Improved three-dimensional sensor network optimization deployment method based on PSO

Abstract

The application discloses a kind of improved PSO three-dimensional sensor network optimization deployment method, the scheme includes: construct the three-dimensional space monitoring model of fusion topography elevation and barrier distribution;Design dynamic multi-objective fitness function of fusion space coverage, energy consumption balance degree and barrier penetration loss coefficient;According to the search weight of topographic gradient dynamic adjustment particle group in Z axis direction;Three-dimensional space is divided into sub-region and particle group is initialized in parallel;Improved PSO algorithm is used to iteratively optimize node position until the termination condition is met.The beneficial effects of the present application are: suitable for mountain monitoring, unmanned aerial vehicle cluster networking, ocean three-dimensional monitoring and other scenarios, can improve three-dimensional coverage by 15%-30%, network life is prolonged by more than 20%, and supports real-time node redeployment in dynamic environment.Through innovative three-dimensional space modeling, multi-objective optimization algorithm and parallel computing architecture, the overall improvement of sensor network deployment quality and efficiency in complex environment is realized.

Description

Technical Field

[0001] This invention relates to the technical field of wireless sensor network optimization, and in particular to an improved PSO method for optimizing the deployment of three-dimensional sensor networks. Background Technology

[0002] In the field of wireless sensor networks (WSNs), node deployment is a critical factor affecting network coverage quality, energy balance, and communication reliability. Traditional node deployment methods are mainly designed for two-dimensional planar scenarios, achieving node placement through static planning or simple random distribution. However, as application scenarios expand into three-dimensional space (such as mountain monitoring, UAV swarm networking, and three-dimensional marine monitoring), existing technologies face the following prominent challenges:

[0003] First, traditional deployment methods suffer from significantly insufficient coverage efficiency in 3D environments. Existing research largely relies on 2D planar assumptions and employs uniform distribution or 2D grid-based optimization strategies, which struggle to adapt to the spatial heterogeneity of complex 3D terrain. For example, in mountainous or urban environments, undulating terrain and obstructions complicate signal propagation paths, resulting in 3D coverage rates of only 40%-60% for traditional methods, leaving numerous monitoring blind spots. While some studies have attempted to introduce particle swarm optimization (PSO) algorithms into node deployment, their convergence speed in 3D space is slow, and they are prone to getting trapped in local optima, making it difficult to meet the real-time requirements of large-scale network deployments.

[0004] Secondly, existing deployment schemes suffer from poor energy consumption balance. In three-dimensional space, due to significant differences in communication distances between nodes, edge nodes often exhaust their energy prematurely due to undertaking excessive data forwarding tasks, leading to a shortened network lifetime. Experiments show that the energy consumption variance of three-dimensional networks optimized by traditional PSO algorithms is 30%-50% higher than that in two-dimensional scenarios, severely restricting the sustainability of long-term monitoring tasks. Although some studies have attempted to alleviate this problem by introducing energy consumption weighting coefficients, the lack of dynamic adaptation to the signal attenuation characteristics in three-dimensional space makes it difficult to achieve true energy consumption balance.

[0005] Secondly, there is insufficient adaptability to dynamic environments. In practical applications, 3D WSNs often face dynamic interference such as terrain changes (e.g., landslides), node movement (e.g., drone nodes), or temporary obstacles (e.g., newly added buildings). Existing methods mostly employ a global redeployment strategy, which incurs high computational overhead and response latency, failing to meet real-time requirements. For example, in drone swarm networking scenarios, the traditional PSO algorithm takes several minutes to complete a global redeployment, far exceeding the second-level response threshold allowed in practical applications.

[0006] To address the problems of existing node deployment methods in the field of Wireless Sensor Networks (WSN), Chinese patent CN 106792750 A, titled "A Node Deployment Method Based on Guided Particle Swarm Optimization Algorithm," proposes spatial segmentation and crossover operations. However, due to the inherent limitations of the particle swarm optimization algorithm, premature convergence occurs. Furthermore, while the patent aims to reduce node energy consumption, nodes still consume significant energy during movement, especially in high-density deployment scenarios. In their paper "Coverage Optimization of 3D Wireless Sensor Networks Based on Fusion Algorithm," scholars Sun Aijing et al. set the simulation environment to an idealized 3D cubic region. However, the performance of the algorithm in complex terrain or dynamic environments (such as mountains or underwater) has not been fully validated.

[0007] In summary, existing technologies in the field of 3D WSN node deployment suffer from core bottlenecks such as low coverage efficiency, uneven energy consumption, and slow dynamic response. There is an urgent need for a novel deployment method that can balance the characteristics of 3D space, multi-objective optimization requirements, and real-time adaptability. This invention effectively solves these problems by innovatively combining multi-objective fitness functions, dynamic PSO algorithms, and parallel computing architecture, providing a reliable technical solution for the efficient deployment of 3D WSNs. Summary of the Invention

[0008] The purpose of this invention is to overcome the shortcomings of existing technologies. It is applicable to scenarios such as mountain monitoring, UAV swarm networking, and marine three-dimensional monitoring. It can improve the three-dimensional coverage by 15%-30%, extend the network lifetime by more than 20%, and support real-time node redeployment in dynamic environments. Through innovative three-dimensional spatial modeling, multi-objective optimization algorithms, and parallel computing architecture, it achieves a comprehensive improvement in the quality and efficiency of sensor network deployment in complex environments. This is an improved PSO three-dimensional sensor network optimization deployment method.

[0009] The objective of this invention is achieved through the following technical solution: an improved method for optimizing the deployment of a three-dimensional sensor network in PSO, comprising the following steps:

[0010] S1. Construct a three-dimensional spatial monitoring model that integrates terrain elevation and obstacle distribution;

[0011] S2. Design a dynamic multi-objective fitness function that integrates spatial coverage, energy consumption balance, and obstacle penetration loss coefficient.

[0012] S3. Dynamically adjust the search weight of the particle swarm in the Z-axis direction based on the terrain gradient;

[0013] S4. Divide the three-dimensional space into sub-regions and initialize the particle swarm in parallel to optimize computational efficiency;

[0014] S5. Iteratively optimize the node position using the improved PSO algorithm until the termination condition is met;

[0015] S6 outputs the optimal 3D node deployment scheme and supports real-time redeployment in dynamic environments.

[0016] Preferably, the three-dimensional spatial monitoring model in step S1 includes: acquiring the digital elevation model (DEM) of the target area and the three-dimensional coordinate data of obstacles; establishing a grid-based three-dimensional discretization model, dividing the monitoring area into several cubic grid units; constructing a signal propagation model that comprehensively considers free space loss, diffraction loss, and penetration loss; and generating a three-dimensional visualization monitoring interface that can display the deployment status of nodes in real time.

[0017] Preferably, the multi-objective fitness function in step S2 is:

[0018] ;

[0019] In the formula, For spatial coverage, For energy consumption balance, The obstacle penetration loss coefficient. , , These are dynamic weighting coefficients.

[0020] Preferably, the search weight adjustment formula in the Z-axis direction in step S3 is:

[0021] ;

[0022] In the formula, Based on the weights, For terrain gradient, This represents the maximum elevation difference.

[0023] Preferably, the sub-region division in step S4 adopts the octree spatial partitioning method, and each sub-region independently initializes the particle swarm and optimizes it synchronously.

[0024] Preferably, the improved PSO algorithm in step S5 includes:

[0025] A dynamic inertia weight adjustment strategy is introduced, employing a nonlinear decreasing formula:

[0026] ;

[0027] In the formula, Indicates the first The inertia weight value at the next iteration These are the initial weight values, ranging from 0.8 to 0.95. The final weight value ranges from 0.3 to 0.5. This represents the current iteration number. The maximum number of iterations, This is the attenuation coefficient, with a value ranging from 0.5 to 2.0. This is a parameter for the shape of the attenuation curve, with a value range of 0.5-3.0;

[0028] The Cauchy mutation operator is used to enhance the local search. Cauchy mutation is applied to the global optimum, and its formula is as follows:

[0029] ;

[0030] In the formula, This is the globally optimal solution. The new solution after mutation For adaptive coefficient of variation, It obeys the position parameter 0 and the scale parameter. The Cauchy distribution has the following probability density function:

[0031] ;

[0032] In the formula, The scale parameter determines the amplitude of the disturbance;

[0033] Apply repulsive force constraints to nodes near obstacles:

[0034] ;

[0035] In the formula, For particles The three-dimensional position coordinates, The repulsive force coefficient, The current Euclidean distance between the node and the obstacle. The preset safe distance threshold, The distance gradient represents the unit direction vector from the node to the obstacle.

[0036] Preferably, the real-time redeployment in the dynamic environment in step S6 is achieved through the following steps:

[0037] Monitoring environmental changes triggers local area re-optimization; the environmental change monitoring model is as follows:

[0038] ;

[0039] In the formula, It is a moment Total environmental change It is the spatiotemporal dynamic environment data field function, the overall spatial volume of the monitoring area, the time interval between two detections, and the three-dimensional spatial volume element;

[0040] Preserving the node positions in unaffected areas is achieved through the following mathematical mechanism:

[0041] ;

[0042] In the formula, For binary identification functions, A collection of areas of environmental change;

[0043] The particle swarm optimization was recalculated within the changed region using an environmental change monitoring model. Identify areas of change Reset Position and velocity of the inner particle:

[0044] ;

[0045] In the formula, For particles new position coordinates For particles The original position coordinates, Use a binary identifier function to determine whether the coordinates belong to a region of change. , In the changing region A function that generates random positions uniformly distributed within the interior;

[0046] With local fitness function To achieve the goal, an improved PSO algorithm is used to iteratively update particle positions until the convergence condition is met:

[0047] ;

[0048] In the formula, the first row represents fitness convergence, which is the default. The second line represents the maximum number of iterations. .

[0049] The present invention has the following advantages:

[0050] 1. This invention constructs a 3D spatial monitoring model that integrates terrain elevation and obstacle distribution, combined with dynamic optimization of multi-objective fitness functions (coverage, energy consumption balance, and penetration loss coefficient), thereby improving network coverage by 15%-30% in complex environments and extending network lifetime by over 20%. Secondly, it innovatively employs a terrain gradient-adaptive Z-axis search weight adjustment mechanism and octree parallel partitioning technology, improving computational efficiency by 40%-60% and solving the problem of low efficiency in 3D spatial search. Thirdly, the improved PSO algorithm, through the synergistic effect of dynamic inertia weights, Cauchy mutation, and obstacle repulsion constraints, improves local optimum avoidance capability by 35% while maintaining convergence speed, achieving an obstacle avoidance success rate exceeding 98%. Finally, based on a real-time redeployment mechanism using an environmental change monitoring model, only local optimization is needed for the changed area (approximately 15% of the total volume), reducing dynamic adjustment response time by 70% and significantly improving the system's adaptability in scenarios such as mobile networking and disaster emergency response. These technological advancements collectively constitute an efficient and reliable 3D WSN deployment solution. Attached Figure Description

[0051] Figure 1 This is a schematic diagram of the process of the present invention;

[0052] Figure 2 This is a diagram of the three-dimensional spatial monitoring model of the present invention;

[0053] Figure 3 This is a diagram showing the dynamic adjustment of the weights of the multi-objective fitness function in this invention.

[0054] Figure 4 This is a diagram showing the relationship between the Z-axis search weights and the terrain gradient in this invention.

[0055] Figure 5 This is a diagram illustrating the octree space partitioning and parallel optimization of the present invention.

[0056] Figure 6 This is a flowchart of the particle swarm optimization process of the present invention;

[0057] Figure 7 A comparison chart showing the optimization effect of node deployment in this invention. Detailed Implementation

[0058] The present invention will be further described below with reference to the accompanying drawings. The scope of protection of the present invention is not limited to the following description:

[0059] like Figure 1 An improved method for optimizing the deployment of 3D sensor networks in PSO includes the following steps:

[0060] S1: Construct a three-dimensional spatial monitoring model that integrates terrain elevation and obstacle distribution;

[0061] S2: Design a multi-objective fitness function that integrates spatial coverage, energy consumption balance, and obstacle penetration loss coefficient;

[0062] S3: Dynamically adjust the search weight of the particle swarm in the Z-axis direction based on the terrain gradient;

[0063] S4: Divide the three-dimensional space into sub-regions and initialize the particle swarm in parallel to optimize computational efficiency;

[0064] S5: Iteratively optimize the node position using the improved PSO algorithm until the termination condition is met;

[0065] S6: Outputs the optimal 3D node deployment scheme and supports real-time redeployment in dynamic environments.

[0066] In step S1, constructing a three-dimensional spatial monitoring model includes acquiring the digital elevation model (DEM) of the target area and the three-dimensional coordinate data of obstacles, establishing a grid-based three-dimensional discretized model, dividing the monitoring area into several cubic grid cells, and labeling each cell with terrain elevation values ​​and obstacle markers; simultaneously, constructing a signal propagation model that comprehensively considers free space loss, diffraction loss, and penetration loss, calculating the multipath effect and attenuation coefficient of obstacles on wireless signals, and finally generating a three-dimensional visualization monitoring interface that can display the node deployment status and network coverage in real time; wherein, the resolution of the digital elevation model is not less than 1 meter × 1 meter, the side length of the cubic grid cell is set to 1 / 5 to 1 / 3 of the node communication radius, and obstacles include physical objects that affect signal propagation, such as buildings, vegetation, and terrain undulations.

[0067] like Figure 2 As shown, digital elevation model (DEM) data (resolution ≥ 1 m × 1 m) and obstacle 3D coordinate data (including entities such as buildings, vegetation, and terrain undulations) of the target area are obtained. After importing the data through the GIS system, a gridded 3D terrain surface (101) as shown in the figure is generated, where the X / Y / Z axes represent spatial coordinates (unit: meters); terrain elevation values ​​are represented by color gradients; and obstacles (102) are marked with red cubes.

[0068] In step S2, an innovative design scheme for a multi-objective fitness function is proposed. This function comprehensively considers key indicators of network deployment quality, and its specific expression is as follows:

[0069] ;

[0070] In the formula, For spatial coverage, For energy consumption balance, The obstacle penetration loss coefficient. , , These are dynamic weighting coefficients.

[0071] like Figure 3 As shown, the weighting coefficients are dynamically adjusted according to the following rules:

[0072] Initial stage (iterations 1-30): Covering weights The initial value is 0.6, and the energy consumption weight is... The initial value is 0.3, and the penetration weight is... The initial value is 0.1;

[0073] Mid-term stage (30-70 iterations): Non-linearity decreases to around 0.3. It increases linearly to around 0.48. Periodic fluctuations;

[0074] Later stage (70-100 iterations): The three factors tend to stabilize. To around 0.2, To around 0.7, Up to around 0.1.

[0075] In step S3, a dynamic adjustment mechanism for the search weights in the Z-axis direction is proposed. This method achieves intelligent search optimization in the vertical direction through the following formula:

[0076] ;

[0077] In the formula, The base weights are set to a range of 0.4-0.6, and are dynamically adjusted according to the network density. Higher values ​​(0.55-0.6) are used in sparse areas, and lower values ​​(0.4-0.45) are used in dense areas. The terrain gradient is calculated using the three-dimensional finite difference method to reflect the degree of undulation of the local terrain. The elevation change rate of the 3×3 neighborhood grid is considered during the calculation and the unit is normalized to the range of [0,1] for easy uniform quantitative comparison. To determine the maximum elevation difference, the elevation range (the difference between the highest and lowest points) within the monitoring area is taken.

[0078] Taking a mountainous monitoring area as an example, the topographic parameters are first obtained: maximum elevation difference. =100 meters, current elevation difference of the 3×3 grid =30 meters. Due to the sparse node distribution in this area (node ​​spacing > 2 times the communication radius), based on the dynamic adjustment rule of this invention, the basic weight is taken. =0.58. Calculated using the weighting formula. Substituting the scaling factor k=0.3, the current Z-axis search weight is calculated. =0.58+0.3×(30 / 100)=0.67.

[0079] In step S4, the sub-region partitioning adopts the octree spatial partitioning method, and each sub-region independently initializes the particle swarm and optimizes it synchronously.

[0080] like Figure 5 As shown, the monitoring area (100m×100m×50m) is recursively divided into 8 sub-cube regions using an octree. Each sub-region has dimensions of 50m×50m×25m (length×width×height), and the segmentation depth is set to 3 layers, ultimately forming 64 minimum units (12.5m×12.5m×6.25m). Particle swarms are initialized independently in each sub-region, with the number of particles distributed according to the volume ratio of the sub-region. The initialization positions follow a three-dimensional uniform distribution, and the 8 sub-regions are initialized synchronously through multi-threading.

[0081] In step S5, the PSO algorithm is improved, including:

[0082] A dynamic inertia weight adjustment strategy is introduced, employing a nonlinear decreasing formula:

[0083] ;

[0084] In the formula, Indicates the first The inertia weight value at the next iteration The initial weight values ​​(ranging from 0.8 to 0.95) The final weight value (ranging from 0.3 to 0.5) is used. This represents the current iteration number. The maximum number of iterations, The attenuation coefficient (range 0.5-2.0) The attenuation curve shape parameter (value range 0.5-3.0);

[0085] The Cauchy mutation operator is used to enhance the local search. Cauchy mutation is applied to the global optimum, and its formula is as follows:

[0086] ;

[0087] In the formula, This is the globally optimal solution. The new solution after mutation For adaptive coefficient of variation, It obeys the position parameter 0 and the scale parameter. The Cauchy distribution has the following probability density function:

[0088] ;

[0089] In the formula, The scale parameter determines the amplitude of the disturbance;

[0090] Apply repulsive force constraints to nodes near obstacles:

[0091] ;

[0092] In the formula, For particles The three-dimensional position coordinates; The repulsive force coefficient, The current Euclidean distance between the node and the obstacle. The preset safe distance threshold, The distance gradient represents the unit direction vector from the node to the obstacle.

[0093] In step S6, real-time redeployment in a dynamic environment is achieved through the following steps:

[0094] Monitoring environmental changes triggers local area re-optimization; the environmental change monitoring model is as follows:

[0095] ;

[0096] In the formula, It is a moment Total environmental change It is a spatiotemporal dynamic environment data field function. It is the overall spatial volume of the monitoring area. It is the time interval between the two tests. It is a three-dimensional spatial volume element;

[0097] Preserving the node positions in unaffected areas is achieved through the following mathematical mechanism:

[0098] ;

[0099] In the formula, For binary identification functions, A collection of areas of environmental change;

[0100] The particle swarm optimization was recalculated within the changed region using an environmental change monitoring model. Identify areas of change Reset Position and velocity of the inner particle:

[0101] ;

[0102] In the formula, For particles new position coordinates For particles The original position coordinates, Use a binary identifier function to determine whether the coordinates belong to a region of change. , In the changing region A function that generates random positions uniformly distributed within the interior;

[0103] With local fitness function To achieve the goal, an improved PSO algorithm is used to iteratively update particle positions until the convergence condition is met:

[0104] ;

[0105] In the formula, the first line in parentheses represents fitness convergence, which is the default. The second line represents the maximum number of iterations. .

[0106] Finally, it should be noted that the above descriptions are merely preferred embodiments of the present invention and are not intended to limit the present invention. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art can still modify the technical solutions described in the foregoing embodiments or make equivalent substitutions for some of the technical features. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. An improved method for optimizing the deployment of a three-dimensional sensor network using PSO, characterized in that: Includes the following steps: S1. Construct a three-dimensional spatial monitoring model that integrates terrain elevation and obstacle distribution; S2. Design a dynamic multi-objective fitness function that integrates spatial coverage, energy consumption balance, and obstacle penetration loss coefficient. S3. Dynamically adjust the search weight of the particle swarm in the Z-axis direction based on the terrain gradient; S4. Divide the three-dimensional space into sub-regions and initialize the particle swarm in parallel to optimize computational efficiency; S5. Iteratively optimize the node position using the improved PSO algorithm until the termination condition is met; The improved PSO algorithm in step S5 includes: A dynamic inertia weight adjustment strategy is introduced, employing a nonlinear decreasing formula: ； In the formula, Indicates the first The inertia weight value at the next iteration These are the initial weight values, ranging from 0.8 to 0.

95. The final weight value ranges from 0.3 to 0.

5. This represents the current iteration number. The maximum number of iterations, This is the attenuation coefficient, with a value ranging from 0.5 to 2.

0. This is a parameter for the shape of the attenuation curve, with a value range of 0.5-3.0; The Cauchy mutation operator is used to enhance the local search. Cauchy mutation is applied to the global optimum, and its formula is as follows: ； In the formula, This is the globally optimal solution. The new solution after mutation For adaptive coefficient of variation, It obeys the position parameter 0 and the scale parameter. The Cauchy distribution has the following probability density function: ； In the formula, The scale parameter determines the amplitude of the disturbance; Apply repulsive force constraints to nodes near obstacles: ； In the formula, For particles The three-dimensional position coordinates, The repulsive force coefficient, The current Euclidean distance between the node and the obstacle. The preset safe distance threshold, The distance gradient represents the unit direction vector from the node to the obstacle; S6. Output the optimal 3D node deployment scheme and support real-time redeployment in dynamic environments; The real-time redeployment in the dynamic environment in step S6 is achieved through the following steps: Monitoring environmental changes triggers local area re-optimization; the environmental change monitoring model is as follows: ； In the formula, It is the environment of a moment Total change It is a spatiotemporal dynamic environment data field function. It is the overall spatial volume of the monitoring area. It is the time interval between the two tests. It is a three-dimensional spatial volume element; Preserving the node positions in unaffected areas is achieved through the following mathematical mechanism: ； In the formula, For binary identification functions, A collection of areas of environmental change; The particle swarm optimization was recalculated within the changed region using an environmental change monitoring model. Identify areas of change Reset Position and velocity of the inner particle: ； In the formula, For particles New position coordinates For particles The original position coordinates, Use a binary identifier function to determine whether the coordinates belong to a region of change. , In the changing region A function that generates random positions uniformly distributed within the interior; With local fitness function To achieve the goal, an improved PSO algorithm is used to iteratively update particle positions until the convergence condition is met: ； In the formula, the first row represents fitness convergence, which is the default. The second line represents the maximum number of iterations. .

2. The improved PSO three-dimensional sensor network optimization deployment method according to claim 1, characterized in that: The three-dimensional spatial monitoring model in step S1 includes: acquiring the digital elevation model (DEM) of the target area and the three-dimensional coordinate data of obstacles; establishing a grid-based three-dimensional discretization model, dividing the monitoring area into several cubic grid units; constructing a signal propagation model that comprehensively considers free space loss, diffraction loss, and penetration loss; and generating a three-dimensional visualization monitoring interface that can display the deployment status of nodes in real time.

3. The improved PSO three-dimensional sensor network optimization deployment method according to claim 1, characterized in that: The multi-objective fitness function in step S2 is: ； In the formula, For spatial coverage, For energy consumption balance, This is the obstacle penetration loss coefficient. These are dynamic weighting coefficients.

4. The improved PSO three-dimensional sensor network optimization deployment method according to claim 1, characterized in that: The formula for adjusting the search weight in the Z-axis direction in step S3 is: ； In the formula, Based on the weights, For terrain gradient, This represents the maximum elevation difference.

5. The improved PSO three-dimensional sensor network optimization deployment method according to claim 1, characterized in that: In step S4, the sub-region partitioning adopts the octree spatial partitioning method, and each sub-region is independently initialized with particle swarm optimization and optimized synchronously.

Images