CN103945401B - Towards the wireless sensor network automatic deployment method of non-homogeneous sensing region - Google Patents

Towards the wireless sensor network automatic deployment method of non-homogeneous sensing region Download PDF

Info

Publication number
CN103945401B
CN103945401B CN201410182101.5A CN201410182101A CN103945401B CN 103945401 B CN103945401 B CN 103945401B CN 201410182101 A CN201410182101 A CN 201410182101A CN 103945401 B CN103945401 B CN 103945401B
Authority
CN
China
Prior art keywords
sensor
load
region
circumscribed circle
minimum circumscribed
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
Application number
CN201410182101.5A
Other languages
Chinese (zh)
Other versions
CN103945401A (en
Inventor
吕琳
梁广会
杨承磊
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Shandong University
Original Assignee
Shandong University
Priority date (The priority date 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 date listed.)
Filing date
Publication date
Application filed by Shandong University filed Critical Shandong University
Priority to CN201410182101.5A priority Critical patent/CN103945401B/en
Publication of CN103945401A publication Critical patent/CN103945401A/en
Application granted granted Critical
Publication of CN103945401B publication Critical patent/CN103945401B/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Landscapes

  • Arrangements For Transmission Of Measured Signals (AREA)

Abstract

The invention discloses towards the wireless sensor network automatic deployment method of non-homogeneous sensing region, it is random to generate sensor initial position；The minimum circumscribed circle heart of the position of movable sensor to its corresponding domination region is until convergence；According to load restriction and the present load of sensor, the load request of setting sensor of sensor, it is each sensor distribution load；Optimize the weight of each sensor, find the division of the load request set in meeting step 3；Judge that all the sensors current location is moved to its correspondence minimum circumscribed circle heart needs mobile average distance, if average distance is more than its condition of convergence, is moved to the correspondence minimum circumscribed circle heart by sensing station, and goes to step 3；The otherwise division in output transducer position, radius and region.The present invention can carry out radio sensing network deployment to non-homogeneous monitored area；During sensor deployment is carried out, each sensor non-overloading is ensure that.

Description

Towards the wireless sensor network automatic deployment method of non-homogeneous sensing region
Technical field
The present invention relates to the automatic deployment method of sensor, more particularly to a kind of wireless biography towards non-homogeneous sensing region Sensor network automatic deployment method.
Background technology
Wireless sensor network (Wireless Sensor Network, WSN) is substantial amounts of in monitored area by being deployed in Cheap microsensor node composition, the network system of the self-organizing of the multi-hop formed by communication, its mesh Be collaboratively to perceive, collection and process the information of perceived object in network's coverage area, and be sent to observer.Sensing Device, perceptive object and observer constitute three key elements of wireless sensor network.Wireless sensor network has numerous types Sensor, it is detectable including earthquake, electromagnetism, temperature, noise, light intensity, pressure, soil constituent, the size of mobile object, speed Diversified phenomenon in degree and the surrounding enviroment such as direction, and the information of the region that can monitor of each sensor or process Amount has a upper limit, referred to as maximum load.Wireless sensor network, can be real-time used as a kind of brand-new information acquisition platform The information of the various detection objects in monitoring and collection network distribution region, and gateway node is sent such information to, with reality Object detecting and tracking in existing complicated specified range, with rapid deployment, survivability is strong the features such as, and be widely used in Environmental monitoring and protection, medical treatment and nursing, military field and industry monitoring etc..
In wireless sensor network, in order to realize the effective monitoring to monitoring objective, monitoring objective is typically required Each position at least in the monitoring range of a sensor.Covering problem in wireless senser includes three classes：Area covers Lid, point are covered and obstacle is covered, and wherein area covering is most widely used.Area covering can be divided into two classes again：Once cover, I.e. each position of monitored area at least can be arrived by a Sensor monitoring；Each position of multi-fold, i.e. monitored area At least can be arrived by two or more Sensor monitorings.In the present invention, we mainly study and once cover, and below referred to as cover Lid.For whole monitored area, we term it non-homogeneous sensing region (Prioritized Sensing Field), each The probability for monitoring information being produced on position or event occurring is incomplete same.In the present invention, we will produce monitoring Information or the probability density function ρ that event occurs are represented.
Have much for area covering problem at present, have already been proposed many methods.Such as：For given Monitored area, in order that area coverage reaches maximum, and uses sensor as few as possible, Jing Li and Hai-ping Huang is based on the Voronoi diagram with border, it is proposed that OCDSN (Optimal coverage in directional sensor networks)(J.Li,R.-c.Wang,H.-p.Huang,and L.-j.Sun,“Voronoi based area coverage optimization for directional sensor networks,”International Symposium in Electronic Commerce and Security, vol.1,2009, pp.488-493, " the oriented biography based on Voronoi diagram Sensor network coverage optimization ", ecommerce and safe international Conference, 2009,488-493).In this approach, often Individual sensor has an adjustable angle, i.e. direction, can increase the face of covering by the direction of adjustment sensor Product.In order to increase the reliability of wireless sensor network, S.Poduri and G.S.Sukhatme proposes a kind of band and limits Covering problem, i.e. individual adjacent node (the S.Poduri and of each sensor at least k (being previously set) G.S.Sukhatme, " Constrained coverage for mobile sensor networks ", IEEE International Conference on Robotics and Automation, 2004, vol.1, pp.165-171 " are moved The restriction of dynamic sensor network is covered ", robot and automation international conference, 2004,165-171).This method ensure that 95% sensor has at least k neighbours, increases the reliability and stability of wireless sensor network to a certain extent.But It is that above two method all has two shortcomings：
1) assume in monitored area, on each position event occur probability be it is identical, but this is in reality It is unreasonable using in.
2) load for assuming each sensor is infinitely great, does not account for sensor overload and may cause to network Infringement.
For the first shortcoming above, monitored area is given, number and the monitoring of probability and sensor that event occurs Radius, Mahboubi are based on multiplicatively weighted Voronoi diagram, by by the position of sensor Gradually to density it is big region movement, reached with this increase monitoring event probability (H.Mahboubi, J.Habibi, A.Aghdam and K.Sayrafian-Pour, " Distributed deployment strategies for improved Coverage in a network of mobile sensors with prioritized sensing field ", IEEE Transactions on Industrial Informatics2013, vol.9,451-461, based on the wireless of covering priority Movable sensor Deployment Algorithm, industrial information, IEEE, 451-461).This method thinks each position event in monitored area The probability of generation is different, and in the final deployment result for obtaining where event occurrence rate height, the distribution of sensor is relative It is intensive, on the contrary then sensor distribution is sparse.But this method has been previously set the radius of each sensor, in its radius Within region will be monitored, do not account for whether sensor overloads.Jorge Cort é s are based on additively Weighted Voronoi partitions propose a kind of Jacobi iterative algorithms (J.Cortes, " Coverage Optimization and spatial load balancing by robotic sensor networks ", IEEE Transactions on Automatic Control, vol.55, no.3, pp.749-754,2010, machine sensor network Coverage optimization and load balance, automatically control, 2010,749-754).This method needs to be input into the negative of each sensor Carry, ensure that the load of all the sensors is equal to the load value being input into, and all the sensors during algorithm iteration All non-overloadings.But there is another drawback in this method：
3) load of each sensor is needed as input, and require that all the sensors load sum is equal to density function Integration on monitored area, so before disposing to sensor, artificial load distribution to be carried out to sensor, it is non- It is often dumb.
The content of the invention
To solve the deficiency that prior art is present, the invention discloses towards the wireless sensor network of non-homogeneous sensing region Network automatic deployment method, the present invention can carry out the side of wireless sensor network disposition according to monitored area event occurrence rate Method.The present invention has and perceives non-homogeneous monitored area event density, and ensures each sensor without departing from maximum load, quickly The features such as automatic deployment of response and sensor.Described in the technical program, sensor refers both to wireless senser.
For achieving the above object, concrete scheme of the invention is as follows：
Towards the wireless sensor network automatic deployment method of non-homogeneous sensing region, comprise the following steps：
Step one：According to given monitored area, number of sensors and sensor maximum load, random generation sensor is initial Position；
Step 2：The minimum circumscribed circle heart of the position of movable sensor to its corresponding domination region is until convergence；
Step 3：According to sensor load limit and sensor present load, the load request of setting sensor, For each sensor distribution load；
Step 4：Optimize the weight of each sensor, find the division of the load request set in meeting step 3；
Step 5：Judge that all the sensors current location is moved to its correspondence minimum circumscribed circle heart needs the average of movement Sensing station, if average distance is more than its condition of convergence, is moved to the correspondence minimum circumscribed circle heart, and goes to step by distance Rapid three；The otherwise division in output transducer position, radius and region.
Sensor initial position collection in the step one is combined intoWherein n is the integer more than 1, monitoring Region is Ω, and the event on the Ω of region occurs density function for ρ, ρ >=0, sensor piPresent load be Ci, maximum load is C0。
Concretely comprising the following steps in the step 2：
21) by each sensor piWeights omegai0 is set to, and constructs sensing stationEnergy diagram；
22) calculate each sensor piDomination region Ωi；
23) calculate each sensor piArranged region ΩiMinimum circumscribed circle OiThe center of circle be ci, radius is ri；
24) calculate the average distance that each sensor is moved to the heart movement of correspondence minimum circumscribed circle IfBy each sensing station piIt is moved to correspondence minimum circumscribed circle heart ci；Otherwise output transducer PositionAnd energy diagram is dividedσ is threshold value.
The construction process of the energy diagram is：OrderFor RmA scatterplot collection in space, each point p in PiQuilt Give a weights ωi>=0, RmAny point p in space in any point p to scatterplot collection PiEnergy distance definition be：
dω(p,pi)=| | p-pi||2-ωi
It is criterion to space R with energy distancemDivided, defined V (pi) be and scatterplot piThe region of association, has：
pjRepresent that scatterplot is concentrated except piAny point.Define V (pi), i=1 ..., the collection of n is combined into the energy of scatterplot collection P Spirogram.
The σ threshold values value is 10-6。
The step 3 is concretely comprised the following steps：
31) for each sensor pi, density function ρ (x) is calculated in its correspondence domination region ΩiOn integration, as which Load Ci, i.e.,：ρ (x) represents the probability that event occurs on monitored area, and x represents appointing on monitored area Meaning one point, i.e. integration variable；
32) by the load of all the sensorsArranged according to descending, then chosen the two of head of the queue and tail of the queue every time Individual sensor is processed, and the load of two sensors is set to CmaxAnd Cmin, temporary variable is CtIf, Cmax≥C0If, Determine C'max=C0, Ct+=Cmax-C0；If Cmin+Ct≥C0, set C'min=C0, Ct=Cmin+Ct-C0, otherwise set C'min+= Ct, Ct=0, the load for exporting all the sensors is limitedC'maxAnd C'minTwo sensors for being expressed as selecting every time set Fixed new load value, will optimize region division and sensing station so that the load of each sensor reaches in step 4 The target that this step is arranged；
The step 4 is concretely comprised the following steps：
41) object function isGiven for one Divide, object function is weight setFunction,X represents integration variable；
42) using Newton method minimization object function.
The step 42) using Newton method minimization object function, detailed process is as follows：
421) gradient descent direction d is solved, is allowed to meet ▽2F (W) d=- ▽ F (W)；
422) determine step-length α for meeting Armijo conditions to meet the decline of F (W+ α d) value；
423) displacement of a α d, i.e. ω are carried out to the weight of sensori'←ωi+ α d, ωi' it is for sensor settings New weight；
424) according to new sensor weight, calculate new energy diagram, when object function with regard toGradient ▽ F (W) >=ε when, return to step 421), otherwise output transducer position and region division；Wherein, ε is convergence Rule of judgment.
The ε is related to areal concentration for convergence Rule of judgment,
The step 5 is concretely comprised the following steps：
51) calculate the minimum circumscribed circle center of circle c that each sensor arranges regioniAnd radius ri；
If 52)By the position p of each sensoriIt is moved to corresponding minimum circumscribed circle center of circle ri And go to step 3, the otherwise position of output transducerRadiusAnd region divisionζ is convergence bar Part.
The ζ is the condition of convergence, and its value is 10-5。
After the division that load is limited is met in the step 4, for current division, in order to reduce covering for maximum All of sensing station is all moved to the minimum circumscribed circle center of circle in correspondence domination region, i.e., by lid radius：pi←ci, i= 1,...,n。
Beneficial effects of the present invention：
1. the present invention can carry out radio sensing network deployment to non-homogeneous monitored area；
2. the present invention ensure that each sensor non-overloading during sensor deployment is carried out；
3. the present invention during optimization weight employs Optimal transport methods, ensure that when monitored In region
During generation event, the response of sensor is obtained as soon as possible；
4. the present invention can carry out automatic sensor deployment to monitored area, it is not necessary to artificial participation.
Description of the drawings
Fig. 1 (a) is the density map of given square monitored area；
Fig. 1 (b) is initial random sensor distribution；
Fig. 2 (a) is the coverage condition of initial random sensor and load distribution condition；
Fig. 2 (b) is that sensor is moved to covering and the load distribution condition that convergence is obtained to minimum circumscribed circle unfaithful intention；
Fig. 3 (a) is any given division and load distribution condition；
Fig. 3 (b) is for Fig. 3 (a) is according to the division and load distribution condition obtained after given load restriction optimization；
Fig. 4 (a)-(f) be respectively for different maximum load limit the wireless senser deployment coverage diagram for obtaining with And load distribution condition.
Fig. 5 is the flow process frame diagram of the present invention.
Specific embodiment：
The present invention is described in detail below in conjunction with the accompanying drawings：
The present invention will be further elaborated with example below in conjunction with the accompanying drawings, it should explanation, and the description below is only In order to explain the present invention, content is not defined.
The invention provides a kind of wireless sensor network automatic deployment method of perceived density, Fig. 5 is according to the present invention Embodiment radio sensing network dispositions method flow process frame diagram, step is as follows：
1) it is random to generate sensor initial position, and energy diagram is generated, such as flow chart 501-502；
2) the minimum circumscribed circle heart of the position of movable sensor to corresponding domination region is until convergence, such as flow chart 503- 504；
3) the load restriction and the present load of sensor according to sensor, is each sensor distribution load, such as flows Journey Figure 50 5-506；
4) optimize the weight of each sensing station, find suitable division, meet the load request for 3) setting, such as flow process Figure 50 7-508；
5) judge that all the sensors current location is moved to its correspondence minimum circumscribed circle heart needs mobile average distance, If average distance is larger, sensing station is moved to into the correspondence minimum circumscribed circle heart, and goes to step 3)；Otherwise output is passed The division in sensor position, radius and region, such as flow chart 509-510.
The step 1) in sensor initial position collection be combined intoWherein n is the integer more than 1, monitoring Region is Ω, and the event on the Ω of region occurs density function for ρ, ρ >=0, sensor piPresent load be Ci, maximum load is C0.If Fig. 1 (a) is unit square monitored area, the density function ρ=x on region2+y2, the gray scale in region represents that density is big Little, more black expression density is bigger, otherwise less.Fig. 1 (b) is the 20 sensing station distributions spread on monitored area at random, Black line represents that the energy diagram to monitored area is divided.
The step 2) concretely comprise the following steps：
21) by each sensor piWeights omegai0 is set to, and constructs sensing stationEnergy diagram；
22) calculate each sensor piDomination region Ωi；
23) calculate each sensor piArranged region ΩiMinimum circumscribed circle Oi, the center of circle is ci, radius is ri；
24) calculateBy each sensing station piIt is moved to correspondence minimum circumscribed circle heart ci, and turn To step 21)；Otherwise output transducer positionAnd energy diagram is dividedσ is threshold value, takes 10-6。
For given load is limited, it is energy diagram (Power diagram) to meet the division for limiting, so in the present invention Using energy diagram dividing to monitored area.Energy diagram is a kind of weighted Voronoi diagrams proposed by Auernhammer Figure.OrderFor RmA scatterplot collection in space, each point p in PiIt is endowed a weights ωi≥0。RmAppoint in space Any point p in 1 point of p to scatterplot collection PiEnergy distance definition be：
dω(p,pi)=| | p-pi||2-ωi
It is criterion to space R with energy distancemDivided, defined V (pi) be and scatterplot piThe region of association, has：
pjRepresent that scatterplot is concentrated except piAny point.Define V (pi), i=1 ..., the collection of n is combined into the energy of scatterplot collection P Spirogram.
Give a closed areaMake ΩiRepresent V (pi) with the common factor part of Ω, have：
Ωi=V (pi)∩Ω
Then all ΩiUnion be region Ω, claim ΩiFor piUnit, define all ΩiIt is formed in P in the Ω of region The domination region of each sensor.
After dividing for given area, each sensor will cover its corresponding domination region, in order to Complete covering domination region, and cause the covering radius of sensor minimum, it is necessary to area is arranged into which in the position of sensor The minimum circumscribed circle center of circle movement in domain, until convergence.So ensure that the radius of sensor is minimum, more save energy.Such as Fig. 2 A () is initial random sensor distribution situation, coverage condition and loading condition.Circle represents the minimum external of correspondence domination region Circle (sensor covering radius), Grey Point represent current sensor position, and black color dots represent the minimum circumscribed circle center of circle.Area grayscale Drawn according to load, region is more black to represent that the sensor load for being responsible for the region is larger, otherwise less.Fig. 2 B () is step 2) as a result, it is apparent that the wherein radius of sensor all very littles, but the load of the sensor in high-density region It is very big, it is possible to overload.So passing through step 3) and step 4) being optimized to result 2).
The step 3) concretely comprise the following steps：
31) for each sensor pi, density function ρ is calculated in its correspondence domination region ΩiOn integration, it is negative as which Carry Ci, i.e.,：ρ (x) represents the probability that event occurs on monitored area, and x represents any on monitored area One point, i.e. integration variable；
32) by the load of all the sensorsArranged according to descending, then chosen the two of head of the queue and tail of the queue every time Individual sensor is processed, and the load of two sensors is set to CmaxAnd Cmin, temporary variable is Ct.If Cmax≥C0If, Determine C'max=C0, Ct+=Cmax-C0；If Cmin+Ct≥C0, set C'min=C0, Ct=Cmin+Ct-C0, otherwise set C'min+= Ct, Ct=0.The load of output all the sensors is limitedC'maxAnd C'minTwo sensors for being expressed as selecting every time set Fixed new load value, will optimize region division and sensing station so that the load of each sensor reaches in step 4 The target that this step is arranged；
Because the load allocation result by sensor in 2) result that step is obtained is very uneven, the biography of high-density region Sensor load is very big, causes sensor overload, and the sensor load very little of density regions, sensor performance cannot be abundant Utilize.In order to ensure all the sensors all non-overloadings, need to limit C according to sensor maximum load0All of sensor is entered Row load is redistributed.The present load of each sensor is calculated first：According to the knowledge of calculus, each sensor it is negative The integration arranged on region in sensor for density function is carried, i.e.,And the load sum of all the sensors Equal to integration of the density function on whole monitored area, i.e.,：Then according to method 32), to load More than C0Sensor, its load is set for C1(C1=C0)；For less sensor is loaded, which is loaded carry out it is certain Supplement and (be less than C0).A sensor is obtained finally for each sensor to limitNext step optimization will be caused Each sensor piLoad be equal to Ci'。
The step 4) concretely comprise the following steps：
41) object function isGiven for one Divide, object function is with regard to weightFunction, W represents the weight set of all the sensors, i.e.,X tables Show integration variable；
42) using Newton method minimization object function, comprise the following steps that：
421) gradient descent direction d is solved, is allowed to meet ▽2F (W) d=▽ F (W)；
422) determine decline step-length α for meeting Armijo conditions so that F (W+ α d) value declines；
423) displacement of a α d, i.e. ω are carried out to the weight of sensing stationi'←ωi+αdi；
424) according to new sensor weight, calculate new energy diagram, when object function with regard toGradient ▽ F (ω) during ＞ ε, return to step 421), otherwise output transducer position and region division；Wherein, ε is convergence Rule of judgment, with Areal concentration is related,
By the 3) step, we are provided with a load and limit to each sensor, that is, require Ci=Ci', i= 1,...,n(1).On the monitored area with density, there are many methods find the division for meeting that load limits (1).For Occur to be responded during event as soon as possible during monitored area can be made, in the present invention, take Optimal transport methods. Optimal transport methods are initially proposed by Gaspard Monge.Give a region, be paved with above sand and Have some websites, Optimal transport initial definition refer in order to by these sands (can only be moved every time) all The method for moving to mobile minimum range on website.Optimal transport are applied in our problem and are just to solve for The minima of following object functions：
Minimize
Subject to:
Middle W represents the weight set of all the sensors, i.e.,X is represented Integration variable.
Above-mentioned optimization is a kind of optimization problem limited with inequality.A kind of method for solving the problem is to utilize Lagrange multiplier methods.Using Lagrange multiplier methods, above-mentioned optimization problem can be converted into：
Minimize
Wherein W represents the weight set of all the sensors, i.e.,λ represents Lagrange multiplier set, i.e.,H'(W, λ) it is with regard to weightWith Lagrange multipliersThe object function of two class variables. In order to the further optimization for simplifying object function, we are by H'(W, λ) carried out further simplification, be reduced to containing only There is weightObject function F (W):
Minimize
F (W) is with regard to weightFunction, to try to achieve the minima of F (W), we only need to try to achieve function F's (W) Stationary point.Weight W minimum in order to try to achieve the value for making F (W)opt, we employ Newton method.In the process of each Newton iteration In, we are required for solving a following linear system：
▽2F (W) d=▽ F (W)
▽2F (W) represents the Hessian matrixes of object function F (W), and ▽ F (W) represent object function F (W) with regard to weight Gradient.
▽ F (W)=(C1'-C1,C'2-C2,...,C'n-Cn)；
F (W) is equal to its Laplacian matrix Δ with regard to the Hessian matrixes of weightwNegative value.I.e.：Wherein eijRepresent the line segment of two sensors i and j of connection, eij *Represent and eijThe side (through border cuts) of the energy diagram of antithesis.
After F (W) is tried to achieve with regard to the Hessian matrixes and gradient of weight, we tried to achieve using Eigen function libraries with regard to The object function descent direction d of weight.After trying to achieve descent direction d, by the method for linear search, finding to make target letter Number meets decline step-length α of Armijo conditions.
Armijo conditions are a kind of stop conditions of linear search.For given object function F (W), Armijo is met Decline step-length α of condition refers to that enabling to F (W) declines enough step-lengths.This condition can be described to lower inequality：
F(W+αd)≤F(W)+t·α·▽F(W)·d
Wherein t is constant, and t ∈ (0,1), t=10 in the present invention-4That is the decline of object function will and be walked Long, descent direction is into certain ratio.
Try to achieve after meeting step-length α of Armijo conditions, ω is updated to weighti'←ωi+αdi, and calculate new energy The gradient ▽ F (W) of spirogram and object function with regard to weight.Judge whether >=ε sets up | | ▽ F (W) | |, proceed to if setting up 421)；Step 5 is entered otherwise).If Fig. 3 (a) is given sensing station and division, the black and white degree in region illustrates biography The load of sensor.If limit the load of each sensor be set to：It is each sensing All, Fig. 3 (b) is to optimize the division result for obtaining, and the color in each region is identical, represents each sensor for the load of device Load it is all identical.
The step 5) concretely comprise the following steps：
51) calculate each sensor piDomination region ΩiMinimum circumscribed circle center of circle ciAnd radius ri；
If 52)By the position p of each sensoriIt is moved to the corresponding minimum circumscribed circle center of circle ciAnd go to step 3), the otherwise position of output transducerRadiusAnd region division
After the division that load is limited is met in the 4) step, for current division, in order to reduce covering for maximum All of sensing station is all moved to the minimum circumscribed circle center of circle in correspondence domination region, i.e., by lid radius：pi←ci, i= 1,...,n
Fig. 4 illustrates the present invention (C under different loads are limited0It is different) final deployment result.Monitoring section in Fig. 4 Domain and areal concentration are that, shown in Fig. 1 (a), initial sensor is distributed as shown in Fig. 1 (b).The maximum load limit of Fig. 4 (a)-(f) C processed0It is respectively：Sensor maximal cover radius is respectively： 0.247,0.214,0.199,0.183,0.166,0.163.The results show, our dispositions method enable to sensor Deployment result embodies monitored area density：Sensor distribution positioned at high-density region is more loaded, and density regions Sensor distribute less load.In actual applications, we can select suitable maximum load limit as the case may be C processed0。
Although the above-mentioned accompanying drawing that combines is described to the specific embodiment of the present invention, not to present invention protection model The restriction enclosed, on the basis of technical scheme, those skilled in the art are done by need not paying creative work The various modifications for going out or deformation are still within protection scope of the present invention.

Claims (7)

1., towards the wireless sensor network automatic deployment method of non-homogeneous sensing region, it is characterized in that, comprise the following steps：
Step one：It is according to given monitored area, number of sensors and sensor maximum load, random to generate sensor initial bit Put；
Step 2：The minimum circumscribed circle heart of the position of movable sensor to its corresponding domination region is until convergence；
Step 3：According to the load restriction and the present load of sensor of sensor, the load request of setting sensor is every Individual sensor distribution load；
Step 4：Optimize the weight of each sensor, find the division of the load request set in meeting step 3；
Step 5：Judge that all the sensors current location is moved to its correspondence minimum circumscribed circle heart needs mobile average departure From if sensing station is moved to the correspondence minimum circumscribed circle heart, and goes to step more than its condition of convergence by average distance Three；The otherwise division in output transducer position, radius and region；
Concretely comprising the following steps in the step 2：
21) by each sensor piWeights omegai0 is set to, and constructs sensing stationEnergy diagram；
22) calculate each sensor piDomination region Ωi；
23) calculate each sensor piArranged region ΩiMinimum circumscribed circle OiThe center of circle be ci, radius is ri；
24) calculate the distance that each sensor is moved to the heart movement of correspondence minimum circumscribed circleIfBy each sensing station piIt is moved to correspondence minimum circumscribed circle heart ci；Otherwise output transducer positionAnd energy diagram is dividedσ is threshold value；
The step 3 is concretely comprised the following steps：
31) for each sensor pi, density function ρ (x) is calculated in its correspondence domination region ΩiOn integration, as its load Ci, i.e.,：ρ (x) represents the probability that event occurs on monitored area, any one point on x monitored areas；
32) by the load of all the sensorsArranged according to descending, then chosen two biographies of head of the queue and tail of the queue every time Sensor is processed, and the load of two sensors is set to CmaxAnd Cmin, temporary variable is CtIf, Cmax≥C0, setting C'max=C0, Ct+=Cmax-C0；If Cmin+Ct≥C0, set C'min=C0, Ct=Cmin+Ct-C0, otherwise set C'min+= Ct, Ct=0, the load for exporting all the sensors is limitedC'maxAnd C'minTwo sensors for being expressed as selecting every time set Fixed load optimized target；
The step 4 is concretely comprised the following steps：
41) object function isThe division given for one, Object function is weight setFunction, W represents sensor weight set, i.e.,X represents integration variable；
42) using Newton method minimization object function；
N is the integer more than 1.
2. as claimed in claim 1 towards the wireless sensor network automatic deployment method of non-homogeneous sensing region, its feature It is that the sensor initial position collection in the step one is combined intoWherein n is the integer more than 1, and monitored area is Density function for ρ in Ω, the event on the Ω of region, ρ >=0 there is, sensor piPresent load be Ci, maximum load is C0。
3. as claimed in claim 1 towards the wireless sensor network automatic deployment method of non-homogeneous sensing region, its feature It is that the construction process of the energy diagram is：OrderFor RmA scatterplot collection in space, each point p in PiIt is endowed One weights ωi>=0, RmAny point p in space in any point p to scatterplot collection PiEnergy distance definition be：
dω(p,pi)=| | p-pi||2-ωi
It is criterion to space R with energy distancemDivided, defined V (pi) be and scatterplot piThe region of association, has：
V ( p i ) = { p ∈ R m | d ω ( p , p i ) ≤ d ω ( p , p j ) , ∀ p j ∈ P }
pjRepresent that scatterplot is concentrated except piAny point；
Define V (pi), i=1 ..., the collection of n is combined into the energy diagram of scatterplot collection P.
4. as claimed in claim 1 towards the wireless sensor network automatic deployment method of non-homogeneous sensing region, its feature It is that the σ threshold values value is 10-6。
5. as claimed in claim 1 towards the wireless sensor network automatic deployment method of non-homogeneous sensing region, its feature It is, the step 42) using Newton method minimization object function, detailed process is as follows：
421) gradient descent direction d is solved, is allowed to meet ▽2F (W) d=- ▽ F (W)；
422) determine step-length α for meeting Armijo conditions to meet the decline of F (W+ α d) value；
423) displacement of a α d, i.e. ω ' are carried out to the weight of sensing stationi←ωi+αd；
424) according to new sensor weight, calculate new energy diagram, when object function with regard toGradient ▽ F (W) >= During ε, return to step 421), otherwise output transducer position and region division；Wherein, ε is convergence Rule of judgment, and the ε is Convergence Rule of judgment is related to areal concentration,
6. as claimed in claim 1 towards the wireless sensor network automatic deployment method of non-homogeneous sensing region, its feature It is that the step 5 is concretely comprised the following steps：
51) calculate the minimum circumscribed circle center of circle c that each sensor arranges regioniAnd radius ri；
If 52)By the position p of each sensoriIt is moved to corresponding minimum circumscribed circle center of circle ciAnd turn To step 3, the otherwise position of output transducerRadiusAnd region divisionζ is the condition of convergence.
7. as claimed in claim 6 towards the wireless sensor network automatic deployment method of non-homogeneous sensing region, its feature It is that the ζ is the condition of convergence, its value is 10-5。
CN201410182101.5A 2014-04-30 2014-04-30 Towards the wireless sensor network automatic deployment method of non-homogeneous sensing region Active CN103945401B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN201410182101.5A CN103945401B (en) 2014-04-30 2014-04-30 Towards the wireless sensor network automatic deployment method of non-homogeneous sensing region

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN201410182101.5A CN103945401B (en) 2014-04-30 2014-04-30 Towards the wireless sensor network automatic deployment method of non-homogeneous sensing region

Publications (2)

Publication Number Publication Date
CN103945401A CN103945401A (en) 2014-07-23
CN103945401B true CN103945401B (en) 2017-03-29

Family

ID=51192833

Family Applications (1)

Application Number Title Priority Date Filing Date
CN201410182101.5A Active CN103945401B (en) 2014-04-30 2014-04-30 Towards the wireless sensor network automatic deployment method of non-homogeneous sensing region

Country Status (1)

Country Link
CN (1) CN103945401B (en)

Families Citing this family (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN107389318B (en) * 2017-08-01 2020-12-01 宿迁市创盈知识产权服务有限公司 LED light source detection method and system
CN109561498B (en) * 2018-12-04 2020-07-10 桂林电子科技大学 Sensor node distributed positioning method based on improved Newton method
CN111132182B (en) * 2019-12-16 2022-08-23 北京农业信息技术研究中心 Method and system for enhancing wireless multimedia network coverage

Citations (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN102014398A (en) * 2010-09-21 2011-04-13 上海大学 Optimal deployment method of large-scale industrial wireless sensor network based on differential evolution algorithm
CN102647726A (en) * 2012-02-17 2012-08-22 无锡英臻科技有限公司 Balancing optimizing strategy for energy consumption of coverage of wireless sensor network
CN103200581A (en) * 2013-03-08 2013-07-10 山东大学 Method and device of wireless sensor network interaction deployment based on three-dimensional scene

Patent Citations (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN102014398A (en) * 2010-09-21 2011-04-13 上海大学 Optimal deployment method of large-scale industrial wireless sensor network based on differential evolution algorithm
CN102647726A (en) * 2012-02-17 2012-08-22 无锡英臻科技有限公司 Balancing optimizing strategy for energy consumption of coverage of wireless sensor network
CN103200581A (en) * 2013-03-08 2013-07-10 山东大学 Method and device of wireless sensor network interaction deployment based on three-dimensional scene

Non-Patent Citations (4)

* Cited by examiner, † Cited by third party
Title
Coverage Optimization and Spatial Load Balancing by Robotic Sensor Networks;Jorge Cortes;《IEEE TRANSACTIONS ON AUTOMATIC CONTROL》;20100331;全文 *
Distributed Deployment Algorithms for Improved Coverage in a Network of Wireless Mobile Sensors;Hamid Mahboubi等;《IEEE TRANSACTIONS ON INDUSTRIAL INFORMATICS》;20140228;全文 *
Distributed Deployment Strategies for Improved Coverage in a Network of Mobile Sensors With Prioritized Sensing Field;Hamid Mahboubi等;《IEEE TRANSACTIONS ON INDUSTRIAL INFORMATICS》;20130228;全文 *
Minimum-Cost Load-Balancing Partitons;Boris Aronov等;《Algorithmica》;20091231;全文 *

Also Published As

Publication number Publication date
CN103945401A (en) 2014-07-23

Similar Documents

Publication Publication Date Title
CN103945401B (en) Towards the wireless sensor network automatic deployment method of non-homogeneous sensing region
CN105260575B (en) A kind of deformation of the surrounding rock in tunnel prediction technique neural network based
CN104875203B (en) Generate the robot simulation device of the path of motion of robot
US20170351936A1 (en) Object detection with neural network
CN108012276B (en) Three-dimensional heterogeneous mobile sensor network autonomous deployment method based on weighted Voronoi diagram
Toda et al. Self-localization based on multiresolution map for remote control of multiple mobile robots
CN108009544A (en) Object detection method and device
CN105786014B (en) The control method and device of unmanned machine
US20220276650A1 (en) Methods for risk management for autonomous devices and related node
CN114638954B (en) Training method of point cloud segmentation model, point cloud data segmentation method and related device
KR102330055B1 (en) Method and system for detecting change of structure using drone
CN103856951A (en) Coverage hole eliminating method based on multi-sensor probability perceptual model
Kan et al. A cyber–physical system (CPS) for planning and monitoring mobile cranes on construction sites
CN115359360B (en) Power field operation scene detection method, system, equipment and storage medium
CN111474925A (en) Path planning method for irregular-shape mobile robot
CN113242562A (en) WSNs coverage enhancement method and system
Arivudainambi et al. Coverage and connectivity-based 3D wireless sensor deployment optimization
Wu et al. Optimizing GPS-guidance transit route for cable crane collision avoidance using artificial immune algorithm
CN111553404B (en) Self-adaptive dynamic planning method for communication network node
CN113645631A (en) Three-dimensional scene-oriented mobile directed sensor network coverage optimization method
CN105050097A (en) Optimization deployment method for image sensor network
Lee et al. Affective surveillance management in virtual emotion based smart complex infrastructure
Waskita et al. A performance study of anomaly detection using entropy method
Stipanović et al. Collision free coverage control with multiple agents
CN103796237A (en) Method for fast detecting uncovered areas of sensor network

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