CN106358210A - Mixed strategy-based heterogeneous directional sensor network dynamic coverage method - Google Patents

Abstract

The invention discloses a mixed strategy-based heterogeneous directional sensor network dynamic coverage method. The method comprises the following steps of: classifying nodes as boundary nodes or non-boundary nodes according to distribution of the nodes, and then adjusting a perception direction of the nodes to reduce coverage redundancy; and further detecting the existence of a coverage hole, calculating optimum positions of the nodes if a coverage hole exists, and finally moving the nodes to the optimum positions. The method has the following beneficial effects: by adopting the mixed strategy-based heterogeneous directional sensor network dynamic coverage method disclosed by the invention, the coverage redundancy is reduced, and redundant nodes are moved to the optimum positions to repair the coverage hole. The result shows that a proposed algorithm, compared with a comparison algorithm, can effectively improve the network coverage rate and reduce energy consumption.

Description

Isomery directional sensor network dynamic coverage method based on mixed strategy

Technical field

The present invention relates to field of intelligent control, more particularly to a kind of isomery directional sensor network based on mixed strategy Dynamic coverage method.

Background technology

In recent years, with the development of hardware technology, the sensor such as video sensor, ultrasonic sensor and infrared sensor Price is more and more cheaper so that the application of directional sensor network is more and more extensive, in video monitoring, smart home and environment prison The occasions such as survey are used widely.

Cover the basic problem controlling as directional sensor network, be the important finger of reflection network service quality Mark, is paid attention to by more and more researcheres in recent years.Different from traditional omnidirectional's sensor node, oriented sensor node Overlay area depends not only on the position of node, but also relevant with perceived direction with the perception angle of node.

The dynamic coverage algorithm of directional sensor network refers to the physics of perceived direction by concept transfer or node The purpose of the network coverage is improved in position to reach.In recent years, the dynamic coverage algorithm of directional sensor network is subject to increasingly The attention of many researcheres.Although studying to directional sensor network covering problem, and achieve certain achievement, existing Achievement in research all assume that participate in cover oriented sensor node type identical, that is, the perception radius of node, perception angle, shifting The parameters such as kinetic force are all identical, have ignored the impact to directional sensor network covering performance for the node isomerism and have The shortcoming of network scalability difference.

Meanwhile, existing dynamic coverage algorithm only relies on merely the adjustment of perceived direction or changing of node physical location Become and then reach the purpose strengthening network covering property, fail fully to combine both the impact considering to covering performance. In addition, in some scenes, the adjustment relying solely on node perceived direction is likely to result in the phenomenon of covering " empty ".

Content of the invention

For solving above technical problem, the present invention provides a kind of isomery directional sensor network based on mixed strategy dynamic Covering method, by adjusting the perceived direction of node, reduces overlapping area in overlay area for the node, by adjusting node Physical location, reduces unlapped region area in overlay area.

Technical scheme is as follows:

A kind of isomery directional sensor network dynamic coverage method based on mixed strategy, it is critical only that and walks including following Rapid:

Step 1, the monitored area of sensor network is divided into n area identical grid, and obtains covering of each grid Lid situation；

Step 2, to the node s in monitored area_iClassified, determined node s_iBe boundary node be also non-boundary node, Node s_iRepresent i-th sensor in monitored area, i=1,2,3 ... h, h are the sensor sum in monitored area；

Step 3, predicate node s_iThe need of adjustment direction, if not needing to adjust, node s_iKeep former direction, enter Step 5；If desired adjust, then enter step 4；

Step 4, predicate node s_iWhether it is redundant node, if not redundant node, then by node s_iIt is adjusted to optimal perceived Direction, enters step 5 afterwards；If redundant node, then enter step 5；

Step 5, makes i=i+1, judges whether i is more than h, if i is more than h, enters step 6, otherwise, return to step 2；

Step 6, judges whether have grid not by any node perceived, if all grids are all by node sense in monitored area Know, then in holding sensing network, the original position of all nodes is constant, and algorithm terminates；Under conditions of there is redundant node, if having Grid by any node perceived, does not then enter step 7；

Step 7, adjusts the physical location of all redundant nodes, selects to make the maximum physical location of areal coverage as superfluous The optimum position of remaining node.

Using said method, by monitored area is divided into several area identical grids, and according to each grid The perceived direction to determine each node for the coverage condition, taken into full account the impact to network covering property for the node isomerism, The autgmentability of network is made to be improved.And the adjustment of perceived direction and the adjustment of node physical location are combined, enhances net Network covering performance, it is to avoid unlapped region occurs.

Further, in step 1, the coverage condition of grid includes the node ID of covering grid, whether grid is coated to The situation whether lid, grid are repeated to cover.

Using said method, it is easy to the coverage condition of each grid is counted.

Further, pass through node s in step 2_iCome to node s with monitored area border nearest Euclidean distance d_i Classified, if Euclidean distance d is less than threshold value q, node s_iFor boundary node, otherwise it is non-boundary node.

Using said method, node can be divided into boundary node and non-boundary node, the node different to two classes be not using Same method carries out perceived direction adjustment, so that node perceived direction is adjusted more accurate.

Further, include determining whether in step 3 that boundary node the need of adjustment perceived direction and judges non-border section Point is the need of adjustment perceived direction:

1) Decision boundaries node is made whether to need to adjust perceived direction:

A, calculates the perception area c of boundary node_i, draw sense by calculating the grid quantity in boundary node coverage Know area c_i；

B, according to below equation Decision boundaries node the need of adjustment perceived direction:

Wherein, p is predetermined threshold value,For node s_iPerception angle,For node s_iThe perception radius, s_boundary=1 Represent node s_iNeed to adjust perceived direction, s_boundary=0 expression does not need to adjust perceived direction；

2) judge that non-boundary node is made whether to need to adjust perceived direction:

Repeat the grid number covering in a, the coverage of the non-boundary node of statistics, that is, count what non-boundary node was covered In grid, repeated the grid sum covering by other nodes；

Whether the grid sum that b, judgement repeat to cover exceedes preset value θ_tIf exceeding, non-boundary node needs to adjust Perceived direction, conversely, then do not need to adjust perceived direction.

Using said method, threshold determination is carried out to boundary node and non-boundary node it is not necessary to adjust perceived direction Node adjusts without travel direction, it is to avoid flog a dead horse.Because the coverage condition of boundary node and non-boundary node is different, institute To be judged to boundary node and non-boundary node using different methods, more accurately.

Further, include determining whether in step 4 whether boundary node is redundant node and whether judges non-boundary node Step for redundant node；

Firstth, whether Decision boundaries node is redundant node:

Step 4.1, calculates marginal coverage effectiveness s of boundary node_a, i.e. a grid number being perceived by boundary node；

Step 4.2, boundary node is rotated a time along same direction of rotation with fixed angle bc, and calculates every time postrotational Marginal coverage effectiveness s_a, a=1,2,3 ... a, a=2 × π/bc；

Step 4.3, judges postrotational marginal coverage effectiveness s every time_aWhether it is more than s_aIf being more than, select institute after rotation There is marginal coverage effectiveness s_aIntermediate value maximum corresponding sensing direction in border is optimal perceived direction；Otherwise, described boundary node is superfluous Remaining node；

Secondth, judge non-boundary node whether as redundant node:

Step 4.1a, marginal coverage effectiveness s of the non-boundary node of calculating_b, that is, only by the grid of non-boundary node itself perception Lattice number；

Step 4.1b, non-boundary node obtain the adjacent segments points of itself, that is, in the communication radius of non-boundary node The quantity of node；

Adjacent segments are counted and are sent aggregation node by step 4.1c, all non-boundary nodes, and aggregation node obtains overlay area Adjacent node sum；

Step 4.1d, aggregation node draw the average nodal number of overlay area according to adjacent node sum and are sent to each Individual non-boundary node；

Step 4.1e, each non-boundary node judge whether average nodal number is compared more than the adjacent segments points of itself Relatively；

If adjacent segments points are more than average nodal number, this non-boundary node is in Node distribution close quarters, described non- Boundary node rotates 2 × π/small_angle time along same direction of rotation with fixed angle small_angle, and calculates each Marginal coverage effectiveness s of non-boundary node after rotation_b1；

B1=1,2,3 ... 2 × π/small_angle；

If adjacent segments points are less than average nodal number, this non-boundary node is in Node distribution sparse region, described non- Boundary node rotates 2 × π/big_angle time along same direction of rotation with fixed angle big_angle, and calculates and rotate every time Marginal coverage effectiveness s of non-boundary node afterwards_b2, b2=1,2,3 ... 2 × π/big_angle；

Step 4.1f, judgement postrotational marginal coverage effectiveness s every time_b1、s_b2Whether it is more than s_bIf being more than, select rotation All marginal coverage effectiveness s after turning_b1、s_b2Middle maximum corresponding sensing direction is optimal perceived direction；Otherwise described non-border Node is redundant node.

Using said method, because boundary node is different to the coverage condition of grid with non-boundary node, using not Same method is screened to the redundant node in boundary node and non-boundary node respectively, and adjusts the perceived direction of node, Make the perceived direction of the node after adjustment more accurate, strengthen the covering performance of network.

Further, judge whether non-boundary node adopts following methods to obtain phase as step 4.1b in redundant node Neighbors number:

Node s_iSend a neighbor_discover message in its communication radius, in message, comprise node s_iOnly One id, physical locationThe perception radiusPerception angleAnd perceived directionReceive neighbor_discover message Node s_jThen recovery of node s_iOne ack message, comprises node s in message_iId and node s_jId, node s_iBy statistics The ack message receiving obtains its neighbor node number.

Using said method, the adjacent segments points of each non-boundary node can be obtained, be easy to the follow-up average section to network Points are calculated.

Further, step 7 determines redundant node s using following methods_kOptimum position, redundant node s_kRepresent kth Individual redundant node:

Step 7.1, coding, to all redundant nodes s by the way of real coding_kPosition coordinateses encoded, obtain To each redundant node s_kThe individuality corresponding to position coordinateses, wherein k=1,2,3..., n_m, n_mSum for redundant node；

Step 7.2, set up initial population, random initializtion carried out to individuality, sets up initial population, its formula is as follows:

$x_{g,j}^0=x_{g,j}^l+rand(x_{g,j}^u-x_{g,j}^l),g=1,2......np,j=1,2......2\×n_m;$

Wherein,It is respectively the maximum of jth dimension and the minima of g-th individuality in initial population,For For the jth dimension element of g-th individuality in initial population, np is individual number in initial population, rand be between (0,1) with Machine number；

Step 7.3, mutation operation, enter row variation using de/rand-to-best/1 variation method to initial population, simultaneously Introduce fictitious force the variation direction of initial population is guided, its formula is as follows:

$u_{k,j}^{t+1}=x_{r1,j}^t+f\×(x_{best,j}^t-x_{r1,j}^t)+f\×(x_{r2,j}^t-x_{r3,j}^t)+{\mathrm{ac}}_1\×r_k^t\×a^t$

a^t=[a^t(1),...,a^t(n_m)]

a^t(k)=[ax^t(k),ay^t(k)]

${\mathrm{ax}}^t\left(k\right)=\frac{f_k^x}{f_k}\×maxstep\×e^{\frac{-1}{f_k}}$

${\mathrm{ay}}_k^t\left(k\right)=\frac{f_k^y}{f_k}\×maxstep\×e^{\frac{-1}{f_k}}$

Wherein,Individuality corresponding to areal coverage best in population during the t time iteration, r1, r2, r3 are to plant The three individual sequence numbers randomly choosing in group, r1 ≠ r2 ≠ r3,WithIt is three randomly choosing in population The value of the individual jth dimension element in the t time iteration, f is zoom factor, ac₁WithIt is constant and the variable controlling fictitious force, a^tRepresent the change of all redundant nodes position in the case of suffered fictitious force,Represent redundant node s_kSuffered is virtual Power f_kComponent in x-axis,Represent described redundant node s_kSuffered fictitious force f_kComponent on the y axis, maxstep is Described redundant node s_kThe ultimate range being moved under the guiding of fictitious force；

Step 7.4, crossover operation, carry out crossover operation using below equation to initial population:

$v_{g,j}^{t+1}=\left\{\begin{array}{cc}{u}_{g,j}^{t+1},& \begin{array}{cc}if& (rand\≤crorj=rand(g\left)\right)\end{array}\\ x_{g,j}^t,& otherwise\end{array}\right.;$

Wherein,After representing crossover operation, the jth of g-th individuality ties up the value of element in initial population, cr be intersect because Son, value is (0,1)；Rand (g) is 1 random integers arriving np；

Step 7.5, according to valueCalculate the areal coverage of monitored area

Step 7.6, selection operation, obtain iterationses using below equation individual for g-th during t+1:

$x_g^{t+1}=\left\{\begin{array}{cc}{v}_g^{t+1},& \begin{array}{cc}if& f\left(v_g^{t+1}\right)<f\left(v_g^t\right)\end{array}\\ x_g^t,& otherwise\end{array}\right.;$

Step 7.7, determine whether to reach maximum iteration time, if not having, return to step 7.2, if reaching, select energy Make the areal coverage of monitored area maximum individual, each redundant node using the coordinate of corresponding position in this individuality as The optimum position of oneself.

Using said method, more extensive position coordinateses can be obtained, then select optimal from these position coordinateses Position, the new position coordinateses of redundant node consider more comprehensive, and the optimum position obtaining also more conforms to practical situation.

Further, fictitious force f in step 7.3_kCalculate according to below equation:

$f_{kj}^{\&prime;}=\left\{\begin{array}{cc}\frac{k_1{m}_k{m}_j}{d{(k,j)}^{l_1}},{\mathrm{\&alpha;}}_{kj}& 0<d(k,j)\&le;r_{s_k}\\ 0& d(k,j)>r_{s_k}\end{array}\right.;$

$f_k=\underset{j\&element;neig\_bor\left(s_k\right)}{\&sigma;}{f_{kj}}^{\&prime;}$

f′_kjRepresent redundant node s_kBy its adjacent node s_jFictitious force, m_kRepresent redundant node s_kPerception area, m_jRepresent adjacent node s_jPerception area, α_kjRepresent unit vector, the direction of instruction fictitious force, that is, by redundant node s_kPoint to Adjacent node s_jDirection,Represent redundant node s_kThe perception radius, l₁For constant, neig_bor (s_k) represent node s_kPhase The set of neighbors, redundant node s_kThe fictitious force being subject to is that its all adjacent node applies fictitious force to it and makes a concerted effort.

Using said method, to guide the choice direction of optimum position using fictitious force, to reduce the feelings of data processing amount It is ensured that the accuracy of optimal location under condition.

Further, areal coverage in step 7.5Calculate in accordance with the following methods:

Step 7.51, one-dimension array coverstatus for n for the initialization length is so as to the value of all elements is 0, one-dimensional Element in array is corresponded with grid；Initialization iteration indicator variable n and k, makes n=1, k=1.

Step 7.52, the central point obtaining grid is to described redundant node s_kApart from d_nCentral point and redundancy with grid Node s_kThe position vector and redundant node s that are formed_kPerceived direction between angle α_n；

Step 7.53, judges apart from d_nWhether it is less than described redundant node s_kThe perception radiusIf not less than perception half FootpathThen make n=n+1, that is, next grid is judged, and return to step 7.52；If being less than the perception radius Then enter step 7.54；

Step 7.54, judges central point and redundant node s of grid_kThe position vector and redundant node s that are formed_kSense Know the angle α between direction_nWhether it is less than redundant node s_kPerception angle half, if being less than, this grid is in redundant node s_kSensing range, in one-dimension array coverstatus, the value of element corresponding with this grid is changed into 1, that is, makes Coverstatus (n)=1；Otherwise, this grid is not by redundant node s_kPerception.

Step 7.55, determines whether that all grids all have been carried out judging, that is, judges whether n is more than n, if not right All of grid is judged, i.e. n < n, then return to step 7.52；Otherwise, illustrate that all grids have completed judgement, then enter Enter step 7.56；

Step 7.56, judges to next redundant node, that is, makes k=k+1, judges whether k is more than n_mIf, k More than n_m, then enter step 7.57；Otherwise, return to step 7.52.

Step 7.57, in statistics one-dimension array coverstatus, element value is 1 element sum, and makes the value of l be equal to this It is worth and according to formulaCalculate the areal coverage of monitored area.

Using said method, the coverage condition of each grid can be drawn, thus calculating the areal coverage of each node, It is easy to the coverage condition evaluation to node.

Beneficial effect: using the isomery directional sensor network dynamic coverage method based on mixed strategy for the present invention, subtract Cover redundancy less, mobile for the redundant node position that extremely optimizes is repaired to covering cavity.Result proves, propose algorithm compared to Comparison algorithm can effectively improve the network coverage and reduce energy expenditure.

Brief description

Fig. 1 is method of the present invention flow chart；

Fig. 2 is the method flow diagram determining optimum position in Fig. 1；

Fig. 3 is the computational methods flow chart of areal coverage；

Fig. 4 is sensor senses model schematic；

Fig. 5 is sensor transmissions apart from schematic diagram；

Fig. 6 is individual UVR exposure schematic diagram；

Fig. 7 is network coverage comparison diagram after perceived direction adjustment；

Fig. 8 is final area coverage rate comparison diagram；

Fig. 9 is coverage overlapping area change comparison diagram；

Figure 10 is Node distribution schematic diagram；

Figure 11 is algorithm Energy Expenditure Levels comparison diagram.

Specific embodiment

With reference to embodiment and accompanying drawing, the invention will be further described.

As shown in figs. 1-11, first by h sensor node random placement in monitored area, following steps pair are then adopted Sensor node in monitored area is adjusted:

Step 1, the monitored area of sensor network is divided into n area identical grid, and obtains covering of each grid Lid situation.The coverage condition of grid includes the node ID of covering grid, whether grid is capped, whether grid is repeated to cover Situations such as.

Step 2, by node s_iCome to node s with monitored area border nearest Euclidean distance d_iClassified, if Europe Formula is less than threshold value q, node s apart from d_iFor boundary node, otherwise it is non-boundary node.Node s_iRepresent i-th in monitored area Sensor, i=1,2,3 ... h, h are the sensor sum in monitored area.

Step 3, predicate node s_iThe need of adjustment direction, if not needing to adjust, node s_iKeep former direction, enter Step 5；If desired adjust, then enter step 4；

Predicate node s_iInclude determining whether boundary node the need of adjustment perceived direction and judgement the need of adjustment direction Non- boundary node is the need of adjustment perceived direction:

1) Decision boundaries node is made whether to need to adjust perceived direction:

A, calculates the perception area c of boundary node_i, draw sense by calculating the grid quantity in boundary node coverage Know area c_i；

B, according to below equation Decision boundaries node the need of adjustment perceived direction:

Wherein, p is predetermined threshold value,For node s_iPerception angle,For node s_iThe perception radius, s_boundary=1 Represent node s_iNeed to adjust perceived direction, s_boundary=0 expression does not need to adjust perceived direction；

2) judge that non-boundary node is made whether to need to adjust perceived direction:

Repeat the grid number covering in a, the coverage of the non-boundary node of statistics, that is, count what non-boundary node was covered In grid, repeated the grid sum covering by other nodes；

Whether the grid sum that b, judgement repeat to cover exceedes preset value θ_tIf exceeding, non-boundary node needs to adjust Perceived direction, conversely, then do not need to adjust perceived direction.

Step 4, predicate node s_iWhether it is redundant node, if not redundant node, then by node s_iIt is adjusted to optimal perceived Direction, enters step 5 afterwards；If redundant node, then enter step 5.

Predicate node s_iWhether it is that redundant node includes determining whether whether boundary node is redundant node and judges non-border section Whether point is the step of redundant node；

Firstth, whether Decision boundaries node is redundant node:

Step 4.1, calculates marginal coverage effectiveness s of boundary node_a, i.e. a grid number being perceived by boundary node；

Step 4.2, boundary node is rotated a time along same direction of rotation with fixed angle bc, and calculates every time postrotational Marginal coverage effectiveness s_a, a=1,2,3 ... a, a=2 × π/bc；

Step 4.3, judges postrotational marginal coverage effectiveness s every time_aWhether it is more than s_aIf being more than, select institute after rotation There is marginal coverage effectiveness s_aIntermediate value maximum corresponding sensing direction in border is optimal perceived direction；Otherwise, described boundary node is superfluous Remaining node；

Secondth, judge non-boundary node whether as redundant node:

Step 4.a, marginal coverage effectiveness s of the non-boundary node of calculating_b, that is, only by the grid of non-boundary node itself perception Number；

Step 4.b, non-boundary node obtain the adjacent segments points of itself, i.e. section in the communication radius of non-boundary node The quantity of point；

Adjacent segments are counted and are sent aggregation node by step 4.c, all non-boundary nodes, and aggregation node obtains overlay area Adjacent node sum；

Step 4.d, aggregation node draw the average nodal number of overlay area according to adjacent node sum and are sent to each Individual non-boundary node；

Step 4.e, each non-boundary node judge whether average nodal number is compared more than the adjacent segments points of itself；

If adjacent segments points are more than average nodal number, this non-boundary node is in Node distribution close quarters, described non- Boundary node rotates 2 × π/small_angle time along same direction of rotation with fixed angle small_angle, and calculates each Marginal coverage effectiveness s of non-boundary node after rotation_b1,

B1=1,2,3 ... 2 × π/small_angle；

If adjacent segments points are less than average nodal number, this non-boundary node is in Node distribution sparse region, described non- Boundary node rotates 2 × π/big_angle time along same direction of rotation with fixed angle big_angle, and calculates and rotate every time Marginal coverage effectiveness s of non-boundary node afterwards_b2, b2=1,2,3 ... 2 × π/big_angle.

The present embodiment adopts following methods to obtain adjacent segments and counts:

Node s_iSend a neighbor_discover message in its communication radius, in message, comprise node s_iOnly One id, physical locationThe perception radiusPerception angleAnd perceived directionReceive neighbor_discover message Node s_jThen recovery of node s_iOne ack message, comprises node s in message_iId and node s_jId, node s_iBy statistics The ack message receiving obtains its neighbor node number.

Step 4.f, judgement postrotational marginal coverage effectiveness s every time_b1、s_b2Whether it is more than s_bIf being more than, select rotation All marginal coverage effectiveness s afterwards_b1、s_b2Middle maximum corresponding sensing direction is optimal perceived direction；Otherwise described non-border section Point is redundant node.

Step 5, makes i=i+1, judges whether i is more than h, if i is more than h, enters step 6, otherwise, return to step 2；

Step 6, judges whether have grid not by any node perceived, if all grids are all by node sense in monitored area Know, then in holding sensing network, the original position of all nodes is constant, and algorithm terminates；Under the conditions of there is redundant node, if there being grid Lattice by any node perceived, then do not enter step 7；

Step 7, adjusts the physical location of all redundant nodes, selects to make the maximum physical location of areal coverage as superfluous The optimum position of remaining node.

The present embodiment determines redundant node s using following methods_kOptimum position, redundant node s_kRepresent k-th redundancy section Point:

Step 7.1, coding, to all redundant nodes s by the way of real coding_kPosition coordinateses encoded, obtain To each redundant node s_kThe individuality corresponding to position coordinateses, wherein the initial value of k be 1, k=1,2,3..., n_m, n_mFor superfluous The sum of remaining node；

Step 7.2, set up initial population, random initializtion carried out to individuality, sets up initial population, its formula is as follows:

$x_{g,j}^0=x_{g,j}^l+rand(x_{g,j}^u-x_{g,j}^l),g=1,2......np,j=1,2......2\&times;n_m;$

Wherein,It is respectively the maximum of jth dimension and the minima of g-th individuality in initial population,For For the jth dimension element of g-th individuality in initial population, np is individual number in initial population, rand be between (0,1) with Machine number；

Step 7.3, mutation operation, enter row variation using de/rand-to-best/1 variation method to initial population, simultaneously Introduce fictitious force the variation direction of initial population is guided, its formula is as follows:

$u_{k,j}^{t+1}=x_{r1,j}^t+f\&times;(x_{best,j}^t-x_{r1,j}^t)+f\&times;(x_{r2,j}^t-x_{r3,j}^t)+{\mathrm{ac}}_1\&times;r_k^t\&times;a^t$

a^t=[a^t(1),...,a^t(n_m)]

a^t(k)=[ax^t(k),ay^t(k)]

${\mathrm{ax}}^t\left(k\right)=\frac{f_k^x}{f_k}\&times;maxstep\&times;e^{\frac{-1}{f_k}}$

${\mathrm{ay}}_k^t\left(k\right)=\frac{f_k^y}{f_k}\&times;maxstep\&times;e^{\frac{-1}{f_k}}$

Wherein,Individuality corresponding to areal coverage best in population during the t time iteration, r1, r2, r3 are to plant The three individual sequence numbers randomly choosing in group, r1 ≠ r2 ≠ r3,WithIt is three randomly choosing in population The value of the individual jth dimension element in the t time iteration, f is zoom factor, ac₁WithIt is constant and the variable controlling fictitious force, a^tRepresent the change of all redundant nodes position in the case of suffered fictitious force,Represent redundant node s_kSuffered is virtual Power f_kComponent in x-axis,Represent described redundant node s_kSuffered fictitious force f_kComponent on the y axis, maxstep is Described redundant node s_kThe ultimate range being moved under the guiding of fictitious force.

Wherein, fictitious force f_gCalculate according to below equation:

$f_{kj}^{\&prime;}=\left\{\begin{array}{cc}\frac{k_1{m}_k{m}_j}{d{(k,j)}^{l_1}},{\mathrm{\&alpha;}}_{kj}& 0<d(k,j)\&le;r_{s_k}\\ 0& d(k,j)>r_{s_k}\end{array}\right.;$

$f_k=\underset{j\&element;neig\_bor\left(s_k\right)}{\&sigma;}{f_{kj}}^{\&prime;}$

f_k′_jRepresent redundant node s_kBy its adjacent node s_jFictitious force, m_kRepresent redundant node s_kPerception area, m_jRepresent adjacent node s_jPerception area, α_kjRepresent unit vector, the direction of instruction fictitious force, that is, by redundant node s_kPoint to Adjacent node s_jDirection,Represent redundant node s_kThe perception radius, l₁For constant, neig_bor (s_k) represent node s_kPhase The set of neighbors, redundant node s_kThe fictitious force being subject to is that its all adjacent node applies fictitious force to it and makes a concerted effort.

Step 7.4, crossover operation, carry out crossover operation using below equation to initial population:

$v_{g,j}^{t+1}=\left\{\begin{array}{cc}{u}_{g,j}^{t+1},& \begin{array}{cc}if& (rand\&le;crorj=rand(g\left)\right)\end{array}\\ x_{g,j}^t,& otherwise\end{array}\right.;$

Wherein,After representing crossover operation, the jth of g-th individuality ties up the value of element in initial population, cr be intersect because Son, value is (0,1)；Rand (g) is 1 random integers arriving np；

Step 7.5, according to valueCalculate the areal coverage of monitored area

Wherein, areal coverageCalculate in accordance with the following methods:

Step 7.51, one-dimension array coverstatus for n for the initialization length is so as to the value of all elements is 0, one-dimensional Element in array is corresponded with grid；Initialization iteration indicator variable n and k, makes n=1, k=1.

Step 7.52, the central point obtaining grid is to described redundant node s_kApart from d_nCentral point and redundancy with grid Node s_kThe position vector and redundant node s that are formed_kPerceived direction between angle α_n.

Step 7.53, judges apart from d_nWhether it is less than described redundant node s_kThe perception radiusIf not less than perception half FootpathThen make n=n+1, that is, next grid is judged, and return to step 7.52；If being less than the perception radius Then enter step 7.54；

Step 7.54, judges central point and redundant node s of grid_kThe position vector and redundant node s that are formed_kSense Know the angle α between direction_nWhether it is less than redundant node s_kPerception angle half, if being less than, this grid is in redundant node s_kSensing range, in one-dimension array coverstatus, the value of element corresponding with this grid is changed into 1, that is, makes Coverstatus (n)=1；Otherwise, this grid is not by redundant node s_kPerception.

Step 7.55, determines whether that all grids all have been carried out judging, that is, judges whether n is more than n, if not right All of grid is judged, i.e. n < n, then return to step 7.52；Otherwise, illustrate that all grids have completed judgement, then enter Enter step 7.56；

Step 7.56, judges to next redundant node, that is, makes k=k+1, judges whether k is more than n_mIf, k More than n_m, then enter step 7.57；Otherwise, return to step 7.52.

Step 7.57, in statistics one-dimension array coverstatus, element value is 1 element sum, and makes the value of l be equal to this It is worth and according to formulaCalculate the areal coverage of monitored area.

Step 7.6, selection operation, obtain iterationses using below equation individual for g-th during t+1:

$x_g^{t+1}=\left\{\begin{array}{cc}{v}_g^{t+1},& \begin{array}{cc}if& f\left(v_g^{t+1}\right)<f\left(v_g^t\right)\end{array}\\ x_g^t,& otherwise\end{array}\right.;$

Step 7.7, determine whether to reach maximum iteration time, if not having, return to step 7.2, if reaching, select energy Make the areal coverage of monitored area maximum individual, each redundant node s_kMade with the coordinate of corresponding position in this individuality Optimum position for oneself.

Various parameters are configured: b_mi=10, bc=π/12, μ=0.5, and p=60, p '=0.9, big_angle=π/ 36, small_angle=π/180.g=200, np=30, cr=0.6, ac₁=1, f obey n (0,1) normal distribution, l₁=2,.

Fig. 5 is certain individual UVR exposure schematic diagram in population, whereinWithFor needing the section of shift position for i-th The x-axis coordinate of point and y-axis coordinate.

Fig. 6 is under different number of nodes, the situation comparison diagram of the network coverage after perceived direction adjustment.Can from figure Go out, with the increase of node deployment density, after perceiving angle adjustment, the coverage rate of network all increased, algorithm presented herein Covering performance is better than comparison algorithm it was demonstrated that the effectiveness of the adjustment of node perceived angle presented herein.

After Fig. 7 is for node location change, the Landfill covering rate comparison diagram of network.It can be seen that with deployment section The increase counted out, the coverage rate of network increases therewith, and algorithm presented herein is substantially excellent in terms of the final network coverage In comparison algorithm.

Fig. 8 nodes cover the situation of change of overlapping area.It can be seen that propose algorithm comparing than two kinds The redundant cover of node can effectively be reduced compared with algorithm.

Fig. 9 is that the oriented sensor node of 60 isomeries is running this paper algorithm posterior nodal point distribution schematic diagram.A () is 60 sections Point, initial distribution, coverage rate 0.69；B () is the network coverage 0.71 after boundary node adjustment；C () is perception angle adjustment after The network coverage 0.80；D () is 0.90 for the coverage rate of network after node physical location change.Can be seen that from four in figures With the operation of algorithm, the distribution between node tends to reasonable, and overlapping area gradually reduces, coverage rate step up it was demonstrated that The effectiveness of algorithm.

The energy expenditure comparison diagram of Figure 10 algorithm.Using the energy expenditure of the method record node of existing document, that is, The energy that node often rotates 180 degree consumption during perception angle adjustment is 1.5j, in physical location moving process, often moves The consumption of dynamic 1m energy is 3.6j, and result is as shown in Figure 7.It can be seen that the algorithm proposing is bright in terms of energy expenditure Aobvious better than comparison algorithm it was demonstrated that proposing algorithm is a kind of covering algorithm of Energy Efficient.

Finally it should be noted that foregoing description is only the preferred embodiments of the present invention, the ordinary skill people of this area Member, under the enlightenment of the present invention, without prejudice on the premise of present inventive concept and claim, can make table as multiple types Show, such conversion each falls within protection scope of the present invention.

Claims (9)

1. a kind of isomery directional sensor network dynamic coverage method based on mixed strategy is it is characterised in that include following walking Rapid:

Step 1, the monitored area of sensor network is divided into n area identical grid, and obtains the covering feelings of each grid Condition；

Step 2, to the node s in monitored area_iClassified, determined node s_iBe boundary node be also non-boundary node, node s_iRepresent i-th sensor in monitored area, i=1,2,3 ... h, h are the sensor sum in monitored area；

Step 3, predicate node s_iThe need of adjustment direction, if not needing to adjust, node s_iKeep former direction, enter step 5；If desired adjust, then enter step 4；

Step 4, predicate node s_iWhether it is redundant node, if not redundant node, then by node s_iIt is adjusted to optimal perceived side To entrance step 5 afterwards；If redundant node, then enter step 5；

Step 5, makes i=i+1, judges whether i is more than h, if i is more than h, enters step 6, otherwise, return to step 2；

Step 6, judges whether have grid not by any node perceived in monitored area, if all grids are all by node perceived, In holding sensing network, the original position of all nodes is constant, and algorithm terminates；Under conditions of there is redundant node, if having grid not By any node perceived, then enter step 7；

Step 7, adjusts the physical location of all redundant nodes, selects to make the maximum physical location of areal coverage as redundancy section The optimum position of point.

2. the isomery directional sensor network dynamic coverage method based on mixed strategy according to claim 1, its feature exists In: in step 1, the coverage condition of grid includes the node ID of covering grid, whether grid is capped, whether grid is repeated Situation about covering.

3. the isomery directional sensor network dynamic coverage method based on mixed strategy according to claim 1, its feature exists In: pass through node s in step 2_iCome to node s with monitored area border nearest Euclidean distance d_iClassified, if European away from It is less than threshold value q, then node s from d_iFor boundary node, otherwise it is non-boundary node.

4. the isomery directional sensor network dynamic coverage method based on mixed strategy according to claim 1, its feature exists In: include determining whether in step 3 that boundary node the need of adjustment perceived direction and judges non-boundary node the need of adjustment sense Know direction:

1) Decision boundaries node is made whether to need to adjust perceived direction:

A, calculates the perception area c of boundary node_i, draw perception face by calculating the grid quantity in boundary node coverage Long-pending c_i；

B, according to below equation Decision boundaries node the need of adjustment perceived direction:

Wherein, p is predetermined threshold value,For node s_iPerception angle,For node s_iThe perception radius, s_boundary=1 expression Node s_iNeed to adjust perceived direction, s_boundary=0 expression does not need to adjust perceived direction；

2) judge that non-boundary node is made whether to need to adjust perceived direction:

Repeat the grid number covering in a, the coverage of the non-boundary node of statistics, that is, count the grid that non-boundary node is covered In, repeated the grid sum covering by other nodes；

Whether the grid sum that b, judgement repeat to cover exceedes preset value θ_tIf exceeding, non-boundary node needs adjustment perception side To conversely, then not needing to adjust perceived direction.

5. the isomery directional sensor network dynamic coverage method based on mixed strategy according to claim 1, its feature exists In: include determining whether in step 4 that whether boundary node is redundant node and judges the step whether as redundant node for the non-boundary node；

Firstth, whether Decision boundaries node is redundant node:

Step 4.1, calculates marginal coverage effectiveness s of boundary node_a, i.e. a grid number being perceived by boundary node；

Step 4.2, boundary node is rotated a time along same direction of rotation with fixed angle bc, and calculates every time postrotational limit Cover effectiveness s_a, a=1,2,3 ... a, a=2 × π/bc；

Step 4.3, judges postrotational marginal coverage effectiveness s every time_aWhether it is more than s_aIf being more than, select all sides after rotation Border covers effectiveness s_aIntermediate value maximum corresponding sensing direction in border is optimal perceived direction；Otherwise, described boundary node is redundancy section Point；

Secondth, judge non-boundary node whether as redundant node:

Step 4.a, marginal coverage effectiveness s of the non-boundary node of calculating_b, that is, only by the grid number of non-boundary node itself perception；

Step 4.b, non-boundary node obtain the adjacent segments points of itself, i.e. node in the communication radius of non-boundary node Quantity；

Adjacent segments are counted and are sent aggregation node by step 4.c, all non-boundary nodes, and aggregation node obtains the adjacent of overlay area Node total number；

Step 4.d, aggregation node draw the average nodal number of overlay area according to adjacent node sum and to be sent to each non- Boundary node；

Step 4.e, each non-boundary node judge whether average nodal number is compared more than the adjacent segments points of itself；

If adjacent segments points are more than average nodal number, this non-boundary node is in Node distribution close quarters, described non-border Node rotates 2 × π/small_angle time along same direction of rotation with fixed angle small_angle, and calculates and rotate every time Marginal coverage effectiveness s of non-boundary node afterwards_b1；

B1=1,2,3 ... 2 × π/small_angle；

If adjacent segments points are less than average nodal number, this non-boundary node is in Node distribution sparse region, described non-border Node rotates 2 × π/big_angle time along same direction of rotation with fixed angle big_angle, and calculate rotate every time after non- Marginal coverage effectiveness s of boundary node_b2, b2=1,2,3 ... 2 × π/big_angle；

Step 4.f, judgement postrotational marginal coverage effectiveness s every time_b1、s_b2Whether it is more than s_bIf being more than, select institute after rotation There is marginal coverage effectiveness s_b1、s_b2Middle maximum corresponding sensing direction is optimal perceived direction；Otherwise described non-boundary node is Redundant node.

6. the isomery directional sensor network dynamic coverage method based on mixed strategy according to claim 5, its feature exists In: judge whether non-boundary node adopts following methods to obtain adjacent segments as step 4.b in redundant node and count:

Node s_iSend a neighbor_discover message in its communication radius, in message, comprise node s_iUnique Id, physical locationThe perception radiusPerception angleAnd perceived directionReceive neighbor_discover message Node s_jThen recovery of node s_iOne ack message, comprises node s in message_iId and node s_jId, node s_iReceived by statistics To ack message obtain its neighbor node number.

7. the isomery directional sensor network dynamic coverage method based on mixed strategy according to claim 1, its feature exists In: step 7 determines redundant node s using following methods_kOptimum position, redundant node s_kRepresent k-th redundant node:

Step 7.1, coding, to all redundant nodes s by the way of real coding_kPosition coordinateses encoded, obtain each Redundant node s_kThe individuality corresponding to position coordinateses, wherein k=1,2,3..., n_m, n_mSum for redundant node；

Step 7.2, set up initial population, random initializtion carried out to individuality, sets up initial population, its formula is as follows:

$x_{g,j}^0=x_{g,j}^l+rand(x_{g,j}^u-x_{g,j}^l),g=1,2......np,j=1,2......2\&times;n_m;$

Wherein,It is respectively the maximum of jth dimension and the minima of g-th individuality in initial population,For first The jth dimension element of g-th individuality in beginning colony, np is individual number in initial population, and rand is random between (0,1) Number；

Step 7.3, mutation operation, enter row variation using de/rand-to-best/1 variation method to initial population, are simultaneously introduced Fictitious force guides to the variation direction of initial population, and its formula is as follows:

$u_{k,j}^{t+1}=x_{r1,j}^t+f\&times;(x_{best,j}^t-x_{r1,j}^t)+f\&times;(x_{r2,j}^t-x_{r3,j}^t)+{\mathrm{ac}}_1\&times;r_k^t\&times;a^t$

a^t=[a^t(1),...,a^t(n_m)]

a^t(k)=[ax^t(k),ay^t(k)]

${\mathrm{ax}}^t\left(k\right)=\frac{f_k^x}{f_k}\&times;maxstep\&times;e^{\frac{-1}{f_k}}$

${\mathrm{ay}}_k^t\left(k\right)=\frac{f_k^y}{f_k}\&times;maxstep\&times;e^{\frac{-1}{f_k}}$

Wherein,Individuality corresponding to best areal coverage in population during the t time iteration, r1, r2, r3 are population The individual sequence number of three of middle random selection, r1 ≠ r2 ≠ r3,WithIt is three randomly choosing in population The value of the individual jth dimension element in the t time iteration, f is zoom factor, ac₁WithIt is constant and the variable controlling fictitious force, a^t Represent the change of all redundant nodes position in the case of suffered fictitious force,Represent redundant node s_kSuffered fictitious force f_kComponent in x-axis,Represent described redundant node s_kSuffered fictitious force f_kComponent on the y axis, maxstep is institute State redundant node s_kThe ultimate range being moved under the guiding of fictitious force；

Step 7.4, crossover operation, carry out crossover operation using below equation to initial population:

$v_{g,j}^{t+1}=\left\{\begin{array}{cc}{u}_{g,j}^{t+1},& if\left(rand\&le;crorj=rand\left(g\right)\right)\\ x_{g,j}^t,& otherwise\end{array}\right.;$

Wherein,In initial population after representing crossover operation, the jth of g-th individuality ties up the value of element, and cr is to intersect the factor, takes It is worth for (0,1)；Rand (g) is 1 random integers arriving np；

Step 7.5, according to valueCalculate the areal coverage of monitored area

Step 7.6, selection operation, obtain iterationses using below equation individual for g-th during t+1:

$x_g^{t+1}=\left\{\begin{array}{cc}{v}_g^{t+1},& iff\left(v_g^{t+1}\right)<f\left(v_g^t\right)\\ x_g^t,& otherwise\end{array}\right.;$

Step 7.7, determine whether to reach maximum iteration time, if not having, return to step 7.2, if reaching, select to make prison The areal coverage surveying region is maximum individual, and each redundant node is using the coordinate of corresponding position in this individuality as oneself Optimum position.

8. the isomery directional sensor network dynamic coverage method based on mixed strategy according to claim 7, its feature exists In: fictitious force f in step 7.3_kCalculate according to below equation:

$\begin{array}{c}{f}_{kj}^{\&prime;}=\left\{\begin{array}{cc}\frac{k_1{m}_k{m}_j}{d{\left(k,j\right)}^{l_1}},{\mathrm{\&alpha;}}_{kj}& 0<d\left(k,j\right)\&le;r_{s_k}\\ 0& d\left(k,j\right)>r_{s_k}\end{array}\right.\\ f_k=\underset{j\&element;neig\_bor\left(s_k\right)}{\mathrm{\&sigma;}}{f_{kj}}^{\&prime;}\end{array};$

f′_kjRepresent redundant node s_kBy its adjacent node s_jFictitious force, m_kRepresent redundant node s_kPerception area, m_jRepresent Adjacent node s_jPerception area, α_kjRepresent unit vector, the direction of instruction fictitious force, that is, by redundant node s_kIt is directed towards saving Point s_jDirection, r_skRepresent redundant node s_kThe perception radius, l₁For constant, neig_bor (s_k) represent node s_kAdjacent node Set, redundant node s_kThe fictitious force f being subject to_kIt is that its all adjacent node applies fictitious force to it and makes a concerted effort.

9. the isomery directional sensor network dynamic coverage method based on mixed strategy according to claim 7, its feature exists In: areal coverage in step 7.5Calculate in accordance with the following methods:

Step 7.51, initializes one-dimension array coverstatus that length is n so as to the value of all elements is 0, one-dimension array In element and grid correspond；Initialization iteration indicator variable n and k, makes n=1, k=1；

Step 7.52, the central point obtaining grid is to described redundant node s_kApart from d_nCentral point and redundant node with grid s_kThe position vector and redundant node s that are formed_kPerceived direction between angle α_n；

Step 7.53, judges apart from d_nWhether it is less than described redundant node s_kThe perception radiusIf being not less than the perception radius Then make n=n+1, that is, next grid is judged, and return to step 7.52；If being less than the perception radiusThen enter Step 7.54；

Step 7.54, judges central point and redundant node s of grid_kThe position vector and redundant node s that are formed_kPerception side To between angle α_nWhether it is less than redundant node s_kPerception angle half, if being less than, this grid is in redundant node s_k's Sensing range, in one-dimension array coverstatus, the value of element corresponding with this grid is changed into 1, that is, makes coverstatus (n)=1；Otherwise, this grid is not by redundant node s_kPerception；

Step 7.55, determines whether that all grids all have been carried out judging, that is, judges whether n is more than n, if not to all Grid judged, i.e. n < n, then return to step 7.52；Otherwise, illustrate that all grids have completed judgement, then enter step Rapid 7.56；

Step 7.56, judges to next redundant node, that is, makes k=k+1, judges whether k is more than n_mIf k is more than n_m, then enter step 7.57；Otherwise, return to step 7.52；

Step 7.57, in statistics one-dimension array coverstatus, element value is 1 element sum, and makes the value of l be equal to this value, And according to formulaCalculate the areal coverage of monitored area.