A 3D Fence Overlay Method Based on Improved CGA for UWSN

Abstract

This invention designs a three-dimensional fence coverage method for UWSN based on an improved CGA. This invention addresses the practical needs of fence coverage algorithms in underwater wireless sensor networks. Secondly, it analyzes and compares existing technologies for reducing network energy consumption and improving network node utilization, proposing a probability evaluation function based on distance and angle. Finally, to accelerate the convergence of the chaotic genetic algorithm, an adaptive function is added to dynamically change the ratio of parent to offspring in each round. This is a technique for ensuring a low-energy-consumption and high-node-utilization fence network in underwater wireless sensor network environments. Applying this invention, a fence network can be effectively established underwater, resulting in a high node utilization rate and low network energy consumption.

Description

Technical Field

[0001] This invention belongs to the field of underwater wireless sensor network coverage, and particularly relates to a three-dimensional fence coverage method for UWSN based on an improved CGA. Background Technology

[0002] Underwater Wireless Sensor Networks (UWSNs) are one of the most important tools for current underwater exploration. Network coverage is a crucial prerequisite for network operation, and fence coverage is a branch of network coverage. The complex underwater environment, node movement, and energy consumption pose significant challenges to UWSNs. To effectively monitor whether dynamic targets have entered a fixed area underwater, the effective deployment of underwater sensor nodes is essential. A good deployment method can reduce deployment costs, improve communication quality during network operation, and extend network lifespan. Therefore, researching a low-energy three-dimensional fence coverage network is of great significance. Summary of the Invention

[0003] To address the shortcomings of existing technologies, this invention provides a three-dimensional fence coverage method for UWSN based on an improved CGA (Conductive Gaussian Algorithm). The method involves layering the network region, performing a three-dimensional to two-dimensional conversion within each layer, and using an improved chaotic genetic algorithm to calculate the nodes within the plane, forming a fence network. An evaluation function is employed to control the distance and angle between nodes in the fence network, reducing the number of nodes in the network, improving node utilization, and lowering network energy consumption. This results in more uniform distances between nodes in the underwater wireless sensor network, reduces the number of nodes in the network, improves node utilization, and ultimately leads to lower energy consumption in the wireless sensor network obtained using this method.

[0004] A 3D fence overlay method based on improved CGA for UWSN specifically includes the following steps:

[0005] Step 1: Define the network space structure and layer it, then map the nodes obtained from the layering; complete the transformation of nodes in the network space from three-dimensional to two-dimensional, simplifying the fence network construction process;

[0006] Step 1.1: Define the network space structure and perform layered processing; the network space structure is: a crisscrossing transportation, communication, and power supply network formed by multiple path connection channels between points in the region;

[0007] Before deploying the fence network, assume the entire space is a cuboid region with length 2L, width W, and height H. Let the cuboid region on the left with length L, width W, and height H be the target region, and the cuboid region on the right with length L, width W, and height H be the monitoring region. Wireless sensor nodes are randomly distributed in the monitoring region. When a monitored target passes through this region and enters the target region, the sensor nodes can effectively sense the monitored target. Since a fence is deployed in this region, this region is called the region to be deployed.

[0008] The layering of the network space refers to dividing the monitoring area into multiple three-dimensional regions at a certain height. The purpose is to simplify the construction process. A fence is established within each layer of the network space. To avoid network holes, the height of the fence in each layer is determined by the optimal distance between nodes. Assume the monitoring area is L in length, W in width, and H in height. The sensor nodes adopt a probabilistic sensing model, r... s +r e r represents the outer radius of the probabilistic sensing model. s -r e This represents the inner radius of the probabilistic sensing model. The optimal distance between nodes is achieved when the spheres containing the inner and outer radii are externally tangent, at which point the distance between the two nodes is d = 2r. s Therefore, the height H of each layer in the network region c The calculation formula is as follows:

[0009] H c =2r s

[0010] Step 1.2: Map the sensor nodes obtained after layering;

[0011] First, in each network layer, take the midpoints of the four heights of that layer and connect them sequentially to form a cross-section. Establish a coordinate system with the lower left corner of this cross-section as the origin, the length of the cross-section as the X-axis, and the width as the Y-axis, so that the entire cross-section lies on the XOY plane. Map the nodes in the network space of this layer onto the cross-section. Take any two nodes A and B in the space, where the coordinates of node A are (x...). A y A , z A The perpendicular distance from the plane is z. A Assume the coordinates of node B are (x... B y B , z B The perpendicular distance from the plane is z. B After mapping, node A is set to form a two-dimensional probability sensing map on a plane with node A' as the center and its coordinates as (x′). A y′ A The radius of its internal sensing disk is r′. A-s -r′A-e Its external sensing disk has a radius of r′ A-s +r′ A-e The coordinates of node B on the cross section after mapping are B' = (x'). B y′ B The radius of its internal sensing disk is r′. B-s -r′ B-e Its external sensing disk has a radius of r′ B-s +r′ B-e The radius of the disk within the cross-section is calculated using the Pythagorean theorem. The calculation process for the radius of the internal sensing disk of A' is shown below:

[0012]

[0013] The calculation process for the internal sensing disk radius of B' is the same as that for A'.

[0014] According to the Pythagorean theorem, the radius of a node within a cross section is determined by its Z-coordinate. Therefore, we can conclude that the closer a node is to the cross section, the smaller its Z-coordinate, and the larger its radius within the cross section. The sensing radius of a node within the cross section changes dynamically with the Z-value. This yields a two-dimensional representation of nodes within the network space.

[0015] Step 2: Use an improved chaotic genetic algorithm to calculate the sensor nodes in the two-dimensional plane and select suitable sensor nodes to form a fence in the two-dimensional plane;

[0016] The improved chaotic genetic algorithm is based on the chaotic genetic algorithm. By adding an adaptive function during the selection process of the offspring population, the ratio of parents to offspring in each generation is dynamically changed, thus solving the problem of non-convergence in the later stages of the chaotic genetic algorithm.

[0017] Step 2.1: Use an improved chaotic genetic algorithm to perform gene encoding operations on each layer of nodes;

[0018] Based on the node's location, generate the corresponding 3D coordinates of the node; traverse all nodes in the network space, assigning a unique ID value to each traversed node, and combine the ID value with the node's location coordinates to generate a key-value pair, which is the node's genotype, as shown in the following formula:

[0019] <ID,(x,y,z)>

[0020] Step 2.2: Determine the gene evaluation function for each node individual;

[0021] Step 2.2.1: Calculate the node distance probability value;

[0022] First, using the coordinates of the node's center point on the plane obtained in step 1.1 and the inner and outer radii of the node's probability perception model, the distance between any two nodes in space is calculated using the Euclidean distance formula; assuming the distance calculation is performed on nodes a and b, as shown below:

[0023]

[0024] Where, x′ a and x′ b Let y' and y' represent the x-coordinates of the two nodes respectively. a and y′ b These represent the ordinates of the two nodes, respectively.

[0025] Based on the distance results, distance probability values ​​are calculated; a probability transformation is performed using the Laplace distribution; the expectation of the Laplace distribution is equivalent to the optimal distance between nodes, and the formula for calculating the optimal distance between nodes is shown below:

[0026] D best =r′ a-s +r′ b-s

[0027]

[0028] Wherein, from formula r′ a-s +r′ a-e Let r′ represent the radius of the external sensing disk after node a is mapped onto the plane. a-s -r′ a-e This represents the radius of the inner sensing disk after node a is mapped onto the plane; and is related to r′. a-s Similarly, we can calculate r′ b-s Therefore, the expectation function of the Laplace distribution with respect to distance is μ. d =D best ;

[0029] Finally, the distance probability values ​​between nodes a and b in the plane are obtained using the Laplace distribution, as shown below:

[0030]

[0031] Step 2.2.2: Calculate the node angle probability value;

[0032] First, the coordinate formula between two nodes is used, and the angle between the two nodes is calculated using the cosine function; assuming the cosine value of the angle between nodes a and b in the two-dimensional plane is cosθ. ab The calculation is performed using the formula shown below:

[0033]

[0034] Where, |x′a -x′ b | represents the absolute value of the difference between the x-coordinates of two nodes in the plane;

[0035] Secondly, based on the angle calculation formula for two nodes, the angle θ between the starting and ending nodes of each fence in the two-dimensional plane is derived. sg The calculation is as follows:

[0036]

[0037] Among them, arccosθ sg x′ is the inverse function of the cosine function representing the angles at the start and end points. s and y′ s This represents the x and y coordinates of the starting node mapped to the two-dimensional plane, x′. g and y′ g This represents the x and y coordinates of the terminating node mapped to the two-dimensional plane.

[0038] Because the cosine function has its value in Decreasing within the range, The Laplace distribution increases within a certain range and peaks at 0 and 2π, so let the expectation function of the Laplace distribution be...

[0039] Based on the cosine values ​​of the two nodes and the angles of the starting and ending nodes of the fence, the phase value of the original cosine function is adjusted, and θ is calculated using the following formula. sg And the angle between the two nodes:

[0040]

[0041] Finally, substituting the expected value into the probability density function of the Laplace distribution, and then using the formula for calculating the angle between two nodes and the formula for adjusting the phase value of the cosine function, the angular probability result between nodes a and b is obtained. The calculation formula is shown below:

[0042]

[0043] Step 2.2.3: Determine the gene node evaluation function;

[0044] Based on the distance and angle probabilities between the two nodes, the final evaluation function for node b is obtained by weighting the two probabilities, as shown below:

[0045] P(b)=ω1P(D ab )+ω2P(cos(θ ab -θ sg ))

[0046] stω1+ω2=1

[0047] Where ω1 and ω2 represent the weights of the distance probability and the angle probability, respectively, and the sum of the two weights must be 1. In general, if there is no particular bias towards distance and angle, then ω1 = ω2 = 0.5.

[0048] Step 2.3: Use random traversal sampling to select the offspring population;

[0049] First, the roulette wheel is determined. The fitness value of each offspring is represented in the roulette wheel as a ratio of the fitness of all individuals in the population. Therefore, the better the individual, the higher its fitness, the larger its ratio to the total fitness of the population, the larger its angle in the roulette wheel, and the higher its chance of being selected. Next, the number of pointers is determined. By adding an adaptive function during the offspring selection process, the ratio of parents to offspring in each generation is dynamically changed, solving the convergence problem in the later stages of the chaotic genetic algorithm. The adaptive function is shown below:

[0050]

[0051] Where F represents the number of pointers, η0 represents the ratio of the parents of each individual in the population to the next generation when the evaluation of each individual in the population is 0; T represents the number of individuals in the population; η max This represents the percentage of the parent population whose evaluation is passed on to the next generation when every individual in the population has the highest possible evaluation. λ represents the average individual rating of a population. max This represents the maximum value of an individual's evaluation.

[0052] Step 2.4: Gene optimization;

[0053] Step 2.4.1: Gene crossover;

[0054] Gene crossover is performed using partial matching. During the crossover process, a gene mapping relationship is established to ensure that no duplicate genes appear in the crossover individuals, thus ensuring the uniqueness of nodes in a fence. The process is as follows: First, two parents are selected, and the start and end genes of the part to be crossovered are determined. Second, the positions of the two gene parts are swapped, and two initial offspring genes are generated after the gene swap. Then, gene conflict detection is performed to check whether the crossover genes conflict with other genes in the current individual. Finally, for conflicting genes, duplicate genes are replaced according to the mapping relationship.

[0055] Step 2.4.2: Gene mutation;

[0056] Genetic mutation in an individual is achieved using basic positional mutations. The process is as follows: First, the genes of the individual are traversed, and a random probability value P is assigned to each gene, satisfying P∈(0,1). The mutation probability ρ is set. When ρ>P, the point is a mutation point; otherwise, the point remains unchanged. Once the point is determined to be a mutation point, a circular region is established with the point as the center and r as the radius. The ID of the mutation point is then randomly mutated to the ID of any gene within the region, ultimately generating a new offspring individual.

[0057] Step 2.5: Chaotic perturbation;

[0058] To increase the genetic diversity of individuals with low evaluation values ​​in the population, suppress premature convergence, and escape local optima, chaotic perturbations are introduced. A logistic mapping chaotic system is employed, and its functional equation is shown below:

[0059] δ k+1 =μδ k (1-δ k )

[0060] Where k represents the population index, δ represents the chaotic variable, and μ represents the attractor of the chaotic system; when 0 < δ < 1 and 3.57 < μ < 4, the system operates in a chaotic state; the process of adding chaotic perturbations is as follows:

[0061] Suppose an individual has n genes, and let them be set as G = (g1, g2, ..., g...). n Take the i-th gene g i For example, first, the gene g i After normalizing the result, and then substituting it into the chaotic system, we obtain δ. i Next, the gene g after chaotic perturbation is obtained using the following formula. i ′;

[0062] g i ′=α i +λ i δ i ′

[0063] Where, α i and λ i All of these are constants, and the purpose is to convert the normalized values ​​into gene values;

[0064] Step 3: After the two-dimensional fence network is constructed, in order to deploy the three-dimensional fence network, the inter-layer fences are selected to ensure that the inter-layer fences can communicate and do not create perception holes.

[0065] First, the selection of inter-layer fences adopts a top-down approach, that is, the position of the lower-layer nodes is determined by the upper-layer fences, and each node only needs to ensure that it can communicate with one of the nodes in the upper or lower layers. Since the position of each fence is determined by the start and end nodes of the fence, the nodes that make up each layer of the fence are divided into two categories: start and end nodes and intermediate nodes. The former determines the position of the fence, while the latter, which are numerous, form the main body of the fence and determine the communication quality of the fence. Second, the start and end nodes of each layer of fences are selected. When the start and end nodes of the inter-layer fence simultaneously satisfy the following formula, the lower-layer fence is considered to be selectable.

[0066]

[0067] in, Let x, y, and y represent the x, y, and y coordinates of node i in the r-th layer, respectively. Let r represent the x-coordinate, y-coordinate, and vertical coordinate of node i in the (r+1)th layer, respectively. s This indicates the distance from the center of the sensing disk inside or outside the node.

[0068] Beneficial technical effects of the present invention:

[0069] This invention first studies the deployment of fencing in underwater wireless sensor networks in recent years, analyzes the shortcomings of different deployment methods, and identifies the research needs for fencing deployment algorithms in underwater wireless sensor networks. Second, based on the analysis of the relevant requirements for fencing deployment algorithms, a network region layering and node mapping method is proposed to obtain the mapping results of nodes in the plane. Based on this result, an improved chaotic genetic algorithm is used to generate a single-layer fencing. By designing the evaluation function, the distance between nodes in the fencing network becomes more uniform. Finally, an inter-layer fencing selection algorithm is used to complete the deployment of the fencing network in a three-dimensional region. Attached Figure Description

[0070] Figure 1 An initial environment diagram of the unlayered deployment area in this embodiment of the invention;

[0071] Figure 2 The optimal distance between two nodes in the network space region according to the embodiments of the present invention;

[0072] Figure 3 A schematic diagram of the layered deployment area according to an embodiment of the present invention;

[0073] Figure 4 A two-dimensional plan view after fence coverage calculation using the improved CGA algorithm;

[0074] Figure 5 A graph showing the change in the number of fence nodes as a function of the number of network sensor nodes.

[0075] Figure 6 Figure showing the variation of energy consumption in a fence network with the number of sensor nodes in the network.

[0076] Figure 7 The graph shows the change in the utilization rate of fence network nodes as the number of network sensor nodes increases. Detailed Implementation

[0077] The present invention will be further described below with reference to the accompanying drawings and embodiments;

[0078] A 3D fence overlay method based on improved CGA for UWSN specifically includes the following steps:

[0079] Step 1: Define the network space structure and layer it, then map the nodes obtained from the layering; complete the transformation of nodes in the network space from three-dimensional to two-dimensional, simplifying the fence network construction process;

[0080] Step 1.1: Define the network space structure and perform layered processing; the network space structure is: a crisscrossing transportation, communication, and power supply network formed by multiple path connection channels between points in the region;

[0081] Before deploying the fence network, assume the entire space is a cuboid region with length 2L, width W, and height H. Let the cuboid region on the left with length L, width W, and height H be the target region, and the cuboid region on the right with length L, width W, and height H be the monitoring region. Wireless sensor nodes are randomly distributed in the monitoring region. When a monitored target passes through this region and enters the target region, the sensor nodes can effectively sense the monitored target. Since a fence is deployed in this region, this region is called the region to be deployed.

[0082] The layering of the network space refers to dividing the monitoring area into multiple three-dimensional regions at a certain height. The purpose is to simplify the construction process. A fence is established within each layer of the network space. To avoid network holes, the height of the fence in each layer is determined by the optimal distance between nodes. Assume the monitoring area is L in length, W in width, and H in height. The sensor nodes adopt a probabilistic sensing model, r... s +r e r represents the outer radius of the probabilistic sensing model. s -r e This represents the inner radius of the probabilistic sensing model. The optimal distance between nodes is achieved when the spheres containing the inner and outer radii are externally tangent, at which point the distance between the two nodes is d = 2r. s Therefore, the height H of each layer in the network region c The calculation formula is as follows:

[0083] H c =2r s

[0084] Step 1.2: Map the sensor nodes obtained after layering;

[0085] First, in each network layer, take the midpoints of the four heights of that layer and connect them sequentially to form a cross-section. Establish a coordinate system with the lower left corner of this cross-section as the origin, the length of the cross-section as the X-axis, and the width as the Y-axis, so that the entire cross-section lies on the XOY plane. Map the nodes in the network space of this layer onto the cross-section. Take any two nodes A and B in the space, where the coordinates of node A are (x...). A y A , z A The perpendicular distance from the plane is z. A Assume the coordinates of node B are (x... B y B , z B The perpendicular distance from the plane is z. B After mapping, node A is set to form a two-dimensional probability sensing map on a plane with node A' as the center and its coordinates as (x′). A y′ A The radius of its internal sensing disk is r′. A-s -r′ A-e Its external sensing disk has a radius of r′ A-s +r′ A-e The coordinates of node B on the cross section after mapping are B' = (x'). B y′ B The radius of its internal sensing disk is r′. B-s -r′ B-e Its external sensing disk has a radius of r′ B-s +r′ B-e The radius of the disk within the cross-section is calculated using the Pythagorean theorem. The calculation process for the radius of the internal sensing disk of A' is shown below:

[0086]

[0087] The calculation process for the internal sensing disk radius of B' is the same as that for A'.

[0088] According to the Pythagorean theorem, the radius of a node within a cross section is determined by its Z-coordinate. Therefore, we can conclude that the closer a node is to the cross section, the smaller its Z-coordinate, and the larger its radius within the cross section. The sensing radius of a node within the cross section changes dynamically with the Z-value. This yields a two-dimensional representation of nodes within the network space.

[0089] The above process transforms nodes in space from three-dimensional to two-dimensional, simplifying the construction of the fence network. However, the node mapping process introduces the problem of changing node radii. This paper proposes a fence covering algorithm based on an improved CGA, which addresses the dynamic change in node radius during node mapping by incorporating an adaptive dynamic distance function into the traditional genetic algorithm.

[0090] Step 2: Use an improved chaotic genetic algorithm to calculate the sensor nodes in the two-dimensional plane and select suitable sensor nodes to form a fence in the two-dimensional plane;

[0091] The improved chaotic genetic algorithm is based on the chaotic genetic algorithm. By adding an adaptive function during the selection process of the offspring population, the ratio of parents to offspring in each generation is dynamically changed, thus solving the problem of non-convergence in the later stages of the chaotic genetic algorithm.

[0092] Step 2.1: Use an improved chaotic genetic algorithm to perform gene encoding operations on each layer of nodes;

[0093] Based on the node's location, generate the corresponding 3D coordinates of the node; traverse all nodes in the network space, assigning a unique ID value to each traversed node, and combine the ID value with the node's location coordinates to generate a key-value pair, which is the node's genotype, as shown in the following formula:

[0094] <ID,(x,y,z)>

[0095] Step 2.2: Determine the gene evaluation function for each node individual;

[0096] Step 2.2.1: Calculate the node distance probability value;

[0097] First, using the coordinates of the node's center point on the plane obtained in step 1.1 and the inner and outer radii of the node's probability perception model, the distance between any two nodes in space is calculated using the Euclidean distance formula; assuming the distance calculation is performed on nodes a and b, as shown below:

[0098]

[0099] Where, x′ a and x′ b Let y' and y' represent the x-coordinates of the two nodes respectively. a and y′ b These represent the ordinates of the two nodes, respectively.

[0100] Based on the distance results, distance probability values ​​are calculated; a probability transformation is performed using the Laplace distribution; the expectation of the Laplace distribution is equivalent to the optimal distance between nodes, and the formula for calculating the optimal distance between nodes is shown below:

[0101] Dbest =r′ a-s +r′ b-s

[0102]

[0103] Wherein, from formula r′ a-s +r′ a-e Let r′ represent the radius of the external sensing disk after node a is mapped onto the plane. a-s -r′ a-e This represents the radius of the inner sensing disk after node a is mapped onto the plane; and is related to r′. a-s Similarly, we can calculate r′ b-s Therefore, the expectation function of the Laplace distribution with respect to distance is μ. d =D best ;

[0104] Finally, the distance probability values ​​between nodes a and b in the plane are obtained using the Laplace distribution, as shown below:

[0105]

[0106] Step 2.2.2: Calculate the node angle probability value;

[0107] First, the coordinate formula between two nodes is used, and the angle between the two nodes is calculated using the cosine function; assuming the cosine value of the angle between nodes a and b in the two-dimensional plane is cosθ. ab The calculation is performed using the formula shown below:

[0108]

[0109] Where, |x′ a -x′ b | represents the absolute value of the difference between the x-coordinates of two nodes in the plane;

[0110] Secondly, based on the angle calculation formula for two nodes, the angle θ between the starting and ending nodes of each fence in the two-dimensional plane is derived. sg The calculation is as follows:

[0111]

[0112] Among them, arccosθ sg x′ is the inverse function of the cosine function representing the angles at the start and end points. s and y′ s This represents the x and y coordinates of the starting node mapped to the two-dimensional plane, x′. g and y′ g This represents the x and y coordinates of the terminating node mapped to the two-dimensional plane.

[0113] Because the cosine function has its value in Decreasing within the range, The Laplace distribution increases within a certain range and peaks at 0 and 2π, so let the expectation function of the Laplace distribution be...

[0114] Based on the cosine values ​​of the two nodes and the angles of the starting and ending nodes of the fence, the phase value of the original cosine function is adjusted, and θ is calculated using the following formula. sg And the angle between the two nodes:

[0115]

[0116] Finally, substituting the expected value into the probability density function of the Laplace distribution, and then using the formula for calculating the angle between two nodes and the formula for adjusting the phase value of the cosine function, the angular probability result between nodes a and b is obtained. The calculation formula is shown below:

[0117]

[0118] Step 2.2.3: Determine the gene node evaluation function;

[0119] Based on the distance and angle probabilities between the two nodes, the final evaluation function for node b is obtained by weighting the two probabilities, as shown below:

[0120] P(b)=ω1P(D ab )+ω2P(cos(θ ab -θ sg ))

[0121] stω1+ω2=1

[0122] Where ω1 and ω2 represent the weights of the distance probability and the angle probability, respectively, and the sum of the two weights must be 1. In general, if there is no particular bias towards distance and angle, then ω1 = ω2 = 0.5.

[0123] Step 2.3: Use random traversal sampling to select the offspring population;

[0124] First, the roulette wheel is determined. The fitness value of each offspring is represented in the roulette wheel as a ratio of the fitness of all individuals in the population. Therefore, the better the individual, the higher its fitness, the larger its ratio to the total fitness of the population, the larger its angle in the roulette wheel, and the higher its chance of being selected. Next, the number of pointers is determined. By adding an adaptive function during the offspring selection process, the ratio of parents to offspring in each generation is dynamically changed, solving the convergence problem in the later stages of the chaotic genetic algorithm. The adaptive function is shown below:

[0125]

[0126] Where F represents the number of pointers, η0 represents the ratio of the parents of each individual in the population to the next generation when the evaluation of each individual in the population is 0; T represents the number of individuals in the population; η max This represents the percentage of the parent population whose evaluation is passed on to the next generation when every individual in the population has the highest possible evaluation. λ represents the average individual rating of a population. max This represents the maximum value of an individual's evaluation.

[0127] Step 2.4: Gene optimization;

[0128] Step 2.4.1: Gene crossover;

[0129] Gene crossover is performed using partial matching. During the crossover process, a gene mapping relationship is established to ensure that no duplicate genes appear in the crossover individuals, thus ensuring the uniqueness of nodes in a fence. The process is as follows: First, two parents are selected, and the start and end genes of the part to be crossovered are determined. Second, the positions of the two gene parts are swapped, and two initial offspring genes are generated after the gene swap. Then, gene conflict detection is performed to check whether the crossover genes conflict with other genes in the current individual. Finally, for conflicting genes, duplicate genes are replaced according to the mapping relationship.

[0130] Step 2.4.2: Gene mutation;

[0131] Genetic mutation in an individual is achieved using basic positional mutations. The process is as follows: First, the genes of the individual are traversed, and a random probability value P is assigned to each gene, satisfying P∈(0,1). The mutation probability ρ is set. When ρ>P, the point is a mutation point; otherwise, the point remains unchanged. Once the point is determined to be a mutation point, a circular region is established with the point as the center and r as the radius. The ID of the mutation point is then randomly mutated to the ID of any gene within the region, ultimately generating a new offspring individual.

[0132] Step 2.5: Chaotic perturbation;

[0133] To increase the genetic diversity of individuals with low evaluation values ​​in the population, suppress premature convergence, and escape local optima, chaotic perturbations are introduced. A logistic mapping chaotic system is employed, and its functional equation is shown below:

[0134] δ k+1 =μδ k (1-δ k )

[0135] Where k represents the population index, δ represents the chaotic variable, and μ represents the attractor of the chaotic system; when 0 < δ < 1 and 3.57 < μ < 4, the system operates in a chaotic state; the process of adding chaotic perturbations is as follows:

[0136] Suppose an individual has n genes, and let them be set as G = (g1, g2, ..., g...). n Take the i-th gene g i For example, first, the gene g i After normalizing the result, and then substituting it into the chaotic system, we obtain δ. i Next, the gene g after chaotic perturbation is obtained using the following formula. i ′;

[0137] g i ′=α i +λ i δ i ′

[0138] Where, α i and λ i All of these are constants, and the purpose is to convert the normalized values ​​into gene values;

[0139] Step 3: After the two-dimensional fence network is constructed, in order to deploy the three-dimensional fence network, the inter-layer fences are selected to ensure that the inter-layer fences can communicate and do not create perception holes.

[0140] First, the selection of inter-layer fences adopts a top-down approach, that is, the position of the lower-layer nodes is determined by the upper-layer fences, and each node only needs to ensure that it can communicate with one of the nodes in the upper or lower layers. Since the position of each fence is determined by the start and end nodes of the fence, the nodes that make up each layer of the fence are divided into two categories: start and end nodes and intermediate nodes. The former determines the position of the fence, while the latter, which are numerous, form the main body of the fence and determine the communication quality of the fence. Second, the start and end nodes of each layer of fences are selected. When the start and end nodes of the inter-layer fence simultaneously satisfy the following formula, the lower-layer fence is considered to be selectable.

[0141]

[0142] in, Let x, y, and y represent the x, y, and y coordinates of node i in the r-th layer, respectively. Let r represent the x-coordinate, y-coordinate, and vertical coordinate of node i in the (r+1)th layer, respectively. s This indicates the distance from the center of the sensing disk inside or outside the node.

[0143] This invention proposes a three-dimensional fence coverage algorithm based on an improved CGA (Computer-Aided Networking) to establish a three-dimensional fence network with low energy consumption and high node utilization. The algorithm first specifies the initial environment of the area to be deployed, as shown in the attached diagram. Figure 1 As shown, the three-dimensional space is further divided into layers based on the optimal distance between sensor nodes, as shown in the attached figure. Figure 2 As shown in the attached figure, the layered deployment areas are as follows. Figure 3 As shown, nodes within each layer are mapped onto a two-dimensional plane, and then 3D-ICGA is used to plan a single fence within that plane. The result is shown in the attached figure. Figure 4 As shown, the next step is to select the inter-layer fences to complete the planning of the 3D fence network. Since this algorithm controls the distance between nodes in the evaluation function, compared with the Greedy algorithm and the 3D-ACO algorithm, the sensor node distribution of the fence network calculated by this algorithm is more uniform, and the increase in the number of fence nodes is minimal as the size of the network sensor nodes increases. The results are shown in the attached figure. Figure 5 As shown in the figure; therefore, the number of nodes forming the fence is more stable, resulting in lower network energy consumption and higher node utilization. The results are shown in the appendix. Figure 6 and attached Figure 7 As shown.

Claims

1. A three-dimensional fence overlay method for UWSN based on improved CGA, characterized in that, Specifically, the following steps are included: Step 1: Define the network space structure and layer it, then map the nodes obtained from the layering; complete the transformation of nodes in the network space from three-dimensional to two-dimensional, simplifying the fence network construction process; Step 2: Use an improved chaotic genetic algorithm to calculate the sensor nodes in the two-dimensional plane and select suitable sensor nodes to form a fence in the two-dimensional plane; The improved chaotic genetic algorithm is based on the chaotic genetic algorithm. By adding an adaptive function during the selection process of the offspring population, the ratio of parents to offspring in each generation is dynamically changed, thus solving the problem of non-convergence in the later stages of the chaotic genetic algorithm. Step 2.1: Use an improved chaotic genetic algorithm to perform gene encoding operations on each layer of nodes; Step 2.2: Determine the gene evaluation function for each node individual; Step 2.2.1: Calculate the node distance probability value; First, using the coordinates of the center point of the node on the plane obtained in step 1, and the inner and outer radii of the node's probabilistic perception model, the distance between any two nodes in space is calculated using the Euclidean distance formula. Assuming the distance is calculated for nodes a and b, as shown below: ； in, and Let x and y represent the x-coordinates of the two nodes, respectively. and These represent the ordinates of the two nodes, respectively. Based on the distance results, distance probability values ​​are calculated; a probability transformation is performed using the Laplace distribution; the expectation of the Laplace distribution is equivalent to the optimal distance between nodes, and the formula for calculating the optimal distance between nodes is shown below: ； ； Wherein, by formula Represents a node The radius of the externally perceived disk after mapping onto the plane. Represents a node The radius of the inner sensing disk after mapping to the plane; and Similar, calculations show that... Therefore, the expectation function of the Laplace distribution with respect to distance is: ; Finally, the nodes are obtained using the Laplace distribution. and The probability value of distance in the plane is calculated as follows: ； Step 2.2.2: Calculate the node angle probability value; First, the coordinate formula between two nodes is used, and the cosine function is used to calculate the angle between the two nodes; assuming that for nodes... and The cosine of the angle in a two-dimensional plane The calculation is performed using the formula shown below: ； in, It represents the absolute value of the difference between the x-coordinates of two nodes in the plane; Secondly, based on the angle calculation formula for two nodes, the angles of the starting and ending nodes of each fence in the two-dimensional plane are derived. The calculation is as follows: ； in, The inverse function of the cosine function representing the angles of the start and end nodes. and This represents the x and y coordinates of the starting node mapped to a two-dimensional plane. and This represents the x and y coordinates of the terminating node mapped to the two-dimensional plane. Because the cosine function has its value in Decreasing within the range, Increasing within the range, and in and The peak value was achieved at [location]. Given the characteristics of the Laplace distribution, let its expectation function be... ; Based on the cosine values ​​of the two nodes and the angles of the starting and ending nodes of the fence, the phase value of the original cosine function is adjusted, and the following formula is used to calculate... And the angle between the two nodes: ； Finally, the expected value is substituted into the probability density function of the Laplace distribution, and then the angle between the two nodes is calculated and adjusted. The formula for the phase value of a function yields the node... and The probability result of the angle between them is calculated using the following formula: ； Step 2.2.3: Determine the gene node evaluation function; Based on the distance and angle probabilities between two nodes, the node is obtained by weighting the two probabilities. The final evaluation function result is shown below: ； ； in, and Let these represent the weights for the distance probability and the angle probability, respectively, and the sum of the two weights must satisfy the condition that... Generally, if there is no particular preference for distance and angle, then take... ; Step 2.3: Use random traversal sampling to select the offspring population; First, the roulette wheel is determined. The fitness value of each offspring is represented in the roulette wheel as a ratio of the fitness of all individuals in the population. Therefore, the better the individual, the higher its fitness, the larger its ratio to the total fitness of the population, the larger its angle in the roulette wheel, and the higher its chance of being selected. Next, the number of pointers is determined. By adding an adaptive function during the offspring selection process, the ratio of parents to offspring in each generation is dynamically changed, solving the convergence problem in the later stages of the chaotic genetic algorithm. The adaptive function is shown below: ； in, Indicates the number of pointers. This represents the percentage of the population whose parents are inherited when the evaluation of each individual in the population is 0. Indicates the number of individuals in a population; This represents the percentage of the parent population whose evaluation is passed on to the next generation when every individual in the population has the highest possible evaluation. Represents the average rating of individuals in a population; This represents the maximum value of an individual's evaluation. Step 2.4: Gene optimization; Step 2.5: Chaotic perturbation; Step 3: After the two-dimensional fence network is constructed, in order to deploy the three-dimensional fence network, the inter-layer fences are selected to ensure that the inter-layer fences can communicate and do not create perception holes.

2. The three-dimensional fence overlay method for UWSN based on improved CGA according to claim 1, characterized in that, Step 1 is as follows: Step 1.1: Define the network space structure and perform layered processing; the network space structure is: a crisscrossing transportation, communication, and power supply network formed by multiple path connection channels between points in the region; Before deploying the fence network, assume the entire space is a cuboid region with length 2L, width W, and height H. Let the cuboid region on the left with length L, width W, and height H be the target region, and the cuboid region on the right with length L, width W, and height H be the monitoring region. Wireless sensor nodes are randomly distributed in the monitoring region. When a monitored target passes through this region and enters the target region, the sensor nodes can effectively sense the monitored target. Since a fence is deployed in this region, this region is called the region to be deployed. The layering of the network space refers to dividing the monitoring area into multiple three-dimensional regions according to a certain height. The purpose is to simplify the construction process. A fence is established within each layer of the network space. To avoid network holes, the height of the fence in each layer is determined by the optimal distance between nodes. Assume the monitoring area is L in length, W in width, and H in height; the sensor nodes adopt a probabilistic sensing model. This represents the outer radius of the probability-aware model. This represents the inner radius of the probabilistic sensing model. The optimal distance between nodes is achieved when the spheres containing the inner and outer radii are externally tangent. At this point, the distance between the two nodes is... Therefore, the height of each layer in the network region is... The calculation formula is as follows: ； Step 1.2: Map the sensor nodes obtained after layering; First, in each network layer, take the midpoints of the four heights of that layer and connect them sequentially to form a cross section; with the lower left corner of this cross section as the origin, the length of this cross section is... Shaft, width is Establish a coordinate system along the axes, so that the entire cross-section is located at... On the plane; map the nodes in the network space of this layer to the cross section, taking any node in the space... , Two nodes, The coordinates of the node are The perpendicular distance from the plane is Assuming The coordinates of the node are The perpendicular distance from the plane is After mapping, set The nodes form a two-dimensional probabilistic sensing graph on a plane. The node is centered, and its coordinates are... Its internal sensing disk radius is Its external sensing disk radius is , The coordinates of the node on the cross section after mapping are Its internal sensing disk radius is Its external sensing disk radius is The radius of the disk within the cross-section is calculated using the Pythagorean theorem. The calculation process for the radius of the internal sensing disk is as follows: ； The calculation process of the internal sensing disk radius is the same as ; According to the Pythagorean theorem, the radius of a node within a cross section is determined by the value of the node. The coordinates determine the distance between the node and the cross section, therefore: the closer the node is to the cross section, the better. The smaller the coordinate, the larger its radius in the cross section; the sensing radius of the nodes within the cross section increases with... The values ​​change dynamically, resulting in a two-dimensional representation of nodes within the network space.

3. The three-dimensional fence overlay method for UWSN based on improved CGA according to claim 1, characterized in that, The specific content of step 2.1 is as follows: Based on the node's location, generate the corresponding 3D coordinates of the node; traverse all nodes in the network space, assigning a unique ID value to each traversed node, and combine the ID value with the node's location coordinates to generate a key-value pair, which is the node's genotype, as shown in the following formula: ； The specific content of step 2.5 is as follows: To increase the genetic diversity of individuals with low evaluation values ​​in the population, suppress premature convergence, and escape local optima, chaotic perturbations are introduced. A logistic mapping chaotic system is employed, and its functional equation is shown below: ； in, Indicates the population index. Represents chaotic variables. Represents the attractor of a chaotic system; when and At that time, the system was operating in a chaotic state; the process of adding chaotic perturbations is as follows: Suppose that there are individuals Each gene is set as a set. Take the first One gene For example, first, the genes After normalizing the result, and then substituting it into the chaotic system, we obtain... Next, the gene after chaotic perturbation is obtained using the following formula. ; ； in, and All of these are constants, and the purpose is to convert the normalized values ​​into gene values.

4. A three-dimensional fence overlay method for UWSN based on improved CGA according to claim 3, characterized in that, Step 2.4 specifically involves: Step 2.4.1: Gene crossover; Gene crossover is performed using partial matching. During the crossover process, a gene mapping relationship is established to ensure that no duplicate genes appear in the crossover individuals, thus ensuring the uniqueness of nodes in a fence. The process is as follows: First, two parents are selected, and the start and end genes of the part to be crossovered are determined. Second, the positions of the two gene parts are swapped, and two initial offspring genes are generated after the gene swap. Then, gene conflict detection is performed to check whether the crossover genes conflict with other genes in the current individual. Finally, for conflicting genes, duplicate genes are replaced according to the mapping relationship. Step 2.4.2: Gene mutation; Genetic variation in an individual is achieved by utilizing fundamental positional variations; The process is as follows: First, the genes of an individual are traversed, and a random probability value is assigned to each gene. And satisfy Set mutation probability ,when If the value is true, then the point is a mutation point; otherwise, the point remains unchanged. Once this point is identified as a mutation point, take this point as the center. If a circular region is established with a radius, the ID of the mutation point will randomly mutate to the ID of any gene within that region, eventually generating a new offspring individual.

5. A three-dimensional fence overlay method for UWSN based on improved CGA according to claim 1, characterized in that, Step 3 specifically involves: First, the selection of inter-layer fences adopts a top-down approach, that is, the position of the lower-layer nodes is determined by the upper-layer fences, and each node only needs to ensure that it can communicate with one of the nodes in the upper or lower layers. Since the position of each fence is determined by the start and end nodes of the fence, the nodes that make up each layer of the fence are divided into two categories: start and end nodes and intermediate nodes. The former determines the position of the fence, while the latter, which are numerous, form the main body of the fence and determine the communication quality of the fence. Second, the start and end nodes of each layer of fences are selected. When the start and end nodes of the inter-layer fence simultaneously satisfy the following formula, the lower-layer fence is considered to be selectable. ； in, , , They represent the first Layer The x-coordinate, y-coordinate, and vertical coordinate of a node. , , They represent the first Layer The x-coordinate, y-coordinate, and vertical coordinate of a node. This indicates the distance from the center of the sensing disk inside or outside the node.

Images