CN107205255A - Towards the method for tracking target of the wireless sensor network based on imaging sensor - Google Patents
Towards the method for tracking target of the wireless sensor network based on imaging sensor Download PDFInfo
- Publication number
- CN107205255A CN107205255A CN201710339773.6A CN201710339773A CN107205255A CN 107205255 A CN107205255 A CN 107205255A CN 201710339773 A CN201710339773 A CN 201710339773A CN 107205255 A CN107205255 A CN 107205255A
- Authority
- CN
- China
- Prior art keywords
- mrow
- node
- msub
- cluster node
- target
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
Classifications
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04W—WIRELESS COMMUNICATION NETWORKS
- H04W40/00—Communication routing or communication path finding
- H04W40/02—Communication route or path selection, e.g. power-based or shortest path routing
- H04W40/04—Communication route or path selection, e.g. power-based or shortest path routing based on wireless node resources
- H04W40/10—Communication route or path selection, e.g. power-based or shortest path routing based on wireless node resources based on available power or energy
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04W—WIRELESS COMMUNICATION NETWORKS
- H04W40/00—Communication routing or communication path finding
- H04W40/02—Communication route or path selection, e.g. power-based or shortest path routing
- H04W40/20—Communication route or path selection, e.g. power-based or shortest path routing based on geographic position or location
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04W—WIRELESS COMMUNICATION NETWORKS
- H04W84/00—Network topologies
- H04W84/18—Self-organising networks, e.g. ad-hoc networks or sensor networks
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02D—CLIMATE CHANGE MITIGATION TECHNOLOGIES IN INFORMATION AND COMMUNICATION TECHNOLOGIES [ICT], I.E. INFORMATION AND COMMUNICATION TECHNOLOGIES AIMING AT THE REDUCTION OF THEIR OWN ENERGY USE
- Y02D30/00—Reducing energy consumption in communication networks
- Y02D30/70—Reducing energy consumption in communication networks in wireless communication networks
Landscapes
- Engineering & Computer Science (AREA)
- Computer Networks & Wireless Communication (AREA)
- Signal Processing (AREA)
- Mobile Radio Communication Systems (AREA)
Abstract
The present invention provides a kind of method for tracking target towards the wireless sensor network based on imaging sensor, comprises the following steps：Using the distributed tracking structure based on cluster, multiple nodes constitute a task cluster and track target simultaneously, and each cluster node obtains target measurement value and passes the result to leader cluster node after carrying out independent calculating；Leader cluster node merges the state estimation that the observed value from different nodes draws target using SRCI, predicts subsequent time target status information；According to the target status information of prediction, the contribution decision-making mechanism by the use of proposition considers the energy expenditure of current cluster node and guard node and it is expected that information gain removes the online suitable node of selection of dynamic as new task cluster node and implements tracing task；Finally there is the dump energy of node to consider to select a suitable leader cluster node according to the distance of node and prediction target in selected cluster node.The present invention can balance the contradiction between the consumption of network energy and target tracking accuracy.
Description
Technical field
The present invention relates to wireless sensor network field, more particularly to it is a kind of towards the wireless sensing based on imaging sensor
The method for tracking target of device network.
Background technology
In numerous applications of wireless sensor network, it is therein to track a mobile target in a monitor area
One important application.Target following is applied to many different scenes, such as monitor, monitor animal habitat, Vehicle Detection with
And the tracking of invader and information report.In target following, when there is one or more target to appear in detection zone,
Some nodes are waken up the status information of detection target, then report these information to aggregation node, are integrated by aggregation node
Command centre is uploaded to after processing.Node detects target by sampling perception information, and these information include light, sound, figure
Picture and video etc..With the development of image procossing and sensor and semiconductor technology, the price of smart image sensors is
Rapid decrease, this has greatly facilitated the application of the wireless sense network based on imaging sensor.Therefore, in the mesh of wireless sense network
Go to obtain, handle and analyze the information on target as a kind of new trend using imaging sensor in mark tracking.So
And this target following technology based on image sensor network has with the target following technology in traditional wireless sense network
The observation of the viewing area limitation of very big difference, such as node, limited energy, relatively low computing capability and directionality
Deng, be required for we for these difference propose suitable for image sensing net trackings.
The Target Tracking System of current main flow is mainly distributed formula (all nodes are fusion center, and status is equal), divided
Dissipate formula (several nodes have a unified fusion center, but calculating is distributed in after each node is carried out and passes to fusion center)
With integrated system (node is only responsible for data being transmitted to Centroid, calculates in the unified progress of Centroid).For requiring nothing
Interrupt for the tracking system with high reliability, distributed and distributed system is most popular.For large-scale network
For distributed system be reliable and possess higher fault-tolerance because measured value between several different nodes constantly
Interaction.But this tracking system generally requires to consume very big energy, since it is desired that constantly processing data and and neighbours
Communication interaction is carried out between node.Because distributing tracking system equilibrium energy can be consumed and tracking accuracy well, so right
It is more favourable for the network of energy constraint.In distributed system, before target arrival, interdependent node can reach into one
All nodes in individual cluster, cluster can detect target, and data are reported into cluster head, be obtained by the different observed value of cluster head unified fusion
Go out the state estimation of target.Therefore distributing tracking needs the dynamic of Measurement fusion algorithm and task cluster node and its cluster head
State is elected and administrative mechanism.In many different blending algorithms, some mutation algorithms based on information filter such as extend letter
Breath filtering (extend information filter, EIF), volume information filtering (cubature information
Filter, CIF) and the filtering of square root volume information (square root cubature information filter,
) etc. SRCIF more suitable for distributing tracking system, because they can disperse computational load and be easier based on many of cluster
Carried out in sensor cooperation follow-up mechanism.In addition, the wireless image sensor network monitored suitable for field, in order to ensure
The validity of Detection task, usual network can be all made up of the node of massive band width and resource-constrained.Therefore, a target is usual
It can be arrived by many nodal tests.Under normal circumstances in distributed information filtering algorithm, more nodes participate in tracking, obtain
The estimation of dbjective state will be more accurate.But too many node participates in tracking, the huge of energy and bandwidth resources can be caused
Consumption, and then reduce the life cycle of tracking network.So, it is necessary to the task section of participation tracking in distributing tracking system
Click through the selection of Mobile state.In addition, in distributing follow-up mechanism, leader cluster node as current task cluster scheduler, it is necessary to
Consume to interact between more energy and its cluster node and communicate and merge the observation data from different nodes.Cause
This one suitable node of selection does cluster head, has for the operation effective, stable with follow-up mechanism is ensured of balance network consumption
Highly important meaning.
The content of the invention
The technical problems to be solved by the invention are that proposition is a kind of effective and stabilization towards based on imaging sensor
The method for tracking target of wireless sensor network, it is intended to balance the contradiction between the consumption of network energy and target tracking accuracy.
To achieve these goals, the technical solution adopted for the present invention to solve the technical problems is：
A kind of method for tracking target towards the wireless sensor network based on imaging sensor, comprises the following steps：
S1, intiating radio sensor network：In all nodes for tracking current goal, a cluster head section is selected
Point and multiple cluster nodes constitute task cluster, remaining node as guard node, wherein, the leader cluster node and cluster node are right respectively
Target is observed and inter-process is carried out to the coordinates of targets measured value each measured, and the cluster node will be respective internal
Obtained data transfer is handled to the leader cluster node；
S2, leader cluster node carries out fusion treatment：The data that the leader cluster node obtains its own inter-process and each institute
The data progress fusion treatment that cluster node inter-process is obtained is stated, to obtain current goal state estimation, and lower a period of time is predicted
The target-like state value at quarter；
S3, alternate node calculates itself expected gain：The leader cluster node will predict the target-like of obtained subsequent time
State value passes to each cluster node and guard node, and the leader cluster node, cluster node and guard node are respectively according to the prediction
The target-like state value of subsequent time calculate the expectation information gain of itself subsequent time, the then cluster node and guard node
Respective expectation information gain and the status information data of itself are passed into the leader cluster node respectively；
S4, selects the cluster node of subsequent time：The leader cluster node makees its own and each cluster node and guard node
For the alternate node of subsequent time, and expectation information gain according to each alternate node subsequent time and respective state letter of itself
Cease data and calculate the expectation contribution decision content of each alternate node subsequent time, and select to expect larger multiple standby of contribution decision content
Node is selected as the cluster node of subsequent time；
S5, selects the leader cluster node of subsequent time：Lower a period of time that presently described leader cluster node is selected from the step S4
In the cluster node at quarter, suitable cluster knot is selected according to the dump energy of the distance between each cluster node and target and each cluster node
Put the leader cluster node as subsequent time.
Further, in the step S1, selection detects the node of target as the leader cluster node, selection at first
It is currently able to detect other nodes in the node of target in addition to the leader cluster node as the task cluster node.
Further, in the step S1, SRCIF algorithms are respectively adopted to respectively testing oneself in the leader cluster node and cluster node
The coordinates of targets measured value obtained carries out inter-process, obtains corresponding information vectorWith square root information matrixIts
In, i represents i-th of node, and Y represents information matrix, k | k represents the estimation to the k moment according to the measured value at current k moment.
Further, in step s 2, the leader cluster node carries out fusion treatment using SRCIF algorithms, obtains final mesh
The information vector of mark stateWith square root information matrix SY, k | k：
Wherein, NcThe node number in current task cluster is represented, Tria represents that ORTHOGONAL TRIANGULAR DECOMPOSITION is operated, and then calculates
The current goal state estimation xk|kWith the covariance matrix P of estimationk|k：
xk|k=Pk|kYk|k (5)。
Further, in the step S2, the leader cluster node is according to the information vector of obtained final goal state
With square root information matrix, the target-like state value of subsequent time is predicted using SRCIF algorithms.
Further, in the step S3, SRCIF calculations are respectively adopted in the leader cluster node, cluster node and guard node
Method predicts the anticipation error covariance matrix Y of itself subsequent timek+1|k, and then calculate the expectation information for obtaining respective subsequent time
Gain Gi,k+1：Gi,k+1=tr (Yi,k+1), wherein, tr represents to take the mark of matrix.
Further, in the step S4, the calculation formula of the expectation contribution decision content of each alternate node subsequent time
It is as follows：
Di,k+1=αi,k+1Gi,k+1-βi,k+1Ci,k+1(6),
In formula (6), αi,k+1For the weight of the i-th point of expectation information gain at the k+1 moment, for representing that target is detected
The probability likelihood value of survey, βi,k+1For the weight of the i-th point of energy expenditure at the k+1 moment, Ci,k+1For i-th point in k+1
Carve the energy expenditure as cluster node.
Further, in the step S5, the corresponding alternate node of optimal solution that will be obtained by formula (7)Under
The leader cluster node at one moment：
Wherein, ψ (ci) it is alternate node ciTo the distance and alternate node current residual of the subsequent time target location of prediction
The weight joint of energy, defines ψ (ci) be：
Herein, ψe、ψdNode c is represented respectivelyiCorresponding normalized relative energy size and normalized section
Point-target range, wherein, eiRepresent alternate node ciDump energy,Represent alternate node ciTo the next of prediction
Moment target location distance, ei min/ei maxMinimum/most corresponding energy of node of dump energy in alternate node is represented respectively
Value,Represent that the subsequent time target location of range prediction in alternate node is farthest/nearest respectively
The distance between subsequent time target location of node and prediction.
As a result of technical scheme proposed above, compared with prior art, the present invention has the following advantages that and accumulated
Pole effect：
The present invention considers an actual imaging sensor sensor model, and prediction target is located at into imaging sensor
Different observation areas, be converted to different probability.When the node activated needed for selecting subsequent time, these probability are also examined
Worry is come in, so that the probability of subsequent time target can be observed by improving selected node, it is ensured that the stabilization of tracking system
Property.In addition, going to select the optimal of the condition of satisfaction using the method for global optimum relative to the dynamic quick pick mechanism of traditional cluster node
Set of node, this technology proposes a greedy on-line decision mechanism.The mechanism is selected according to the definition for proposing contribution decision content
Suitable node, this definition has considered expectation observation gain that sensor node results in it as institute after cluster node
The Resources Consumption of consumption.Furthermore, it is understood that contribution decision content assigns observation gain the weights different with resource consumption respectively, especially
It is weight of the resource consumption of node shared by contribution decision content with the dump energy of present node and tracks in candidate cluster
The change of the relative size of the dump energy of other both candidate nodes and change.Relative energy is bigger, and resource consumption is in contribution decision-making
Shared weight is just smaller in amount, on the contrary then bigger.So go to consider using energy as a dynamic amount, can be prevented effectively from surplus
The few node of complementary energy is selected as cluster head, and the node more than those dump energies can not be selected as cluster node.So as to balance net
Network dump energy, improves the life cycle of tracking network system.Furthermore, relative to some existing cluster head quick pick mechanisms, such as
Select near the target location of prediction node or select the node for possessing most dump energies as next task cluster
The method of cluster head, this technology has done a compromise between node to future position and the dump energy of node, equally takes
Obtained effect well.Finally, this technology proposes the above-mentioned different mechanisms of tracking scheme desintegration.The tracking scheme is not
Existing tracking scheme, any information of the leader cluster node without knowing other nodes cluster Nei in advance are same as, but cluster node is obtained
Internally calculated after observation and obtain oneself result, and the status information by the result together with oneself passes to cluster head, for cluster
Head node makes a policy.So the follow-up mechanism has very big advantage, Er Qiegeng on computation complexity and space complexity
Meet the actual conditions that node is often difficult to obtain other nodal informations in advance.
Brief description of the drawings
Fig. 1 is a kind of principle of method for tracking target towards the wireless sensor network based on imaging sensor of the invention
Schematic diagram；
Fig. 2 is the tracking error comparison diagram that fusion treatment is carried out using different methods；
Fig. 3 is the tracking numerical stability comparison diagram that fusion treatment is carried out using different methods；
Fig. 4 A are with using other task cluster node pickout apparatus using the task cluster node quick pick mechanism proposed in the present invention
Tracing positional mean error comparison diagram of the method for system；
Fig. 4 B are with using other task cluster node pickout apparatus using the task cluster node quick pick mechanism proposed in the present invention
The average tracking energy consumption comparison figure of the method for system；
Fig. 5 A are the residual energy using the cluster head quick pick mechanism proposed in this method and other common cluster head selection mechanisms
Measure the comparison diagram of distribution standard deviation；
Fig. 5 B are using the cluster head quick pick mechanism proposed in this method and the tracking of other common cluster head selection mechanisms
Target loses the comparison diagram of number of times.
Embodiment
With reference to instantiation, the present invention is expanded on further.It should be understood that this example be merely to illustrate the present invention and without
In limitation the scope of the present invention.In addition, it is to be understood that after the content of the invention lectured has been read, those skilled in the art can be with
The present invention is made various changes or modifications, these equivalent form of values equally fall within the model that the application appended claims are limited
Enclose.
A kind of method for tracking target towards the wireless sensor network based on imaging sensor of the present invention, such as Fig. 1 institutes
Show, comprise the following steps：
S1, intiating radio sensor network：
In all nodes for tracking current goal, selection one detects the node of target as cluster head section at first
Point, selects to be currently able to other nodes in the node for detecting target in addition to leader cluster node as cluster node, remaining node is made
For guard node.Wherein, leader cluster node and cluster node composition task cluster cooperation tracking target, and the target to each measuring respectively
Coordinates measurements carry out inter-process (realizing the renewal part for being preferred to use SRCIF algorithms) to obtain corresponding information vectorWith square root information matrix(i represents i-th of node, and Y represents information vector matrix, k | k was represented according to the current k moment
Estimation of the measured value to the k moment), then the data transfer after respective inter-process is given current cluster cephalomere by each cluster node respectively
Point.Now, not selected guard node enters dormancy or alarm state to save energy expenditure.
S2, leader cluster node carries out fusion treatment：
Leader cluster node is received after the packet that each cluster node is sent, using SRCIF algorithms renewal part by its own
The data that the data that inter-process is obtained are obtained with each cluster node inter-process carry out fusion treatment, obtain final goal state
Information vectorWith square root information matrixSpecific formula for calculation is as follows：
Wherein, NcThe node number in current task cluster being represented, Tria represents that ORTHOGONAL TRIANGULAR DECOMPOSITION is operated, k | k-1 is represented
The prediction to the current K moment carried out at the K-1 moment.And then calculate current goal state estimation xk|k(include the position of target
Put and velocity information) with estimation covariance matrix Pk|k：
xk|k=Pk|kYk|k (5)。
Then, leader cluster node is according to the information vector of obtained dbjective stateWith square root information matrix SY,k|k, perform
The predicted portions of SRCIF algorithms, predict the target-like state value and square root information matrix of subsequent time.
S3, alternate node calculates itself expected gain：Leader cluster node will predict the target-like state value of obtained subsequent time
Each cluster node and each guard node are passed to, then leader cluster node, cluster node and guard node perform the update section of SRCIF algorithms
Point, the anticipation error covariance matrix Y of itself subsequent time is predicted according to the target-like state value of the subsequent time of prediction respectivelyk+1|k
(k+1 | k represents subsequent time --- the prediction at K+1 moment carried out at the current K moment), and then calculating obtains each lower a period of time
The expectation information gain G at quarteri,k+1=tr (Yi,k+1) (wherein, tr (Yi,k+1) represent to take matrix Yk+1|kMark), its be used for weigh meter
Calculate the accuracy of result, it is more accurate that information gain means to predict the outcome more greatly, and by the expectation information gain and each itself
Status information data (including current remaining, target detection probability, own coordinate position all pass to cluster head section) transmission
Give current cluster head node.
S4, selects the cluster node of subsequent time：Using each cluster node and guard node as subsequent time alternate node, when
Preceding leader cluster node is received after the data of cluster node and guard node, according to the expectation information gain and the shape of itself of each alternate node
State information data calculates the expectation contribution decision content of each alternate node subsequent time, and selects to expect the portion that contribution decision content is larger
Divide alternate node as the cluster node of subsequent time, wherein, expect to contribute the calculation formula of decision content as follows：
Di,k+1=αi,k+1Gi,k+1-βi,k+1Ci,k+1(6),
In formula (6), αi,k+1For the weight of the i-th point of expectation information gain at the k+1 moment, for representing that target is detected
The probability likelihood value of survey, βi,k+1For the weight of the i-th point of energy expenditure at the k+1 moment, Ci,k+1For i-th point in k+1
Carve the energy expenditure as cluster node.
S5, selects the leader cluster node of subsequent time：The cluster knot for the subsequent time that current cluster head node is selected from step S4
In point, suitable cluster node is selected to be lower a period of time according to the dump energy of the distance between each cluster node and target and each cluster node
The leader cluster node at quarter, is used as the scheduler of subsequent time task.
In the present embodiment, optimal problem will be converted into the problem of selecting leader cluster node：
Wherein, ψ (ci) it is distance and alternate node current residual of the alternate node to the subsequent time target location of prediction
The weight joint of energy, defines ψ (ci) be：
Herein, Node c is represented respectivelyiNormalized relative energy size and normalized node destination away from
From, wherein, eiRepresent alternate node ciDump energy,Represent alternate node ciTo the subsequent time target of prediction
The distance of position, ei min/ei maxMinimum/most corresponding energy value of node of dump energy in alternate node is represented respectively,Farthest/nearest node in the subsequent time target location of range prediction in alternate node is represented respectively
The distance between with the subsequent time target location of prediction.
In the aforementioned embodiment, step S1 uses target following structure of the distributing based on cluster.Utilized in step S2
Square root volume information filtering (SRCIF) algorithm of numerical stability carries out fusion treatment, is carried while can reducing amount of calculation
High tracking accuracy and stability.The target data of prediction make use of to go to calculate in step S3, alternate node calculates oneself under
One moment issuable information gain and energy expenditure, can so disperse the computational load of leader cluster node.Proposed in step S4
Contribution decision content estimates dynamic and selects subsequent time task cluster node online.This estimates the expectation letter for having considered node
Gain and energy expenditure are ceased, them are assigned different weights respectively.Energy expenditure has been assigned dynamic weight, works as node energy
When fewer relative to other nodes, energy expenditure weight shared in contribution decision content will be bigger, so as to avoid the node
It is chosen as cluster node.The cluster head selection algorithm that is proposed in step S5 has considered node and the distance and node of prediction target
Dump energy selects a suitable cluster head, balance network energy expenditure and avoids cluster head from not observing target.
It should be noted that in the above-described embodiments, leader cluster node uses square root volume information filtering algorithm and entered
Row fusion treatment, in addition, can also carry out fusion treatment using Extended information filter algorithm or volume information filtering algorithm.
Wherein, Fig. 2 show using square root volume information filtering algorithm carry out fusion treatment with using Extended information filter algorithm and
Volume information filtering algorithm carries out the tracking error comparison diagram of fusion treatment, and Fig. 3 is shown using the filtering of square root volume information
Algorithm carries out the tracking numerical value that fusion treatment carries out fusion treatment with being calculated using Extended information filter algorithm and volume information filtering
Stability comparison diagram.From figures 2 and 3, it will be seen that carrying out the tracking of fusion treatment using square root volume information filtering algorithm
Error is smaller, and tracking numerical stability is more preferably.One is built below in the region that the length of side is 500 meters, is passed by 8000 images
Sensor node lays the simulation model for carrying out target following at random, has additionally been carried out of the invention with existing preferable algorithm pair
Than to further illustrate the advantage of the present invention.The energy value of each node is initially being uniformly distributed between [0,1].In order to
Simplify emulation, it is 100 to set whole number of times per secondary tracking, the time interval between two secondary trackings is 1s.In this emulation,
Using the task cluster node quick pick mechanism (Algorithm 1) proposed in the present invention with using other task cluster node quick pick mechanisms
Method tracing positional mean error with average tracking energy consumption comparison as illustrated in figures 4 a and 4b.Wherein, the side that M1 is represented
Method be also according to cluster node effectiveness how much go selection activation node, but this method consider node absolute energy without
It is the weighted value that node is consumed in the relative energy that both candidate nodes are concentrated as node energy, this method increases all information in addition
Benefit is all elected to be task node rather than a selected section node of the invention as task cluster node more than the node of energy consumption；M2 is represented
Method be the larger node of dump energy for directly selecting fixed number as cluster node；The method that M3 is represented is directly ought
Preceding all nodes that can observe target are all elected to be task cluster node.Cluster node quick pick mechanism of the present invention
The cluster knot points selected in (Algorithm 1) and algorithm M2 are 9.
Referring to Fig. 4 A and 4B, compared with other three kinds of cluster node systems of selection, although this programme tracking error (1.7257m)
It is higher by 23% or so than algorithm M1 (1.3839m) and M3 (1.3984m), but this tracking accuracy is entirely capable of in general
Enough meet demand.And in terms of the energy expenditure of length systems life cycle is directly connected in wireless sensor network, this hair
The bright algorithm energy (5.436J) that tracing task is consumed each time is than algorithm M1 (11.716J) and algorithm M3 (11.517J)
Respectively will low 53.6% and 52.8%.Also it is better than in tracking accuracy compared with the algorithm M2 for being fixed as 9 even if with nodes
Algorithm M2 (1.8543m).
For leader cluster node quick pick mechanism (Algorithm 2) proposed by the invention, illustrate in terms of following two：
One is, leader cluster node can consume most energy as the scheduler of tracing task during tracking, so leader cluster node is needed
Possess more energy, and the more balanced whole tracking network of Energy distribution of each node in network is understood according to pertinent literature
Life cycle also will be longer.So whether the Energy distribution that the selection of leader cluster node have impact on each node in cluster is balanced, and then
Also the life cycle of tracking network can be influenceed；Two are, inaccurate due to subsequent time target prodiction, so the cluster of selection
Head has a probability that can't detect subsequent time target, and such case can often occur in actual application scenarios, have impact on
Effective operation of track system, occurs so to reduce this state as far as possible.In summary, present invention determine that current cluster network is remained
The standard deviation of complementary energy distribution (shows whether the dump energy distribution of each cluster node is balanced, by being used as cluster head in a certain node
The standard deviation for calculating the dump energy of whole cluster interior nodes afterwards is realized) it can not detect target with cluster head in each tracing task
As weighing, cluster head selection is fine or not to be estimated probability.
Fig. 5 A and 5B are using the cluster head quick pick mechanism (Algorithm 2) proposed in the present invention and other common cluster heads
The dump energy distribution standard deviation of selection mechanism and the target of tracking lose the contrast of number of times.Wherein C1 represents to select near pre-
The cluster node for surveying subsequent time target location is selected as cluster head, and C2 represents to select dump energy in the cluster node selected most
Node as subsequent time tracking cluster cluster head.As shown in Figure 5 A and 5B, compared with algorithm C2, cluster head proposed by the invention is chosen
Select algorithm (Algorithm2) almost identical with its in terms of residue of network organization energy scale difference, but the present invention averagely every time with
Nearly 60% is reduced on the situation frequency that the on target of track can not be observed by current cluster head：3.1440 secondary contrast
7.8060 it is secondary.Compared with algorithm C1, algorithm (Algorithm 2) proposed by the present invention is although the cluster head selected can't detect currently
The number of times of target is somewhat more (2.2740 contrast 3.1440), but the cluster head selection algorithm proposed by the present invention in energy balance
Perform better than, further increase the life cycle of tracking network.
It is seen that, the various different mechanisms that method for tracking target proposed by the present invention is included, relative to existing one
A little methods achieve preferable effect in comprehensive improve in tracking accuracy and reduction network energy consumption performance indications, are particularly suitable for use in
The target following of the large-scale dense wireless sense network based on imaging sensor is using upper.
Claims (8)
1. a kind of method for tracking target towards the wireless sensor network based on imaging sensor, it is characterised in that including with
Lower step：
S1, intiating radio sensor network：In all nodes for tracking current goal, one leader cluster node of selection and
Multiple cluster nodes constitute task clusters, remaining node as guard node, wherein, the leader cluster node and cluster node are respectively to target
It is observed and inter-process is carried out to the coordinates of targets measured value that each measures, and the cluster node is by respective inter-process
Obtained data transfer gives the leader cluster node；
S2, leader cluster node carries out fusion treatment：The data that the leader cluster node obtains its own inter-process and each cluster
The data that intra-node processing is obtained carry out fusion treatment, to obtain current goal state estimation, and predict subsequent time
Target-like state value；
S3, alternate node calculates itself expected gain：The leader cluster node will predict the target-like state value of obtained subsequent time
Each cluster node and guard node are passed to, and the leader cluster node, cluster node and guard node are respectively according under the prediction
The target-like state value at one moment calculates the expectation information gain of itself subsequent time, then the cluster node and guard node difference
Respective expectation information gain and the status information data of itself are passed into the leader cluster node；
S4, selects the cluster node of subsequent time：The leader cluster node is using its own and each cluster node and guard node under
The alternate node at one moment, and expectation information gain and the respective status information number of itself according to each alternate node subsequent time
Decision content is contributed according to the expectation for calculating each alternate node subsequent time, and selects to expect multiple alternative sections that contribution decision content is larger
Put the cluster node as subsequent time；
S5, selects the leader cluster node of subsequent time：The subsequent time that presently described leader cluster node is selected from the step S4
In cluster node, suitable cluster node is selected to make according to the dump energy of the distance between each cluster node and target and each cluster node
For the leader cluster node of subsequent time.
2. the method for tracking target according to claim 1 towards the wireless sensor network based on imaging sensor, its
It is characterised by, in the step S1, selection detects the node of target as the leader cluster node at first, and selection is currently able to
Other nodes in the node of target in addition to the leader cluster node are detected as the task cluster node.
3. the method for tracking target according to claim 1 towards the wireless sensor network based on imaging sensor, its
It is characterised by, in the step S1, SRCIF algorithms are respectively adopted to the target that each measures in the leader cluster node and cluster node
Coordinates measurements carry out inter-process, obtain corresponding information vectorWith square root information matrixWherein, i represents
I node, Y represents information matrix, k | k represents the estimation to the k moment according to the measured value at current k moment.
4. the method for tracking target according to claim 3 towards the wireless sensor network based on imaging sensor, its
It is characterised by, in step s 2, the leader cluster node carries out fusion treatment using SRCIF algorithms, obtains final goal state
Information vectorWith square root information matrix SY,k|k：
<mrow>
<msub>
<mover>
<mi>y</mi>
<mo>^</mo>
</mover>
<mrow>
<mi>k</mi>
<mo>|</mo>
<mi>k</mi>
</mrow>
</msub>
<mo>=</mo>
<msub>
<mover>
<mi>y</mi>
<mo>^</mo>
</mover>
<mrow>
<mi>k</mi>
<mo>|</mo>
<mi>k</mi>
<mo>-</mo>
<mn>1</mn>
</mrow>
</msub>
<mo>+</mo>
<munderover>
<mo>&Sigma;</mo>
<mrow>
<mi>i</mi>
<mo>=</mo>
<mn>1</mn>
</mrow>
<msub>
<mi>N</mi>
<mi>c</mi>
</msub>
</munderover>
<msubsup>
<mi>y</mi>
<mrow>
<mi>k</mi>
<mo>|</mo>
<mi>k</mi>
</mrow>
<mrow>
<mo>(</mo>
<mi>i</mi>
<mo>)</mo>
</mrow>
</msubsup>
<mo>-</mo>
<mo>-</mo>
<mo>-</mo>
<mrow>
<mo>(</mo>
<mn>1</mn>
<mo>)</mo>
</mrow>
<mo>,</mo>
</mrow>
<mrow>
<msub>
<mi>S</mi>
<mrow>
<mi>Y</mi>
<mo>,</mo>
<mi>k</mi>
<mo>|</mo>
<mi>k</mi>
</mrow>
</msub>
<mo>=</mo>
<mi>T</mi>
<mi>r</mi>
<mi>i</mi>
<mi>a</mi>
<mrow>
<mo>(</mo>
<mfenced open = "[" close = "]">
<mtable>
<mtr>
<mtd>
<msub>
<mover>
<mi>S</mi>
<mo>^</mo>
</mover>
<mrow>
<mi>y</mi>
<mo>,</mo>
<mi>k</mi>
<mo>|</mo>
<mi>k</mi>
<mo>-</mo>
<mn>1</mn>
</mrow>
</msub>
</mtd>
<mtd>
<msubsup>
<mi>S</mi>
<mrow>
<mi>Y</mi>
<mo>,</mo>
<mi>k</mi>
<mo>|</mo>
<mi>k</mi>
</mrow>
<mrow>
<mo>(</mo>
<mn>1</mn>
<mo>)</mo>
</mrow>
</msubsup>
</mtd>
<mtd>
<mn>...</mn>
</mtd>
<mtd>
<msubsup>
<mi>S</mi>
<mrow>
<mi>Y</mi>
<mo>,</mo>
<mi>k</mi>
<mo>|</mo>
<mi>k</mi>
</mrow>
<mrow>
<mo>(</mo>
<msub>
<mi>N</mi>
<mi>c</mi>
</msub>
<mo>)</mo>
</mrow>
</msubsup>
</mtd>
</mtr>
</mtable>
</mfenced>
<mo>)</mo>
</mrow>
<mo>-</mo>
<mo>-</mo>
<mo>-</mo>
<mrow>
<mo>(</mo>
<mn>2</mn>
<mo>)</mo>
</mrow>
<mo>,</mo>
</mrow>
Wherein, NcThe node number in current task cluster is represented, Tria represents that ORTHOGONAL TRIANGULAR DECOMPOSITION is operated, and then calculates described
Current goal state estimation xk|kWith the covariance matrix P of estimationk|k：
<mrow>
<msub>
<mi>Y</mi>
<mrow>
<mi>k</mi>
<mo>|</mo>
<mi>k</mi>
</mrow>
</msub>
<mo>=</mo>
<msub>
<mi>S</mi>
<mrow>
<mi>Y</mi>
<mo>,</mo>
<mi>k</mi>
<mo>|</mo>
<mi>k</mi>
</mrow>
</msub>
<msubsup>
<mi>S</mi>
<mrow>
<mi>Y</mi>
<mo>,</mo>
<mi>k</mi>
<mo>|</mo>
<mi>k</mi>
</mrow>
<mi>T</mi>
</msubsup>
<mo>-</mo>
<mo>-</mo>
<mo>-</mo>
<mrow>
<mo>(</mo>
<mn>3</mn>
<mo>)</mo>
</mrow>
<mo>,</mo>
</mrow>
<mrow>
<msub>
<mi>P</mi>
<mrow>
<mi>k</mi>
<mo>|</mo>
<mi>k</mi>
</mrow>
</msub>
<mo>=</mo>
<msubsup>
<mi>Y</mi>
<mrow>
<mi>k</mi>
<mo>|</mo>
<mi>k</mi>
</mrow>
<mrow>
<mo>-</mo>
<mn>1</mn>
</mrow>
</msubsup>
<mo>=</mo>
<msup>
<mrow>
<mo>(</mo>
<msub>
<mi>S</mi>
<mrow>
<mi>Y</mi>
<mo>,</mo>
<mi>k</mi>
<mo>|</mo>
<mi>k</mi>
</mrow>
</msub>
<msubsup>
<mi>S</mi>
<mrow>
<mi>Y</mi>
<mo>,</mo>
<mi>k</mi>
<mo>|</mo>
<mi>k</mi>
</mrow>
<mi>T</mi>
</msubsup>
<mo>)</mo>
</mrow>
<mrow>
<mo>-</mo>
<mn>1</mn>
</mrow>
</msup>
<mo>-</mo>
<mo>-</mo>
<mo>-</mo>
<mrow>
<mo>(</mo>
<mn>4</mn>
<mo>)</mo>
</mrow>
<mo>,</mo>
</mrow>
1
xk|k=Pk|kYk|k (5)。
5. the method for tracking target according to claim 4 towards the wireless sensor network based on imaging sensor, its
It is characterised by, in the step S2, the leader cluster node is according to the information vector and square root of obtained final goal state
Information matrix, the target-like state value of subsequent time is predicted using SRCIF algorithms.
6. the method for tracking target according to claim 5 towards the wireless sensor network based on imaging sensor, its
It is characterised by, in the step S3, the prediction of SRCIF algorithms is respectively adopted certainly in the leader cluster node, cluster node and guard node
The anticipation error covariance matrix Y of body subsequent timek+1|k, and then calculate the expectation information gain for obtaining respective subsequent time
Gi,k+1：Gi,k+1=tr (Yi,k+1), wherein, tr represents to take the mark of matrix.
7. the method for tracking target according to claim 6 towards the wireless sensor network based on imaging sensor, its
It is characterised by, in the step S4, the calculation formula of the expectation contribution decision content of each alternate node subsequent time is as follows：
Di,k+1=αi,k+1Gi,k+1-βi,k+1Ci,k+1(6),
In formula (6), αi,k+1For the weight of the i-th point of expectation information gain at the k+1 moment, it is detected for representing target
Probability likelihood value, βi,k+1For the weight of the i-th point of energy expenditure at the k+1 moment, Ci,k+1Make for i-th point at the k+1 moment
For the energy expenditure of cluster node.
8. the method for tracking target according to claim 7 towards the wireless sensor network based on imaging sensor, its
It is characterised by, in the step S5, the corresponding alternate node of optimal solution that will be obtained by formula (7)It is used as subsequent time
Leader cluster node：
<mrow>
<mfenced open = "{" close = "">
<mtable>
<mtr>
<mtd>
<mrow>
<msubsup>
<mi>c</mi>
<mrow>
<mi>k</mi>
<mo>+</mo>
<mn>1</mn>
</mrow>
<mi>h</mi>
</msubsup>
<mo>=</mo>
<munder>
<mrow>
<mi>m</mi>
<mi>a</mi>
<mi>x</mi>
</mrow>
<mrow>
<mo>&ForAll;</mo>
<msub>
<mi>c</mi>
<mi>i</mi>
</msub>
</mrow>
</munder>
<mi>&psi;</mi>
<mrow>
<mo>(</mo>
<msub>
<mi>c</mi>
<mi>i</mi>
</msub>
<mo>)</mo>
</mrow>
</mrow>
</mtd>
</mtr>
<mtr>
<mtd>
<mtable>
<mtr>
<mtd>
<mrow>
<mi>s</mi>
<mo>.</mo>
<mi>t</mi>
<mo>.</mo>
</mrow>
</mtd>
<mtd>
<mrow>
<msub>
<mi>e</mi>
<mi>i</mi>
</msub>
<mo>&GreaterEqual;</mo>
<msub>
<mi>E</mi>
<mrow>
<mi>a</mi>
<mi>s</mi>
</mrow>
</msub>
</mrow>
</mtd>
</mtr>
</mtable>
</mtd>
</mtr>
<mtr>
<mtd>
<mrow>
<msub>
<mover>
<mi>T</mi>
<mo>^</mo>
</mover>
<mrow>
<mi>k</mi>
<mo>+</mo>
<mn>1</mn>
</mrow>
</msub>
<mo>&Element;</mo>
<msubsup>
<mi>Z</mi>
<mi>i</mi>
<mn>2</mn>
</msubsup>
</mrow>
</mtd>
</mtr>
<mtr>
<mtd>
<mrow>
<msub>
<mi>c</mi>
<mi>i</mi>
</msub>
<mo>&Element;</mo>
<msub>
<mi>l</mi>
<mrow>
<mi>k</mi>
<mo>+</mo>
<mn>1</mn>
</mrow>
</msub>
<mo>,</mo>
</mrow>
</mtd>
</mtr>
</mtable>
</mfenced>
<mo>-</mo>
<mo>-</mo>
<mo>-</mo>
<mrow>
<mo>(</mo>
<mn>7</mn>
<mo>)</mo>
</mrow>
<mo>,</mo>
</mrow>
Wherein, ψ (ci) it is alternate node ciTo the distance and alternate node current remaining of the subsequent time target location of prediction
Weight joint, define ψ (ci) be：
<mrow>
<mi>&psi;</mi>
<mrow>
<mo>(</mo>
<msub>
<mi>c</mi>
<mi>i</mi>
</msub>
<mo>)</mo>
</mrow>
<mo>=</mo>
<msub>
<mi>&theta;&psi;</mi>
<mi>e</mi>
</msub>
<mrow>
<mo>(</mo>
<msub>
<mi>e</mi>
<mi>i</mi>
</msub>
<mo>)</mo>
</mrow>
<mo>+</mo>
<mrow>
<mo>(</mo>
<mn>1</mn>
<mo>-</mo>
<mi>&theta;</mi>
<mo>)</mo>
</mrow>
<msub>
<mi>&psi;</mi>
<mi>d</mi>
</msub>
<mrow>
<mo>(</mo>
<mi>d</mi>
<mo>(</mo>
<mrow>
<msub>
<mi>c</mi>
<mi>i</mi>
</msub>
<mo>,</mo>
<msub>
<mover>
<mi>T</mi>
<mo>^</mo>
</mover>
<mrow>
<mi>k</mi>
<mo>+</mo>
<mn>1</mn>
</mrow>
</msub>
</mrow>
<mo>)</mo>
<mo>)</mo>
</mrow>
<mo>-</mo>
<mo>-</mo>
<mo>-</mo>
<mrow>
<mo>(</mo>
<mn>8</mn>
<mo>)</mo>
</mrow>
<mo>,</mo>
</mrow>
Herein,
ψe、ψdNode c is represented respectivelyiCorresponding normalized relative energy size and normalized node-target range, wherein, eiTable
Show alternate node ciDump energy,Represent alternate node ciTo the distance of the subsequent time target location of prediction,
ei min/ei maxMinimum/most corresponding energy value of node of dump energy in alternate node is represented respectively,Farthest/nearest node in the subsequent time target location of range prediction in alternate node is represented respectively
The distance between with the subsequent time target location of prediction.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201710339773.6A CN107205255A (en) | 2017-05-15 | 2017-05-15 | Towards the method for tracking target of the wireless sensor network based on imaging sensor |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201710339773.6A CN107205255A (en) | 2017-05-15 | 2017-05-15 | Towards the method for tracking target of the wireless sensor network based on imaging sensor |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN107205255A true CN107205255A (en) | 2017-09-26 |
Family
ID=59905245
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201710339773.6A Pending CN107205255A (en) | 2017-05-15 | 2017-05-15 | Towards the method for tracking target of the wireless sensor network based on imaging sensor |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN107205255A (en) |
Cited By (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111680713A (en) * | 2020-04-26 | 2020-09-18 | 中国科学院上海微系统与信息技术研究所 | Unmanned aerial vehicle ground target tracking and approaching method based on visual detection |
| CN112929991A (en) * | 2021-02-10 | 2021-06-08 | 上海工程技术大学 | Sensor management method, device, equipment and storage medium |
| CN115278819A (en) * | 2022-06-13 | 2022-11-01 | 中国科学院上海微系统与信息技术研究所 | Node scheduling method based on wireless sensor network |
| CN115278819B (en) * | 2022-06-13 | 2024-06-07 | 中国科学院上海微系统与信息技术研究所 | Node scheduling method based on wireless sensor network |
Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101458325A (en) * | 2009-01-08 | 2009-06-17 | 华南理工大学 | Wireless sensor network tracking method based on self-adapting prediction |
| CN102123473A (en) * | 2011-01-06 | 2011-07-13 | 山东大学 | Dynamic clustering mechanism-based target tracking method for wireless sensor network |
| KR20140075270A (en) * | 2012-12-11 | 2014-06-19 | 국방과학연구소 | Target Following Clustering Algorism |
| CN104469875A (en) * | 2014-11-26 | 2015-03-25 | 北京邮电大学 | Prediction-based target tracking method and system in wireless sensor network |
| CN105611626A (en) * | 2015-12-28 | 2016-05-25 | 南京信息工程大学 | Target tracking algorithm based on regional clustering of perceptron in wireless sensor network |
-
2017
- 2017-05-15 CN CN201710339773.6A patent/CN107205255A/en active Pending
Patent Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101458325A (en) * | 2009-01-08 | 2009-06-17 | 华南理工大学 | Wireless sensor network tracking method based on self-adapting prediction |
| CN102123473A (en) * | 2011-01-06 | 2011-07-13 | 山东大学 | Dynamic clustering mechanism-based target tracking method for wireless sensor network |
| KR20140075270A (en) * | 2012-12-11 | 2014-06-19 | 국방과학연구소 | Target Following Clustering Algorism |
| CN104469875A (en) * | 2014-11-26 | 2015-03-25 | 北京邮电大学 | Prediction-based target tracking method and system in wireless sensor network |
| CN105611626A (en) * | 2015-12-28 | 2016-05-25 | 南京信息工程大学 | Target tracking algorithm based on regional clustering of perceptron in wireless sensor network |
Non-Patent Citations (2)
| Title |
|---|
| 周伟: "基于无线传感器网络的定位跟踪技术研究", 《中国优秀博士论文数据库》 * |
| 陈彦明，赵清杰，刘岩宇: "一种适用于分布式摄像机网络的SCIWCF算法", 《电子学报》 * |
Cited By (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111680713A (en) * | 2020-04-26 | 2020-09-18 | 中国科学院上海微系统与信息技术研究所 | Unmanned aerial vehicle ground target tracking and approaching method based on visual detection |
| CN111680713B (en) * | 2020-04-26 | 2023-11-03 | 中国科学院上海微系统与信息技术研究所 | Unmanned aerial vehicle ground target tracking and approaching method based on visual detection |
| CN112929991A (en) * | 2021-02-10 | 2021-06-08 | 上海工程技术大学 | Sensor management method, device, equipment and storage medium |
| CN115278819A (en) * | 2022-06-13 | 2022-11-01 | 中国科学院上海微系统与信息技术研究所 | Node scheduling method based on wireless sensor network |
| CN115278819B (en) * | 2022-06-13 | 2024-06-07 | 中国科学院上海微系统与信息技术研究所 | Node scheduling method based on wireless sensor network |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN110489223A (en) | Method for scheduling task, device and electronic equipment in a kind of isomeric group | |
| Sarker et al. | Optimal selection of crowdsourcing workers balancing their utilities and platform profit | |
| Zhang et al. | Achieving democracy in edge intelligence: A fog-based collaborative learning scheme | |
| Gao et al. | A UAV-assisted multi-task allocation method for mobile crowd sensing | |
| Dai et al. | Delay-sensitive energy-efficient UAV crowdsensing by deep reinforcement learning | |
| Gao et al. | Budgeted unknown worker recruitment for heterogeneous crowdsensing using cmab | |
| US11468284B2 (en) | Space utilization measurement and modeling using artificial intelligence | |
| CN107205255A (en) | Towards the method for tracking target of the wireless sensor network based on imaging sensor | |
| CN109376919A (en) | A kind of prediction technique of coal mine fully-mechanized mining working gas emission | |
| Lu et al. | Deep reinforcement learning-based multi-objective edge server placement in Internet of Vehicles | |
| Liu et al. | Incentive design for air pollution monitoring based on compressive crowdsensing | |
| Zhu et al. | Incentive mechanism for cooperative localization in wireless networks | |
| Weikert et al. | Multi-objective task allocation for wireless sensor networks | |
| Luo et al. | A node depth adjustment method with computation-efficiency based on performance bound for range-only target tracking in UWSNs | |
| Ding et al. | Dynamic delayed-decision task assignment under spatial-temporal constraints in mobile crowdsensing | |
| Song et al. | Generalized Model and Deep Reinforcement Learning-Based Evolutionary Method for Multitype Satellite Observation Scheduling | |
| Jia et al. | A high accuracy localization algorithm with DV-Hop and fruit fly optimization in anisotropic wireless networks | |
| CN111082878B (en) | Target tracking method based on underwater mobile wireless sensor network | |
| Moradbeikie et al. | A cost-effective LoRaWAN-based IoT localization method using fixed reference nodes and dual-slope path-loss modeling | |
| Yan et al. | Service caching for meteorological emergency decision-making in cloud-edge computing | |
| CN109151760A (en) | It is measured based on square root volume and weights consistent distributions filtering method | |
| Zhu et al. | Path planning for adaptive CSI map construction with A3C in dynamic environments | |
| CN109362049A (en) | Consistent square root volume filtering method is weighted based on mixed information | |
| CN109493065A (en) | A kind of fraudulent trading detection method of Behavior-based control incremental update | |
| CN109688598B (en) | WSAN-based complex pipe network distributed data acquisition system and transmission optimization method |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| WD01 | Invention patent application deemed withdrawn after publication |
Application publication date: 20170926 |
|
| WD01 | Invention patent application deemed withdrawn after publication |