A beam direction-based adaptive field strength prediction method and system

Abstract

The application relates to a beam direction-based adaptive field strength prediction method and system, wherein the method comprises acquiring scene data of a target area, and determining a transmitting end position, a receiving end position and beam codebook information; according to the receiving end position and a preset observation direction, a directional ray from the receiving end position to the observation direction is defined, and a space region passed through by the directional ray is discretized into a plurality of voxel points, so that voxel positions of the voxel points are obtained; the scene data, the transmitting end position, the observation direction, the voxel positions and the beam codebook information are input into a codebook-aware neural radiated field model, and propagation properties of the voxel points are output; and the propagation properties of the voxel points are rendered and aggregated along the directional ray, so that field strength prediction values of the receiving end position under different beam conditions are obtained. The method solves the problem of how to improve the flexibility of field strength prediction under a given beam condition in related technologies.

Description

Technical Field

[0001] This application relates to the field of field strength prediction, and in particular to an adaptive field strength prediction method and system based on beam direction. Background Technology

[0002] In the field of wireless communication, received signal strength prediction and channel knowledge graph construction are fundamental to coverage analysis, network planning, and beam management. Existing technologies mainly fall into three categories: traditional interpolation-based schemes, deep learning schemes based on two-dimensional image reconstruction, and model-driven schemes based on ray tracing or Neural Radiation Fields (NeRF). Among these, NeRF, by learning the field properties of spatial voxels and combining them with rendering aggregation, can express occlusion and multipath effects in complex environments with finer granularity, providing a new high-fidelity prediction path for this field.

[0003] However, the existing solutions still have inherent drawbacks: First, explicit modeling with location or environmental information as the main input makes it difficult for the model to distinguish the field strength differences of the same transmit / receive location under different beam conditions, resulting in low flexibility and prediction results that cannot directly support beam selection. Second, traditional two-dimensional methods are insufficient in expressing three-dimensional propagation mechanisms, while ray tracing or conventional NeRF, although capable of characterizing fine-grained propagation, struggle to simultaneously handle complex environment modeling and beam direction difference modeling. Third, if online beam scanning is used to compensate for missing information, system overhead increases; if all beam conditions are mapped separately, the data scale, computational complexity, and training costs increase significantly, making it difficult to meet the needs of rapid querying and deployment updates.

[0004] Currently, there is no effective solution in related technologies for improving the flexibility of field strength prediction under given beam conditions in beam direction field strength prediction methods based on codebook-sensing neural radiation fields. Summary of the Invention

[0005] This application provides an adaptive field strength prediction method and system based on beam direction, which at least solves the problem in related technologies of how to improve the flexibility of field strength prediction under given beam conditions.

[0006] In a first aspect, embodiments of this application provide an adaptive field strength prediction method based on beam direction, the method comprising: Acquire scene data of the target area and determine the location of the transmitter, the location of the receiver, and the beam codebook information; Based on the receiver position and the preset observation direction, a directional ray is defined from the receiver position to the observation direction, and the spatial region traversed by the directional ray is discretized into multiple voxel points to obtain the voxel position of each voxel point. The scene data, the transmitter position, the observation direction, the voxel position, and the beam codebook information are input into the codebook-sensing neural radiation field model, and the propagation attributes of each voxel point are output. The propagation properties of each voxel point are rendered and aggregated along the directional ray to obtain the predicted field strength value of the receiving end position under different beam conditions.

[0007] In one embodiment, acquiring scene data of the target area and determining the transmitter location, receiver location, and beam codebook information includes: Construct scene data for the target area, wherein the scene data is used to characterize environmental information affecting wireless propagation within the target area; Based on the scene data, determine the location of the transmitter and multiple receivers within the target area; Based on the transmitter location, beam codebook information corresponding to the transmitter location is determined. The beam codebook information is used to characterize the beam vector or codeword corresponding to different beam directions.

[0008] In one embodiment, the spatial region traversed by the directional ray is discretized into multiple voxel points to obtain the voxel position of each voxel point, including: Based on the receiver position, the observation direction, and the radial distance along the observation direction, multiple voxel points are obtained, and based on the multiple voxel points, the voxel position of each voxel point is obtained, wherein the voxel points are used to characterize the spatial contribution points of the signal during the ray propagation along the direction to the receiver position.

[0009] In one embodiment, the propagation properties include the attenuation properties and re-radiation contribution of the voxel point; wherein the attenuation properties characterize the propagation loss of the signal when it propagates through the current voxel point, and the re-radiation contribution characterizes the degree of contribution of the voxel point to the received signal strength under the current viewing direction and beam conditions.

[0010] In one embodiment, the rendering and aggregation of the propagation properties of each voxel point along the directional ray to obtain the predicted field strength value of the receiver position under different beam conditions includes: For each voxel point along the observation direction and the directional ray, the attenuation factors of each voxel point between the receiving end position and the current voxel point are multiplied together to obtain the cumulative attenuation coefficient from the current voxel point to the receiving end position. The cumulative attenuation coefficient is then multiplied by the re-radiation contribution of the current voxel point to obtain the effective signal contribution of the current voxel point. The effective signal contribution of each voxel point is weighted and summed according to the radial sampling interval to obtain the directional signal component corresponding to the observation direction. The received power corresponding to each observation direction is obtained by taking the square of the modulus of each of the signal components in each direction, and the received power is summed in the same linear power domain to obtain the predicted field strength of the receiving end position under the current beam conditions. The above processing is performed for different beam conditions to obtain the predicted field strength values ​​of the receiving end position under different beam conditions.

[0011] In one embodiment, after obtaining the predicted field strength values ​​for the receiver location under different beam conditions, the method further includes: The predicted field strength values ​​are used to train and optimize the codebook-sensing neural radiation field model. The trained codebook-based neural radiation field model is used to predict the field strength, obtain the predicted field strength value under the corresponding beam conditions, and generate the field strength distribution map of the target area based on multiple predicted field strength values.

[0012] In one embodiment, training and optimizing the codebook-sensing neural radiation field model using the predicted field strength includes: Obtain the actual measured field strength value corresponding to the predicted field strength value; Expand and concatenate the real and imaginary parts of the complex beam vector corresponding to the current beam to obtain the real-valued beam feature vector; Based on the attenuation attribute prediction branch and re-radiation contribution prediction branch of the codebook-sensory neural radiation field model, the real-valued beam feature vector is concatenated with the intermediate features of the corresponding branch to obtain joint features. The joint features are input into a fully connected mapping network for nonlinear mapping to generate scaling and offset vectors corresponding to the channel dimensions of the intermediate features. The scaling vector is multiplied element-wise with the intermediate feature, and the multiplication result is added element-wise with the offset vector to obtain the fused feature; Based on the fused features, the network parameters of the codebook-perceptual neural radiation field model are adjusted, and a loss function is constructed between the predicted field strength value and the actual measured field strength value. The network parameters are iteratively optimized based on the loss function until the preset convergence condition is met, thereby reducing the difference between the predicted field strength value and the actual measured field strength value.

[0013] In one embodiment, the step of predicting field strength using the trained codebook-sensing neural radiation field model to obtain predicted field strength values ​​under corresponding beam conditions, and generating a field strength distribution map of the target region based on multiple predicted field strength values, includes: Obtain the locations of multiple receivers to be predicted and target beam information within the target area; The scene data, the transmitter position, the receiver positions, the preset observation direction, and the target beam information are input into the trained codebook-aware neural radiation field model, and the propagation attributes of each voxel are output. Based on the propagation attributes, rendering aggregation processing is performed to obtain the field strength prediction value of each receiver position under the target beam condition. Based on the predicted field strength values, a field strength distribution map of the target region under the target beam condition is generated.

[0014] In one embodiment, before acquiring the multiple receiver locations to be predicted and the target beam information within the target area, the method further includes: The target area is divided into key areas and non-key areas, and the key areas are those where the field strength prediction accuracy is higher than a preset threshold. Based on the key area, each beam in the beam codebook information is filtered to obtain a filtered beam set. The filtered beam set is used as the target beam information.

[0015] Secondly, embodiments of this application provide an adaptive field strength prediction system based on beam direction. The system is used to execute the above-described method and includes a basic data acquisition module, a voxel position acquisition module, a propagation attribute acquisition module, and a field strength prediction value acquisition module, wherein: The basic data acquisition module is used to acquire scene data of the target area and determine the transmitter location, receiver location, and beam codebook information. The voxel position acquisition module is used to define a directional ray from the receiving end position to the observation direction based on the receiving end position and the preset observation direction, and to discretize the spatial region traversed by the directional ray into multiple voxel points to obtain the voxel position of each voxel point. The propagation attribute acquisition module is used to input the scene data, the transmitter position, the observation direction, the voxel position, and the beam codebook information into the codebook perception neural radiation field model, and output the propagation attributes of each voxel point; The field strength prediction module is used to render and aggregate the propagation properties of each voxel point along the directional ray to obtain the field strength prediction value of the receiving end position under different beam conditions.

[0016] Thirdly, embodiments of this application provide a computer device, including a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, it implements an adaptive field strength prediction method based on beam direction as described in the first aspect above.

[0017] Fourthly, embodiments of this application provide a computer-readable storage medium storing a computer program thereon, which, when executed by a processor, implements an adaptive field strength prediction method based on beam direction as described in the first aspect above.

[0018] The adaptive field strength prediction method and system based on beam direction provided in this application embodiment have at least the following technical effects.

[0019] By introducing beam codebook information as an explicit conditional variable into the neural radiation field modeling framework, the model input simultaneously includes information from three dimensions: spatial location, scene environment, and beam configuration. The output is the voxel propagation attributes corresponding to the beam conditions, which are then rendered and aggregated to obtain the final field strength prediction value. This expands the traditional low-dimensional mapping from position to field strength into a multi-dimensional mapping from position, direction, and beam conditions to field strength. This allows for differentiated field strength prediction results for the same transmit / receive location under different beam conditions, enabling adaptive prediction based on beam direction and improving flexibility. This addresses the problem of improving the flexibility of field strength prediction under given beam conditions in related technologies.

[0020] Details of one or more embodiments of this application are set forth in the following drawings and description to make other features, objects and advantages of this application more readily apparent. Attached Figure Description

[0021] The accompanying drawings, which are included to provide a further understanding of this application and form part of this application, illustrate exemplary embodiments and are used to explain this application, but do not constitute an undue limitation of this application. In the drawings: Figure 1 This is a flowchart of an adaptive field strength prediction method based on beam direction; Figure 2 This is a general flowchart illustrating beam-direction adaptive field strength prediction according to an exemplary embodiment. Figure 3 This is a schematic diagram illustrating a voxel discretization, attenuation accumulation, and reradiative aggregation process based on an exemplary embodiment. Figure 4 This is a structural block diagram of an electronic device according to an exemplary embodiment. Detailed Implementation

[0022] To make the objectives, technical solutions, and advantages of this application clearer, the application is described and illustrated below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the scope of this application. All other embodiments obtained by those skilled in the art based on the embodiments provided in this application without inventive effort are within the scope of protection of this application.

[0023] Obviously, the accompanying drawings described below are merely some examples or embodiments of this application. Those skilled in the art can apply this application to other similar scenarios based on these drawings without any inventive effort. Furthermore, it is understood that although the efforts made in this development process may be complex and lengthy, for those skilled in the art related to the content disclosed in this application, any changes to design, manufacturing, or production based on the technical content disclosed in this application are merely conventional technical means and should not be construed as insufficient disclosure of the content of this application.

[0024] In this application, the reference to "embodiment" means that a specific feature, structure, or characteristic described in connection with an embodiment may be included in at least one embodiment of this application. The appearance of this phrase in various places in the specification does not necessarily refer to the same embodiment, nor is it a separate or alternative embodiment that is mutually exclusive with other embodiments. It will be explicitly and implicitly understood by those skilled in the art that the embodiments described in this application may be combined with other embodiments without conflict.

[0025] Unless otherwise defined, the technical or scientific terms used in this application shall have the ordinary meaning understood by one of ordinary skill in the art to which this application pertains. The terms “a,” “an,” “an,” “the,” and similar words used in this application do not indicate quantity limitation and may indicate singular or plural. The terms “comprising,” “including,” “having,” and any variations thereof used in this application are intended to cover non-exclusive inclusion; for example, a process, method, system, product, or device that includes a series of steps or modules (units) is not limited to the listed steps or units, but may also include steps or units not listed, or may include other steps or units inherent to these processes, methods, products, or devices. The terms “connected,” “linked,” “coupled,” and similar words used in this application are not limited to physical or mechanical connections, but may include electrical connections, whether direct or indirect. “Multiple” used in this application refers to two or more. “And / or” describes the relationship between related objects, indicating that three relationships may exist; for example, “A and / or B” can represent: A alone, A and B simultaneously, and B alone. The character " / " generally indicates that the preceding and following objects are in an "or" relationship. The terms "first," "second," and "third" used in this application are merely to distinguish similar objects and do not represent a specific ordering of the objects.

[0026] In the field of wireless communication, received signal strength prediction and channel knowledge graph construction are fundamental technologies supporting coverage analysis, network planning, and beam management. Existing methods typically include traditional interpolation-based methods, deep learning methods based on two-dimensional image reconstruction, and model-driven methods based on ray tracing or wireless neural radiation fields. For general field strength prediction tasks, these methods can establish a correspondence between spatial location and wireless propagation results to some extent. However, when the system uses a finite beamcodebook for beam management, the received signal strength at the same spatial location may vary significantly under different beam conditions. If the modeling objective is still simplified to a low-dimensional mapping from location to channel or from location to field strength, it is difficult to directly characterize the impact of beam direction on propagation results, and it is also difficult to make the prediction results directly serve beam selection and coverage assessment.

[0027] To improve modeling capabilities in complex propagation environments, existing technologies have proposed incorporating neural radiation fields into wireless propagation modeling. These methods borrow from the concepts of 3D radiation field representation and rendering, avoiding direct neural network regression of received signal strength. Instead, they learn the field properties of spatial voxels and combine this with directional accumulation or rendering aggregation processes to obtain the final received result. Compared to simple 2D map prediction, these methods can express occlusion, multipath, and directional propagation characteristics in complex environments with finer granularity, providing a new technical path for high-fidelity field strength prediction.

[0028] Despite the progress made by the above technologies, existing technologies still exhibit the following inherent shortcomings in beam direction adaptive field strength prediction tasks for finite beam codebooks: First, the expression of beam conditions is insufficient: Most existing field strength prediction or channel knowledge graph methods still take location or location and environment information as the main input, lack explicit modeling of beam vectors or codewords, and it is difficult to accurately distinguish the field strength difference of the same transmit and receive location under different beam conditions, which makes it difficult to directly use the prediction results for beam management and beam selection.

[0029] Second, it is difficult to simultaneously consider both three-dimensional propagation mechanisms and beam differences: traditional two-dimensional map restoration methods are better at learning spatial distribution results, but have limited ability to express complex three-dimensional propagation mechanisms; while conventional ray tracing or wireless neural radiation field methods can describe more fine-grained propagation processes, they often do not introduce beam conditions as a unified modeling variable, making it difficult to simultaneously consider complex environment propagation modeling and beam direction difference modeling.

[0030] Third, the data and computational overhead is relatively large: if a large number of beam scans are still needed to make up for the missing information in the map during the online phase, the effect of field strength map in reducing overhead will be weakened; if all beam conditions are to be included in the mapping separately, the data scale, computational complexity and model training cost will increase rapidly, making map updates, deployment and fast querying more difficult.

[0031] Therefore, there is an urgent need in this field for a method that can simultaneously incorporate spatial location, propagation direction, and beam conditions, while also considering the complex three-dimensional propagation mechanism and rapid field strength prediction capabilities. Specifically, this requires an adaptive field strength prediction method and system based on beam direction to achieve adaptive field strength prediction for finite beam codebooks, addressing the challenge of improving the flexibility of field strength prediction under given beam conditions in related technologies. This would provide more direct technical support for subsequent coverage analysis, scheme comparison, and beam management.

[0032] In this document, it should be understood that the terms used may be technical means used to implement part of this application or other summary technical terms. For example, terms may include: Beam codebook information refers to a set of candidate beam information predefined by the transmitter (such as a base station). In wireless communication systems such as 5G / 6G, the transmitter forms beams with different directions and shapes by configuring different antenna phases and amplitude weights. These predefined beam configuration parameter sets constitute the beam codebook. In this application, the beam codebook information is used as one of the input conditions of the model to distinguish the influence of different beam directions on the field strength prediction results.

[0033] Beam vector or codeword: refers to the precoded weight vector or codeword index corresponding to each candidate beam in the beamcodebook. The beam vector typically contains the amplitude and phase configuration parameters of each antenna element, used to generate a beam pointing in a specific direction in the digital domain; the codeword is the quantization index identifier of the vector. In this application, the beam vector or codeword is input into the model as a specific data form of the beamcodebook information.

[0034] Observation direction: refers to the spatial direction in which the receiving end (such as a terminal device) receives the incoming wave. In this application, it is specifically represented as a unit vector pointing from the receiving end position towards the signal source direction. This direction is used to define the direction ray, that is, the model samples spatial voxels and infers attributes along this direction.

[0035] A voxel is a discrete spatial location point obtained by discretizing a spatial region along a directional ray. "Voxel" is short for "volume pixel," representing a sampling point or the smallest volume unit in three-dimensional space. In this application, each voxel represents a spatial location on the signal propagation path. The model outputs the attenuation properties and re-radiation contribution for each voxel, thereby characterizing the cumulative effect of the signal along the propagation path.

[0036] Codebook-aware neural radiation field model: refers to the improved neural radiation field model proposed in this application. Based on the standard neural radiation field, this model introduces beam codebook information (beam vector or codeword) as an explicit conditional variable into the input layer, enabling the model to distinguish the influence of different beam conditions on the propagation results while learning the propagation properties of spatial voxels, thereby achieving "beam direction adaptive" field strength prediction.

[0037] Rendering aggregation: refers to the process of accumulating and calculating the propagation properties (attenuation properties and re-radiation contribution) at the voxel level along the directional ray and converting them into a predicted field strength value at the receiver level. In this application, it specifically refers to the process where the re-radiation signal of each voxel point is affected by the attenuation of voxels along the way during propagation to the receiver, and then accumulated along the directional ray to finally obtain the received signal strength at the receiver.

[0038] Beam conditions: refer to the configuration parameters or status related to the beam. In this application, it specifically refers to the specific beam configuration determined by the beam codebook information (such as the beam pointing and beamwidth corresponding to a certain beam vector or codeword).

[0039] In a first aspect, embodiments of this application provide an adaptive field strength prediction method based on beam direction. Figure 1 This is a flowchart of an adaptive field strength prediction method based on beam direction, such as... Figure 1 As shown, the method includes: Step S101: Obtain scene data of the target area and determine the location of the transmitter, the location of the receiver, and the beam codebook information.

[0040] Step S102: Based on the receiver position and the preset observation direction, define a directional ray from the receiver position to the observation direction, and discretize the spatial region traversed by the directional ray into multiple voxel points to obtain the voxel position of each voxel point.

[0041] Step S103: Input the scene data, transmitter position, observation direction, voxel position and beam codebook information into the codebook perception neural radiation field model, and output the propagation attributes of each voxel point.

[0042] Step S104: Render and aggregate the propagation properties of each voxel point along the directional ray to obtain the predicted field strength value of the receiver position under different beam conditions.

[0043] In summary, this application provides an adaptive strong field strength prediction method based on beam direction. By introducing beam codebook information as an explicit conditional variable into the neural radiation field modeling framework, the model input simultaneously includes information in three dimensions: spatial location, scene environment, and beam configuration. The output is the voxel propagation attributes corresponding to the beam conditions, which are then rendered and aggregated to obtain the final field strength prediction value. This expands the traditional low-dimensional mapping from position to field strength into a multi-dimensional mapping from position, direction, and beam conditions to field strength. This allows for differentiated field strength prediction results at the same transmit / receive location under different beam conditions, enabling adaptive prediction based on beam direction and improving flexibility. This addresses the problem in related technologies of how to improve the flexibility of field strength prediction under given beam conditions.

[0044] Figure 2 This is a general flowchart illustrating beam direction adaptive field strength prediction according to an exemplary embodiment, such as... Figure 2 As shown, in one embodiment, step S101 involves acquiring scene data of the target area and determining the transmitter location, receiver location, and beam codebook information. Specifically, this includes the following steps: Step S1011: Construct scene data for the target area. The scene data is used to characterize environmental information affecting wireless propagation within the target area. Step S1012: Based on the scene data, determine the location of the transmitter and multiple receivers within the target area; Step S1013: Based on the transmitter location, determine the beam codebook information corresponding to the transmitter location. The beam codebook information is used to characterize the beam vector or codeword corresponding to different beam directions.

[0045] Optionally, scene data for the target area is first constructed. Scene data is used to characterize environmental information affecting wireless propagation within the target area, and can take the form of building distribution, occlusion information, grid maps, 3D scene models, or other data formats capable of characterizing the spatial propagation environment. Simultaneously, the transmitter location, receiver location set, and beam codebook information corresponding to the transmitter are determined. The beam codebook information is used to characterize the beam vectors or codewords corresponding to different beam directions.

[0046] In one embodiment, step S102 involves defining a directional ray from the receiver position to the observation direction based on the receiver position and a preset observation direction, and discretizing the spatial region traversed by the directional ray into multiple voxel points to obtain the voxel position of each voxel point. Specifically, this includes: Based on the receiver position, the observation direction, and the radial distance along the observation direction, multiple voxel points are obtained, and based on the multiple voxel points, the voxel position of each voxel point is obtained. The voxel points are used to characterize the spatial contribution points of the signal as it propagates along the direction ray to the receiver position.

[0047] Optionally, for a given receiver location and observation direction Define a direction ray associated with the receiver. This direction ray originates from the receiver and travels along the direction... Extending, its spatial position can be represented as

[0048] in, Indicates along direction radial distance, A point in space along this direction is called a voxel.

[0049] In the direction of observation Below, the spatial region traversed by the ray is discretized into multiple voxel points to characterize the spatial contribution of the signal as it propagates along this direction to the receiver. Each voxel on the ray in this direction can be considered a new virtual transmitter, and the receiver starts from the direction... The received signal is the accumulation of the signals re-radiated from these voxels toward the receiver; among them, the re-radiated signal of any voxel is also affected by the attenuation of other voxels between it and the receiver during the propagation process.

[0050] To adapt to different scene scales and propagation complexities, the number of voxels in this direction can be adjusted based on scene boundaries, propagation distance, or computational configuration. For regions where propagation conditions vary significantly, a higher density of voxels can be used to enhance the model's ability to represent complex propagation processes.

[0051] In one embodiment, Figure 3 This is a schematic diagram illustrating the voxel discretization, attenuation accumulation, and re-radiative aggregation process based on an exemplary embodiment, as shown below. Figure 3 As shown, step S103 involves inputting scene data, transmitter location, observation direction, voxel location, and beam codebook information into the codebook-sensing neural radiation field model, and outputting the propagation attributes of each voxel point. Specifically, this includes: The propagation properties include the voxel attenuation property and the re-radiation contribution; the attenuation property is used to characterize the propagation loss of the signal when it passes through the current voxel, and the re-radiation contribution is used to characterize the degree of contribution of the voxel to the received signal strength under the current observation direction and beam conditions.

[0052] Optionally, scene environment information, voxel positions, transmitter positions, viewing direction, and the beam vector or codeword corresponding to the current beam are input into the codebook-based perceptual neural radiation field model. The model is used to learn the influence of each voxel on the wireless propagation process under the current environment and beam conditions, and outputs attributes related to the propagation process.

[0053] In this application, the model output attributes include voxel attenuation attributes and re-radiation contribution. The attenuation attribute characterizes the propagation loss of the signal as it passes through the current voxel, while the re-radiation contribution characterizes the degree to which the voxel contributes to the final received signal strength under the current viewing direction and beam conditions. By explicitly introducing beam vectors or codewords, different field properties and final received signal strength results can be generated for the same spatial location under different beam conditions.

[0054] At the implementation level, the beam vector can be lightweight encoded and then fused with the original voxel position, transmitter position, and observation direction information, enabling the model to explicitly receive and distinguish observations under different beam and codeword conditions. In this way, the model can not only learn the spatial propagation laws but also learn the modulation effect of different beam conditions on the propagation results, thus providing a unified modeling entry point for subsequent beam-oriented field strength map construction.

[0055] In one embodiment, step S104 involves rendering and aggregating the propagation properties of each voxel point along the directional ray to obtain the predicted field strength value of the receiver location under different beam conditions. Specifically, this includes the following steps: Step S1041: For each voxel point along the observation direction and the directional ray, multiply the attenuation factors of each voxel point between the receiver position and the current voxel point to obtain the cumulative attenuation coefficient from the current voxel point to the receiver position, and multiply the cumulative attenuation coefficient by the re-radiation contribution of the current voxel point to obtain the effective signal contribution of the current voxel point.

[0056] Step S1042: Weight the effective signal contribution of each voxel point according to the radial sampling interval to obtain the directional signal component corresponding to the observation direction.

[0057] Step S1043: Take the square of the modulus of each signal component in each direction to obtain the received power corresponding to each observation direction, and sum the received power in the same linear power domain to obtain the predicted field strength value of the receiver position under the current beam conditions.

[0058] Step S1044: Perform the above steps S1041-S1043 for different beam conditions to obtain the field strength prediction value of the receiver position under different beam conditions.

[0059] Optionally, after obtaining the attenuation properties and re-radiation contributions of each voxel, for each of the preset observation directions and each voxel on the corresponding directional ray, the attenuation factors of each voxel along the path between the voxel and the receiver position are multiplied sequentially to obtain the cumulative attenuation coefficient from the voxel to the receiver. This cumulative attenuation coefficient is then multiplied by the re-radiation contribution of the voxel to obtain the effective signal contribution of the voxel to the receiver. Next, for each observation direction, the effective signal contributions of each voxel are weighted and summed along the corresponding directional ray at radial sampling intervals to synthesize the directional signal components received by the receiver from each spatial voxel in that direction. Subsequently, the modulus of each directional signal component is squared to convert it into the corresponding received power for each direction, and the received power is summed within the same linear power domain to synthesize the signal energy from different directions, obtaining the predicted field strength of the receiver under the current beam conditions. Finally, the above rendering and aggregation process is repeated for different beam conditions to obtain the predicted field strength of the receiver position under different beam conditions.

[0060] The aggregation process essentially transforms voxel-level propagation properties into receiver-level field strength predictions: the final received signal near the receiver is determined by multiple voxels in that direction, with each voxel's contribution related to its reradiative contribution and cumulative attenuation along the path. Given the transmitter location, receiver location, and beam conditions, the predicted received signal strength under the current conditions can be obtained by rendering and aggregating the voxel contributions in relevant directions. Since the inner-layer field function is explicitly modulated by the beam conditions, the outer-layer rendering / aggregation results naturally reflect the field strength differences under different beam conditions.

[0061] In one embodiment, after obtaining the predicted field strength values ​​of the receiver location under different beam conditions in step S104, the method further includes: Step S105: Train and optimize the codebook-sensing neural radiation field model using the predicted field strength values. Specifically, this includes: Obtain the actual measured field strength corresponding to the predicted field strength value; Expand and concatenate the real and imaginary parts of the complex beam vector corresponding to the current beam to obtain the real-valued beam feature vector; Based on the attenuation attribute prediction branch and re-radiation contribution prediction branch of the codebook-sensory neural radiation field model, the real-valued beam feature vector is concatenated with the intermediate features of the corresponding branch to obtain the joint feature. The joint features are input into a fully connected mapping network for nonlinear mapping, generating scaling and offset vectors corresponding to the channel dimensions of the intermediate features; The scaling vector is multiplied element-wise with the intermediate features, and the result of the multiplication is added element-wise with the offset vector to obtain the fused features. Based on the fused features, the network parameters of the codebook-perceived neural radiation field model are adjusted, and a loss function is constructed between the predicted field strength and the actual measured field strength. The network parameters are iteratively optimized based on the loss function until the preset convergence condition is met, thereby reducing the difference between the predicted field strength and the actual measured field strength.

[0062] Optionally, during the training phase, the received signal strength predicted by the model is compared with the actual received signal strength label to construct a loss function, and the codebook-based perceptual neural radiation field model is trained using a parameter optimization algorithm. Through training, the model gradually learns the wireless propagation patterns under different scene environments, spatial locations, viewing directions, and beam conditions, improving the consistency between the predicted field strength and the actual value. The loss function can be mean squared error loss, absolute error loss, or other loss forms suitable for field strength regression prediction tasks. To further improve the model's efficiency in utilizing beam conditions, corresponding feature extraction and fusion mechanisms can be introduced on the original network structure, allowing beam information to more fully influence intermediate features and the final output, thereby improving training convergence speed, prediction error, and generalization performance.

[0063] The training optimization process embeds beam vectors into the intermediate features of the model in a scaling and offset manner, enabling beam information to participate deeply in the forward propagation calculation. Compared with simply splicing beam information at the input layer, this enhances the model's ability to represent the differences in propagation under different beam conditions, which is beneficial to improving training convergence speed and prediction accuracy.

[0064] Step S106: Using the trained codebook-based perceptual neural radiation field model, predict the field strength to obtain the predicted field strength value under the corresponding beam conditions, and generate a field strength distribution map of the target area based on multiple predicted field strength values. Specifically, this includes: Acquire the locations of multiple receivers to be predicted within the target area and the target beam information; The scene data, transmitter position, receiver positions, preset observation direction and target beam information are input into the trained codebook perception neural radiation field model, which outputs the propagation attributes of each voxel point. Based on the propagation attributes, rendering aggregation processing is performed to obtain the field strength prediction value of each receiver position under the target beam condition. Based on the predicted field strength values, a field strength distribution map of the target area under the target beam condition is generated.

[0065] Optionally, during the inference phase, for multiple receiver locations to be predicted within the target area, corresponding scene environment information, transmitter location, receiver location, observation direction, and target beam information are input respectively. The trained codebook-aware neural radiation field model outputs the predicted value of the received signal strength. Repeating the above process for multiple receiver locations within the same area generates a field strength distribution map under given beam conditions. Furthermore, by traversing multiple candidate beams for the same transmitter, field strength distribution results under multiple beam directions can be obtained, thereby realizing multi-beam field strength map construction, coverage effect comparison, and fast field strength query. In this way, the original low-dimensional field strength prediction can be extended to beam direction adaptive field strength prediction results under finite beam codebook conditions, more directly serving tasks such as coverage assessment, scheme selection, and beam management.

[0066] By introducing beam vectors or codewords as explicit conditional variables into the wireless neural radiation field modeling process, an adaptive field strength prediction process for beam direction under finite beam codebook conditions was established. This process not only learns the field properties of spatial voxels in complex propagation environments but also aggregates the received signal strength predictions corresponding to different beam conditions through outer rendering, thereby improving the adaptability of the field strength prediction results to beam management, coverage analysis, and scheme comparison tasks.

[0067] In one embodiment, before acquiring the multiple receiver locations to be predicted within the target area and the target beam information, the method further includes: The target area is divided into key areas and non-key areas. Key areas are those where the field strength prediction accuracy is higher than a preset threshold. Based on key areas, each beam in the beam codebook information is filtered to obtain a filtered beam set. The filtered set of beams is used as the target beam information.

[0068] Optionally, under constrained deployment conditions, an application process for adaptive beam direction field strength prediction based on codebook-aware neural radiation fields is implemented. This embodiment is suitable for scenarios where computational resources are limited during the online inference phase, response latency requirements are high, or it is difficult to predict with equal accuracy for the complete candidate beam set and all spatial locations. Under these conditions, adaptive beam direction field strength prediction for the target area can still be achieved through candidate beam pre-screening and spatial discrete resolution adjustment.

[0069] First, scene data for the target area is still constructed, and the locations of the transmitter and receiver, as well as the beamcodebook information corresponding to the transmitter, are determined. In this embodiment, it is not necessary to perform high-precision prediction on all candidate beams one by one during the online phase. Instead, based on service requirements, historical experience, or preset rules, candidate beams can be pre-screened from the complete beamcodebook, and subsequent fine-grained predictions are only performed on the screened candidate beams. In this way, the computational burden during the online inference phase can be reduced while retaining the beam condition modeling capability.

[0070] Subsequently, for a given receiver location and observation direction, the space is discretized using voxels. To adapt to computational constraints under limited deployment conditions, a lower spatial discretization resolution can be used, or a higher discretization resolution can be used only for the target area, key coverage areas, and spatial regions along the key action directions of candidate beams, while a lower discretization resolution is used for non-key areas. By adapting the spatial discretization resolution, the number of voxels and the cumulative computational overhead along the directional rays can be reasonably controlled according to the importance of the target area and computational resource conditions.

[0071] During the modeling phase, scene environment information, voxel positions, transmitter positions, observation directions, and beam vectors or codewords corresponding to pre-selected candidate beams are still input into the codebook-aware neural radiation field model to obtain the attenuation properties and re-radiation contributions of the corresponding voxels. The model structure itself remains unchanged, so this embodiment still retains the codebook-aware modeling capability, only adapting the input scale and spatial discretization accuracy to the constraints of deployment conditions.

[0072] After obtaining the attenuation properties and re-radiation contributions for each voxel, the outer rendering / aggregation operator accumulates the voxel contributions along the directional ray corresponding to the observation direction to obtain the predicted received signal strength at the receiver under candidate beam conditions. Repeating this process for multiple receiver locations within the target area generates the corresponding field strength distribution results within the pre-selected candidate beam range, thus meeting the requirements for rapid querying and near real-time evaluation.

[0073] In this embodiment, model training can still utilize relatively complete training data for parameter optimization during the offline phase; during the online deployment phase, a simplified strategy of candidate beam pre-screening and spatial discrete resolution adjustment is adopted to reduce the burden of real-time inference. For situations requiring further improvement in local prediction accuracy, local refinement predictions can be performed on key locations or key beams based on the lower spatial resolution prediction results, thereby achieving a balance between prediction efficiency and prediction accuracy.

[0074] Unlike existing technologies that degenerate into low-dimensional field strength prediction or simple static estimation when computational resources are limited, this embodiment retains the key features of the invention under constrained deployment conditions: explicit introduction of beam conditions, voxel-level field property modeling, and rendering aggregation of directional rays. By adapting the candidate beam range and spatial discretization resolution, this embodiment enables the invention to not only perform high-precision prediction under standard deployment conditions but also maintain availability and engineering adaptability under resource-constrained conditions, thereby expanding the practical application scope of the invention.

[0075] In summary, the adaptive field strength prediction method based on beam direction provided in this application introduces beam codebook information as an explicit conditional variable into the neural radiation field model. This allows the model to establish a "beam-space-signal" relationship while learning the propagation properties of spatial voxels, thereby changing the inherent pattern of "same location, same field strength" in traditional field strength prediction. This enables the same receiver location to obtain differentiated field strength prediction values ​​under different beam conditions, achieving adaptive prediction based on beam direction. This design brings significant flexibility improvements: during the network planning phase, there is no need to train a model separately for each beam direction or repeatedly perform highly complex ray tracing calculations, nor is it necessary to obtain coverage information for different beams through numerous beam scans during the online phase. Simply switching the beam codebook information in the input during inference allows for quick querying of field strength prediction results for any beam direction.

[0076] This application also has the following beneficial effects: This invention achieves beam-direction adaptive field strength prediction under finite beam codebook conditions: Existing field strength prediction or channel knowledge graph methods mostly focus on low-dimensional mapping from location to channel, which is difficult to directly reflect the propagation differences under different beam conditions; while this invention introduces beam vectors or codewords as explicit conditional variables into the model input, so that the same set of transmit and receive positions can correspond to different field properties and final received signal strength results under different beam conditions, thereby enabling the prediction results to more directly serve tasks such as beam selection, coverage analysis and beam management.

[0077] This invention improves the expressive power and interpretability of field strength prediction in complex 3D propagation environments: Instead of directly regressing the received signal strength via a neural network, it first learns the attenuation properties and re-radiation contributions of voxels in the inner network, and then accumulates the voxel contributions along the directional rays using the outer rendering / aggregation operator to obtain the final field strength result. This preserves the fine-grained expressive power for multipath, occlusion, and directional differences in complex propagation environments, while giving the prediction process a clearer physical meaning. Compared to purely data-driven direct regression methods, this approach is more conducive to explaining the sources of field strength variations under different spatial locations and beam conditions.

[0078] This invention enhances the model's ability to utilize beam information, which is beneficial for improving training convergence and prediction accuracy. By introducing beam vectors, this invention not only uses them as additional input but also employs corresponding feature extraction and fusion mechanisms to enable the model to more effectively utilize beam information and establish a correlation between "beam-space-signal strength." Compared to simply splicing beams together, this approach is more effective in improving the model's ability to represent propagation differences under different beam conditions, thereby improving training convergence speed, prediction error, and generalization performance.

[0079] This invention supports the construction and rapid querying of multi-beam field strength maps, demonstrating significant engineering application value. After training, it can traverse multiple candidate beams for the same transmitter, rapidly generating field strength distribution results corresponding to different beam directions. This enables the construction of multi-beam field strength maps, comparison of coverage effects, and rapid field strength querying. Compared to repeatedly performing highly complex ray tracing calculations or large-scale field measurements, this method helps reduce the time cost of field strength prediction and planning evaluation, improves prediction efficiency and application flexibility, and is suitable for wireless network planning and simulation evaluation scenarios such as 5G / 6G and WiFi. Furthermore, it is easily packaged into a core algorithm module of related software systems.

[0080] Secondly, embodiments of this application provide an adaptive field strength prediction system based on beam direction, including a basic data acquisition module, a voxel position acquisition module, a propagation attribute acquisition module, and a field strength prediction value acquisition module, wherein: The basic data acquisition module is used to acquire scene data of the target area and determine the transmitter location, receiver location, and beam codebook information; The voxel position acquisition module is used to define a directional ray from the receiver position to the observation direction based on the receiver position and the preset observation direction, and to discretize the spatial region traversed by the directional ray into multiple voxel points to obtain the voxel position of each voxel point. The propagation attribute acquisition module is used to input scene data, transmitter position, observation direction, voxel position and beam codebook information into the codebook perception neural radiation field model, and output the propagation attributes of each voxel point; The field strength prediction module is used to render and aggregate the propagation properties of each voxel point along the directional ray to obtain the field strength prediction value of the receiver position under different beam conditions.

[0081] In summary, this application provides an adaptive field strength prediction system based on beam direction. By introducing beam codebook information as an explicit conditional variable into the neural radiation field modeling framework, the model input simultaneously includes information in three dimensions: spatial location, scene environment, and beam configuration. The output is the voxel propagation attributes corresponding to the beam conditions, which are then rendered and aggregated to obtain the final predicted field strength value. This expands the traditional low-dimensional mapping from position to field strength into a multi-dimensional mapping from position, direction, and beam conditions to field strength. This allows for differentiated field strength prediction results for the same transmit / receive location under different beam conditions, enabling adaptive prediction based on beam direction and improving flexibility. It addresses the problem in related technologies of how to improve the flexibility of field strength prediction under given beam conditions.

[0082] It should be noted that the adaptive field strength prediction system based on beam direction provided in this embodiment is used to implement the above-described implementation method, and details already described will not be repeated. As used above, the terms "module," "unit," "subunit," etc., can refer to a combination of software and / or hardware that performs a predetermined function. Although the apparatus described in the above embodiments is preferably implemented in software, hardware implementation, or a combination of software and hardware, is also possible and contemplated.

[0083] Thirdly, embodiments of this application provide an electronic device, Figure 4 This is a block diagram illustrating an electronic device according to an exemplary embodiment. (e.g.) Figure 4 As shown, the electronic device may include a processor 41 and a memory 42 storing computer program instructions.

[0084] Specifically, the processor 41 may include a central processing unit (CPU), an application specific integrated circuit (ASIC), or one or more integrated circuits that can be configured to implement the embodiments of this application.

[0085] The memory 42 may include a large-capacity memory for data or instructions. For example, and not limitingly, the memory 42 may include a hard disk drive (HDD), a floppy disk drive, a solid-state drive (SSD), flash memory, an optical disk drive, a magneto-optical disk drive, magnetic tape, or a Universal Serial Bus (USB) drive, or a combination of two or more of these. Where appropriate, the memory 42 may include removable or non-removable (or fixed) media. Where appropriate, the memory 42 may be internal or external to a data processing device. In a particular embodiment, the memory 42 is non-volatile memory. In a particular embodiment, the memory 42 includes read-only memory (ROM) and random access memory (RAM). Where appropriate, the ROM may be a mask-programmed ROM, a programmable read-only memory (PROM), an erasable programmable read-only memory (EPROM), an electrically erasable programmable read-only memory (EEPROM), an electrically alterable read-only memory (EAROM), or flash memory, or a combination of two or more of these. Where appropriate, the RAM can be Static Random-Access Memory (SRAM) or Dynamic Random-Access Memory (DRAM). DRAM can be Fast Page Mode Dynamic Random-Access Memory (FPMDRAM), Extended Data Out Dynamic Random-Access Memory (EDODRAM), Synchronous Dynamic Random-Access Memory (SDRAM), etc.

[0086] The memory 42 can be used to store or cache various data files that need to be processed and / or used for communication, as well as possible computer program instructions executed by the processor 41.

[0087] The processor 41 reads and executes the computer program instructions stored in the memory 42 to implement any of the adaptive field strength prediction methods based on beam direction in the above embodiments.

[0088] In one embodiment, an adaptive field strength prediction device based on beam direction may further include a communication interface 43 and a bus 40. Wherein, as Figure 4 As shown, the processor 41, memory 42, and communication interface 43 are connected through bus 40 and complete communication with each other.

[0089] The communication interface 43 is used to enable communication between the various modules, devices, units, and / or equipment in the embodiments of this application. The communication interface 43 can also enable data communication with other components such as external devices, image / data acquisition devices, databases, external storage, and image / data processing workstations.

[0090] Bus 40 includes hardware, software, or both, that couples together components of a beam-direction-based adaptive field strength prediction device. Bus 40 includes, but is not limited to, at least one of the following: data bus, address bus, control bus, expansion bus, and local bus. For example, and not as a limitation, bus 40 may include an Accelerated Graphics Port (AGP) or other graphics bus, an Extended Industry Standard Architecture (EISA) bus, a Front Side Bus (FSB), a HyperTransport (HT) interconnect, an Industry Standard Architecture (ISA) bus, an InfiniBand interconnect, a Low Pin Count (LPC) bus, a memory bus, a Micro Channel Architecture (MCA) bus, a Peripheral Component Interconnect (PCI) bus, a PCI-Express (PCI-X) bus, a Serial Advanced Technology Attachment (SATA) bus, a Video Electronics Standards Association Local Bus (VLB) bus, or other suitable buses, or a combination of two or more of these. Where appropriate, bus 40 may include one or more buses. Although specific buses are described and illustrated in the embodiments of this application, this application considers any suitable bus or interconnection.

[0091] Fourthly, embodiments of this application provide a computer-readable storage medium having a program stored thereon, which, when executed by a processor, implements the adaptive field strength prediction method based on beam direction provided in the first aspect.

[0092] The readable storage medium may be more specifically adopted, including but not limited to: portable disk, hard disk, random access memory, read-only memory, erasable programmable read-only memory, optical storage device, magnetic storage device, or any suitable combination thereof.

[0093] In a possible implementation, the present invention can also be implemented as a program product comprising program code, which, when the program product is run on a terminal device, causes the terminal device to perform steps to implement the adaptive field strength prediction method based on beam direction provided in the first aspect.

[0094] The program code for executing the present invention can be written in any combination of one or more programming languages. The program code can be executed entirely on the user device, partially on the user device, as a standalone software package, partially on the user device and partially on a remote device, or entirely on a remote device.

[0095] The technical features of the above embodiments can be combined in any way. For the sake of brevity, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0096] The embodiments described above are merely illustrative of several implementation methods of this application, and while the descriptions are relatively specific and detailed, they should not be construed as limiting the scope of the invention patent. It should be noted that those skilled in the art can make various modifications and improvements without departing from the concept of this application, and these all fall within the protection scope of this application. Therefore, the protection scope of this patent application should be determined by the appended claims.

Claims

1. An adaptive field strength prediction method based on beam direction, characterized in that, The method includes: Acquire scene data of the target area and determine the location of the transmitter, the location of the receiver, and the beam codebook information; Based on the receiver position and the preset observation direction, a directional ray is defined from the receiver position to the observation direction, and the spatial region traversed by the directional ray is discretized into multiple voxel points to obtain the voxel position of each voxel point. The scene data, the transmitter position, the observation direction, the voxel position, and the beam codebook information are input into the codebook-sensing neural radiation field model, and the propagation attributes of each voxel point are output. The propagation properties of each voxel point are rendered and aggregated along the directional ray to obtain the predicted field strength value of the receiving end position under different beam conditions.

2. The adaptive field strength prediction method based on beam direction according to claim 1, characterized in that, The propagation properties include the attenuation properties and re-radiation contribution of the voxel point; wherein, the attenuation properties characterize the propagation loss of the signal when it propagates through the current voxel point, and the re-radiation contribution characterizes the degree of contribution of the voxel point to the received signal strength under the current observation direction and beam conditions.

3. The adaptive field strength prediction method based on beam direction according to claim 2, characterized in that, The process of rendering and aggregating the propagation properties of each voxel point along the ray along the stated direction to obtain the predicted field strength value of the receiver location under different beam conditions includes: For each voxel point along the observation direction and the directional ray, the attenuation factors of each voxel point between the receiving end position and the current voxel point are multiplied together to obtain the cumulative attenuation coefficient from the current voxel point to the receiving end position. The cumulative attenuation coefficient is then multiplied by the re-radiation contribution of the current voxel point to obtain the effective signal contribution of the current voxel point. The effective signal contribution of each voxel point is weighted and summed according to the radial sampling interval to obtain the directional signal component corresponding to the observation direction. The received power corresponding to each observation direction is obtained by taking the square of the modulus of each of the signal components in each direction, and the received power is summed in the same linear power domain to obtain the predicted field strength of the receiving end position under the current beam conditions. The above processing is performed for different beam conditions to obtain the predicted field strength values ​​of the receiving end position under different beam conditions.

4. The adaptive field strength prediction method based on beam direction according to claim 1, characterized in that, After obtaining the predicted field strength values ​​for the receiver location under different beam conditions, the method further includes: The predicted field strength values ​​are used to train and optimize the codebook-sensing neural radiation field model. The trained codebook-based neural radiation field model is used to predict the field strength, obtain the predicted field strength value under the corresponding beam conditions, and generate the field strength distribution map of the target area based on multiple predicted field strength values.

5. The adaptive field strength prediction method based on beam direction according to claim 4, characterized in that, The step of training and optimizing the codebook-sensing neural radiation field model using the predicted field strength includes: Obtain the actual measured field strength value corresponding to the predicted field strength value; Expand and concatenate the real and imaginary parts of the complex beam vector corresponding to the current beam to obtain the real-valued beam feature vector; Based on the attenuation attribute prediction branch and re-radiation contribution prediction branch of the codebook-sensory neural radiation field model, the real-valued beam feature vector is concatenated with the intermediate features of the corresponding branch to obtain joint features. The joint features are input into a fully connected mapping network for nonlinear mapping to generate scaling and offset vectors corresponding to the channel dimensions of the intermediate features. The scaling vector is multiplied element-wise with the intermediate feature, and the multiplication result is added element-wise with the offset vector to obtain the fused feature; Based on the fused features, the network parameters of the codebook-perceptual neural radiation field model are adjusted, and a loss function is constructed between the predicted field strength value and the actual measured field strength value. The network parameters are iteratively optimized based on the loss function until the preset convergence condition is met, thereby reducing the difference between the predicted field strength value and the actual measured field strength value.

6. The adaptive field strength prediction method based on beam direction according to claim 4, characterized in that, The step involves using the trained codebook-based perceptual neural radiation field model to predict the field strength, obtaining predicted field strength values ​​under corresponding beam conditions, and generating a field strength distribution map of the target region based on multiple predicted field strength values, including: Obtain the locations of multiple receivers to be predicted and target beam information within the target area; The scene data, the transmitter position, the receiver positions, the preset observation direction, and the target beam information are input into the trained codebook-aware neural radiation field model, and the propagation attributes of each voxel are output. Based on the propagation attributes, rendering aggregation processing is performed to obtain the field strength prediction value of each receiver position under the target beam condition. Based on the predicted field strength values, a field strength distribution map of the target region under the target beam condition is generated.

7. The adaptive field strength prediction method based on beam direction according to claim 6, characterized in that, Before acquiring the multiple receiver locations to be predicted and the target beam information within the target area, the method further includes: The target area is divided into key areas and non-key areas, and the key areas are those where the field strength prediction accuracy is higher than a preset threshold. Based on the key area, each beam in the beam codebook information is filtered to obtain a filtered beam set. The filtered beam set is used as the target beam information.

8. The adaptive field strength prediction method based on beam direction according to claim 1, characterized in that, Discretize the spatial region traversed by the directional ray into multiple voxel points to obtain the voxel positions of each voxel point, including: Based on the receiver position, the observation direction, and the radial distance along the observation direction, multiple voxel points are obtained, and based on the multiple voxel points, the voxel position of each voxel point is obtained, wherein the voxel points are used to characterize the spatial contribution points of the signal during the ray propagation along the direction to the receiver position.

9. The adaptive field strength prediction method based on beam direction according to claim 1, characterized in that, The acquisition of scene data of the target area and determination of the transmitter location, receiver location, and beam codebook information include: Construct scene data for the target area, wherein the scene data is used to characterize environmental information affecting wireless propagation within the target area; Based on the scene data, determine the location of the transmitter and multiple receivers within the target area; Based on the transmitter location, beam codebook information corresponding to the transmitter location is determined. The beam codebook information is used to characterize the beam vector or codeword corresponding to different beam directions.

10. An adaptive field strength prediction system based on beam direction, characterized in that, The system is used to perform the method according to any one of claims 1 to 9, and the system includes a basic data acquisition module, a voxel location acquisition module, a propagation attribute acquisition module, and a field strength prediction value acquisition module, wherein: The basic data acquisition module is used to acquire scene data of the target area and determine the transmitter location, receiver location, and beam codebook information. The voxel position acquisition module is used to define a directional ray from the receiving end position to the observation direction based on the receiving end position and the preset observation direction, and to discretize the spatial region traversed by the directional ray into multiple voxel points to obtain the voxel position of each voxel point. The propagation attribute acquisition module is used to input the scene data, the transmitter position, the observation direction, the voxel position, and the beam codebook information into the codebook perception neural radiation field model, and output the propagation attributes of each voxel point; The field strength prediction module is used to render and aggregate the propagation properties of each voxel point along the directional ray to obtain the field strength prediction value of the receiving end position under different beam conditions.

Images