CN102647726A - Balancing optimizing strategy for energy consumption of coverage of wireless sensor network - Google Patents

Abstract

In a wireless sensor network, the coverage degree of a target area and the energy consumption of the network are important indexes for measuring the property of the wireless sensor network. The reasonable arrangement of nodes is in favor of ensuring the network coverage and balancing the energy consumption of the network. Specific to the wireless sensor network with an adjustable node sensing distance, the invention provides a balancing optimizing strategy for the energy consumption of the coverage of the wireless sensor network. According to the strategy, meeting a certain coverage quality of network area is taken as a condition, the balancing of the energy consumption of the coverage is taken as an optimizing target, and a particle swarm optimization is adopted for firstly dynamically optimizing the layout of the nodes in the network and then accordingly and reasonably adjusting the node sensing distance, thereby causing the property of the energy consumption of the coverage of the network to be optimal. A simulated result shows that compared with a traditional energy-saving coverage scheme, the strategy has the advantages that the sensing overlapping area and the sensing blind area are effectively reduced, the coverage quality of network area is increased and the energy consumption of the network is reduced.

Description

A kind of wireless sensor network covers energy consumption balance optimizing strategy

Technical field

The power management technique of the technical field of the energy consumption management method of wireless sensing network system, particularly mobile node.

Background technology

In wireless sensor network, covering and energy consumption are one group of contradiction, and promptly along with the level of coverage raising, energy consumption also increases thereupon, and how research reduces the network total energy consumption when satisfied regional covering requires, and it is significant promptly to study covering energy consumption balance optimizing.

Usually the covering energy consumption balance optimizing method that adopts has: the redundant deployment node, according to the target real time status, make the node energy-conservation covering scheme of " enlivening " and " dormancy " by turns, and this is a cost to drop into a large amount of redundant nodes; Carry out dynamic layout optimization through the adjustment node location,, still can produce unnecessary energy consumption if still have more perception overlapping region after the node location adjustment; For the stationary nodes of disposing at random, reach energy-conservation covering through the perceived distance of regulating node, if but the node skewness will reduce energy-conservation coverage effect.

Summary of the invention

Wireless sensor network is made up of the sensor node that is deployed in a large number in the observing environment, and layout is flexible, the associated treatment ability is strong, is widely used in fields such as target following, environmental monitoring, Industry Control.Network is to weigh the important indicator of its performance to the level of coverage and the network energy consumption of target area.Generally, the network coverage and energy consumption are one group of contradiction, and promptly along with level of coverage improves, energy consumption is increase thereupon also, so how research cuts down the consumption of energy when satisfied zone covers requirement, i.e. research covering energy consumption balance optimizing is significant.

Usually in wireless sensor network, adopt the redundant deployment node,, make the node energy-conservation covering scheme of " enlivening " and " dormancy " by turns according to the target real time status; In recent years, the dynamic layout optimization of wireless sensor network obtains extensive concern, and these class methods are optimization aim according to monitoring of environmental to improve the network area coverage rate, dynamically realize the wireless sensor network layout optimization, reduce the network energy consumption.Appearance along with the adjustable sensor node of perceived distance; The method that covers energy consumption based on the balance of node perceived distance adjustment also draws attention gradually; These class methods mainly are through setting rational node perceived distance, making network when satisfying the covering requirement, cut down the consumption of energy.

The present invention is directed to the adjustable wireless sensor network of node perceived distance, proposes a kind ofly, realize the dynamic network layout optimization strategy of covering, energy consumption balance based on particle cluster algorithm.At first optimize the layout of node in the network, reduce perception overlay region and perception blind area, improve regional covering quality through particle cluster algorithm; Under the prerequisite that guarantees certain regional covering quality,, further reduce the perception overlay region then, reduce the network energy consumption, prolong network life through adjustment node perceived distance.

Description of drawings

Fig. 1 is emulation experiment 1 result's a sketch map

Fig. 2 is emulation experiment 2 sketch map as a result

Fig. 3 is emulation experiment 3 sketch map as a result

Fig. 4 is the The simulation experiment result sketch map of three kinds of optimization methods

Embodiment

With regard to this tactful practical implementation process and through accompanying drawing and concrete emulation experiment the present invention is done further detailed explanation below.

Step 1: based on the dynamic layout optimization of the wireless sensor network of particle cluster algorithm.

Process is following:

1) intiating radio sensor network sensor node position, the perception radius r of each node _{I (i=1..., N)}=r, r ∈ R, and be a fixed value, the initial coverage rate of computing network.

2) establishing population scale is m, generates the initial position and the speed of each particle, and the individual adaptive optimal control value of particle and the initial coverage rate that global optimum's adaptive value is network are set.

3) calculate each particle adaptive value according to wireless sensor network covering performance evaluation method.

4) current adaptive value and individual adaptive optimal control value, the global optimum's adaptive value with each particle compares, and upgrades individual optimal location p according to formula (11)-(12) _Id, the position p of global optimum _Gd, present speed v _IdWith current location x _Id

5) if reach stop condition (reaching preset maximum iteration time MaxDT or preset adaptive value) then stop, the output operation result, otherwise return step 3).

Step 2: cover energy consumption balance optimizing algorithm design.

On to the wireless sensor network cloth basis that dynamically office optimizes, cover the energy consumption balance optimizing.If sensor node is optimized the completing place migration according to network topology, the adjusting range of each node perceived radius is 0-r (r is a fixed value that is provided with in the dynamic layout optimization of network), maximal rate limit v _Maxr=0.1 * r works as v _Idr＞v _Maxr, make v _Idr=v _MaxrWork as v _Idr＜v _MaxrThe time, then make v _Idr=-v _MaxrWhen particle leaves the search volume, if x _Idr＞r establishes x _IdrIf=r is x _Idr＜r then establishes x _Idr=0.Wireless sensor network covering energy consumption balance optimizing algorithm design based on particle cluster algorithm is following:

1) initialization sensor node perception radius for guaranteeing certain initial coverage rate, avoids being absorbed in local optimum, and it is the random number between the r-r/2 that the node perceived radius is set, and coverage rate threshold value C is set _Th, it is 0～r apart from adjustable extent R that node perceived is set.

2) calculate behind dynamic layout optimization, cover energy consumption equilibrium valve φ.

3) set population scale m, generate the initial position and the speed of each particle, individual adaptive optimal control value and global optimum's adaptive value of particle is set, these two values are the covering energy consumption equilibrium valve φ of network behind dynamic layout optimization.

4) covering and cover the energy optimization method of evaluating performance according to wireless sensor network calculates each particle coverage rate and covers energy consumption balance optimizing adaptive value.

5) with each particle coverage rate and C _ThRelatively, with the current adaptive value of each particle and individual adaptive optimal control value, global optimum's adaptive value relatively, and upgrade individual optimal location P according to formula (11)-(12) _Idr, the position P of global optimum _Gdr, present speed v _IdrWith current location x _Idr

6) if reach stop condition (reaching preset maximum iteration time MaxDTr or preset adaptive value) then stop, the output operation result, otherwise return step 4).

Step 3: cover the experiment of energy consumption balance optimizing emulation.

Suppose in the length of side to be to dispose N the movable sensor node that perceived distance is adjustable in the square monitored area of 20m, the adjustable range R of perceived distance is 0～r (establishing r=5), each node perceived radius r _i∈ R works as r _i=0 o'clock, node was in resting state.Reliable measurement property parameter r _e=0.5r _i, radio sensor network monitoring areal coverage threshold value C _Th=0.9, the probabilistic model measurement parameter is α ₁=1, α ₂=0, β ₁=1, β ₂=0.5, accelerated factor c ₁=c ₂=1, adjustment coefficient η=1, θ=1, MaxDT=500, MaxDTr=300, adopting dominant frequency is PC emulation under the Matlab environment of 2.26GHz.

Experiment 1.If N=20, the initial perception radius r of each node _{I (i=1..., N)}=r, initialization is disposed sensor node position, back at random shown in Fig. 1 (a), the prime area coverage rate Q of network _Area=86%, cover energy consumption equilibrium valve φ=0.0086.Adopting particle cluster algorithm to carry out dynamic layout optimization, is optimization aim with the areal coverage of network, adjustment sensor node position; On this basis, satisfying Q _Area>=C _ThPrerequisite under, with φ _OptBe optimization aim, adopt particle cluster algorithm to cover the energy consumption balance optimizing, adjustment sensor node perceived distance.The layout situation of optimizing sensor node in the monitored area, back is shown in Fig. 1 (b); Node comparatively is evenly distributed in the monitored area; Wherein have 8 nodes to be in resting state, other 12 nodes are in the perception state and perceived distance has nothing in common with each other, the coverage rate Q of network _Area=90%, cover energy consumption balance optimizing value φ _Opt=0.0162.

Experiment 2.If N=25, the initial perception radius r of each node _{I (i=1..., N)}=r, initialization is disposed sensor node position, back at random shown in Fig. 2 (a), the prime area coverage rate Q of network _Area=87.81%, cover energy consumption equilibrium valve φ=0.0070.Cover behind the energy consumption balance optimizing in the monitored area layout situation of sensor node shown in Fig. 2 (b); Have 14 nodes and be in resting state; The node that is in resting state has increased by 6 than experiment 1, and 11 nodes are in the perception state of different perceived distance, the coverage rate Q of network _Area=90%, cover energy consumption balance optimizing value φ _Opt=0.0151.

Experiment 3.If N=30, the initial perception radius r of each node _{I (i=1..., N)}=r, initialization is disposed sensor node position, back at random shown in Fig. 3 (a), the prime area coverage rate Q of network _Area=89.6%, cover energy consumption equilibrium valve φ=0.0060.Cover behind the energy consumption balance optimizing in the monitored area layout situation of sensor node shown in Fig. 3 (b); Have 18 nodes and be in resting state; The node that is in resting state has increased by 4 again than experiment 2, and 12 nodes are in the perception state of different perceived distance, the coverage rate Q of network _Area=90%, cover energy consumption balance optimizing value φ _Opt=0.0158.

Visible from above experimental result, adopt covering energy consumption balance optimizing can effectively reduce perception blind area and perception overlay region, promote and cover the energy consumption equilibrium valve.

For further checking covers the validity of energy consumption balance optimizing, adopt following three kinds of optimization methods, separately network energy consumption when relatively reaching the same area coverage rate.

The inventive method (method 1).

According to Lin Zhuliang; Feng Yuanjing.Optimization strategy of wireless sensor networks coverage based on particle swarm algorithm [J] .Computer Simulation; 2009,26 (4): the wireless sensor network layout optimization thought that 190-193. proposes, adopt particle cluster algorithm to carry out dynamic layout optimization; Through adjustment sensor node position, improve the coverage rate (method 2) in network monitor zone.Sensor node perception radius r (r ∈ R) according to the areal coverage target that will reach set.

According to Wu J and Yang S.Coverage issue in sensor networks with adjustable ranges [C] .International Conferences on Parallel Processing Workshops; Montreal; Quebec, the energy-conservation coverage optimization thought based on the node perceived distance adjustment that August2004:61-68. proposes is for the stationary nodes of disposing at random; Through regulating the perceived distance of node, make the coverage rate Q in network monitor zone _Area>=C _ThPrerequisite under, network energy consumes minimum (method 3).

In the length of side is to dispose N the movable sensor node that perceived distance is adjustable in the square monitored area of 20m; Experiment parameter is with 4.1; When N=20, N=25, N=30, adopt above-mentioned three kinds of optimization methods to carry out the independent optimization emulation experiment 20 times respectively, experimental result is as shown in Figure 4.

Visible by Fig. 4, when disposing sensor node number less (N=20) in the monitored area, there is more perception blind area in node, Q for disposing at random in the method 3 _AreaMaximum can only reach about 90%, and method 2 is effective than method 3; Deployment sensor node number increase in the monitored area (N=25, N=30), when areal coverage requires to hang down; The perceived distance adjustment of method 3 can effectively reduce the perception overlapping region, reduce the network energy consumption; Cover energy consumption balance optimizing value and be superior to method 2, along with the increase of areal coverage, the existence of in the method 3 because perception blind area; Advantage progressively disappears, and method 2 is superior to method 3; Employing method 1 can effectively reduce perception overlapping region and perception blind area, and it covers energy consumption balance optimizing value and is higher than conventional method (method 1, method 2) all the time, is the coverage rate Q in zone owing to cover energy consumption balance optimizing value _AreaWith the ratio of network total energy consumption E, so when areal coverage was identical, the network total energy consumption E of method 1 was minimum, thereby proved should strategy validity.

Claims (2)

1. a wireless sensor network covers energy consumption balance optimizing strategy; It is characterized in that in this specific wireless sensing network system; Stationary nodes in the network provides a kind of preactivate mechanism, and mobile node is guaranteed normal communication simultaneously through the lifetime that the machine-processed sleep cycle of regulating self of this preactivate obtains maximum.

Assumed wireless sensor network monitoring zone A is a two dimensional surface, on this zone, throws in N identical sensor node of parameter, the node perceived radius r, and communication range C, sensor node is expressed as K{k ₁, k ₂..., k _N.If sensor node k _iThe position be (x _i, y _i), monitoring objective M be positioned at (x, y), then monitoring objective M and sensor node k _iDistance do

In 0～1 sensor model, node is to being 1 with the perceived quality in the disc area that is radius as the center, with the perceived distance, and the perceived quality outer to disk is 0, k _iMonitoring Probability p to target _Xy(k _i) be:

In the practical application, because monitoring of environmental and noise jamming, the sensor node measurement model should be the probability distribution [9-10] of certain characteristic, that is:

$p_{\mathrm{xy}}\left(k_i\right)=\left\{\begin{array}{cc}0,& r+r_e\&le;d(k_i,M)\\ e^{(-{\mathrm{\&alpha;}}_1{{\mathrm{\&lambda;}}_1}^{{\mathrm{\&beta;}}_1})/{{\mathrm{\&lambda;}}_2}^{{\mathrm{\&beta;}}_2}+{\mathrm{\&alpha;}}_2},& r-r_e<d(k_i,M)<r+r_e\\ 1,& d(k_i,M)\&le;r-r_e\end{array}\right.---\left(2\right)$

Wherein, r _e(0＜r _e＜r) be sensor node reliable measurement property parameter, α ₁, α ₂, β ₁, β ₂Be the measurement parameter relevant with the sensor node characteristic, λ ₁And λ ₂Be input parameter:

λ ₁＝r _e-r+d(k _i，M) (3)

λ ₁＝r _e+r-d(k _i，M) (4)

For improving the target measurement probability, need to adopt a plurality of sensor nodes measurement target simultaneously.The combined measurement probability is following:

$P_{\mathrm{xy}}\left(k_{\mathrm{ov}}\right)=1-\underset{k_i\&Element;k_{\mathrm{ov}}}{\mathrm{\&Pi;}}(1-P_{\mathrm{xy}}\left(k_i\right))---\left(5\right)$

For estimating the areal coverage of wireless sensor network; Regional A to be measured is divided into m * n grid; Its granularity (being the distance between adjacent mesh) is determined by solving precision; Experiment shows that when granule size was the 4%-0.25% of area size to be measured, the absolute deviation between calculated value and the exact value was about 0.5%-0.1%.The areal coverage Q of network _Arer(K) weigh by the combined measurement probability of each grid, Q _Arer(K) be defined as the combined measurement probability sum of each grid and the ratio of grid number, that is:

$Q_{\mathrm{arer}}\left(K\right)=\frac{\mathrm{\&Sigma;}{P}_{\mathrm{xy}}\left(k_{\mathrm{ov}}\right)}{m\&times;n}---\left(6\right)$

Generally speaking, the energy consumption of sensor node is by monitoring energy consumption e _M, calculate energy consumption e _CWith the energy consumption e that communicates by letter _TThree parts are formed, monitoring energy consumption e _MBe the function of perceived distance, functional form is determined by the node characteristic.The monitoring energy consumption of node is the θ power function of perceived distance, θ＞0, node k _iPerceived distance be designated as r _i, node k so _iMonitoring energy consumption and perceived distance satisfy following relation:

$e_M\left(r_i\right)=\mathrm{\&eta;}\&times;r_i^{\mathrm{\&theta;}}---\left(7\right)$

Wherein η is adjustment coefficient, η＞0.The perceived distance of node is big more, and the monitoring energy consumption of node is many more.

Formula (7) has provided the energy consumption model of individual node, and the monitoring energy consumption sum that whole radio sensor network monitoring energy consumption is each node in the network is expressed as following functional form:

$E_M(r_i,...,r_N)=\underset{i=1}{\overset{N}{\mathrm{\&Sigma;}}}{e}_M\left(r_i\right)---\left(8\right)$

The aim that this patent covers energy optimization is through setting rational node location and perceived distance, make the coverage rate in radio sensor network monitoring zone be not less than under the prerequisite of certain predetermined value, and the network monitor energy consumption is minimum.Here adopt φ to weigh wireless sensor network and cover energy consumption balance selection scheme:

$\mathrm{\&phi;}=\frac{Q_{\mathrm{arer}}\left(K\right)}{E_M(r_i,...,r_N)},r_i\&Element;R,i=1,...,N---\left(9\right)$

Wherein R representes the adjustable extent of perceived distance.If use C _ThExpression radio sensor network monitoring areal coverage threshold value, wireless sensor network covers energy consumption balance optimizing problem and can be abstracted into like drag:

${\mathrm{\&phi;}}_{\mathrm{opt}}=\max \left[\mathrm{\&phi;}\right]=\max \left[\frac{Q_{\mathrm{arer}}\left(K\right)}{E_M(r_i,...,r_N)}\right],Q_{\mathrm{area}}\left(K\right)\&GreaterEqual;C_{\mathrm{th}}---\left(10\right)$

Be simplified model, do following hypothesis to the wireless sensor network characteristics:

Step 1: wireless sensor network comprises a sink node, has stronger disposal ability, is used for realizing covering the energy consumption balance optimizing.

Step 2: all the sensors node can obtain self-position, also can in interval R, regulate perceived distance continuously.

Step 3: sensor node possesses locomotivity and perceived distance adjustment capability.

Step 4: particle cluster algorithm is a kind of optimized Algorithm based on iteration pattern, and algorithm is simple, speed is fast, separate the quality height, robustness is good.In particle cluster algorithm, each individuality is called one " particle ", and each particle is being represented potential separating.Each particle all has following a few category information: the current present position of particle, and the optimal location of promptly up to the present finding (Pbest) by oneself, this information is regarded as self flying experience of particle; Up to the present the optimal location (Gbest) (Gbest is the optimal value of Pbest) that whole colony finds, this can be considered, and the companion shares flying experience in the population; The movement velocity of each particle receives self and the influence of the historical movement state information of colony; And come the current direction of motion and movement velocity are influenced with self and colony historical optimal location, coordinated the relation between particle displacement and the group movement well.

If colony is made up of m particle, the coordinate of each particle in the D dimension space can be expressed as x _i=(x _I1, x _I2... x _Id, x _ID), particle i (i=1,2 ..., speed definition m) is the distance that particle moves in each iteration, uses v _i=(v _I1, v _I2... v _Id... v _ID) expression.Particle is according to following formula renewal speed and position in each iteration:

v _id＝ωv _id+c ₁r ₁(p _id-x _id)+c ₂r ₂(p _gd-x _id) (11)

x _id＝x _id+v _id (12)

In the formula (11): r ₁And r ₂Be the random number between [0,1], these two parameters are used for keeping colony's diversity; c ₁And c ₂Be accelerated factor; p _IdBe the historical optimal location of current particle, p _GdHistorical optimal location for whole population; ω is an inertia weight.

Adopt the areal coverage of gridding method evaluating network, suppose that monitored area A is the square area of L * L, becomes equal and opposite in direction, area to be the grid of (L/20) * (L/20) this area dividing.At N node of regional A deployed, random initializtion sensor node position, the prime area coverage rate of network is:

$Q_s\left(K\right)=\frac{\mathrm{\&Sigma;}{p}_{\mathrm{xys}}\left(k_{\mathrm{ov}}\right)}{L\&times;L}---\left(13\right)$

In the formula (13), ∑ P _Xys(k _Ov) be the combined measurement probability sum of each grid under the initial condition.

Particle cluster algorithm is incorporated in the dynamic layout optimization search volume dimension D=2 * N of algorithm.Q (K) is the adaptive value function of algorithm, establishes Q (K)=Q when initial _s(K), each particle is the body one by one that has D dimension sensor node, and different individualities has different positions, and different positions is corresponding to different areal coverage Q _Arer(K), particle's velocity and position are able to upgrade through continuous iteration, make node location distribute and tend to be reasonable combined measurement probability sum ∑ P _Xy(k _Ov) increase areal coverage Q _Arer(K) increase thereupon.Work as Q in each iteration _Arer(K)＞during Q (K), the adaptive value function is adjusted into Q (K)=Q _Arer(K), so move in circles, make formula (14) set up:

$\left\{\begin{array}{c}Q\left(K\right)=\max \left(Q_{\mathrm{arer}}\left(K\right)\right)\\ Q\left(K\right)>Q_s\left(K\right)\end{array}\right.---\left(14\right)$

This moment and Q (K) corresponding population colony's optimal location (Gbest) are the position of sensor node behind the dynamic layout optimization.

For algorithm can comparatively fast be focused near the globally optimal solution in the early stage, the later stage then can be made as a linear in time function that reduces with the ω in the formula (11) in local convergence to globally optimal solution:

$\mathrm{\&omega;}\left(t\right)=1.0-\frac{t}{\mathrm{MaxDT}}\&times;0.4---\left(15\right)$

Wherein, MaxDT is total iterations of algorithm, and t is the current iteration number of times.

If v _MaxBe the maximal rate limit of certain dimension, v _MaxBeing worth too conference causes particle to be skipped preferably separating the too little insufficient search that can cause again the space of value.Here establish v _Max=0.1 * x _Max, work as v _Id＞v _MaxThe time, make v _Id=v _MaxWork as v _Id＜-v _MaxThe time, then make v _Id=-v _MaxThe position is updated to when particle leaves the search volume:

Following based on the dynamic layout optimization process of the wireless sensor network of particle cluster algorithm:

Step 1: intiating radio sensor network sensor node position, the perception radius r of each node _{I (i=1..., N)}=r, r ∈ R, and be a fixed value, the initial coverage rate of computing network.

Step 2: establishing population scale is m, generates the initial position and the speed of each particle, and the individual adaptive optimal control value of particle and the initial coverage rate that global optimum's adaptive value is network are set.

Step 3: calculate each particle adaptive value according to wireless sensor network covering performance evaluation method.

Step 4: current adaptive value and individual adaptive optimal control value, global optimum's adaptive value of each particle are compared, and upgrade individual optimal location p according to formula (11)-(12) _Id, the position p of global optimum _Gd, present speed v _IdWith current location x _Id

Step 5: if reach stop condition (reaching preset maximum iteration time MaxDT or preset adaptive value) then stop, the output operation result, otherwise return step 3).

2. on the basis that dynamic office optimizes to wireless sensor network cloth, cover the energy consumption balance optimizing.If sensor node is optimized the completing place migration according to network topology, the adjusting range of each node perceived radius is 0-r (r is a fixed value that is provided with in the dynamic layout optimization of network), maximal rate limit v _Maxr=0.1 * r works as v _Idr＞v _Maxr, make v _Idr=v _MaxrWork as v _Idr＜v _MaxrThe time, then make v _Idr=-v _MaxrWhen particle leaves the search volume, if x _Idr＞r establishes x _IdrIf=r is x _Idr＜r then establishes x _Idr=0.Wireless sensor network covering energy consumption balance optimizing algorithm design based on particle cluster algorithm is following:

Step 1: initialization sensor node perception radius, for guaranteeing certain initial coverage rate, avoid being absorbed in local optimum, it is the random number between the r-r/2 that the node perceived radius is set, and coverage rate threshold value V is set _Th, it is 0～r apart from adjustable extent R that node perceived is set.

Step 2: calculate behind dynamic layout optimization, cover energy consumption equilibrium valve φ.

Step 3: set population scale m, generate the initial position and the speed of each particle, individual adaptive optimal control value and global optimum's adaptive value of particle is set, these two values are the covering energy consumption equilibrium valve φ of network behind dynamic layout optimization.

Step 4: cover and cover the energy optimization method of evaluating performance according to wireless sensor network and calculate each particle coverage rate and cover energy consumption balance optimizing adaptive value.

Step 5: with each particle coverage rate and C _ThRelatively, with the current adaptive value of each particle and individual adaptive optimal control value, global optimum's adaptive value relatively, and upgrade individual optimal location P according to formula (11)-(12) _Idr, the position P of global optimum _Gdr, present speed v _IdrWith current location x _Idr

Step 6: if reach stop condition (reaching preset maximum iteration time MaxDTr or preset adaptive value) then stop, the output operation result, otherwise return step 4).

Images