CN109784465A - The effective dynamic coverage method of forest fire monitoring system node based on Internet of Things - Google Patents
The effective dynamic coverage method of forest fire monitoring system node based on Internet of Things Download PDFInfo
- Publication number
- CN109784465A CN109784465A CN201811570336.6A CN201811570336A CN109784465A CN 109784465 A CN109784465 A CN 109784465A CN 201811570336 A CN201811570336 A CN 201811570336A CN 109784465 A CN109784465 A CN 109784465A
- Authority
- CN
- China
- Prior art keywords
- forest fire
- vector
- fire monitoring
- node
- internet
- 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.)
- Withdrawn
Links
Abstract
The effective dynamic coverage method of the forest fire monitoring system node that the invention discloses a kind of based on Internet of Things, first setting parameter: population scale NP, scale factor F, crossover probability CR, maximum evolutionary generation Gmax；Then the initial population that population scale is NP is generated at random, seeks the fitness value of NP individual；Judge whether to meet stop criterion, DE mutation operation, DE crossover operation, DE selection operation are carried out to current population at individual；Local search algorithm SQI is executed again；Finally export the best result acquired.The present invention solves the problems, such as that the effective coverage cost of forest fire monitoring Internet of things node existing in the prior art is big, low efficiency.
Description
Technical field
The invention belongs to forest fire monitoring system technical fields, and in particular to a kind of forest fire prison based on Internet of Things
The effective dynamic coverage method of examining system node.
Background technique
With being constantly progressive for MEMS, sensor technology and the communication technology, Internet of Things and modern industrial production and
Connection in daily life is increasingly close.For disaster environment monitoring, forest fires early warning etc. on a large scale monitoring, Internet of Things
Coverage rate become measure Internet of Things monitoring region can normal range of operation important indicator.
Situations such as forest fire monitoring Internet of Things overlay strategy is according to monitoring environment or whether known region, can be divided into
Certainty covering and random covering.When monitoring in situation known to environment or region, Internet of Things covering problem is then relatively easy, needs
The problems such as carrying out node setting and route planning, this coverage mode are known as certainty covering.When monitoring environment or region are unknown
In the case where, Internet of things node covering problem then becomes more intractable, but such case often more tallies with the actual situation, in this feelings
The Internet of things node covering completed under condition is known as random covering.Forest fire monitoring Internet of Things cover according to the node of use whether
Static covering and dynamic coverage can be also divided into locomotivity.When Internet of things node by for the first time disposed after, no longer have
The ability of shift position, that is, its position in a network is fixed after disposing, then is known as static covering.When Internet of things node is first
After secondary deployment, still there is certain locomotivity, to improve its position in a network, to improve the network coverage, enhance
Network performance, this covering are then known as dynamic coverage.Dynamic coverage is covered compared to static state, due to the mobility of node, is made
Internet of Things in coverage rate, network transmission capacity, many aspects such as node life span all embody more significant advantage, day
Benefit becomes the focus of people's research.
The covering deployment of Internet of things node is generated along with the appearance of Internet of Things, is Internet of Things Overall Acquisition peripheral information
Necessary condition.The main research of covering deployment of Internet of things node such as how less number of nodes, is realized to target area
It is completely covered, or makes overlay area area maximum, and make network link and quality that there is reliable guarantee.Under some cases, it is
Internet of Things is realized to the abundant sensing capability of peripheral information, the deployment of node often takes the mode of redundancy to carry out, i.e., with guarantor
Premised on demonstrate,proving coverage goal region, carried out by the way of adjacent node overlapped coverage.Redundant cover realizes node and obtains
The target of peripheral information is taken, but also brings many problems, the peripheral information as node is obtained and transmitted repeats, and occupies network
Resource leads to the increase of nodes energy consumption, reduces the life span etc. of node in a network, therefore, in Internet of Things section
Unnecessary redundant cover should be reduced to the greatest extent when point deployment.
Summary of the invention
The object of the present invention is to provide a kind of effective dynamic coverage side of forest fire monitoring system node based on Internet of Things
Method, solves that the effective coverage cost of forest fire monitoring Internet of things node existing in the prior art is big, low efficiency asks
Topic.
The technical scheme adopted by the invention is that the effective dynamic coverage of forest fire monitoring system node based on Internet of Things
Method is specifically implemented according to the following steps:
Step 1, parameter setting: population scale NP, scale factor F, crossover probability CR, maximum evolutionary generation Gmax；
Step 2, the random population scale that generates seek the fitness value of NP individual for the initial population of NP；
Step 3 judges whether to meet stop criterion, if it is satisfied, then algorithm terminates, turns to step 6；Otherwise step is turned to
4；
Step 4 carries out DE mutation operation, DE crossover operation, DE selection operation to current population at individual；
Step 5 executes local search algorithm SQI；
The best result that step 6, output acquire.
The features of the present invention also characterized in that
Step 1 is specific as follows: setting population scale NP=40, scale factor F=0.5, crossover probability CR=0.9, maximum
Evolutionary generation Gmax=300.
Step 2 is specific as follows:
Step 2.1, in order to guarantee the node coordinate after forest fire monitoring node initializing entirely monitoring regional scope
Interior, that is, the object vector range initialized can cover entire solution space, and the object vector after initialization is expressed as shown in following formula:
Xji,G=Xj,min+randj,i(0,1)·(Xj,max-Xj,min)
In formula, rand (0,1) is the equally distributed random number of obedience that computer random generates in (0,1) section,
Xj,min={ X1,min,X2,min,…,XD,minIndicate to tie up the lower boundary of object vector in continuous Real-value space, i.e. forest fire in D
Monitor the lower boundary in region；Xj,max={ X1,max,X2,max,…,XD,maxIndicate to tie up object vector in continuous Real-value space in D
Coboundary, the i.e. coboundary in forest fire monitoring region；Xi,GIndicate i-th of individual vector or the object vector in G generation:
Xi,G=(X1i,G,X2i,G,Xji,G,…,XDi,G)
In formula, G=0,1 ..., Gmax, G indicates algebra belonging to the population, and G=0 indicates initialization population vector, GmaxFor
Maximum algebra, i=1,2 ..., NP indicate that i-th of individual vector, D indicate D dimension space, i.e., tie up continuous real number value parameter in the D
Space solves globally optimal solution；
Step 2.2, the fitness function for calculating the object vector after initialization, i.e., under current state, according to fitness
Function calculates the fitness function value of forest fire monitoring node, chooses forest fire monitoring node in the coverage rate in monitoring region
Function is as follows as the fitness function:
Step 3 is specific as follows: judging whether to meet forest fire monitoring according to the fitness function value that step 2 is calculated
The requirement of coverage rate turns to step 6 if it is satisfied, then algorithm terminates；Otherwise step 4 is turned to.
Step 4 is specific as follows:
Step 4.1, DE mutation operation: it after forest fire monitoring node coordinate completes initialization as object vector, adopts
With " DE/rand/1 " strategy generating variation vector, i.e. the variation vector V of forest fire monitoring node coordinatei,G=(V1i,G,V2i,G,
Vji,G,…,VDi,G):
Vi,G=Xr1,G+F·(Xr2,G-Xr3,G)
In formula, r1 ≠ r2 ≠ r3, and r1, r2, r3 are the positive integers that computer randomly selects from [1, NP], in Forest Fire
One variation vector of every generation in the mutation operation of calamity monitoring node, r1, r2, r3 can be generated once by computer at random, and
It is also not identical as object vector i, therefore the initial population number NP of forest fire monitoring node should indicate G generation not less than 4, G
Vector, F indicate that the scale factor of control forest fire monitoring node variation vector scaling, the scale factor are positive real constant, and F takes
Being worth range is F ∈ [0,2]；
In the dynamic coverage method of forest fire monitoring node, when it is desirable that make a variation vector population show diversity,
When avoiding falling into local optimum, F should take relative larger value；When being desired with local search, when realizing fast convergence, F should take phase
To smaller value；
Step 4.2, DE crossover operation: the diversity in order to guarantee node coordinate vector individual, by forest fire monitoring section
The variation vector V of point coordinate vectori,GWith the object vector X of forest fire monitoring node coordinate vectori,GThe parameter for being included, is adopted
The trial vector U of new forest fire monitoring node coordinate vector is generated with binomial intersectioni,G, Ui,G=(U1i,G,U2i,G,
Uji,G,…,UDi,G), Ui,GIn variation vector V of at least one parameter from forest fire monitoring node coordinate vectori,G,
In formula, CR is to intersect the factor, indicates real value crossover probability constant, CR ∈ [0,1], rand (0,1) are that computer exists
(0,1) the equally distributed random number of the obedience generated at random in range；K is used to guarantee that newly-generated forest fire monitoring node is sat
Mark the trial vector U of vectori,GIn variation vector V of at least one parameter from forest fire monitoring node coordinate vectori,G,
K value range is k ∈ [1, D], is chosen by computer random, CR value is bigger, in crossover operation in newly-generated experiment vector
From variation vector Vi,GParameter it is more；Conversely, CR value is smaller, from change in newly-generated experiment vector in crossover operation
Incorgruous amount Vi,GParameter it is fewer；
Step 4.3, DE selection operation: the new forest fire generated after vector variation and crossover operation before this is supervised
Survey the trial vector U of nodei,G, carry out the calculating of fitness value, and by calculated value and forest fire monitoring node coordinate vector
Object vector Xi,GThe value for carrying out fitness calculating is compared, thus under selecting in the two calculated result more preferably individual becoming
Generation individual vector:
In formula, f () indicates fitness function, i.e.,At this point, Xi,G+1As optimum point, f
(Xi,G+1) it is that the smallest point of fitness function value in NP point is similarly denoted as most not good enough w, the point pair by optimum point fitness value
The fitness function value answered is worst fitness value f (w).
Step 5 is specific as follows:
Step 5.1 randomly selects forest fire monitoring node coordinate vector x from NP point2And x3, x2=(x21,x22,
x2i,…x2D), x3=(x31,x32,x3i,…x3D) and meet x2≠x3≠Xi,G+1, enable x1=Xi,G+1；
Step 5.2, the new testing site p=(p for calculating forest fire monitoring node coordinate vector·1,p·2,p·j,…,pD)T,
Wherein
I=1,2 ... D；
Wherein, xj=(xj1,xj2,…xjD)T, j=1,2,3 ..., D indicate the testing site of forest fire monitoring D dimension space；f
() is the effective dynamic coverage adaptation of methods degree function of forest fire monitoring system node；
Step 5.3 calculates f (p), if fp> fbPoint p substitution point b is then used, as forest fire monitoring node coordinate vector
Next-generation individual vector, is not otherwise replaced.
Step 6 is specific as follows:
Enable forest fire monitoring node coordinate vector Xi,G+1=(X, Y), output forest fire monitoring node coordinate matrix (X,
Y) to get the optimal Node distribution coordinates matrix value for the dynamic coverage method for arriving forest fire monitoring node.
The invention has the advantages that based on the effective dynamic coverage side of forest fire monitoring system node based on Internet of Things
Method, this method propose the hybrid differential evolution algorithm based on quadratic interpolation on the basis of standard DE algorithm, supervise to forest fire
Survey Internet of things node, which effectively covers, to be optimized.In the case where not increasing other hardware devices of network, calculation is further reduced
The calculating cost of method improves the efficiency of node optimization deployment.Simulation result shows that this chapter algorithm has preferable node deployment
Effect.
Detailed description of the invention
Fig. 1 is the schematic diagram of region overlay；
Fig. 2 is a schematic diagram for covering；
Fig. 3 is fence covering schematic diagram；
Fig. 4 is grid covering schematic diagram；
Fig. 5 (a) is that closed curve region is initialization set in greedy covering；
Fig. 5 (b) is that closed curve region is Internet of things node set in greedy covering；
Fig. 6 (a) is schematic diagram when Internet of things node e and f carve normal work at the beginning；
Fig. 6 (b) is that Internet of things node e and f schematic diagram when blind spot situation occur；
Fig. 7 is worst case coverage diagram；
Fig. 8 (a) is ancestor node distribution map under DE optimization algorithm；
Fig. 8 (b) is the 100th generation node distribution map under DE optimization algorithm；
Fig. 8 (c) is the 200th generation node distribution map under DE optimization algorithm；
Fig. 8 (d) is the 300th generation node distribution map under DE optimization algorithm；
Fig. 8 (e) is to improve mesomere point diagram under DE optimization algorithm；
Fig. 8 (f) is to improve the 100th generation node distribution map under DE optimization algorithm；
Fig. 8 (g) is to improve the 200th generation node distribution map under DE optimization algorithm；
Fig. 8 (h) is to improve the 300th generation node distribution map under DE optimization algorithm；
Fig. 9 is that the coverage rate of two kinds of algorithms under different number of nodes compares figure；
Figure 10 is that the coverage rate of two kinds of algorithms under different the number of iterations compares figure.
Specific embodiment
The following describes the present invention in detail with reference to the accompanying drawings and specific embodiments.
According to Internet of Things to the difference in coverage goal region, usually Internet of things node can be covered deployment and be divided into: covered in region
Lid, point covering, fence cover three kinds.
Internet of Things region overlay is using a certain region as target, in the target area, with Internet of Things section as few as possible
It counts, each point in coverage goal region realizes target area covering to guarantee the reliable communication between regional nodes
It maximizes, reduces network construction cost and expense.Region overlay requires greatly covering surface, and quality requirement is high, is typically employed in weight
Point region, or to the occasion that monitoring quality has higher requirements.Fig. 1 is the schematic diagram of region overlay, to be completed to the region
Covering needs 6 Internet of things node.
Point covering, with Internet of things node number as few as possible, guarantees coverage goal point using a certain group of discrete point as target, with
Just in the case where target point is capped, the reliable communication of Internet of things node is realized.Point covering covers mesh compared to region overlay
Mark significantly reduces, and does not need to cover a certain region all the points, therefore, network construction is relatively easy, and cost is relatively low.It is applicable in
In the occasion relatively low to monitoring quality requirement.But when in target area, range is larger or large scale network deployment scenario,
Often available point covers to be indicated the covering of Internet of Things target area.The covering of Internet of Things target area is covered on research side with point
In method and research process, all exist many similar, difference is, point covering algorithm it should be understood that target point distributed intelligence and object
In addition to this network topology structure of networking, region overlay will further recognize that the information such as target area geometry.As shown in Fig. 2,
The round Internet of things node for indicating to need to dispose in figure is rectangular to indicate to need capped target point.As shown in Figure 2, it is needing
Three Internet of things node are disposed around capped target point, and the acquisition of information to target point can be realized.
Fence is covered using mobile target unknown in Internet of Things as monitoring object, when there is unknown mobile target to attempt to pass through
When Internet of things node deployment region, i.e., found by network monitor.In fence covering, situation about being not relevant in target area,
The maximum probability that the node deployment of Internet of Things is monitored to using unknown mobile target is target.Fence covering generally with strip,
Band-like, closing or the shapes such as semiclosed occur, and compared with region overlay, do not need all positions in overlay area, thus
Required number of nodes greatly reduces, more suitable for the occasion to mobile target monitoring.As shown in figure 3, curve indicates examination in figure
Movement track of the figure by the unknown mobile target of Internet of things node deployment region, the Internet of Things section that circle expression is disposed
Point.
To the level of coverage of target area after being disposed according to Internet of things node, Internet of things node covering algorithm can be divided into
Complete coverage algorithm and non-fully covering algorithm.In algorithm is completely covered, and two kinds of situations are divided into, a kind of situation is to work as
It is referred to as k- covering (k >=2) that all the points in target area are all covered by k sensor simultaneously, is otherwise 1- covering.It covers completely
Lid algorithm network construction cost is high, is generally used for emergency management and rescue, the environment that battlefield surroundings etc. have higher requirements to covering.It is non-complete
All standing can generally be monitored since interstitial content is relatively fewer in water quality of river, and atmosphere environment supervision etc. requires not covering
Too high environment.
According to the processing mode of Internet of things node covering algorithm, Internet of things node covering algorithm can be divided into centralized covering
Algorithm and Distributed coverage algorithm.In centralized covering algorithm, by the central node or monitoring center root being deployed in Internet of Things
According to the global information of network, after running certain specific covering algorithm, result is fed back into the node in network, i.e. covering algorithm
It does not calculate on each node.And in Distributed coverage algorithm, by each node by the local message of network, and saved with closing on
Point cooperation, is calculated using the hardware resource of itself and executes covering algorithm.
Whether connectivity is considered according to Internet of things node covering algorithm, and can be divided into ensures that the Internet of things node of network-in-dialing covers
Lid algorithm and the Internet of things node covering algorithm unrelated with network connectivty.Pertinent literature shows, it is assumed that all nodes of Internet of Things
Structure is identical, and when the sensing range of node is round, if the communication radius distance of node is greater than twice of the perception radius at this time
Apart from when, can be ensured of between Internet of things node interconnected.
According to the working method of Internet of Things interior joint, covering algorithm can be divided into, grouping Internet of things node covering algorithm and
Round Internet of things node covering algorithm.In grouping Internet of things node covering algorithm, it is small that all nodes are divided into several first
Group, node uses within the network life time, by successively scheduling as work group node, in this approach, in network life
Interior algorithm Exactly-once.And in round Internet of things node covering algorithm, algorithm is executed once within a certain period, then root
According to certain algorithm, selected section node is activated from Internet of things node, becomes working node use, other node dormancies, period
After, the method continued the above, in this approach, algorithm performs repeatedly in network life.
In the Internet of things node covering algorithm based on grid, Internet of things node and target coverage point are all made of grid-shaped
Formula indicates, and indicates whether target coverage point is covered by Internet of things node with Boolean Model, then each target coverage point is by object
The case where networked node covers can be represented as an array.
As shown in figure 4, target coverage point 8 can be represented as the case where covering by Internet of things node energy vectors (0,0,
1,1,0,0).Each target point can be no less than one Internet of things node covering in figure, therefore, to each target coverage
At least one is 1 in the array of point, i.e., realizes and be completely covered to target coverage point.But when hardware resource deficiency, Internet of Things
Net interstitial content is less, cannot achieve when being completely covered, then to carry out relevant node arrangement according to covering cost upper limit.This
When, grid Internet of things node covering problem can be converted to the minimization problem for solving distance mistake, to obtain Internet of Things section
The optimization covering method of point.
Greedy connection Internet of things node covering method
The algorithm belongs to communication path covering and certainty millet cake cover type in connectivity covering.
In greedy connection Internet of things node covering method, it is assumed that initialize the Internet of things node collection M that selects for comprising
One set of part Internet of things node.Other Internet of things node that there is the overlay area that partly overlaps with M in network, by it
Referred to as both candidate nodes.After algorithm starts, a random node set M is selected first, is then selected from the node of node set M
Larger range of path can be covered with both candidate nodes by selecting, and the M that initialization is added in the Internet of things node in this path be gathered, shape
At a new Internet of things node set M ', then algorithm continues previous step, until the target area of required covering or target
Point is completely covered by node set M '.
It is illustrated in figure 5 the operational process figure of greedy connection Internet of things node covering method, in Fig. 5 (a), closed curve
Region is initialization set M, algorithm is run start after, after randomly choosing set M, bigger overlay area in order to obtain, than
Path b is obtained compared with selection, to obtain the Internet of things node set M ' in Fig. 5 (b).
Rotation is active/covering method of suspend mode Internet of things node
Rotation it is active/covering method of suspend mode Internet of things node in, algorithm can by the suspend mode part within a certain period
The Internet of things node substituted, to extend the life span of entire Internet of Things.After algorithm starts, first by each Internet of things node
Peripherad all neighbor nodes send broadcast message, inform own location information and id information, then each Internet of things node root
According to the broadcast message of the neighbor node received, check that the overlay area of itself could be covered by other neighbor nodes, if it can,
State advertisement message is then sent, then enters Internet of things node suspend mode, other Internet of things node then work on mould
Formula.
This method passes through the Internet of things node suspend mode that substitute part can, when substantially prolonging the existence of entire Internet of Things
Between.But this mechanism has a problem that, if that is, when the Internet of Things neighbor node of surrounding finds the overlay area of itself simultaneously
It can be covered by other neighbor nodes, and after then entering Internet of things node suspend mode together, it is originally capped to will lead to part
Region there is blind area or blind spot.
As shown in Fig. 6 (a), Fig. 6 (b), when Internet of things node e and f are carved at the beginning, the region covered can be all
The Internet of things node a, b, c enclosed, d is covered, when Internet of things node e and f subsequently enter suspend mode, then capped in script
Region in there is the as shown in the figure blind area that cannot be capped or blind spot, i.e. shadow region in Fig. 6 (b).In order to effective
Prevent the generation of this phenomenon, can Internet of things node check itself overlay area could by other neighbor nodes cover before,
A random-backoff time is added, Internet of things node is accordingly checked again after this random-backoff time, and can in part
Before the node substituted enters suspend mode, wait for a period of time at random come the state for the neighbor node checked in Internet of Things, with
This is avoided the appearance of blind area in as shown in the figure.
Worst case Internet of things node covering method
Worst case Internet of things node covering method, belongs to fence cover type, can also be divided into certainty Internet of Things
Path/region overlay.Algorithm is using mobile target unknown in Internet of Things as monitoring object, when there is unknown mobile target to attempt to lead to
It when crossing Internet of things node deployment region, i.e., is found by network monitor, and by the Internet of things node capturing information in passage path.It calculates
Path is broken through by defining the maximum of unknown mobile target in method, to indicate that the unknown mobile target is not monitored by Internet of Things
The probability minimum arrived, the i.e. the worst situation of Internet of Things network.The maximum path of breaking through of unknown mobile target is by Vornoi
Line segment in figure is constructed.As shown in fig. 7, the maximum path of breaking through from S to D is each side of Delaunay triangle in figure
The dotted line with the arrow that perpendicular bisector is constituted.
Covering problem is a major issue of forest fire monitoring Internet of Things.Under a wide range of environment based on Internet of Things
Forest fire intelligent monitor system in, shell projects forest fire monitoring node or the mode broadcasted sowing of aircraft carries out head taking
When secondary random placement, part of node is the node with locomotivity that caterpillar robot or intelligent carriage carry, other
Node is ordinary node, in favor of realizing the optimization covering of forest fire monitoring node by the method for dynamic adjustment.
This problem is described the present invention by such as drag.It is assumed that there are Target monitoring area A, and it is one two
Target monitoring area A is divided into m × n mesh point by grid by dimensional plane, meanwhile, the target monitoring area can be covered
Forest fire monitoring Internet of things node on domain, with set expression:
S={ s1,s2,…,sn}
In formula, in forest fire monitoring Internet of things system node set S, si={ xi,yi,riBe used to describe Forest Fire
Calamity monitors a certain node s in Internet of things systemiOverlay model, (xi,yi) indicate system node siSeat in a two-dimensional plane
Mark, riFor Internet of things node siThe perception radius.Here r is definedcFor node siCommunication radius, then there is rc≥2ri, to protect
The network that card forest fire monitoring Internet of things system node is formed after deployment is the network that can be interconnected, and wherein
The location information of all forest fire monitoring Internet of things node all can accurately be obtained by the location algorithm in this paper chapter 3.
For any grid point p in two-dimensional surface in forest fire monitoring target coverage areaj, with forest fire monitoring
Internet of things node siBetween Euclidean distance, indicated using formula are as follows:
Here two-valued function mould is used to the coverage condition of any grid point in forest fire monitoring target coverage area
Type indicates, i.e., as forest fire monitoring Internet of things node siWith grid point pjEuclidean distance d (si,pj) in node siCovering
When in range, i.e., its value is less than node siThe perception radius riWhen, then it represents that the grid point in two-dimensional surface in target coverage area
pj, can be by node siCovering, i.e. coverage function value are 1, are otherwise 0, are indicated by formula are as follows:
The network coverage of forest fire monitoring Internet of things node is represented by target coverage area, is supervised by forest fire
The region area of Internet of things node covering and the ratio of entire Target monitoring area area are surveyed, is indicated using mathematical formulae are as follows:
In formula, cb(pj) indicate all Internet of things node in two-dimensional surface in target coverage area in grid point pjPlace takes
The total coverage function value obtained.The formula is also as the subsequent fitness function used in text.
Therefore, fitness function value is bigger, and coverage rate is bigger, and the coverage effect of forest fire monitoring network is better.
By calculating the maximized optimal solution of fitness function, the location estimation of optimization posterior nodal point can be obtained, improve entire Forest Fire
The coverage rate of calamity monitoring Internet of things node.
Genetic evolution of the essential idea of differential evolution algorithm in biotic population, and simulate this process and produce
It is raw, it mainly include fitness function, algorithm operating, three parts of algorithm parameter setting in algorithm.Wherein, fitness
Evaluation function is the foundation and direction that difference algorithm specifically executes, and reflects the final goal of algorithm genetic evolution, is for commenting
Whether valence particular problem has obtained the pointer of optimal solution, and the maximum or minimum of a certain function are expressed as generally according to practical problem
Value.Algorithm operating makes the mathematical operations process inside difference algorithm, does not need to introduce a certain problem others information, detailed process
It is similar to other bionic Algorithms, it such as makes a variation, intersects, the processes such as selection, difference is that the concrete operations sequence of algorithm is variant.
DE algorithm is a kind of simple and direct and strong search capability evolution algorithm.Standard test functions and Practical Project are asked
Topic, DE algorithm are better than other evolution algorithms with the solution of fast convergence rate and high quality.Although compared with other evolution algorithms, DE
Algorithm has relatively good performance, but some complicated optimum problems for engineering in practice, DE algorithm can show to solve
The deficiencies of of low quality or convergence rate is slow.The reason of causing this result is that DE algorithm cannot be effectively utilized and acquired
Objective function information.
For forest fire monitoring Internet of things node coverage optimization problem, it is mixed based on quadratic interpolation that this paper presents a kind of
Close differential evolution algorithm.The algorithm is to be embedded under the frame of DE algorithm and simplify quadratic interpolation (SQI) method to improve original DE
The overall performance of algorithm.SQI method it is intended that improve algorithm local search ability, reduce the calculating cost of algorithm.
Simplified quadratic interpolattion is a kind of searching method of curve matching, this method have it is easy to use, without former letter
The characteristics of several derivative informations, operation difficulty is greatly reduced, arithmetic speed is improved, when being used as heuristic search operator
With preferable performance.Herein as local searching operator, it is embedded into DE algorithm, accelerates the convergence speed of standard DE algorithm
Degree, improves the precision of solution.
The present invention is based on the effective dynamic coverage methods of the forest fire monitoring system node of Internet of Things, specifically according to following step
It is rapid to implement:
Step 1, parameter setting: population scale NP, scale factor F, crossover probability CR, maximum evolutionary generation Gmax, wherein
Population scale NP=40, scale factor F=0.5, crossover probability CR=0.9, maximum evolutionary generation Gmax=are set
300；
Step 2, the random population scale that generates seek the fitness value of NP individual for the initial population of NP, specific as follows:
Step 2.1, in order to guarantee the node coordinate after forest fire monitoring node initializing entirely monitoring regional scope
Interior, that is, the object vector range initialized can cover entire solution space, and the object vector after initialization is expressed as shown in following formula:
Xji,G=Xj,min+randj,i(0,1)·(Xj,max-Xj,min)
In formula, rand (0,1) is the equally distributed random number of obedience that computer random generates in (0,1) section,
Xj,min={ X1,min,X2,min,…,XD,minIndicate to tie up the lower boundary of object vector in continuous Real-value space, i.e. forest fire in D
Monitor the lower boundary in region；Xj,max={ X1,max,X2,max,…,XD,maxIndicate to tie up object vector in continuous Real-value space in D
Coboundary, the i.e. coboundary in forest fire monitoring region；Xi,GIndicate i-th of individual vector or the object vector in G generation:
Xi,G=(X1i,G,X2i,G,Xji,G,…,XDi,G)
In formula, G=0,1 ..., Gmax, G indicates algebra belonging to the population, and G=0 indicates initialization population vector, GmaxFor
Maximum algebra, i=1,2 ..., NP indicate that i-th of individual vector, D indicate D dimension space, i.e., tie up continuous real number value parameter in the D
Space solves globally optimal solution；
Step 2.2, the fitness function for calculating the object vector after initialization, i.e., under current state, according to fitness
Function calculates the fitness function value of forest fire monitoring node, chooses forest fire monitoring node in the coverage rate in monitoring region
Function is as follows as the fitness function:
Step 3 judges whether to meet stop criterion, if it is satisfied, then algorithm terminates, turns to step 6；Otherwise step is turned to
4, it is specific as follows: to be judged whether to meet wanting for forest fire monitoring coverage rate according to the fitness function value that step 2 is calculated
It asks, if it is satisfied, then algorithm terminates, turns to step 6；Otherwise step 4 is turned to；
Step 4 carries out DE mutation operation, DE crossover operation, DE selection operation to current population at individual, specific as follows:
Step 4.1, DE mutation operation: it after forest fire monitoring node coordinate completes initialization as object vector, adopts
With " DE/rand/1 " strategy generating variation vector, i.e. the variation vector V of forest fire monitoring node coordinatei,G=(V1i,G,V2i,G,
Vji,G,…,VDi,G):
Vi,G=Xr1,G+F·(Xr2,G-Xr3,G)
In formula, r1 ≠ r2 ≠ r3, and r1, r2, r3 are the positive integers that computer randomly selects from [1, NP], in Forest Fire
One variation vector of every generation in the mutation operation of calamity monitoring node, r1, r2, r3 can be generated once by computer at random, and
It is also not identical as object vector i, therefore the initial population number NP of forest fire monitoring node should indicate G generation not less than 4, G
Vector, F indicate that the scale factor of control forest fire monitoring node variation vector scaling, the scale factor are positive real constant, and F takes
Being worth range is F ∈ [0,2]；
In the dynamic coverage method of forest fire monitoring node, when it is desirable that make a variation vector population show diversity,
When avoiding falling into local optimum, F should take relative larger value；When being desired with local search, when realizing fast convergence, F should take phase
To smaller value；
Step 4.2, DE crossover operation: the diversity in order to guarantee node coordinate vector individual, by forest fire monitoring section
The variation vector V of point coordinate vectori,GWith the object vector X of forest fire monitoring node coordinate vectori,GThe parameter for being included, is adopted
The trial vector U of new forest fire monitoring node coordinate vector is generated with binomial intersectioni,G, Ui,G=(U1i,G,U2i,G,
Uji,G,…,UDi,G), Ui,GIn variation vector V of at least one parameter from forest fire monitoring node coordinate vectori,G,
In formula, CR is to intersect the factor, indicates real value crossover probability constant, CR ∈ [0,1], rand (0,1) are that computer exists
(0,1) the equally distributed random number of the obedience generated at random in range；K is used to guarantee that newly-generated forest fire monitoring node is sat
Mark the trial vector U of vectori,GIn variation vector V of at least one parameter from forest fire monitoring node coordinate vectori,G,
K value range is k ∈ [1, D], is chosen by computer random, CR value is bigger, in crossover operation in newly-generated experiment vector
From variation vector Vi,GParameter it is more；Conversely, CR value is smaller, from change in newly-generated experiment vector in crossover operation
Incorgruous amount Vi,GParameter it is fewer；
Step 4.3, DE selection operation: the new forest fire generated after vector variation and crossover operation before this is supervised
Survey the trial vector U of nodei,G, carry out the calculating of fitness value, and by calculated value and forest fire monitoring node coordinate vector
Object vector Xi,GThe value for carrying out fitness calculating is compared, thus under selecting in the two calculated result more preferably individual becoming
Generation individual vector:
In formula, f () indicates fitness function, i.e.,At this point, Xi,G+1As optimum point, f
(Xi,G+1) it is that the smallest point of fitness function value in NP point is similarly denoted as most not good enough w, the point pair by optimum point fitness value
The fitness function value answered is worst fitness value f (w)；
Step 5 executes local search algorithm SQI, specific as follows:
Step 5.1 randomly selects forest fire monitoring node coordinate vector x from NP point2And x3, x2=(x21,x22,
x2i,…x2D), x3=(x31,x32,x3i,…x3D) and meet x2≠x3≠Xi,G+1, enable x1=Xi,G+1；
Step 5.2, the new testing site p=(p for calculating forest fire monitoring node coordinate vector·1,p2,p·j,…,p.D)T,
Wherein
I=1,2 ... D；
Wherein, xj=(xj1,xj2,…xjD)T, j=1,2,3 ..., D indicate the testing site of forest fire monitoring D dimension space；f
() is the effective dynamic coverage adaptation of methods degree function of forest fire monitoring system node；
Step 5.3 calculates f (p), if fp> fbPoint p substitution point b is then used, as forest fire monitoring node coordinate vector
Next-generation individual vector, is not otherwise replaced；
The best result that step 6, output acquire, specific as follows:
Enable forest fire monitoring node coordinate vector Xi,G+1=(X, Y), output forest fire monitoring node coordinate matrix (X,
Y) to get the optimal Node distribution coordinates matrix value for the dynamic coverage method for arriving forest fire monitoring node.
Simulating, verifying and analysis:
Simulation parameter setting:
The present invention uses 7.0 environment of MATLAB, carries out the analog simulation verifying of forest fire monitoring node, the kind of setting
Group number of individuals NP is 40, and maximum evolutionary generation is 300, scaling factor F=0.5, crossover probability CR=0.9.Setting sensing
The effectively perceive radius of device node is 12m, in the simulating forest fire monitoring region of 100m × 100m, to 30 forest fires
The distribution of monitoring mobile sensor node is emulated.As a result as shown in Fig. 8 (a)~Fig. 8 (h).
Fig. 8 (a)~Fig. 8 (d) and Fig. 8 (e)~Fig. 8 (h) is shown respectively in standard difference evolution algorithm and after improving
Two kinds of differential evolution algorithm in the case of, simulate 30 forest fire monitoring Internet of things node covering dynamic changing process.By
Figure can be seen, and carve at the beginning, the random distribution of forest fire monitoring Internet of things node is simultaneously uneven, a part of region Forest Fire
Calamity monitoring Internet of things node is more intensive, and another part region forest fire monitoring Internet of things node is then more loose, so that whole
A Target monitoring area shows part blind area.Algorithm is after operation iteration 100 times, 200 times, 300 times, forest fire monitoring object
Networked node is distributed the more reasonable of performance Target monitoring area, and most Target monitoring area is supervised by forest fire
Internet of things node is surveyed to cover.The Node distribution area coverage that improved algorithm obtains is bigger, and final operation result compares original error
Divide the operation result of evolution algorithm more preferable.
Simulation analysis:
Under different forest fire monitoring Internet of things node numbers and different the number of iterations, observation is based on standard respectively
Fig. 9 and figure can be obtained by MATLAB emulation in the coverage rate of the forest fire monitoring node of optimization algorithm after DE algorithm and improvement
10。
(1) in different number of nodes, if maximum evolutionary generation is 120 times, forest fire monitoring node it is effective
The perception radius is respectively 12m and 8m, other simulation parameters are the same, and emulation can obtain Fig. 9.
As can be seen from Figure 9, under identical forest fire monitoring coverage rate, the improved distribution optimization algorithm of this chapter compares standard
The forest fire monitoring number of nodes that DE algorithm needs is few.When forest fire monitoring node effectively perceive radius is 12m, the number of iterations
It is 120 times, when coverage rate requires to reach 95% or more, improved DE optimization algorithm is than the Forest Fire that primary standard DE algorithm needs
Calamity monitoring node number is 10 few；When forest fire monitoring node effectively perceive radius is 8m, the number of iterations is 120 times, coverage rate
It is required that when reaching 80% or more, improved DE optimization algorithm is fewer than the forest fire monitoring number of nodes that primary standard DE algorithm needs
13.The improved DE optimization algorithm of this chapter compares primary standard DE algorithm, in different forest fire monitoring node communication radius
In the case of number of nodes, better coverage rate can get.
(2) in different the number of iterations, if forest fire monitoring number of nodes is 30, the effectively perceive half of node
Diameter is respectively 12m and 8m, other simulation parameters are the same, and emulation can obtain Figure 10.
As can be seen from Figure 10, when forest fire monitoring Internet of things node number is 30, and effectively perceive radius is 12m, this chapter
Improved DE optimization algorithm tends to restrain after iteration 200 times, and standard DE optimization algorithm tends to restrain after iteration 400 times；
When forest fire monitoring Internet of things node number is 30, and effectively perceive radius is 8m, the improved DE optimization algorithm of this chapter exists
Tend to restrain after iteration 320 times, standard DE optimization algorithm tends to restrain after iteration 420 times；The improved DE distribution of this chapter is excellent
Change algorithm faster than standard DE distribution optimization algorithm the convergence speed, forest fire monitoring coverage rate is higher.
Forest fire monitoring Internet of things node coverage optimization is a major issue in forest fire monitoring system, and
The Research foundation of forest fire monitoring contingency question.The covering of forest fire monitoring Internet of things node is optimized, can more be closed
Reason ground distribution Internet resources, are all of crucial importance environment sensing and acquisition of information precision etc..This chapter proposes one
Forest fire monitoring Internet of things node effective covering method of the kind based on improved differential evolution, base of this method in standard DE algorithm
On plinth, it is excellent effectively to cover progress to forest fire monitoring Internet of Things to propose the hybrid differential evolution algorithm based on quadratic interpolation
Change.Simulation result shows that the improvement of algorithm improves the local search ability of former algorithm, reduces the calculating cost of algorithm, adds
Speed standard DE convergence speed of the algorithm, improves the precision of understanding, answers the node optimization deployment of forest fire monitoring Internet of Things
With with certain reference value.
Claims (7)
1. the effective dynamic coverage method of forest fire monitoring system node based on Internet of Things, which is characterized in that specifically according to
Lower step is implemented:
Step 1, parameter setting: population scale NP, scale factor F, crossover probability CR, maximum evolutionary generation Gmax；
Step 2, the random population scale that generates seek the fitness value of NP individual for the initial population of NP；
Step 3 judges whether to meet stop criterion, if it is satisfied, then algorithm terminates, turns to step 6；Otherwise step 4 is turned to；
Step 4 carries out DE mutation operation, DE crossover operation, DE selection operation to current population at individual；
Step 5 executes local search algorithm SQI；
The best result that step 6, output acquire.
2. the effective dynamic coverage method of the forest fire monitoring system node according to claim 1 based on Internet of Things,
Be characterized in that, the step 1 is specific as follows: setting population scale NP=40, scale factor F=0.5, crossover probability CR=0.9,
Maximum evolutionary generation Gmax=300.
3. the effective dynamic coverage method of the forest fire monitoring system node according to claim 1 based on Internet of Things,
It is characterized in that, the step 2 is specific as follows:
Step 2.1, in order to guarantee the node coordinate after forest fire monitoring node initializing entirely monitoring regional scope in, i.e.,
The object vector range of initialization can cover entire solution space, and the object vector after initialization is expressed as shown in following formula:
Xji,G=Xj,min+randj,i(0,1)·(Xj,max-Xj,min)
In formula, rand (0,1) is the equally distributed random number of obedience that computer random generates in (0,1) section, Xj,min=
{X1,min,X2,min,…,XD,minIndicate to tie up the lower boundary of object vector in continuous Real-value space, i.e. forest fire monitoring area in D
The lower boundary in domain；Xj,max={ X1,max,X2,max,…,XD,maxIndicate to tie up the coboundary of object vector in continuous Real-value space in D,
That is the coboundary in forest fire monitoring region；Xi,GIndicate i-th of individual vector or the object vector in G generation:
Xi,G=(X1i,G,X2i,G,Xji,G,…,XDi,G)
In formula, G=0,1 ..., Gmax, G indicates algebra belonging to the population, and G=0 indicates initialization population vector, GmaxFor maximum
Algebra, i=1,2 ..., NP indicate that i-th of individual vector, D indicate D dimension space, i.e., tie up continuous real number value parameter space in the D
Solve globally optimal solution；
Step 2.2, the fitness function for calculating the object vector after initialization, i.e., under current state, according to fitness function
The fitness function value of forest fire monitoring node is calculated, chooses forest fire monitoring node in the coverage rate function in monitoring region
It is as follows as the fitness function:
4. the effective dynamic coverage method of the forest fire monitoring system node according to claim 3 based on Internet of Things,
It is characterized in that, the step 3 is specific as follows: judging whether to meet Forest Fire according to the fitness function value that step 2 is calculated
Calamity monitors the requirement of coverage rate, if it is satisfied, then algorithm terminates, turns to step 6；Otherwise step 4 is turned to.
5. the effective dynamic coverage method of the forest fire monitoring system node according to claim 4 based on Internet of Things,
It is characterized in that, the step 4 is specific as follows:
Step 4.1, DE mutation operation: it after forest fire monitoring node coordinate completes initialization as object vector, uses
" DE/rand/1 " strategy generating variation vector, i.e. the variation vector V of forest fire monitoring node coordinatei,G=(V1i,G,V2i,G,
Vji,G,…,VDi,G):
Vi,G=Xr1,G+F·(Xr2,G-Xr3,G)
In formula, r1 ≠ r2 ≠ r3, and r1, r2, r3 are the positive integers that computer randomly selects from [1, NP], are supervised in forest fire
One variation vector of every generation in the mutation operation of node is surveyed, r1, r2, r3 can be generated once by computer at random, and and mesh
It is also not identical to mark vector i, thus the initial population number NP of forest fire monitoring node should be indicated not less than 4, G G for vector,
F indicates that the scale factor of control forest fire monitoring node variation vector scaling, the scale factor are positive real constant, F value model
It encloses for F ∈ [0,2]；
In the dynamic coverage method of forest fire monitoring node, when it is desirable that make a variation vector population show diversity, avoid
When falling into local optimum, F should take relative larger value；When being desired with local search, when realizing fast convergence, F should be taken relatively
Small value；
Step 4.2, DE crossover operation: the diversity in order to guarantee node coordinate vector individual sits forest fire monitoring node
Mark the variation vector V of vectori,GWith the object vector X of forest fire monitoring node coordinate vectori,GThe parameter for being included, using two
Item formula intersects to generate the trial vector U of new forest fire monitoring node coordinate vectori,G, Ui,G=(U1i,G,U2i,G,
Uji,G,…,UDi,G), Ui,GIn variation vector V of at least one parameter from forest fire monitoring node coordinate vectori,G,
In formula, CR is to intersect the factor, indicates real value crossover probability constant, CR ∈ [0,1], rand (0,1) are computers in (0,1)
The equally distributed random number of the obedience generated at random in range；K be used to guarantee newly-generated forest fire monitoring node coordinate to
The trial vector U of amounti,GIn variation vector V of at least one parameter from forest fire monitoring node coordinate vectori,G, k takes
Being worth range is k ∈ [1, D], is chosen by computer random, CR value is bigger, comes from newly-generated experiment vector in crossover operation
Make a variation vector Vi,GParameter it is more；Conversely, CR value is smaller, in crossover operation in newly-generated experiment vector from variation to
Measure Vi,GParameter it is fewer；
Step 4.3, DE selection operation: the new forest fire monitoring section that will be generated after vector variation and crossover operation before this
The trial vector U of pointi,G, carry out the calculating of fitness value, and by the target of calculated value and forest fire monitoring node coordinate vector
Vector Xi,GThe value for carrying out fitness calculating is compared, to select calculated result in the two more preferably individual as the next generation
Individual vector:
In formula, f () indicates fitness function, i.e.,At this point, Xi,G+1As optimum point, f (Xi,G+1)
The smallest point of fitness function value in NP point is similarly denoted as most not good enough w for optimum point fitness value, the point is corresponding suitable
Response functional value is worst fitness value f (w).
6. the effective dynamic coverage method of the forest fire monitoring system node according to claim 5 based on Internet of Things,
It is characterized in that, the step 5 is specific as follows:
Step 5.1 randomly selects forest fire monitoring node coordinate vector x from NP point2And x3, x2=(x21,x22,x2i,…
x2D), x3=(x31,x32,x3i,…x3D) and meet x2≠x3≠Xi,G+1, enable x1=Xi,G+1；
Step 5.2, the new testing site p=(p for calculating forest fire monitoring node coordinate vector·1,p·2,p·j,…,p·D)T, wherein
Wherein, xj=(xj1,xj2,…xjD)T, j=1,2,3 ..., D indicate the testing site of forest fire monitoring D dimension space；f(·)
For the effective dynamic coverage adaptation of methods degree function of forest fire monitoring system node；
Step 5.3 calculates f (p), if fp> fbPoint p substitution point b is then used, as the next of forest fire monitoring node coordinate vector
Generation individual vector, is not otherwise replaced.
7. the effective dynamic coverage method of the forest fire monitoring system node according to claim 6 based on Internet of Things,
It is characterized in that, the step 6 is specific as follows:
Enable forest fire monitoring node coordinate vector Xi,G+1=(X, Y) is exported forest fire monitoring node coordinate matrix (X, Y),
Obtain the optimal Node distribution coordinates matrix value of the dynamic coverage method of forest fire monitoring node.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201811570336.6A CN109784465A (en) | 2018-12-21 | 2018-12-21 | The effective dynamic coverage method of forest fire monitoring system node based on Internet of Things |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201811570336.6A CN109784465A (en) | 2018-12-21 | 2018-12-21 | The effective dynamic coverage method of forest fire monitoring system node based on Internet of Things |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN109784465A true CN109784465A (en) | 2019-05-21 |
Family
ID=66497990
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201811570336.6A Withdrawn CN109784465A (en) | 2018-12-21 | 2018-12-21 | The effective dynamic coverage method of forest fire monitoring system node based on Internet of Things |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN109784465A (en) |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110691126A (en) * | 2019-09-25 | 2020-01-14 | 中南大学 | Reliable code distribution strategy for improving code coverage rate in Internet of things |
| CN111210586A (en) * | 2020-01-15 | 2020-05-29 | 浙江大虫科技股份有限公司 | Forest fire prevention system |
Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107909139A (en) * | 2017-11-13 | 2018-04-13 | 东南大学 | A kind of adjustable differential evolution algorithm of the crossover probability factor |
| CN108053597A (en) * | 2018-01-11 | 2018-05-18 | 深圳市晟达机械设计有限公司 | A kind of effective forest fire monitoring system |
| CN108401013A (en) * | 2018-01-23 | 2018-08-14 | 沈阳工程学院 | A kind of smart home Internet of Things communication device |
-
2018
- 2018-12-21 CN CN201811570336.6A patent/CN109784465A/en not_active Withdrawn
Patent Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107909139A (en) * | 2017-11-13 | 2018-04-13 | 东南大学 | A kind of adjustable differential evolution algorithm of the crossover probability factor |
| CN108053597A (en) * | 2018-01-11 | 2018-05-18 | 深圳市晟达机械设计有限公司 | A kind of effective forest fire monitoring system |
| CN108401013A (en) * | 2018-01-23 | 2018-08-14 | 沈阳工程学院 | A kind of smart home Internet of Things communication device |
Non-Patent Citations (1)
| Title |
|---|
| WENLI LEI 等: "Research on an Improved Wireless Sensor Networks Effective Coverage Method", 《INTERNATIONAL JOURNAL OF FUTURE GENERATION COMMUNICATION AND NETWORKING》 * |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110691126A (en) * | 2019-09-25 | 2020-01-14 | 中南大学 | Reliable code distribution strategy for improving code coverage rate in Internet of things |
| CN111210586A (en) * | 2020-01-15 | 2020-05-29 | 浙江大虫科技股份有限公司 | Forest fire prevention system |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN103354642B (en) | A kind of method improving mobile sensor network coverage rate | |
| He et al. | Collaborative sensing in internet of things: A comprehensive survey | |
| Yue et al. | Review and empirical analysis of sparrow search algorithm | |
| CN110062390A (en) | Based on the wireless sensor network node Optimization deployment method for improving wolf pack algorithm | |
| Liu et al. | Dynamic path planning for mobile robot based on improved genetic algorithm | |
| Pradhan et al. | Connectivity constrained wireless sensor deployment using multiobjective evolutionary algorithms and fuzzy decision making | |
| Wang et al. | Wireless sensor network deployment of 3D surface based on enhanced grey wolf optimizer | |
| CN103281704B (en) | Based on the wireless sensor network certainty spatial portion arranging method of three-dimensional perception | |
| Derr et al. | Wireless sensor network configuration—Part II: Adaptive coverage for decentralized algorithms | |
| CN111565372B (en) | Directed sensor network optimized deployment system and method | |
| CN109743740A (en) | Directional sensor network target monitoring method based on 3D-Voronoi subdivision | |
| CN109784465A (en) | The effective dynamic coverage method of forest fire monitoring system node based on Internet of Things | |
| Hajjej et al. | An efficient deployment approach for improved coverage in wireless sensor networks based on flower pollination algorithm | |
| Wang et al. | Virtual force-directed particle swarm optimization for dynamic deployment in wireless sensor networks | |
| Hideg et al. | Data collection for widely distributed mass of sensors | |
| Wang et al. | Multi-strategy enhanced grey wolf algorithm for obstacle-aware WSNs coverage optimization | |
| Yanmaz et al. | Area coverage with unmanned vehicles: A belief-based approach | |
| Zhu et al. | Brain storm optimization algorithm for full area coverage of wireless sensor networks | |
| CN103237312A (en) | Method for optimizing coverage of nodes of wireless sensor network | |
| Razafimandimby et al. | Neuro-Dominating set scheme for a fast and efficient robot deployment in internet of robotic things | |
| Feng | Patch-based hybrid modelling of spatially distributed systems by using stochastic HYPE-ZebraNet as an example | |
| CN114980134A (en) | Virtual force-based dynamic coverage method for multiple moving targets | |
| Wang et al. | Coverage-All Targets Algorithm for 3D Wireless Multimedia Sensor Networks Based on the Gravitational Search Algorithm | |
| Mathur et al. | Energy-efficient deployment of distributed mobile sensor networks using fuzzy logic systems | |
| Stewart et al. | A Voronoi diagram based population model for social species of wildlife |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| WW01 | Invention patent application withdrawn after publication |
Application publication date: 20190521 |
|
| WW01 | Invention patent application withdrawn after publication |