CN103813450A - Optimized mobile wireless sensor network node positioning method - Google Patents

Abstract

The invention relates to an optimized mobile wireless sensor network node positioning method which comprises the steps of calculating anchor node coverage area weighting mass center, producing a sampling area, filtering sampling particles and confirming position estimation of a node to be positioned. In the method, the anchor node weighting mass center serves as a circle center, the radius is confirmed according to the estimated position of the node to be positioned at a previous moment, an anchor box is obtained and then a corresponding sampling area is confirmed. The method comprises the steps of calculating the anchor node coverage area weighting mass center, producing the sampling area and positioning the node and is a specific implementing method. Compared with a traditional MCB positioning method, the method can effectively improve the node positioning accuracy in a network and improve the positioning coverage rate to the most degree.

Description

A kind of mobile radio sensor network node localization method of optimization

Technical field

The present invention relates to a kind of mobile radio sensor network node localization method of optimization, belong to the technical field of wireless sensor network.

Background technology

Wireless sensor network (Wireless Sensor Networks, be called for short WSN) form by being deployed in cheap microsensor nodes a large amount of in monitored area, form the ad hoc network system of a multi-hop by communication, the information of perceptive object in the perception, acquisition and processing network's coverage area of cooperation, and send to observer.It has a wide range of applications in civilian, military, industry and other some commercial fields.Node locating technique, as one of key technology of wireless sensor network, not only has important effect to the basic application of wireless sensor network, and is the basis of target monitoring and tracking.

Current location, wireless senser networking can be divided into traditional location, static wireless senser networking (Static Wireless Sensor Network, and mobile wireless sensor network location (Mobile Wireless Sensor Network, MWSN) SWSN).Location, so-called static wireless senser networking refers to that sensor node keeps static node locating, and mobile wireless sensor network location refers to the location of sensor node under motion conditions.

The measuring distance according to whether, location, static wireless senser networking can be divided into two large classes: (Range-based) based on range finding and the location algorithm without find range (Range-free).The former uses trilateration, triangulation or Maximum Likelihood Estimation Method etc. by distance or the angle information of point-to-point between measured node, can calculate the position of node self; The latter is without distance or angle information, only according to information realization node locating such as network connectivties.According to account form, location, static wireless senser networking can be divided into centralized compute location and Distributed Calculation location.Centralized calculating refers to needed information is sent to Centroid, and the mode positioning at this Centroid; Distributed Calculation refers to internodal information interchange and coordination, the mode that node is located voluntarily of relying on.There are its pluses and minuses centralized and location distributed computing, respectively has its advantage in different application scenarios.

Location, mobile radio sensor networking (also referred to as the mobile wireless sensor network location based on statistical method) and non-statistical method (other mobile wireless sensor network location) mobile wireless sensor network that current representational mobile wireless sensor network location can be divided into based on sequential Monte-Carlo method are located two kinds.Mobile WSN Position Research based on sequence Monte Carlo is a focus at present, and the location algorithm of this aspect mainly contains MCL, MCB, MMCL, MSL, range-based-MCL etc.

MCL algorithm is first Monte Carlo Range-free location algorithm for WSN mobile node, the location accuracy of this algorithm is not subject to the impact of node motion state, utilize on the contrary the mobility of node to improve the accuracy of node locating, reduced the cost of locating simultaneously.But MCL location exists sampling efficiency low, amount of calculation is large, and anchor density requires the shortcomings such as high, aspect acquisition sample, is needing to improve.

Monte Carlo box (MCB, Monte-Carlo Localization Boxed) location algorithm is to put forward on the basis of MCL algorithm, this algorithm is limited in sample area in a sampling box by definition anchor box and sampling box, improved and be sampled into power and positioning precision, Performance Ratio MCL algorithm has had larger improvement.

Also there is in actual applications certain defect in MCB algorithm.According to statistics, be approximately under 10% prerequisite at anchor density, MCB positioning precision can be higher, and the position error of node is approximately the communication radius of node.But in the time that anchor node is not enough, the sample area that MCB algorithm obtains does not have enough minimizings, filtration stage calculates the cycle of iteration will be increased, and positioning precision will be had a greatly reduced quality.If compare with traditional MCL algorithm, now the superiority of algorithm does not just embody, and locating effect is still can not bring into play design advantage.As can be seen here, this algorithm still has very high requirement to the density of anchor node.

Summary of the invention

For the deficiencies in the prior art, the invention provides a kind of mobile radio sensor network node localization method of optimization.In mobile wireless sensor network position fixing process, there will be the sampling anchor box that generation is larger in the situation that anchor density is low, thereby this category node is because the large sample area of sampling anchor box is large, can cause the increase of position error; Thereby even there is the too low situation that cannot get sampling anchor box of anchor density.For above problem, the present invention proposes a kind of mobile radio sensor network node localization method of optimization.

Technical scheme of the present invention is as follows:

A mobile radio sensor network node localization method for optimization, comprises the generation of calculating, the sample area of anchor node overlay area weighted mass center, sampling particle is carried out filtering and determines the location estimation of node to be positioned:

Step 1: the calculating of anchor node overlay area weighted mass center comprises the following first step to the three steps

The first step: the anchor node of all location awares in network is the information of seed node inundation oneself in the whole network, this information by inundation comprises (Xseed, Yseed), IDseed, TTL, wherein (Xseed, Yseed) coordinate figure of expression seed node, IDseed represents the sequence number of this seed, and TTL is maximum inundation jumping figure, and initial value is set to 2;

Second step: Location-Unknown in network, etc. ordinary node to be positioned be node node after receiving the inundation information of seed node, record the signal strength signal intensity RSSI of coordinate figure, sequence number and the reception of seed node; For the information frame that guarantees anchor node is only forwarded once, if TTL is 2, is set to 1 and is also forwarded, if TTL is 1, its zero setting is no longer forwarded, the anchor node number simultaneously monitoring under nodes records, is designated as M;

The 3rd step: anchor node information inundation is complete, calculates node node anchor box; Be divided into again following three kinds of situations according to the difference that receives anchor node kind:

(1) while only having a jumping anchor node

Draw the distance d of each anchor node to node to be positioned according to the signal strength signal intensity RSSI receiving _i, obtain weighted mass center according to formula (I):

$X_e=\frac{\underset{i=1}{\overset{M}{\mathrm{\Σ}}}{w}_i{X}^i\mathrm{seed}}{\underset{i=1}{\overset{M}{\mathrm{\Σ}}}{w}_i};Y_e=\frac{\underset{i=1}{\overset{M}{\mathrm{\Σ}}}{w}_i{Y}^i\mathrm{seed}}{\underset{i=1}{\overset{M}{\mathrm{\Σ}}}{w}_i}$ （I）

In formula (I): w _irepresent the weights of each anchor node, these weights are functions that unknown node arrives the distance of anchor node, here get

w _i＝1/d _i （II）

Wherein, d _irepresent the distance of a node node and i jumping anchor node, the signal strength signal intensity RSSI being received by node node draws, specific formula for calculation is:

$\mathrm{PL}\left(d\right)=\mathrm{PL}\left(d_O\right)+10\mathrm{nlg}\left(\frac{d}{d_O}\right)+X_{\mathrm{\δ}}$ （III）

Wherein: PL (d) is through the power loss apart from after d; PL (d _o) be to be d apart from the distance of sending node _othe signal power received of reference node; D is that sending node is to the distance between receiving node; N is signal attenuation parameter, and relevant with actual environment, scope is between 2～5; X _δfor the random numbers of Gaussian distribution are that average is 0, its standard deviation scope is 4～10;

From formula (III), the signal strength signal intensity that unknown node is received is:

P(d)=Pt-PL(d) （IV）

Wherein, P (d) is received signal strength; Pt is transmit signal strength, obtains the corresponding relation of RSSI and distance based on above-mentioned principle;

(2) while only having double bounce anchor node

Because unknown node and double bounce anchor node cannot direct communications, therefore can not directly use received signal strength to ask distance, cannot obtain weighted mass center, so use the barycenter of all double bounce anchor nodes position here, formula is as follows

$X_e=\frac{\underset{i=1}{\overset{M}{\mathrm{\Σ}}}{X}^i\mathrm{seed}}{M};Y_e=\frac{\underset{i=1}{\overset{M}{\mathrm{\Σ}}}{Y}^i\mathrm{seed}}{M}$ （V）

(3) when an existing jumping anchor node has again double bounce anchor node

Obtain barycenter and whole barycenter of anchor nodes of weighted mass center, the double bounce anchor node of a jumping anchor node, suc as formula (VI) (VII) shown in (VIII):

One jumps the barycenter of anchor node:

$X_{1e}=\frac{\underset{i=1}{\overset{M_1}{\mathrm{\Σ}}}{w}_i{X}^{i}1\mathrm{seed}}{\underset{i=1}{\overset{M_1}{\mathrm{\Σ}}}{w}_i};Y_{1e}=\frac{\underset{i=1}{\overset{M_1}{\mathrm{\Σ}}}{w}_i{Y}^{i}1\mathrm{seed}}{\underset{i=1}{\overset{M_1}{\mathrm{\Σ}}}{w}_i}$ （VI）

The barycenter of double bounce anchor node:

$X_{2e}=\frac{\underset{i=1}{\overset{M_2}{\mathrm{\Σ}}}{X}^{i}2\mathrm{seed}}{M_2};Y_{2e}=\frac{\underset{i=1}{\overset{M_2}{\mathrm{\Σ}}}{Y}^{i}2\mathrm{seed}}{M_2}$ （VII）

All barycenter of anchor node:

$X_e=\frac{\underset{i=1}{\overset{M_1}{\mathrm{\Σ}}}{X}^{i}1\mathrm{seed}+\underset{i=1}{\overset{M_2}{\mathrm{\Σ}}}{X}^{i}2\mathrm{seed}}{M_1+M_2}$

$Y_e=\frac{\underset{i=1}{\overset{M_1}{\mathrm{\Σ}}}{Y}^{i}1\mathrm{seed}+\underset{i=1}{\overset{M_2}{\mathrm{\Σ}}}{Y}^{i}2\mathrm{seed}}{M_1+M_2}$ （VIII）；

Step 2: the generation of sample area comprises the following first step to the three steps

The first step: according to estimated position and the position error in a moment on node node, determine the half of the anchor box length of side, length is R, and its expression formula is:

$R=\sqrt{{(X_e-X_{\mathrm{node}\_g}(k-1,i))}^2}+\mathrm{error}(k-1,i)$ （IX）

In formula: x _{node_g}(k-1, i) is the estimated position in a moment on node; Error (k-1, i) is the position error of a moment node on node to be positioned;

Second step: determine anchor box, be now divided into three kinds of situations:

(1), while only having a jumping anchor node, according to formula (VI), can obtain anchor box suc as formula shown in (X)

x _1min＝（x _1e-R）x _1max＝（x _1e+R）

y _1min＝（y _1e-R）y _1max＝（x _1e+R） （X）

(2), while only having double bounce anchor node, according to formula (VII), can obtain anchor box suc as formula shown in (XI)

x _2min＝（x _2e-R）x _2max＝（x _2e+R）

y _2min＝（y _2e-R）y _2max＝（x _2e+R） （XI）

(3) when an existing jumping anchor node has again double bounce anchor node

Calculate respectively the anchor box of a jumping anchor node and double bounce anchor node, shown in (XII), (XIII):

One jumps the anchor box of anchor node:

x _1min＝（x _1e-R）x _1max＝（x _1e+R）

y _1min＝（y _1e-R）y _1max＝（x _1e+R） （XII）

The anchor box of double bounce anchor node:

x _2min＝（x _2e-R）x _2max＝（x _2e+R）

y _2min＝（y _2e-R）y _2max＝（x _2e+R） （XIII）

Judge whether the two has common factor:

If there is common factor, final anchor box is

x _min＝max（x _1min，x _2min）x _max＝min（x _1max，x _2max）

y _min＝max（y _1min，y _2min）y _max＝min（y _1max，y _2max） （XIV）

If without common factor, final anchor box is

x _min＝（x _e-R） x _max＝（x _e+R）

y _min＝（y _e-R） y _max＝（x _e+R） （XV）

The 3rd step: determine sampling box, Box ⁱ _t=(x ⁱ _min, x ⁱ _max, y ⁱ _min, y ⁱ _max), sampling box is anchor box and border box, on node centered by the estimated position in a moment, with 2 × V _maxfor the common factor of the square area of the length of side, computational methods are as follows:

x ⁱ _min=max（x _min，x ⁱ _t-1-v _max） x ⁱ _max=min（x _max，x ⁱ _t-1+v _max）

y ⁱ _min=max（y _min，y ⁱ _t-1-v _max） y ⁱ _max=min（y _max，y ⁱ _t-1+v _max）（XVI）

Wherein (x ⁱ _t-1, y ⁱ _t-1) represent the sampling particle l of this node node in (t-1) moment ⁱ _t-1coordinate, V _maxstare at fourth and represent the maximum translational speed of node, after obtaining sampling box, sampling obtains the particle l in t moment ⁱ _t;

Step 3: sampling particle is carried out to filtering

p(o _t|l ⁱ _t)＝1，

tr≤d(l _t,s)≤2×tr

p(o _t|l ⁱ _t)＝0，otherwise

Wherein o _trepresent observation information, S is the set of a jumping anchor node, and T is the set of double bounce anchor node, d (l _t, s) be sampling particle l _teuclidean distance with anchor node;

After filtering completes, if the needed number N of the not enough location estimation of population is carried out resampling, and repeated filtering, until reach hits N or reach maximum sampling number;

Step 4: determine the location estimation value of this node node to be positioned,

Location estimation value

Advantage of the present invention is:

Although MCB can obtain good locating effect, owing to comprising many uncertain factors in its sampling process, we can find that this algorithm also exists many problems:

1.MCB algorithm is larger to the dependence of anchor density.In the time that anchor density is lower, the number of nodes that unknown node can be detected is also fewer, but sample area does not have enough reducing, and will cause iteration cycle to increase, and positioning precision is inevitable undesirable.

2.MCB algorithm is thought a changeless value the communication radius of node, communication range is regarded as to a regular border circular areas, but in reality, be subject to the impact of external environment, the communication radius of node and communication range can not be changeless, and this is also a considerable problem.

For above problem, the optimized algorithm that the present invention proposes is in the time of definite anchor box, get rid of the possibility that anchor node communication radius may be affected by extraneous factor, introduce higher RSSI algorithm and the centroid algorithm based on distance of precision, unknown node is limited in a more accurate anchor box, the sampling box obtaining is more tallied with the actual situation, and then improved the positioning precision at whole networking.

Accompanying drawing explanation

Fig. 1 is the distributed model of node in network.

Fig. 2 is that a jumping anchor node anchor box obtains schematic diagram.

Fig. 3 is that double bounce anchor node anchor box obtains schematic diagram.

Fig. 4 is the generation schematic diagram of sample area.

Fig. 5 is FB(flow block) of the present invention.

Embodiment

Below in conjunction with embodiment, the present invention will be further described.

Embodiment:

As Figure 1-5.

A mobile radio sensor network node localization method for optimization, comprises the generation of calculating, the sample area of anchor node overlay area weighted mass center, sampling particle is carried out filtering and determines the location estimation of node to be positioned:

Step 1: the calculating of anchor node overlay area weighted mass center comprises the following first step to the three steps

The first step: the anchor node of all location awares in network is the information of seed node inundation oneself in the whole network, this information by inundation comprises (Xseed, Yseed), IDseed, TTL, wherein (Xseed, Yseed) coordinate figure of expression seed node, IDseed represents the sequence number of this seed, and TTL is maximum inundation jumping figure, and initial value is set to 2;

Second step: Location-Unknown in network, etc. ordinary node to be positioned be node node after receiving the inundation information of seed node, record the signal strength signal intensity RSSI of coordinate figure, sequence number and the reception of seed node; For the information frame that guarantees anchor node is only forwarded once, if TTL is 2, is set to 1 and is also forwarded, if TTL is 1, its zero setting is no longer forwarded, the anchor node number simultaneously monitoring under nodes records, is designated as M;

The 3rd step: anchor node information inundation is complete, calculates node node anchor box; Be divided into again following three kinds of situations according to the difference that receives anchor node kind:

(1) while only having a jumping anchor node

Draw the distance di of each anchor node to node to be positioned according to the signal strength signal intensity RSSI receiving, obtain weighted mass center according to formula (I):

$X_e=\frac{\underset{i=1}{\overset{M}{\mathrm{\Σ}}}{w}_i{X}^i\mathrm{seed}}{\underset{i=1}{\overset{M}{\mathrm{\Σ}}}{w}_i};Y_e=\frac{\underset{i=1}{\overset{M}{\mathrm{\Σ}}}{w}_i{Y}^i\mathrm{seed}}{\underset{i=1}{\overset{M}{\mathrm{\Σ}}}{w}_i}$ （I）

In formula (I): w _irepresent the weights of each anchor node, these weights are functions that unknown node arrives the distance of anchor node, here get

w _i＝1/d _i （II）

Wherein, d _irepresent the distance of a node node and i jumping anchor node, the signal strength signal intensity RSSI being received by node node draws, specific formula for calculation is:

$\mathrm{PL}\left(d\right)=\mathrm{PL}\left(d_O\right)+10\mathrm{nlg}\left(\frac{d}{d_O}\right)+X_{\mathrm{\δ}}$ （III）

Wherein: PL (d) is through the power loss apart from after d; PL (d _o) be to be d apart from the distance of sending node _othe signal power received of reference node; D is that sending node is to the distance between receiving node; N is signal attenuation parameter, and relevant with actual environment, scope is between 2～5; X _δfor the random numbers of Gaussian distribution are that average is 0, its standard deviation scope is 4～10;

From formula (III), the signal strength signal intensity that unknown node is received is:

P(d)=Pt-PL(d) （IV）

Wherein, P (d) is received signal strength; Pt is transmit signal strength, obtains the corresponding relation of RSSI and distance based on above-mentioned principle;

(2) while only having double bounce anchor node

Because unknown node and double bounce anchor node cannot direct communications, therefore can not directly use received signal strength to ask distance, cannot obtain weighted mass center, so use the barycenter of all double bounce anchor nodes position here, formula is as follows

$X_e=\frac{\underset{i=1}{\overset{M}{\mathrm{\Σ}}}{X}^i\mathrm{seed}}{M};Y_e=\frac{\underset{i=1}{\overset{M}{\mathrm{\Σ}}}{Y}^i\mathrm{seed}}{M}$ （V）

(3) when an existing jumping anchor node has again double bounce anchor node

Obtain barycenter and whole barycenter of anchor nodes of weighted mass center, the double bounce anchor node of a jumping anchor node, suc as formula (VI) (VII) shown in (VIII):

One jumps the barycenter of anchor node:

$X_{1e}=\frac{\underset{i=1}{\overset{M_1}{\mathrm{\Σ}}}{w}_i{X}^{i}1\mathrm{seed}}{\underset{i=1}{\overset{M_1}{\mathrm{\Σ}}}{w}_i};Y_{1e}=\frac{\underset{i=1}{\overset{M_1}{\mathrm{\Σ}}}{w}_i{Y}^{i}1\mathrm{seed}}{\underset{i=1}{\overset{M_1}{\mathrm{\Σ}}}{w}_i}$ （VI）

The barycenter of double bounce anchor node:

$X_{2e}=\frac{\underset{i=1}{\overset{M_2}{\mathrm{\Σ}}}{X}^{i}2\mathrm{seed}}{M_2};Y_{2e}=\frac{\underset{i=1}{\overset{M_2}{\mathrm{\Σ}}}{Y}^{i}2\mathrm{seed}}{M_2}$ （VII）

All barycenter of anchor node:

$X_e=\frac{\underset{i=1}{\overset{M_1}{\mathrm{\Σ}}}{X}^{i}1\mathrm{seed}+\underset{i=1}{\overset{M_2}{\mathrm{\Σ}}}{X}^{i}2\mathrm{seed}}{M_1+M_2}$

$Y_e=\frac{\underset{i=1}{\overset{M_1}{\mathrm{\Σ}}}{Y}^{i}1\mathrm{seed}+\underset{i=1}{\overset{M_2}{\mathrm{\Σ}}}{Y}^{i}2\mathrm{seed}}{M_1+M_2}$ （VIII）；

Step 2: the generation of sample area comprises the following first step to the three steps

The first step: according to estimated position and the position error in a moment on node node, determine the half of the anchor box length of side, length is R, and its expression formula is:

$R=\sqrt{{(X_e-X_{\mathrm{node}\_g}(k-1,i))}^2}+\mathrm{error}(k-1,i)$ （IX）

In formula: x _{node_g}(k-1, i) is the estimated position in a moment on node; Error (k-1, i) is the position error of a moment node on node to be positioned;

Second step: determine anchor box, be now divided into three kinds of situations:

(1), while only having a jumping anchor node, according to formula (VI), can obtain anchor box suc as formula shown in (X)

x _1min＝（x _1e-R） x _1max＝（x _1e+R）

y _1min＝（y _1e-R） y _1max＝（x _1e+R） （X）

(2), while only having double bounce anchor node, according to formula (VII), can obtain anchor box suc as formula shown in (XI)

x _2min＝（x _2e-R） x _2max＝（x _2e+R）

y _2min＝（y _2e-R） y _2max＝（x _2e+R） （XI）

(3) when an existing jumping anchor node has again double bounce anchor node

Calculate respectively the anchor box of a jumping anchor node and double bounce anchor node, shown in (XII), (XIII):

One jumps the anchor box of anchor node:

x _1min＝（x _1e-R） x _1max＝（x _1e+R）

y _1min＝（y _1e-R） y _1max＝（x _1e+R） （XII）

The anchor box of double bounce anchor node:

x _2min＝（x _2e-R） x _2max＝（x _2e+R）

y _2min＝（y _2e-R） y _2max＝（x _2e+R） （XIII）

Judge whether the two has common factor:

If there is common factor, final anchor box is

x _min＝max（x _1min，x _2min） x _max＝min（x _1max，x _2max）

y _min＝max（y _1min，y _2min） y _max＝min（y _1max，y _2max） （XIV）

If without common factor, final anchor box is

x _min＝（x _e-R） x _max＝（x _e+R）

y _min＝（y _e-R） y _max＝（x _e+R） （XV）

The 3rd step: determine sampling box, Box ⁱ _t=(x ⁱ _min, x ⁱ _max, y ⁱ _min, y ⁱ _max), sampling box is anchor box and border box, on node centered by the estimated position in a moment, with 2 × V _maxfor the common factor of the square area of the length of side, computational methods are as follows:

x ⁱ _min=max（x _min，x ⁱ _t-1-v _max） x ⁱ _max=min（x _max，x ⁱ _t-1+v _max）

y ⁱ _min=max（y _min，y ⁱ _t-1-v _max） y ⁱ _max=min（y _max，y ⁱ _t-1+v _max）（XVI）

Wherein (x ⁱ _t-1, y ⁱ _t-1) represent the sampling particle l of this node node in (t-1) moment ⁱ _t-1coordinate, V _maxrepresent the maximum translational speed of node, after obtaining sampling box, sampling obtains the particle l in t moment ⁱ _t;

Step 3: sampling particle is carried out to filtering

p(o _t|l ⁱ _t)＝1，

tr≤d(l _t,s)≤2×tr

p(o _t|l ⁱ _t)＝0，otherwise

Wherein o _trepresent observation information, S is the set of a jumping anchor node, and T is the set of double bounce anchor node, d (l _t, s) be sampling particle l _teuclidean distance with anchor node;

After filtering completes, if the needed number N of the not enough location estimation of population is carried out resampling, and repeated filtering, until reach hits N or reach maximum sampling number;

Step 4: determine the location estimation value of this node node to be positioned,

Location estimation value

Claims (1)

1. the mobile radio sensor network node localization method of an optimization, it is characterized in that, this localization method comprises the generation of calculating, the sample area of anchor node overlay area weighted mass center, sampling particle is carried out filtering and determines the location estimation of node to be positioned:

Step 1: the calculating of anchor node overlay area weighted mass center comprises the following first step to the three steps

The first step: the anchor node of all location awares in network is the information of seed node inundation oneself in the whole network, this information by inundation comprises (Xseed, Yseed), IDseed, TTL, wherein (Xseed, Yseed) coordinate figure of expression seed node, IDseed represents the sequence number of this seed, and TTL is maximum inundation jumping figure, and initial value is set to 2;

Second step: Location-Unknown in network, etc. ordinary node to be positioned be node node after receiving the inundation information of seed node, record the signal strength signal intensity RSSI of coordinate figure, sequence number and the reception of seed node; For the information frame that guarantees anchor node is only forwarded once, if TTL is 2, is set to 1 and is also forwarded, if TTL is 1, its zero setting is no longer forwarded, the anchor node number simultaneously monitoring under nodes records, is designated as M;

The 3rd step: anchor node information inundation is complete, calculates node node anchor box; Be divided into again following three kinds of situations according to the difference that receives anchor node kind:

(1) while only having a jumping anchor node

Draw the distance d of each anchor node to node to be positioned according to the signal strength signal intensity RSSI receiving _i, obtain weighted mass center according to formula (I):

$X_e=\frac{\underset{i=1}{\overset{M}{\mathrm{\Σ}}}{w}_i{X}^i\mathrm{seed}}{\underset{i=1}{\overset{M}{\mathrm{\Σ}}}{w}_i};Y_e=\frac{\underset{i=1}{\overset{M}{\mathrm{\Σ}}}{w}_i{Y}^i\mathrm{seed}}{\underset{i=1}{\overset{M}{\mathrm{\Σ}}}{w}_i}$ (I)

In formula (I): w _irepresent the weights of each anchor node, these weights are functions that unknown node arrives the distance of anchor node, here get

w _i＝1/d _i (II)

Wherein, d _irepresent the distance of a node node and i jumping anchor node, the signal strength signal intensity RSSI being received by node node draws, specific formula for calculation is:

$\mathrm{PL}\left(d\right)=\mathrm{PL}\left(d_O\right)+10\mathrm{nlg}\left(\frac{d}{d_O}\right)+X_{\mathrm{\δ}}$ (III)

Wherein: PL (d) is through the power loss apart from after d; PL (d _o) be to be d apart from the distance of sending node _othe signal power received of reference node; D is that sending node is to the distance between receiving node; N is signal attenuation parameter, and relevant with actual environment, scope is between 2～5; X _δfor the random numbers of Gaussian distribution are that average is 0, its standard deviation scope is 4～10;

From formula (III), the signal strength signal intensity that unknown node is received is:

P (d)=Pt mono-PL (d) (IV)

Wherein, F (d) is received signal strength; Pt is transmit signal strength, obtains the corresponding relation of RSSI and distance based on above-mentioned principle;

(2) while only having double bounce anchor node

Because unknown node and double bounce anchor node cannot direct communications, therefore can not directly use received signal strength to ask distance, cannot obtain weighted mass center, so use the barycenter of all double bounce anchor nodes position here, formula is as follows

$X_e=\frac{\underset{i=1}{\overset{M}{\mathrm{\Σ}}}{X}^i\mathrm{seed}}{M};Y_e=\frac{\underset{i=1}{\overset{M}{\mathrm{\Σ}}}{Y}^i\mathrm{seed}}{M}$ (V)

(3) when an existing jumping anchor node has again double bounce anchor node

Obtain barycenter and whole barycenter of anchor nodes of weighted mass center, the double bounce anchor node of a jumping anchor node, suc as formula (VI) (VII) shown in (VIII):

One jumps the barycenter of anchor node:

$X_{1e}=\frac{\underset{i=1}{\overset{M_1}{\mathrm{\Σ}}}{w}_i{X}^{i}1\mathrm{seed}}{\underset{i=1}{\overset{M_1}{\mathrm{\Σ}}}{w}_i};Y_{1e}=\frac{\underset{i=1}{\overset{M_1}{\mathrm{\Σ}}}{w}_i{Y}^{i}1\mathrm{seed}}{\underset{i=1}{\overset{M_1}{\mathrm{\Σ}}}{w}_i}$ (VI)

The barycenter of double bounce anchor node:

$X_{2e}=\frac{\underset{i=1}{\overset{M_2}{\mathrm{\Σ}}}{X}^{i}2\mathrm{seed}}{M_2};Y_{2e}=\frac{\underset{i=1}{\overset{M_2}{\mathrm{\Σ}}}{Y}^{i}2\mathrm{seed}}{M_2}$ (VII)

All barycenter of anchor node:

$X_e=\frac{\underset{i=1}{\overset{M_1}{\mathrm{\Σ}}}{X}^{i}1\mathrm{seed}+\underset{i=1}{\overset{M_2}{\mathrm{\Σ}}}{X}^{i}2\mathrm{seed}}{M_1+M_2}$

$Y_e=\frac{\underset{i=1}{\overset{M_1}{\mathrm{\Σ}}}{Y}^{i}1\mathrm{seed}+\underset{i=1}{\overset{M_2}{\mathrm{\Σ}}}{Y}^{i}2\mathrm{seed}}{M_1+M_2}$ (VIII)；

Step 2: the generation of sample area comprises the following first step to the three steps

The first step: according to estimated position and the position error in a moment on node node, determine the half of the anchor box length of side, length is R, and its expression formula is:

$R=\sqrt{{(X_e-X_{\mathrm{node}\_g}(k-1,i))}^2}+\mathrm{error}(k-1,i)$ (IX)

In formula: x _{node_g}(k-1, i) is the estimated position in a moment on node; Error (k-1, i) is the position error of a moment node on node to be positioned;

Second step: determine anchor box, be now divided into three kinds of situations:

(1), while only having a jumping anchor node, according to formula (VI), can obtain anchor box suc as formula shown in (X)

x _1min＝(x _1e-R) x _1max＝(x _1e+R)

y _1min＝(y _1e-R) y _1max＝(x _1e+R) (X)

(2), while only having double bounce anchor node, according to formula (VII), can obtain anchor box suc as formula shown in (XI)

x _2min＝(x _2e-R) x _2max＝(x _2e+R)

y _2min＝(y _2e-R) y _2max＝(x _2e+R) (XI)

(3) when an existing jumping anchor node has again double bounce anchor node

Calculate respectively the anchor box of a jumping anchor node and double bounce anchor node, shown in (XII), (XIII):

One jumps the anchor box of anchor node:

x _1min＝(x _1e-R) x _1max＝(x _1e+R)

y _1min＝(y _1e-R) y _1max＝(x _1e+R) (XII)

The anchor box of double bounce anchor node:

x _2min＝(x _2e-R) x _2max＝(x _2e+R)

y _2min＝(y _2e-R) y _2max＝(x _2e+R) (XIII)

Judge whether the two has common factor:

If there is common factor, final anchor box is

x _min＝max(x _1min，x _2min) x _max＝min（x _1max，x _2max）

y _min＝max（y _1min，y _2min） y _max＝min（y _1max，y _2max） （XIV）

If without common factor, final anchor box is

x _min＝（x _e-R） x _max＝（x _e+R）

y _min＝（y _e-R） y _max＝（x _e+R） （XV）

The 3rd step: determine sampling box, Box ⁱ _t=(x ⁱ _min, x ⁱ _max, y ⁱ _min, y ⁱ _max), sampling box is anchor box and border box, on node centered by the estimated position in a moment, with 2 × V _maxfor the common factor of the square area of the length of side, computational methods are as follows:

x ⁱ _min=max（x _min，x ⁱ _t-1-v _max） x ⁱ _max=min（x _max，x ⁱ _t-1+v _max）

y ⁱ _min=max（y _min，y ⁱ _t-1-v _max） y ⁱ _max=min（y _max，y ⁱ _t-1+v _max）（XVI）

Wherein (x ⁱ _t-1, y ⁱ _t-1) represent the sampling particle l of this node node in (t-1) moment ⁱ _t-1coordinate, V _maxrepresent the maximum translational speed of node, after obtaining sampling box, sampling obtains the particle l in t moment ⁱ _t;

Step 3: sampling particle is carried out to filtering

p(o _t|l ⁱ _t)＝1，

tr≤d(l _t,s)≤2×tr

p(o _t|l ⁱ _t)＝0，otherwise

Wherein o _trepresent observation information, S is the set of a jumping anchor node, and T is the set of double bounce anchor node, d (l _t, s) be sampling particle l _teuclidean distance with anchor node;

After filtering completes, if the needed number N of the not enough location estimation of population is carried out resampling, and repeated filtering, until reach hits N or reach maximum sampling number;

Step 4: determine the location estimation value of this node node to be positioned,

Location estimation value

Images