CN105263149B - Mobile wireless sensor network interior joint adapter distribution weight clustering method - Google Patents
Mobile wireless sensor network interior joint adapter distribution weight clustering method Download PDFInfo
- Publication number
- CN105263149B CN105263149B CN201510638551.5A CN201510638551A CN105263149B CN 105263149 B CN105263149 B CN 105263149B CN 201510638551 A CN201510638551 A CN 201510638551A CN 105263149 B CN105263149 B CN 105263149B
- Authority
- CN
- China
- Prior art keywords
- node
- cluster
- moment
- class
- particle
- 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.)
- Active
Links
Classifications
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04W—WIRELESS COMMUNICATION NETWORKS
- H04W16/00—Network planning, e.g. coverage or traffic planning tools; Network deployment, e.g. resource partitioning or cells structures
- H04W16/24—Cell structures
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04W—WIRELESS COMMUNICATION NETWORKS
- H04W24/00—Supervisory, monitoring or testing arrangements
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04W—WIRELESS COMMUNICATION NETWORKS
- H04W40/00—Communication routing or communication path finding
-
- 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
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 mobile wireless sensor network interior joint adapter distribution weight clustering methods, including in the mobile wireless sensor network for having clustered cluster, node is accurately estimated the position in moving process each moment itself according to the data of inertial sensor and using dead reckoning and particle filter algorithm;When each reunion class period starts, node moves the position of followed motion model and current time according to it, reasonable prediction is carried out to subsequent time self-position, non- leader cluster node is allowed to carry out reunion class according to the location information of nodes other in network at each cluster inner boundary, it thereby may be ensured that node can be always in than better suited cluster in moving process, communication distance i.e. between the leader cluster node of its affiliated cluster is maintained in reasonable range, so that being able to maintain higher data when communication therebetween is sent to rate, therefore it can guarantee the service quality of mobile wireless sensor network.
Description
Technical field
The present invention relates to node weight clustering methods, and in particular to mobile wireless sensor network interior joint adapter distribution
Weight clustering method.
Background technique
With the fast development of mobile wireless sensor network technology, in numerous applications such as environmental monitoring, target following
It is widely used in field, while about sides such as the life cycles and service quality for improving mobile wireless sensor network
The research in face is also receive more and more attention, such as using the method for cluster in mobile wireless sensor network.By nothing
Line sensor node (hereinafter referred to as node) selects suitable leader cluster node, cluster according to the information clusters cluster such as its position
Remaining interior node is non-leader cluster node.In each cluster, collected information etc. is sent leader cluster node, cluster head by non-leader cluster node
The information that node is responsible for receive is integrated, and converging information point or base station are sent to, and is responsible in cluster to non-cluster head
Node distributes information.Collected information is sent directly to converging information point or base station compared to each node, using poly-
The method of class can greatly reduce the energy consumption of non-leader cluster node, and can reduce the transmission to redundancy, and cluster head section
Energy consumption difference between point and non-leader cluster node can be eliminated in such a way that leader cluster node rotates.Since node energy is limited simultaneously
And be inconvenient to supplement, thus the reduction of node energy consumption can extend the life cycle of entire mobile wireless sensor network, simultaneously
Less redundancy is also beneficial to the raising of mobile wireless sensor network service quality.Therefore, it is passed based on mobile wireless
In the application of sensor network, the method for cluster has great development prospect.
However mobile wireless sensor network interior joint has certain mobility, this will be to clustered mobile wireless
The topological structure of sensor network causes dynamic, random influence, will lead to and is overlapped between cluster, so that node may
It is not in and is most suitable in its cluster, i.e., non-leader cluster node needs are communicated with apart from its farther away leader cluster node, this will
Data are sent to the reduction of rate when bringing the increase and communication of communication consumption energy.Therefore, the movement of node how is effectively eliminated
Property influence to Clustering Effect be that must face and solve the problems, such as in current mobile wireless sensor network application.
Through the literature search of existing technologies, J.Baek, S.K.An and P.Fisher are in 2010 years in IEEE
" Dynamic has been delivered in transactions on consumer electronics (IEEE consumer electronics periodical)
cluster header selection and conditional re-clustering for wireless sensor
Networks " (selection of dynamic cluster head and condition weight clustering method in wireless sensor network), transduces for the wheel of leader cluster node
The non-leader cluster node caused communicates with distance and increases this problem, meets again to non-leader cluster node with proposing a kind of conditionity
The method (hereinafter referred to as SISR) of class, the conditionity refer to needs carry out reunion class non-leader cluster node it is only brand new in target
The cluster head of one wheel, which rotates, can just carry out reunion class when starting.SISR avoids its cluster head section with affiliated cluster to a certain extent
Communication distance between point is too long.But in mobile wireless sensor network, weight clustering problem caused by the mobility of node
Both it is more complicated, keep reasonable communication distance with leader cluster node when SISR is unable to ensure node motion, therefore be unable to ensure
Between data when communicating be sent to rate, and then influence the service quality of mobile wireless sensor network.
Summary of the invention
In view of the deficiencies of the prior art, the present invention proposes mobile wireless sensor network interior joint adapter distribution weights
Clustering method.
A kind of mobile wireless sensor network interior joint adapter distribution weight clustering method provided according to the present invention, packet
Include following steps:
Step A:In the mobile wireless sensor network for having clustered cluster, node in moving process, node according to
The sensing data of inertial sensor estimates the self-position at each moment, obtains the estimated location at each moment;
Step B:When each reunion class period starts, node is according to itself followed motion model of movement and itself works as
The position at preceding moment predicts the position of subsequent time itself, obtains predicted position;
Step C:Estimated location obtained in step A is sent to the leader cluster node of affiliated cluster, cluster head section by non-leader cluster node
The estimated location that point integrates cluster interior nodes is communicated with the leader cluster node of adjacent clusters later, and will be obtained from the leader cluster node of adjacent clusters
Estimated location be sent to cluster inner boundary node, boundary node assesses the class that itself whether needs to meet again;
Step D:Predicted position obtained in step B is sent to the leader cluster node of affiliated cluster, non-cluster head by non-leader cluster node
Need to meet again in node class boundary node by reunion class request be sent to belonging to cluster leader cluster node, leader cluster node obtain cluster in
It is communicated with the leader cluster node of adjacent clusters after the predicted position and reunion class solicited message of node, and will be from the cluster head section of adjacent clusters
The predicted position of predicted position, the request of reunion class and itself affiliated cluster interior nodes that point obtains, the request of reunion class are sent to and need
The boundary node of the boundary node for class of meeting again, the class that needs to meet again completes reunion class.
Preferably, each node operation particle filter algorithm estimates the position at moment itself each in moving process
Meter executes the position estimation procedure of mobile node by step A, specific as follows:
The step A includes the following steps:
Step A1:Initial time, corresponding k=0, the position of node is it is known that N is randomly generated near node location0It is a
Particle;Indicate i-th of particle of the corresponding initial time of k=0, i ∈ { 1 ..., N0,Weighted value be
Step A2:Obtain the update of -1 moment of kth particle position or node estimated location into k-th of etching process
Equation, specially:
When k-th of moment, wherein k >=1 obtains particle position or node estimated location more according to dead reckoning
Newly equation is
Wherein, skParticle position or node estimated location when indicating k-th of moment, sk-1When indicating -1 moment of kth
Particle position or node estimated location, t indicate adjacent moment between time interval length, wherein between assuming each time
Constant, the v every interior nodes velocity amplitude and the direction of motionk-1Indicate the node speed value in -1 time interval of kth, dk-1Indicate kth-
Movement direction of nodes in 1 time interval, Δ dkIndicate -1 time interval of kth to k-th of time interval joint movements side
To change value;- 1 time interval of the kth refers to the time interval between -1 moment of kth and k-th of moment;
Step A3:At -1 moment of kth to the particle state renewal process, including particle at k-th of moment itself position, grain
The update of the weighted value of son and particle develop NevolveA new particle replaces itself;I-th of new grain at k-th of moment
SonNon-normalized weighted valueForNew particleNormalized weight valueForWherein, NkIndicate the total number of particles including all new particles, when N () is by -1 moment of kth
ParticleDevelop new particleProbability, N () obeyFor mean value,
For the Gaussian Profile of variance,Indicate particleNormalized weight value, NevolveFor preset positive integer, particleIndicate i-th of particle when -1 moment of kth, particleIndicate i-th of new particle when k-th of moment;μkWhen indicating k-th
The mean value of the Gaussian Profile, δ when quarterkIndicate the variance of Gaussian Profile when k-th of moment,When indicating k-th of moment
J-th of new particle;
Step A4:Calculate effective particle threshold at k-th of moment
Step A5:Calculate the number of effective particles mesh at k-th of momentWherein,IfThen in order to avoid weighted value concentrates on certain particles, the case where introducing the noise for obeying Xue Shengshi t distribution
Lower resamplingA particle, that is, filter outWeighted value is the smallest in a particleA particle,
And it will be with highest weight weight valuesA particle regenerates;
Step A6:Calculate the estimated location of k-th of moment nodeWherein,
Preferably, the step A4 includes the following steps:
Step A41:K-th of moment calculates all particlesPosition and current state lower node position between Europe
Family name's distance, wherein the position p of k-th of moment nodekBy the estimated location based on -1 moment node of kthInertia sensing
The sensing data of device is calculated, these Euclidean distances are carried out ascending sort, obtain ascending order set Sd, wherein inertia sensing
The sensing data of device includes the node speed value v in -1 time interval of kthk-1, joint movements in -1 time interval of kth
Direction dk-1, change value Δ d of -1 time interval of kth to k-th of time interval movement direction of nodesk, pkCalculation formula be:
Step A42:Calculate ascending order set SdThe average value of middle all elementsAnd by ascending order set SdIn it is all be greater than it is flat
Mean valueEuclidean distance be put into subset according to former sequenceAscending order set SdIn remaining element be then put into subset according to former sequence
Step A43:Calculate separately subsetMiddle all elements and subsetIn preceding i element average valueWith
AndIn surplus element except preceding i element average value
Step A44:Calculate the corresponding effective particle threshold of i and iOptimal value, calculation formula is respectively:
Wherein,ForThe number of middle element.
Preferably, node predicts the position of subsequent time itself, executes node location by step B and predicted
Journey, step B include the following steps:
Step B1:According to the motion conditions that the motion model of node and node actual capabilities occur, and combine kth -1
Node speed value v in a time intervalk-1With movement direction of nodes dk-1, k-th of time interval internal segment spot speed can be obtained
It is worth variation range VrangeAnd relative to movement direction of nodes dk-1Movement direction of nodes variation range Drange;By VrangeDeng between
Isolation dispersion is NvA velocity amplitude, by DrangeIt is discrete at equal intervals to turn to NdA direction change value;
Wherein, Nv、NdFor positive integer, Vrange=[vmin,vmax], Drange=[- θ1,θ2], vminIndicate that node speed value becomes
Change the lower limit of range, vmaxIndicate the upper limit of node speed value variation range, θ1Indicate that the direction of motion changes most along clockwise direction
Big value, θ2Indicate that the direction of motion changes maximum value in the counterclockwise direction;
Step B2:Formula is calculated as follows:
According to the estimated location of k-th of moment nodeK-th i-th of time interval interior nodes possible velocity amplitude vi,
vi∈Vrange,i∈NvAnd j-th of possible direction of motion changing value θj,θj∈Drange,j∈Nd, each pair of v is calculatediWith θj
Combination corresponding to+1 moment node possible position s of kthij, obtain by each couple of viWith θjCombination corresponding to kth+1
Moment node possible position sijThe possible position set S of composition;By each pair of vi、θjProbability value p (vi)、p(θj), it is corresponded to
sijProbability value p (the s of appearanceij), p (sij)=p (vi)·p(θj), it obtains general correspondingly with element in possible position set S
Rate value set P, wherein probability value set P is by each couple of viWith θjCorresponding probability value p (sij) constitute;
Step B3:N is randomly selected from possible position set SrandA node possible position, Nrand≤Nv, and record correspondence
NrandThe set P for the probability value that a node possible position occursrand;By PrandIn probability value sequence, select probability value is highest
NselA probability value is set Psel, and obtain NselThe corresponding possible position set S of a probability valuesel;NselFor positive integer;
Step B4:Normalize set PselIn probability value obtain set Pnorm, then according to set SselThe possibility of interior joint
Predicted position of the position normalization probability calculation node corresponding with possible position at+1 moment of kth:
Wherein, sk+1Indicate predicted position of the node at+1 moment of kth, piIndicate set PnormIn i-th of element, si
Indicate set SselIn i-th of element.
Preferably, boundary node assesses the class that itself whether needs to meet again, and is met again by step C exercise boundary node
Class evaluation process, step C include the following steps:
Step C1:The estimated location that non-leader cluster node will obtain in step AIt is sent to current affiliated cluster CnowCluster head section
Point, leader cluster node integrate all non-leader cluster nodes and the estimated location of itself in cluster;
Step C2:Cluster CnowLeader cluster node communicated with the leader cluster node of adjacent clusters, by cluster CnowInterior all nodes
Estimated location is sent to the leader cluster node of adjacent clusters, and by all nodes in the adjacent clusters obtained from the leader cluster node of adjacent clusters
Estimated location be sent to cluster CnowInterior boundary node;
Step C3:Boundary node is met again class by node according to the estimated locations of all nodes in the adjacent clusters received
Whether appraisal procedure needs to carry out reunion class to current time itself and assesses, that is, calculate itself respectively with cluster Cnow, it is neighbouring
Hypotaxis degree between cluster;
The node reunion class appraisal procedure, specially:
Hypotaxis degree ψ (i, C) calculation between definition node i and cluster C is
Wherein, Euclidean distance of the d (i, j) between node i and node j, node j belong to cluster C;
For each boundary node, will there is the cluster of maximum hypotaxis degree to be most suitable for the cluster being added as current time, be denoted as
Cluster Cnow-opt;If cluster Cnow-optWith cluster CnowDifference then determines that the current time boundary node needs to carry out reunion class.
Preferably, mobile node realizes the reunion class of itself, executes reunion class process by step D, step D includes as follows
Step:
Step D1:In cluster CnowIn, non-leader cluster node is to predicted position obtained in leader cluster node sending step B, non-cluster head
The boundary node for needing to carry out reunion class in node sends the request of reunion class to leader cluster node, and the request of reunion class includes target cluster
Cnow-optInformation;
Step D2:Cluster CnowLeader cluster node by the predicted position of the adjacent clusters interior nodes received, reunion class request and
Predicted position in itself cluster is integrated integrated information after, communicated again with the leader cluster node of adjacent clusters,
To the leader cluster node of adjacent clusters send it is described integrated information, and by the predicted position of the adjacent clusters interior nodes received, meet again
Class request and itself affiliated cluster CnowThe predicted position of interior nodes, the request of reunion class are sent to cluster CnowInterior needs carry out reunion class
Boundary node;
Step D3:The boundary node for needing to carry out reunion class is requested using the node predicted position and reunion class obtained,
Using the node reunion class appraisal procedure, subsequent time itself class that whether needs to meet again is assessed, subsequent time is obtained
It is most suitable for the cluster C being addednext-opt;If Cnext-optWith CnowDifference then illustrates that table tennis is not present in the reunion class process of the boundary node
Pang effect, the boundary node can meet again class to cluster Cnow-optIn, it enters step D4 and continues to execute;Otherwise, illustrate currently to meet again
The boundary node not can be carried out reunion class in the class period, then the boundary node is without reunion class, cluster CnowLeader cluster node will neglect
The reunion class request slightly sent before the boundary node;
Step D4:The boundary node is to cluster CnowLeader cluster node send reunion class confirmation message, to cluster Cnow-optCluster head
Node sends the information that request is added, and is detached from cluster C at+1 moment of kthnowAnd cluster C is addednow-opt。
Compared with prior art, the present invention has the advantages that:
Confirmed by a large amount of Computer Simulation and actual experiment, the present invention by the position to node each moment into
The accurate estimation of row, and when each reunion class period starts, reasonable prediction is carried out in conjunction with the subsequent time position to node, with
And the location information according to nodes other in mobile wireless sensor network, it can be realized the reunion class to node.It is this to section
The method that point carries out reunion class carries out reunion class to node according to certain reunion class period, can guarantee that non-leader cluster node always can
It is enough to be in than better suited cluster, it can so that the communication distance between non-leader cluster node and the leader cluster node of its affiliated cluster
It is maintained in reasonable range, data with higher are sent to rate when so as to guarantee to communicate therebetween.Therefore, of the invention
The heavy clustering method of proposition can eliminate influence of the node mobility to Clustering Effect in mobile wireless sensor network, can
Improve the performance of mobile wireless sensor network.
Detailed description of the invention
Upon reading the detailed description of non-limiting embodiments with reference to the following drawings, other feature of the invention,
Objects and advantages will become more apparent upon:
Fig. 1 is the general frame of mobile wireless sensor network interior joint adapter distribution weight clustering method;
Fig. 2 is that node location is predicted subsequent time position schematically illustrate;
Fig. 3 is that data when communicating between non-leader cluster node and leader cluster node in specific example are sent to rate result.
Specific embodiment
The present invention is described in detail combined with specific embodiments below.Following embodiment will be helpful to the technology of this field
Personnel further understand the present invention, but the invention is not limited in any way.It should be pointed out that the ordinary skill of this field
For personnel, without departing from the inventive concept of the premise, several changes and improvements can also be made.These belong to the present invention
Protection scope.
The present invention provides mobile wireless sensor network interior joint adapter distribution weight clustering methods, are included in
In the mobile wireless sensor network for clustering cluster, node is according to the data of inertial sensor and utilizes dead reckoning and particle
Filtering algorithm accurately estimates the position at moment itself each in moving process;When each reunion class period starts, section
Point moves followed motion model (such as Gauss Markov motion model, walk random motion model) according to it, and
The motion conditions that node actual capabilities occur (are likely to occur in such as reasonable locomotion speed value fluctuation range and adjacent time inter
Direction of motion variation range) and current time position, to subsequent time self-position carry out reasonable prediction, each cluster inner edge
Non- leader cluster node is allowed to carry out reunion class according to the location information of nodes other in network at boundary, thereby may be ensured that node exists
Can be always in than better suited cluster in moving process, i.e., the communication distance between the leader cluster node of its affiliated cluster is protected
It holds in reasonable range, so that being able to maintain higher data when communication therebetween is sent to rate, therefore can guarantee to move
The service quality of wireless sensor network.
A kind of mobile wireless sensor network interior joint adapter distribution weight clustering method provided according to the present invention, packet
Include following steps:
Step A:In the mobile wireless sensor network for having clustered cluster, node in moving process, node according to
The sensing data of inertial sensor estimates the self-position at each moment, obtains the estimated location at each moment;
Step B:When each reunion class period starts, node is according to itself followed motion model of movement and itself works as
The position at preceding moment predicts the position of subsequent time itself, obtains predicted position;
Step C:Estimated location obtained in step A is sent to the leader cluster node of affiliated cluster, cluster head section by non-leader cluster node
The estimated location that point integrates cluster interior nodes is communicated with the leader cluster node of adjacent clusters later, and will be obtained from the leader cluster node of adjacent clusters
Estimated location be sent to cluster inner boundary node, boundary node assesses the class that itself whether needs to meet again;
Step D:Predicted position obtained in step B is sent to the leader cluster node of affiliated cluster, non-cluster head by non-leader cluster node
Need to meet again in node class boundary node by reunion class request be sent to belonging to cluster leader cluster node, leader cluster node obtain cluster in
It is communicated with the leader cluster node of adjacent clusters after the predicted position and reunion class solicited message of node, and will be from the cluster head section of adjacent clusters
The predicted position of predicted position, the request of reunion class and itself affiliated cluster interior nodes that point obtains, the request of reunion class are sent to and need
The boundary node of the boundary node for class of meeting again, the class that needs to meet again completes reunion class.
Preferably, each node operation particle filter algorithm estimates the position at moment itself each in moving process
Meter executes the position estimation procedure of mobile node by step A, specific as follows:
The step A includes the following steps:
Step A1:Initial time, corresponding k=0, the position of node is it is known that N is randomly generated near node location0It is a
Particle;Indicate i-th of particle of the corresponding initial time of k=0, i ∈ { 1 ..., N0,Weighted value be
Step A2:Obtain the update of -1 moment of kth particle position or node estimated location into k-th of etching process
Equation, specially:
When k-th of moment, wherein k >=1 obtains particle position or node estimated location more according to dead reckoning
Newly equation is
Wherein, skParticle position or node estimated location when indicating k-th of moment, sk-1When indicating -1 moment of kth
Particle position or node estimated location, t indicate adjacent moment between time interval length, wherein between assuming each time
Constant, the v every interior nodes velocity amplitude and the direction of motionk-1Indicate the node speed value in -1 time interval of kth, dk-1Indicate kth-
Movement direction of nodes in 1 time interval, Δ dkIndicate -1 time interval of kth to k-th of time interval joint movements side
To change value;- 1 time interval of the kth refers to the time interval between -1 moment of kth and k-th of moment;
Step A3:At -1 moment of kth to the particle state renewal process, including particle at k-th of moment itself position, grain
The update of the weighted value of son and particle develop NevolveA new particle replaces itself;I-th of new grain at k-th of moment
SonNon-normalized weighted valueForNew particleNormalized weight valueForWherein, NkIndicate the total number of particles including all new particles, grain when N () is by -1 moment of kth
SonDevelop new particleProbability, N () obeyFor mean value,For
The Gaussian Profile of variance,Indicate particleNormalized weight value, NevolveFor preset positive integer, particle
Indicate i-th of particle when -1 moment of kth, particleIndicate i-th of new particle when k-th of moment;μkWhen indicating k-th of moment
The mean value of the Gaussian Profile, δkIndicate the variance of Gaussian Profile when k-th of moment,It indicates when k-th of moment j-th
New particle;
Step A4:Calculate effective particle threshold at k-th of moment
Step A5:Calculate the number of effective particles mesh at k-th of momentWherein,IfThen in order to avoid weighted value concentrates on certain particles, the case where introducing the noise for obeying Xue Shengshi t distribution
Lower resamplingA particle, that is, filter outWeighted value is the smallest in a particleA particle, and
It will be with highest weight weight valuesA particle regenerates;
Step A6:Calculate the estimated location of k-th of moment nodeWherein,
Preferably, the step A4 includes the following steps:
Step A41:K-th of moment calculates all particlesPosition and current state lower node position between Europe
Family name's distance, wherein the position p of k-th of moment nodekBy the estimated location based on -1 moment node of kthInertia sensing
The sensing data of device is calculated, these Euclidean distances are carried out ascending sort, obtain ascending order set Sd, wherein inertia sensing
The sensing data of device includes the node speed value v in -1 time interval of kthk-1, joint movements in -1 time interval of kth
Direction dk-1, change value Δ d of -1 time interval of kth to k-th of time interval movement direction of nodesk, pkCalculation formula be:
Step A42:Calculate ascending order set SdThe average value of middle all elementsAnd by ascending order set SdIn it is all be greater than it is flat
Mean valueEuclidean distance be put into subset according to former sequenceAscending order set SdIn remaining element be then put into subset according to former sequence
Step A43:Calculate separately subsetMiddle all elements and subsetIn preceding i element average valueWith
AndIn surplus element except preceding i element average value
Step A44:Calculate the corresponding effective particle threshold of i and iOptimal value, calculation formula is respectively:
Wherein,ForThe number of middle element.
Preferably, node predicts the position of subsequent time itself, executes node location by step B and predicted
Journey, step B include the following steps:
Step B1:According to the motion conditions that the motion model of node and node actual capabilities occur, and combine kth -1
Node speed value v in a time intervalk-1With movement direction of nodes dk-1, k-th of time interval internal segment spot speed can be obtained
It is worth variation range VrangeAnd relative to movement direction of nodes dk-1Movement direction of nodes variation range Drange;By VrangeDeng between
Isolation dispersion is NvA velocity amplitude, by DrangeIt is discrete at equal intervals to turn to NdA direction change value;
Wherein, Nv、NdFor positive integer, Vrange=[vmin,vmax], Drange=[- θ1,θ2], vminIndicate that node speed value becomes
Change the lower limit of range, vmaxIndicate the upper limit of node speed value variation range, θ1Indicate that the direction of motion changes most along clockwise direction
Big value, θ2Indicate that the direction of motion changes maximum value in the counterclockwise direction;
Step B2:Formula is calculated as follows:
According to the estimated location of k-th of moment nodeK-th i-th of time interval interior nodes possible velocity amplitude vi,
vi∈Vrange,i∈NvAnd j-th of possible direction of motion changing value θj,θj∈Drange,j∈Nd, each pair of v is calculatediWith θj
Combination corresponding to+1 moment node possible position s of kthij, obtain by each couple of viWith θjCombination corresponding to kth+1
Moment node possible position sijThe possible position set S of composition;By each pair of vi、θjProbability value p (vi)、p(θj), it is corresponded to
sijProbability value p (the s of appearanceij), p (sij)=p (vi)·p(θj), it obtains general correspondingly with element in possible position set S
Rate value set P, wherein probability value set P is by each couple of viWith θjCorresponding probability value p (sij) constitute;
Step B3:N is randomly selected from possible position set SrandA node possible position, Nrand≤Nv, and record correspondence
NrandThe set P for the probability value that a node possible position occursrand;By PrandIn probability value sequence, select probability value is highest
NselA probability value is set Psel, and obtain NselThe corresponding possible position set S of a probability valuesel;NselFor positive integer;
Step B4:Normalize set PselIn probability value obtain set Pnorm, then according to set SselThe possibility of interior joint
Predicted position of the position normalization probability calculation node corresponding with possible position at+1 moment of kth:
Wherein, sk+1Indicate predicted position of the node at+1 moment of kth, piIndicate set PnormIn i-th of element, si
Indicate set SselIn i-th of element.
Preferably, boundary node assesses the class that itself whether needs to meet again, and is met again by step C exercise boundary node
Class evaluation process, step C include the following steps:
Step C1:The estimated location that non-leader cluster node will obtain in step AIt is sent to current affiliated cluster CnowCluster head
Node, leader cluster node integrate all non-leader cluster nodes and the estimated location of itself in cluster;
Step C2:Cluster CnowLeader cluster node communicated with the leader cluster node of adjacent clusters, by cluster CnowInterior all nodes
Estimated location is sent to the leader cluster node of adjacent clusters, and by all nodes in the adjacent clusters obtained from the leader cluster node of adjacent clusters
Estimated location be sent to cluster CnowInterior boundary node;
Step C3:Boundary node is met again class by node according to the estimated locations of all nodes in the adjacent clusters received
Whether appraisal procedure needs to carry out reunion class to current time itself and assesses, that is, calculate itself respectively with cluster Cnow, it is neighbouring
Hypotaxis degree between cluster;
The node reunion class appraisal procedure, specially:
Hypotaxis degree ψ (i, C) calculation between definition node i and cluster C is
Wherein, Euclidean distance of the d (i, j) between node i and node j, node j belong to cluster C;
For each boundary node, will there is the cluster of maximum hypotaxis degree to be most suitable for the cluster being added as current time, be denoted as
Cluster Cnow-opt;If cluster Cnow-optWith cluster CnowDifference then determines that the current time boundary node needs to carry out reunion class.
Preferably, mobile node realizes the reunion class of itself, executes reunion class process by step D, step D includes as follows
Step:
Step D1:In cluster CnowIn, non-leader cluster node is to predicted position obtained in leader cluster node sending step B, non-cluster head
The boundary node for needing to carry out reunion class in node sends the request of reunion class to leader cluster node, and the request of reunion class includes target cluster
Cnow-optInformation;
Step D2:Cluster CnowLeader cluster node by the predicted position of the adjacent clusters interior nodes received, reunion class request and
Predicted position in itself cluster is integrated integrated information after, communicated again with the leader cluster node of adjacent clusters,
To the leader cluster node of adjacent clusters send it is described integrated information, and by the predicted position of the adjacent clusters interior nodes received, meet again
Class request and itself affiliated cluster CnowThe predicted position of interior nodes, the request of reunion class are sent to cluster CnowInterior needs carry out reunion class
Boundary node;
Step D3:The boundary node for needing to carry out reunion class is requested using the node predicted position and reunion class obtained,
Using the node reunion class appraisal procedure, subsequent time itself class that whether needs to meet again is assessed, subsequent time is obtained
It is most suitable for the cluster C being addednext-opt;If Cnext-optWith CnowDifference then illustrates that table tennis is not present in the reunion class process of the boundary node
Pang effect, the boundary node can meet again class to cluster Cnow-optIn, it enters step D4 and continues to execute;Otherwise, illustrate currently to meet again
The boundary node not can be carried out reunion class in the class period, then the boundary node is without reunion class, cluster CnowLeader cluster node will neglect
The reunion class request slightly sent before the boundary node;
Step D4:The boundary node is to cluster CnowLeader cluster node send reunion class confirmation message, to cluster Cnow-optCluster head
Node sends the information that request is added, and is detached from cluster C at+1 moment of kthnowAnd cluster C is addednow-opt。
More specifically, as shown in Figure 1, in the network for having clustered cluster, the initial position of each node is true
It is fixed and for known to node itself.In the moving process of node, inertia sensing on initial position and node based on node
The data of device are accurately estimated the position at each moment itself using dead reckoning and particle filter algorithm in each node
Meter;When the new reunion class period starts, the motion model followed when node is according to itself movement, to subsequent time itself position
Set carry out reasonable prediction.The estimated location information at itself current time is sent to leader cluster node, leader cluster node by non-leader cluster node
Integrate the location information of cluster interior nodes and communicated with the leader cluster node of adjacent clusters so that each cluster head to obtain its neighbouring
The location information of cluster interior nodes.Cluster inner boundary node most probable occurs needing the case where carrying out reunion class, thus being allowed to can be with
Reunion class.Leader cluster node is sent in this cluster of current time to boundary node and the location information of adjacent clusters interior nodes, boundary node
According to these information to current time itself whether need to meet again class carry out entry evaluation.Non- leader cluster node is sent to leader cluster node
The predicted position of its subsequent time needs the boundary node for carrying out reunion class also to need to send the weight including target cluster information
Cluster request, leader cluster node integrate cluster interior nodes predicted position and reunion class request after again with the leader cluster node of adjacent clusters
It is communicated, the predicted position and reunion class for obtaining adjacent clusters interior nodes request situation;Then need to carry out weight into affiliated cluster
The predicted position and reunion class of cluster and the node in adjacent clusters request where the boundary node of cluster is sent, to be used for these nodes
The subsequent time class that whether needs to meet again is assessed;If occurring without ping-pong, i.e., subsequent time is most suitable for addition
Cluster and current affiliated cluster are not same, then in subsequent time reunion class to current time most suitable cluster.Reunion class mistake
Cheng Zhong, the communication carried out between node introduces corresponding energy consumption, therefore can choose the suitable reunion class period with equal weight
Increased communication distance between energy and non-leader cluster node and leader cluster node consumed by being communicated between cluster process interior joint.
In specific implementation, the characteristics of motion of node can be indicated by motion model, the motion model followed based on node
And data of time interval before, motion conditions of the node in following time interval such as speed and the direction of motion can be changed
Rational prediction is carried out, and then the reasonable prediction of the position to node subsequent time may be implemented.As shown in Fig. 2, according to node
In -1 time interval of kth, (motion model that motion conditions and node in length t) are followed, can obtain node
Variation range [the v of movement velocity in k-th of time intervalmin,vmax] and relative to -1 time interval of kth the direction of motion
dk-1Variation range [- θ1,θ2], by the speed and direction of motion variation discretization in variation range, and taken by the difference of the two
The position s of value combination and the node at k momentk, node can be determined in the possible position region at+1 moment of kth, i.e. Fig. 2
In shadow region.The probability occurred to friction speed value and direction of motion changing value is sampled analysis, can obtain a certain
Group valued combinations have maximum probability, then mutually should be used as in available shadow region, i.e.+1 moment of kth is most probable
Node location, the prediction as node current time to subsequent time position.
It is adaptive in order to evaluate mobile wireless sensor network interior joint proposed by the present invention in emulation and experimentation
The performance for answering distributed weight clustering method, selected data when being communicated between non-leader cluster node and leader cluster node be sent to rate as
Index.
Mobile wireless sensor network interior joint adapter distribution weight clustering method proposed by the present invention is being embodied
When, the different reunion class periods can generate different effects.It is enumerated in Fig. 3 in different reunion class cycle Ts (T=1,4,16)
When, when a length of 1000 time intervals moving process in, non-leader cluster node sends 200 byte lengths to leader cluster node every time
Data packet, therebetween communication when data be sent to rate cumulative distribution (Cumulative distribution function,
CDF).It can be seen from the figure that data are sent to the CDF curve of rate gradually to reduced direction with the increase in reunion class period
Mobile, i.e., data when communicating between non-leader cluster node and leader cluster node are sent to rate and gradually decrease, therefore influence mobile wireless and pass
The service quality of sensor network.In addition, the mobile wireless sensor network interior joint adapter distribution that will be proposed in the present invention
Weight clustering method is compared with SISR, and corresponding data are sent to rate CDF curve and are also contained in Fig. 3.Obviously, relatively
Condition weight clustering method in SISR, using mobile wireless sensor network interior joint adapter distribution proposed by the present invention
Weight clustering method simultaneously chooses suitable reunion class period, number when can significantly improve between non-leader cluster node and leader cluster node
According to rate is sent to, this illustrates that the present invention has positive effect, can be improved the service quality of mobile wireless sensor network.
More specifically, in a preference of the invention, the present invention is achieved by the following technical solutions, this hair
It is bright to include the following steps:
The first step:In the mobile wireless sensor network for having clustered cluster, each node reads inertial sensor
Data, the position at itself each moment is accurately estimated using dead reckoning and particle filter algorithm, obtain phase
The estimated location information answered;
Second step:When the new reunion class period starts, each node moves followed motion model according to itself, by
The movement feelings being likely to occur in the latter time interval at the motion conditions acquisition current time of the previous interval at current time
Condition;According to the position at current time, it can get the possible location sets of node subsequent time using dead reckoning, and by each
The corresponding velocity amplitude in position and the probability value of direction of motion change value calculate that egress subsequent time position is the probability of the position,
By way of sampling, the maximum possible position of probability value in sample, the predicted position as node subsequent time are selected;
Third step:Non- leader cluster node sends the estimated location at current time itself, cluster head section to the leader cluster node of affiliated cluster
Point communicates after integrating to the estimated location information of cluster interior nodes with the leader cluster node of adjacent clusters, is sent to it in cluster
The estimated location information of node, while the estimated location at adjacent clusters interior nodes current time is also obtained, and be sent to cluster inner boundary
Node;
4th step:The estimated location information of boundary node its affiliated cluster and adjacent clusters interior nodes according to current time and work as
The distribution situation of preceding moment each cluster obtains its cluster for being most suitable for being added to itself whether needing to carry out reunion class to assess,
If the cluster and current affiliated cluster are inconsistent, illustrate that the boundary node needs to carry out reunion class;
5th step:Non- leader cluster node sends the predicted position at current time itself to the leader cluster node of affiliated cluster, if node
To need to carry out the boundary node of reunion class, then also need to send the request of reunion class to leader cluster node.Leader cluster node is integrated in cluster
The predicted position and reunion class of node are requested, and are communicated with the leader cluster node of adjacent clusters, and the predicted position of cluster interior nodes is sent
The predicted position requested with reunion class, while also obtaining adjacent clusters interior nodes is given with reunion class request Concurrency meets again
The boundary node of class;
6th step:Need to carry out the boundary node of reunion class according to the predicted position of affiliated cluster and adjacent clusters interior nodes with again
The distribution situation of the case where cluster request and current time cluster, is commented whether itself subsequent time needs to carry out reunion class
Estimate, obtains its subsequent time and be most suitable for the cluster being added;
7th step:If desired the boundary node subsequent time for carrying out reunion class is most suitable for the cluster being added and current affiliated cluster
It is identical, then illustrate ping-pong occurred, therefore current time not can be carried out reunion class;If not identical, which is working as
The preceding moment can carry out reunion class and is most suitable in the cluster of its addition to current time.Need the boundary node for carrying out reunion class to working as
The leader cluster node of cluster belonging to preceding sends reunion class confirmation message, and sends to the leader cluster node for the cluster that will be added and request is added,
Then reunion class is completed in subsequent time.
Specific embodiments of the present invention are described above.It is to be appreciated that the invention is not limited to above-mentioned
Particular implementation, those skilled in the art can make a variety of changes or modify within the scope of the claims, this not shadow
Ring substantive content of the invention.In the absence of conflict, the feature in embodiments herein and embodiment can any phase
Mutually combination.
Claims (1)
1. a kind of mobile wireless sensor network interior joint adapter distribution weight clustering method, which is characterized in that including as follows
Step:
Step A:In the mobile wireless sensor network for having clustered cluster, node is in moving process, and node is according to inertia
The sensing data of sensor estimates the self-position at each moment, obtains the estimated location at each moment;
Step B:When each reunion class period starts, node according to itself move followed motion model and itself it is current when
The position at quarter predicts the position of subsequent time itself, obtains predicted position;
Step C:Estimated location obtained in step A is sent to the leader cluster node of affiliated cluster by non-leader cluster node, and leader cluster node is whole
The estimated location for closing cluster interior nodes is communicated with the leader cluster node of adjacent clusters later, and is estimated what is obtained from the leader cluster node of adjacent clusters
Meter position is sent to cluster inner boundary node, and boundary node assesses the class that itself whether needs to meet again;
Step D:Predicted position obtained in step B is sent to the leader cluster node of affiliated cluster, non-leader cluster node by non-leader cluster node
The request of reunion class is sent to the leader cluster node of affiliated cluster by the boundary node of the middle class that needs to meet again, and leader cluster node obtains cluster interior nodes
Predicted position and reunion class solicited message after communicated with the leader cluster node of adjacent clusters, and will be obtained from the leader cluster node of adjacent clusters
The predicted position of the predicted position, the request of reunion class and itself affiliated cluster interior nodes that obtain, the request of reunion class are sent to and need weight
The boundary node of the boundary node of cluster, the class that needs to meet again completes reunion class;
Each node operation particle filter algorithm estimates the position at moment itself each in moving process, passes through step A
The position estimation procedure of mobile node is executed, it is specific as follows:
The step A includes the following steps:
Step A1:Initial time, corresponding k=0, the position of node is it is known that N is randomly generated near node location0A particle;Indicate i-th of particle of the corresponding initial time of k=0, i ∈ { 1 ..., N0,Weighted value be
Step A2:Obtain -1 moment of kth update side of particle position or node estimated location into k-th of etching process
Journey, specially:
When k-th of moment, wherein k >=1 obtains the update side of particle position or node estimated location according to dead reckoning
Cheng Wei:
Wherein, skParticle position or node estimated location when indicating k-th of moment, sk-1Indicate grain when -1 moment of kth
Sub- position or node estimated location, t indicates the length of time interval between adjacent moment, wherein assuming in each time interval
Node speed value and the direction of motion are constant, vk-1Indicate the node speed value in -1 time interval of kth, dk-1Indicate kth -1
Movement direction of nodes in time interval, Δ dkIndicate -1 time interval of kth to k-th time interval movement direction of nodes
Change value;- 1 time interval of the kth refers to the time interval between -1 moment of kth and k-th of moment;
Step A3:At -1 moment of kth to the particle state renewal process at k-th of moment, including particle position itself, particle
The update of weighted value and particle develop NevolveA new particle replaces itself;I-th of new particle at k-th of moment
Non-normalized weighted valueForNew particleNormalized weight valueForWherein, NkIndicate the total number of particles including all new particles, grain when N () is by -1 moment of kth
SonDevelop new particleProbability, N () obeyFor mean value,
For the Gaussian Profile of variance,Indicate particleNormalized weight value, NevolveFor preset positive integer, particle
Indicate i-th of particle when -1 moment of kth, particleIndicate i-th of new particle when k-th of moment;μkWhen indicating k-th of moment
The mean value of the Gaussian Profile, δkIndicate the variance of Gaussian Profile when k-th of moment,It indicates when k-th of moment j-th
New particle;
Step A4:Calculate effective particle threshold at k-th of moment
Step A5:Calculate the number of effective particles mesh at k-th of momentWherein,If
Then in order to avoid weighted value concentrates on certain particles, the resampling in the case where introducing the noise for obeying Xue Shengshi t distributionA particle, that is, filter outWeighted value is the smallest in a particleA particle, and will have most
High weighted valueA particle regenerates;
Step A6:Calculate the estimated location of k-th of moment nodeWherein,
The step A4 includes the following steps:
Step A41:K-th of moment calculates all particlesPosition and current state lower node position between Euclidean away from
From, wherein the position p of k-th of moment nodekBy the estimated location based on -1 moment node of kthInertial sensor
Sensing data is calculated, these Euclidean distances are carried out ascending sort, obtain ascending order set Sd, wherein inertial sensor
Sensing data includes the node speed value v in -1 time interval of kthk-1, movement direction of nodes in -1 time interval of kth
dk-1, change value Δ d of -1 time interval of kth to k-th of time interval movement direction of nodesk, pkCalculation formula be:
Step A42:Calculate ascending order set SdThe average value of middle all elementsAnd by ascending order set SdIn it is all be greater than average value
Euclidean distance be put into subset according to former sequenceAscending order set SdIn remaining element be then put into subset according to former sequence
Step A43:Calculate separately subsetMiddle all elements and subsetIn preceding i element average valueAnd
In surplus element except preceding i element average value
Step A44:Calculate the corresponding effective particle threshold of i and iOptimal value, calculation formula is respectively:
Wherein,ForThe number of middle element;
Node predicts the position of subsequent time itself, executes node location by step B and predicts process, step B includes
Following steps:
Step B1:According to the motion conditions that the motion model of node and node actual capabilities occur, and when combining kth -1
Between interval in node speed value vk-1With movement direction of nodes dk-1, k-th of time interval interior nodes velocity amplitude can be obtained and become
Change range VrangeAnd relative to movement direction of nodes dk-1Movement direction of nodes variation range Drange;By VrangeAt equal intervals from
Dispersion is NvA velocity amplitude, by DrangeIt is discrete at equal intervals to turn to NdA direction change value;
Wherein, Nv、NdFor positive integer, Vrange=[vmin,vmax], Drange=[- θ1,θ2], vminIndicate that node speed value changes model
The lower limit enclosed, vmaxIndicate the upper limit of node speed value variation range, θ1Indicate that the direction of motion changes maximum along clockwise direction
Value, θ2Indicate that the direction of motion changes maximum value in the counterclockwise direction;
Step B2:Formula is calculated as follows:
According to the estimated location of k-th of moment nodeK-th i-th of time interval interior nodes possible velocity amplitude vi,vi∈
Vrange,i∈NvAnd j-th of possible direction of motion changing value θj,θj∈Drange,j∈Nd, each pair of v is calculatediWith θj's
Corresponding+1 moment node possible position s of kth of combinationij, obtain by each couple of viWith θjCombination corresponding to kth+1 when
Carve node possible position sijThe possible position set S of composition;By each pair of vi、θjProbability value p (vi)、p(θj), obtain corresponding sij
Probability value p (the s of appearanceij), p (sij)=p (vi)·p(θj), it obtains and the one-to-one probability of element in possible position set S
Value set P, wherein probability value set P is by each couple of viWith θjCorresponding probability value p (sij) constitute;
Step B3:N is randomly selected from possible position set SrandA node possible position, Nrand≤Nv, and record corresponding Nrand
The set P for the probability value that a node possible position occursrand;By PrandIn probability value sequence, the highest N of select probability valuesel
A probability value is set Psel, and obtain NselThe corresponding possible position set S of a probability valuesel;NselFor positive integer;
Step B4:Normalize set PselIn probability value obtain set Pnorm, then according to set SselThe possible position of interior joint with
Predicted position of the corresponding normalization probability calculation node of possible position at+1 moment of kth:
Wherein, sk+1Indicate predicted position of the node at+1 moment of kth, piIndicate set PnormIn i-th of element, siIt indicates
Set SselIn i-th of element;
Boundary node assesses the class that itself whether needs to meet again, by step C exercise boundary node reunion class evaluation process,
Step C includes the following steps:
Step C1:The estimated location that non-leader cluster node will obtain in step AIt is sent to current affiliated cluster CnowLeader cluster node,
Leader cluster node integrates all non-leader cluster nodes and the estimated location of itself in cluster;
Step C2:Cluster CnowLeader cluster node communicated with the leader cluster node of adjacent clusters, by cluster CnowThe estimation of interior all nodes
Position is sent to the leader cluster node of adjacent clusters, and all nodes in the adjacent clusters obtained from the leader cluster node of adjacent clusters are estimated
Meter position is sent to cluster CnowInterior boundary node;
Step C3:Boundary node passes through node weight Cluster Evaluation according to the estimated location of all nodes in the adjacent clusters received
Whether method needs to carry out reunion class to current time itself and assesses, that is, calculate itself respectively with cluster Cnow, adjacent clusters it
Between hypotaxis degree;
The node reunion class appraisal procedure, specially:
Hypotaxis degree ψ (i, C) calculation between definition node i and cluster C is
Wherein, Euclidean distance of the d (i, j) between node i and node j, node j belong to cluster C;
For each boundary node, will there is the cluster of maximum hypotaxis degree to be most suitable for the cluster being added as current time, be denoted as cluster
Cnow-opt;If cluster Cnow-optWith cluster CnowDifference then determines that the current time boundary node needs to carry out reunion class;
Mobile node realizes the reunion class of itself, executes reunion class process by step D, step D includes the following steps:
Step D1:In cluster CnowIn, non-leader cluster node is to predicted position obtained in leader cluster node sending step B, non-leader cluster node
The boundary node that middle needs carry out reunion class sends the request of reunion class to leader cluster node, and the request of reunion class includes target cluster Cnow-opt
Information;
Step D2:Cluster CnowLeader cluster node by the predicted position of the adjacent clusters interior nodes received, reunion class request and itself
Predicted position in cluster is integrated integrated information after, communicated again with the leader cluster node of adjacent clusters, to neighbour
The leader cluster node of nearly cluster has integrated information described in sending, and the predicted position of the adjacent clusters interior nodes received, reunion class are asked
It asks and itself affiliated cluster CnowThe predicted position of interior nodes, the request of reunion class are sent to cluster CnowInterior needs carry out the side of reunion class
Boundary's node;
Step D3:The boundary node for needing to carry out reunion class is requested using the node predicted position and reunion class obtained, is utilized
The node reunion class appraisal procedure assesses subsequent time itself class that whether needs to meet again, it is most suitable to obtain subsequent time
Close the cluster C being addednext-opt;If Cnext-optWith CnowDifference then illustrates that table tennis effect is not present in the reunion class process of the boundary node
It answers, which can meet again class to cluster Cnow-optIn, it enters step D4 and continues to execute;Otherwise, illustrate current reunion class week
The interim boundary node not can be carried out reunion class, then the boundary node is without reunion class, cluster CnowLeader cluster node will ignore this
The reunion class request sent before boundary node;
Step D4:The boundary node is to cluster CnowLeader cluster node send reunion class confirmation message, to cluster Cnow-optLeader cluster node
The information that request is added is sent, and is detached from cluster C at+1 moment of kthnowAnd cluster C is addednow-opt。
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201510638551.5A CN105263149B (en) | 2015-09-29 | 2015-09-29 | Mobile wireless sensor network interior joint adapter distribution weight clustering method |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201510638551.5A CN105263149B (en) | 2015-09-29 | 2015-09-29 | Mobile wireless sensor network interior joint adapter distribution weight clustering method |
Publications (2)
Publication Number | Publication Date |
---|---|
CN105263149A CN105263149A (en) | 2016-01-20 |
CN105263149B true CN105263149B (en) | 2018-11-16 |
Family
ID=55102617
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201510638551.5A Active CN105263149B (en) | 2015-09-29 | 2015-09-29 | Mobile wireless sensor network interior joint adapter distribution weight clustering method |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN105263149B (en) |
Families Citing this family (7)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN109313841B (en) * | 2016-05-09 | 2021-02-26 | 塔塔咨询服务有限公司 | Method and system for implementing adaptive clustering in sensor networks |
CN106507275B (en) * | 2016-10-14 | 2019-11-08 | 清华大学 | A kind of robust Distributed filtering method and apparatus of wireless sensor network |
CN108566663B (en) * | 2018-01-10 | 2021-06-15 | 重庆邮电大学 | SDWSN energy consumption balance routing method based on disturbance particle swarm optimization |
CN110823212B (en) * | 2018-08-14 | 2023-04-07 | 北京自动化控制设备研究所 | SINS/DR integrated navigation system position tracking determination method based on particle filtering |
CN108816525B (en) * | 2018-09-12 | 2023-08-18 | 齐鲁工业大学 | Soft measuring device and method for concentration of stirring tank of copper ore flotation machine |
CN115952426B (en) * | 2023-03-10 | 2023-06-06 | 中南大学 | Distributed noise data clustering method based on random sampling and user classification method |
CN117941673A (en) * | 2024-03-26 | 2024-04-30 | 中国科学院水生生物研究所 | Acoustic driving network intelligent monitoring driving method based on multi-source sounder |
Citations (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN101639535A (en) * | 2009-08-28 | 2010-02-03 | 茂名学院 | Wireless sensor network multi-target tracking method for fuzzy clustering particle filtering |
CN102711210A (en) * | 2012-06-04 | 2012-10-03 | 广州市香港科大***研究院 | Data collecting method based on wireless sensor network adopting dynamic clustering routing algorithm |
-
2015
- 2015-09-29 CN CN201510638551.5A patent/CN105263149B/en active Active
Patent Citations (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN101639535A (en) * | 2009-08-28 | 2010-02-03 | 茂名学院 | Wireless sensor network multi-target tracking method for fuzzy clustering particle filtering |
CN102711210A (en) * | 2012-06-04 | 2012-10-03 | 广州市香港科大***研究院 | Data collecting method based on wireless sensor network adopting dynamic clustering routing algorithm |
Non-Patent Citations (2)
Title |
---|
Dynamic Cluster Header Selection and Conditional Re-clustering;Jinsuk Baek et al.;《IEEE Transactions on Consumer Electronics》;20101130;全文 * |
Survey on Data Collection Protocols in Wireless;Ranadeep Mukherjee et al.;《Computing for Sustainable Global Development(INDIACom)》;20150313;全文 * |
Also Published As
Publication number | Publication date |
---|---|
CN105263149A (en) | 2016-01-20 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN105263149B (en) | Mobile wireless sensor network interior joint adapter distribution weight clustering method | |
Chen et al. | Label-less learning for traffic control in an edge network | |
CN109544911B (en) | Urban road network traffic state prediction method based on LSTM-CNN | |
US11551154B1 (en) | Predictive power management in a wireless sensor network | |
Pathirana et al. | Node localization using mobile robots in delay-tolerant sensor networks | |
Cao et al. | Swarm intelligence-based performance optimization for mobile wireless sensor networks: survey, challenges, and future directions | |
CN102685255B (en) | Distributed opportunistic network community division method | |
Manweiler et al. | Predicting length of stay at wifi hotspots | |
WO2021211282A1 (en) | Architecture for machine learning (ml) assisted communications networks | |
Nadembega et al. | A destination prediction model based on historical data, contextual knowledge and spatial conceptual maps | |
CN105228159B (en) | Wireless sense network coverage enhancement algorithm based on gridding and improvement particle swarm algorithm | |
CN105050095B (en) | A kind of topological construction method of the heterogeneous wireless sensor net based on energy predicting | |
CN101848529A (en) | Method for compressing multiple principle component analysis data of wireless sensor network | |
CN103079270B (en) | A kind of mobile node based on wireless location assembles cognitive method | |
CN110267292A (en) | Cellular network method for predicting based on Three dimensional convolution neural network | |
CN101610567A (en) | A kind of dynamic group scheduling method based on wireless sensor network | |
US20230209419A1 (en) | Machine learning handover prediction based on sensor data from wireless device | |
CN106793031A (en) | Based on the smart mobile phone energy consumption optimization method for gathering competing excellent algorithm | |
Chen et al. | Smartphone-based indoor fingerprinting localization using channel state information | |
CN103971384A (en) | Node cooperation target tracking method of wireless video sensor | |
Wala et al. | Energy efficient data collection in smart cities using iot | |
Li et al. | Mobile user location prediction based on user classification and markov model | |
Parvin et al. | Gravitational search algorithm based mobile aggregator sink nodes for energy efficient wireless sensor networks | |
CN110049500B (en) | UAV energy compensation method in wireless chargeable sensor network based on simulated annealing algorithm | |
CN106102148B (en) | A kind of base station dormancy method and device |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
C06 | Publication | ||
PB01 | Publication | ||
C10 | Entry into substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
GR01 | Patent grant | ||
GR01 | Patent grant |