A Monte Carlo Integration-Based Indoor Wireless Signal Coverage Prediction Method

Abstract

This invention discloses an indoor wireless signal coverage prediction method based on Monte Carlo integration. A transmitting antenna and receiver are deployed within an indoor scene. The geometric model of the indoor scene and the attribute parameters of the transmitting antenna are acquired and input into a computer to build a simulation model. Rays are generated to describe electromagnetic wave propagation for simulation. Through multiple iterative processing, an indoor wireless signal coverage map is predicted. This process is repeated, applying random rotations for multiple simulations, and the final indoor wireless signal coverage prediction result is obtained by combining the results. This invention avoids performing path search and field strength calculation for each receiver point individually, significantly improving computational efficiency. It also avoids the exponential growth problem of ray number caused by ray splitting in traditional ray tracing, thus significantly reducing memory usage and improving computational efficiency. This enhances the stability and accuracy of signal coverage prediction, enabling rapid acquisition of signal coverage information in complex indoor scenes while maintaining prediction accuracy.

Description

Technical Field

[0001] This invention relates to the field of wireless communication and radio wave propagation simulation technology, and specifically to a signal coverage prediction method based on Monte Carlo integration for indoor scenarios. Background Technology

[0002] With the dense deployment of systems such as wireless LANs and the Internet of Things (IoT) in indoor environments such as office buildings, shopping malls, campuses, and factories, the requirements for signal coverage quality in indoor wireless communication services are increasing. In order to rationally deploy antennas and evaluate the overall signal coverage during the network planning phase, it is necessary to accurately predict the received power distribution in the target area.

[0003] In existing technologies, ray tracing methods based on geometric optics and uniform diffraction theory are widely used in indoor channel modeling and coverage prediction. Traditional ray tracing methods typically employ a ray splitting strategy: when a ray intersects an object's surface, it simultaneously generates reflected and refracted rays, which then continue tracing along their respective directions. To achieve high simulation accuracy, a large initial number of rays and a large maximum path depth are required. However, in complex indoor environments, the number of rays increases exponentially with path depth, leading to a rapid increase in memory usage and a significant extension of computation time, making it difficult to meet the demands of rapid coverage prediction in large-scale scenarios. On the other hand, traditional methods often perform ray tracing and field strength calculations separately for each receiver point. When the number of receiver points is large or a high-resolution coverage map is required, this approach incurs enormous computational overhead, hindering its widespread adoption in engineering practice.

[0004] Therefore, existing technologies lack an indoor wireless signal coverage prediction method that can significantly reduce memory usage, improve computational efficiency, and quickly provide the overall signal coverage distribution of the target area while ensuring prediction accuracy. Summary of the Invention

[0005] To address the problems in the background art, such as the need for path searching and field strength calculation at each receiving point when using traditional ray tracing methods for signal coverage prediction in large-scale complex indoor scenarios, resulting in high computational load, high memory consumption, and long running time, as well as the exponential increase in the number of rays with path depth when using ray splitting strategies, making it difficult to achieve high-resolution indoor wireless signal coverage prediction with limited computing resources, this invention provides an indoor wireless signal coverage prediction method based on Monte Carlo integration.

[0006] This invention proposes an indoor wireless signal coverage prediction method based on Monte Carlo integration. This method does not require searching the propagation path between the transmitting antenna and each receiving point one by one, thus achieving high computational efficiency. It utilizes Fibonacci lattice points for spherical uniform sampling and employs a Fresnel coefficient-based ray sampling strategy when rays interact with the scene, obtaining signal coverage prediction results for complex indoor scenes with relatively low memory overhead and computational complexity.

[0007] The specific technical solution adopted in this invention is as follows:

[0008] Step 1: Use a 3D model acquisition device to acquire a geometric model of the indoor scene with transmitting antennas and receivers.

[0009] Preferably, the 3D model acquisition device is, for example, a lidar.

[0010] The geometric model of the indoor scene was obtained by a laser scanner, and the geometric model is a triangular mesh model.

[0011] The indoor scene described is a home environment, containing various objects, walls for partitioning, and basic indoor facilities such as windows and doors. A transmitting antenna and a receiver are also installed within this environment. Specifically, the transmitting antenna is a router or wireless transmitter with Wi-Fi functionality, and the receiver is a device such as a mobile phone or tablet.

[0012] Step 2: The 3D model acquisition device combines the geometric model of the indoor scene with its own attribute parameters sent by the transmitting antenna to the computer. A simulation model with a channel gain simulation plane is established in the computer, and a finite number of rays are generated in the simulation space to describe the propagation of electromagnetic waves. The simulation is then performed based on the finite interactive propagation tracking iteration of reflection and transmission probability sampling to predict and obtain the indoor wireless signal coverage map.

[0013] Step 3: Repeat Step 2 above, apply random rotation, and perform multiple simulations. Combine the indoor wireless signal coverage maps from multiple simulations to obtain the final indoor wireless signal coverage prediction result.

[0014] In step two, the simulation process is carried out in the computer according to the following procedure:

[0015] S1. In the geometric model of the indoor scene, set a simulation plane at the height of the receiving antenna, divide the simulation plane into several cells of the same size, and set the channel gain for each cell as a signal position to draw a signal coverage map.

[0016] S2. Using the position of the transmitting antenna as the starting point, generate several initial rays simulating electromagnetic wave emission using a uniform sampling method;

[0017] S3. For each ray, perform intersection tracking in the geometric model of the indoor scene combined with the simulation plane, use Monte Carlo integration to calculate the path of all rays, and then update the channel gain of each cell in the signal coverage diagram.

[0018] S4. For each ray, perform interactive ray tracking in the geometric model of the indoor scene to obtain a new ray after the interaction, and return to step S3 according to the new ray after the interaction to process and update the signal coverage map.

[0019] S5. Repeat steps S3 to S4 until the ray reaches the preset maximum path depth or the ray no longer intersects with the geometric model of the indoor scene. End ray tracing and use the channel gain of each cell on the simulation plane obtained in the last iteration as the predicted indoor wireless signal coverage map.

[0020] Step S3 specifically includes:

[0021] S31. Determine whether each ray intersects with the simulation plane using ray tracing:

[0022] When a ray intersects the simulation plane, the cell containing the intersection point and its index are found as the intersection point information.

[0023] S32. The intersection information of all rays is superimposed onto the channel gain of the cell containing the intersection point through a unified integration operation of Monte Carlo integration, thereby updating the channel gain of the cell containing the intersection point.

[0024] This allows for the updating of the signal coverage map by iterating through all intersecting rays and updating the channel gain of the cells containing all intersection points.

[0025] Step S32 specifically includes: establishing the path of each ray relative to the starting point of the transmitting antenna, and establishing a sphere with the starting point of the transmitting antenna as the center. By using the Monte Carlo integration method on the sphere to perform unified integration processing on the paths of all rays combined with the intersection information, the received electric field of each ray at the intersection point of the simulation plane is obtained. Then, the channel gain of the cell where the intersection point is located is updated according to the magnitude of the received electric field intensity.

[0026] Step S4 specifically includes:

[0027] When the ray intersects the first object in the geometric model of the indoor scene along its own path, an interaction type is randomly selected from reflection and transmission based on the probability model. The new ray propagation direction is calculated based on the selected interaction type to obtain the new ray after the interaction. Then, the process returns to step S3 to continue ray tracing.

[0028] In step S4, the random selection probability of the reflection and transmission interaction type is determined based on the Fresnel reflection coefficient and transmission coefficient of the intersecting object surfaces, thereby constructing a probability model of the path interaction type sequence.

[0029] The specific steps of S4 are as follows:

[0030] S41. First, based on the Fresnel reflection coefficient and transmission coefficient of the surface material at the intersection point of the ray and the object, construct the following probability model to calculate and determine the reflection probability q(R) and transmission probability q(T):

[0031] q(R) = (|R ⊥ |²+|R ∥ |²) / (|R ⊥ |²+| R ∥ |²+| T ⊥ |²+| T ∥ |²)

[0032] q(T) = (|T ⊥ |²+|T ∥ |²) / (|R ⊥ |²+| R ∥ |²+| T ⊥ |²+| T ∥ |²)

[0033] Among them, R ⊥ R ∥ T represents the vertically polarized reflection coefficient and the horizontally polarized reflection coefficient, respectively. ⊥ T ∥ q(R) and q(T) represent the vertical polarization transmission coefficient and the horizontal polarization transmission coefficient, respectively; the values ​​of reflection probability q(R) and transmission probability q(T) are between 0 and 1.

[0034] S42. Then the computer generates a random number, which takes the value between 0 and 1. If the random number is less than or equal to q(R), then reflection is selected as the interaction type of the current ray at the object. If the random number is greater than q(R), then transmission is selected as the interaction type of the current ray at the object.

[0035] S43. After determining the interaction type, use the Fresnel reflection coefficient or transmission coefficient corresponding to the current interaction type, combined with the electric field of the current ray, to calculate the electric field of the next ray using the geometric optics method.

[0036] In this invention, the electric field of the next ray is calculated using geometric optics based on the electric field of the previous ray for the propagation of adjacent rays.

[0037] The maximum path depth refers to the number of interactions in step S4 experienced by each ray reaching a preset fixed number. The geometric models of the indoor scene no longer intersect typically means that rays pass through windows or doors in the geometric model of the indoor scene in a transmission interaction manner, allowing them to move away from the geometric model of the indoor scene without obstruction.

[0038] In step S2, a sphere is established with the position of the transmitting antenna as the center. Each sampling point is generated using a spherical sampling method based on Fibonacci lattice points. The direction of the line connecting the position of the transmitting antenna and each sampling point is set as the direction of each initial ray, forming an initial ray set.

[0039] Step three specifically includes applying random rotation to all generated initial ray directions each time step two is repeated, repeating the indoor wireless signal coverage map construction process, and averaging the indoor wireless signal coverage maps obtained from multiple rounds of simulation to obtain an average signal coverage map, which is used as the final indoor wireless signal coverage prediction result.

[0040] In step three, the initial ray direction of each round of multi-round simulation is transformed by applying a random rotation matrix to the sphere while keeping the number of samples constant, so as to obtain a set of initial ray directions different from the previous round, and the indoor wireless signal coverage map obtained from each round of simulation is averaged.

[0041] The method of this invention constructs a signal coverage map on the plane where the receiving point / receiver is located, avoiding the need for separate path search and field strength calculation for each receiving point. Since the method does not require obtaining the precise path from the transmitting antenna to each receiving point, it can significantly improve computational efficiency.

[0042] Furthermore, this invention uses a spherical sampling method based on Fibonacci lattice points to generate the initial ray direction, and avoids the problem of exponential growth in the number of rays caused by ray splitting in traditional ray tracing through a random sampling strategy, thereby significantly reducing memory usage and improving computational efficiency. At the same time, by randomly rotating the initial ray direction and averaging multiple simulation results, the stability and accuracy of signal coverage prediction are improved.

[0043] This invention can quickly obtain signal coverage information in complex indoor scenarios while ensuring prediction accuracy, and is suitable for the planning and optimization of indoor wireless networks.

[0044] The beneficial effects of this invention are:

[0045] This invention directly estimates the channel gain of cells using Monte Carlo integration on the simulation plane. It only requires one global ray tracing of the initial ray set to obtain the signal coverage map of the entire region, avoiding the need to perform path search and field strength calculation for each receiving point separately, thus significantly reducing the computational load.

[0046] This invention performs probability sampling of reflection and transmission based on Fresnel coefficients at the intersection of the ray and the scene, retaining only one path to continue tracking. This avoids the exponential growth of the number of rays with the path depth, significantly reducing memory usage and computational complexity, and is more suitable for parallel acceleration on GPU platforms.

[0047] This invention uses Fibonacci lattice points to achieve uniform sampling of the sphere, and applies random rotation to the initial ray set to average the simulation results from multiple rounds, effectively improving the accuracy of signal coverage prediction. This allows for the acquisition of indoor wireless signal coverage prediction results that are highly consistent with measured results under limited computing resources, and has the advantages of high computational efficiency, wide applicability, and ease of engineering promotion. Attached Figure Description

[0048] Figure 1 This is a flowchart of the method of the present invention;

[0049] Figure 2 These are schematic diagrams of spherical sampling, where (a) is a schematic diagram with 1000 sampling points and (b) is a schematic diagram with 10000 sampling points;

[0050] Figure 3 This is a schematic diagram of the path from the transmitting antenna to the simulation plane;

[0051] Figure 4 It is a 3D indoor scene image;

[0052] Figure 5 It is a simulated signal coverage map of an indoor scene. Detailed Implementation

[0053] The present invention will be further described in detail below with reference to the accompanying drawings and specific embodiments.

[0054] like Figure 1 As shown, the method of the present invention includes, but is not limited to, the following steps:

[0055] Step 1: Deploy the transmitting antenna and receiver within the indoor scene. Acquire the geometric model of the indoor scene using 3D model acquisition equipment. Obtain the attribute parameters of the transmitting antenna. The attribute parameters of the transmitting antenna include its position in the indoor scene, operating frequency, polarization, transmission power, antenna pattern, and other parameters.

[0056] Step 2: The 3D model acquisition device sends the geometric model of the indoor scene to the computer, and the computer combines the attribute parameters of the transmitting antenna to build a simulation model with a channel gain simulation plane. In the simulation space, a finite number of rays to describe the propagation of electromagnetic waves are generated for simulation processing based on finite interactive propagation tracking iteration based on reflection and transmission probability sampling, and the indoor wireless signal coverage map is predicted.

[0057] S1. In the computer simulation model, a simulation plane is set at the height of the receiving antenna in the geometric model of the indoor scene. The simulation plane covers the area to be evaluated and is divided into several cells of the same size. Each cell is used as a signal location to set the channel gain and then draw a signal coverage map. Specifically, the simulation plane is divided into several closely spaced cells according to the preset number and size of cells. Each cell corresponds to a received power value.

[0058] The simulation plane is set on a horizontal plane at the height of the receiving antenna. The center position and plane size of the simulation plane are selected according to the architectural layout of the indoor scene to completely cover the area to be evaluated. The cell size of the simulation plane is set according to the required resolution.

[0059] The height of the simulation plane is set according to the actual installation height of the receiving antenna. The center and size of the simulation plane are selected according to the indoor scene to cover typical areas such as corridors and rooms. The size of the cells is set according to the required spatial resolution to balance simulation accuracy and computational complexity.

[0060] S2. Starting from the position of the transmitting antenna in the indoor scene, generate several initial rays simulating electromagnetic wave emission using a uniform sampling method, and use the direction of each initial ray as the emission direction of the signal propagation path; thus forming an initial ray set.

[0061] In step S2, a sphere is established with the position of the transmitting antenna in the indoor scene as the center. The direction of the initial ray is generated by generating each sampling point using a spherical sampling method based on Fibonacci lattice points to ensure the uniform distribution of the spherical sampling points. The direction of the line connecting the position of the transmitting antenna and each sampling point is set as the direction of each initial ray, forming an initial ray set.

[0062] The Fibonacci lattice method calculates the coordinates of sampling points based on a preset number of rays. The direction of the line connecting the transmitting antenna position and the sampling points is normalized and used as the initial direction of the ray. This sampling method can obtain a uniformly distributed ray direction, improving the calculation accuracy of Monte Carlo integrals.

[0063] Step 3: Repeat Step 2 above, apply random rotation, and perform multiple simulations. Combine the indoor wireless signal coverage maps from multiple simulations to obtain the final indoor wireless signal coverage prediction result.

[0064] S3. For each ray in the geometric model of the indoor scene, perform intersection tracking judgment in combination with the simulation plane, use Monte Carlo integration to calculate the path of all rays, and then update the channel gain of each cell in the signal coverage diagram.

[0065] S31. First, determine whether each ray intersects with the simulation plane using ray tracing:

[0066] When a ray intersects the simulation plane, find the cell containing the intersection point of the ray and the simulation plane and its index;

[0067] S32. The intersection information of all rays is superimposed onto the channel gain of the cell containing the intersection point through a unified integration operation of Monte Carlo integration, thereby updating the channel gain of the cell containing the intersection point.

[0068] Step S32 specifically includes: establishing the path of each ray relative to the starting point of the transmitting antenna, and establishing a sphere with the starting point of the transmitting antenna as the center. The received electric field at the intersection point of each ray is obtained by uniformly integrating the paths of all rays with the intersection information using the Monte Carlo integration method on the sphere. The received electric field is then updated by superimposing the magnitude of the received electric field onto the channel gain of the cell containing the intersection point, specifically calculated according to the following formula:

[0069] g i,j = (λ² / (4πN s |C|))×Σ n (E) n ²×Ⅱ(t n ∈C i,j ) ×(r n ² / (Pr(χ n cosθ n )))

[0070] Among them, g i,j It is cell C in row i and column j. i,j The corresponding channel gain, where λ is the wavelength of the wireless signal, and N s It is the number of emitted rays, |C| is the area of ​​the cell, and E n t is the electric field strength of the received electric field at the intersection of the nth path and the simulated plane. n It is the intersection of the nth path and the simulation plane, r n It is the length of the nth path, Pr(χ) n ) is the interaction type sequence χ corresponding to the nth path.n The probability, θ n It is the angle between the nth path and the normal to the simulation plane; Ⅱ() is the indicator function, when t n ∈C i,j Ⅱ(t) n ∈C i,j ) equals 1, otherwise II(t) n ∈C i,j ) equals 0.

[0071] If the ray does not intersect with the simulation plane, the signal coverage map will not be updated.

[0072] Thus, by iterating through all intersecting rays and updating the channel gain of the cells containing all intersection points, the signal coverage map can be updated.

[0073] For the initial ray, the electric field of the initial ray is calculated based on the transmitting power of the transmitting antenna and the antenna pattern of the transmitting antenna.

[0074] S4. For each ray, perform interactive ray tracing in the geometric model of the indoor scene to obtain a new ray after the interaction. Return to step S3 to process and update the signal coverage map according to all the new rays after the interaction. When the ray intersects the first object in the geometric model of the indoor scene along its own path, randomly select one of the two interaction types, reflection and transmission, according to the set probability model. Calculate the new ray propagation direction and corresponding electric field according to the selected interaction type to obtain a new ray after the interaction, which is used to continue ray tracing.

[0075] Specifically:

[0076] S41. First, based on the Fresnel reflection coefficient and transmission coefficient of the surface material at the intersection point of the ray and the object, construct the following probability model to calculate and determine the reflection probability q(R) and transmission probability q(T):

[0077] q(R) = (|R ⊥ |²+|R ∥ |²) / (|R ⊥ |²+| R ∥ |²+| T ⊥ |²+| T ∥ |²)

[0078] q(T) = (|T ⊥ |²+|T ∥ |²) / (|R ⊥ |²+| R ∥ |²+| T ⊥ |²+| T ∥ |²)

[0079] Among them, R ⊥ R ∥ T represents the vertically polarized reflection coefficient and the horizontally polarized reflection coefficient, respectively. ⊥ T ∥ These represent the vertical polarization transmission coefficient and the horizontal polarization transmission coefficient, respectively.

[0080] S42. Then the computer generates a random number, which takes the value between 0 and 1. If the random number is less than or equal to q(R), then reflection is selected as the interaction type of the current ray at the object. If the random number is greater than q(R), then transmission is selected as the interaction type of the current ray at the object.

[0081] S43. After determining the interaction type, use the Fresnel reflection coefficient or transmission coefficient corresponding to the current interaction type, combined with the electric field of the current ray, to calculate the electric field of the next ray using the geometric optics method.

[0082] In steps S3 and S4, when calculating the channel gain, the weight of each ray path is related to the path length, the angle between the path and the normal to the simulation plane when the path finally reaches the simulation plane, and the probability of the path interaction type sequence, so as to make an unbiased estimate of the contribution of different paths.

[0083] In the above processing, the probability of random selection of path interaction type is determined by the Fresnel reflection coefficient and transmission coefficient of the object surface material. By randomly sampling between reflection and transmission, instead of the ray splitting strategy of generating reflected and transmitted rays at the intersection, the number of rays does not increase exponentially with the path depth.

[0084] S5. Repeat steps S3 to S4 until the ray reaches the preset maximum path depth or the ray no longer intersects with the geometric model of the indoor scene. End ray tracing and use the channel gain of each cell on the simulation plane obtained in the last iteration as the predicted indoor wireless signal coverage map.

[0085] Finally, the channel gain of each cell in the signal coverage map is multiplied by the transmit power of the transmitting antenna to obtain the final signal coverage prediction map result. In other words, the channel gain prediction result is multiplied by the transmit power to obtain the received power value of each cell, thus forming a visualized indoor wireless signal coverage map for subsequent indoor wireless network planning and optimization.

[0086] Each time step two above is repeated, a random rotation is applied to all generated initial ray directions, so that all generated initial ray directions are rotated by the random rotation angle. The above indoor wireless signal coverage map construction process is repeated, and the indoor wireless signal coverage maps obtained from multiple rounds of simulation are averaged to obtain an average signal coverage map, which is used as the final indoor wireless signal coverage prediction result that is stable and converged.

[0087] In multi-round simulations, the initial ray direction of each round is transformed by applying a random rotation matrix to the sphere while keeping the number of samples constant, so as to obtain a set of initial ray directions that are different from the previous round. Averaging the indoor wireless signal coverage map obtained from each round of simulation can significantly improve the simulation accuracy.

[0088] In practice, the method is implemented on a graphics processing unit (GPU). The generation of the initial ray, the calculation of the intersection between the ray and the indoor scene, the calculation of the electric field, and the construction of the signal coverage map are accelerated in parallel. The number of initial rays and the maximum path depth are set according to the GPU memory capacity and the target prediction accuracy. Under the premise that the memory resources allow, the running time is further shortened by increasing the number of rays in a single simulation round and appropriately reducing the number of iteration rounds.

[0089] After obtaining the signal coverage map, the method can determine the cell index of the receiving point based on its location, extract the predicted received power value of any cell location from the coverage map, and realize the prediction of indoor wireless signal strength.

[0090] The overall flowchart of a specific embodiment of the present invention is as follows: Figure 1 As shown, it includes the following steps:

[0091] 1) This invention first constructs a three-dimensional geometric model containing a typical furniture interior layout, and sets a simulation plane parallel to the ground at the height of the receiving antenna. The simulation plane covers the area to be evaluated and is divided into several cells of uniform size according to the target resolution, with each cell corresponding to a channel gain value. The surface of each object in the interior three-dimensional geometric model is divided into several triangular facets for subsequent ray intersection calculations.

[0092] Initially, the initial channel gain at each cell is set to 0, thus generating an initial signal coverage map. The channel gain at each cell ranges from 0 to 1.

[0093] 2) This invention uses the Fibonacci lattice point spherical sampling method to generate the initial ray set, such as... Figure 2 As shown. Specifically, with the transmitting antenna position as the center of the sphere, the polar angle and azimuth angle of each ray are calculated according to the preset number of rays, and mapped to a sampling point on the sphere. Then, the direction of the line connecting the transmitting antenna position and the sampling point is normalized and used as the initial propagation direction of the ray.

[0094] In practice, the initial number of rays is usually set to 10. 6 ~10 9 Approximately 1000 rays. When 1000 rays map to form 1000 sampling points, as... Figure 2As shown in (a), when 10,000 rays are mapped to form 10,000 sampling points, as Figure 2 As shown in (b).

[0095] Compared to traditional partitioning methods, this method generates sampling points that are more evenly distributed on the sphere, effectively improving the calculation accuracy of Monte Carlo integrals. For each ray, the corresponding initial electric field is calculated based on the transmitting antenna pattern.

[0096] 3) This invention performs ray tracing based on a ray sampling strategy for each initial ray, such as... Figure 3 As shown, during signal propagation:

[0097] First, check if the ray intersects the simulation plane: if there is a direct path from the transmitting antenna to the simulation plane, calculate the electric field at the intersection point of the direct path and update the channel gain of the corresponding cell.

[0098] Then, the intersection operation between the ray and the indoor triangular facet is performed to obtain the nearest intersection point and its normal vector.

[0099] The incident angle is calculated based on the incident direction and normal vector of the ray. Then, the reflection probability q(R) and transmission probability q(T) are obtained by constructing the Fresnel reflection coefficient and transmission coefficient of the surface material at the intersection point according to the above formula.

[0100] This invention employs a ray sampling strategy to select either reflection or transmission as the actual path interaction type: when reflection is selected, the reflection direction is calculated based on the law of reflection; when transmission is selected, the refraction direction is calculated based on Snell's law.

[0101] Simultaneously, the reflection and transmission fields are calculated using geometric optics. Compared to traditional ray splitting strategies, this invention retains only one ray direction for continued tracking, effectively avoiding the exponential growth of the number of rays with depth and significantly reducing memory usage. The ray continues to propagate forward in the scene, repeating the intersection and ray sampling process described above.

[0102] 4) When the ray intersects the simulation plane during propagation, the channel gain of the cell where the intersection point is located is updated based on the path length, the angle between the path and the normal of the simulation plane when the path finally reaches the simulation plane, and the probability of the path interaction type sequence.

[0103] Ray propagation terminates when any of the following conditions are met: the ray no longer intersects with the scene, the ray path depth reaches a preset maximum value, or the ray path energy decays to below a threshold.

[0104] After all ray tracing is completed, the present invention calculates the received power at the corresponding position based on the channel gain and transmit power of the transmitting antenna for each cell.

[0105] This invention implements the ray tracing and signal coverage map construction process on a graphics processing unit (GPU). Leveraging the massively parallel computing capabilities of GPUs, this method can efficiently handle large-scale ray tracing tasks, thereby rapidly generating signal coverage maps.

[0106] 5) To further improve prediction accuracy, this invention applies a random rotation matrix to the initial ray direction set to generate different ray direction sets, repeats the above ray tracing process, and averages the results of multiple simulations at the cell level to make the prediction results of the signal coverage map more accurate.

[0107] Specific indoor scenes such as Figure 4 As shown, this scenario is a typical office area, consisting of corridors and multiple rooms, with a relatively complex spatial structure, demonstrating the impact of wall obstruction and multipath propagation on signal coverage. The initial ray count of the antenna transmitter is set to 2×102. 7 The maximum depth of the signal propagation path was set to 20, and the number of iterations was set to 5. The simulated signal coverage map is shown below. Figure 5 As shown, it can be observed that areas with higher received power usually appear in locations with a direct path, while areas with lower received power are mostly distributed in areas where there are multiple walls blocking the source and the signal needs to be reflected multiple times to reach it.

[0108] Overall, the signal coverage map can intuitively reflect the impact of indoor spatial layout and obstacle distribution on wireless signal coverage. For a specific receiving point, its cell index can be determined based on the receiving point's location, and the corresponding predicted received power value can be extracted from the coverage map.

[0109] The above specific embodiments are used to explain and illustrate the present invention, but not to limit the present invention. Any modifications and changes made to the present invention within the spirit and scope of the claims shall fall within the protection scope of the present invention.

[0110] The above description is only a preferred embodiment of the present invention. Therefore, all equivalent changes or modifications made to the structure, features and principles described in the claims of this patent application are included in the scope of this patent application.

Claims

1. A method for predicting indoor wireless signal coverage based on Monte Carlo integration, characterized in that, The method includes the following steps: Step 1: Use a 3D model acquisition device to acquire a geometric model of the indoor scene with transmitting antennas and receivers. Step 2: The 3D model acquisition device combines the geometric model of the indoor scene with its own attribute parameters sent by the transmitting antenna to the computer. A simulation model with a channel gain simulation plane is established in the computer, and a finite number of rays are generated in the simulation space to describe the propagation of electromagnetic waves. The simulation is then performed based on the finite interactive propagation tracking iteration of reflection and transmission probability sampling to predict and obtain the indoor wireless signal coverage map. Step 3: Repeat Step 2 above, apply random rotation, and perform multiple simulations. Combine the indoor wireless signal coverage maps from the multiple simulations to obtain the final indoor wireless signal coverage prediction result. In step two, the simulation process is carried out in the computer according to the following procedure: S1. In the geometric model of the indoor scene, set a simulation plane at the height of the receiving antenna, divide the simulation plane into several cells of the same size, and set the channel gain for each cell as a signal position to draw a signal coverage map. S2. Using the position of the transmitting antenna as the starting point, generate several initial rays simulating electromagnetic wave emission using a uniform sampling method; S3. For each ray, perform intersection tracking in the geometric model of the indoor scene combined with the simulation plane, use Monte Carlo integration to calculate the path of all rays, and then update the channel gain of each cell in the signal coverage diagram. S4. For each ray, perform interactive ray tracking in the geometric model of the indoor scene to obtain the new ray after the interaction. Then, return to step S3 to process and update the signal coverage map according to all the new rays after the interaction. S5. Repeat steps S3 to S4 until the ray reaches the preset maximum path depth or the ray no longer intersects with the geometric model of the indoor scene. End ray tracing and use the channel gain of each cell on the simulation plane obtained in the last iteration as the predicted indoor wireless signal coverage map. Step S4 specifically includes: When the ray intersects the first object in the geometric model of the indoor scene along its own path, an interaction type is randomly selected from reflection and transmission based on the probability model. The new ray propagation direction is calculated based on the selected interaction type to obtain the new ray after the interaction, and the process returns to step S3.

2. The indoor wireless signal coverage prediction method based on Monte Carlo integration according to claim 1, characterized in that: Step S3 specifically includes: S31. Determine whether each ray intersects with the simulation plane using ray tracing: When a ray intersects the simulation plane, the cell containing the intersection point and its index are found as the intersection point information. S32. The intersection information of all rays is superimposed onto the channel gain of the cell containing the intersection point through a unified integration operation of Monte Carlo integration, thereby updating the channel gain of the cell containing the intersection point. This allows for the updating of the signal coverage map by iterating through all intersecting rays and updating the channel gain of the cells containing all intersection points.

3. The indoor wireless signal coverage prediction method based on Monte Carlo integration according to claim 2, characterized in that: Step S32 specifically includes: establishing the path of each ray relative to the starting point of the transmitting antenna, and establishing a sphere with the starting point of the transmitting antenna as the center. By using the Monte Carlo integration method on the sphere to perform unified integration processing on the paths of all rays combined with the intersection information, the received electric field of each ray at the intersection point of the simulation plane is obtained. Then, the channel gain of the cell where the intersection point is located is updated according to the magnitude of the received electric field intensity.

4. The indoor wireless signal coverage prediction method based on Monte Carlo integration according to claim 1, characterized in that: The specific steps of S4 are as follows: S41. First, based on the Fresnel reflection coefficient and transmission coefficient of the surface material at the intersection point of the ray and the object, construct the following probability model to calculate and determine the reflection probability q(R) and transmission probability q(T): q(R) = (|R ⊥ |²+|R ∥ |²) / (|R ⊥ |²+| R ∥ |²+| T ⊥ |²+| T ∥ |²) q(T) = (|T ⊥ |²+|T ∥ |²) / (|R ⊥ |²+| R ∥ |²+| T ⊥ |²+| T ∥ |²) Among them, R ⊥ R ∥ T represents the vertically polarized reflection coefficient and the horizontally polarized reflection coefficient, respectively. ⊥ T ∥ These represent the vertical polarization transmission coefficient and the horizontal polarization transmission coefficient, respectively. S42. Then the computer generates a random number, which takes the value between 0 and 1. If the random number is less than or equal to q(R), then reflection is selected as the interaction type of the current ray at the object. If the random number is greater than q(R), then transmission is selected as the interaction type of the current ray at the object. S43. After determining the interaction type, use the Fresnel reflection coefficient or transmission coefficient corresponding to the current interaction type, combined with the electric field of the current ray, to calculate the electric field of the next ray using the geometric optics method.

5. The indoor wireless signal coverage prediction method based on Monte Carlo integration according to claim 1, characterized in that: In step S2, a sphere is established with the position of the transmitting antenna as the center. Each sampling point is generated using a spherical sampling method based on Fibonacci lattice points. The direction of the line connecting the position of the transmitting antenna and each sampling point is set as the direction of each initial ray, forming an initial ray set.

6. The indoor wireless signal coverage prediction method based on Monte Carlo integration according to claim 1, characterized in that: Step three specifically includes applying random rotation to all generated initial ray directions each time step two is repeated, repeating the indoor wireless signal coverage map construction process, and averaging the indoor wireless signal coverage maps obtained from multiple rounds of simulation to obtain an average signal coverage map, which is used as the final indoor wireless signal coverage prediction result.

7. The indoor wireless signal coverage prediction method based on Monte Carlo integration according to claim 1, characterized in that: In step three, the initial ray direction of each round of multi-round simulation is transformed by applying a random rotation matrix to the sphere while keeping the number of samples constant, so as to obtain a set of initial ray directions different from the previous round, and the indoor wireless signal coverage map obtained from each round of simulation is averaged.

Images