CN103152824B - Positioning method of node in wireless sensor network - Google Patents

Abstract

The invention discloses a positioning method of a node in a wireless sensor network. The positioning method comprises the following steps of: establishing a rectangular coordinate system to obtain position coordinates of each anchor node in each wireless sensor network; enabling a node to be positioned to acquire ID (Identification) and position coordinates of all the anchor nodes as well as hop counts of the node to be positioned and the corresponding anchor node in the wireless sensor network through a distance vector exchange agreement, and establishing a corresponding anchor node information list for the node to be positioned; inquiring a self anchor node information list by the node to be positioned; when the number of the anchor nodes is more than or equal to 3 in a 1 hop range, carrying out ranging by using the node to be positioned to acquire a ranging sample between the node to be positioned and each anchor node; and processing the ranging sample by using a voting average combined filtering algorithm to establish a CSS (Chirp Spread Spectrum) ranging mode to acquire accurate ranging information between the node to be positioned and each anchor node.

Description

A kind of wireless sensor network interior joint localization method

Technical field

The present invention relates to a kind of wireless sensor network node positioning field, particularly a kind of based on filtering algorithm and CSS(Chirp Spread Spectrum, linear frequency modulation spread spectrum) the WSN(Wireless Sensor Network of model of finding range, wireless sensor network) node positioning method.

Background technology

In numerous application of wireless sensor network (Wireless Sensor Network, WSN), the positional information of sensor node is vital, because do not have the Monitoring Data of node location information nonsensical often.The accurate location of node self is to provide the prerequisite of monitoring event position information, is also the basis realizing Moving objects location, track following prediction, network topology control and network route optimization.

Existing wireless sensor network locating method is divided into according to the need of range finding: based on the localization method of range finding and the localization method without the need to range finding.Localization method based on range finding has: the location based on TOA, the location based on TDOA, the location based on AOA and the location based on RSSI.Wherein, the location based on TOA needs strict time synchronization between nodes, is difficult to be applied to massive wireless sensor location; Location based on TDOA is limited to the non-straight length problem of ultrasonic transmission Distance geometry; Location based on AOA needs extra hardware, and hardware size and power consumption are difficult to meet practical application; Location based on RSSI meets the requirement of low-power, low cost, but has larger position error.Without the need to the localization method of range finding as centroid algorithm, DV-Hop(Distance Vector-Hop) etc., have low cost, the advantage such as little affected by environment, but there is range error accumulation problem, position error is also larger simultaneously.

Recently, linear frequency modulation spread spectrum (Chirp Spread Spectrum, CSS) technology adopt by IEEE802.15.4a standard, become a kind of new technique being applied to WSN node locating.This technology not only supports low energy ezpenditure, also supports precision ranging, is applicable to very much being applied to wireless sensor network node location.But this technology is applied to WSN node locating still to be existed following problem and needs solution badly: (1), because around adverse circumstances are on the impact of wireless signal, CSS sample of finding range exists random measurement noise and abnormal point distance measurement pollution problem, severe exacerbation range accuracy; (2) due to sensor node finite energy, how the efficient filtering algorithm of simplicity of design improves positioning performance is a great problem.

Summary of the invention

Goal of the invention: technical problem to be solved by this invention is for the deficiencies in the prior art, provides a kind of wireless sensor network interior joint localization method.

In order to solve the problems of the technologies described above, the invention discloses a kind of wireless sensor network interior joint localization method, containing following step:

Set up rectangular coordinate system, obtain the positional information of each anchor node in wireless sensor network; And by distance vector exchange agreement, make the ID of anchor node in node to be positioned acquisition network, position coordinates and corresponding jumping figure, and set up corresponding anchor node information table; Anchor node is the node that in wireless sensor network, self-position is known;

The anchor node information table of the querying node to be positioned self in network, when the anchor node number had within the scope of its 1 jumping is greater than or equal to 3, this node to be positioned starts range finding, obtains the range finding sample between node to be positioned and each anchor node;

Design ballot-average Federated filter algorithm processes range finding sample, and sets up CSS range finding model to obtain the precision ranging information between node to be positioned and anchor node.

The present invention utilizes CSS technology (can see H.-S.Ahn, H.Hur, W.-S.Choi, One-way Ranging Technique for CSS-based Indoor Localization, IEEE International Conference on Industrial Informatics, Daejeon, Korea, 2008.) distance between repetitive measurement self and anchor node, obtains N number of range finding sample, uses represent, wherein represent the secondary sample of finding range of kth between this node to be positioned and anchor node j.

Ballot described in step 3-average Federated filter algorithm comprises the steps:

(1) formula (1) is utilized to ask for the sample average μ of N number of range finding sample;

$\mathrm{\μ}=\frac{d_j^1+d_j^2+...+d_j^k}{N}---\left(1\right),$

(2) experimentally simulation definition standard deviation sigma, Voting Model adopts gaussian probability distribution function, and through type (2) calculates the ballot probability of each range finding sample

$P\left(d_j^k\right|\mathrm{\μ})=\frac{1}{\sqrt{2\mathrm{\π}}\·\mathrm{\σ}}{e}^{-\frac{{(d_j^k-\mathrm{\μ})}^2}{{2\mathrm{\σ}}^2}}---\left(2\right),$

The span of Plays difference σ of the present invention is 0.05≤σ≤0.5, σ value is less, be more conducive to suppressing exception range finding sample, but σ value is too small, the ballot probability of part normal range finding sample also can be too small, thus cause part normal range finding sample to be deleted by mistake.

(3) define the value of the confidence λ according to the actual requirements, range finding sample is voted, when the ballot probability of range finding sample when being less than the value of the confidence λ, just abandoning this range finding sample, otherwise just retain.The range finding sample number M retained represents.Advise in the present invention that the span of the value of the confidence λ is 0.8≤λ≤0.5, λ value is larger, be more conducive to suppressing exception range finding sample, but λ value is excessive, and part normal range finding sample is easily deleted by mistake.

(4) then retain M range finding sample to be averaged filtering according to formula (3), to obtain range finding estimated value

${\hat{d}}_j=\frac{1}{M}\underset{k=1}{\overset{M}{\mathrm{\Σ}}}{d}_j^k---\left(3\right),$

CSS range finding model described in step 3 is:

$d_j=\frac{{\hat{d}}_j-k_2}{k_1}---\left(4\right),$

Wherein: represent the range finding estimated value between node to be positioned and anchor node j obtained through ballot-average Federated filter process, k ₁and k ₂for Slope Parameters and the intercept parameter of CSS range finding model, d _jfor the accurate distance between node to be positioned to be asked for and anchor node j.

Obtained successively by step 3 node to be positioned and range information within the scope of its 1 jumping between anchor node, adopt maximum-likelihood estimation to obtain the position coordinates of node to be positioned.

The present invention, first by internodal information exchange, makes the ID of anchor node in node to be positioned acquisition network, position coordinates and corresponding jumping figure, and sets up anchor node information table; Node to be positioned initiatively initiates CSS distance measurement request according to the number of around anchor node, obtains N number of range finding sample; Then the precision ranging information that ballot-average Federated filter algorithm and CSS find range between model acquisition and anchor node is utilized; The position coordinates of node to be positioned is obtained finally by maximum-likelihood estimation.

Beneficial effect: ballot of the present invention-average Federated filter algorithm reduces the impact that surrounding environment is found range on CSS, realizes the internodal precision distance measurement of adjacent sensors by elimination random measurement noise and rejecting abnormalities point distance measurement; Propose CSS range finding model according to actual experiment simultaneously, improve CSS range accuracy further, achieve the WSN node positioning method of a kind of high position precision, low complex degree, low-yield expense on this basis.

Accompanying drawing explanation

To do the present invention below in conjunction with the drawings and specific embodiments and further illustrate, above-mentioned and/or otherwise advantage of the present invention will become apparent.

Fig. 1 is linear frequency modulation spread spectrum ranging process flow chart.

Fig. 2 is the process of filtering algorithm and the precision ranging of range finding model realization.

Embodiment

The invention discloses a kind of wireless sensor network interior joint localization method, contain following step: set up rectangular coordinate system, obtain the positional information of each anchor node in wireless sensor network; And by distance vector exchange agreement, make the ID of anchor node in node to be positioned acquisition network, position coordinates and corresponding jumping figure, and set up corresponding anchor node information table; Anchor node is the node that in wireless sensor network, self-position is known; The anchor node information table of the querying node to be positioned self in network, when the anchor node number had within the scope of its 1 jumping is greater than or equal to 3, this node to be positioned starts range finding, obtains the range finding sample between node to be positioned and each anchor node; Design ballot-average Federated filter algorithm processes range finding sample, and sets up CSS range finding model to obtain the precision ranging information between node to be positioned and anchor node.

Node to be positioned is found range, and the range finding sample obtained between node to be positioned and each anchor node comprises: utilize CSS technology to measure distance between node to be positioned self and anchor node for N time, obtain N number of range finding sample, use represent, wherein represent the secondary sample of finding range of kth between this node to be positioned and anchor node j.

Described use average Federated filter algorithm of voting carries out process comprise the steps: range finding sample

Following formula is utilized to ask for the sample average μ of N number of range finding sample:

$\mathrm{\μ}=\frac{d_j^1+d_j^2+...+d_j^k}{N},$

Established standards difference σ, Voting Model adopts gaussian probability distribution function, is calculated the ballot probability of each range finding sample by following formula $P\left(d_j^k\right|\mathrm{\μ}):$

$P\left(d_j^k\right|\mathrm{\μ})=\frac{1}{\sqrt{2\mathrm{\π}}\·\mathrm{\σ}}{e}^{-\frac{{(d_j^k-\mathrm{\μ})}^2}{{2\mathrm{\σ}}^2}},$

Setting the value of the confidence λ, votes, when the ballot probability of range finding sample to range finding sample when being less than the value of the confidence λ, just abandon this range finding sample, otherwise just retain, the range finding sample number M retained represents.

Then retain M range finding sample is averaged filtering according to the following formula, obtains range finding estimated value

${\hat{d}}_j=\frac{1}{M}\underset{k=1}{\overset{M}{\mathrm{\Σ}}}{d}_j^k.$

Described CSS range finding model is:

$d_j=\frac{{\hat{d}}_j-k_2}{k_1}$

Wherein: represent the range finding estimated value between node to be positioned and anchor node j, k ₁and k ₂for Slope Parameters and the intercept parameter of CSS range finding model, d _jfor the accurate distance between node to be positioned to be asked for and anchor node j.

Node to be positioned obtains and range information within the scope of its 1 jumping between anchor node successively, adopts maximum-likelihood estimation to obtain the position coordinates of node to be positioned.

Embodiment

The present embodiment comprises the following steps:

1. set up rectangular coordinate system, definition obtains the positional information of each anchor node in wireless sensor network; And by distance vector exchange agreement, make the ID of anchor node in node to be positioned acquisition network, position coordinates and corresponding jumping figure, and set up corresponding anchor node information table;

2. the anchor node information table of the querying node to be positioned self in network, when the anchor node number had within the scope of its 1 jumping is greater than or equal to 3, this node to be positioned starts range finding, and utilizes the distance between CSS technology repetitive measurement self and anchor node, obtain N number of range finding sample, use represent, wherein represent the secondary sample of finding range of kth between this node to be positioned and anchor node j.

CSS ranging process between node i to be positioned and anchor node j as shown in Figure 1, first i sends a ranging data bag to j, when j receives the ranging data bag from i, a response packet is sent at once to i, after i receives the response packet that j sends it back, just complete first time bidirectional ranging, obtain the propagation delay time T from i to j ₁with the data processing delay time T of j ₂.And then, initiated to carry out second time bidirectional ranging by j.Last j sends the result of the bilateral range finding of second time to i, thus i obtains the propagation delay time T from j to i ₃with the data processing time delay T of i ₄.Utilize T ₁, T ₂, T ₃and T ₄, the one-way transmission time T of node i to be positioned to anchor node j can be obtained _tOF, and then obtain range information between node i to be positioned and anchor node j

$T_{\mathrm{TOF}}=\frac{T_1-T_2+T_3-T_4}{4}$

$d_j^k=T_{\mathrm{TOF}}*c$

The wherein speed aloft transmitted of c wireless signal and the light velocity.Node duplicate measurements to be positioned N time, obtains N number of corresponding range finding sample.

N=20 in the present embodiment.

Design ballot-average Federated filter algorithm processes range finding sample, and sets up CSS range finding model to obtain the precision ranging information between node to be positioned and anchor node.The process of filtering algorithm and the precision ranging of range finding model realization as shown in Figure 2.

Wherein vote-average Federated filter algorithm process step is as follows:

(1) to N number of range finding sample ask sample average, computing formula as shown in the formula: sample average μ;

$\mathrm{\μ}=\frac{d_j^1+d_j^2+...+d_j^k}{N},$

Wherein μ obtains sample average for calculating.

(2) experimentally simulation definition standard deviation sigma, Voting Model adopts gaussian probability distribution function, is calculated the ballot probability of each range finding sample by following formula

$P\left(d_j^k\right|\mathrm{\μ})=\frac{1}{\sqrt{2\mathrm{\π}}\·\mathrm{\σ}}{e}^{-\frac{{(d_j^k-\mathrm{\μ})}^2}{{2\mathrm{\σ}}^2}},$

σ=0.2 in the present embodiment.

(3) define the value of the confidence λ according to the actual requirements, range finding sample is voted, when the ballot probability of range finding sample when being less than the value of the confidence λ, just abandoning this range finding sample, otherwise just retain.The range finding sample number M retained represents.

λ=0.7 in the present embodiment, when the ballot probability of range finding sample time, retain this range finding sample, otherwise just abandon.The range finding sample retained is

(4) then to M the range finding sample retained be averaged filtering, and average filter formula is as follows, and the range finding estimated value obtained after filtering is

${\hat{d}}_j=\frac{1}{M}\underset{k=1}{\overset{M}{\mathrm{\Σ}}}{d}_j^k,$

The CSS range finding model set up is shown below:

$d_j=\frac{{\hat{d}}_j-k_2}{k_1},$

Wherein: represent the range finding estimated value between node to be positioned and anchor node j obtained through ballot-average Federated filter process, k ₁and k ₂for Slope Parameters and the intercept parameter of CSS range finding model, least square curve fit can be adopted to obtain, d _jfor the accurate distance between node to be positioned to be asked for and anchor node j.

K in the present embodiment ₁=1.0247, k ₂=1.2655.Wherein k ₁, k ₂adopt least square curve fit to obtain according to the internodal range finding sample of two sensors under direct-view atmosphere experimental data, detailed process is as follows:

M that obtains sample data of finding range Wei (x ₁, y ₁), (x ₂, y ₂) ..., (x _m, y _m), and the functional relation defining x and y is y=k ₁x+k ₂, by as shown in the formula asking for k ₁, k ₂:

$\min {\mathrm{\Σ}}_{i=1}^m{[k_1\·x_i+k_2-y_i]}^2.$

Can be obtained successively by above-mentioned steps three node to be positioned and range information within the scope of its 1 jumping between anchor node, adopt maximum-likelihood estimation to obtain the position coordinates of node to be positioned.The detailed process of maximal possibility estimation is as follows:

If the coordinate of a known n anchor node is respectively (x ₁, y ₁), (x ₂, y ₂) ..., (x _n, y _n), the distance of they and node M to be positioned is respectively d ₁, d ₂..., d _n.The coordinate supposing node M for (, y), then there is following relation:

$\left\{\begin{array}{c}{(x_1-x)}^2+{(y_1-y)}^2=d_1^2\\ {(x_2-x)}^2+{(y_2-y)}^2=d_2^2\\ \·\·\·\·\·\·\\ {(x_n-x)}^2+{(y_n-y)}^2=d_n^2\end{array}\right.,$

Above formula system of linear equations can be expressed as:

AX=d，

Wherein each several part is respectively:

$A=2\left\{\begin{array}{cc}{x}_1-x_n& y_1-y_n\\ \·\·\·\·\·\·& \·\·\·\·\·\·\\ x_{n-1}-x_n& y_{n-1}-y_n\end{array}\right\},$

$d=\left\{\begin{array}{c}{x}_1^2-x_n^2+y_1^2-y_n^2+d_n^2-d_1^2\\ \·\·\·\·\·\·\\ x_{n-1}^2-x_n^2+y_{n-1}^2-y_n^2+d_n^2-d_{n-1}^2\end{array}\right\},$

$X=\left(\begin{array}{c}x\\ y\end{array}\right),$

The coordinate finally adopting Minimum Mean Squared Error estimation method can calculate node M to be positioned is: $\hat{X}=$ ${\left(A^{T}A\right)}^{-1}{A}^{T}d.$

Simulated environment is the region of 100m × 100m, 5 anchor nodes and 45 even random placements of node to be positioned.To typically carrying out Experimental comparison without the need to distance-measuring and positioning method DV-Hop, typical distance-measuring and positioning method RSSI and the present embodiment localization method in simulated environment, the range finding of three kinds of localization methods and the result of location as shown in table 1:

The experiment of table 1 three kinds of localization methods is compared

	DV-Hop localization method	Based on the localization method of RSSI	The present embodiment localization method
				Average range error	1.62m	1.14m	0.35m
Average localization error	2.17m	1.35m	0.41m

As can be seen from the above table, localization method of the present invention is all significantly better than existing method in positioning precision and range accuracy.

The invention provides a kind of thinking and method of wireless sensor network interior joint localization method; the method and access of this technical scheme of specific implementation is a lot; the above is only the preferred embodiment of the present invention; should be understood that; for those skilled in the art; under the premise without departing from the principles of the invention, can also make some improvements and modifications, these improvements and modifications also should be considered as protection scope of the present invention.The all available prior art of each part not clear and definite in the present embodiment is realized.

Claims (1)

1. a wireless sensor network interior joint localization method, is characterized in that, comprises the following steps:

Set up rectangular coordinate system, obtain the position coordinates of each anchor node in wireless sensor network; By distance vector exchange agreement, make node to be positioned obtain the jumping figure of the ID of all anchor nodes in wireless sensor network, position coordinates and node to be positioned and respective anchors node, node to be positioned sets up the anchor node information table of correspondence;

The anchor node information table of querying node to be positioned self, when the anchor node number had within the scope of its 1 jumping is greater than or equal to 3, this node to be positioned is found range, and obtains the range finding sample between node to be positioned and each anchor node;

Use the average Federated filter algorithm of ballot to process range finding sample, and set up CSS model of finding range and obtain accurate ranging information between node to be positioned and anchor node;

Node to be positioned is found range, and the range finding sample obtained between node to be positioned and each anchor node comprises: utilize CSS technology to measure distance between node to be positioned self and anchor node for N time, obtain N number of range finding sample, use represent, wherein represent the secondary sample of finding range of kth between this node to be positioned and anchor node j;

Described use average Federated filter algorithm of voting carries out process comprise the steps: range finding sample

Following formula is utilized to ask for the sample average μ of N number of range finding sample:

$\mathrm{\μ}=\frac{d_j^1+d_j^2+...+d_j^k}{N},$

Established standards difference σ, Voting Model adopts gaussian probability distribution function, is calculated the ballot probability of each range finding sample by following formula

$P\left(d_j^k\right|\mathrm{\μ})=\frac{1}{\sqrt{2\mathrm{\π}}\·\mathrm{\σ}}{e}^{-\frac{{(d_j^k-\mathrm{\μ})}^2}{2{\mathrm{\σ}}^2}},$

Setting the value of the confidence λ, votes, when the ballot probability of range finding sample to range finding sample when being less than the value of the confidence λ, just abandon this range finding sample, otherwise just retain, the range finding sample number M retained represents;

Then retain M range finding sample is averaged filtering according to the following formula, obtains range finding estimated value

${\hat{d}}_j=\frac{1}{M}\underset{k=1}{\overset{M}{\mathrm{\Σ}}}{d}_j^k;$

Described CSS range finding model is:

$d_j=\frac{{\hat{d}}_j-k_2}{k_1}$

Wherein: represent the range finding estimated value between node to be positioned and anchor node j, k ₁and k ₂for Slope Parameters and the intercept parameter of CSS range finding model, d _jfor the accurate distance between node to be positioned to be asked for and anchor node j;

Node to be positioned obtains and range information within the scope of its 1 jumping between anchor node successively, adopts maximum-likelihood estimation to obtain the position coordinates of node to be positioned.