CN103220746B - A kind of wireless sensor network node method for self-locating - Google Patents

Abstract

The present invention relates to a kind of node self-localization method of wireless sensor network, comprise following steps: adopt distance vector exchange agreement, flooded broadcast, to network, makes all nodes get the jumping hop count order of all anchor nodes; Cluster is carried out to anchor node, forms several subregions.According to the jumping segment distance information between anchor node, factor pair current sub network network topological structure judges to utilize network topology structure to judge; Set interior element is screened one by one, finds out the node meeting promotion condition and the reference anchor node that it is positioned.The node meeting decision condition is promoted; Again travel through all nodes, perform and promote strategy, until all unknown node complete location.The present invention compared with prior art, DV-distance algorithm basis is added node and promotes strategy, reduce the distance accumulation error of route vector between nonuniform network interior joint, improve overall positioning precision.

Description

A kind of wireless sensor network node method for self-locating

Technical field

The present invention relates to a kind of wireless sensor network node method for self-locating, be applicable to the sensor network system with Self-organization.

Background technology

Wireless sensor network have disguised strong, can rapid deployment, can self-organizing, cost low, the advantage such as can work under severe and particular surroundings, at present, we is also in the starting stage to the research of wireless sensor network, still there is key technology problems and needs solution badly.Massive wireless sensor be usually all broadcasted sowing by aircraft, the mode such as rocket ejection is at random dispersed in area to be monitored, the positional information of node has very large uncertainty.And for not determining the node of positional information, its data message collected almost does not have any using value.Therefore in the types of applications of wireless sensor network, the monitoring of positional information to network itself of node is extremely important, can determine that locale or the node location released news are all one of basic functions of wireless sensor network, key effect is played to the practicality of sensor network and validity.Such as in the comprehensive safety protecting of important goal or sensitive target, need based on sensor node positional information, determine abnormal particular location occurs, and the alarm grade of invasion; In target following, sensor node perceives the speed of moving target, needs the positional information in conjunction with oneself, could monitor the moving line of target, and target of prediction movement locus, monitors to have immeasurable meaning to the precision strike of target and situation of battlefield.

Massive wireless sensor location is general adopts range-free localization method, and network topology structure brings a lot of uncertain factor to range-free localization algorithm.For regular network, existing algorithm well can realize node locating, but reality is due to the impact of other objective factors such as landform, environment and arrangement, and scramble network ubiquity, causes many location algorithms not meet the demands.Owing to there is " node cavity " and " blind area " in scramble network, namely empty in route at a certain region memory, cause the multihop routing vector between this region interior nodes to form an arc route, the direct impact that arc route is brought is exactly cause internodal estimated distance will much larger than its actual value.Traditional route Vector operation euclidean distance between node pair method is difficult to adapt to this topological structure, must improve.

Summary of the invention

Goal of the invention: the present invention is directed in traditional range-free localization process and to convert the distance obtained based on hop count information, is easily subject to network topology structure impact, especially in scramble network medial error this problem more obvious, proposes a kind of new localization method,

Technical scheme: the technical solution adopted in the present invention is as follows:

A node self-localization method for wireless sensor network, comprises following steps:

1) adopt distance vector exchange agreement, flooded broadcast, to network, makes all nodes get the jumping hop count order of all anchor nodes;

2) according to the distribution of network size and anchor node, suitably cluster is carried out to anchor node, form several subregions.In each sub regions, according to the jumping segment distance information between anchor node, factor delta judges current sub network network topological structure to utilize network topology structure to judge;

3) as judgement factor delta >2, to the jumping figure of anchor node, ascending sort is carried out to each unknown node, then chooses the set of anchor node corresponding to all jumping figures, be denoted as { set:Hop, Num, ID}, and the Hop item in this set is initialized as { min _hopsset, wherein Hop representative is by the jumping figure of ascending sort, the anchor node number that Num is corresponding under representing current jumping figure, and ID represents the anchor node numbering under current jumping figure;

4) to set, { set:Hop, Num, ID} interior element screens one by one, finds out the node meeting promotion condition and the reference anchor node positioned it.

5) node meeting decision condition is promoted;

6) again travel through all nodes, perform and promote strategy, until all unknown node complete location.

Described network topology structure judges the factor $\mathrm{\δ}=\frac{\underset{\mathrm{ij}}{\mathrm{\Σ}}\underset{x=1}{\overset{{\mathrm{hop}}_{\mathrm{ij}}}{\mathrm{\Σ}}}{\mathrm{hopdist}}_x}{\underset{\mathrm{ij}}{\mathrm{\Σ}}{\mathrm{dist}}_{\mathrm{ij}}},$ Wherein ${\mathrm{dist}}_{\mathrm{ij}}=\sqrt{{(x_i-x_j)}^2+{(y_i-y_j)}^2},$ (x _i, y _i), (x _j, y _j) be the coordinate of node i, j, hop _ijfor jumping hop count, hopdist _xfor every section is jumped distance.

In step 4, element screening step is as follows:

1) a threshold value Num is chosen in set set _thas decision condition;

2) using Hop item corresponding with it as { min _hopsset upgrade;

3) judgement of range difference least square is carried out to the anchor node of ID each under this set, namely calculate and whether meet

$\underset{i}{\mathrm{\Σ}}\underset{j}{\mathrm{\Σ}}{({\mathrm{hop}}_i-{\mathrm{hop}}_j)}^2=\min ;$

Num element number sum >=3 in described set set.

Beneficial effect: the present invention adds node and promotes strategy on DV-distance algorithm basis, adopts DV-distance to locate, the node completing location is promoted as anchor node in the isotropism dense network of local.Reducing the jumping hop count of route vector by increasing the way promoting anchor node, reducing the possibility of arc route appearance with this, reaching the effect improving large radian multihop routing vector.Along with the carrying out of position fixing process, anchor density constantly increases, and reduces the distance accumulation error of route vector between nonuniform network interior joint, improves overall positioning precision.

Accompanying drawing explanation

Fig. 1 is flow chart of the present invention.

Fig. 2 is regular network topological diagram.

Fig. 3 is C type scramble network topological diagram.

Fig. 4 is algorithm flow chart of the present invention.

Embodiment

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

As shown in Figure 1, wireless sensor network method for self-locating provided by the invention, comprises following steps:

Step one: adopt distance vector exchange agreement, flooded broadcast, to network, makes all nodes get the jumping hop count order of all anchor nodes.

Step 2: according to the distribution of network size and anchor node, suitably cluster is carried out to anchor node, form several subregions.In each sub regions, according to the jumping segment distance information between anchor node, factor delta judges current sub network network topological structure to utilize network topology structure to judge.

For massive wireless sensor, nodes number is too much, if adopt global calculation, first cause amount of calculation very large, next brings network topology structure identification inaccurate.Region dividing can be carried out to global network, be divided into several sub-networks, then Distributed Calculation be carried out to each sub-network.

Because anchor node is the known node in position, dynamic clustering can be carried out to them, form the sub-network that several are organized into by anchor node, then Topology identification be carried out to each sub-network.

C-means Method, as the classic algorithm of dynamic clustering, implements more convenient, fast convergence rate, can obtain good Clustering Effect.Weak point must specify classification number before being cluster.This algorithm basis is error sum of squares criterion, and core concept is as follows:

(1) according to particular problem, from sample set, N number of proper sample is selected by rule of thumb as initial cluster center;

(2) get a sample, be classified in that class of the cluster centre nearest with it, recalculate sample average, upgrade cluster centre.Then a sample is taken off, repetitive operation, until all samples are included in respective class;

(3) whether cluster is reasonable, unreasonable to adopt error sum of squares criterion function to judge, revises classification.Circulation is carried out judging, revising until reach algorithm end condition.

In the very large situation of network size, when anchor node distributes C-shaped network, arranging cluster number is 3; When distribution is in O type network, cluster number is set to 4.

Step 3: for regular network, traditional APS algorithm can solve orientation problem very well, and node does not need to promote.For irregular network, then node is needed to promote.How recognition network topological structure can judge as follows:

The topological structure of analysis rule network (Fig. 2 shown in) and irregular network (being that representative graph 3 is shown with C-network), the actual range between can finding out according to anchor node and the ratio between measuring distance, can suitably distinguish.If anchor node i, j coordinate is respectively (x _i, y _i), (x _j, y _j), the actual range between them ${\mathrm{dist}}_{\mathrm{ij}}=\sqrt{{(x_i-x_j)}^2+{(y_i-y_j)}^2},$ Jumping hop count is hop _ij, every section is jumped distance is hopdist _x.Note δ is that network topology structure judges the factor, $\mathrm{\δ}=\frac{\underset{\mathrm{ij}}{\mathrm{\Σ}}\underset{x=1}{\overset{{\mathrm{hop}}_{\mathrm{ij}}}{\mathrm{\Σ}}}{\mathrm{hopdist}}_x}{\underset{\mathrm{ij}}{\mathrm{\Σ}}{\mathrm{dist}}_{\mathrm{ij}}}.$ In actual conditions, the cumulative measurement distance between anchor node is all greater than air line distance, i.e. δ >1, but the δ value in regular network much smaller than the δ value in irregular network, will arrange δ=2 for decision threshold in the present invention.When δ≤2 are judged to be regular network, otherwise it is irregular network.

As judgement factor delta >2, known current network is irregular network, and node should be adopted to promote strategy.First to the jumping figure of anchor node, ascending sort is carried out to each unknown node, then choose the set of anchor node corresponding to all jumping figures, be denoted as set:Hop, Num, ID}, and the Hop item in this set is initialized as { min _hopsset, wherein Hop representative is by the jumping figure of ascending sort, the anchor node number that Num is corresponding under representing current jumping figure, and ID represents the anchor node numbering under current jumping figure.

Step 4: when recognition network topological structure, by the δ value between test anchor node, can find out between them whether there is multi-hop arc route.Such as, in Fig. 3 δ ₁₂, δ ₁₃will much larger than δ ₂₃, then UNICOM A1, the localized network of A2, A3 is irregular certainly.If X node in figure can be promoted as anchor node, then by the localized network approximate regulation network of X, A2, A3 UNICOM, can be positioned by APS algorithm.In fact, X node uniquely can be determined by following method:

If unknown node X to be promoted is respectively hop to the jumping figure of anchor node i, j _i, hop _j, ensure hop _i, hop _j∈ { min _hops, and namely meet X minimum to the range difference quadratic sum of each anchor node, then uniquely can determine X.Wherein { min _hopsrepresenting the set that less hop forms, the node that can promote is always preferential searches for anchor node in this set, then utilizes range difference Least square-fit, filters out concrete anchor node and position.

Can illustrate that the node locating precision of above selection is the highest theoretically according to the Maximum-likelihood estimation in node location evaluation method, be the optimum node promoted.Suppose that the coordinate of n point is respectively (x ₁, y ₁), (x ₂, y ₂), (x ₃, y ₃) ..., (x _n, y _n), corresponding distance is d ₁, d ₂, d ₃..., d _n, the coordinate of unknown node X is (x, y), and so formula (1) is always set up:

(x-x _i) ²+(y-y _i) ²=d _i ²,i=1,2,...,n…(1)

Employing maximum-likelihood method solves, and can obtain the coordinate of X , wherein

$A=2\left[\begin{array}{cc}{x}_1-x_2& y_1-y_2\\ \·& \·\\ \·& \·\\ \·& \·\\ x_1-x_n& y_1-y_n\end{array}\right],$ $b=\left[\begin{array}{c}{x_1}^2-{x_2}^2+{y_1}^2-{y_2}^2+{d_1}^2-{d_2}^2\\ \·\\ \·\\ \·\\ {x_1}^2-{x_n}^2+{y_1}^2-{y_n}^2+{d_1}^2-{d_n}^2\end{array}\right],$ $\hat{X}=\left[\begin{array}{c}x\\ y\end{array}\right]\·\·\·\left(2\right)$

According to formula (2), if error is minimum, then d _i ²-d _j ²minimum.And d _i ²-d _j ²=(d _i+ d _j) (d _i-d _j), wherein (d _i-d _j)=|| hop _i-hop _j||, obtain (hop _i-hop _j) ²minimum.Thus the node promoted demonstrating selection is optimum node.

To set, { set:Hop, Num, ID} interior element screens one by one.First ensure that the Num number sum of set set interior element at least equals three with the demand of satisfied location, a threshold value Num can be chosen _th>=3 as decision condition, and using the Hop item of correspondence as { min _hopsset upgrade, then the judgement of range difference least square is carried out to the anchor node of ID each under this set, namely calculates and whether meet by this process, the node meeting promotion condition and the reference anchor node that it is positioned can be found out.

Step 5: for the node meeting decision condition, can think that this node meets strict positioning requirements, position error Local Minimum, allow to be promoted.So far, first round node is promoted complete.

Meet the unknown node of promotion condition, due to the condition that itself demand fulfillment is strict, so in the middle of all unknown node, their positioning precision is the highest.In figure 3, X node meets promotion condition, and the preferential maximum-likelihood method that adopts is located.Although the estimated position obtained and actual position deviation to some extent, for the location of follow-up unknown node, compare and select A1 anchor node as the arc route accumulated error brought with reference to node, this departure is negligible.

Under normal circumstances, the impact of multihop routing accumulated error significantly can be reduced after the anchor node promoted replaces part actual anchors node, now select to promote anchor node as reference node, coordinate other actual anchors nodes, adopt DV-distance algorithm to obtain range information, then utilize the optimized algorithm such as maximum-likelihood method or simulated annealing to position.If unknown node is found to promote the jumping hop count of anchor node when differing very little with the jumping hop count to actual anchors node, consider and promote the position error that node exists self after all, should abandon utilizing promotion anchor node to select actual anchors node to replace it as with reference to node.

Step 6: again travel through all nodes, performs and promotes strategy, until all unknown node complete location.

If exist in network and take turns promotion more, need the anchor node setting priority to promoting, the anchor node priority attained promotion at first is the highest.Promote anchor node as with reference to when node in selection, first will select the node that priority is the highest, can reduce like this in the accumulated error selecting to bring when promoting anchor node location.As shown in Figure 4, the node that the present invention proposes promotes the complete flow chart of strategy.

Claims (2)

1. a node self-localization method for wireless sensor network, is characterized in that, comprise following steps:

1) adopt distance vector exchange agreement, flooded broadcast, to network, makes all nodes get the jumping hop count order of all anchor nodes;

2) according to the distribution of network size and anchor node, suitably cluster is carried out to anchor node, form several subregions, in each sub regions, according to the jumping segment distance information between anchor node, factor delta judges current sub network network topological structure to utilize network topology structure to judge; Described network topology structure judges the factor (x _i, y _i), (x _j, y _j) be the coordinate of node i, j, hop _ijfor jumping hop count, hopdist _xfor every section is jumped distance;

3) as judgement factor delta >2, to the jumping figure of anchor node, ascending sort is carried out to each unknown node, then chooses the set of anchor node corresponding to all jumping figures, be denoted as { set:Hop, Num, ID}, and the Hop item in this set is initialized as { min _hopsset, wherein Hop representative is by the jumping figure of ascending sort, the anchor node number that Num is corresponding under representing current jumping figure, and ID represents the anchor node numbering under current jumping figure;

4) to set, { set:Hop, Num, ID} interior element screens one by one, finds out the node meeting promotion condition and the reference anchor node positioned it;

5) node meeting decision condition is promoted;

6) again travel through all nodes, perform and promote strategy, until all unknown node complete location.

2. method for self-locating according to claim 1, is characterized in that: in step 4, element screening step is as follows:

1) a threshold value Num is chosen in set set _thas decision condition;

2) using Hop item corresponding with it as { min _hopsset upgrade;

3) judgement of range difference least square is carried out to the anchor node of ID each under this set, namely calculate and whether meet

Num element number sum >=3 in described set set.