CN102724681B - Sensor network coverage hole detection method combining with energy efficiency - Google Patents

Abstract

The invention provides a sensor network coverage hole detection method combining with energy efficiency. The sensor network coverage hole detection method combining with the energy efficiency includes: calculating residual energy of sensor nodes in a sensor network after the sensor network works for one period; judging whether sensor nodes become invalid because of energy depletion according to the residual energy: on yes judgment, calculating detection probability of the sensor network, and executing a step four; on no judgment, returning to the step two; judging whether coverage holes occur according to the detection probability: on yes judgment, calculating positions of the coverage holes, and executing a step five; on no judgment, returning to the step two; screening current effective sensor nodes combining with the energy efficiency, and deleting sensor nodes which cannot meet energy efficiency requirements; and calculating the positions of the sensor network coverage holes for the residual nodes which can meet the energy efficiency requirements. The sensor network coverage hole detection method combining with the energy efficiency guarantees that energy waste of the sensor network is small, prolongs service life of repaired network, reduces network repair cost, and improves guiding significance of hole detection results to actual repair.

Description

A kind of sensor network of combination energy efficiency covers empty detection method

Technical field

The present invention relates to wireless sensor network field, the sensor network that is specifically related to a kind of combination energy efficiency covers empty detection method.

Background technology

Wireless sensor network forms by being deployed in the cheap microsensor nodes with perception and computing capability a large amount of in monitored area, form multihop self-organizing network by wireless mode, in order to complete the task of perceptive object information in perception, acquisition and processing monitored area.The revolution that wireless sensor network has brought information Perception with its low-power consumption, low cost, feature distributed and self-organizing, and be more and more applied to the fields such as military affairs, disaster relief, environmental monitoring, medical treatment and nursing, intelligent building.But in the practical application of wireless sensor network, for the sensor network of having disposed, node may be because environmental factor, External Force Acting, electronic device damage etc. are former thereby inefficacy; On the other hand, due to the finite energy of node, in the time of depleted of energy, failure node also can impact the covering of network.The covering cavity being produced by these reasons can impact network work quality.Thus, researcher enters on how the covering cavity of sensor network being detected.The key issue that sensor network covers empty detection method is, obtains fast and accurately the covering cavity position of sensor network, and makes to have good repairing effect based on the repair of this testing result.

Moment sensor network cavity detection method is more, mainly comprises the method based on computational geometry, the method based on Voronoi figure and the method based on simplicial complex etc.Method based on computational geometry be by calculate a certain node be adjacent the angle forming between node judge whether exist cover cavity, can accurately provide the empty number in certain region, but cannot accurately describe covering empty position.And the classical method of utilizing Voronoi figure is according to the positional information of network node, utilize Voronoi figure that overlay area is divided into some unit, in each unit, only comprise a bit.According to the principle of Voronoi figure, the Voronoi region of each node is the convex domain nearest apart from this point, in a certain specific Voronoi region, if the region that exists node corresponding to this region to cover, other nodes also cannot cover so, and this region is a covering cavity.This method can easy, quick discovery cover cavity, represents cavity position, and can not accurately describe empty position and shape but can only scheme polygonal summit with Voronoi.Surveying and cover empty method based on simple reinspection, is by setting up maximum simplicial complex subnet, and the polygon being connected into by the covering edge intersection point that covers empty marginal node is described cavity; This method is more accurate compared with Voronoi figure method, but irregular often owing to covering empty shape, describes cavity still can have error with polygon.

In sum, the theoretical foundation of traditional covering cavity searching method based on different is searched for the covering cavity of sensor network, but all the geographical location information such as the node location that utilizes network and internodal distance without exception carries out cavity search, and the basis of repairing as further cavity using this Search Results.But the impact of the capacity volume variance that these methods have all been ignored each working node of empty when search on cavity search, because these work on hand nodes energy that depleted of energy produces after repair in cavity cavity problem may exert an influence to the coverage rate of network, and then in short time after reparation, again occur covering cavity, affect the repairing effect of network.

Summary of the invention

The deficiency existing for prior art, the sensor network that the invention provides a kind of combination energy efficiency covers empty detection method, and the method cavity searching position is accurate, and the sensor network based on after this cavity testing result reparation has the longer operating time.

The sensor network of a kind of combination energy efficiency of the present invention covers empty detection method, comprises sensor network working node heat-supplied and screening based on energy efficiency, and the covering cavity position based on probability sensor model and combined detection probability detects.Particularly: the sensor network energy consumption models based on definite, the dump energy situation of each sensor node after the every work one-period of computing network.If occur, sensor node dump energy is less than the situation of the primary power of this sensor node, illustrate that this sensor node lost efficacy, calculate the detection probability of Zhong Ge position, monitored area based on the concept of sensor node probability sensor model and combined detection probability, compare with the coverage rate index requiring, obtain the coverage condition of each point, judge whether to occur covering cavity.Cover cavity if do not occur, sensor network works on, and covers cavity if occur, need to the position in this covering cavity be determined and be repaired.In the time covering cavity search, need to screen shorter working node of life-span in sensor network in conjunction with energy efficiency.The ratio of the shorter shared residue working node of working node number of life-span that need to filter out can obtain by drawing the curve that network restoration cost and network energy waste change with this ratio, and the corresponding ratio of focus of two curves is determined ratio value.After the ratio of shorter node of the definite life-span that need to delete by screening, can calculate the node number that needs deletion.Screen by calculating the concrete surplus working life of each residue working node the shorter node of life-span ascertaining the number afterwards.Finally utilize the working node that meets energy efficiency requirement, the detection probability of network monitor region each point is calculated, obtain the coverage condition of each point.And by the noise reduction process of binary conversion treatment and opening operation and closed operation, extract empty edge, obtain sensor network and cover empty position.

Technical scheme of the present invention is achieved in that

The sensor network that the inventive method adopts, comprises some sensor nodes and a Sink node (being aggregation node), and each sensor node is isomorphism, has identical running parameter and primary power.Sensor node and Sink node random placement, each sensor node adopts single-hop mode to transmit data, be that each sensor node directly sends data to Sink node, sensor network is with periodic manner work, and it is one-period that each sensor node completes the time that a perception and data transmission work experiences.A node completes the data acquisition to monitored target in one-period, and obtained packet is mail to base station.

The sensor network of combination energy efficiency covers an empty detection method, comprises the steps:

Step 1: sensor network is started working;

Step 2: after sensor network work one-period, the dump energy of each sensor node in calculating sensor network;

Adopt the dump energy of each sensor node in energy consumption model calculating sensor network:

Energy consumption model is described below:

Distance between transmitting terminal and receiving terminal is d, sets a threshold value d ₀, short range transmission is d<d ₀time, adopt free space model (d ²energy loss) calculating energy consumption; When larger Distance Transmission is d>=d ₀time, due to transmitting terminal and all ground proximitys of receiving terminal, disturb greatlyr, barrier is more, and energy loss sharply increases along with the increase of distance, therefore establish the biquadratic d of communication energy consumption and distance ⁴be directly proportional, now adopt multichannel attenuation model (d ⁴energy loss) calculating energy consumption.The energy that transmitting terminal consumes in the time that the receiving terminal apart from d sends l Bit data is E _tx(l, d):

$E_{\mathrm{Tx}}(l,d)=E_{\mathrm{Tx}-\mathrm{elec}}\left(l\right)+E_{\mathrm{Tx}-\mathrm{amp}}(l,d)=\left\{\begin{array}{cc}{\mathrm{lE}}_{\mathrm{elec}}+{\mathrm{l\&epsiv;}}_{\mathrm{fs}}{d}^2,& d<d_0\\ {\mathrm{lE}}_{\mathrm{elec}}+{\mathrm{l\&epsiv;}}_{\mathrm{mp}}{d}^4,& d\&GreaterEqual;d_0\end{array}\right.---\left(1\right)$

Wherein, E _tx-elec(l) be radiating circuit loss gross energy, E _tx-amp(l, d) is power amplification loss gross energy, and d is the distance of transmitting terminal to receiving terminal.Calculate the distance of each sensor node and Sink node and determine each energy parameter, energy parameter comprises electron energy E _elec, amplifier energy parameter ε under free space model _fs, many in amplifier energy parameter ε under attenuation model _mp, distance threshold d ₀, substitution energy consumption model formula (1), determines the energy E consuming after the every work one-period of each sensor node _w(l, d), and then calculate the dump energy E of each sensor node _r, E _r0for current period starts the dump energy of front nodal point.

E _r=E _r0-E _w(l,d) （2）

E _r0be that 0 node is failure node, not being 0 is current effective node.

Step 3: according to the dump energy of the each sensor node calculating, judged whether sensor node depleted of energy and lost efficacy: be, the detection probability of calculating sensor network, and perform step four; No, sensor network works on, and returns to step 2;

Sensor network nodes has certain primary power E in the time disposing ₀, the every work one-period of network, the dump energy of sensor network nodes just reduces to some extent.In the time that dump energy reduces to 0, this node energy exhausts, and node no longer works on.

The detection probability of calculating sensor network, specific as follows:

The probability sensor model calculating sensor network detection probability adopting, according to the description of this model, the probability that sensor node can be found to the event that its distance is d is p (d):

$p\left(d\right)=\left\{\begin{array}{c}1,d\&le;R_1\\ e^{-\mathrm{\&lambda;}{(d-R_1)}^{\mathrm{\&gamma;}}},R_{\max }>d>R_1\\ 0,d\&GreaterEqual;R_{\max }\end{array}\right.---\left(3\right)$

Wherein, R ₁for this sensor node perception starts to become uncertain sensing range lower limit, the perceptual parameters that λ and γ are sensor node, the value of λ and γ is determined by the physical characteristic of sensor node.R _maxit is the maximum sensing range upper limit of this node.Set R here, ₁=0, γ=1, simplifies model, and the probability sensor model after simplification becomes:

$p\left(d\right)=\left\{\begin{array}{c}{e}^{-\mathrm{\&lambda;d}},R_{\max }>d\&GreaterEqual;0\\ 0,d\&GreaterEqual;R_{\max }\end{array}\right.---\left(4\right)$

Calculate the detection probability of each sensor node for sensor network monitoring region each point by this probability sensor model.

In sensor network, certain monitored area often can be by the perception simultaneously of multiple sensor nodes, so be the synergistic results of these sensor nodes to the detection of this monitored area.Therefore when, the detection probability of each point calculates in to monitored area, should adopt joint probability.

Suppose S={s ₁, s ₂..., s _nbe all can be to sensor network monitoring region mid point P _jcarry out all sensor node set of perception, put P _jthe probability that can be detected by sensor network is:

$C\left(P_j\right)=1-\underset{i\&Element;S}{\mathrm{\&Pi;}}(1-p(i,j))---\left(5\right)$

Wherein, C (P _j) be that sensor network is to a P _jdetection probability, p (i, j) is sensor node S _ito a P _jdetection probability.According to above formula, can the detection probability of monitored area be distributed and be calculated.

Step 4: judge whether to occur covering cavity according to the detection probability of the sensor network calculating: be to calculate and cover empty position; No, sensor network works on, and returns to step 2;

Judge which region is to cover empty region, can be by judging whether the detection probability of every bit of monitored area reaches coverage requirement and realize.Supposing has unified coverage requirement to monitored area, and the detection probability of and if only if certain monitored area reaches detection probability and requires C _thtime, this region can effectively be surveyed, otherwise, represent this region in cover hole region, cannot effectively be surveyed by network, and covering hole region is regarded as in this region.When the combined detection probability of monitored area meets C _thtime, this search coverage is not to cover cavity, the C (P calculating _j) be the detection probability of this monitored area; When being less than detection probability, the combined detection probability of monitored area requires C _thtime, this region is for covering cavity, and the detection probability in this region sets to 0, and covering empty judge mode can illustrate by following formula:

$C\left(P_j\right)=\left\{\begin{array}{c}C\left(P_j\right),C\left(P_j\right)\&GreaterEqual;C_{\mathrm{th}}\\ 0,C\left(P_j\right)<C_{\mathrm{th}}\end{array}\right.---\left(6\right)$

Detection probability to each point in sensor network monitoring region calculates, and requires to judge that by detection probability existence covers behind cavity, and the covering cavity position of sensor network is calculated.Concrete account form is: gridding is carried out in the monitored area of sensor network, be divided into m × n grid, gridding can be carried out in the overlay area of network after determining the step-length of network transverse and longitudinal coordinate.For the each lattice point after gridding, first calculate the detection probability of each sensor node for this lattice point, then utilize the detection probability of these detection probability calculating sensor networks to this lattice point, finally require C with the detection probability of sensor network _thwhether compare, just can judge the coverage condition of this lattice point, be to cover cavity.The coverage condition process that judges each lattice point of network's coverage area is the same, just can calculate the coverage condition of network entirety, i.e. the coverage of network.

Suppose that sensor network monitoring region carries out being divided into N lattice point after gridding, the coverage p of definition sensor network is:

$p=\frac{\underset{k=1}{\overset{N}{\mathrm{\&Sigma;}}}\mathrm{GridNum}\left(k\right)}{N}---\left(7\right)$

GridNum in formula (k) characterizes k grid point and meets the degree of detection probability requirement, if meet the demands, is 1; Otherwise be 0.

Step 5: in conjunction with energy efficiency, current effective sensor node is screened, delete the sensor node that does not meet energy efficiency requirement;

So-called energy efficiency, refers to the energy situation of each sensor node in sensor network.The energy state of in running order sensor node can distinguish the detection of covering cavity by the concept of energy efficiency time.The concrete evaluation method of energy efficiency is the concrete surplus working life of working node.

According to concrete surplus working life, each working node in network is screened, delete the working node that does not meet energy efficiency, delete shorter working node of life-span, these nodes are regarded as to failure node, in the time that coverage rate is calculated, do not consider the impact of these nodes, while reparation in cavity, the caused covering of these node failures cavity is repaired simultaneously;

When sensor network is covered to cavity detection, in running order sensor node screens, and deletes shorter terminal node of life-span, deletes shorter sensor node of life-span, specifically carries out as follows:

Step 1: the variation of sensor network rehabilitation cost when computing node deletion ratio changes.

The account form of the rehabilitation cost of sensor network is: suppose that sensor network occurs that covering the empty time is t _w, the residue working node number covering outside hole region is n, and knot removal ratio is d%, and needing the node number of deleting is n × d%.And then can calculate and delete the covering cavity area that sensor network need to be repaired after these nodes, knot removal ratio is the ratio that needs the shorter shared current effective sensor node of sensor node of life-span of deleting.

Covering empty area can calculate by following formula:

$H=M\&times;p=M\&times;\frac{\underset{k=1}{\overset{N}{\mathrm{\&Sigma;}}}\mathrm{GridNum}\left(k\right)}{N}---\left(8\right)$

Wherein, H is for covering empty area, and M is network monitor region area, and p is sensor network coverage.

Covering cavity area after each sensor node that calculating meets energy efficiency lost efficacy, if covering empty area changes, when repairing in cavity, the covering cavity producing when this node failure is described do not consider, sensor network after reparation has produced new cavity, need to repair sensor network.The time t losing efficacy with this sensor node _ndeduct sensor network and occur covering the empty time, be the working life t of sensor network after repairing _f; Suppose that the cost of often once repairing is C, the rehabilitation cost of whole network is C/t _f.Can calculate the rehabilitation cost variation of deleting ratio lower network in difference according to the method.

Step 2: the variation of sensor network waste energy when computing node deletion ratio changes.Suppose that needing the nodes of deleting is K, the dump energy of these nodes that will delete while covering cavity monitoring by calculating, and these dump energies are summed up, can calculate the energy E that network is wasted _c.Computing formula is:

$E_c=\underset{i=1}{\overset{K}{\mathrm{\&Sigma;}}}{E}_r=\underset{i=1}{\overset{K}{\mathrm{\&Sigma;}}}[E_i-E_w(l,d)]=\underset{i=1}{\overset{K}{\mathrm{\&Sigma;}}}[E_i-\frac{t_w}{T}{E}_{\mathrm{Tx}}(l,d)]---\left(9\right)$

Wherein E _rfor the dump energy of node, E _ifor the primary power of node, E _w(l, d) is the node work energy of loss, the work period that T is node, t _wfor there is covering the empty time in sensor network.Can calculate the variation of deleting ratio lower sensor energy that network is wasted in difference according to the method.

Step 3: the sensor network rehabilitation cost calculating and sensor network waste energy are normalized.

Step 4: draw sensor network rehabilitation cost and sensor network energy waste change curve according to the data after normalization under the same coordinate system, get abscissa value that two intersections of complex curve are corresponding as adopted knot removal ratio, and then determine and need the node number of deleting.

Obtain after the shorter sensor node number of life-span that needs to delete, by calculating the concrete surplus working life of each residue working sensor node, determine because the life-span is compared with short each sensor node that needs deletion.The residual life t of node _rcomputing formula is as follows:

$t_r=T\&times;\frac{E_r}{E_{\mathrm{Tx}}(l,d)}---\left(10\right)$

The node number of deleting as required, deletes the shortest a few thing node of life-span, and after deleting, remaining sensor node is the residue working node that meets energy efficiency.

Step 6: to meeting the position in residue node calculating sensor network coverage cavity of energy efficiency requirement.

Recalculate the detection probability of the each lattice point in sensor network monitoring region after deletion of node, and require C by detectivity _thjudge the coverage condition of each node, be 1 if meet coverage rate requirement, otherwise be 0.Just can be converted into each lattice point value be 0 or 1 matrix to the coverage rate situation of sensor network thus.Then this matrix is converted into two-dimensional monochromatic image, and by the noise reduction process of opening operation and closed operation, removes incoherent structure in image, extract empty edge.

Just obtain thus covering cavity in conjunction with the sensor network of energy efficiency.

Beneficial effect

The inventive method has been considered the impact that energy problem detects sensor network, in the situation that ensureing that sensor network energy waste is less, extend greatly the working life of network after repairing, improve the reliability of network work, and reduce the cost of network restoration, improve the directive significance of empty Search Results to actual repair work, made the network after repairing can avoid occurring at short notice covering cavity, there is longer working life.

Brief description of the drawings

The method flow diagram of Fig. 1 specific embodiment of the invention;

Fig. 2 specific embodiment of the invention detection probability distribution graphics;

Fig. 3 specific embodiment of the invention detection probability distribution contour map;

Fig. 4 specific embodiment of the invention sensor network covers cavity position schematic diagram;

Fig. 5 specific embodiment of the invention sensor network rehabilitation cost and energy dissipation are with the change curve of knot removal ratio;

The each sensor node energy consumption of Fig. 6 specific embodiment of the invention sensor network schematic diagram;

Fig. 7 specific embodiment of the invention covers empty detection position schematic diagram in conjunction with the sensor network of energy efficiency;

After the reparation of Fig. 8 specific embodiment of the invention, sensor network covers empty area change comparison diagram.

Embodiment

Below in conjunction with accompanying drawing, specific embodiment of the invention is elaborated.

The sensor network that present embodiment adopts, comprises 80 sensor nodes and a Sink node, and each sensor node is isomorphism, has identical running parameter and primary power.Sensor node and Sink node random placement are in the plane domain of 300 × 300 units, each sensor node adopts single-hop mode to transmit data, be that each sensor node directly sends data to Sink node, ensure all standing of sensor network, dispose sensor node at edge, monitored area, do not consider the energy consumption of these nodes, to avoid the impact of edge effect simultaneously.Sensor network is with periodic manner work.

The sensor network of the combination energy efficiency of present embodiment covers empty detection method, and software environment is WINDOWS7 system, and simulated environment is MATLAB2009, and flow process as shown in Figure 1, comprises the steps:

Step 1: sensor network is started working;

Step 2: after sensor network work one-period, the dump energy of each sensor node in calculating sensor network;

Adopt the dump energy of each sensor node in energy consumption model calculating sensor network:

Suppose that sensor network works 20 cycles every day, each sensor node sends 20 secondary data to sink node every day, and the length of each packet is 1000byte, and communication energy parameter is as shown in table 1.The setting parameter of Energy-aware model is: λ=0.05, R _max=50m, and setting network detection probability requires C _th=0.6.

Table 1 sensor network communication energy parameter

By energy parameter substitution energy consumption model formula (1), determine the energy E consuming after the every work one-period of each sensor node _w(l, d), and then calculate the dump energy E of each sensor node _r, E _r0for current period starts the dump energy of front nodal point.

E _r=E _r0-E _w(l,d) （2）

E _r0be that 0 node is failure node, not being 0 is current effective node.

Step 3: according to the dump energy of the each sensor node calculating, judged whether sensor node depleted of energy and lost efficacy: being, the detection probability of calculating sensor network; No, sensor network works on, and returns to step 2;

Sensor network nodes has certain primary power E in the time disposing ₀, the every work one-period of network, the dump energy of sensor network nodes just reduces to some extent.In the time that dump energy reduces to 0, this node energy exhausts, and node no longer works on.

The detection probability of calculating sensor network, specific as follows:

The probability sensor model calculating sensor network detection probability adopting, according to the description of this model, the probability that sensor node can be found to the event that its distance is d is:

$p\left(d\right)=\left\{\begin{array}{c}1,d\&le;R_1\\ e^{-\mathrm{\&lambda;}{(d-R_1)}^{\mathrm{\&gamma;}}},R_{\max }>d>R_1\\ 0,d\&GreaterEqual;R_{\max }\end{array}\right.---\left(3\right)$

Wherein, R ₁for this sensor node perception starts to become uncertain sensing range lower limit, the perceptual parameters that λ and γ are sensor node, the value of λ and γ is determined by the physical characteristic of sensor node.R _maxit is the maximum sensing range upper limit of this node.Set R here, ₁=0, γ=1, simplifies model, and the probability sensor model after simplification becomes:

$p\left(d\right)=\left\{\begin{array}{c}{e}^{-\mathrm{\&lambda;d}},R_{\max }>d\&GreaterEqual;0\\ 0,d\&GreaterEqual;R_{\max }\end{array}\right.---\left(4\right)$

Calculate the detection probability of each sensor node for sensor network monitoring region each point by this probability sensor model.

In sensor network, certain monitored area often can be by the perception simultaneously of multiple sensor nodes, so be the synergistic results of these sensor nodes to the detection of this monitored area.Therefore when, the detection probability of each point calculates in to monitored area, should adopt joint probability.

Suppose S={s ₁, s ₂..., s _nbe all can be to sensor network monitoring region mid point P _jcarry out all sensor node set of perception, put P _jthe probability that can be detected by sensor network is:

$C\left(P_j\right)=1-\underset{i\&Element;S}{\mathrm{\&Pi;}}(1-p(i,j))---\left(5\right)$

Wherein, C (P _j) be that sensor network is to a P _jdetection probability, p (i, j) is sensor node S _ito a P _jdetection probability.According to above formula, can the detection probability of monitored area be distributed and be calculated.

Step 4: judge whether to occur covering cavity according to the detection probability of the sensor network calculating: be to calculate and cover empty position; No, sensor network works on, and returns to step 2;

It is C (P that the detection probability of and if only if certain monitored area reaches that detection probability requires _j)>=C _thtime, this region can effectively be surveyed, on the contrary C (P _j) <C _thtime, represent this region in cover hole region, cannot effectively be surveyed by network, and covering hole region is regarded as in this region.When the detection probability of monitored area meets C _thtime, this search coverage is not to cover cavity, the C (P calculating _j) be the detection probability of this monitored area; When being less than detection probability, the detection probability of monitored area requires C _thtime, this region is for covering cavity, and the detection probability in this region sets to 0, and covering empty judge mode can illustrate by following formula:

$C\left(P_j\right)=\left\{\begin{array}{c}C\left(P_j\right),C\left(P_j\right)\&GreaterEqual;C_{\mathrm{th}}\\ 0,C\left(P_j\right)<C_{\mathrm{th}}\end{array}\right.---\left(6\right)$

Covering cavity position to sensor network calculates: the step-length of getting horizontal stroke, ordinate is 2, gridding is carried out in the network monitor region of 300 × 300 units, be divided into 150 × 150 lattice points, first calculate the detection probability of each sensor node for this lattice point, then utilize these detection probabilities to calculate the detection probability of this lattice point, be the detection probability of sensor network to this lattice point, finally require C with setting network detection probability _th=0.6 compares, if be less than detection probability requirement, illustrates that this lattice point representative region exists covering cavity.Calculate after the coverage value of these 150 × 150 lattice points, just can determine in network and whether exist and cover cavity.

Sensor network monitoring region carries out being divided into N lattice point after gridding, N=150 × 150, and the coverage of definition sensor network is:

$p=\frac{\underset{k=1}{\overset{N}{\mathrm{\&Sigma;}}}\mathrm{GridNum}\left(k\right)}{N}---\left(7\right)$

GridNum in formula (k) characterizes k grid point and meets the degree of detection probability requirement, if meet the demands, is 1; Otherwise be 0.

By calculating, in the time there is the 8th failure node in sensor network, there is covering cavity in sensor network.By calculating the out-of-service time of the 8th node, can obtain network and occur covering the empty time, be the 21.85th day.

The primary power of supposing sensor node is E ₀, the work period of sensor network is T, the working life t of sensor node _lcomputing formula is as follows:

$t_1=T\&times;\frac{E_0}{E_{\mathrm{Tx}}(l,d)}$

The 8th sensor node is brought in above formula to distance and the network communication parameters of Sink node, can obtain sensor network and occur covering cavity at the 17th all after date of the 21st day, now in sensor network, existing 8 sensor nodes lost efficacy, and remained 72 working sensor nodes.Fig. 2,3 is respectively now graphics and the contour map of sensor network detection probability.According to detection probability requirement, can obtain the covering cavity position of sensor network.Now be necessary this covering cavity to repair.Shown in Fig. 4, be the covering cavity position of current appearance.

Step 5: in conjunction with energy efficiency, current effective sensor node is screened, delete the sensor node that does not meet energy efficiency requirement;

According to concrete surplus working life, each working node in network is screened, delete the working node that does not meet energy efficiency, delete shorter working node of life-span, these nodes are regarded as to failure node, in the time that coverage rate is calculated, do not consider the impact of these nodes, while reparation in cavity, the caused covering of these node failures cavity is repaired simultaneously;

(in the time being less than 5%, the working life of sensor network is too short from 5% ~ 30% variation to set knot removal ratio; In the time being greater than 30%, have the waste of a large amount of sensor network energy, therefore only consider the situation in 5% ~ 30% interior variation), get a bit every 5%, calculate respectively the rehabilitation cost of network while getting each ratio value and the energy that network is wasted.

When sensor network is covered to cavity detection, in running order sensor node screens, and deletes shorter terminal node of life-span, deletes shorter sensor node of life-span, specifically carries out as follows:

Step 1: the variation of sensor network rehabilitation cost when computing node deletion ratio changes.

The account form of the rehabilitation cost of sensor network is: suppose that sensor network occurs that covering the empty time is t _w, the residue working node number covering outside hole region is n, and knot removal ratio is d%, and needing the node number of deleting is n × d%.And then can calculate and delete the covering cavity area that sensor network need to be repaired after these nodes, covering empty area can calculate by following formula:

$H=M\&times;p=M\&times;\frac{\underset{k=1}{\overset{N}{\mathrm{\&Sigma;}}}\mathrm{GridNum}\left(k\right)}{N}$

Wherein, H is for covering empty area, and M is network monitor region area, and p is sensor network coverage.

Covering cavity area after each sensor node that calculating meets energy efficiency lost efficacy, if covering empty area changes, when repairing in cavity, the covering cavity producing when this node failure is described do not consider, sensor network after reparation has produced new cavity, need to repair sensor network.The time t losing efficacy with this sensor node _ndeduct sensor network and occur covering the empty time, be the working life t of sensor network after repairing _f; Suppose that the cost of often once repairing is C, the rehabilitation cost of whole network is C/t _f.Can calculate the rehabilitation cost variation of deleting ratio lower network in difference according to the method.

Be taken as 5% as example taking knot removal ratio, the rehabilitation cost of calculating sensor network.In the time that knot removal ratio is taken as 5%, needing the shorter sensor node number of life-span of deleting is 72 × 5%=3.6, is 4 after rounding up.By calculating the concrete surplus working life of each residue working node, can obtain the position of 4 the shortest nodes of this life-span.These 4 nodes are deleted from sensor network residue working node, calculated the current sensor network of repairing that needs and cover empty area, can obtain empty area is 1612.When in evaluation work node, next node loses efficacy afterwards, the covering cavity size of sensor network, changed if occur, and illustrate in network and occurred new covering cavity, the sensor network cisco unity malfunction after reparation, the working life of sensor network finishes.For this embodiment, while calculating the cavity size of sensor network after the 5th node failure that the life-span is shorter, find that empty size becomes 3620, illustrate in sensor network and occurred new covering cavity, the working life of the sensor network after the knot removal ratio reparation based on 5% finishes.Calculating the time of the 5th node failure, is the 27.85th day, deducts network and carries out the time that repair in cavity, and 21.8 days, after repairing can get this ratio time, the work number of days of network was 6.05 days.The rehabilitation cost of sensor network is C/6.05.

Network restoration cost during taking same method calculating ratio respectively as 10% to 30%, result of calculation is as shown in table 2.

Table 2 network restoration cost is with knot removal ratio delta data

Need the ratio of deletion of node	5%	10%	15%	20%	25%	30%
							Need the node number of deleting	4	7	11	14	18	22
Network restoration cost	C/6.05	C/17	C/22.7	C/26.2	C/33.7	C/41.9

Step 2: the variation of sensor network waste energy when computing node deletion ratio changes.Suppose that needing the nodes of deleting is K, the dump energy of these nodes that will delete while covering cavity monitoring by calculating, and these dump energies are summed up, can calculate the energy E that network is wasted _c.Computing formula is:

$E_c=\underset{i=1}{\overset{K}{\mathrm{\&Sigma;}}}{E}_r=\underset{i=1}{\overset{K}{\mathrm{\&Sigma;}}}[E_i-E_w(l,d)]=\underset{i=1}{\overset{K}{\mathrm{\&Sigma;}}}[E_i-\frac{t_w}{T}{E}_{\mathrm{Tx}}(l,d)]$

Wherein E _rfor the dump energy of node, E _ifor the primary power of node, E _wfor the energy of node work loss, the work period that T is node.Can calculate the variation of deleting ratio lower sensor energy that network is wasted in difference according to the method.

Be taken as 5% as example taking knot removal ratio, the energy that calculating sensor network is wasted.The dump energy of each residue working node when calculating sensor network occurs covering cavity respectively.Taking shorter node of the life-span of the 1st deletion as example, calculate the distance of this node to Sink node, and by distance value with covering empty time of occurrence, 21.85, and in sensor network communication energy parameter substitution following formula:

$E_r=E_i-E_w(l,d)=E_i-\frac{t_w}{T}{E}_{\mathrm{Tx}}(l,d)$

The dump energy that can calculate this node is 0.0186.Calculate the shorter node of life-span of other deletions, and the residual energy value of each node is summed up, the waste energy value that can obtain sensor network is 0.2846.

The energy that while getting other knot removal ratios with same method calculating, sensor network is wasted, result of calculation is as shown in table 3

Table 3 network energy is wasted with knot removal ratio delta data

Need the ratio of deletion of node	5%	10%	15%	20%	25%	30%
							Need the node number of deleting	4	7	11	14	18	22
Network restoration cost	0.2846	0.76	1.7076	2.4836	3.6345	4.8918

Step 3: the sensor network rehabilitation cost calculating and sensor network waste energy are normalized.

Obtaining after sensor network rehabilitation cost and the real data of sensor network waste energy with the variation of knot removal ratio, for both variation tendencies are represented under the same coordinate system, these two groups of data are normalized.Adopt the normalized function mapminmax () in Matlab to be normalized, the concrete form of calling is Y=mapminmax (X, 0,1), X is normalized to [0,1] interval, Y is the data after the normalization of acquisition, and the data after normalization are as shown in table 4

The normalization data of table 4 network restoration cost and network energy waste

Network restoration cost	C/6.05	C/17	C/22.7	C/26.2	C/33.7	C/41.9
							Network restoration cost (normalization)	1	0.2472	0.1427	0.1011	0.0411	0
Network energy waste (J)	0.2846	0.76	1.7076	2.4836	3.6345	4.8918
							Network energy waste (normalization)	0	0.1032	0.3089	0.4773	0.7271	1

Step 4: draw sensor network rehabilitation cost and sensor network energy waste change curve according to the data after normalization under the same coordinate system, get abscissa value that two intersections of complex curve are corresponding as adopted knot removal ratio, and then determine and need the node number of deleting.

After obtaining normalization data, the change curve of sensor network rehabilitation cost and sensor network waste energy is being plotted under the same coordinate system, X-axis is knot removal ratio, Y-axis is [0,1], article two, the corresponding X-axis coordinate of change curve focus is required knot removal ratio, and this change curve as shown in Figure 5.In the present embodiment, this deletion ratio is 12.5%, and needing the nodes of deleting is 72 × 12.5%=9 node.

Obtain after the shorter sensor node number of life-span that needs to delete, by calculating the concrete surplus working life of each residue working sensor node, determine because the life-span is compared with short each sensor node that needs deletion.The residual life t of sensor node _rcomputing formula is as follows:

$t_r=T\&times;\frac{E_r}{E_{\mathrm{Tx}}(l,d)}$

The node number of deleting as required, deletes the shortest a few thing node of life-span, and after deleting, remaining sensor node is the residue working node that meets energy efficiency.

Calculate the concrete surplus working life of each working node, 9 the shortest working life nodes are deleted from residue working node.Utilize the covering cavity position of the remaining working node computing network that meets energy efficiency requirement.The Energy Expenditure Levels of each sensor node when Fig. 6 represents to occur cavity, is divided into the node of depleted of energy (being failure node) at work, working life is shorter and needs are deleted node (not meeting the node of energy efficiency requirement) and working life compared with working node (meeting the node of energy efficiency requirement) three classes of growing and retaining because meeting energy efficiency requirement by the sensor node of disposing in sensor network.

Step 6: to meeting the position in residue node calculating sensor network coverage cavity of energy efficiency requirement.

Utilize the working node retaining to recalculate the detection probability of the each lattice point in sensor network monitoring region after deletion of node, and require C by detectivity _thjudging the detection probability of each node, be 1, otherwise be 0 if meet detection probability requirement, is 0 or 1 150 × 150 matrix notation by value by the detection probability of sensor network.Just can be converted into each lattice point value be 0 or 1 matrix to the detection probability of sensor network thus.Then by the image () statement in Matlab, this matrix is converted into two-dimensional monochromatic image, utilize imopen () and imclose () statement to carry out opening operation and closed operation, remove incoherent structure in image, extract empty edge, Fig. 7 is in conjunction with the sensor network of energy efficiency and covers empty detection position.

Just obtain thus covering cavity in conjunction with the sensor network of energy efficiency.

Comparison diagram 5,7 can be seen, the energy efficiency of combined sensor node covers after the analysis of cavity sensor network, cover cavity position and shape variation has occurred, what need reparation is not only the covering cavity being produced by current failure node, also carried out the reparation of predictability to repairing the covering cavity that the rear life-span produces when shorter node failure, made the sensor network after repairing can have longer working life.

Comparison diagram 5,7 can be seen, when we have considered after the energy problem of node, there is variation in cavity position and shape, what need reparation is not only the covering cavity being produced by current failure node, also carried out the reparation of predictability to repairing the cavity that the rear life-span produces when shorter node failure, made the network after repairing can have longer working life.

In order further this method to be compared with the covering cavity testing result of conventional method, this method has been calculated from sensor network and has been occurred for the first time covering cavity and having carried out empty reparation, (suppose that repair repaired current cavity completely, and the coverage rate in other regions except restoring area is not exerted an influence) within 40 days, cover afterwards the variation tendency of empty area to sensor network work, as shown in Figure 8.

As can be seen from Figure 8, cover empty area, within a couple of days behind cavity appears in sensor network for the first time, variation has occurred, at the 26th day, cavity area has become 1144, at the 27th day, and empty Area Growth to 4920, during by the 38th day, the area in cavity has reached 6928.If this explanation is according to the covering cavity position of traditional current appearance of empty searching method calculating sensor network, and the words of repairing, in a few days after reparation (after 5 of this emulation days), covering empty area will change, so just have to the covering cavity of sensor network to carry out reparation again.Therefore,, although traditional empty searching method can provide current covering cavity position information, the covering cavity repair of carrying out according to the result of this method is inefficiency in the long term.And the sensor network of the combination energy efficiency that the present invention proposes covers empty detection method, the impact of the covering cavity still having produced in the time of the low-yield node failure of work while having fully taken into account empty search on network coverage degree, the cavity producing after these node failures is carried out to the calculating of predictability, the cavity that result is carried out is thus repaired and can be made network within the longer time, avoid occurring covering cavity, needn't carry out frequently cavity and repair, improve the quality of empty repair.The emulation of carrying out taking present embodiment is example, the calculating of predictability account for cover 12.5% the node (9) that remains working node number when cavity is detected and lost efficacy after the empty positional information of covering, can in subsequently 20 days, there is not covering cavity according to the sensor network after this result reparation, until the 44.5th day (being produced the time in new covering cavity by the life-span compared with long working node inefficacy), just there is new covering cavity to occur, need to repair.

To sum up, cover empty detection method in conjunction with the sensor network of energy efficiency and realized the covering cavity testing of considering energy efficiency problem.The method, in ensureing that energy dissipation is less, has extended the working life of network after repairing greatly, has improved the reliability of network work, and has reduced the cost of network restoration.

Claims (2)

1. the sensor network in conjunction with energy efficiency covers empty detection method, the sensor network that the method adopts, comprises some sensor nodes and a Sink node, and Sink node is aggregation node, each sensor node is isomorphism, has identical running parameter and primary power; Sensor node and Sink node random placement, each sensor node adopts single-hop mode to transmit data, be that each sensor node directly sends data to Sink node, sensor network is with periodic manner work, and it is one-period that each sensor node completes the time that a perception and data transmission work experiences;

It is characterized in that: method comprises the steps:

Step 1: sensor network is started working;

Step 2: after sensor network work one-period, the dump energy of each sensor node in calculating sensor network;

Step 3: according to the dump energy of the each sensor node calculating, judged whether sensor node depleted of energy and lost efficacy: be, the detection probability of calculating sensor network, and perform step four; No, sensor network works on, and returns to step 2;

Step 4: judge whether to occur covering cavity according to the detection probability of the sensor network calculating: be to calculate and cover empty position, and perform step five; No, sensor network works on, and returns to step 2;

Step 5: in conjunction with energy efficiency, current effective sensor node is screened, delete the sensor node that does not meet energy efficiency requirement;

Described combination energy efficiency screens current effective sensor node, deletes the sensor node that does not meet energy efficiency requirement, specifically carries out as follows:

Step 1: the variation of sensor network rehabilitation cost when computing node deletion ratio changes;

Knot removal ratio is the ratio that needs the shared current effective sensor node of the sensor node that does not meet energy efficiency of deleting;

Step 2: the variation of sensor network waste energy when computing node deletion ratio changes;

Step 3: the sensor network rehabilitation cost calculating and sensor network waste energy are normalized;

Step 4: draw sensor network rehabilitation cost and sensor network energy waste change curve according to the data after normalization under the same coordinate system, get abscissa value that two intersections of complex curve are corresponding as adopted knot removal ratio, and then determine and need the node number of deleting;

Step 6: to meeting the position in residue node calculating sensor network coverage cavity of energy efficiency requirement.

2. the sensor network of combination energy efficiency according to claim 1 covers empty detection method, it is characterized in that: the dump energy of each sensor node in the calculating sensor network described in step 2, adopts energy consumption model.