CN105682171B - Spatio-temporal clustering method for compressive data gathering - Google Patents
Spatio-temporal clustering method for compressive data gathering Download PDFInfo
- Publication number
- CN105682171B CN105682171B CN201610132885.XA CN201610132885A CN105682171B CN 105682171 B CN105682171 B CN 105682171B CN 201610132885 A CN201610132885 A CN 201610132885A CN 105682171 B CN105682171 B CN 105682171B
- Authority
- CN
- China
- Prior art keywords
- node
- cluster
- data
- nodes
- neighborhood
- 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.)
- Expired - Fee Related
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
- H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
- H04L45/00—Routing or path finding of packets in data switching networks
- H04L45/14—Routing performance; Theoretical aspects
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
- H04L45/00—Routing or path finding of packets in data switching networks
- H04L45/46—Cluster building
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04W—WIRELESS COMMUNICATION NETWORKS
- H04W40/00—Communication routing or communication path finding
- H04W40/24—Connectivity information management, e.g. connectivity discovery or connectivity update
- H04W40/32—Connectivity information management, e.g. connectivity discovery or connectivity update for defining a routing cluster membership
-
- 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
- H04W84/20—Master-slave selection or change arrangements
-
- 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
Abstract
The invention discloses a spatio-temporal clustering method for compressive data gathering. The spatio-temporal clustering method comprises the steps of: 1, an initialization stage; 2, spatio-temporal correlation measurement; 3, network clustering; 4, compressive data gathering; 5, data reconstruction. In the network clustering step, aggregation nodes obtain clustering information according to a calculated value of spatio-temporal correlation and broadcast the information to the whole network, and intra-cluster nodes select cluster head nodes dynamically. By adopting the steps, the spatio-temporal clustering method for compressive data gathering reduces data transmission times under the condition of same reconfiguration error, balances network energy consumption, prevents a certain node from dying too early due to energy consumption, and prolongs network lifetime.
Description
Technical field:The present invention provides a kind of wireless sensor network compressed data based on temporal correlation sub-clustering and collects
Method, it belongs to wireless communication technology field.
Background technology
Wireless sensor network (Wireless Sensor Networks, WSNs) is by a large amount of low-power consumption, low cost
Sensor node deployment sets up into network in the way of wireless Ad Hoc to relevant range.These nodes can be cooperated jointly, week
Phase property ground is perceived and gathered data, and is delivered to aggregation node in a multi-hop fashion.How Data Collection task is efficiently completed,
I.e. under conditions of same data collection precision is ensured, maximize and extend life-span of network and be one and important study a question.
Cluster Networks have topology clear, the features such as energy efficient, are widely used in compressed data and collect in agreement.Conventional wireless is sensed
The problem of device network cluster dividing agreement, such as LEACH, HEED scheduling algorithm, the main selection for considering leader cluster node and sub-clustering number, and compared with
The data characteristicses of monitor area are considered less.
Compressed sensing is an emerging digital signal processing method, and the method is considered as to classical Shannon-
One kind of Nyqist sampling thheorems surmounts.Nyquist sampling theorem thinks that, in order to recover original signal exactly, sample frequency is extremely
If 2 times of signal highest frequency less.And compressive sensing theory is thought, by obtaining a small amount of of sparse signal or compressible signal
Linear projection just can reconstruct the signal exactly.
Due to there is substantial amounts of redundancy in initial data, and the number of the less consideration sensing region of traditional cluster algorithm
According to temporal correlation, in same cluster, data dependence differs greatly, it is impossible to carry out rarefaction representation well, reconstructs in identical
In the case of error, more observation frequencies are needed.The increase of data transmission times, wastes the network bandwidth, reduces network
Life-span.
The content of the invention
For deficiency of the prior art, it is an object of the invention to provide a kind of wireless biography based on temporal correlation sub-clustering
Sensor Web compression method of data capture (Spatiotemporal Clustering Method for Compressive
Data Gathering, STC-CDG), a kind of method of tolerance part area data dependency is specifically proposed, and is applied to net
Network sub-clustering, has considered the data dependence of the dump energy and cluster interior nodes of leader cluster node.
The problem to be solved in the present invention is illustrated visually here by an example.Assume that sensor network is responsible for monitoring
The intensity of illumination in one region, the different intensity of illumination of the region representation of dotted line in Fig. 1.As shown in Fig. 1 .a, traditional net
Network Clustering protocol only considers that the dump energy of communication cost and leader cluster node in cluster defines 7 clusters.Then for No. 3 clusters, No. 5 clusters
For, the dependency very little of the data of cluster interior nodes, data be not it is sparse, or can not effectively rarefaction representation, thus
Data collection can not be carried out using the method for compressed sensing.From unlike Fig. 1 .a, when Fig. 1 .b are according to part area data
Empty dependency carries out sub-clustering, data height correlation in cluster, can be very good to carry out rarefaction representation.From compressive sensing theory,
This cluster-dividing method can in significantly less cluster data pendulous frequency, reduce energy expenditure, extend network life.
A kind of wireless sensor network compressed data collection method based on temporal correlation sub-clustering of the present invention, it comprising with
Lower step:
Step one, initial phase:
In this stage, each sensor acquisition data, and the initial data for collecting is sent in a multi-hop fashion
Aggregation node;Aggregation node preserves sampled data of each sensor node in each time slot;When aggregation node receives data
Afterwards, carry out the temporal correlation tolerance stage of data;During specific use, as aggregation node preserves each node
Historical data, when data dependency change less than setting threshold value when, sub-clustering information need not change, and directly be compressed
Perception data is collected;
Step 2, temporal correlation tolerance:
Aggregation node according to the data collected in step one, according to area data relativity measurement side proposed by the present invention
Method, calculates the data temporal correlation between any two sensor node.
Make x(k)∈RN,xi∈RTRespectively the data vector and i-th sensor of network kth time sampling is in T time slot
Vector of samples, N for nodes number.It is the ultimate range in neighborhood between two nodes that we define Gmax_dst,
Or maximum hop count, when the distance in net between any two node is less than Gmax_dst, a neighborhood N can be classified asc.We are only
Consider in neighborhood NcCorrelation calculations of the interior joint to data, had so both reduced the computation complexity of aggregation node, had avoided again
There is distant company side, be more conform with practical situation.
It is specific as follows,
Step 2.1:Calculate the temporal correlation of two nodes in any neighborhood, if the arbitrary node in neighborhood to (i,
J) existence time dependency, then need to meet
ρm(i,j)>ρm_min,
Wherein ρm_minFor self-defining temporal correlation threshold value,
In formula, NcNeighborhood defined in expression is above-mentioned,WithRepresent i-th node respectively in kth time and kth+m
Secondary sampled value, xiRepresent vector of samples of i-th sensor in T time slot, E [xi] andI-th sensing is represented respectively
Sampling average and sample variance of the device in T time slot, are understood in the same mannerE[xj] andExpression implication.
Step 2.2:Calculate the spatial coherence of two nodes in any neighborhood, if the arbitrary node in neighborhood to (i,
J) Existential Space dependency, then need to meet
ρs(i,j)>ρs_min,
Wherein ρs_minFor self-defining spatial coherence threshold value,
In formula, NcNeighborhood defined in expression is above-mentioned,Represent sampled value of i-th node in kth time, xiRepresent i-th
Vector of samples of the individual sensor in T time slot, E [xi] andRepresent that i-th sensor is equal in the sampling of T time slot respectively
Value and standard for manual sampling are poor, understand in the same mannerE[xj] andExpression implication.
Step 2.3:Calculate the temporal correlation of two nodes in any neighborhood, if the arbitrary node in neighborhood to (i,
J) there is temporal correlation, then need to meet
ρms(i,j)>ρmsmin,
Wherein ρmsminFor self-defining spatial coherence threshold value,
ρms(i, j)=ρm(i,j)*ρs(i,j)
In formula, ρm(i, j) and ρsThe implication of (i, j) is given in step 2.1 and step 2.2 respectively.
Step 3, network cluster:Aggregation node obtains sub-clustering information backward according to the temporal correlation value of calculation between node
The whole network broadcast message, cluster interior nodes dynamically select leader cluster node again.
It is specific as follows,
Step 3.1:According to the temporal correlation data that step 2 is obtained, given threshold thresholding ρmsminIf, neighborhood interior nodes
Data dependence is more than threshold value thresholding, is considered as the presence of company side (u, v) between the two nodes, constructs graph model G, figure
Summit is exactly sensor node in network, figure while be defined as it is above-mentioned even while (u, v);
Step 3.2:Sub-clustering and leader cluster node are determined according to graph model G interior joints angle value and residue energy of node;
Step 3.3:Aggregation node confirms the position that cluster internal segment is counted out with leader cluster node, and is determined by compressive sensing theory
In cluster, compressed sensing number of samples is no less than Cklog (n/k), and wherein C is decimation factor, and k is the degree of rarefication of signal in cluster, and n is
Cluster internal segment is counted out.
Step 4, compressed data are collected:
Step 4.1:Cluster interior nodes and leader cluster node use identical pseudorandom number generator, generate measurement coefficient Φi, institute
Need not be transmitted in a network with measurement coefficient.Cluster interior nodes using compressed sensing data collection method, by initial data with
The linear operation result of measurement coefficient passes to leader cluster node;
Step 4.2:Calculation matrix ΦiFor the submatrix of i-th cluster of correspondence, CHiThe leader cluster node of i-th cluster is represented, is responsible for
Collect all data x in clusteri, then in single cluster, data collection is expressed as yi,yi=Φixi;
Step 4.3:Leader cluster node CHiUsing minimum range square spanning tree algorithm, by measured value yiIt is transferred to convergence section
Point.
Step 5, data reconstruction:Aggregation node receives the data from cluster head, constitutes measured value vectorThen converge
Poly- node produces identical random matrix Φ, and primary collection data are reconstructed, it is assumed that is divided into 5 clusters in network, then weighs
Structure formula meets:
In formula, measurement coefficient ΦiFor the submatrix of i-th cluster of correspondence, xiRepresent the vector of samples in i-th cluster, yiRepresent
The compressed sensing measured value of i-th cluster of correspondence.
By above step, based on the wireless sensor network compressed data collection method of temporal correlation sub-clustering, in phase
In the case of with reconstructed error, data transmission times are reduced, balanced network energy consumption extends network lifetime.
Compare with the radio sensor network data collection method of compressive sensing theory with existing combination network cluster dividing, this
The advantage of invention is:
1st, temporal correlation measure proposed by the present invention, in can effectively calculating network, data dependence is larger
Neighbor node, and the dump energy of the temporal correlation and node for considering data carries out sub-clustering, the cluster interior nodes for obtaining
Data height correlation, can effectively carry out rarefaction representation, and then in the case of identical reconstructed error, reduce data transfer time
Number.
2nd, due to having considered the dump energy of node during cluster, the balanced network energy consumption of this method, it is to avoid
Certain node extends network lifetime because of energy consumption problem premature death.
Description of the drawings
Fig. 1 a are the cluster-dividing method of the present invention and traditional cluster-dividing method effect diagram.
Fig. 1 b are the cluster-dividing method of the present invention and traditional cluster-dividing method effect diagram.
Fig. 2 is the Hierarchical network topological model schematic diagram of the present invention.
Fig. 3 is the methods described flow chart of the present invention.
Fig. 4 is the cluster-dividing method and random cluster-dividing method data collection observation frequency of the present invention and Between Signal To Noise Ratio contrast
Figure.
Fig. 5 is the method for data capture of the present invention and shortest path method of data capture energy consumption comparison figure.
Specific embodiment
See Fig. 1 a- Fig. 5, a kind of wireless sensor network compressed data based on temporal correlation sub-clustering proposed by the present invention
Collection method, with reference to specific implementation method, the present invention is described in detail.
The network model of the lower present invention is introduced first.When network size is larger, Node distribution interior in a big way, we
Network can be layered, set up meromixis center.As shown in Fig. 2 if network is divided into each sub-network, each subnet
Network has respective local aggregation node, and final each local aggregation node is directly transmitted to data the sink centers of network again.I
Consider the data collection situation of single localized network here, and single localized network is made the following assumptions:
(1) node is all that isomorphism, i.e. node have identical communication capacity and primary power;
(2) net interior nodes periodically collect same physical message, and the data of diverse location different time there may be
Larger difference;
(3) in positive direction region of the length for L, N number of sensor node is evenly distributed on equally spaced grid, is converged
Outside monitor area, all nodes keep fixed position motionless after placing to node;
(4) cluster interior nodes and leader cluster node direction communication, leader cluster node adopt minimum range square tree algorithm (SSDST) and
Cluster head communicates;
(5) in net, each sensor node and Sink node have identical pseudorandom number generator, produce random matrix
Coefficient, so projection matrix need not be transmitted in a network.
Aggregation node has the energy supply of abundance compared with the ordinary node in network, and computing capability is stronger.Commonly
Node receives the scheduling of aggregation node, performs and perceives and data transfer task.Aggregation node is responsible for whole network data and collects, performs net
The reconstruct of network Clustering protocol and perception data.
A kind of wireless sensor network compressed data collection method based on temporal correlation sub-clustering of the present invention, such as Fig. 3 institutes
Show, specifically comprise the steps of:
Step one, initial phase:
In this stage, each sensor acquisition data, and the initial data for collecting is sent in a multi-hop fashion
Aggregation node.Aggregation node possesses larger memory space, saves hits of each sensor node in each time slot
According to.When aggregation node receives enough data, the temporal correlation tolerance stage of data is carried out.Process is being used specifically
In, as aggregation node preserves the historical data of each node, when the dependency change of data is less, sub-clustering information is not required to
Change, can directly be compressed perception data collection as needed.
Step 2, temporal correlation tolerance:
Aggregation node according to the data collected in step one, according to area data relativity measurement side proposed by the present invention
Method, calculates the data temporal correlation between any two sensor node.
Make x(k)∈RN,xi∈RTRespectively the data vector and i-th sensor of network node kth time sampling is at T
Between groove vector of samples, N for nodes number.We define Gmax_dst for the maximum between two nodes in neighborhood away from
From, or maximum hop count, when the distance in net between any two node is less than Gmax_dst, a neighborhood N can be classified asc.I
Only consider in neighborhood NcSimilarity Measure of the interior joint to data, had so both reduced the computation complexity of aggregation node, and
Avoid the presence of distant company side, be more conform with practical situation.
It is specific as follows,
Step 2.1:Calculate the temporal correlation of two nodes in any neighborhood, if the arbitrary node in neighborhood to (i,
J) existence time dependency, then need to meet
ρm(i,j)>ρm_min,
Wherein ρm_minFor self-defining temporal correlation threshold value,
In formula, NcNeighborhood defined in expression is above-mentioned,WithRepresent i-th node respectively in kth time and kth+m
Secondary sampled value, xiRepresent vector of samples of i-th sensor in T time slot, E [xi] andI-th sensing is represented respectively
Sampling average and sample variance of the device in T time slot, are understood in the same mannerE[xj] andExpression implication.
Step 2.2:Calculate the spatial coherence of two nodes in any neighborhood, if the arbitrary node in neighborhood to (i,
J) Existential Space dependency, then need to meet
ρs(i,j)>ρs_min,
Wherein, ρs_minFor self-defining spatial coherence threshold value,
In formula, NcNeighborhood defined in expression is above-mentioned,Represent sampled value of i-th node in kth time, xiRepresent i-th
Vector of samples of the individual sensor in T time slot, E [xi] andRepresent that i-th sensor is equal in the sampling of T time slot respectively
Value and standard for manual sampling are poor, understand in the same mannerE[xj] andExpression implication.
Step 2.3:Calculate the temporal correlation of two nodes in any neighborhood, if the arbitrary node in neighborhood to (i,
J) there is temporal correlation, then need to meet
ρms(i,j)>ρmsmin,
Wherein, ρmsminFor self-defining spatial coherence threshold value,
ρms(i, j)=ρm(i,j)*ρs(i,j)
In formula, ρm(i, j) and ρsThe implication of (i, j) is given in step 2.1 and step 2.2 respectively.
Step 3, network cluster:Aggregation node obtains sub-clustering information backward according to the temporal correlation value of calculation between node
The whole network broadcast message, cluster interior nodes dynamically select leader cluster node again.
It is specific as follows,
Step 3.1:According to the temporal correlation data that step 2 is obtained, given threshold thresholding ρmsminIf, neighborhood interior nodes
Data dependence is more than threshold value thresholding, is considered as the presence of company side (u, v) between the two nodes, constructs graph model G, figure
Summit is exactly sensor node in network, figure while be defined as it is above-mentioned even while (u, v);
Step 3.2:Sub-clustering and leader cluster node are determined according to graph model G interior joints angle value and residue energy of node;
Step 3.3:Aggregation node confirms the position that cluster internal segment is counted out with leader cluster node, and is determined by compressive sensing theory
In cluster, compressed sensing number of samples is no less than Cklog (n/k), and wherein C is decimation factor, and k is the degree of rarefication of signal in cluster, and n is
Cluster internal segment is counted out.The network Clustering Algorithm of this step describes false code, referring to table 1.
1 STC Clustering Algorithms of table
Step 4, compressed data are collected:
Step 4.1:Cluster interior nodes and leader cluster node use identical pseudorandom number generator, generate measurement coefficient Φi, institute
Need not be transmitted in a network with measurement coefficient.Cluster interior nodes using compressed sensing data collection method, by initial data with
The linear operation result of measurement coefficient passes to leader cluster node;
Step 4.2:Measurement coefficient ΦiFor the submatrix of i-th cluster of correspondence, CHiThe leader cluster node of i-th cluster is represented, is responsible for
Collect all data x in clusteri, then in single cluster, data collection is expressed as yi,yi=Φixi;
Step 4.3:Leader cluster node CHiUsing minimum range square spanning tree algorithm, by measured value yiIt is transferred to convergence section
Point.
Step 5, data reconstruction:Aggregation node receives the data from cluster head, constitutes measured value vectorThen converge
Poly- node produces identical random matrix Φ=[Φ1 Φ2 Φ3 Φ4], and primary collection data are reconstructed, reconstruct formula
Meet (in assuming network, be divided into 5 clusters):
In formula, measurement coefficient ΦiFor the submatrix of i-th cluster of correspondence, xiRepresent the vector of samples in i-th cluster, yiRepresent
The compressed sensing measured value of i-th cluster of correspondence.
For checking the effectiveness of the method, the present invention carries out emulation experiment using Matlab emulation platforms, by 20 nodes
In the region of 100m × 100m, aggregation node is located at outside monitor area random placement.The energy that node is consumed is according to the following formula
Calculate:
Wherein ETxThe energy that 1 bit data is consumed, E are sent by transtation mission circuitRx1 bit data is received for receiving circuit
The energy for being consumed, EAmpFor sending the energy consumption of amplifying circuit.Energy parameter used in emulation is shown in Table 2:
2 STC-CDG method data collection simulation parameters of table
For convenience of contrasting, we carry out sub-clustering using following two methods:First, the random sub-clustering of network;Second, using this
The sub-clustering based on temporal correlation that invention is proposed.Test every time, it is ensured that observation frequency equally, each test repeats 50 times,
In two kinds of cluster-dividing methods of contrast, the relation of observation frequency and signal to noise ratio.Signal to noise ratio (the Signal Noise of reconstruction signal
Ratio, SNR) it is defined as follows:
In formula, | | x | |2With | | e | |2The l of primary signal and noise signal is represented respectively2Norm.
From Fig. 4, it is apparent that under same observation frequency, being recovered using the signal of the cluster-dividing method in the present invention
Precision is significantly larger than random cluster-dividing method.
In network node normal working hourss, the analysis of experiments method of the present invention and shortest path data collection is based on
The power consumption of method, as a result as shown in Figure 5.As can be seen from Figure 5, under same data reconstruction precision, dividing based on temporal correlation
Cluster data collection method, hence it is evident that reduce network energy consumption.
Claims (1)
1. a kind of wireless sensor network compressed data collection method based on temporal correlation sub-clustering, it is characterised in that:It wraps
Containing following steps:
Step one, initial phase:
In this stage, each sensor acquisition data, and the initial data for collecting is sent to convergence in a multi-hop fashion
Node;Aggregation node preserves sampled data of each sensor node in each time slot;
Step 2, temporal correlation tolerance:
After aggregation node receives data, the temporal correlation tolerance stage of data is carried out;Aggregation node is received according in step one
The data for collecting, according to the number of regions that the wireless sensor network compressed data collection method based on temporal correlation sub-clustering is proposed
According to relativity measurement method, the data temporal correlation between any two sensor node is calculated;
Make x(k),xiRespectively vector of samples of the data vector and i-th sensor of network kth time sampling in T time slot, N
For the number of nodes;It is the ultimate range in neighborhood between two nodes, i.e. maximum hop count to define Gmax_dst, is appointed in net
When distance between two nodes of meaning is less than Gmax_dst, a neighborhood N is classified asc;Only consider in neighborhood NcPhase of the interior joint to data
Closing property is calculated, to tally with the actual situation;
The area data that the wireless sensor network compressed data collection method based on temporal correlation sub-clustering is proposed is related
Property measure, it is specific as follows,
Step 2.1:The temporal correlation of two nodes in any neighborhood is calculated, if the arbitrary node in neighborhood is deposited to (i, j)
In temporal correlation,
ρm(i,j)>ρm_min,
ρm(i, j) represents the temporal correlation of two nodes in neighborhood;
Wherein ρm_minFor self-defining temporal correlation threshold value,
In formula, NcNeighborhood defined in expression is above-mentioned,WithRepresent i-th node adopting in kth time and kth+m time respectively
Sample value, xiRepresent vector of samples of i-th sensor in T time slot, E [xi] andRepresent i-th sensor at T respectively
The sampling average of time slot and sample variance, are understood in the same mannerE[xj] andExpression implication;
Step 2.2:The spatial coherence of two nodes in any neighborhood is calculated, if the arbitrary node in neighborhood is deposited to (i, j)
In spatial coherence, then need to meet
ρs(i,j)>ρs_min,
ρs(i, j) represents the spatial coherence of two nodes in neighborhood;
Wherein ρs_minFor self-defining spatial coherence threshold value,
In formula, NcNeighborhood defined in expression is above-mentioned,Represent sampled value of i-th node in kth time, xiRepresent i-th biography
Vector of samples of the sensor in T time slot, E [xi] andRepresent respectively i-th sensor T time slot sampling average and
Standard for manual sampling is poor, understands in the same mannerE[xj] andExpression implication;
Step 2.3:The temporal correlation of two nodes in any neighborhood is calculated, if the arbitrary node in neighborhood is deposited to (i, j)
In temporal correlation, then need to meet
ρms(i,j)>ρmsmin,
ρms(i, j) represents the temporal correlation of two nodes in neighborhood;
Wherein ρmsminFor self-defining spatial coherence threshold value,
ρms(i, j)=ρm(i,j)*ρs(i,j)
In formula, ρm(i, j) and ρsThe implication of (i, j) is given in step 2.1 and step 2.2 respectively;
Step 3, network cluster:Aggregation node obtains sub-clustering according to the data temporal correlation between any two sensor node
To the whole network broadcast message after information, cluster interior nodes dynamically select leader cluster node again;
It is specific as follows,
Step 3.1:According to self-defining spatial coherence threshold value ρ of settingmsminWith the temporal correlation number obtained in step 2
According to, if neighborhood interior nodes data dependence is more than threshold value thresholding, it is considered as between the two nodes, the presence of company side (u, v),
Construction graph model G, the summit of figure is exactly sensor node in network, figure when above-mentioned company is defined as (u, v);
Step 3.2:Sub-clustering and leader cluster node are determined according to graph model G interior joints angle value and residue energy of node;According to step 3.1
Frontier juncture system of company in the graph model G for obtaining, if all nodes are all connected in region, is divided in a cluster, cluster with probability Pch=
Eresidual/EmaxSelect leader cluster node, wherein EresidualRepresent the dump energy of node, EmaxRepresent the primary power of node, Pch
Represent that node is elected as the probability of cluster head;
Step 3.3:Aggregation node confirms the position that cluster internal segment is counted out with leader cluster node, and is determined in cluster by compressive sensing theory
Compressed sensing number of samples is no less than C kln (n/k), and wherein C is decimation factor, and k is the degree of rarefication of signal in cluster, and n is in cluster
Interstitial content;C kln (n/k) show the index with natural number e as bottom;
Step 4, compressed data are collected:
As aggregation node preserves the historical data of each node, when the dependency change of data is less than the threshold value for setting,
Sub-clustering information need not change, and directly be compressed perception data collection;The compressed data collection method is specific as follows:
Step 4.1:Cluster interior nodes and leader cluster node use identical pseudorandom number generator, generate measurement coefficient Φi, so surveying
Coefficient of discharge need not be transmitted in a network;Method of the cluster interior nodes using compressed sensing data collection, by initial data and measurement
The linear operation result of coefficient passes to leader cluster node;Cluster interior nodes carry out data transfer according to step 4.2;Cluster interior nodes are adopted
Compressed sensing method of data capture specifically as described in step 4.2;
Step 4.2:Calculation matrix ΦiFor the submatrix of i-th cluster of correspondence, CHiThe leader cluster node of i-th cluster is represented, is responsible for collection
All data x in clusteri, then in single cluster, data collection is expressed as yi,yi=Φixi;
Step 4.3:Leader cluster node CHiUsing minimum range square spanning tree algorithm, by measured value yiIt is transferred to aggregation node;
Step 5, data reconstruction:
Aggregation node receives the data from cluster head, constitutes measured value vectorThen aggregation node produces the random square of identical
Battle array Φ, and primary collection data are reconstructed, it is assumed that it is divided into 5 clusters in network, then reconstructs formula and meet:
In formula, measurement coefficient ΦiFor the submatrix of i-th cluster of correspondence, xiRepresent the vector of samples in i-th cluster, yiRepresent correspondence
The compressed sensing measured value of i-th cluster.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201610132885.XA CN105682171B (en) | 2016-03-09 | 2016-03-09 | Spatio-temporal clustering method for compressive data gathering |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201610132885.XA CN105682171B (en) | 2016-03-09 | 2016-03-09 | Spatio-temporal clustering method for compressive data gathering |
Publications (2)
Publication Number | Publication Date |
---|---|
CN105682171A CN105682171A (en) | 2016-06-15 |
CN105682171B true CN105682171B (en) | 2017-03-22 |
Family
ID=56308094
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201610132885.XA Expired - Fee Related CN105682171B (en) | 2016-03-09 | 2016-03-09 | Spatio-temporal clustering method for compressive data gathering |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN105682171B (en) |
Families Citing this family (9)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN106211256B (en) * | 2016-07-08 | 2019-10-01 | 西北大学 | A kind of Unmanned Aerial Vehicle Data collection method based on data critical node |
CN107544017B (en) * | 2017-07-12 | 2020-06-16 | 清华大学 | Low-power-consumption weighted pseudo-random test method based on vector compression and related equipment |
CN107766422A (en) * | 2017-09-12 | 2018-03-06 | 中国科学院信息工程研究所 | A kind of mapping method and equipment of data of registering |
CN108494758B (en) * | 2018-03-14 | 2020-10-27 | 南京邮电大学 | Perceptual big data hierarchical perceptual compression coding method |
CN108682140B (en) * | 2018-04-23 | 2020-07-28 | 湘潭大学 | Enhanced anomaly detection method based on compressed sensing and autoregressive model |
CN109587651B (en) * | 2018-12-26 | 2021-11-02 | 中国电建集团河南省电力勘测设计院有限公司 | Wireless sensor network data aggregation method |
US11275366B2 (en) | 2019-05-07 | 2022-03-15 | Cisco Technology, Inc. | Adaptively calibrated spatio-temporal compressive sensing for sensor networks |
CN112702710A (en) * | 2020-12-22 | 2021-04-23 | 杭州电子科技大学 | Opportunistic routing optimization method based on link correlation in low duty ratio network |
CN116489709B (en) * | 2023-06-20 | 2023-11-17 | 中电科新型智慧城市研究院有限公司 | Node scheduling policy determination method, terminal equipment and storage medium |
Family Cites Families (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
DE602007005389D1 (en) * | 2007-12-14 | 2010-04-29 | Ntt Docomo Inc | Method and device for operating a wireless network for collecting data |
CN101909330B (en) * | 2010-07-27 | 2013-01-30 | 杭州电子科技大学 | Sensor network data compression method based on near-optimal clustering and local virtual coordinates |
CN105025498B (en) * | 2015-06-08 | 2018-08-21 | 南京邮电大学 | Compression method when sensing network clustering sky based on network code and compressed sensing |
CN105120469B (en) * | 2015-07-06 | 2016-09-07 | 湘潭大学 | The low information density method of data capture of quality scalable based on compressed sensing |
-
2016
- 2016-03-09 CN CN201610132885.XA patent/CN105682171B/en not_active Expired - Fee Related
Also Published As
Publication number | Publication date |
---|---|
CN105682171A (en) | 2016-06-15 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN105682171B (en) | Spatio-temporal clustering method for compressive data gathering | |
CN107786959B (en) | Compressed data collection method in wireless sensor network based on adaptive measuring | |
CN102438291B (en) | Data aggregation method for increasing capacity of wireless sensor network | |
CN104618947A (en) | Compressive sensing based dynamic clustering wireless sensor network data collecting method and device | |
CN102594904A (en) | Method for detecting abnormal events of wireless sensor network in distributed way | |
Ling et al. | Decentralized support detection of multiple measurement vectors with joint sparsity | |
CN107708133A (en) | Underwater wireless sensor network multi-dimensional optimization node deployment method based on mesh generation | |
CN107332631A (en) | A kind of method of use multi-attribute group decision making theoretical evaluation link-quality assessment models | |
CN103684634B (en) | Based on the compression frequency spectrum sensing method of locating information in heterogeneous wireless sensor net | |
CN101835233B (en) | Wireless sensor network oriented random route selecting method | |
CN103888957B (en) | The node method of diffusion of the loss of signal prediction based on corn growth situation | |
CN102724681A (en) | Sensor network coverage hole detection method combining with energy efficiency | |
CN103702282B (en) | A kind of multiple types multiple target passive type localization method based on migration compressed sensing | |
Akbar et al. | Performance optimization of the IEEE 802.15. 4-based link quality protocols for WBASNS/IOTS in a hospital environment using fuzzy logic | |
CN103957544A (en) | Method for improving survivability of wireless sensor network | |
CN101808383B (en) | Method for selecting matrix wireless sensor network-oriented random routing | |
CN101895957B (en) | Random routing method facing matrix type wireless sensor network distribution weighing | |
Ammar et al. | A new approach based on wireless sensor network and fuzzy logic for forest fire detection | |
CN110012488A (en) | A kind of compressed data collection method of mobile wireless sensor network | |
CN109474904A (en) | A kind of wireless sensor network compressed data collection method considering energy consumption and covering | |
CN105357676A (en) | Matrix filling-based indoor idle frequency spectrum detection method | |
Wang et al. | Topology analysis of wireless sensor networks for sandstorm monitoring | |
CN109348445A (en) | Soil environment high-precision real-time monitoring system for precision agriculture | |
CN103868552A (en) | Wireless sensor network node with high practicality | |
Camilo et al. | Biomedical signal monitoring using wireless sensor networks |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
C06 | Publication | ||
PB01 | Publication | ||
C10 | Entry into substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
C14 | Grant of patent or utility model | ||
GR01 | Patent grant | ||
CF01 | Termination of patent right due to non-payment of annual fee |
Granted publication date: 20170322 Termination date: 20180309 |
|
CF01 | Termination of patent right due to non-payment of annual fee |