A WSN deployment method based on a swarm intelligence optimization algorithm

Abstract

The application belongs to the field of wireless sensor network, and particularly relates to a WSN deployment method based on a swarm intelligence optimization algorithm, which comprises the following steps: constructing a WSN coverage model, and constructing a WSN target function according to the WSN coverage model; improving the swarm intelligence optimization algorithm according to an elite reverse learning method, a Levy flight strategy and a Brown motion strategy, and calculating a WSN deployment strategy according to the WSN target function by using the improved swarm intelligence optimization algorithm; the WSN is a sensor node; the application improves the swarm intelligence algorithm by combining many advantage strategies to optimize the WSN coverage, guarantees the convergence speed and precision of optimization, avoids falling into local optimization at the same time, and realizes the minimization of the sensor node and the maximization of the coverage rate.

Description

TECHNICAL FIELD

[0001] The application belongs to the field of wireless sensor networks, and particularly relates to a WSN deployment method based on a swarm intelligence optimization algorithm. BACKGROUND

[0002] With the continuous progress of the Internet and artificial intelligence technology, the Internet of Things has become a hot research target in today's technology field. As one of the core supporting technologies of the Internet of Things, wireless sensor networks play a supporting role in the Internet of Things. Wireless sensor network (WSN) is a network composed of multiple wireless sensor nodes, which detects and collects different types of environmental data, including temperature, humidity, air pressure, light and sound, and then transmits the data to one or more base stations or central nodes. Wireless sensor networks have been widely used in target tracking, environmental monitoring, military applications, medical applications, etc., which can help people better understand the environmental conditions, improve productivity, reduce resource waste, and improve the quality of life.

[0003] Due to its increasing popularity, the performance of wireless sensor networks itself has attracted more and more attention. In particular, the reliability of network applications and the quality of network services. The nodes deployed in the designated monitoring area and the guarantee of connection can be used to explain the challenge of coverage optimization of wireless sensor networks. In order to meet the coverage requirements, people usually randomly scatter a large number of sensor nodes, which often causes larger coverage blind area or node redundancy due to the performance limitations of sensor node power and stability, shortens the network life, reduces the network reliability, and causes a large amount of resource waste in terms of energy consumption and cost, so coverage optimization is an important research topic of wireless sensor networks.

[0004] The coverage optimization challenge in wireless sensor networks can be better solved by swarm intelligence optimization methods. Many studies use swarm intelligence optimization methods to limit the coverage of wireless sensor networks. For example, particle swarm optimization (PSO) is introduced into WSN to optimize network coverage, and the concept of combinatorial mathematics is combined to improve PSO to achieve shorter time consumption and higher coverage rate; the whale optimization algorithm (WOA) is combined with WSN to effectively realize the coverage optimization of wireless sensor networks, but its redundancy is high; Khalaf invented an artificial bee colony technology, which provides more coverage rate compared with genetic algorithm, while using fewer system resources; Chen W proposed a sensor deployment strategy based on ant colony optimization (ALO), which provides more uniform sensor distribution and significantly reduces the number of nodes.

[0005] Although the optimization technology such as swarm intelligence is introduced at present for the coverage optimization in the wireless sensor network, most algorithms can only have good optimization effect for a certain type of special problem, and there is no algorithm for the WSN coverage optimization problem, so there is more exploratory for this problem. And many swarm intelligence algorithms currently have the disadvantages of slow convergence speed and low convergence accuracy and easy to fall into local optimum, and the key of design and improvement lies in balancing the relationship between the diversified exploration and the centralized mining. The diversified exploration allows the algorithm to explore the global optimal solution in the largest possible feasible domain to avoid falling into the local optimum, but it takes a long time and the accuracy of the optimal solution obtained is poor. The centralized mining allows the algorithm to use the accumulated experience and the knowledge in the search process to carry out detailed and in-depth mining in a certain area, so as to obtain the optimal solution faster and more accurately, but the mining also makes the algorithm easily fall into the local optimum.

[0006] Therefore, how to balance the relationship between the diversified exploration and the centralized mining and find better and more stable algorithms for more and more complex WSN coverage optimization problems becomes a problem to be solved by the current swarm intelligence optimization algorithm. SUMMARY

[0007] In order to solve the above prior art problems, the application adopts a WSN deployment method based on a swarm intelligence optimization algorithm, which comprises the following steps: constructing a WSN coverage model, constructing a WSN target function according to the WSN coverage model; improving the swarm intelligence optimization algorithm according to an elite reverse learning method, a Levy flight strategy and a Brownian motion strategy, and calculating a WSN deployment strategy according to the WSN target function by using the improved swarm intelligence optimization algorithm; and the WSN is a sensor node.

[0008] The construction of the WSN coverage model comprises the following steps: constructing a sensor node group: Node{node1,...,node N}, wherein the sensor node node i =(x i ,y i ,r), (x i ,y i ) is the position of the sensor node, and r is the sensing radius of the sensor node; and calculating the coverage area of the sensor node according to the sensor node group.

[0009] The WSN target function is as follows:

[0010]

[0011] wherein CoverRatio is the coverage area of the sensor node set, P cov is the coverage area of the sensor node, and m and n are the length and width of the target monitoring area.

[0012] The improved swarm intelligence optimization algorithm is used to calculate the WSN deployment strategy according to a WSN objective function, and the WSN deployment strategy comprises the following steps:

[0013] S1, an elite reverse learning strategy is used for population initialization; the population comprises a plurality of individuals, and each individual represents a sensor node position;

[0014] S2, a motion strategy under Levy flight and a motion strategy under Brown motion are used to update the initialized population individuals in a first stage, an avoidance mechanism is used to update the population individuals updated in the first stage in a second stage, and the population individuals updated in the second stage are obtained;

[0015] S3, the WSN objective function values corresponding to the population individuals updated in the second stage are calculated, and the population individual with the minimum WSN objective function value is selected as the best population individual, i.e., the best sensor node position;

[0016] S4, when the maximum iteration number T is reached, the final population individual, i.e., the final sensor node position, is obtained; otherwise, the step S2 is returned to, and the best population individual is updated.

[0017] The first stage update of the population individual comprises the following steps:

[0018] The WSN objective function value corresponding to each population individual X t [i,j] is calculated, and an elite node is selected from the population individual X t [i,j] according to the WSN objective function value; wherein i is the index of the population individual, j is the index of the dimension of the population individual, and t is the current iteration number;

[0019] An adaptive parameter alpha is calculated; if the alpha is greater than 0.5, a new population individual X is calculated according to the elite node by using the motion strategy under Levy flight; otherwise, a new population individual X is calculated according to the elite node by using the motion strategy under Brown motion.

[0020] The elite node is updated, and the WSN objective function corresponding to the updated elite node is calculated. The first stage updated population individual is obtained according to the WSN objective function corresponding to the updated elite node.

[0021] The calculation method of the adaptive parameter alpha is as follows:

[0022]

[0023] Wherein, rand() is a random function, and T is the maximum iteration number.

[0024] calculating new population individuals using a motion strategy under Levy flight comprising:

[0025]

[0026] wherein X t [i,j] represents a population individual, r is a random number from 0 to 1, X t leader represents an elite node, L is a function of Levy flight, and sinθ, cosθ represent two-dimensional components of plane coordinates.

[0027] calculating new population individuals using a motion strategy under Brownian motion comprising:

[0028]

[0029] wherein X t [i,j] represents a population individual, r is a random number from 0 to 1, X t leader represents an elite node, and RB is a Brownian motion strategy.

[0030] obtaining first-stage updated population individuals according to a WSN objective function comprising:

[0031]

[0032] wherein, is the first-stage updated population individual, F new1 is a value of an objective function of the new population individual, F leader is a value of an objective function of the elite node, and X t [i,j] is a population individual.

[0033] second-stage updating the first-stage updated population individuals according to an evading mechanism comprises:

[0034] setting a warning value and calculating new population individuals according to the warning value

[0035]

[0036] calculating population individuals and population individuals corresponding to WSN objective functions, obtaining second-stage updated population individuals according to the WSN objective functions corresponding to population individuals and population individuals corresponding to WSN objective functions;

[0037] wherein, For the first stage updated population individual, R is a random number subject to normal distribution, D is a matrix, each element of D is 1, T is the maximum iteration number, t is the current iteration number, r is the sensing radius, alpha is the adaptive parameter, and WV is the warning value.

[0038] Beneficial effects:

[0039] The application combines the elite reverse learning strategy, the Levy flight strategy and the Brown motion strategy to improve the swarm intelligence algorithm for WSN coverage optimization, avoids falling into local optimum while ensuring the convergence speed and precision of optimization, and realizes minimization of sensor nodes and maximization of coverage rate. BRIEF DESCRIPTION OF DRAWINGS

[0040] Figure 1 A flowchart of a WSN deployment method based on a swarm intelligence optimization algorithm provided by the embodiment of the application is shown in the figure;

[0041] Figure 2 A parameter diagram of wireless sensor coverage and swarm intelligence algorithm provided by the embodiment of the application is shown in the figure;

[0042] Figure 3 A dynamic change diagram of the operator alpha provided by the embodiment of the application is shown in the figure;

[0043] Figure 4 Optimized coverage results of the swarm intelligence algorithm and other algorithms provided by the embodiment of the application are shown in the figure;

[0044] Figure 5 The node distribution after optimization of the swarm intelligence algorithm and other algorithms provided by the embodiment of the application is shown in the figure. DETAILED DESCRIPTION

[0045] The technical solutions in the embodiments of the application will be described clearly and completely below with reference to the drawings in the embodiments of the application. Obviously, the described embodiments are only part of the embodiments of the application, rather than all the embodiments of the application. Based on the embodiments in the application, all other embodiments obtained by those skilled in the art without creative labor fall within the protection scope of the application.

[0046] I. Constructing a WSN coverage model

[0047] The sensing range of a sensor node in a two-dimensional plane is a circular area with the node as the center and the radius as Rn. The circular area is usually referred to as the sensing disc of the node, and Rn is referred to as the sensing radius of the sensor node, which is related to the physical characteristics of the built-in sensor device of the node. Assuming that the position coordinates of node n are (x n , y n ), in the 0-1 sensing model, for any point p(x p,y p ), the node n monitors the event occurrence probability of the point p in the area as:

[0048]

[0049] The number of homogeneous structure sensor nodes placed on the area A is N, and the target monitoring area is considered as a two-dimensional plane. The position coordinate value and sensing radius r of each node are initially assigned. Therefore, the sensor node group is represented as: Node{node1,...,node N} where node i =(x i ,y i ,r) represents a circle with the node (x i ,y i ) as the center and r as the monitoring radius. It is assumed that the monitoring area Area is discretized into m*n pixel points, the coordinates of the pixel points are (x, y), and the distance between the target pixel point and the sensor node is:

[0050]

[0051] The event that the pixel point in the target area is covered by the sensor node is defined as c i . The probability of the event is the probability that the pixel (x, y) is covered by the sensor node i:

[0052]

[0053] The present application defines the coverage area CoverRatio of all sensor nodes in the target monitoring environment as the ratio of the area covered by the sensor node set to the area of the monitoring area, as shown in the following equation:

[0054]

[0055] The objective of the intelligent optimization algorithm is to find the minimum value. Therefore, the present application defines the fitness function of the intelligent optimization algorithm as the minimum value of the uncovered, so that the final result is the maximum coverage range. The expression is as follows:

[0056]

[0057] II. Using the swarm intelligence optimization algorithm to calculate the WSN deployment strategy according to the WSN target function F

[0058] As shown in Figure 1 , the steps of using the swarm intelligence optimization algorithm to calculate the WSN deployment strategy according to the WSN target function F include:

[0059] (I) Initialize the population

[0060] As Figure 2 shown, the solution space is first determined: X i = LB + rand x (UB-LB)

[0061] Where UB and LB are the upper and lower bounds of the solution space, respectively, and rand is a random number in [0, 1]. Assuming that each population individual represents a candidate solution in the problem search space, the entire population solution space can be represented as:

[0062]

[0063] Where Dim represents the dimension of the variable, and pop is the total size. In order to expand the initial node diversity and speed up the convergence, the present application adopts the elite inverse learning strategy in initialization. The main idea is to calculate and identify the candidate solution and the matching inverse solution at the same time, and then select the best candidate solution from the two according to the fitness value as the initial population.

[0064] 1. Inverse solution

[0065] Let the feasible solution of the current node in D-dimensional space be X = (x1, x2,..., x D ), Then its inverse solution is a j is the minimum value of the individual, b j is the maximum value of the individual, and r is a random number following a uniform distribution in [0, 1].

[0066] 2. Elite inverse solution

[0067] Define the extreme point of the average individual in the population, and define the extreme point of the population as the corresponding elite individual in the population Then its elite inverse solution is Defined as:

[0068]

[0069] Where δ is a random number in [0, 1]; lb j = min X i,j , i.e. the minimum value of the individual, ub j = max X i,j , i.e. the maximum value of the individual.

[0070] In elite inverse learning, the feasible solution is used to create the inverse solution of the current problem, increasing the search space and population diversity. Then the optimal solution is selected from the population of the current solution and its inverse solution, and the individual is updated to improve the quality of the population.

[0071] (II) First stage update

[0072] AsFigure 3 、 Figure 4 At this stage, the present application proposes an adaptive parameter alpha that fluctuates with the number of iterations to correct the interaction between local development and global search, expressed as:

[0073]

[0074] It can be seen that as the number of iterations increases, the alpha fluctuation decreases, and the present application uses 0.5 as a dividing point, half of which is used for global development and half for local exploration.

[0075] 1. Motion strategy under Levy flight

[0076] At the beginning of the update iteration, when alpha is greater than 0.5, the movement step of the operator is simulated by using Levy flight. In the random walk process, since Levy flight is a kind of wandering, the step probability distribution is heavy-tailed, so there is a greater possibility to take a longer step. This matches the fact that the node is in a groping state in the early iteration, with a high probability of quickly crossing to other positions, while avoiding the algorithm quickly falling into a local optimum. At the same time, the direction is adjusted and the angle is changed during the iteration process, which can expand the neighborhood of the current local solution and develop as many optimal solution regions as possible. Therefore, the present application generates a mathematical representation of the new position for the next iteration:

[0077]

[0078] where X t [i,j] represents the population individual of the tth iteration, r is a random number from 0 to 1, X t leader represents the elite node of the tth iteration, L is the function of Levy flight, and sinθ, cosθ represent the two-dimensional components of the plane coordinates, θ belongs to [0, 2π].

[0079] 2. Motion strategy under Brownian motion

[0080] When the iteration reaches the middle and late stages, i.e., alpha is less than 0.5, Brownian motion is performed. Brownian motion is also a kind of random wandering, which is more stable than Levy flight, and the function obeys a normal distribution with a mean of 0 and a variance of σ. This is consistent with the general law of optimization, because in the middle and late stages, there is no longer a need for a large-scale search, but rather a need to avoid local extremes and to stably explore the global optimum. Its mathematical expression is:

[0081]

[0082] When the new position of the new node makes the value of the objective function superior to the position of the elite node, the application makes the node with better position the new elite node, and makes the rest of the nodes follow it to move to the new position.

[0083]

[0084] wherein, is the population individual after the first stage update, F new1 is the objective function value of the new population individual, F leader is the objective function value of the elite node, X t [i,j] is the population individual of the tth iteration.

[0085] (Three), second stage update

[0086] In the later iterations, the most important problem that the application needs to consider is that the node falls into the trap of local optimum. Therefore, the application sets the warning value WV=0.5. When the dynamic parameter alpha is less than the warning value, the node can continue to explore the area, and once the alpha is greater than the warning value, the elite node needs to guide the rest of the nodes to the safe area in the next iteration:

[0087]

[0088] wherein, is the population individual after the first stage update, R is a random number subject to normal distribution, D is a 1xd matrix, wherein each element is 1, T is the maximum number of iterations, r is the sensing radius, alpha is the adaptive parameter, and WV is the warning value.

[0089] Then, the final position X t+1 [i,j] is updated according to the fitness function.

[0090]

[0091] wherein, F new1 is the population individual corresponding to the WSN objective function value, F new2 is the population individual corresponding to the WSN objective function value.

[0092] The population individual X t [i,j] of the tth iteration is calculated, the WSN objective function value corresponding to the population individual is selected, and the population individual with the minimum WSN objective function value is selected as the best population individual, that is, the best sensor node position;

[0093] (Four) calculation of the final sensor node position

[0094] Set the maximum number of iterations T, let the algorithm loop execute the algorithm above the equation to perform each stage, until greater than the set number of iterations, then exit the loop, get the final population individual, that is, the final sensor node position.

[0095] III. Simulation Performance Analysis

[0096] In order to comprehensively test the improved swarm intelligence algorithm of the application-Tibetan Antelope Optimization (TAO) for optimizing the performance of wireless sensor network node coverage, the application selects five other algorithms for comparison. In addition, the simulation experiment settings used in different comparison techniques are the same. 30 simulation experiments were conducted for each comparison algorithm to improve the accuracy of the results. Then the average coverage is used for comparison.

[0097] As Figure 5 shown, the final sensor distribution after applying TAO, Ant Lion Optimizer (ALO), Marine Predators Algorithm (MPA), Particle Swarm Optimization (PSO), Sparrow Search Algorithm (SSA), and Whale Optimization Algorithm (WOA) is depicted respectively. It can be seen that TAO has the best coverage, the least redundant area, and the least blank coverage. TAO also has the most uniform sensor distribution and the least sensor redundancy.

[0098] The above examples further illustrate the purpose, technical solutions and advantages of the application. It should be understood that the above examples are only preferred embodiments of the application and are not intended to limit the application. Any modifications, equivalent replacements, improvements, etc. made to the application within the spirit and principles of the application shall be included in the protection scope of the application.

Claims

1. A WSN deployment method based on a swarm intelligence optimization algorithm, characterized in that, The application relates to a WSN deployment method. The WSN deployment method comprises the following steps: S1, initializing a population by using an elite reverse learning strategy; the population comprises a plurality of individuals, and each individual represents the position of a sensor node; WSN objective function is: ； wherein, is a coverage area of the set of sensor nodes, is a coverage area of the sensor node, , is a length and a width of the target monitoring area; S2, updating the initialized population individuals in a first stage by using a motion strategy under Levy flight and a motion strategy under Brown motion, and updating the population individuals in a second stage by using an avoidance mechanism to obtain the population individuals in the second stage; S3, calculating the WSN target function value corresponding to the population individuals in the second stage, and selecting the population individual with the minimum WSN target function value as the best population individual, i.e. the best sensor node position; S4, when the maximum iteration number T is reached, the final population individual, i.e. the final sensor node position, is obtained; otherwise, returning to step S2 to update the best population individual. The WSN deployment method comprises the following steps: computing each population individual a corresponding WSN objective function value, selecting an elite node from the population individuals according to the WSN objective function values; wherein, is an index of the population individual, is an index of the dimension of the population individual, is a current iteration number; Computing adaptive parameters ; if greater than 0.5, then compute new population individuals from the elite nodes using a motion strategy under Levy flight ; otherwise, compute new population individuals from the elite nodes using a motion strategy under Brownian motion ; updating the elite node, calculating and the WSN objective function corresponding to the updated elite node, according to obtaining the first-stage updated population individual from the WSN objective function corresponding to the updated elite node; Adaptive parameters The calculation is as follows: ； wherein is a random function, is the maximum number of iterations; calculating new population individuals using a levin flying motion strategy comprising: ； where r is a random number from 0 to 1, denotes an elite node, is a function of the Levy flight, , denotes the two-dimensional component of the planar coordinate; calculating new population individuals using a motion strategy under brownian motion comprising: ； wherein is a Brownian motion strategy; According to the WSN objective function, the population individuals in the first stage are updated comprising: ； wherein, is the population individual after the first stage update, is the objective function value of the new population individual, is the objective function value of the elite node; S1, initializing a population by using an elite reverse learning strategy; the population comprises a plurality of individuals, and each individual represents the position of a sensor node; setting a warning value, calculating new population individuals based on the warning value : ； population individuals and population individuals corresponding WSN objective function, according to population individuals and population individuals corresponding WSN objective function to obtain the second-stage updated population individuals wherein, is a random number that is normally distributed, is a matrix, each element of which is 1, is a warning value; S2, updating the initialized population individuals in a first stage by using a motion strategy under Levy flight and a motion strategy under Brown motion, and updating the population individuals in a second stage by using an avoidance mechanism to obtain the population individuals in the second stage; S3, calculating the WSN target function value corresponding to the population individuals in the second stage, and selecting the population individual with the minimum WSN target function value as the best population individual, i.e. the best sensor node position; S4, when the maximum iteration number T is reached, the final population individual, i.e. the final sensor node position, is obtained; otherwise, returning to step S2 to update the best population individual. 2.The WSN deployment method based on swarm intelligence optimization algorithm according to claim 1, characterized in that, The WSN coverage model is constructed, comprising: constructing a sensor node group: wherein the sensor node , is a position of the sensor node, is a sensing radius of the sensor node; and calculating a coverage area of the sensor node according to the sensor node group.

Images