS priori storehouse under the monitored area built in step 2 to carry out the location of target to be measured.First, select the ballot link in data processing cycle t, the candidate lattices engraved when then selecting a certain in t, use the votes that multilink position voting method is determined on each candidate lattices, select votes maximum be target location to be measured.

Further, described step 1 comprises the steps:

Step 10: dispose wireless sensor node, aggregation node and a PC in monitored area, can communicate between adjacent wireless sensor node; The overlay area S setting every communication link is the rectangular area of long A=l, wide B=l/2, and l is communication link length, 0m < l < 12m;

Step 11: from monitored area often kind of length communication link in select arbitrary link, as the sample link under this length;

Step 12: the overlay area square net of communication link corresponding for selected every bar sample link is divided into multiple grid; Choosing the nodal distance that in monitored area, adjacent node spacing is minimum is reference distance, and other nodal distances are migration distance.

Further, described step 2 comprises the steps:

Step 20, the RSS value set of all grids under collection reference distance, the grid with reference to the overlay area of the communication link of distance correspondence is divided into the grid of reference and non-reference grid, builds the multiple linear regression model about the grid of reference and non-reference grid.

Step 21, according to the grid of reference in communication link overlay area corresponding under reference distance and non-reference grid, determines the grid of reference in the communication link overlay area that migration distance L` is corresponding and non-reference grid;

Step 22, measure the RSS value set of the grid of reference in communication link overlay area corresponding to migration distance L`, utilize the multiple linear regression model that step 20 obtains, obtain the RSS value set of the non-reference grid in communication link overlay area corresponding to migration distance L`;

Step 23, the RSS value set of the grid of reference and the RSS value set of non-reference grid in the communication link overlay area that the migration distance obtained according to step 22 is corresponding, obtain the RSS value set of all the other grids under migration distance L` (grid namely under migration distance L` outside the grid of reference and non-reference grid);

Step 24, according to the grid RSS value set of sample links all in monitored area, obtains the grid RSS value set of the non-sample link of other and sample link equal length, identical with sample link; In monitored area, the RSS value set of the grid of all communication links forms grid RSS priori storehouse under monitored area.

Further, described step 20 comprises the following steps:

Step 200, the RSS value set of all grids under collection reference distance;

Step 201, the grid with reference to the overlay area of the communication link of distance correspondence is divided into the grid of reference and non-reference grid, builds the multiple linear regression model about the grid of reference and non-reference grid; Specifically comprise two steps:

A) with reference to the overlay area S of communication link corresponding to distance L ^lgrid be divided into the grid of reference and non-reference grid.

The all grids chosen under reference distance in communication link sighting distance and on the cross region vertical with link sighting distance are the grid of reference, and remaining grid is non-reference grid; Q (q=0,1 ..., n1) the RSS value set of the individual grid of reference is kth (k=1,2 ..., n2) individual non-reference grid RSS set be

B) multiple linear regression model about the grid of reference and non-reference grid is built;

The xth time RSS value of the xth time RSS value of all grids of reference under the reference distance L gathered, all non-reference grids is substituted into formula 1, obtains the multiple linear regression coefficient matrix that xth is secondary; X=1,2 ..., X; Obtain X multiple linear regression coefficient matrix altogether; Thus obtain the multiple linear regression model of the grid of reference and non-reference grid:

$\left\{\begin{array}{c}{r}_{{\mathrm{nrg}}_1}^L={\mathrm{\&alpha;}}_0^1+{\mathrm{\&alpha;}}_1^1{r}_{{\mathrm{rg}}_1}^L+\mathrm{...}+{\mathrm{\&alpha;}}_q^1{r}_{{\mathrm{rg}}_q}^L+\mathrm{...}+{\mathrm{\&alpha;}}_{n1}^1{r}_{{\mathrm{rg}}_{n1}}^L+{\mathrm{\&xi;}}_1,\\ r_{{\mathrm{nrg}}_2}^L={\mathrm{\&alpha;}}_0^2+{\mathrm{\&alpha;}}_1^2{r}_{{\mathrm{rg}}_1}^L+\mathrm{...}+{{\mathrm{\&alpha;}}_q^{2}r}_{{\mathrm{rg}}_q}^L+\mathrm{...}+{\mathrm{\&alpha;}}_{n1}^2{r}_{{\mathrm{rg}}_{n1}}^L+{\mathrm{\&xi;}}_2,\\ \mathrm{...}\\ r_{{\mathrm{nrg}}_k}^L={\mathrm{\&alpha;}}_0^k+{\mathrm{\&alpha;}}_1^k{r}_{{\mathrm{rg}}_1}^L+\mathrm{...}+{\mathrm{\&alpha;}}_q^k{r}_{{\mathrm{rg}}_q}^L+\mathrm{...}+{\mathrm{\&alpha;}}_{n1}^k{r}_{{\mathrm{rg}}_{n1}}^L+{\mathrm{\&xi;}}_k,\\ \mathrm{...}\\ r_{{\mathrm{nrg}}_{n2}}^L={\mathrm{\&alpha;}}_0^{n2}+{\mathrm{\&alpha;}}_1^{n2}{r}_{{\mathrm{rg}}_1}^L+\mathrm{...}+{\mathrm{\&alpha;}}_q^{n2}{r}_{{\mathrm{rg}}_q}^L+\mathrm{...}+{\mathrm{\&alpha;}}_{n1}^{n2}{r}_{{\mathrm{rg}}_{n1}}^L+{\mathrm{\&xi;}}_{n2},\end{array}\right.$

In formula for certain RSS measured value once of a kth non-reference grid, k=1,2 ..., n2; be the RSS measured value of the correspondence time of q the grid of reference, q=0,1 ..., n1; for the multiple linear regression matrix coefficient of corresponding time; ξ _kfor noise jamming, ignore.

Further, the concrete operations of described step 21 are as follows:

Step 210, determines the grid of reference in the communication link overlay area that migration distance L` is corresponding;

With reference to the grid of reference under distance L from top to bottom from left to right number consecutively be 1,2 ..., n1}; By the cross area grid under migration distance L` from top to bottom from left to right number consecutively be 1,2 ..., n3};

The grid of reference under formula 2 computation migration distance L` is utilized to number

${\mathrm{Index}}_{rg}^{L^{`}}=\{1+j\&times;\&lsqb;n3/n1\&rsqb;,0\&le;j<n1\},$ (formula 2)

In formula, n1 is the number of the grid of reference under reference distance L; N3 is the number of the cross area grid under migration distance L`;

Step 211, determines the non-reference grid in the communication link overlay area that migration distance L` is corresponding;

With reference to the non-cross region under distance L grid from left to right snakelike number consecutively be 1,2 ..., n2}; By the non-cross area grid under migration distance L` from left to right snakelike number consecutively be; Non-reference grid numbering under utilizing formula 3 to calculate migration distance L`

${\mathrm{lndex}}_{nrg}^{L^{`}}=\{1+u\&times;\&lsqb;n4/n2\&rsqb;,0\&le;u<n2\},$ (formula 3)

In formula, n2 is the number of the non-reference grid under reference distance L; N4 is the number of the non-cross area grid under migration distance L`.

Further, described step 22 is specific as follows:

Gather respectively q the grid of reference under X migration distance L` RSS value (q=0,1 ... n1), obtain the RSS value set of q the grid of reference under migration distance L` by the xth of all grids of reference under the migration distance L` that records time (x=1,2, X) RSS value substitutes into the multiple linear regression model under the multiple linear regression coefficient matrix of the correspondence that obtains of step 20 time, calculates the value of the RSS of the kth non-reference grid under the migration distance L` of xth time finally obtain the RSS value set on non-reference grid k

$r_{{\mathrm{nrg}}_k}^{L^{`}}={\mathrm{\&alpha;}}_0^k+{\mathrm{\&alpha;}}_1^k{r}_{{\mathrm{rg}}_1}^{L`}+\mathrm{...}+{\mathrm{\&alpha;}}_q^k{r}_{{\mathrm{rg}}_q}^{L`}+\mathrm{...}+{\mathrm{\&alpha;}}_{n1}^k{r}_{{\mathrm{rg}}_{n1}}^{L`}+{\mathrm{\&xi;}}_k,$ (formula 4)

Wherein: for q the corresponding secondary RSS measured value of the grid of reference under migration distance, q=0,1 ..., n1; for the multiple linear regression matrix coefficient of corresponding time, k=1,2 ..., n2; ξ _kfor noise jamming, ignore.

Further, described step 23 comprises the steps:

Step 230, determines interpenetration network;

Using the grid outside the grid of reference in communication link overlay area corresponding for migration distance L` and non-reference grid as interpenetration network;

Step 231, calculates the RSS value set of grid each in interpenetration network by B-spline interpolation model; Comprise following two steps:

A) with the communication link overlay area S that migration distance L` is corresponding <sup>l`</sup>on a line grid be a B-spline interpolation operation, all row of communication link overlay area corresponding for migration distance L` are divided into two classes: the first kind be row in have the grid of reference and non-reference grid simultaneously; Another kind of is without any the grid of reference and non-reference grid in row;

B) the RSS value set of the interpenetration network in first kind row is calculated by B-spline interpolation model:

For every a line of the first kind, certain RSS value of grid of reference each in this row and the corresponding secondary RSS value of each non-reference grid are substituted into B-spline interpolation model, obtains the corresponding secondary RSS value of all interpenetration networks on this row; Repeat above-mentioned steps X time, obtain the RSS value set of each interpenetration network on this row;

C) by carrying out B-spline interpolation model to each row of overlay area, the set of RSS migration distance remaining interpenetration network is calculated.

Further, described step 3 comprises the steps:

Step 30, when recording driftlessness in monitored area, RSS value on arbitrary communication link, is called RSS value ideally

Step 31, the RSS value of all sampling instants on the every communication link in image data treatment cycle t, obtains the RSS value set of this communication link in data processing cycle t;

Step 32, selects d moment t _dall ballot links in upper monitored area, inscribe the fluctuation set on every bar ballot link, t when determining this _d∈ t;

Step 33, at t _don every bar ballot link on select the candidate lattices of target to be measured, thus select t _don all ballot links on the candidate lattices of target to be measured;

Step 34, uses multilink position voting method, under all ballot links in d moment target to be measured candidate lattices in select the grid at target place to be measured, obtain target location to be measured.

Further, described step 32 comprises the steps:

A. d moment t _dunder, all communication link p in monitored area _iin all ballot link vp of engraving when selecting this _i;

The communication link p meeting formula 5 will be engraved during td _ias t _dtime the ballot link vp that engraves _i, and the link vp that will vote _ion RSS value be defined as

$|r^{P_i}\left(t_d\right)-r_{state}^{P_i}|\&GreaterEqual;\mathrm{\&epsiv;}$ (formula 5)

Wherein, threshold epsilon=2; for communication link p _iat t _dthe RSS value in moment;

The fluctuation set on each article of ballot link is inscribed when B. determining d;

Determine t _dmoment votes link vp _iafter, definition t _dtime the ballot link vp that engraves _ithe RSS value set gathered in data processing cycle t is $R^{{\mathrm{vp}}_i}=\&lsqb;r^{{\mathrm{vp}}_i}\left(t_1\right),r^{{\mathrm{vp}}_i}\left(t_2\right),\mathrm{...},r^{{\mathrm{vp}}_i}\left(t_d\right),\mathrm{...},r^{{\mathrm{vp}}_i}\left(t_j\right)\&rsqb;,$ J is the ballot link vp gathered in data processing cycle t ⁱthe number of RSS value, d ∈ [1,2,3 ..., j], t _dfor d the moment in t; Then will gather in each value substitute into formula 6 successively, will formula 6 be met and comprise moment t _dcontinuous element as fluctuation set element;

$|r^{{\mathrm{vP}}_i}\left(t_d\right)-r_{state}^{{\mathrm{vP}}_i}|\&GreaterEqual;\mathrm{\&epsiv;}$ (formula 6)

Wherein, threshold epsilon=2; for the link vp that votes when driftlessness occurs in monitored area _irSS value;

Further, described step 34 comprises the steps:

Step 340, based on long distance ballot link ballot, obtain target rough position; Concrete steps are as follows:

A. utilize formula 9, obtain long distance ballot link the RSS degree of fluctuation engraved when td;

${\mathrm{\&Delta;r}}^{{\mathrm{vp}}_i^{lo}}=|r_{object}^{{\mathrm{vp}}_i^{lo}}\left(t_d\right)-r_{state}^{{\mathrm{vP}}_i}|/\left|r_{state}^{{\mathrm{vP}}_i}\right|$ (formula 9)

Wherein, refer to t _dtime engrave long distance ballot link on RSS value, long distance ballot link refers in step 32 and obtains t _dtime the ballot link that engraves in length l meet the link of 6m < l < 12m; for the RSS value of duration distance ballot link appears in driftlessness in monitored area;

B. formula 10 is utilized, calculated candidate grid po _hat t _dtime inscribe long distance ballot link on the first pre-votes:

${\mathrm{Vote}}_{{\mathrm{vp}}_i^{lo}}{(t_d,{\mathrm{po}}_h)}_p=\left|\right|{\mathrm{\&Delta;r}}^{{\mathrm{vp}}_i^{lo}}|{|}^2$ (formula 10);

C. candidate lattices po is calculated _hthe second pre-votes in the td moment:

${\mathrm{Vote}}_{{\mathrm{vp}}_i^{lo}}{(t_d,{\mathrm{po}}_h)}_e={\mathrm{Vote}}_{{\mathrm{vp}}_i^{lo}}(t_{d-1},{\mathrm{po}}_h)e^{-\mathrm{\&lambda;}(t_d-t_{d-1})}$ (formula 11)

Wherein, refer to t _d-1candidate lattices po under moment long distance ballot link _hthe votes obtained; Cooling ratio λ=0.025;

D. according to the first pre-votes and the second pre-votes, formula 12 is utilized to obtain t _dmoment candidate lattices po _hin long-distance link on the votes that obtains:

${\mathrm{Vote}}_{{\mathrm{vp}}_i^{lo}}(t_d,{\mathrm{po}}_h)=({\mathrm{Vote}}_{{\mathrm{vp}}_i^{lo}}{\left(t_d,{\mathrm{po}}_h\right)}_p-{\mathrm{Vote}}_{{\mathrm{vp}}_i^{lo}}{\left(t_d,{\mathrm{po}}_h\right)}_e)\&times;p+{\mathrm{Vote}}_{{\mathrm{vp}}_i^{lo}}{(t_d,{\mathrm{po}}_h)}_e$ (formula 12)

Wherein, p refers to the fluctuating quantity of signal, utilizes formula 13 to calculate:

$p=\frac{num}{{\max }_{num}}=\frac{2L\&times;size\left(R_{\&GreaterEqual;}^{{\mathrm{vp}}_i^{lo}}\right)}{vf}$ (formula 13)

Wherein, num is fluctuation set the number of element; max _numformula 14 is utilized to obtain:

${\max }_{num}=\frac{2l/v}{f}$ (formula 14)

Wherein, v refers to the speed of target movement to be measured, is set as known quantity; For the length of long distance ballot link; F refers to the sample frequency of node, i.e. the inverse in sampling period;

If do not exist, i.e. t _dfor the first poll moment in t, then calculate according to formula 15

${\mathrm{Vote}}_{{\mathrm{vp}}_i^{lo}}(t_d,{\mathrm{po}}_h)={\mathrm{Vote}}_{{\mathrm{vp}}_i^{lo}}{(t_d,{\mathrm{po}}_h)}_p$ (formula 15)

E. repeat steps A-D, obtain td moment candidate lattices po _hthe votes in monitored area, each long-distance link obtained:

F. formula 16 is utilized, t step e obtained _dmoment candidate lattices po _hin monitored area, the votes of all long distance ballot links is added, and obtains t _dmoment candidate lattices po _htotal votes on long distance ballot link;

$Vote(t_d,{\mathrm{po}}_h)={\mathrm{\&Sigma;}}_i{\mathrm{Vote}}_{{\mathrm{vp}}_i^{lo}}(t_d,{\mathrm{po}}_h)$ (formula 16)

G. repeat the candidate lattices process of steps A-F to target to be measured on td moment all ballots link, obtain total votes of each candidate lattices on long distance ballot link; Select total votes maximum time corresponding candidate lattices as region, target location, this region comprises multiple candidate lattices, is rough target location;

Step 341, in the target rough position region that step G obtains, chooses short distance ballot link, performs step 340, obtain target location after then the long distance ballot link in step 340 being replaced with described short distance ballot link; Short distance ballot link refers in step 32 and obtains t _dtime the ballot link that engraves in length l meet the link of 0m < l≤6m.

The present invention compared with prior art has following advantage:

1. localization method of the present invention does not need special installation, with low cost, has wide range of applications, and is especially applicable to large scale scene application.

2. the present invention is a dynamic migration algorithm that can adapt to different nodal distance, the migration of nodal distance from L to L` is completed with minimum cost by building multiple linear regression model, obtain the priori storehouse on distance L`, there is very high flexibility and practicality.Compare with existing localization method, greatly can reduce the cost in node deployment and acquisition priori storehouse.

3. the present invention uses the accurate location of multiple distance and position Voting Algorithm realize target, and utilizes the temporal correlation of target movement to reduce the interference of ambient noise, improves positioning precision.

4. the low-cost and high-precision passive type localization method of the present invention target following of large scale scene of surviving for wild animal is very applicable.

5. obtain the RSS priori storehouse of all grids in monitored area under present invention can be implemented in low measurement cost, and can higher positioning precision be ensured.

Accompanying drawing explanation

Fig. 1 is the stress and strain model schematic diagram under the different nodal distance of single-link respectively, wherein, Fig. 1 (a) is reference under reference distance and non-reference stress and strain model schematic diagram, and Fig. 1 (b) is reference under migration distance and non-reference stress and strain model schematic diagram; Fig. 1 (c) is the reference under migration distance, non-reference and interpenetration network divide schematic diagram;

Fig. 2 is ballot location algorithm example schematic, determines the position of target in the ti moment by long-distance link ballot and the ballot of short-distance link ballot zero degree;

Fig. 3 is that the positioning precision chosen under the different grid of reference compares schematic diagram;

Fig. 4 is that the priori storehouse that obtains with reference to distance L=4m is directly as the priori storehouse contrast locating comparatively schematic diagram of migration distance L`=6m, 8m;

Fig. 5 is the priori storehouse that obtains with reference to distance L=4m through the comparison schematic diagram that the process of this patent method obtains migration distance L`=6m, then the priori storehouse of 8m positions;

Fig. 6 is that the time cost under different nodal distance compares schematic diagram;

Fig. 7 is the comparison schematic diagram under different localization method in positioning precision;

Fig. 8 is that the lower deployment cost of different monitoring scale compares schematic diagram.

Below in conjunction with the drawings and specific embodiments, further explanation is explained to the present invention.

Embodiment

Applicant is in the conservation of wildlife; in order to study the mechanics of wild animal in wild environment; need to obtain the positional information of wild animal in the wild in region, therefore, propose of the present inventionly to distribute low cost apart from adaptive Sensor Network passive type localization method based on RSS.Basic ideas are: the deployment of (1) wireless sensor network and data acquisition; (2) the RSS value of all grids under recording reference distance; (3) multiple linear regression model is built by the RSS value recorded; (4) the distribution RSS of all grids utilizing multiple linear regression model and B-spline interpolation model to obtain under migration distance distributes, thus obtains the RSS priori storehouse on monitored area; (4) use multi-path Voting Algorithm localizing objects position, comprising: the ballot of long link and the ballot on short chain road, finally obtain a comparatively accurate positioning result; (5) obtain the preferred version of this method by experiment, and itself and existing localization method are contrasted with regard to node deployment cost and positioning precision, evaluate validity of the present invention.

One, the distance adaptive radio sensing network passive type localization method based on RSS distribution of the present invention, comprises the following steps:

Step 1, disposes wireless sensor node in monitored area, chooses sample link, by sample link coverage area grid division, and according to the length determination reference distance of sample link and migration distance.Concrete operations are as follows:

Step 10: dispose wireless sensor node

In the C of monitored area, dispose M wireless sensor node, can communicate between adjacent wireless sensor node; F communication link is formed altogether in the C of monitored area; The overlay area S setting every communication link is the rectangular area of long A=l, wide B=l/2, and l is communication link length, l ∈ D (D is the set of adjacent node distance); Record through test, when nodal distance is greater than 12m, the RSS value on grid can accurately not reflect the situation of target, therefore specifies 0m < l < 12m; Arrange all wireless sensor nodes distance ground level and be H, the repetition test according to applicant draws, has good signal propagation characteristics as nodal distance ground height H=0.95m.

Each wireless sensor node is forwarded to aggregation node after receiving data; Aggregation node is less than 70 meters from the distance of monitored area C, carries out preserving, analyze and processing for the PC by network, Serial Port Transmission or GPRS data being reached base.

In order to avoid packet loss or node conflict, through test, determine that the sampling period (namely node receives and dispatches a secondary data every a sampling period) of wireless sensor node is 2s; PC interval ts (the desirable arbitrary value being greater than 2 of t, as 50) processes the RSS data once collected, and t is data processing cycle.

Step 11: choose sample link

Because RSS value is relevant with communication link length, so for the communication link of often kind of length in the C of monitored area, only need select arbitrary link under this length as the sample link under this length, below use sample link to build the RSS priori storehouse of grid in this length, and non-sample link distribute identical with the RSS of the sample link of its equal length.

Step 12: by sample link coverage area grid division, choose the reference distance on monitored area and migration distance;

The square net of the overlay area S length of side ω=0.5m of communication link corresponding for selected every bar sample link is divided into N number of grid; Choosing the nodal distance (length on the most short chain road namely in sample link) that in the C of monitored area, adjacent node spacing is minimum is reference distance, and other nodal distances length of other links (namely in sample link) are migration distance.

Follow above-mentioned steps, inventor is on the playground of school's spaciousness, the open area choosing 28m × 25m carries out the target localization experiment of real scene as monitored area, 26 wireless sensor nodes (Micaz node) are deployed in monitored area, distance (i.e. the length of a communication link) between any pair adjacent communication node is l ∈ D{4m, 6m, 8m}.Setting nodal distance (length on the most short chain road in the sample link) l=L=4m that in this monitored area, adjacent node spacing is minimum is reference distance, and L`=6m, 8m are migration distance.

Step 2, the grid with reference to the overlay area of the communication link of correspondence is divided into the grid of reference and non-reference grid; According to the grid of reference under reference distance and non-reference grid, determine the grid of reference under migration distance and non-reference grid; Try to achieve the RSS value set of the grid of reference under migration distance and non-reference grid; Obtain the RSS value set of all the other grids under migration distance; The RSS set of non-sample link is obtained according to the RSS value set of all grids under migration distance.

From above, the RSS value set R of grid under different nodal distance l in monitored area only need be obtained ^lthe priori storehouse under this region can be built.But in actual large scale deployment, in monitored area, the set D of all communication node spacing l is very huge, want to record respectively the RSS distribution of each grid under different nodal distance l, cost is very huge, and the RSS priori storehouse that how can obtain all grids in monitored area under low measurement cost is a challenge.The concrete operations of step 2 are as follows:

Step 20, the RSS value set of all grids under collection reference distance, the grid with reference to the overlay area of the communication link of distance correspondence is divided into the grid of reference and non-reference grid, builds the multiple linear regression model about the grid of reference and non-reference grid; Specifically comprise the following steps:

Step 200, the RSS value set of all grids under collection reference distance;

Target stands in each grid under reference distance successively, gathers X measurement result as corresponding grid for same grid ( s=1,2 ..., N, Po ^lset for the grid under reference distance) RSS value set thus the RSS value set of all grids under obtaining reference distance; In this experiment, X=80;

Step 201, the grid with reference to the overlay area of the communication link of distance correspondence is divided into the grid of reference and non-reference grid, builds the multiple linear regression model about the grid of reference and non-reference grid; Specifically comprise following two steps:

A) with reference to the overlay area S of communication link corresponding to distance L ^lgrid be divided into the grid of reference and non-reference grid.

When target is in node link sighting distance, its corresponding RSS value change obviously, therefore all grids chosen under reference distance in communication link sighting distance and on the cross region vertical with link sighting distance are that the grid of reference is (as cross region grayish in Fig. 1 (a), be numbered 1,2..., n1), remaining grid is that non-reference grid (as the region of Fig. 1 (a) Oxford gray, is numbered 1,2..., n2); Q (q=0,1 ..., n1) the RSS value set of the individual grid of reference is (this is because get X measurement result for same grid, so comprise X RSS value, in experiment, X=80), kth (k=1,2 ..., n2) the RSS set of individual non-reference grid is ( comprise X RSS value, in experiment, X=80).

If be not high especially to positioning accuracy request, also can only choose in link sighting distance or only choose vertical with link sighting distance on grid be the grid of reference, to reduce lower deployment cost.

B) multiple linear regression model about the grid of reference and non-reference grid is built;

The xth time RSS value of the xth time RSS value of all grids of reference under reference distance L step 20 gathered, all non-reference grids substitutes into formula 1, obtains the multiple linear regression coefficient matrix that xth is secondary; X=1,2 ..., X; Obtain X multiple linear regression coefficient matrix altogether; Thus obtain the multiple linear regression model of the grid of reference and non-reference grid.

$\left\{\begin{array}{c}{r}_{{\mathrm{nrg}}_1}^L={\mathrm{\&alpha;}}_0^1+{\mathrm{\&alpha;}}_q^1{r}_{{\mathrm{rg}}_q}^L+\mathrm{...}+{\mathrm{\&alpha;}}_q^1{r}_{{\mathrm{rg}}_q}^L+\mathrm{...}+{\mathrm{\&alpha;}}_{n1}^1{r}_{{\mathrm{rg}}_{n1}}^L+{\mathrm{\&xi;}}_1,\\ r_{{\mathrm{nrg}}_2}^L={\mathrm{\&alpha;}}_0^2+{\mathrm{\&alpha;}}_1^2{r}_{{\mathrm{rg}}_1}^L+\mathrm{...}+{\mathrm{\&alpha;}}_q^2{r}_{{\mathrm{rg}}_q}^L+\mathrm{...}+{\mathrm{\&alpha;}}_{n1}^2{r}_{{\mathrm{rg}}_{n1}}^L+{\mathrm{\&xi;}}_2,\\ ...\\ r_{{\mathrm{nrg}}_k}^L={\mathrm{\&alpha;}}_0^k+{\mathrm{\&alpha;}}_1^k{r}_{{\mathrm{rg}}_1}^L+\mathrm{...}+{\mathrm{\&alpha;}}_q^k{r}_{{\mathrm{rg}}_q}^L+\mathrm{...}+{\mathrm{\&alpha;}}_{n1}^k{r}_{{\mathrm{rg}}_{n1}}^L+{\mathrm{\&xi;}}_k,\\ ...\\ r_{{\mathrm{nrg}}_{n2}}^L={\mathrm{\&alpha;}}_0^{n2}+{\mathrm{\&alpha;}}_1^{n2}{r}_{{\mathrm{rg}}_1}^L+\mathrm{...}+{\mathrm{\&alpha;}}_q^{n2}{r}_{{\mathrm{rg}}_q}^L+\mathrm{...}+{\mathrm{\&alpha;}}_{n1}^{n2}{r}_{{\mathrm{rg}}_{n1}}^L+{\mathrm{\&xi;}}_{n2},\end{array}\right.$ (formula 1)

In above formula, for certain RSS measured value once of a kth non-reference grid, k=1,2 ..., n2; be the RSS measured value of the correspondence time of q the grid of reference, q=0,1 ..., n1; for the multiple linear regression matrix coefficient of corresponding time; ξ _kfor interference such as noises, ξ in the present invention _kignore;

Step 21, according to the grid of reference in communication link overlay area corresponding under reference distance and non-reference grid, determines the grid of reference in the communication link overlay area that migration distance L` is corresponding and non-reference grid;

Because migration distance > reference distance, and the grid length and width divided are identical, thus divide under migration distance the lattice number obtained be greater than reference distance L under lattice number.In the present invention, conveniently utilize multiple linear regression model, the grid of reference number chosen under regulation migration distance and the number of non-reference grid equal the number of grid of reference number under reference distance and non-reference grid respectively.

Concrete operations are as follows:

Step 210, determines the grid of reference in the communication link overlay area that migration distance L` is corresponding;

With reference to the grid of reference (grid in cross region) under distance L from top to bottom from left to right number consecutively be 1,2 ..., n1}; By the cross area grid (not being the grid of reference entirely) under migration distance L` from top to bottom from left to right number consecutively for 1,2 ..., n3} (as Suo Shi Fig. 1 (c));

The grid of reference under formula 2 computation migration distance L` is utilized to number

${\mathrm{Index}}_{rg}^{L^{`}}=\{1+j\&times;\&lsqb;n3/n1\&rsqb;,0\&le;j<n1\},$ (formula 2)

In formula, n1 is the number of the grid (grid of reference) in cross region under reference distance L; N3 is the number of the cross area grid under migration distance L`;

Step 211, determines the non-reference grid in the communication link overlay area that migration distance L` is corresponding;

With reference to the non-cross region under distance L grid (non-reference grid) from left to right snakelike number consecutively be 1,2 ..., n2}; By the non-cross area grid (not being the non-grid of reference entirely) under migration distance L` from left to right snakelike number consecutively for 1,2 ..., n4} (as Suo Shi Fig. 1 (c)); Non-reference grid numbering under utilizing formula 3 to calculate migration distance L`

${\mathrm{Index}}_{nrg}^{L^{`}}=\{1+u\&times;\&lsqb;n4/n2\&rsqb;,0\&le;u<n2\},$ (formula 3)

In formula, n2 is the number of the grid (non-reference grid) in non-cross region under reference distance L; N4 is the number of the non-cross area grid under migration distance L`.

Step 22, measure the RSS value set of the grid of reference in communication link overlay area corresponding to migration distance L`, utilize the multiple linear regression model that step 20 obtains, obtain the RSS value set of the non-reference grid in communication link overlay area corresponding to migration distance L`;

Gather respectively q the grid of reference under X migration distance L` RSS value (q=0,1 ... n1), obtain the RSS value set of q the grid of reference under migration distance L` (owing to getting X measurement result for same grid, so comprise X RSS value, in experiment, X=80), by the xth of all grids of reference under the migration distance L` that records time (x=1,2,, X) and RSS value substitutes into multiple linear regression model under the multiple linear regression coefficient matrix of the correspondence that obtains of step 20 time, calculates the value of the RSS of the kth non-reference grid under the migration distance L` of xth time finally obtain the RSS value set on non-reference grid k

$r_{{\mathrm{nrg}}_k}^{L^{`}}={\mathrm{\&alpha;}}_0^k+{\mathrm{\&alpha;}}_1^k{r}_{{\mathrm{rg}}_1}^{L^{`}}+\mathrm{...}+{\mathrm{\&alpha;}}_q^{k^{`}}{r}_{{\mathrm{rg}}_q}^{L^{`}}+\mathrm{...}+{\mathrm{\&alpha;}}_{n1}^k{r}_{{\mathrm{rg}}_{n1}}^{L^{`}}+{\mathrm{\&xi;}}_k,$ (formula 4)

Wherein: for q the corresponding secondary RSS measured value of the grid of reference under migration distance, q=0,1 ..., n1; for the multiple linear regression matrix coefficient of corresponding time, k=1,2 ..., n2; ξ _kfor interference such as noises, ignore in the present invention;

Step 23, the RSS value set of the grid of reference and the RSS value set of non-reference grid in the communication link overlay area that the migration distance obtained according to step 22 is corresponding, obtain the RSS value set of all the other grids under migration distance L` (grid namely under migration distance L` outside the grid of reference and non-reference grid);

The RSS distribution situation of the grid of reference under migration distance L` and non-reference grid has been obtained according to step 22, but because reference distance L< migration distance L`, in the migration distance L` that can not be drawn by above-mentioned steps total-grid RSS distribution, that how to obtain migration distance L` be left co-net lattice RSS distribution be another need solve problem.In order to solve this problem, the present invention introduces B-spline interpolation model.In B-spline interpolation model, curve has the continuous print gradient and curvature at tie point place, and segmentation polynomial of lower degree has the function interpolation of certain slickness in segmentation place.Because the distribution of RSS is in the horizontal direction and the vertical direction sectionally smooth, and tie point is also smooth, therefore can use the RSS value of all the other grids under B-spline interpolation model computation migration distance L`.

Step 230, determines interpenetration network;

Entirety due to grid under migration distance L` can be regarded as and insert K=S in the horizontal direction and vertical direction of communication link overlay area corresponding to reference distance L <sub>2</sub>× T <sub>2</sub>-S <sub>1</sub>× T <sub>1</sub>individual grid obtains (as Suo Shi Fig. 1 (c)); S <sub>1</sub>, T <sub>1</sub>be respectively the level of communication link overlay area corresponding to reference distance and the grid number of vertical direction, S <sub>2</sub>, T <sub>2</sub>be respectively the level of communication link overlay area corresponding to migration distance and the grid number of vertical direction, K is the lattice number outside the grid of reference of communication link overlay area corresponding to migration distance and non-reference grid; Therefore, the present invention using the grid outside the grid of reference in communication link overlay area corresponding for migration distance L` and non-reference grid as interpenetration network (in Fig. 1 (c) be interpenetration network with the grid of oblique line);

Step 231, calculates the RSS value set of grid each in interpenetration network by B-spline interpolation model; Specifically comprise following two steps:

A) with the communication link overlay area S that migration distance L` is corresponding <sup>l`</sup>on a line grid be a B-spline interpolation operation, as Fig. 1 (c), all row of communication link overlay area corresponding for migration distance L` are divided into two classes: the first kind be row in have the grid of reference and non-reference grid simultaneously; Another kind of is without any the grid of reference and non-reference grid in row.

B) the RSS value set of the interpenetration network in first kind row is calculated by B-spline interpolation model;

For every a line of the first kind, certain RSS value of grid of reference each in this row and the corresponding secondary RSS value of each non-reference grid are substituted into B-spline interpolation model, obtains the corresponding secondary RSS value of all interpenetration networks on this row; Repeat above-mentioned steps X time, obtain the RSS value set (in this experiment X=80) of each interpenetration network on this row.

C) by carrying out B-spline interpolation model to each row of overlay area, the set of RSS migration distance remaining interpenetration network is calculated;

Capable for above-mentioned Equations of The Second Kind, due in such row without any the grid of reference and non-reference grid, so can not directly apply B-spline interpolation model on such row; So we pass through overlay area S ^lin each row carry out the calculating of B-spline interpolation model, the RSS value set remaining interpenetration network in this region can be obtained.Each is classified as a B-spline interpolation operation; Substitute into B-spline interpolation model by a certain row by the RSS value of the non-reference grid of time grid of reference of the correspondence on the interpenetration network obtained in step B RSS value and this row and correspondence time, the corresponding secondary RSS value of all residue interpenetration networks on these row can be obtained; Repeat above-mentioned steps X time, the RSS value set (times of collection X=80) of the interpenetration network of all the unknowns on these row can be obtained, thus grid RSS complete under obtaining migration distance L` distributes.

By above-mentioned steps, obtain the RSS value set of all grids under migration distance;

Step 24, according to the grid RSS value set of sample links all in monitored area, obtains the grid RSS value set of the non-sample link of other and sample link equal length, identical with sample link; In monitored area, the RSS value set of the grid of all communication links (comprising sample link and non-sample link) forms grid RSS priori storehouse under monitored area.

Constructed the priori storehouse of grid under monitored area above by step one and step 2, below the present invention according to priori storehouse, use Voting Algorithm to position the target to be measured entered in monitored area.

Step 3, utilizes grid RSS priori storehouse under the monitored area built in step 2 to carry out the location of target to be measured.First, select the ballot link in data processing cycle t, the candidate lattices engraved when then utilizing Bayes classifier to select a certain in t, uses the votes that multilink position voting method is determined on each candidate lattices, select votes maximum be target location to be measured.Specific as follows:

Step 30, when recording driftlessness in monitored area, RSS value on arbitrary communication link, is called RSS value ideally

For a communication link p any in the C of monitored area _i(i ∈ F), when in monitored area, driftlessness occurs, maintenance is stablized by the RSS value of this communication link, and claim this one-phase to be perfect condition here, the RSS value that now this communication link is corresponding is obtain the RSS value that F communication link is corresponding altogether.

Step 31, the RSS value of all sampling instants on the every communication link in image data treatment cycle t, obtains the RSS value set of this communication link in data processing cycle t.

I-th communication link P in definition monitored area _ion the RSS value set that collects in data processing cycle t be $R^{p_i}=\&lsqb;r^{p_i}\left(t_1\right),r^{p_i}\left(t_2\right),\mathrm{...},r^{p_i}\left(t_d\right),\mathrm{...},r^{p_i}\left(t_j\right)\&rsqb;,$ J is the number of the RSS value of the i-th communication link gathered in data processing cycle t, d ∈ 1,2,3 ..., j ］, t _dfor d the moment in t;

Step 32, selects d moment t _d(t _d∈ t) all ballot links in upper monitored area, inscribe the fluctuation set on every bar ballot link when determining this; Concrete steps are as follows:

A. d moment t _dunder, all communication link p in monitored area _iin all ballot link vp of engraving when selecting this _i;

The communication link p meeting formula 5 will be engraved during td _ias t _dtime the ballot link vp that engraves _i, and the link vp that will vote _ion RSS value be defined as

$|r^{p_i}\left(t_d\right)-r_{state}^{p_i}|\&GreaterEqual;\mathrm{\&epsiv;}$ (formula 5)

Wherein, threshold epsilon=2; for communication link p _iat t _dthe RSS value in moment;

The fluctuation set on each article of ballot link is inscribed when B. determining d;

Determine t _dmoment votes link vp _iafter, definition t _dtime the ballot link vp that engraves _ithe RSS value set gathered in data processing cycle t is $R^{{\mathrm{vp}}_i}=\&lsqb;r^{{\mathrm{vp}}_i}\left(t_1\right),r^{{\mathrm{vp}}_i}\left(t_2\right),\mathrm{...},r^{{\mathrm{vp}}_i}\left(t_d\right),\mathrm{...},r^{{\mathrm{vp}}_i}\left(t_j\right)\&rsqb;,$ J is the ballot link vp gathered in data processing cycle t _ithe number of RSS value, d ∈ [1,2,3 ..., j], t _dfor d the moment in t; Then will gather in each value substitute into formula 6 successively, will formula 6 be met and comprise moment t _dcontinuous element as fluctuation set element; (fluctuation is integrated in step 320 hereafter (B) and uses)

$|r^{{\mathrm{vP}}_i}\left(t_d\right)-r_{state}^{{\mathrm{vP}}_i}|\&GreaterEqual;\mathrm{\&epsiv;}$ (formula 6)

Wherein, threshold epsilon=2; for (perfect condition) when in monitored area, driftlessness occurs votes link vp _irSS value;

Step 33, at t _dtime the every bar ballot link that engraves on select the candidate lattices of target to be measured, thus select t _dtime all ballot links of engraving on the candidate lattices of target to be measured;

By the Bayes classifier shown in formula 7, target to be measured can be calculated at t _dtime the ballot link vp that engraves _igrid in overlay area interior probability po ^lballot link vp _ithe set of upper grid; S=1,2 ..., N, (l represents ballot link vp _ilength); Relatively ballot link vp _ilower target to be measured appears at the probability of each grid size, select maximum probability be t by mesh definition corresponding for maximum probability _dtime engrave this ballot link vp _ithe candidate lattices of lower target to be measured

$P\left({\mathrm{po}}_{g_s}^{{\mathrm{vp}}_i}\right|r_{object}^{{\mathrm{vp}}_i}\left(t_d\right))=\frac{P\left(r_{object}^{{\mathrm{vp}}_i}\right(t_d\left)\right|{\mathrm{po}}_{g_s}^{{\mathrm{vp}}_i})P\left({\mathrm{po}}_{g_s}^{{\mathrm{vp}}_i}\right)}{P\left(r_{object}^{{\mathrm{vp}}_i}\right(t_b\left)\right)}$ (formula 7)

Wherein, t _dthe ballot link vp in moment _ion RSS value; represent grid rSS set in comprise probability; RSS gathers refer in priori storehouse with ballot link vp _iequal length sample link on corresponding grid g _srSS value set; refer to that target to be measured appears at grid g _sprobability;

Hypothetical target is at the probability of each grid equal, then it is constant; Because with once relatively in identical, then also identical, therefore, can be by when target more to be measured appears at the probability of different grid with ignore, then maximum probability can directly represent with formula 8:

$\arg \max P\left({\mathrm{po}}_{g_s}^{{\mathrm{vp}}_i}\right|r_{object}^{{\mathrm{vp}}_i}\left(t_d\right))=\arg \max P\left(r_{object}^{{\mathrm{vp}}_i}\right(t_d\left)\right|{\mathrm{po}}_{g_s}^{{\mathrm{vp}}_i})$ (formula 8);

In addition, from diffraction theory, the distribution of the RSS value on a pair Symmetric Mesh is almost identical, so more than one of the grid possibility that under this ballot link, maximum probability is corresponding.

In d the moment t that step 32 obtains _d(t _d∈ t) every bar ballot link in all ballot links in upper monitored area repeats step 33, obtain the candidate lattices po of the target to be measured under all ballot links in d moment _h, h=1,2 ..., H, H are the number of candidate lattices.

Step 34, uses multilink position voting method, under all ballot links in d moment target to be measured candidate lattices in select the grid at target place to be measured, obtain target location to be measured.

Distribution due to the RSS value on a pair Symmetric Mesh is almost identical, the ballot of known single-link can only determine the Position Approximate of target to be measured, the exact position of target can't be determined, so the present invention obtains the accurate location of target by the method voted in multilink position.

Multilink position voting method comprises two stage ballots, namely based on the ballot of long distance with based on short-range ballot:

Step 340, based on long distance ballot link ballot, obtain target rough position; Concrete steps are as follows:

A. utilize formula 9, obtain long distance ballot link the RSS degree of fluctuation engraved when td;

${\mathrm{\&Delta;r}}^{{\mathrm{vp}}_i^{lo}}=|r_{object}^{{\mathrm{vp}}_i^{lo}}\left(t_d\right)-r_{state}^{{\mathrm{vp}}_i}|/\left|r_{state}^{{\mathrm{vp}}_i}\right|$ (formula 9)

Wherein, refer to t _dtime engrave long distance ballot link on RSS value, long distance ballot link refers in step 32 and obtains t _dtime the ballot link that engraves in length l meet the link of 6m < l < 12m; for the RSS value of (perfect condition) when driftlessness occurs in monitored area long distance ballot link;

B. formula 10 is utilized, calculated candidate grid po _hat t _dtime inscribe long distance ballot link on the first pre-votes:

${\mathrm{Vote}}_{{\mathrm{vp}}_i^{lo}}{(t_d,{\mathrm{po}}_h)}_p=\left|\right|{\mathrm{\&Delta;r}}^{{\mathrm{vp}}_i^{lo}}|{|}^2$ (formula 10)

Can find out, larger, the pre-votes of ballot link is larger;

We know, in actual location, random noise disturbance is can not be uncared-for.Random noise disturbance is not considered when step B.If t _dtime engrave candidate lattices po _hthe votes of gained is not that target moves and causes, but is subject to the impact of random noise, then can cause location error.Because the RSS change of ballot link meets " cooling procedure of RSS value ", i.e. t _dthe votes in moment and t _dthe votes in-1 moment there is certain relation, therefore, we can according to t _dthe votes in-1 moment judges t _dtime the votes that engraves whether because random noise causes.

C. utilize formula 11, calculate candidate lattices po _hthe second pre-votes in the td moment:

${\mathrm{Vote}}_{{\mathrm{vp}}_i^{lo}}{(t_d,{\mathrm{po}}_h)}_e={\mathrm{Vote}}_{{\mathrm{vp}}_i^{lo}}(t_{d-1},{\mathrm{po}}_h)e^{-\mathrm{\&lambda;}(t_d-t_{d-1})}$ (formula 11)

Wherein, refer to t _d-1candidate lattices po under moment long distance ballot link _hthe votes obtained; Cooling ratio λ=0.025;

D. according to the first pre-votes and the second pre-votes, formula 12 is utilized to obtain t _dmoment candidate lattices po _hin long-distance link on the votes that obtains:

${\mathrm{Vote}}_{{\mathrm{vp}}_i^{lo}}(t_d,{\mathrm{po}}_h)=({\mathrm{Vote}}_{{\mathrm{vp}}_i^{lo}}{\left(t_d,{\mathrm{po}}_h\right)}_p-{\mathrm{Vote}}_{{\mathrm{vp}}_i^{lo}}{\left(t_d,{\mathrm{po}}_h\right)}_e)\&times;p+{\mathrm{Vote}}_{{\mathrm{vp}}_i^{lo}}{(t_d,{\mathrm{po}}_h)}_e$ (formula 12)

Wherein, p refers to the fluctuating quantity of signal, utilizes formula 13 to calculate:

$p=\frac{num}{{\max }_{num}}=\frac{2L\&times;size\left(R_{\&GreaterEqual;}^{{\mathrm{vp}}_i^{lo}}\right)}{vf}$ (formula 13)

Wherein, num is fluctuation set the number of element; max _numformula 14 is utilized to obtain:

${\max }_{num}=\frac{2l/v}{f}$ (formula 14)

Wherein, v refers to the speed of target movement to be measured, is set as known quantity; For the length of long distance ballot link; F refers to the sample frequency of node, i.e. the inverse in sampling period;

If do not exist, i.e. t _dfor the first poll moment in t, then calculate according to formula 15

${\mathrm{Vote}}_{{\mathrm{vp}}_i^{lo}}(t_d,{\mathrm{po}}_h)={\mathrm{Vote}}_{{\mathrm{vp}}_i^{lo}}{(t_d,{\mathrm{po}}_h)}_p$ (formula 15)

E. repeat steps A-D, obtain td moment candidate lattices po _hthe votes in monitored area, each long-distance link obtained:

F. formula 16 is utilized, t step e obtained _dmoment candidate lattices po _hin monitored area, the votes of all long distance ballot links is added, and obtains t _dmoment candidate lattices po _htotal votes on long distance ballot link;

$Vote(t_d,{\mathrm{po}}_h)={\mathrm{\&Sigma;}}_i{\mathrm{Vote}}_{{\mathrm{vp}}_i^{lo}}(t_d,{\mathrm{po}}_h)$ (formula 16)

G. repeat the candidate lattices process of steps A-F to target to be measured on td moment all ballots link, obtain total votes of each candidate lattices on long distance ballot link; Select total votes maximum time corresponding candidate lattices as region, target location, this region comprises multiple candidate lattices, is rough target location.

Step 341, in the target rough position region that step G obtains, chooses short distance ballot link, performs step 340, obtain target location after then the long distance ballot link in step 340 being replaced with described short distance ballot link.Short distance ballot link refers in step 32 and obtains t _dtime the ballot link that engraves in length l meet the link of 0m < l≤6m.

As shown in Figure 2, link p1-p3 is ballot link, and its link p1 and link p2 is the ballot link of long distance; Link p3 is short-range ballot link, and position po1-po6 is target candidate position, and real is target actual position.On the ballot link p1 of long distance, because target location candidate lattices po1 and po2 of larger RSS degree of fluctuation obtains the highest ballot simultaneously, and candidate lattices po3 and po4 will obtain lower ballot.Long distance ballot link p2 also obtains identical result.Long-distance link ballot eventually through the first stage will obtain a candidate region (comprising candidate lattices po1 and po2).Second stage (namely voting based on short-range link) will be launched on this candidate region, finally determine target location (in candidate lattices po2) accurately.

Two, the inventive method henchnmrk test and the contrast experiment with other algorithms

1. evaluation index

Minimum position coordinates unit is done by the length of grid, if utilized, this method gained target location and target actual position are maximum differs K-l grid, then this position error is Kl, wherein K is a positive integer, l is the length of grid, defining an acceptable percentage error is error tolerance, within tolerance, be called location Success Ratio.

2. comparison other

What propose for checking the present invention distributes low cost apart from the superiority of adaptive Sensor Network passive type localization method (JRD method) in location and lower deployment cost based on RSS, by it compared with existing two kinds of algorithms, i.e. mid-point algorithm and crossover algorithm, and the correlation ratio of migration positioning performance is comparatively.

3. the assessment (carrying out under experiment scene X) of target location accuracy and performance thereof

(1) grid of reference choosing method is on the impact of positioning precision

Experiment one: grid of reference choosing method is on the impact of positioning precision

In this experiment, applicant compared for and only chooses that transverse grid is the grid of reference, only to choose longitudinal grid be that the grid of reference and both direction are all chosen for the grid of reference three kinds of situations.Fig. 3 shows the result of location.More high-precision location is obtained under the grid choosing both direction.But the quantity chosen with reference to grid is more, and cost is higher, and the positioning precision obtained under only selecting first two situation is enough.

(2) migration algorithm Performance comparision

Experiment two: migration algorithm compares location precision

In this experiment, it is under the scene of 6m and 8m that the priori distributed with reference to the RSS obtained under distance L=4m is directly applied to migration nodal distance.Fig. 4 can find out, under any tolerance, JRD method has very high success rate.When tolerance is 2m, success rate is under 95.7%, 6m and 8m be 32.4% and 23.7% respectively.This is obvious, and when nodal distance changes RSS changes in distribution very greatly, positioning performance declines.This means the localization method needing the different nodal distance of self adaptation.

Estimate the positioning precision under JRD method below.Adopt migration algorithm in experiment, under migration nodal distance is respectively 6m to 8m, target is positioned.Fig. 5 shows experimental result.It is obvious that this success rate increases, and maximum raising is respectively 70.85% and 60.1%.It is effective that experimental result shows moving method of the present invention.

Experiment three: migration algorithm compares (calculating) time cost impact

Target Station is on the grid i of L at nodal distance, then measures corresponding RSS value.Suppose that nodal distance is l, monitored area area is l × l/2, and this region is divided into the grid that several length and width are 0.5m.For each grid, collect 80 continuous print RSS values, each bag sends needs two minutes.Then want the RSS distribution needs obtaining grids all in monitored area the result of Fig. 6 display.React the time cost that number of nodes increase is clearly obviously to increase.JRD method locator time cost is used to be smaller, because use JRD technology only need to collect the RSS value of all grids under original scene and move the RSS value of specifiable lattice in scene.

(3) with existing two kinds of algorithms---mid-point algorithm is compared with crossover algorithm.

Experiment four: positioning precision compares

Test under nodal distance is 4m and fixed knot is counted.Stochastic choice 200 experiment sample inside experimental result.As the display of Fig. 7 result, the JRD method proposed is located Success Ratio and is better than mid-point algorithm and crossover algorithm far away in identical error tolerance.Especially when error is fault-tolerant be 1.5 meters time JRD method accuracy rate can be improved 50%.Result shows, and under the condition that nodes is certain, JRD method can obtain the precision of a higher location.

Experiment five: lower deployment cost compares (calculating)

Change the area of monitored area and then estimate required deployment nodes, suppose monitored area to be a length be d square and in average nodal distance under also can provide XX precision, then need individual coverage monitored area.Show in Fig. 8, along with monitored area constantly increases under mid-point algorithm and crossover algorithm, node deployment cost sharply increases, and adopts its deployment cost growth degree of JRD method of the present invention to be far smaller than first two method.

Claims (10)

1., based on a distance adaptive radio sensing network passive type localization method for RSS distribution, it is characterized in that, comprise the following steps:

Step 1, disposes wireless sensor node in monitored area, chooses sample link, by sample link coverage area grid division, and according to the length determination reference distance of sample link and migration distance;

Step 2, the grid with reference to the overlay area of the communication link of correspondence is divided into the grid of reference and non-reference grid; According to the grid of reference under reference distance and non-reference grid, determine the grid of reference under migration distance and non-reference grid; Try to achieve the RSS value set of the grid of reference under migration distance and non-reference grid; Obtain the RSS value set of all the other grids under migration distance; Obtain the RSS set of non-sample link according to the RSS value set of all grids under migration distance, build grid RSS priori storehouse under monitored area;

Step 3, utilizes grid RSS priori storehouse under the monitored area built in step 2 to carry out the location of target to be measured.First, select the ballot link in data processing cycle t, the candidate lattices engraved when then selecting a certain in t, use the votes that multilink position voting method is determined on each candidate lattices, select votes maximum be target location to be measured.

2., as claimed in claim 1 based on the distance adaptive radio sensing network passive type localization method of RSS distribution, it is characterized in that, described step 1 comprises the steps:

Step 10: dispose wireless sensor node, aggregation node and a PC in monitored area, can communicate between adjacent wireless sensor node; The overlay area S setting every communication link is the rectangular area of long A=l, wide B=l/2, and l is communication link length, 0m < l < 12m;

Step 11: from monitored area often kind of length communication link in select arbitrary link, as the sample link under this length;

Step 12: the overlay area square net of communication link corresponding for selected every bar sample link is divided into multiple grid; Choosing the nodal distance that in monitored area, adjacent node spacing is minimum is reference distance, and other nodal distances are migration distance.

3., as claimed in claim 1 based on the distance adaptive radio sensing network passive type localization method of RSS distribution, it is characterized in that, described step 2 comprises the steps:

Step 20, the RSS value set of all grids under collection reference distance, the grid with reference to the overlay area of the communication link of distance correspondence is divided into the grid of reference and non-reference grid, builds the multiple linear regression model about the grid of reference and non-reference grid.

Step 21, according to the grid of reference in communication link overlay area corresponding under reference distance and non-reference grid, determines the grid of reference in the communication link overlay area that migration distance L` is corresponding and non-reference grid;

Step 22, measure the RSS value set of the grid of reference in communication link overlay area corresponding to migration distance L`, utilize the multiple linear regression model that step 20 obtains, obtain the RSS value set of the non-reference grid in communication link overlay area corresponding to migration distance L`;

Step 23, the RSS value set of the grid of reference and the RSS value set of non-reference grid in the communication link overlay area that the migration distance obtained according to step 22 is corresponding, obtain the RSS value set of all the other grids under migration distance L` (grid namely under migration distance L` outside the grid of reference and non-reference grid);

Step 24, according to the grid RSS value set of sample links all in monitored area, obtains the grid RSS value set of the non-sample link of other and sample link equal length, identical with sample link; In monitored area, the RSS value set of the grid of all communication links forms grid RSS priori storehouse under monitored area.

4., as claimed in claim 3 based on the distance adaptive radio sensing network passive type localization method of RSS distribution, it is characterized in that, described step 20 comprises the following steps:

Step 200, the RSS value set of all grids under collection reference distance;

Step 201, the grid with reference to the overlay area of the communication link of distance correspondence is divided into the grid of reference and non-reference grid, builds the multiple linear regression model about the grid of reference and non-reference grid; Specifically comprise two steps:

A) with reference to the overlay area S of communication link corresponding to distance L ^lgrid be divided into the grid of reference and non-reference grid.

The all grids chosen under reference distance in communication link sighting distance and on the cross region vertical with link sighting distance are the grid of reference, and remaining grid is non-reference grid; Q (q=0,1 ..., n1) the RSS value set of the individual grid of reference is kth (k=1,2 ..., n2) individual non-reference grid RSS set be

B) multiple linear regression model about the grid of reference and non-reference grid is built;

The xth time RSS value of the xth time RSS value of all grids of reference under the reference distance L gathered, all non-reference grids is substituted into formula 1, obtains the multiple linear regression coefficient matrix that xth is secondary; X=1,2 ..., X; Obtain X multiple linear regression coefficient matrix altogether; Thus obtain the multiple linear regression model of the grid of reference and non-reference grid:

In formula, for certain RSS measured value once of a kth non-reference grid, k=1,2 ..., n2; be the RSS measured value of the correspondence time of q the grid of reference, q=0,1 ..., n1; for the multiple linear regression matrix coefficient of corresponding time; ξ _kfor noise jamming, ignore.

5., as claimed in claim 3 based on the distance adaptive radio sensing network passive type localization method of RSS distribution, it is characterized in that, the concrete operations of described step 21 are as follows:

Step 210, determines the grid of reference in the communication link overlay area that migration distance L` is corresponding;

With reference to the grid of reference under distance L from top to bottom from left to right number consecutively be 1,2 ..., n1}; By the cross area grid under migration distance L` from top to bottom from left to right number consecutively be 1,2 ..., n3};

The grid of reference under formula 2 computation migration distance L` is utilized to number

(formula 2)

In formula, n1 is the number of the grid of reference under reference distance L; N3 is the number of the cross area grid under migration distance L`;

Step 211, determines the non-reference grid in the communication link overlay area that migration distance L` is corresponding;

With reference to the non-cross region under distance L grid from left to right snakelike number consecutively be 1,2 ..., n2}; By the non-cross area grid under migration distance L` from left to right snakelike number consecutively be; Non-reference grid numbering under utilizing formula 3 to calculate migration distance L`

(formula 3)

In formula, n2 is the number of the non-reference grid under reference distance L; N4 is the number of the non-cross area grid under migration distance L`.

6., as claimed in claim 3 based on the distance adaptive radio sensing network passive type localization method of RSS distribution, it is characterized in that, described step 22 is specific as follows:

Gather respectively q the grid of reference under X migration distance L` RSS value (q=0,1 ... n1), obtain the RSS value set of q the grid of reference under migration distance L` by the xth of all grids of reference under the migration distance L` that records time (x=1,2, X) RSS value substitutes into the multiple linear regression model under the multiple linear regression coefficient matrix of the correspondence that obtains of step 20 time, calculates the value of the RSS of the kth non-reference grid under the migration distance L` of xth time finally obtain the RSS value set on non-reference grid k

(formula 4)

Wherein: for q the corresponding secondary RSS measured value of the grid of reference under migration distance, q=0,1 ..., n1; for the multiple linear regression matrix coefficient of corresponding time, k=1,2 ..., n2; ξ _kfor noise jamming, ignore.

7., as claimed in claim 3 based on the distance adaptive radio sensing network passive type localization method of RSS distribution, it is characterized in that, described step 23 comprises the steps:

Step 230, determines interpenetration network;

Using the grid outside the grid of reference in communication link overlay area corresponding for migration distance L` and non-reference grid as interpenetration network;

Step 231, calculates the RSS value set of grid each in interpenetration network by B-spline interpolation model; Comprise following two steps:

A) with the communication link overlay area S that migration distance L` is corresponding <sup>l`</sup>on a line grid be a B-spline interpolation operation, all row of communication link overlay area corresponding for migration distance L` are divided into two classes: the first kind be row in have the grid of reference and non-reference grid simultaneously; Another kind of is without any the grid of reference and non-reference grid in row;

B) the RSS value set of the interpenetration network in first kind row is calculated by B-spline interpolation model:

For every a line of the first kind, certain RSS value of grid of reference each in this row and the corresponding secondary RSS value of each non-reference grid are substituted into B-spline interpolation model, obtains the corresponding secondary RSS value of all interpenetration networks on this row; Repeat above-mentioned steps X time, obtain the RSS value set of each interpenetration network on this row;

C) by carrying out B-spline interpolation model to each row of overlay area, the set of RSS migration distance remaining interpenetration network is calculated.

8., as claimed in claim 1 based on the distance adaptive radio sensing network passive type localization method of RSS distribution, it is characterized in that, described step 3 comprises the steps:

Step 30, when recording driftlessness in monitored area, RSS value on arbitrary communication link, is called RSS value ideally

Step 31, the RSS value of all sampling instants on the every communication link in image data treatment cycle t, obtains the RSS value set of this communication link in data processing cycle t;

Step 32, selects d moment t _dall ballot links in upper monitored area, inscribe the fluctuation set on every bar ballot link, t when determining this _d∈ t;

Step 33, at t _don every bar ballot link on select the candidate lattices of target to be measured, thus select t _don all ballot links on the candidate lattices of target to be measured;

Step 34, uses multilink position voting method, under all ballot links in d moment target to be measured candidate lattices in select the grid at target place to be measured, obtain target location to be measured.

9., as claimed in claim 8 based on the distance adaptive radio sensing network passive type localization method of RSS distribution, it is characterized in that, described step 32 comprises the steps:

A. d moment t _dunder, all communication link p in monitored area _iin all ballot link vp of engraving when selecting this _i;

The communication link p meeting formula 5 will be engraved during td _ias t _dtime the ballot link vp that engraves _i, and the link vp that will vote _ion RSS value be defined as

(formula 5)

Wherein, threshold epsilon=2; for communication link p _iat t _dthe RSS value in moment;

The fluctuation set on each article of ballot link is inscribed when B. determining d;

Determine t _dmoment votes link vp _iafter, definition t _dtime the ballot link vp that engraves _ithe RSS value set gathered in data processing cycle t is j is the ballot link vp gathered in data processing cycle t _ithe number of RSS value, d ∈ [1,2,3 ..., j], t _dfor d the moment in t; Then will gather in each value substitute into formula 6 successively, will formula 6 be met and comprise moment t _dcontinuous element as fluctuation set unit system;

(formula 6)

Wherein, threshold epsilon=2; for the link vp that votes when driftlessness occurs in monitored area _irSS value.

10. as claimed in claim 8 based on the distance adaptive radio sensing network passive type localization method of RSS distribution, it is characterized in that, described step 34 step 340, based on long distance ballot link ballot, obtain target rough position; Concrete steps are as follows:

A. utilize formula 9, obtain long distance ballot link the RSS degree of fluctuation engraved when td;

(formula 9)

Wherein, refer to t _dtime engrave long distance ballot link on RSS value, long distance ballot link refers in step 32 and obtains t _dtime the ballot link that engraves in length 1 meet the link of 6m < l < 12m; for the RSS value of duration distance ballot link appears in driftlessness in monitored area;

B. formula 10 is utilized, calculated candidate grid po _hat t _dtime inscribe long distance ballot link on the first pre-votes:

(formula 10);

C. candidate lattices po is calculated _hthe second pre-votes in the td moment:

(formula 11)

Wherein, refer to t _d-1candidate lattices po under moment long distance ballot link _hthe votes obtained; Cooling ratio λ=0.025;

D. according to the first pre-votes and the second pre-votes, formula 12 is utilized to obtain t _dmoment candidate lattices po _hin long-distance link on the votes that obtains:

(formula 12)

Wherein, p refers to the fluctuating quantity of signal, utilizes formula 13 to calculate:

(formula 13)

Wherein, num is fluctuation set the number of element; max _numformula 14 is utilized to obtain:

(formula 14)

Wherein, v refers to the speed of target movement to be measured, is set as known quantity; L is the length of long distance ballot link; F refers to the sample frequency of node, i.e. the inverse in sampling period;

If do not exist, i.e. t _dfor the first poll moment in t, then calculate according to formula 15

(formula 15)

E. repeat steps A-D, obtain td moment candidate lattices po _hthe votes in monitored area, each long-distance link obtained:

F. formula 16 is utilized, t step e obtained _dmoment candidate lattices po _hin monitored area, the votes of all long distance ballot links is added, and obtains t _dmoment candidate lattices po _htotal votes on long distance ballot link;

(formula 16)

G. repeat the candidate lattices process of steps A-F to target to be measured on td moment all ballots link, obtain total votes of each candidate lattices on long distance ballot link; Select total votes maximum time corresponding candidate lattices as region, target location, this region comprises multiple candidate lattices, is rough target location;

Step 341, in the target rough position region that step G obtains, chooses short distance ballot link, performs step 340, obtain target location after then the long distance ballot link in step 340 being replaced with described short distance ballot link; Short distance ballot link refers in step 32 and obtains t _dtime the ballot link that engraves in length 1 meet the link of 0m < l≤6m.