A WSN node localization method based on hybrid improved compressed particle swarm optimization

By introducing a hybrid improved compressed particle swarm optimization algorithm into the wireless sensor network, the problems of high resource loss, low accuracy and poor stability in node positioning are solved, and more efficient and stable positioning performance is achieved.

CN118265137BActive Publication Date: 2024-11-22SHENZHEN HONGYUE TECHNOLOGY CO LTD
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202410327287.2
Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
Filing Date
2024-03-21
Publication Date
2024-11-22
Estimated Expiration
2044-03-21

AI Technical Summary

Technical Problem

The existing DV-Hop algorithms have problems such as excessive resource loss, insufficient accuracy, and poor algorithm positioning stability when positioning nodes in wireless sensor networks.

Method used

The hybrid and improved compressed particle swarm optimization (HICPSO) algorithm is used, combined with the inertial weight dynamic change strategy and the Gaussian mutation mechanism, and the position information in the DV-Hop algorithm is optimized. Through iterative optimization, it gradually approaches the position of unknown nodes, and improves positioning accuracy. and stability.

Benefits of technology

It improves the node positioning accuracy in wireless sensor networks, reduces resource losses, and enhances the positioning stability of the algorithm, achieving more efficient positioning performance.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN118265137B_ABST
    Figure CN118265137B_ABST
Patent Text Reader

Abstract

The present invention proposes a WSN node positioning method based on hybrid improved compressed particle swarm optimization, including: S1, inputting fixed parameter variables of DV-Hop positioning system; S2, calculating the hop counts between unknown nodes and each anchor node; S3, obtaining the minimum hop counts from unknown nodes to each anchor node in the hop count set within the communication range, and then calculating the actual distance from each unknown node to the anchor node; S4, initializing local optimal and global optimal information; S5, HICPSO algorithm initializing unknown node position information; using the following positioning error model to initialize the initial position information and positioning error of unknown nodes; S6, if t≤T, iterating; S7, for all unknown nodes, updating the positioning speed; S8, updating the local optimal value and local optimal solution of the population; S9, t=t+1, then jumping to execute step S6; S10, updating the global optimal value and global optimal solution of the population, and finally obtaining the position of the WSN node. The present invention can improve positioning accuracy, effectively reduce resource loss and improve algorithm positioning stability.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The invention relates to the technical field of wireless sensor networks, and in particular to a WSN node positioning method based on hybrid improved compressed particle swarm optimization. Background Art

[0002] With the rapid development of science and technology, the node positioning problem in wireless sensor networks (WSN) has attracted much attention from the academic community due to its wide range of application fields. Wireless sensor networks (WSN) are composed of multiple small sensors with computing, communication and perception capabilities. With the advantages of easy deployment, low cost and strong monitoring capabilities, they are used to solve problems such as target tracking, target coverage, and target positioning. They have shown great advantages in application fields such as link loss distributed positioning, drone path planning and regional environmental monitoring. With the substantial increase in the demand for intelligent scientific computing and engineering application optimization, the further improvement of node positioning capabilities has become an optimization research problem that needs to be solved urgently in many industrial and agricultural fields.

[0003] The commonly used algorithms for target positioning problems are mainly divided into two categories: ranging-based and non-ranging-based algorithms. Ranging-based algorithms include RSSI, TOA, RTOF and other positioning technologies, while non-ranging algorithms include APIT, centroid algorithm and DV-Hop. Among them, RSSI and DV-Hop algorithms have attracted extensive attention from relevant scholars due to their advantages such as low implementation cost, strong optimizability and simple model. However, the DV-Hop algorithm has problems such as excessive resource loss, insufficient accuracy and poor algorithm positioning stability in wireless sensor network (WSN) node positioning. Summary of the invention

[0004] The present invention aims to at least solve the technical problems existing in the prior art, and in particular innovatively proposes a WSN node positioning method based on hybrid improved compressed particle swarm optimization.

[0005] In order to achieve the above-mentioned object of the present invention, the present invention provides a WSN node positioning method based on hybrid improved compressed particle swarm optimization, comprising the following steps:

[0006] S1, input the fixed parameter variables of the DV-Hop positioning system, the fixed parameter variables include network area, total number of nodes, number of anchor nodes, number of unknown nodes, communication range and communication radius; the network area is used as the search area for WSN node positioning.

[0007] S2, calculate the number of hops between the unknown node and each anchor node. The formula is as follows:

[0008]

[0009] Among them, (x i ,yi ) is the coordinate information of the unknown node, (x j ,y j ) is the coordinate information of the anchor node, h j The default value for the initial number of entries is 1, n i represents the i-th unknown node;

[0010] S3, within the communication range, obtain the minimum number of hops from the unknown node to each anchor node in the hop count set, and then calculate the actual distance from each unknown node to the anchor node;

[0011] S4, passing the data structure packet consisting of the communication range, the number of anchor nodes, the actual distance from the unknown node to the anchor node, the number of unknown nodes, the location information of all nodes, etc. into the HICPSO algorithm to initialize the local optimal and global optimal information (including the solution set and the objective function value);

[0012] S5, HICPSO algorithm initializes the unknown node position information; the following positioning error model is used to initialize the unknown node initialization position information and positioning error;

[0013]

[0014] Among them, (x, y) is the final estimated position of the unknown node after each iteration;

[0015] (x i ,y i ) is the coordinate information of the unknown node;

[0016] d i is the distance between the true position and the estimated position of each unknown node;

[0017] unnode is the total number of unknown nodes;

[0018] R is the communication radius;

[0019] S6, if t≤T, then iterate; where t is the current iteration number and T is the maximum iteration number;

[0020] S7, for all unknown nodes, the positioning speed is updated in combination with the compression factor:

[0021]

[0022] And use Gaussian mutation mechanism to update the positioning position:

[0023]

[0024] in, represents the updated velocity vector of the i-th individual in the j-th dimension of the velocity vector of the population at the t+1th iteration;

[0025] represents the updated velocity vector of the i-th individual in the j-th dimension of the velocity vector of the population at the t-th iteration;

[0026] X ij t+1 represents the updated position vector of the i-th individual in the j-th dimension of the position vector of the population at the t+1th iteration;

[0027] X ij t represents the updated position vector of the i-th individual in the j-th dimension of the position vector of the population at the t-th iteration;

[0028] K is the compression factor adjustment coefficient;

[0029] c is the nonlinear dynamic learning factor;

[0030] Gauss represents the Gaussian mutation operator;

[0031] w represents the iteration parameter;

[0032] pbest ij t represents the local optimal position of the i-th individual in the j-th dimension at the t-th iteration;

[0033] gbest j t represents the global optimal position of the jth dimension at the tth iteration;

[0034] S8, update the local optimal value and local optimal solution of the population;

[0035] S9, t=t+1, then jump to step S6;

[0036] S10, update the global optimal value and global optimal solution of the population, and finally obtain the location of the WSN node.

[0037] A hybrid improved compressed particle swarm algorithm (HICPSO) is introduced to optimize the least square method (LSM) position information deviation in DV-Hop, making the position information closer to the real position and effectively overcoming the common defects of DV-Hop algorithm such as insufficient positioning accuracy and poor positioning stability.

[0038] Furthermore, the inertia weight dynamic change strategy is used to adjust the parameter w of each iteration so that it decreases linearly in the interval [0.3, 0.8]. The calculation formula of the parameter w is as follows;

[0039]

[0040] By adopting the inertia weight dynamic change strategy to adaptively adjust w for each iteration, the convergence process can be accelerated and the algorithm can be prevented from falling into local extreme values ​​in the early stage of algorithm optimization, and the optimization process can be continued in the later stage.

[0041] Furthermore, the calculation formula of the compression factor adjustment coefficient K is as follows:

[0042]

[0043] The calculation formula of the nonlinear dynamic learning factor c is as follows:

[0044]

[0045] Where f(x i ) means bringing the i-th individual (unknown node) into f(x), where f(x) is the formula error;

[0046] f(x) min 、f(x) mean Respectively represent the minimum and average values ​​of all individual objective function values;

[0047] c max and c min is the adjustment interval of the learning factor.

[0048] The compression coefficient K in formula 15) can effectively reduce the evolution direction of the next velocity vector of the population, and can accurately approach a more accurate theoretical optimal position in the later stage of algorithm optimization. The function fitness value calculated by population X is used as the prompt information of the nonlinear dynamic mechanism to comprehensively adjust the learning factor c of PSO, which is more adaptable than the fixed parameter settings of the traditional local learning factor c1 and the global learning factor c2.

[0049] Furthermore, it also includes:

[0050] When the position vector after the positioning position update exceeds the search space solution, the k-th dimension best variable in the global best position is selected to perform boundary processing position update:

[0051] X ij t =gbest k 17)

[0052] Among them, gbest k represents the global optimal position of the kth dimension;

[0053] k follows a uniform distribution to randomly select a position information within the range of D for assignment, where D is the dimension.

[0054] During the iteration process, the model search will have a certain probability of generating an unreasonable solution set. At this time, the best selection strategy is used to replace the solutions that exceed the search space with the global optimal solution in the corresponding dimension. This abnormal position boundary processing solution can effectively speed up the search process.

[0055] In summary, due to the adoption of the above technical solutions, the present invention proposes a hybrid HICPSO and DV-Hop efficient node positioning method, including a HICPSO+DV-Hop node positioning algorithm for WSN. Under the conditions of the original sensor equipment, the algorithm adopts the adaptive search of the HICPSO intelligent algorithm, and gradually approaches the position of the unknown node through iterative optimization, so that the accuracy of the DV-Hop algorithm in estimating the unknown position through the jump distance reaches a new level, providing a theoretical basis and strong guarantee for new positioning needs. By optimizing the DV-Hop algorithm in WSN through a better improved PSO algorithm, the positioning accuracy can be improved, and the resource loss can be effectively reduced and the positioning stability of the algorithm can be improved. Specifically, the main contributions of the patent research work of the present invention are as follows:

[0056] 1) Aiming at the shortcomings of the particle swarm optimization model, a new hybrid improved compressed particle swarm algorithm (HICPSO) is proposed. By compressing the velocity vector and mixing the mutation of the individual vector, the best selection strategy is used to boundary the population, thus obtaining the HICPSO algorithm. The efficient and stable search advantages of the algorithm are verified through a test function set.

[0057] 2) Through the analysis of the advantages of the DV-Hop algorithm and the HICPSO algorithm, an adaptive search approximation model for unknown nodes is designed. The communication range is expanded by increasing the cooperative communication of known nodes to improve the efficient positioning capability of the HICPSO+DV-Hop algorithm.

[0058] 3) The positioning performance of our method is verified using two criteria: positioning error and control variable. The experimental results show that under the same conditions, the proposed positioning algorithm is more competitive.

[0059] Additional aspects and advantages of the present invention will be given in part in the following description and in part will be obvious from the following description, or will be learned through practice of the present invention. BRIEF DESCRIPTION OF THE DRAWINGS

[0060] The above and / or additional aspects and advantages of the present invention will become apparent and easily understood from the description of the embodiments in conjunction with the following drawings, in which:

[0061] Figure 1 It is a schematic diagram of the comparison of the convergence curves of the algorithm of the present invention. Figure 1 (a)~(h) are schematic diagrams showing the comparison of convergence curves under the F1~F8 test functions.

[0062] Figure 2 It is a schematic diagram of the comparison of the convergence curves of the algorithm of the present invention. Figure 2 (a)~(h) represent F9~F16 test functions respectively.

[0063] Figure 3 It is a schematic diagram of the comparison of the convergence curves of the algorithm of the present invention. Figure 3 (a)~(f) represent F17~F22 test functions respectively.

[0064] Figure 4 It is the spatial node distribution diagram of the present invention.

[0065] Figure 5 is the unknown node positioning error distribution diagram of the present invention, Figure 5 (a) to (f) respectively represent the error distribution comparison between each classic positioning algorithm and the algorithm proposed in this patent.

[0066] Figure 6 It is a schematic diagram of a positioning error curve of a change in communication radius of the present invention.

[0067] Figure 7 It is a schematic diagram of a positioning error curve with changes in the number of anchor nodes of the present invention. DETAILED DESCRIPTION

[0068] Embodiments of the present invention are described in detail below, examples of which are shown in the accompanying drawings, wherein the same or similar reference numerals throughout represent the same or similar elements or elements having the same or similar functions. The embodiments described below with reference to the accompanying drawings are exemplary and are only used to explain the present invention, and cannot be understood as limiting the present invention.

[0069] 1 HICPSO+DV-Hop algorithm

[0070] This section will introduce the PSO, DV-Hop basic algorithm model and the improvement of the hybrid improved compressed particle swarm algorithm (HICPSO). At the same time, it will introduce the HICPSO algorithm into the DV-Hop algorithm to minimize the error and optimize the positioning model, so as to construct a hybrid and efficient WSN node positioning algorithm.

[0071] 1.1 Basic Algorithm

[0072] 1.1.1 Particle Swarm Optimization

[0073] In 1995, Kennedy and Eberhart studied and analyzed the cooperative foraging behavior of bird groups and implemented a particle swarm algorithm that can be used for nonlinear function optimization and neural network training transplantation in multiple research paradigms. For example, the classic swarm intelligence optimization algorithms include the artificial bee colony algorithm (ABC) and the cuckoo search algorithm (CS). The particle swarm algorithm (PSO) is also a meta-heuristic swarm intelligence optimization algorithm, which has the characteristics of simple algorithm implementation principle, significant optimization effect, and few control parameters. Through optimization or improvement research, it has been proved that the computational performance advantages of its PSO variant can be used in multiple engineering application fields. In the optimization research and engineering application of swarm intelligence algorithms, the PSO algorithm has many advantages such as few control parameters, strong portability, and good optimization effect.

[0074] The basic principle of the particle swarm algorithm is: in the specified search space, it is assumed that there are N particles as a group to solve the optimization problem of D-dimensional scale in the search space, record the optimal objective function value fvalue and the corresponding individual particle position gbest each time, and iterate and optimize by continuously using the PSO update model. The update model is as follows.

[0075]

[0076]

[0077] in, represents the updated velocity vector of the i-th individual in the j-th dimension of the velocity vector of the population at the t+1th iteration;

[0078] X ij t+1 Represents the update velocity vector of the i-th individual in the j-th dimension of the position vector of the population at the t+1th iteration;

[0079] pbest ij t represents the local optimal position of the i-th individual in the j-th dimension at the t-th iteration;

[0080] gbest j t It represents the global optimal position of the jth dimension in the tth iteration, that is, the optimal individual is found, where j = 1…D, D is the dimension;

[0081] X and V represent the position vector and velocity vector of the population respectively. In the tth iteration, the update velocity vector of the i-th individual and the j-th dimension is calculated by formula (1). c1 and c2 represent the learning factors of the update model. r1 and r2 are pseudo-random numbers that obey uniform distribution. pbest is the local optimal position of each iteration. Formula (2) is used to search and update each iteration.

[0082] Subsequently, PSO proposed an inertia weight w to adjust the velocity vector of the update model on this basis, and obtained a particle swarm optimization algorithm with inertia weight adjustment. Formula (3) is a PSO velocity vector update variant.

[0083]

[0084] Wherein, w represents the inertia weight; the present invention will adopt the PSO of formula (2) and formula (3) as the basic algorithm to carry out the PSO improvement work.

[0085] 1.1.2 DV-Hop Algorithm

[0086] Distance Vector-Hop (DV-Hop) is a widely used node positioning algorithm. In recent years, the non-distance-based DV-Hop algorithm is similar to the principle of the classic routing protocol positioning algorithm. It has been used in high-precision node positioning problems with its advantages of low hardware overhead, simple implementation principle and wide application range. The DV-Hop positioning algorithm can be summarized into three steps. First, calculate the number of hops between the anchor node and the unknown node; second, calculate the hop distance between each unknown node and the anchor node by the Euclidean distance calculation method; finally, use the three-sided positioning algorithm to calculate the estimated coordinates of the unknown node through three or more anchor nodes within the communication range. The calculation of the first step is shown in formula (4).

[0087]

[0088] In formula (4), n i represents the i-th unknown node, (x i ,y i ) and (x j ,y j ) represent the coordinate information of unknown nodes and anchor nodes respectively, h i is the number of hops between anchor nodes in the data set. Using the Euclidean distance formula, the distance calculation equation of formula 6 can be listed.

[0089] d i =HopSize i ×hop i 5)

[0090]

[0091] Where HopSize i Indicates the step length from the anchor node to each unknown node, hop i represents the number of hops of each unknown node, d iis the distance between the true position and the estimated position of each unknown node, and (x, y) is the final estimated position of the unknown node after each iteration. The simultaneous equations (4)-(6) can be rewritten into the following operation equations by matrix transformation to solve the unknowns.

[0092] AX=b 7)

[0093]

[0094]

[0095]

[0096] Finally, the unknown node position information (x, y) is obtained by the least squares method.

[0097]

[0098] However, using the DV-Hop algorithm to obtain the unknown node position and the real positioning information will produce a certain error, resulting in low positioning accuracy. Therefore, we will introduce an improved particle swarm algorithm with efficient search effect, and iterate the optimization through error optimization to approximate the real position of the unknown node.

[0099] 1.2 Hybrid Improved Compressed Particle Swarm Optimization Strategy

[0100] First, the inertia weight dynamic change strategy is used to adjust the parameter w of each iteration so that it decreases linearly in the interval [0.3, 0.8], as shown in formula (12).

[0101]

[0102] Where T is the maximum number of iterations of the algorithm. Secondly, the compression factor and Gaussian mutation schemes are used to perform hybrid improvements on the velocity vector and position vector of the update model, and then the PSO update model variants (13) and (14) are obtained.

[0103]

[0104]

[0105] in represents the updated velocity vector of the i-th individual in the j-th dimension of the velocity vector of the population at the t+1th iteration;

[0106] X ij t+1 represents the updated position vector of the i-th individual in the j-th dimension of the position vector of the population at the t+1th iteration;

[0107] Gauss represents the Gaussian mutation operator.

[0108] K is the compression factor adjustment coefficient, c is the nonlinear dynamic learning factor, and their specific implementation principles are as follows.

[0109]

[0110]

[0111] Where f(x i ) represents the objective function value of the i-th individual;

[0112] f(x) min 、f(x) mean Represent the minimum function value and the average value of the entire population respectively.

[0113] c max and c min is the adjustment interval of the learning factor. Finally, since there may be unreasonable solutions (i.e., solutions beyond the search space) for the updated position vector of this model, the best selection scheme is used for boundary processing.

[0114] X ij t =gbest k 17)

[0115] gbest k represents the global optimal position of the k-th dimension;

[0116] When a position in the population exceeds the search space, the best variable of the kth dimension in the global best position is selected for boundary processing and position update, where k follows a uniform distribution to randomly select a position information within the range of D for assignment. Combined with the above PSO hybrid optimization work, a hybrid improved compressed particle swarm algorithm (HICSPO) with multiple search advantages is proposed. The pseudo code of the algorithm is as follows.

[0117]

[0118]

[0119] 1.3 HICPSO+DV-Hop hybrid optimization mechanism

[0120] Based on the relationship between the estimated distance of unknown nodes and anchor nodes listed in formula (6), a hybrid error analysis model of swarm intelligence optimization algorithm for DV-Hop node positioning has been proven to be feasible and effective. The principle is that the error between the estimated distance and the unknown node obtained by formula (6) can establish a minimum error optimization classification model using swarm intelligence optimization algorithm. The specific equation is shown below.

[0121] |r i -d i |<ε i 18)

[0122]

[0123] Among them, r i represents the estimated distance of a single unknown node i determined by the DV-Hop positioning model (the distance refers to the distance from the unknown node to the anchor node), d i represents the distance between the true position and the estimated position of each unknown node, ε i Represents the error between the estimated distance and the actual distance. At this time, an unknown node position data packet can be established through the HICPSO algorithm. The positioning optimization model of the anchor node can be transformed into an optimization problem for solution. An error optimization objective function can be established as follows:

[0124]

[0125] Where (x i ,y i ) is the coordinate information of the unknown node;

[0126] f(x,y) is the objective function;

[0127] R is the communication radius;

[0128] (x, y) is the node position during the HICPSO algorithm search process;

[0129] unnode is the total number of unknown nodes;

[0130] By continuously obtaining smaller errors during the HICPSO optimization process, the closest position of each unknown node is determined to enhance the accuracy of unknown node positioning. Finally, the HICPSO+DV-Hop hybrid positioning algorithm proposed in the present invention is formed, and the execution pseudo code of the algorithm is as follows.

[0131]

[0132]

[0133] 2 Intelligent Optimization Algorithm Performance Verification

[0134] 2.1 Experimental Setup

[0135] In order to verify the comprehensiveness of the optimization performance of the HICPSO algorithm, 22 classic test functions are used to carry out system experiments, among which F1-F9 are unimodal functions and F10-F22 are multimodal functions, which are used to verify the calculation speed and convergence performance of the HICPSO algorithm respectively. The test information of the specific functions is shown in Table 1.

[0136] Table 1. Benchmark function information

[0137] Function Name Search Space feature Function Name Search Space feature F1:Sphere <![CDATA[[-100,100] D ]]> Single Peak F12:NCRastrigin <![CDATA[[-5.12,5.12] D ]]> Multi-peak F2:Elliptic <![CDATA[[-100,100] D ]]> Single Peak F13:Griewank <![CDATA[[-600,600] D ]]> Multi-peak F3:SumSquare <![CDATA[[-10,10] D ]]> Single Peak F14: Schwefer2.26 <![CDATA[[-500,500] D ]]> Multi-peak F4:SumPower <![CDATA[[-1,1] D ]]> Single Peak F15:Ackley <![CDATA[[-50,50] D ]]> Multi-peak F5: Schwefel 2.22 <![CDATA[[-10,10] D ]]> Single Peak F16:Penalized 1 <![CDATA[[-100,100] D ]]> Multi-peak F6: Schwefel 2.21 <![CDATA[[-100,100] D ]]> Single Peak F17:Penalized 2 <![CDATA[[-100,100] D ]]> Multi-peak F7:Step <![CDATA[[-100,100] D ]]> Single Peak F18:Alpine <![CDATA[[-10,10] D ]]> Multi-peak F8:Exponential <![CDATA[[-10,10] D ]]> Single Peak F19:Levy <![CDATA[[-10,10] D ]]> Multi-peak F9:Quartic <![CDATA[[-1.28,1.28] D ]]> Single Peak F20: Weierstrass <![CDATA[[-1,1] D ]]> Multi-peak F10: Rosenbrock <![CDATA[[-5,10] D ]]> Multi-peak F21: Himmelblau <![CDATA[[-5,5] D ]]> Multi-peak F11: Rastrigin <![CDATA[[-5.12,5.12] D ]]> Multi-peak F22:Michalewicz <![CDATA[[0,π] D ]]> Multi-peak

[0138] The experiment includes two parts: analysis of optimization performance experimental results and comparison of search accuracy in high-dimensional space. CS, ABC, PSO, IAGA, BBPSO and other algorithms are used for performance comparison. The system test software is MATLAB2018b. The consistent parameter settings of each comparison algorithm are: 30 independent runs of the algorithm, 500 iterations, 20 population size, and 30 dimensions.

[0139] 2.2 Experimental Results and Analysis

[0140] Through system simulation experiments, the maximum value, minimum value, average value and standard deviation of each comparison algorithm were statistically analyzed for experimental data comparison, as shown in Tables 2 to 4.

[0141] Table 2 Algorithm data comparison (F1-F8)

[0142]

[0143]

[0144] Table 3 Algorithm data comparison (F9-F16)

[0145]

[0146]

[0147] Table 4 Algorithm data comparison (F17-F22)

[0148]

[0149]

[0150] The statistical data in Tables 2 to 4 were obtained by searching for the optimal value by running the algorithm 30 times. The above experimental statistical data show that all test functions of the HICPSO algorithm except F22 have similar or higher calculation accuracy than the comparison algorithm, and the overall optimization effect accounts for up to 90% of the total test functions. At the same time, there are more than 30 orders of magnitude optimizations in the test functions such as F1-F7, F11-F15, F18, F20-F21, etc., accounting for 68%. In particular, the theoretical optimal values ​​were calculated in the test functions such as F7, F11-F15, F20 and F21. It can be seen intuitively from the standard deviation that the statistical data of HICPSO are smaller than the standard deviation of the comparison algorithm, showing 100% search stability. It fully verifies that HICPSO has overcome the common defects of PSO and exhibits super computational search in the search process of the test function. In addition, the convergence curve comparison of each algorithm is as follows Figure 1-Figure 3 shown.

[0151] from Figure 1-Figure 3 It can be clearly seen from the comparison of the convergence curves of various algorithms that the HICPSO algorithm converges faster in the comparison diagrams of F1-F6, F9-F13 and F15-F19, and shows higher calculation accuracy in most test function searches. It can be seen that the HICPSO algorithm has excellent search performance in overcoming the problem that the PSO algorithm is prone to fall into early convergence.

[0152] In summary, the hybrid intelligent optimization solution proposed in the patent of this invention has great search advantages and strong computing performance in solving the problems of insufficient calculation accuracy, slow search speed and poor algorithm robustness in the calculation process of the intelligent optimization algorithm.

[0153] 3 Research and results analysis based on WSN node positioning application

[0154] After the performance of the HICPSO algorithm is verified, it is introduced into the DV-Hop in the WSN. The positioning error is minimized and optimized through the objective function (19), so that the position of the unknown node is continuously approximated. A simulation positioning experiment of the wireless sensor network is carried out under the same experimental environment setting, and the positioning situation of the HICPSO+DV-Hop algorithm is analyzed by controlling variables.

[0155] The system environment of the WSN node positioning experiment is set as follows: the sensor distribution space is 100*100m 2 , the total number of nodes (the sum of unknown nodes and known nodes) is 250, the number of anchor nodes (known nodes) is 50-90, the communication radius is 20-50, and the number of independent algorithm runs is 30 times. The positioning optimization parameter settings of the six algorithms are shown in Table 5.

[0156] Table 5 Parameter settings of the comparison algorithms

[0157]

[0158] When the communication radius R and the number of anchor nodes AnNode are both 50, and other control parameters remain unchanged, we can get Figure 4 The node distribution of the network. Under this node distribution, the positioning accuracy and stability of each algorithm are verified, as shown in Table 6.

[0159] Table 6 Comparison of positioning accuracy results

[0160]

[0161]

[0162] From the statistical data, we can know that the positioning algorithm introduced by HICPSO is 10% more accurate than the traditional DV-Hop algorithm. It is also better than the positioning algorithm of the three basic algorithms mixed together, and has a lower standard deviation than the six comparison algorithms, which verifies that our method has a very good positioning advantage in WSN target positioning. At the same time, the error distribution diagram of the unknown node of the HICPSO hybrid positioning algorithm and other comparison algorithms is visualized, as shown in the figure below: Figure 5 .

[0163] from Figure 5 It can be clearly seen that the positioning error distribution diagram of the HICPSO+DV-Hop algorithm with higher positioning accuracy has a more stable and smaller positioning error than the six comparisons, especially the positioning optimization capability of DV-Hop is very competitive.

[0164] Finally, the control variable method is used to control R in the interval [20,50] and AnNode in the interval [50,90], and the positioning error curves of each algorithm are compared, as shown in the figure below. Figure 6 and Figure 7 .

[0165] from Figure 6 It can be seen that the hybrid positioning algorithm that introduces the intelligent optimization algorithm has a more stable positioning error than the DV-Hop algorithm. At the same time, the HICPSO+DV-Hop algorithm has a higher positioning accuracy when the communication radius is between 40 and 50. From the perspective of the DV-Hop positioning error, the error of the HICPSO+DV-Hop algorithm gradually decreases with the increase of the communication radius, which is in line with the objective fact that theoretically increasing the communication radius improves the positioning accuracy.

[0166] like Figure 7,When the number of anchor nodes varies between 50 and 90, the multiple proposed positioning ,methods are significantly superior to other comparison algorithms, and show consistent ,comprehensive positioning performance at different numbers of anchor nodes. ,Therefore, the positioning error curve shows that our method can obtain ,higher WSN node positioning effect than other algorithms by using less resource ,loss in the same type of positioning applications, which provides a theoretical ,basis and practical possibility for the commissioning of wireless sensor ,networks.

[0167] In summary, the present invention proposes a hybrid improved compressed particle swarm optimization WSN node positioning algorithm. Compression factor, Gaussian variation, nonlinear adjustment mechanism, optimal selection and dynamic inertia weight are introduced to hybrid improve the PSO algorithm. The high efficiency of the HICPSO algorithm in local development and global exploration is verified through experiments of 22 test function sets, which effectively overcomes the defects of intelligent optimization algorithms in terms of calculation accuracy, search speed and algorithm robustness. Finally, the HICPSO+DV-Hop algorithm is compared with some classic intelligent optimization algorithms for WSN node positioning performance. The proposed positioning scheme shows excellent positioning ability in terms of positioning accuracy and algorithm stability. At the same time, the reduction of anchor nodes within a certain range has almost no effect on the positioning performance, which proves that the HICPSO+DV-Hop algorithm can save a lot of sensor equipment in the actual application process, and has the advantage and potential of not affecting the positioning performance. In the next step of research, the HICPSO+DV-Hop algorithm will be considered for practical engineering applications such as smart agriculture and environmental monitoring, so that HICPSO+DV-Hop can be further improved and developed.

[0168] Although the embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions and variations may be made to the embodiments without departing from the principles and spirit of the present invention, and that the scope of the present invention is defined by the claims and their equivalents.

Claims

1. A WSN node localization method based on hybrid improved compressed particle swarm optimization, characterized in that: The following steps are involved: S1, input fixed parameter variables of the DV-Hop positioning system, the fixed parameter variables include network area, total number of nodes, number of anchor nodes, number of unknown nodes, communication range and communication radius; S2, calculate the number of hops between the unknown node and each anchor node; S3, within the communication range, obtain the minimum number of hops from the unknown node to each anchor node in the hop count set, and then calculate the actual distance from each unknown node to the anchor node; S4, the data structure packet consisting of the communication range, the number of anchor nodes, the actual distance from the unknown node to the anchor node, the number of unknown nodes, and the location information of all nodes is passed into the HICPSO algorithm to initialize the local optimal and global optimal information; S5, HICPSO algorithm initializes the unknown node position information; the following positioning error model is used to initialize the unknown node initialization position information and positioning error; Among them, (x, y) is the final estimated position of the unknown node after each iteration; (x i ,y i ) is the coordinate information of the unknown node; d i is the distance between the true position and the estimated position of each unknown node; unnode is the total number of unknown nodes; R is the communication radius; S6, if t≤T, then iterate; where t is the current iteration number and T is the maximum iteration number; S7, for all unknown nodes, the positioning speed is updated in combination with the compression factor: And use Gaussian mutation mechanism to update the positioning position: X ij t+1 =Gauss*X ij t +V ij t+1 in, represents the updated velocity vector of the i-th individual in the j-th dimension of the velocity vector of the population at the t+1th iteration; represents the updated velocity vector of the i-th individual in the j-th dimension of the velocity vector of the population at the t-th iteration; X ij t+1 represents the updated position vector of the i-th individual in the j-th dimension of the position vector of the population at the t+1th iteration; X ij t represents the updated position vector of the i-th individual in the j-th dimension of the position vector of the population at the t-th iteration; K is the compression factor adjustment coefficient; c is the nonlinear dynamic learning factor; Gauss represents the Gaussian mutation operator; w represents the iteration parameter; pbest ij t represents the local optimal position of the i-th individual in the j-th dimension at the t-th iteration; gbest j t represents the global optimal position of the jth dimension at the tth iteration; S8, update the local optimal value and local optimal solution of the population; S9, t=t+1, then jump to step S6; S10, update the global optimal value and global optimal solution of the population, and finally obtain the location of the WSN node.

2. According to claim 1, a WSN node positioning method based on hybrid improved compressed particle swarm optimization is characterized in that: The inertia weight dynamic change strategy is used to adjust the parameter w of each iteration so that it decreases linearly in the interval [0.3, 0.8]. The calculation formula of the parameter w is as follows; 3. The WSN node positioning method based on hybrid improved compressed particle swarm optimization according to claim 1 is characterized in that: The calculation formula of the compression factor adjustment coefficient K is as follows: The calculation formula of the nonlinear dynamic learning factor c is as follows: Where f(x i ) means to bring the ith individual into f(x), where f(x) is the formula error; f(x) min 、f(x) mean Respectively represent the minimum and average values ​​of all individual objective function values; c max and c min is the adjustment interval of the learning factor.

4. The WSN node positioning method based on hybrid improved compressed particle swarm optimization according to claim 1 is characterized in that: Also includes: When the position vector after the positioning position update exceeds the search space solution, the k-th dimension best variable in the global best position is selected to perform boundary processing position update: X ij t =gbest k Among them, gbest k represents the global optimal position of the kth dimension; k follows a uniform distribution to randomly select a position information within the range of D for assignment, where D is the dimension.