A multi-scale wireless channel propagation loss prediction method based on a finite element method

Abstract

The application discloses a multi-scale wireless channel propagation loss prediction method based on a finite element method, S1. Forming an environment description data set; S2. Forming a macro-scale finite element model; S3. Generating a macro-scale ray tracing simulation result; S4. Forming a local-scale finite element submodel; S5. Multi-scale information fusion is carried out on the wireless channel propagation loss data obtained from the macro-scale ray tracing simulation result and the local-scale finite element submodel, a fused wireless channel propagation loss prediction model is formed, and the fusion process realizes mutual correction of global propagation trends and local detail effects; S6. Based on the fused wireless channel propagation loss prediction model, the wireless signal propagation loss in a target environment is calculated, a wireless channel propagation loss prediction result is generated, and post-processing is carried out on the wireless channel propagation loss prediction result. The application realizes dynamic balance of local-scale and macro-scale information through a scale coupling function.

Description

Technical Field

[0001] This invention relates to the field of wireless communication technology, and in particular to a multi-scale wireless channel propagation loss prediction method based on the finite element method. Background Technology

[0002] With the rapid development of wireless communication technology, wireless channel propagation loss prediction plays a crucial role in mobile communication, the Internet of Things, and intelligent transportation. The propagation loss of wireless signals is affected by various factors during the propagation process, which leads to complex spatial variation characteristics. Therefore, accurate prediction of wireless signal propagation loss is of great significance for optimizing network deployment, improving communication quality, and reducing energy consumption.

[0003] Currently, the mainstream methods for predicting wireless channel propagation loss mainly include empirical models, ray tracing methods, and finite element method. Empirical models obtain the statistical characteristics of channel propagation loss through regression analysis of a large amount of measured data, and are computationally simple. However, the applicability of the model is constrained by the scenario and it is difficult to accurately characterize the signal propagation characteristics in complex environments. Ray tracing methods are based on geometric optics theory and use the reflection, diffraction, and transmission characteristics of rays to simulate the wireless signal propagation path, which can accurately simulate the wireless channel propagation process. However, the computational complexity is high, especially in large-scale scenarios where the computational load increases dramatically, making it difficult to meet the needs of real-time prediction. In addition, ray tracing methods are easily affected by occlusion effects in complex environments, leading to increased prediction errors.

[0004] In contrast, the finite element method (FEM) has received widespread attention from researchers in recent years due to its numerical computational advantages in solving electromagnetic wave propagation problems. The FEM can perform refined analysis of electromagnetic environments at different spatial scales through discretized modeling, thereby improving the accuracy of propagation loss prediction. However, traditional single-scale FEM modeling methods face a dilemma between computational efficiency and accuracy: macroscopic-scale FEM models are computationally efficient but struggle to accurately depict local environmental details, such as diffraction and scattering effects of small-scale building structures; while high-precision local modeling methods can improve prediction accuracy, their computational cost is too high, making it difficult to meet the application needs of large-scale scenarios. Furthermore, traditional FEM methods still have certain limitations in handling the multipath effects of wireless signals, making it difficult to effectively couple global propagation trends with local detail changes, thus limiting the applicability of the prediction model.

[0005] In summary, existing technologies for predicting wireless channel propagation loss suffer from the following main shortcomings: empirical models have limited applicability and are difficult to accurately describe propagation characteristics in complex environments; ray tracing methods involve large computational loads and are difficult to achieve efficient computation in large-scale scenarios; traditional finite element methods struggle to balance computational efficiency and accuracy and cannot effectively handle multi-scale wireless channel propagation characteristics. These technological bottlenecks severely impact the planning, deployment, and optimization of wireless communication systems, necessitating a new method that can efficiently and accurately predict wireless channel propagation loss to overcome the limitations of existing technologies. Summary of the Invention

[0006] One objective of this invention is to propose a multi-scale wireless channel propagation loss prediction method based on the finite element method. This invention achieves a dynamic balance between local and macroscopic scale information through a scale coupling function.

[0007] A multi-scale wireless channel propagation loss prediction method based on the finite element method according to an embodiment of the present invention includes the following steps:

[0008] S1. Acquire target wireless channel propagation environment data, and preprocess the environment data to form an environment description dataset;

[0009] S2. Based on the finite element method, the environmental description dataset is discretized at the macro scale to form a macro scale finite element model;

[0010] S3. Use the ray tracing algorithm to simulate the wireless signal propagation path of the macroscopic finite element model and generate macroscopic ray tracing simulation results;

[0011] S4. Based on the macroscopic ray tracing simulation results, identify local regions with significant changes in propagation loss, and refine the model of these local regions using the finite element method to form a local-scale finite element sub-model.

[0012] S5. Multi-scale information fusion is performed on the macro-scale ray tracing simulation results and the wireless channel propagation loss data obtained from the local-scale finite element sub-model to form a fused wireless channel propagation loss prediction model. The fusion process achieves mutual correction between global propagation trends and local detail effects.

[0013] S6. Based on the fused wireless channel propagation loss prediction model, calculate the wireless signal propagation loss in the target environment, generate wireless channel propagation loss prediction results, and post-process the wireless channel propagation loss prediction results to output a wireless channel propagation loss prediction map containing signal coverage, attenuation distribution and loss indicators of key areas.

[0014] Optionally, step S1 includes the following steps:

[0015] S11. Collect target wireless channel propagation environment data, including terrain information, building geometry, building material properties, and electromagnetic environment parameters, and construct an environment description dataset D. env :

[0016]

[0017] Where, d i T represents the i-th target wireless channel propagation environment data unit, N is the total number of target wireless channel propagation environment data units collected, and T i For terrain information, including surface elevation, surface roughness, and slope parameters, B i For the building's geometry, including building height, floor area, shape outline, and relative coordinate information, M i For building material properties, including dielectric constant, conductivity, and absorption coefficient, E i These are electromagnetic environment parameters, including background noise intensity, location of electromagnetic interference sources, and spectrum information.

[0018] S12. For the environmental description dataset D env Perform data format conversion, unify data coordinate system, set global coordinate system as standard coordinate reference, convert target wireless channel propagation environment data into standard coordinate representation, and obtain the transformed standardized environment description dataset.

[0019] S13. Perform a data integrity check on the standardized environmental description dataset, remove incomplete data units, and use interpolation to fill in missing values ​​to form a complete environmental description dataset;

[0020] S14. Invalid data is removed from the complete environment description dataset. Invalid data units are filtered according to the set target wireless channel propagation environment data validity criteria to obtain the final filtered environment description dataset D. final :

[0021]

[0022] Where, N final T′ is the number of data units in the target wireless channel propagation environment after removing invalid data. i B′ i M′ i and E′ i This includes filtered terrain information, building geometry, building material properties, and electromagnetic environment parameters.

[0023] Optionally, step S2 includes the following steps:

[0024] S21. Based on the final environment description dataset Dfinal Describe the dataset unit d′ for each target wireless channel propagation environment i Define the corresponding local computation region Ω i and its boundaries The importance of local areas is evaluated by using the spatial gradient of electromagnetic environment parameters in the environmental description dataset, and a weighted index function χ² is constructed. i (x):

[0025]

[0026] Among them, E′ i (x) represents the standardized electromagnetic environment parameter at point x. Let α be its spatial gradient, and α be a positive definite coefficient for adjusting the sensitivity to local changes, based on the weighted index function χ. i (x) After setting the threshold τ, define the overall computational region Ω, and determine the boundary of the computational region based on the overall computational region.

[0027] S22. Within the overall computational domain Ω, an adaptive multi-objective mesh reconstruction algorithm is used for discretization, dividing the overall computational domain into a set of finite element mesh elements G. macro :

[0028]

[0029] Among them, e j V represents the j-th grid cell. j f is the node coordinate vector. j N represents local medium properties. e ω represents the total number of grid cells. j The grid weights are based on the estimated initial propagation loss. and its spatial gradient Sure:

[0030]

[0031] Where β is the normalization coefficient, δ is a small constant to prevent division by zero, and h j The size of the mesh cell is determined by the weight ω. j Adaptive adjustment:

[0032]

[0033] Among them, h max κ is the maximum allowable cell size, and κ is the control parameter.

[0034] S23. In each grid cell e j Internally, based on local medium property f jWe construct the finite element control equations for wireless channel propagation and introduce multi-scale correction terms to achieve coupling between global and local information:

[0035]

[0036] Where E is the electric field intensity vector; μ r,j With ε r,j Let be the local relative permeability and relative permittivity of the j-th element, respectively; ω be the signal angular frequency; J be the excitation source current density; and λ be the local relative permeability and relative permittivity of the j-th element. j For correction factors:

[0037]

[0038] Where η is the positive definite adjustment coefficient;

[0039] S24. Finite element basis functions are used to locally interpolate the electric field, while scale coupling functions are introduced to fuse multi-scale information:

[0040]

[0041] in, For traditional finite element basis functions, ξ j (x) is the scale coupling function:

[0042]

[0043] Where σ is the coupling strength coefficient, N b,j E represents the number of basis functions in the j-th grid cell. k,j The coefficients to be solved;

[0044] S25. A feedback iteration mechanism is used to dynamically correct the deviation between the preliminary calculation results and the target propagation loss, thereby achieving adaptive coupling between the global and local models:

[0045]

[0046] Among them, the loss function Defined as:

[0047]

[0048] in, The propagation loss calculated for the j-th grid cell in the nth iteration. To determine the target propagation loss based on the environmental description dataset and prior information, α iter This is the iteration step size;

[0049] The iterative process simultaneously adjusts the grid weights ω in each update cycle. jBy combining the coupling function parameters, a macroscopic-scale finite element model is finally obtained, forming a wireless channel propagation loss prediction dataset D. macro .

[0050] Optionally, step S3 includes the following steps:

[0051] S31. Calculate dataset D based on macroscopic-scale finite element model. macro Construct the ray tracing path search space Ω R Based on the location of the target wireless signal source S0 = (x0, y0, z0) and the receiving point R m =(x m ,y m ,z m Define the global ray path set:

[0052]

[0053] Among them, P m,n This indicates the distance from signal source S0 to receiver R. m The nth possible propagation path, L k,n N is the k-th reflection or diffraction point on the path. p The number of all possible paths;

[0054] S32. For each receiving point R m Calculate its direct path P to the signal source S0. m,0 The path length is d m,0 If the path is not blocked by buildings in the environment, then calculate the propagation loss of the direct path. If the path is blocked, the direct path propagation loss is not calculated, and the multipath propagation path calculation is performed instead.

[0055] S33. For all building surfaces B j ∈D macro Perform boundary analysis to calculate the set of propagation paths that satisfy the Fresnel reflection condition.

[0056]

[0057] Where, θ i and θ r Let the incident angle and the reflection angle represent the incident angle and the reflection angle, respectively, and calculate the propagation loss along the reflection path:

[0058]

[0059] Where R(θ) i () represents the reflection coefficient;

[0060] S34. Calculating the set of diffraction paths based on the geometric features of building edges Define the propagation loss of the diffraction path:

[0061]

[0062] Where D(ν) is the diffraction attenuation factor and ν is the diffraction angle;

[0063] S35. Calculate the propagation loss at the receiving point R, considering the propagation losses of the direct path, reflected path, and diffracted path. m Total signal power at:

[0064]

[0065] Among them, P m,n For path P m,n The initial signal power during propagation;

[0066] S36. Summarize the propagation loss data from all receiving points to form a macroscopic ray tracing simulation results dataset:

[0067]

[0068] Among them, L m For receiving point R m Total propagation loss at the point, M represents the received signal power, and M represents the total number of receiving points.

[0069] Optionally, step S4 includes the following steps:

[0070] S41. Dataset D based on macroscopic-scale ray tracing simulation results RT Define the propagation loss gradient field Used to assess the spatial rate of change of propagation loss and to calculate the mean square rate of change Γ of the gradient field. m :

[0071]

[0072] Among them, L m For receiving point R m The total propagation loss at point (x, y, z) is the spatial coordinate;

[0073] And set an adaptive threshold τ Γ To identify local regions where the change in propagation loss exceeds a threshold:

[0074]

[0075] S42. For the local region Ω local Discretization is performed using a non-uniform adaptive mesh refinement strategy, and a local-scale finite element mesh set G is defined. local :

[0076]

[0077] Among them, e′ k Let v′ represent the k-th local finite element mesh. k f′ represents the coordinates of the grid nodes. k For the medium properties within the mesh, ω′ k h′ represents the weight of the local finite element mesh. k The size of the local finite element mesh;

[0078] S43. In each local finite element mesh element e′ k Internally, based on the medium property f′ within the mesh k Construct the finite element control equations for local-scale wireless signal propagation;

[0079] S44. An adaptive coupling interpolation method is used to fuse local and macroscopic scale information to correct the electric field;

[0080] S45. Summarize the calculation results of all local meshes to form a local-scale finite element sub-model dataset:

[0081] D local ={(e′ k ,E′ k ,L′ k |k=1,2,…,N l},

[0082] Among them, E′ k L′ represents the local electric field intensity within a local finite element mesh. k For local propagation loss, N l The number of local finite element mesh elements is the local scale finite element method.

[0083] Optionally, step S5 includes the following steps:

[0084] S51. Dataset D based on macroscopic-scale ray tracing simulation results RT With local scale finite element sub-model dataset D local Construct a multi-scale propagation loss dataset D multi =D RT ∪D local And define the multi-scale propagation loss computation domain Ω multi =Ω macro ∪Ω local ;

[0085] Among them, Ω macro For the macroscopic propagation domain, Ω local For local scale propagation domain;

[0086] S52. Define a multi-scale fusion weighting function Ψ(x) to describe the contribution ratio of local and macro-scale data, and establish a fusion loss calculation model:

[0087]

[0088] in, and These are the loss gradients at the local and macroscopic scales, respectively; the closer Ψ(x) is to 1, the greater the contribution at the local scale, and the closer it is to 0, the greater the contribution at the macroscopic scale.

[0089] S53. Using multi-scale loss fusion to calculate the loss fusion between global propagation trend and local detail effects:

[0090] L fused (x)=(1-Ψ(x))·L macro (x)+Ψ(x·L local (x);

[0091] Among them, L fused (x) represents the propagation loss of the fused wireless channel, L macro (x) represents the propagation loss at the macroscopic scale, L local (x) represents the local scale propagation loss;

[0092] S55. Based on the multi-scale propagation loss dataset D multi And by fusing the loss calculation model, a dataset of fused wireless channel propagation loss prediction models is generated:

[0093] D fused ={(x,L fused (x),P fused (x))∣x∈Ω multi},

[0094] Among them, P fused (x) represents the power of the received signal after fusion.

[0095] The beneficial effects of this invention are:

[0096] (1) By combining the finite element method and the ray tracing algorithm, this invention proposes a multi-scale modeling strategy that can construct the overall propagation environment at the macro scale and at the local scale, refine the modeling of the region where the propagation loss changes drastically. Through dynamic adaptive region division, the importance of the local region is evaluated by using the propagation loss gradient, and a more refined mesh division and solution method is adopted in the local region, so that the overall prediction model can improve the accuracy of loss calculation while ensuring computational efficiency.

[0097] (2) This invention combines the ray tracing method with the finite element method. The ray tracing algorithm is used to quickly calculate the channel propagation path at the macro scale to determine the region of sudden change in propagation loss. Then, the finite element method is introduced in the key region for fine calculation. The region with large change in propagation loss is screened by ray tracing, and the finite element model is used to refine the model in the local region, which reduces unnecessary calculation overhead and reduces the overall calculation complexity.

[0098] (3) This invention proposes an adaptive multi-objective grid reconstruction algorithm. Based on the changes in the spatial gradient and propagation loss of the electromagnetic environment, the computational grid is dynamically adjusted so that the key propagation area adopts a finer grid division, while the area with stable propagation changes maintains a larger grid scale to reduce unnecessary calculations. The adaptive grid technology can improve the computational accuracy of the key area while ensuring the optimization of computational resources. In addition, the multi-scale information fusion method uses the global propagation trend to correct the local propagation calculation error and achieves a dynamic balance between local scale and macro scale information through the scale coupling function. Attached Figure Description

[0099] The accompanying drawings are provided to further illustrate the invention and form part of the specification. They are used in conjunction with embodiments of the invention to explain the invention and do not constitute a limitation thereof. In the drawings:

[0100] Figure 1 This is a flowchart of a multi-scale wireless channel propagation loss prediction method based on the finite element method proposed in this invention. Detailed Implementation

[0101] The present invention will now be described in further detail with reference to the accompanying drawings. These drawings are simplified schematic diagrams, illustrating only the basic structure of the invention, and therefore only show the components relevant to the invention.

[0102] refer to Figure 1 A multi-scale wireless channel propagation loss prediction method based on the finite element method includes the following steps:

[0103] S1. Acquire target wireless channel propagation environment data and preprocess the environment data to form an environment description dataset;

[0104] S2. Based on the finite element method, the environmental description dataset is discretized at the macro scale to form a macro scale finite element model;

[0105] S3. Use the ray tracing algorithm to simulate the wireless signal propagation path of the macroscopic finite element model and generate macroscopic ray tracing simulation results;

[0106] S4. Based on the macroscopic ray tracing simulation results, identify local regions with significant changes in propagation loss, and refine the model of these local regions using the finite element method to form a local-scale finite element sub-model.

[0107] S5. Multi-scale information fusion is performed on the macro-scale ray tracing simulation results and the wireless channel propagation loss data obtained from the local-scale finite element sub-model to form a fused wireless channel propagation loss prediction model. The fusion process achieves mutual correction between global propagation trends and local detail effects.

[0108] S6. Based on the fused wireless channel propagation loss prediction model, calculate the wireless signal propagation loss in the target environment, generate wireless channel propagation loss prediction results, and post-process the wireless channel propagation loss prediction results to output a wireless channel propagation loss prediction map containing signal coverage, attenuation distribution and loss indicators of key areas.

[0109] In this embodiment, S1 includes the following steps:

[0110] S11. Collect target wireless channel propagation environment data, including terrain information, building geometry, building material properties, and electromagnetic environment parameters, and construct an environment description dataset D. env :

[0111]

[0112] Where, d i T represents the i-th target wireless channel propagation environment data unit, N is the total number of target wireless channel propagation environment data units collected, and T i For terrain information, including surface elevation, surface roughness, and slope parameters, B i For the building's geometry, including building height, floor area, shape outline, and relative coordinate information, M i For building material properties, including dielectric constant, conductivity, and absorption coefficient, E i These are electromagnetic environment parameters, including background noise intensity, location of electromagnetic interference sources, and spectrum information.

[0113] S12. Environment description dataset D env Perform data format conversion, unify data coordinate system, set global coordinate system as standard coordinate reference, convert target wireless channel propagation environment data into standard coordinate representation, and obtain the transformed standardized environment description dataset.

[0114] S13. Perform a data integrity check on the standardized environmental description dataset, remove incomplete data units, and use interpolation to fill in missing values ​​to form a complete environmental description dataset;

[0115] S14. Remove invalid data from the complete environment description dataset. Filter invalid data units according to the set target wireless channel propagation environment data validity criteria to obtain the final filtered environment description dataset D. final :

[0116]

[0117] Where, N final T′ is the number of data units in the target wireless channel propagation environment after removing invalid data. i B′ i M′ i and E′ i This includes filtered terrain information, building geometry, building material properties, and electromagnetic environment parameters.

[0118] In this embodiment, S2 includes the following steps:

[0119] S21. Based on the final environment description dataset D final Describe the dataset unit d′ for each target wireless channel propagation environment i Define the corresponding local computation region Ω i and its boundaries The importance of local areas is evaluated by using the spatial gradient of electromagnetic environment parameters in the environmental description dataset, and a weighted index function χ² is constructed. i (x):

[0120]

[0121] Among them, E′ i (x) represents the standardized electromagnetic environment parameter at point x. Let α be its spatial gradient, and α be a positive definite coefficient for adjusting the sensitivity to local changes, based on the weighted index function χ. i (x) After setting the threshold τ, define the overall computational region Ω, and determine the boundary of the computational region based on the overall computational region.

[0122] S22. An adaptive multi-objective mesh reconstruction algorithm is used to discretize the overall computational domain Ω, dividing the overall computational domain into a set of finite element mesh elements G. macro :

[0123]

[0124] Among them, e j V represents the j-th grid cell. j f is the node coordinate vector. j N represents local medium properties. e ω represents the total number of grid cells.j The grid weights are based on the estimated initial propagation loss. and its spatial gradient Sure:

[0125]

[0126] Where β is the normalization coefficient, δ is a small constant to prevent division by zero, and h j The size of the mesh cell is determined by the weight ω. j Adaptive adjustment:

[0127]

[0128] Among them, h max κ is the maximum allowable cell size, and κ is the control parameter.

[0129] S23. In each grid cell e j Internally, based on local medium property f j We construct the finite element control equations for wireless channel propagation and introduce multi-scale correction terms to achieve coupling between global and local information:

[0130]

[0131] Where E is the electric field intensity vector; μ r,j With ε r,j Let be the local relative permeability and relative permittivity of the j-th element, respectively; ω be the signal angular frequency; J be the excitation source current density; and λ be the local relative permeability and relative permittivity of the j-th element. j For correction factors:

[0132]

[0133] Where η is the positive definite adjustment coefficient;

[0134] S24. Finite element basis functions are used to locally interpolate the electric field, while scale coupling functions are introduced to fuse multi-scale information:

[0135]

[0136] in, For traditional finite element basis functions, ξ j (x) is the scale coupling function:

[0137]

[0138] Where σ is the coupling strength coefficient, N b,j E represents the number of basis functions in the j-th grid cell. k,j The coefficients to be solved;

[0139] S25. A feedback iteration mechanism is used to dynamically correct the deviation between the preliminary calculation results and the target propagation loss, thereby achieving adaptive coupling between the global and local models:

[0140]

[0141] Among them, the loss function Defined as:

[0142]

[0143] in, The propagation loss calculated for the j-th grid cell in the nth iteration. To determine the target propagation loss based on the environmental description dataset and prior information, α iter This is the iteration step size;

[0144] The iterative process simultaneously adjusts the grid weights ω in each update cycle. j By combining the coupling function parameters, a macroscopic-scale finite element model is finally obtained, forming a wireless channel propagation loss prediction dataset D. macro .

[0145] In this embodiment, S3 includes the following steps:

[0146] S31. Calculate dataset D based on macroscopic-scale finite element model. macro Construct the ray tracing path search space Ω R Based on the location of the target wireless signal source S0 = (x0, y0, z0) and the receiving point R m =(x m ,y m ,z m Define the global ray path set:

[0147]

[0148] Among them, P m,n This indicates the distance from signal source S0 to receiver R. m The nth possible propagation path, L k,n N is the k-th reflection or diffraction point on the path. p The number of all possible paths;

[0149] S32. For each receiving point R m Calculate its direct path P to the signal source S0. m,0 The path length is d m,0 If the path is not blocked by buildings in the environment, then calculate the propagation loss of the direct path. If the path is blocked, the direct path propagation loss is not calculated, and the multipath propagation path calculation is performed instead.

[0150] S33. For all building surfaces B j ∈D macro Perform boundary analysis to calculate the set of propagation paths that satisfy the Fresnel reflection condition.

[0151]

[0152] Where, θ i and θ r Let the incident angle and the reflection angle represent the incident angle and the reflection angle, respectively, and calculate the propagation loss along the reflection path:

[0153]

[0154] Where R(θ) i () represents the reflection coefficient;

[0155] S34. Calculating the set of diffraction paths based on the geometric features of building edges Define the propagation loss of the diffraction path:

[0156]

[0157] Where D(ν) is the diffraction attenuation factor and ν is the diffraction angle;

[0158] S35. Calculate the propagation loss at the receiving point R, considering the propagation losses of the direct path, reflected path, and diffracted path. m Total signal power at:

[0159]

[0160] Among them, P m,n For path P m,n The initial signal power during propagation;

[0161] S36. Summarize the propagation loss data from all receiving points to form a macroscopic ray tracing simulation results dataset:

[0162]

[0163] Among them, L m For receiving point R m Total propagation loss at the point, M represents the received signal power, and M represents the total number of receiving points.

[0164] In this embodiment, S4 includes the following steps:

[0165] S41. Dataset D based on macroscopic-scale ray tracing simulation results RT Define the propagation loss gradient field Used to assess the spatial rate of change of propagation loss and to calculate the mean square rate of change Γ of the gradient field. m :

[0166]

[0167] Among them, L m For receiving point R m The total propagation loss at point (x, y, z) is the spatial coordinate;

[0168] And set an adaptive threshold τ Γ To identify local regions where the change in propagation loss exceeds a threshold:

[0169]

[0170] S42. For the local region Ω local Discretization is performed using a non-uniform adaptive mesh refinement strategy, and a local-scale finite element mesh set G is defined. local :

[0171]

[0172] Among them, e′ k Let v′ represent the k-th local finite element mesh. k f′ represents the coordinates of the grid nodes. k For the medium properties within the mesh, ω′ k h′ represents the weight of the local finite element mesh. k The size of the local finite element mesh;

[0173] S43. In each local finite element mesh element e′ k Internally, based on the medium property f′ within the mesh k Construct the finite element control equations for local-scale wireless signal propagation;

[0174] S44. An adaptive coupling interpolation method is used to fuse local and macroscopic scale information to correct the electric field;

[0175] S45. Summarize the calculation results of all local meshes to form a local-scale finite element sub-model dataset:

[0176] D local ={(e′ k ,E′ k ,L′ k |k=1,2,…,N l},

[0177] Among them, E′ k L′ represents the local electric field intensity within a local finite element mesh. k For local propagation loss, Nl The number of local finite element mesh elements is the local scale finite element method.

[0178] In this embodiment, S5 includes the following steps:

[0179] S51. Dataset D based on macroscopic-scale ray tracing simulation results RT With local scale finite element sub-model dataset D local Construct a multi-scale propagation loss dataset D multi =D RT ∪D local And define the multi-scale propagation loss computation domain Ω multi =Ω macro ∪Ω local ;

[0180] Among them, Ω macro For the macroscopic propagation domain, Ω local For local scale propagation domain;

[0181] S52. Define a multi-scale fusion weighting function Ψ(x) to describe the contribution ratio of local and macro-scale data, and establish a fusion loss calculation model:

[0182]

[0183] in, and These are the loss gradients at the local and macroscopic scales, respectively; the closer Ψ(x) is to 1, the greater the contribution at the local scale, and the closer it is to 0, the greater the contribution at the macroscopic scale.

[0184] S53. Using multi-scale loss fusion to calculate the loss fusion between global propagation trend and local detail effects:

[0185] L fused (x)=(1-Ψ(x))·L macro (x)+Ψ(x·L local (x);

[0186] Among them, L fused (x) represents the propagation loss of the fused wireless channel, L macro (x) represents the propagation loss at the macroscopic scale, L local (x) represents the local scale propagation loss;

[0187] S55. Based on the multi-scale propagation loss dataset D multi And by fusing the loss calculation model, a dataset of fused wireless channel propagation loss prediction models is generated:

[0188] D fused ={(x,L fused (x),Pfused (x))∣x∈Ω multi},

[0189] Among them, P fused (x) represents the power of the received signal after fusion.

[0190] Example 1:

[0191] On August 15, 2024, at Zhongguancun East Road in Zone B, a mobile communication company was conducting 5G base station optimization tests. During the tests, the engineers found that the signal coverage in the shadow areas of some high-rise buildings was unstable, especially during the peak hours of 12:30-14:00 noon. Users reported a significant increase in network lag. In order to solve this problem, the engineers decided to use the method of this invention to accurately predict and optimize the wireless channel propagation loss in this area.

[0192] The test area is located between Zhongguancun East Road and Zhichun Road, and includes multiple high-rise office buildings, commercial plazas and underground parking garages. The building heights range from 30 meters to 150 meters. In order to construct a channel propagation environment dataset, engineers used drones equipped with LiDAR (Light Detection and Ranging) to scan the entire area. At the same time, they combined GIS (Geographic Information System) data to obtain information on terrain, building structure and material properties, and generated a detailed environmental description dataset.

[0193] During the data acquisition process, the engineers recorded the following key parameters:

[0194] Target area coordinates: 39.9816°N, 116.3276°E; Number of buildings: 47; Average building height: 75 meters; Building height: 150 meters; Test base station location: Rooftop of a shopping mall (35 meters); Base station frequency: 3.5GHz; Antenna gain: 15dBi; Test equipment: Rohde & Schwarz FSW signal analyzer; Number of measurement points: 500 (including streets, high-rise buildings, and underground parking garages).

[0195] In the actual deployment, the engineers first used the traditional ray tracing method to calculate the channel loss in the test area and obtained the preliminary prediction results at around 15:00 on August 15. However, they found that the prediction error was large in several building shadow areas (east side of Building A and north side of Building B), and the measured signal strength deviated from the calculation results by 5.8dB, which made it impossible to adjust the optimization scheme accurately.

[0196] To address this issue, the engineers decided to apply the multi-scale finite element modeling method of this invention. First, based on preliminary ray tracing results, they identified local regions where the propagation loss gradient changed drastically.

[0197] East side of Building A: Area with a sharp increase in signal attenuation (gradient change: 9.2dB / m);

[0198] North side of Building B: Diffraction effect at the building edge (gradient change: 8.5dB / m);

[0199] Underground parking garage entrance: Significant multipath effect (gradient change: 7.3 dB / m);

[0200] Subsequently, the system introduced a more refined finite element mesh model in these key areas and adopted an adaptive mesh reconstruction algorithm to optimize the computational area, making the loss calculation at the local scale more accurate. After optimization, the engineers found that the prediction error in the building edge area decreased from 5.8dB to 3.2dB, and the overall channel prediction error was reduced by 44.8%.

[0201] To verify the advantages of the method of this invention, engineers conducted comparative tests in the same environment using the traditional ray tracing method, the single-scale finite element method, and the multi-scale method of this invention, and recorded the channel loss prediction errors of different methods at multiple key measurement points.

[0202]

[0203] The test data above shows that the prediction error of the method of the present invention is significantly lower than that of the traditional method at all measurement points. In complex environments such as underground garage entrances and building shadow areas, the prediction error of the method of the present invention is reduced by more than 50%, demonstrating higher accuracy.

[0204] In addition to improving prediction accuracy, engineers also compared the computational efficiency of the three methods, and statistically analyzed the computation time of each method:

[0205] method Calculate time (seconds) Traditional ray tracing methods 4125 Single-scale finite element method 2789 This invention's multi-scale method 1647

[0206] The results show that the method of the present invention reduces the computation time by 60.1% compared with the traditional ray tracing method and by 40.9% compared with the single-scale finite element method, which greatly improves the computational efficiency and makes large-scale channel propagation loss prediction more feasible in practical engineering applications.

[0207] This embodiment conducts a wireless channel propagation loss prediction experiment in a real urban environment, verifying the advantages of the method of the present invention in terms of computational accuracy and efficiency. Through multi-scale finite element modeling, adaptive mesh optimization, and ray tracing-assisted calculation, the method of the present invention can accurately predict wireless signal propagation loss in complex environments, effectively reduce prediction errors, and significantly improve computational efficiency. Finally, the method of the present invention is used to optimize the deployment scheme of 5G base stations, making the signal coverage optimization scheme in the shadow area of ​​high-rise buildings more accurate, and providing efficient and reliable technical support for 5G network optimization.

[0208] This invention proposes a multi-scale modeling strategy that combines the finite element method and ray tracing algorithm to construct the overall propagation environment at the macro scale while performing refined modeling of regions with drastic changes in propagation loss at the local scale. By dynamically and adaptively dividing the region, the importance of the local region is evaluated using the propagation loss gradient, and a more refined mesh division and solution method is adopted in the local region. This allows the overall prediction model to improve the accuracy of loss calculation while ensuring computational efficiency.

[0209] This invention combines ray tracing with the finite element method. The ray tracing algorithm is used to quickly calculate the channel propagation path at the macro scale to identify regions of abrupt changes in propagation loss. Then, the finite element method is introduced in key regions for refined calculation. By using ray tracing to filter out regions with large changes in propagation loss and employing finite element modeling in local regions, unnecessary computational overhead is reduced, thus lowering the overall computational complexity.

[0210] This invention proposes an adaptive multi-objective grid reconstruction algorithm. Based on the changes in the spatial gradient and propagation loss of the electromagnetic environment, the computational grid is dynamically adjusted so that the critical propagation region adopts a finer grid division, while the region with stable propagation changes maintains a larger grid scale, reducing unnecessary computation. The adaptive grid technology can improve the computational accuracy of the critical region while ensuring the optimization of computational resources. In addition, the multi-scale information fusion method uses the global propagation trend to correct local propagation calculation errors and achieves a dynamic balance between local and macro-scale information through the scale coupling function.

[0211] The above description is only a preferred embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any equivalent substitutions or modifications made by those skilled in the art within the scope of the technology disclosed in the present invention, based on the technical solution and inventive concept of the present invention, should be covered within the scope of protection of the present invention.

Claims

1. A multi-scale wireless channel propagation loss prediction method based on the finite element method, characterized in that, Includes the following steps: S1. Acquire target wireless channel propagation environment data, and preprocess the environment data to form an environment description dataset; S2. Based on the finite element method, the environmental description dataset is discretized at the macro scale to form a macro scale finite element model; S3. Use the ray tracing algorithm to simulate the wireless signal propagation path of the macroscopic finite element model and generate macroscopic ray tracing simulation results; S4. Based on the macroscopic ray tracing simulation results, identify local regions with significant changes in propagation loss, and refine the model of these local regions using the finite element method to form a local-scale finite element sub-model. S5. Multi-scale information fusion is performed on the macro-scale ray tracing simulation results and the wireless channel propagation loss data obtained from the local-scale finite element sub-model to form a fused wireless channel propagation loss prediction model. The fusion process achieves mutual correction between global propagation trends and local detail effects. S6. Based on the fused wireless channel propagation loss prediction model, calculate the wireless signal propagation loss in the target environment, generate wireless channel propagation loss prediction results, and post-process the wireless channel propagation loss prediction results to output a wireless channel propagation loss prediction map containing signal coverage, attenuation distribution and loss indicators of key areas.

2. The multi-scale wireless channel propagation loss prediction method based on the finite element method according to claim 1, characterized in that, S1 includes the following steps: S11. Collect target wireless channel propagation environment data, including terrain information, building geometry, building material properties, and electromagnetic environment parameters, and construct an environment description dataset. : in, Indicates the first A target wireless channel propagation environment data unit, The total number of target wireless channel propagation environment data units collected. This includes terrain information, such as surface elevation, surface roughness, and slope parameters. This refers to the building's geometric structure, including its height, floor area, shape, and relative coordinates. These are the material properties of a building, including dielectric constant, conductivity, and absorption coefficient. These are electromagnetic environment parameters, including background noise intensity, location of electromagnetic interference sources, and spectral information. S12. The environmental description dataset... Perform data format conversion, unify data coordinate system, set global coordinate system as standard coordinate reference, convert target wireless channel propagation environment data into standard coordinate representation, and obtain the transformed standardized environment description dataset. S13. Perform a data integrity check on the standardized environmental description dataset, remove incomplete data units, and use interpolation to fill in missing values ​​to form a complete environmental description dataset; S14. Invalid data is removed from the complete environment description dataset. Invalid data units are filtered according to the set target wireless channel propagation environment data validity standard to obtain the final filtered environment description dataset. : in, The number of data units in the target wireless channel propagation environment after removing invalid data. , , and This includes filtered terrain information, building geometry, building material properties, and electromagnetic environment parameters.

3. The multi-scale wireless channel propagation loss prediction method based on the finite element method according to claim 2, characterized in that, S2 includes the following steps: S21. Based on the final environment description dataset Dataset unit describing the propagation environment of each target wireless channel Define the corresponding local computation region and its boundaries The importance of local areas is evaluated by using the spatial gradient of electromagnetic environment parameters in the environmental description dataset, and a weighted index function is constructed. : in, Indicates at point Standardized electromagnetic environment parameters at the location, Its spatial gradient, To adjust the positive definite coefficients of sensitivity to local changes, a weighted index function is used. Set threshold Then, define the overall computational region. The boundary of the computational domain is determined by the overall computational domain. S22. In the overall calculation area The internal discretization uses an adaptive multi-objective mesh reconstruction algorithm to divide the overall computational domain into a set of finite element mesh elements. : in, Indicates the first One grid cell, Let the node coordinate vector be... Indicates local medium properties, The total number of grid cells. The grid weights are based on the estimated initial propagation loss. and its spatial gradient Sure: in, The normalization coefficient is... To prevent small constants from being divided by zero, The size of the grid cell is determined by the weight. Adaptive adjustment: in, For the maximum allowed cell size, For adjustment parameters; S23. In each grid cell Internally, based on local medium properties We construct the finite element control equations for wireless channel propagation and introduce multi-scale correction terms to achieve coupling between global and local information: in, The electric field intensity vector; and The first Local relative permeability and relative permittivity within each unit; The signal angular frequency, For the excitation source current density, For correction factors: ； in, It is a positive definite adjustment coefficient; S24. Finite element basis functions are used to locally interpolate the electric field, while scale coupling functions are introduced to fuse multi-scale information: in, For traditional finite element basis functions, For scale coupling function: in, The coupling strength coefficient is... For the first The number of basis functions in each grid cell The coefficients to be solved; S25. A feedback iteration mechanism is used to dynamically correct the deviation between the preliminary calculation results and the target propagation loss, thereby achieving adaptive coupling between the global and local models: Among them, the loss function Defined as: in, For the first The grid cell in the first... The propagation loss calculated in the next iteration To determine the target propagation loss based on the environmental description dataset and prior information, This is the iteration step size; The iterative process simultaneously adjusts the grid weights in each update cycle. By combining the coupling function parameters, a macroscopic-scale finite element model is finally obtained, and a macroscopic-scale finite element model computational dataset is formed. .

4. The multi-scale wireless channel propagation loss prediction method based on the finite element method according to claim 1, characterized in that, S3 includes the following steps: S31. Calculating Datasets Based on Macro-Scale Finite Element Model Constructing the ray tracing path search space Based on the location of the target wireless signal source and receiving point Define the global ray path set: in, Indicates from the signal source to the receiving point The Possible transmission paths, For the first on the path A point of reflection or diffraction, The number of all possible paths; S32. For each receiving point Calculate its relationship with the signal source direct path The path length is If the path is not blocked by buildings in the environment, then calculate the propagation loss of the direct path. If the path is blocked, the direct path propagation loss is not calculated, and the multipath propagation path calculation is performed instead. S33. For all building surfaces Perform boundary analysis to calculate the set of propagation paths that satisfy the Fresnel reflection condition. : in, and Let the incident angle and the reflection angle represent the incident angle and the reflection angle, respectively, and calculate the propagation loss along the reflection path: in, The reflection coefficient; S34. Calculating the set of diffraction paths based on the geometric features of building edges Define the propagation loss of the diffraction path: in, The diffraction attenuation factor, The angle of diffraction; S35. Calculate the propagation loss at the receiving point, taking into account the propagation losses of the direct path, reflected path, and diffracted path. Total signal power at: in, For along the path The initial signal power during propagation; S36. Summarize the propagation loss data from all receiving points to form a macroscopic ray tracing simulation results dataset: in, For receiving point Total propagation loss at the point, To receive signal power, This represents the total number of receiving points.

5. The multi-scale wireless channel propagation loss prediction method based on the finite element method according to claim 1, characterized in that, S4 includes the following steps: S41. Dataset of macroscopic ray tracing simulation results Define the propagation loss gradient field Used to assess the spatial rate of change of propagation loss and calculate the mean square rate of change of the gradient field. : in, For receiving point Total propagation loss at the point, Spatial coordinates; And set an adaptive threshold To identify local regions where the change in propagation loss exceeds a threshold: S42. For local areas Discretization is performed using a non-uniform adaptive mesh refinement strategy, defining a local-scale finite element mesh set. : in, Indicates the first A local finite element mesh element, For grid node coordinates, For the medium properties within the mesh, For local finite element mesh element weights, The size of the local finite element mesh; S43. In each local finite element mesh element Internally, based on the medium properties within the mesh Construct the finite element control equations for local-scale wireless signal propagation; S44. An adaptive coupling interpolation method is used to fuse local and macroscopic scale information to correct the electric field; S45. Summarize the calculation results of all local meshes to form a local-scale finite element sub-model dataset: in, The local electric field intensity within a local finite element mesh element. For local propagation loss, The number of local finite element mesh elements is the local scale finite element method.

6. The multi-scale wireless channel propagation loss prediction method based on the finite element method according to claim 1, characterized in that, S5 includes the following steps: S51. Dataset based on macroscopic-scale ray tracing simulation results With local scale finite element sub-model dataset Construct a multi-scale propagation loss dataset And define the multi-scale propagation loss computation domain. in, For macroscopic propagation domain, For local scale propagation domain; S52. Define the multi-scale fusion weighting function A fusion loss calculation model is established to describe the contribution ratio of local-scale and macro-scale data: in, and These are the loss gradients at the local and macroscopic scales, respectively. The closer it is to 1, the greater the contribution at the local scale; the closer it is to 0, the greater the contribution at the macro scale. S53. Using multi-scale loss fusion to calculate the loss fusion between global propagation trend and local detail effects: in, For the propagation loss of the fused wireless channel, For macroscopic propagation loss, For local scale propagation loss; S55. Based on a multi-scale propagation loss dataset And by fusing the loss calculation model, a dataset of fused wireless channel propagation loss prediction models is generated: in, This represents the power of the received signal after fusion.

Images