CN106211256A - A kind of Unmanned Aerial Vehicle Data collection method based on data critical node - Google Patents

Abstract

The invention discloses a kind of Unmanned Aerial Vehicle Data collection method based on data critical node, by being mounted on unmanned plane by sink node, improve sink node motion limitation problem；By to the collection of some key node data in network, whole network data being carried out the completion of science, it is ensured that unmanned plane normally completes the collection work to whole network data situation under limited electricity, it is ensured that monitored area is long-term and effective data collection effort.The theory of the data gathered with the whole sensing node of data critical node on behalf that the present invention proposes, can not only be applied to Unmanned Aerial Vehicle Data collection aspect, can also well be applied under large-scale network environment, be played minimizing cost, be improved the effect of work efficiency.

Description

A kind of Unmanned Aerial Vehicle Data collection method based on data critical node

Technical field

The invention belongs to data collection techniques field, relate to a kind of Unmanned Aerial Vehicle Data based on data critical node and collect Method.

Background technology

Data collection is the most important task of Sensor Network, and it is to connect sensing layer and the bridge of application layer.Effective Method of data capture can reduce the live load of sensing node, it is to avoid network " regional dead ", extends the existence week of network Phase.

On the whole, whether fix according to sink node in Sensor Network, data main in current Sensor Network can be received Diversity method is divided into two classes: the first kind is method of data capture based on fixing sink node, and Equations of The Second Kind is based on mobile sink Method of data capture.Wherein data method based on mobile sink refers to that sink node according to certain rule or exists freely Move in monitored area, and collect the data that data perception node is gathered.This kind of method can avoid Energy volution to ask completely Topic, and then extend Network morals.

But this kind of method is discussion sink node moves in monitored area when, has all given tacit consent to sink node permissible Move freely in monitored area, and application itself can allow sink node to move in monitored area.But multiple In the environment of miscellaneous, existing method of data capture based on mobile sink node all have ignored " the mobile restricted problem of sink " and " application scenarios limitation problem ".Therefore, in actual environment the application of this kind of method by critical constraints.The hugest in network size In the case of great, sink node to move collection receipt at the whole network, and the time is all huge with the cost of energy, carries some For the mobile device of sink node, the hugest time energy cost, it may be difficult to bear, thus again limit such Method application in reality scene.

Summary of the invention

The problem existed for above-mentioned prior art or defect, it is an object of the invention to, it is provided that a kind of based on data pass The Unmanned Aerial Vehicle Data collection method of key node, by being mounted on unmanned plane by sink node, improves sink node motion Limitation problem；By to the collection of some key node data in network, whole network data being carried out the completion of science, it is ensured that Unmanned plane normally completes the collection work to whole network data situation under limited electricity, it is ensured that monitored area is long-term and effective Data collection effort.

To achieve these goals, the present invention adopts the following technical scheme that

A kind of Unmanned Aerial Vehicle Data collection method based on data critical node, comprises the following steps:

Step one: it is all that all nodes in acquisition radio sensing network are gathered in certain time period [t, t+ α] Data；

Step 2: judge any two node n in all nodes_iAnd n_jThe most similar, obtain in radio sensing network all The similar situation of node；Judge any two node n_iAnd n_jThe most similar method is as follows:

Node n_iAnd n_jThe data sequence collected within the time period [t, t+ α] isWithIf data sequence WithBetween distanceIt is less thanThen node n_iWith node n_j[t, t+ α] is similar , otherwise, node n_iWith node n_j[t, t+ α] is dissimilar；Wherein, ε is application error；

Step 3: according to the similar situation of all nodes that step 2 obtains, in radio sensing network in all nodes Select data critical node；

Step 4: unmanned plane node key to data carries out data acquisition, obtains in whole radio sensing network all The data cases of node.

Specifically, described step 3 selects data critical node in radio sensing network in all nodes, its Specifically include following steps:

Step 3.1, the historical data set D of all nodes in given radio sensing network_preAnd application error ε, pass through Given history node data D_preDTW algorithm is utilized to be calculated the figure Sim of similar node in network with application error ε_graph；

Step 3.2, for the figure Sim of similar node_graphIn all nodes carry out being calculated the worth of each node Attribute；

Step 3.3, to figure Sim_graphIn each node travel through, find out the node that wherein worth attribute is maximum, this joint Point is a key node of data, is added to by this node in set cnodeset{}, simultaneously at figure Sim_graphIn find The neighbor node Sim of this data critical node_graph[cn] .neibor, at figure Sim_graphNode set in, delete data Key node and its neighbor node, obtain new figure Sim_graph；

Step 3.4, to the new figure Sim obtained_graphRepeat step 3.2～step 3.4, until figure Sim_graphInterior joint Collection be combined into sky；

Step 3.5, the node in set cnodeset{} is all of data critical node obtained.

Specifically, the formula that in described step 3.2, calculating worth attribute is used is:

${\mathrm{Sim}}_{graph}\[v\].worth=\frac{{\mathrm{Sim}}_{graph}\[v\]\·\left|neibor\right|}{dis({\mathrm{Sim}}_{graph}\[v\],cnodeset)}---\left(6\right)$

Wherein, v represents node, and the molecule of formula represents the neighbor node number of node v, and denominator represents that node v is with existing The meansigma methods of all data critical nodal distances.

Compared with prior art, the invention have the advantages that

1, by the analysis to perception data, pick out low volume data and be worth bigger sensing node, i.e. data critical Node, the data that they gather jointly, the data that whole sensing node gathers can be represented under the error requirements of application.So After, the conventional target that data critical node is collected as Unmanned Aerial Vehicle Data, thus highlight unmanned plane in data collection mistake Key task in journey, reduces unmanned plane workload in data-gathering process, finally improves the work efficiency of unmanned plane.

2, the theory of the data that the whole sensing node of present invention data critical node on behalf gathers, is possible not only to application In terms of Unmanned Aerial Vehicle Data collection, can also well be applied under large-scale network environment, be played minimizing cost, be improved The effect of work efficiency.

Accompanying drawing explanation

Fig. 1 is the flow chart of the present invention；

Fig. 2 is to be the physical distribution figure of data critical node；

Fig. 3 is application error outdoor data collection correspondence data critical Node distribution figure when being 0.01.

With detailed description of the invention the solution of the present invention done below in conjunction with the accompanying drawings and explain in further detail and illustrate.

Detailed description of the invention

Defer to technique scheme, see Fig. 1, the Unmanned Aerial Vehicle Data collection side based on data critical node of the present invention Method, comprises the following steps:

Step one: it is all that all nodes in acquisition radio sensing network are gathered in certain time period [t, t+ α] Data.Wherein, α represents the length of this time period.

Step 2: judge any two node n in all nodes in radio sensing network_iAnd n_jThe most similar, obtain nothing The similar situation of all nodes in line sensing network.The most similar method of decision node is as follows:

Node n_iAnd n_jThe data sequence with sequential meaning collected within the time period [t, t+ α] isWith If data sequenceWithBetween distanceIt is less thanThat is: Then node n_iWith node n_j[t, t+ α] is similar, otherwise, node n_iWith node n_j[t, t+ α] is dissimilar.

Wherein,Wherein d_i,tRepresent node n_iIn the data that t gathers, and t ＜ t₁ ＜ ... ＜ t+ α.

ε is application error, and it is defined as, and applies acceptable error in data.If application error ε=a, receive when final Collect to data differ between [-a, a] with the data truly collected time, apply be all acceptable.Setting for ε Do not have fixing rule to follow, mainly set according to the requirement of application.

Within in the time period [t, t+ α], node n_iWith node n_jWhen clock system is Complete Synchronization, internodal distanceComputing formula is:

${\mathrm{dis}}_{t,t+\mathrm{\λ}\×\mathrm{\θ}}^{i,j}=|d_{i,t}-d_{j,t}|+|d_{i,t+\mathrm{\θ}}-d_{j,t+\mathrm{\θ}}|+...+|d_{i,t+\mathrm{\λ}\×\mathrm{\θ}}-d_{j,t+\mathrm{\λ}\×\mathrm{\θ}}|---\left(1\right)$

If within the time period [t, t+ α], node n_iWith node n_jClock system does not has a Complete Synchronization, or occurs in that and lose Lose the situation of data, then utilize DTW algorithm to calculateWithBetween distanceAnd the essence of DTW algorithm is to tie up Spending shorter vector to have carried out " stretching ", then longer with dimension vector carries out distance and solves.

Below to node n_iWith node n_jWhen clock system is Complete Synchronization, the similarity determination methods of two nodes is just Really property proves:

If within the time period [t, t+ α], node n_iWith node n_jWhen clock system is Complete Synchronization and node n_jClock system System is Complete Synchronization, and does not the most lose data, and the data that two nodes collect the most within the time period are counting It is equal, i.e. in amountAssume within the time period [t, t+ α], a total of H collection period.If β represents On each collection period in time period [t, t+ α], for node n_iWith node n_jThe data collected, difference between them The set that absolute value is constituted.

β={ β₁,β₂,...,β_H}(2)

${\mathrm{\β}}_i=|d_{j,{\mathrm{\Δt}}_i}-d_{i,{\mathrm{\Δt}}_i}|---\left(3\right)$

${\mathrm{dis}}_{t,t+\mathrm{\α}}^{i,j}=\underset{{\mathrm{\β}}_i\∈\mathrm{\β}}{\Σ}{\mathrm{\β}}_i\≤\mathrm{\ϵ}\×max\left(\right|l_{t,t+\mathrm{\α}}^i|,|l_{t,t+\mathrm{\α}}^j\left|\right)=\mathrm{\ϵ}\×H---\left(4\right)$

Assuming that the element number being less than ε in β is z, the element number more than ε is Q.

As Q=0, then all elements in β is both less than ε.It represented within the time period [t, t+ α] in the physical sense In any identical collection period, for node n_iWith node n_jThe data gathered, the difference between them is all at the error model of application In enclosing, n the most at any time_iThe data gathered can replace n_jThe data gathered, so node n_iWith node n_jIt is similar 's.

As z ＞ ＞ Q, then the nearly all element in β is both less than ε.It represents at time period [t, t+ in the physical sense α] in most collection period in, for node n_iWith node n_jThe data gathered, the difference between them is all in application In range of error, in this case it is also believed that n_iThe data gathered can substitute for n_jThe data gathered, so node n_iWith Node n_jIt is similar.

As z ＜ ＜ Q, then the nearly all element in β is both greater than ε.Assume that in β, all element sums less than ε are β_smallThen:

$\underset{Q\→H}{\lim }\frac{\mathrm{\ϵ}\×H-{\mathrm{\β}}_{small}}{Q}-\mathrm{\ϵ}=0---\left(5\right)$

The physical significance of its formula (5) is, although now the nearly all element in β is both greater than ε, but they and ε Between differ the least by almost 0.In this case it is also believed that n_iThe data gathered can substitute for n_jThe data gathered, So node n_kWith node n_jIt is similar.

When the sizableness of z Yu Q, then the element number more than ε in β is essentially identical with the element number less than ε.By In in stable environment, do not have the data that node collects and situation fluctuated occur, thus can only be half even In the continuous time, n_iThe data gathered and n_jThe data gathered differ all in the range of application error, and in second half time, both adopt The data difference that collection arrives is not the most in the range of error of application.Therefore, when the sizableness of z Yu Q, its physical significance is joint Point n_jThe data gathered and node n_i, within first half section (second half section) time, between them, there is similarity in the data gathered, and Second half section, (first half section) was in the time, there is not similarity between them.Because the environment residing for node is in most cases Stable, do not have a great change, therefore it is unlikely that above-mentioned situation.

In sum, if within the time period [t, t+ α], node n_iWith node n_jClock system is Complete Synchronization, and The most do not lose data, and data acquisition sequenceWithBetween distance less than ε × H, then node n_iAnd node n_jIt is similar.

If within the time period [t, t+ α], node n_iWith node n_jClock system does not has a Complete Synchronization, or occurs in that and lose Lose the situation of data, then utilize DTW algorithm to calculateWithBetween distance.And the essence of DTW algorithm is that dimension is shorter Vector carried out " stretching ", then longer with dimension vector carries out distance and solves.Therefore, in like manner can obtain: if number According to acquisition sequenceWithBetween distance be less thanThen node n_iWith node n_jIt is similar.

If node n_iThe data gathered can replace node n_jThe data gathered, then claim node n_iAnd n_jAt time period [t, t+ α] It is inside similar, also referred to as node n_iAnd n_jSimilarity relation is there is within the time period [t, t+ α].If the data in [t, t+ α], can The most sufficiently to represent the environmental change around node, then claim node n_iAnd n_jIt is all similar in whole life cycle.? The lower part of the present invention assumes that the data in the selected time period the most sufficiently represent the change of environment.

Step 3: according to the similar situation of nodes all in radio sensing network, selects all joints in radio sensing network The key node of point；Concrete grammar is as follows:

Step 3.1, the historical data set D of all nodes in given radio sensing network_preAnd it is acceptable in algorithm Error in data i.e. application error ε.By given history node data D_preDTW algorithm is utilized to be calculated with application error ε, The figure Sim of similar node in network_graph；

Step 3.2, the figure Sim to similar node_graphIn all nodes carry out being calculated the worth of each node and belong to Property, the concrete formula calculating worth attribute is:

${\mathrm{Sim}}_{graph}\[v\].worth=\frac{{\mathrm{Sim}}_{graph}\[v\].\left|neibor\right|}{dis({\mathrm{Sim}}_{graph}\[v\],cnodeset)}---\left(6\right)$

Wherein, v represents node, and the molecule of formula represents the neighbor node number of node v, and denominator represents that node v is with existing The meansigma methods of all data critical nodal distances.And worth attribute i.e. data value, data value is each data perception The attribute that node is had, specifically represents the use value that the data gathered by it are had.In the present invention we assume that: just Often under situation, the data that all sensing nodes are gathered have identical use value, for simplicity it is set to 1.Cause This, for data critical node n_kIf corresponding coverage scope is Γ, then its data value is | Γ |.Because | Γ | represents in the physical sense: in the range of error of application, and the data that node gathers can replace set as representative The data that in Γ, all nodes gather, are employed to be used, so the data value of node is exactly the data of all nodes in Γ It is worth sum, is | Γ |.Therefore, the worth attribute of node is directly proportional to the number of its neighbor node, crucial with available data The distance of property node is inversely proportional to.

Step 3.3, to figure Sim_graphIn each node travel through, find out the node that wherein worth attribute is maximum, this joint Point is a key node of data, is added to by this node in set cnodeset{}, simultaneously at figure Sim_graphIn find The neighbor node Sim of this data critical node_graph[cn] .neibor, at figure Sim_graphNode set in, delete data Key node and its neighbor node, obtain new figure Sim_graph。

Step 3.4, to the new figure Sim obtained_graphRepeat step 3.2～step 3.4, until figure Sim_graphInterior joint Collection be combined into sky.

Step 3.5, the node in set cnodeset{}, it is all of data critical node finally given.

The idiographic flow of said method and use can be summarized as below graph:

The a collection of node N picked out from network_xAs data critical node: it meets following condition: 1, ensure net Remaining each node in network, at least with N_xIn a node be similar；2, N is worked as_xIn all nodes be in one straight During line, the minimum (overall length of line between critical length that is two node of farthest two the internodal critical length of physical distance Degree)；Work as N_xIn all nodes when being not on straight line, when ensureing by N_xIn external figure corresponding to all nodes Critical area minimum (between critical area i.e. node, line forms the area of figure)；3, condition one and condition two are met for any Node set, leave out the most unnecessary node, it is ensured that N_xIt is that element number is minimum and meet the set of node of two above condition Close.

According to the definition of data critical node, the select permeability of data critical node is converted to mathematical problem:

Assuming that whole network is a figure G, all sensing nodes in network are the summits in figure G.If by sense Primary data analysis is learnt, under the error requirements of application, two nodes exist similarity relation, between two the most corresponding summits There is limit, if there is no similarity relation, between two the most corresponding summits, there is not limit.Figure G is referred to as the whole network joint by us The similarity graph of a relation of point.How from network, all of data perception node to select data critical node, now may be used To be interpreted as finding a collection of summit N in figure G_xSo that in network, an at least neighbours summit, remaining each summit is at N_x In, it is ensured that N_xIn vertex number minimum, and critical length corresponding to these summits or critical area minimum.

The select permeability of this data critical node is the problem of minimum dominant set in graph theory, and it solves difficulty is NP- hard.Pertinent literature is that the algorithm using Greedy strategy solves to the solution of minimum dominant set problem, and the present invention is by right Greedy algorithm carries out some amendments, reasonably solves the select permeability of data critical node.

1) prove and solution formula

When the problem that the select permeability of data critical node is minimum dominant set in graph theory, it solves difficulty and is The proof of NP-hard

Prove: the minimum dominant set problem in graph theory, it solves difficulty is NP-hard.It is concrete is defined as follows: for One non-directed graph G (V, E), finds out a collection of node h, h ∈ V from G so that the summit in h all exists limit with the summit in V-h Relation, ensures that the vertex number in h is minimum simultaneously.If for being in each node, with network of network edge other The most there is not similarity relation in node.At this moment, these are in the node of edge zone, are data critical node certainly, institute simultaneously It is also certain for having the critical length corresponding to data critical node or critical area.Therefore, at this point for data critical The selection work of node is it is to be understood that find out minimum a collection of node N in a network_xSo that remaining each joint in network A point at least neighbor node is in N_xIn, and this minimum dominant set problem in graph theory just.Therefore, it can be seen that it is minimum Dominating set problem is a special case of the select permeability of data critical node in network.So, data critical joint in network The select permeability of point is NP-hard.

2) utilize algorithm to carry out the solution procedure of data critical node, be mainly made up of two parts.Part I is profit Use DTW algorithm, build the similarity graph of a relation of the whole network node.Owing to, during building similarity graph of a relation, judge every Whether there is similarity between individual node and other all nodes, therefore its time complexity is O (n²), it is assumed that node is corresponding The average length of data acquisition sequence is a, then the time complexity that Part I is overall is O (a²*n²).Part II is basis Similarity graph of a relation selects data critical node, from algorithm it can be seen that its core is to carry out all nodes twice Cycle criterion process.Therefore, the time complexity of Part II is O (n²).So all in all the time complexity of algorithm 1 is O(a²*n²), although the time complexity of algorithm 1 is O (n on the whole²) rank, but if the length of data acquisition sequence compares Greatly, i.e. the constant coefficient of value bigger then algorithm 1 time complexity of a will be very big, the most how to select suitable data acquisition Collection sequence is most important.

Owing to data critical node is the conventional destination node that unmanned plane carries out data collection, so its meeting and unmanned plane Communicate frequently.If therefore data critical node is the most constant, then its energy can quickly exhaust, so that network enters " seemingly-dead " state.For this problem, solution in this paper is: arrange an energy threshold for data critical node, When its energy is less than this threshold value, then find in the range of its coverage with existing data critical node average distance That little node, sets it to new data critical node.

If according to the similarity between node, the whole network node similarity relationships figure Sim constituted_graphIn, there is N number of top Point and M bar limit.The number assuming the data critical node finally calculated by algorithm is κ, from pertinent literatureTherefore, if the destination node that data critical node is collected as Unmanned Aerial Vehicle Data, then unmanned The task amount of machine data collection is minimum can be reduced

Step 4: unmanned plane carries out data acquisition to the data critical node in network, thus obtains in whole network The data cases of all nodes.

The effect of the present invention can be further illustrated by following emulation.

Emulation 1, verifies data critical node correctness in the inventive method.

Program in whole experiment and emulation is write by oneself completely, and development language is c++, the parameter of all simulated environment All arrange with reference to the parameter in true environment.This emulation use data set respectively, (1) outdoor data collection: be deployed in elm 41 data sensing nodes in the North Platform Ruins of Great Wall of woods town are gathered between 20 days at 1 day to 2015 December of December in 2015 Data.These 41 node bottoms use CC2530 hardware configuration, are responsible for gathering the temperature of the internal 15cm degree of depth of Ruins of Great Wall soil Angle value.(2) house data collection: between February 28 to 5 days April in 2004 in 2004, be deployed in Intel Berkeley The data that 54 data sensing nodes of Research laboratory are collected.These 54 node bottoms use TinyOS to operate system System, gathers temperature, humidity, illumination and the voltage data of self in surrounding.In experiment herein, for collect 4 kinds of data, only utilize the temperature data that node gathers.Utilize outdoor and house data collection separately below, to step 3c) middle proposition The correctness of selection algorithm of data critical node verify.The experimental result of emulation is described as follows:

When table 1 is that application error is set to 0.01, data critical node corresponding to outdoor data collection and corresponding Coverage scope.From table 1, can be clearly seen that whole network has 6 key nodes of data, therefore, in nothing In the case of abnormality sensing data, the task amount in whole data-gathering process will reduce 85.37%.These 6 data critical In property node, the coverage scope corresponding to 56,493, the 495 key nodes of number is bigger, is 19,12,7 to account for respectively Arrive whole number of network node purpose 92.67%.

Outdoor data collection correspondence data critical node situation during table 1 application error 0.01

Fig. 3 is then the true distribution situation of data critical node and coverage scope thereof: five-pointed star shape table in figure Registration represents the ordinary node of data according to key node, circle.

Fig. 3 represents different nodes with distinct symbols (A, B and C), the joint of the most corresponding key node of different pieces of information Point coverage:

● A node represents, the coverage scope corresponding to the 56 key nodes of number.These nodes are mainly distributed : outside western wall, south outside, the inner side of Dong Qiang within the walls.

● B node represents, the coverage scope corresponding to the 493 key nodes of number.These nodes are mainly distributed : outside side within the walls, west and east wall.

● C node represents, the coverage scope corresponding to the 495 key nodes of number.These nodes are mainly distributed : the outside of north wall.

For the coverage scope corresponding to the 56 key nodes of number.It is every that they are placed exactly in the sun in position It place irradiated.Therefore, the soil moisture in these places is the highest.So the data that these nodes are gathered have The biggest similarity.

For the coverage scope corresponding to the 493 key nodes of number.Because North Platform western wall physical features in town is higher, East wall physical features ratio is relatively low.The sun has the most all been blocked in the outside of Xi Qiang and the inner side of Dong Qiang, and body of wall is the thickest, so at Xi Qiang Inner side and Dong Qiang outside gather soil moisture value all ratios relatively low.So the data of these node collections have huge phase Like property.

For the coverage scope corresponding to the 495 key nodes of number.Due to the Shi Tiao road, outside of north wall, so Outside north wall, the temperature of soil is by ectocine is relatively big and other bodies of wall have certain difference.So the number that these nodes gather According to having huge similarity.

Table 2 is application error when being 0.0001, data critical node corresponding to house data collection and saving accordingly Point coverage.As can be seen from Table 2: a total of 7 key nodes of data, the node of the 45 key nodes of number covers Lid scope is 42 to the maximum, has accounted for the 70.77% of whole network node sum.

House data collection correspondence data critical node situation during table 2 application error 0.0001

Fig. 2 is the physical distribution figure of data critical node, as can clearly see from the figure the 45 key joint of number The coverage scope of point, is almost distributed in each corner of network, and the coverage of other data critical nodes is the most scattered The lower left corner being distributed in monitored area and the upper right corner.The first and last coverage model of data critical node and its correspondence Enclose, geographical distribution does not has any significant feature.This is because under indoor environment, environment is highly stable and indoor Air be also mutually circulation, be the most almost all identical at each local air temperature value, therefore basic between data It is all similar, so just there will be the above results.

Claims (3)

1. a Unmanned Aerial Vehicle Data collection method based on data critical node, it is characterised in that comprise the following steps:

Step one: obtain all data that all nodes in radio sensing network are gathered in certain time period [t, t+ α]；

Step 2: judge any two node n in all nodes_iAnd n_jThe most similar, obtain all nodes in radio sensing network Similar situation；Judge any two node n_iAnd n_jThe most similar method is as follows:

Node n_iAnd n_jThe data sequence collected within the time period [t, t+ α] isWithIf data sequenceWithBetween distance be less thanThen node n_iWith node n_j[t, t+ α] is similar, no Then, node n_iWith node n_j[t, t+ α] is dissimilar；Wherein, ε is application error；

Step 3: according to the similar situation of all nodes that step 2 obtains, select in all nodes in radio sensing network Data critical node；

Step 4: unmanned plane node key to data carries out data acquisition, obtains all nodes in whole radio sensing network Data cases.

2. Unmanned Aerial Vehicle Data collection method based on data critical node as claimed in claim 1, it is characterised in that described Selecting data critical node in radio sensing network in all nodes in step 3, it specifically includes following steps:

Step 3.1, the historical data set D of all nodes in given radio sensing network_preAnd application error ε, by given History node data D_preDTW algorithm is utilized to be calculated the figure Sim of similar node in network with application error ε_graph；

Step 3.2, for the figure Sim of similar node_graphIn all nodes carry out being calculated the worth of each node and belong to Property；

Step 3.3, to figure Sim_graphIn each node travel through, find out the node that wherein worth attribute is maximum, this node is i.e. It is a key node of data, this node is added in set cnodeset{}, simultaneously at figure Sim_graphIn find this number Neighbor node Sim according to key node_graph[cn] .neibor, at figure Sim_graphNode set in, delete data critical Property node and its neighbor node, obtain new figure Sim_graph；

Step 3.4, to the new figure Sim obtained_graphRepeat step 3.2～step 3.4, until figure Sim_graphThe set of interior joint For sky；

Step 3.5, the node in set cnodeset{} is all of data critical node obtained.

3. Unmanned Aerial Vehicle Data collection method based on data critical node as claimed in claim 2, it is characterised in that described Step 3.2 in calculate the formula that used of worth attribute and be:

${\mathrm{Sim}}_{graph}\[v\].worth=\frac{{\mathrm{Sim}}_{graph}\[v\].\left|neibor\right|}{dis({\mathrm{Sim}}_{graph}\[v\],cnodeset)}$

Wherein, v represents node, and the molecule of formula represents the neighbor node number of node v, and denominator represents node v and existing institute There is the meansigma methods of data critical nodal distance.