Optimal handover beam pair search method and apparatus

Abstract

The application discloses an optimal switching beam pair search method and device. The method comprises the following steps: obtaining target beam pair information of a plurality of beam pairs between a base station and a target unmanned aerial vehicle in the coverage range of the base station at a first time; analyzing the target beam pair information by using a search range prediction model to obtain a corresponding target optimal beam pair search range, wherein the search range prediction model is constructed based on an improved deep residual network; and determining an optimal switching beam pair between the base station and the unmanned aerial vehicle at a second time from the target optimal beam pair search range by using a heuristic algorithm, wherein the second time is the next time of the first time. The application solves the technical problem that the communication between the base station and the unmanned aerial vehicle is delayed or interrupted due to the fact that the related art cannot quickly find the optimal switching beam pair in a plurality of candidate beam pairs for the communication between the base station and the unmanned aerial vehicle.

Description

Technical Field

[0001] This application relates to the field of wireless communication technology, and more specifically, to an optimal switching beam pair search method and apparatus. Background Technology

[0002] In low-altitude communication scenarios such as drone inspections and live streaming, images need to be transmitted in real time, which places high demands on the communication speed and stability of the airspace network. However, due to the high speed of drones, base stations need to update their beams in real time to ensure high-speed and stable communication with drones and avoid communication delays or interruptions caused by high latency. Therefore, how to quickly switch to the optimal beam when the base station communicates with the drone is a problem that urgently needs to be solved.

[0003] Currently, base station deployment frequency bands suffer from high path loss, and millimeter-wave frequency band sites have already been deployed. In the post-5G era, AAUs will need to rely on ultra-large-scale antenna arrays to obtain beamforming gain to counteract path loss. The more antennas there are, the more numerous and narrower the beams become, requiring more time to search for the optimal beam pair to maintain stable communication with high speed and low latency. Therefore, existing solutions cannot meet the requirements for rapid beam pair switching between base stations and drones. Summary of the Invention

[0004] This application provides an optimal switching beam pair search method and apparatus to at least solve the technical problem that communication between the base station and the drone is delayed or interrupted because related technologies have difficulty in quickly finding the optimal switching beam pair among multiple candidate beam pairs for communication between the base station and the drone.

[0005] According to one aspect of the embodiments of this application, an optimal switching beampair search method is provided, comprising: acquiring a target beampair information map of multiple beampairs between a base station and a target UAV within the coverage area of ​​the base station at a first time moment; analyzing the target beampair information map using a pre-trained search range prediction model to obtain the corresponding target optimal beampair search range, wherein the search range prediction model is constructed based on an improved deep residual network, and the improved deep residual network includes at least: a shallow feature extraction module, a deep feature extraction module based on a residual nested structure, and a prediction output module; and determining the optimal switching beampair between the base station and the UAV at a second time moment from the target optimal beampair search range using a heuristic algorithm, wherein the second time moment is the time moment following the first time moment.

[0006] Optionally, acquiring a target beam pair information map of multiple beam pairs between a base station and a target UAV within the base station's coverage area at a first moment includes: acquiring communication quality parameters of each of the multiple beam pairs between the base station and the target UAV at the first moment, wherein the communication quality parameters include at least one of the following: spectral efficiency and signal-to-dryness ratio; acquiring first communication status information of the base station at the first moment and second communication status information of the target UAV at the first moment, wherein the first communication status information includes at least: first beam status information of each of the multiple beams of the base station and base station status information, and the second communication status information includes at least: second beam status information of each of the multiple beams of the target UAV and UAV status. The information includes, among other things, the first and second beam status information, which at least include: beam azimuth, beam elevation, beamwidth, and beam gain. For each beam pair, the first and second beam status information of each of the two beams within the beam pair, the communication quality parameters of the beam pair, the base station status information, and the UAV status information are used as multiple features of the beam pair. These features form the feature vector of the beam pair, which is then standardized. The multiple beams of the base station and the multiple beams of the target UAV are arranged in ascending order of beam number, forming multiple matrix rows and columns. The standardized feature vectors of each beam pair are used as matrix elements to construct the target beam pair information map.

[0007] Optionally, the training process of the search range prediction model includes: constructing a deep learning model including a shallow feature extraction module, a deep feature extraction module, and a prediction output module; obtaining a training sample set and a sample label set, wherein the training sample set includes multiple beampuppet information maps between the base station and the UAV as training samples, and the sample label set includes the true optimal beampuppet search range corresponding to each beampuppet information map as a sample label; and iteratively training the deep learning model using the training sample set and the sample label set to obtain the search range prediction model.

[0008] Optionally, the shallow feature extraction module includes at least one convolutional layer and is used to extract low-level image features of the beampuppet information map; the deep feature extraction module includes multiple residual blocks connected by short skips, each residual block including multiple convolutional layers, batch normalization layers and activation function layers connected by residuals, and is used to extract high-level image features based on low-level image features; the prediction output module includes two fully connected layers and a Dropout layer, and is used to convert high-level image features into a prediction search range.

[0009] Optionally, the deep learning model is iteratively trained using the training sample set and the sample label set to obtain a search range prediction model. This includes: for each training batch in the iterative training process, inputting each training sample of the training batch into the deep learning model to obtain the search range of each predicted optimal beam pair output by the deep learning model; constructing a target loss function using the search range of the predicted optimal beam pair and the corresponding sample label; and adjusting the model parameters of the deep learning model according to the target loss function until the model parameters converge to obtain the search range prediction model.

[0010] Optionally, the target loss function includes: an angle loss function, a search radius loss function, and a physical constraint loss function, and the expression for the target loss function is: In the formula, Represents the angle loss function. This represents the search radius loss function. Represents the physical constraint loss function. , , These represent the adjustment coefficients for the angle loss function, search radius loss function, and physical constraint loss function, respectively, where: the angle loss function The expression is: In the formula, N represents the total number of training samples in the training batch. This represents the true beam azimuth angle of the i-th training sample. This represents the predicted beam azimuth angle of the i-th training sample. This represents the true beam elevation angle of the i-th training sample. This represents the predicted beam elevation angle of the i-th training sample. This indicates the preset angle threshold. These represent the weighting coefficients and the search radius loss function, respectively. The expression is: In the formula, N represents the total number of training samples in the training batch. This represents the true search radius of the i-th training sample. This represents the predicted search radius of the i-th training sample. This indicates the preset maximum allowed search range. Represents the adjustment coefficient; physical constraint loss function The expression is: In the formula, Let represent the beam directional gradient constraint loss function, and let represent the beam directional gradient constraint loss function. It is obtained by calculating the beam pattern function of the base station. The energy coverage constraint loss function is defined such that when the energy coverage of the predicted optimal beam pair corresponding to the training sample is less than a preset energy threshold, the energy coverage constraint loss function... The value of increases, and when the energy coverage of the predicted optimal beam pair search range corresponding to the training sample is not less than the energy threshold, the energy coverage constraint loss function... The value decreases.

[0011] Optionally, the target optimal beam pair search range is determined by the center beam pair and the search radius. The optimal switching beam pair between the base station and the UAV at the second moment is determined from the target optimal beam pair search range using a heuristic algorithm, including initializing the temperature. and cooling coefficient In the first round of search, the center beam pair is taken as the current optimal beam pair in the current search round. Within the search range of the target optimal beam pair, multiple first neighboring beam pairs adjacent to the current optimal beam pair are determined, and the spectral efficiency change rate of each first neighboring beam pair compared to the current optimal beam pair is calculated according to the following formula: In the formula, Indicates the current optimal beam pair. Indicates the first neighboring beam pair, and When the rate of change of the spectral efficiency of the first target neighboring beam pair compared to the current optimal beam pair exceeds a preset first threshold, the transfer probability of the first target neighboring beam is calculated according to the following formula: In the formula, Indicates the predicted search radius. Indicates the maximum allowed search radius. This represents the adjustment coefficient; when the transfer probability of the beam near the first target is higher than the preset second threshold, the beam near the first target is taken as the optimal beam pair in the next search round, and the temperature of the current search round is updated according to the following formula: In the second and subsequent search rounds, the neighboring beam pairs switched in the previous search round are taken as the current optimal beam pairs in the current search round. Multiple second neighboring beam pairs adjacent to the current optimal beam pair are re-determined within the search range of the target optimal beam pair, and the rate of change of spectral efficiency of each second neighboring beam pair compared to the current optimal beam pair is determined. When the rate of change of spectral efficiency of the second target neighboring beam pair compared to the current optimal beam pair is higher than a first threshold, the transfer probability of the second target neighboring beam is calculated according to the following formula: In the formula, This represents the rate of change of the spectral efficiency of the beam adjacent to the second target. This represents the rate of change of the spectral efficiency of the current optimal beam pair in the current search round. Let represent the temperature in the k-th round, and , K represents the maximum number of iterations for the search. When the transfer probability of the second target's neighboring beam is higher than the second threshold, the second target's neighboring beam is taken as the optimal beam pair in the next search process, and the temperature of the current search round is updated. The above steps are repeated, and the optimal beam pair of the target search round with a temperature lower than the preset third threshold is taken as the optimal switching beam pair between the base station and the UAV at the second moment.

[0012] According to another aspect of the embodiments of this application, an optimal switching beampair search device is also provided, comprising: an acquisition module, configured to acquire a target beampair information map of multiple beampairs between a base station and a target UAV within the coverage area of ​​the base station at a first time moment; a prediction module, configured to analyze the target beampair information map using a pre-trained search range prediction model to obtain the corresponding target optimal beampair search range, wherein the search range prediction model is constructed based on an improved deep residual network, and the improved deep residual network includes at least: a shallow feature extraction module, a deep feature extraction module based on a residual nesting structure, and a prediction output module; and a search module, configured to determine the optimal switching beampair between the base station and the UAV at a second time moment from the target optimal beampair search range using a heuristic algorithm, wherein the second time moment is the time moment following the first time moment.

[0013] According to another aspect of the embodiments of this application, a computer program product is also provided, the computer program product comprising: a computer program, wherein the computer program, when executed by a processor, implements the above-described optimal switching beam pair search method.

[0014] According to another aspect of the embodiments of this application, an electronic device is also provided, the electronic device including: a memory and a processor, wherein the memory stores a computer program, and the processor is configured to execute the above-described optimal switching beam pair search method through the computer program.

[0015] In this embodiment, a target beam pair information map of multiple beam pairs between a base station and a target UAV within the base station's coverage area is obtained at a first moment. A search range prediction model based on an improved deep residual network is then used to analyze the target beam pair information map, obtaining the corresponding optimal target beam pair search range. This significantly reduces the search space and improves search efficiency. A heuristic algorithm is used to determine the optimal switching beam pair between the base station and the UAV at a second moment from within the optimal target beam pair search range, ensuring continuous high-quality communication and low latency. This achieves efficient and stable communication, thus solving the technical problem of communication delays or interruptions between the base station and the UAV due to the difficulty of quickly finding the optimal switching beam pair among multiple candidate beam pairs in base station-UAV communication using related technologies. Attached Figure Description

[0016] 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:

[0017] Figure 1 This is a schematic diagram of beampair gain under an optional line-of-sight communication method according to an embodiment of this application;

[0018] Figure 2 This is a schematic diagram of beam pair gain under an optional non-line-of-sight communication method according to an embodiment of this application;

[0019] Figure 3 This is a flowchart illustrating an optional optimal switching beam pair search method according to an embodiment of this application;

[0020] Figure 4 This is a schematic diagram of an optional optimal beam pair search range according to an embodiment of this application;

[0021] Figure 5 This is a schematic diagram illustrating an optional comparison of search counts according to an embodiment of this application;

[0022] Figure 6 This is a schematic diagram of an optional optimal switching beam pair search device according to an embodiment of this application;

[0023] Figure 7 This is a schematic diagram of an optional electronic device according to an embodiment of this application. Detailed Implementation

[0024] To enable those skilled in the art to better understand the present application, the technical solutions in the embodiments of the present application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present application, and not all embodiments. Based on the embodiments in the present application, all other embodiments obtained by those of ordinary skill in the art without creative effort should fall within the scope of protection of the present application.

[0025] It should be noted that the terms "first," "second," etc., used in the specification, claims, and drawings of this application are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate so that the embodiments of this application described herein can be implemented in orders other than those illustrated or described herein. Furthermore, the terms "comprising" and "having," and any variations thereof, are intended to cover non-exclusive inclusion; for example, a process, method, system, product, or apparatus that comprises a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units not explicitly listed or inherent to such processes, methods, products, or apparatus.

[0026] To better understand the embodiments of this application, the following is a translation and explanation of some nouns or terms that appear in the description of the embodiments of this application:

[0027] Line-of-sight (LOS) communication refers to the transmission of signals between transmitting and receiving antennas at a distance where they can see each other; that is, radio waves propagate directly from the transmitting point to the receiving point (generally including reflected waves from the ground). Typically, the distance for line-of-sight propagation is 20–50 km.

[0028] Non-line-of-sight (NLOS) communication refers to non-direct point-to-point communication between a receiver and a transmitter. The most direct explanation is that the line of sight between the two points is blocked, and they cannot see each other.

[0029] Spectral efficiency, also known as system capacity or bandwidth utilization, is defined as the ratio of the effective information rate transmitted by a system to the bandwidth of the communication channel. In other words, it represents the number of bits that can be transmitted per second on a given bandwidth channel, indicating the system's utilization rate of spectrum resources.

[0030] Signal-to-Interference-plus-Noise Ratio (SINR): Also known as the signal-to-dryness ratio, SINR is the ratio of the strength of the received useful signal to the strength of the received dry signal (including noise and interference). It measures the signal strength relative to the strength of the dry source. Therefore, a higher SINR indicates better communication quality. Improving the SINR is thus one of the main tasks for enhancing communication quality.

[0031] Simulated Annealing (SA) is a probabilistic global optimization heuristic algorithm inspired by the annealing process of solid materials. In physical annealing, the material is first heated to a high temperature, causing the internal particles to be in a highly disordered state. It is then slowly cooled; as the temperature decreases, the particle bonds tend to align in an ordered manner, eventually reaching the lowest energy (ground state) at room temperature. The simulated annealing algorithm abstracts this physical process into a strategy for solving an optimization problem.

[0032] Deep Residual Networks (ResNet) address the vanishing gradient problem in traditional deep neural networks by introducing residual learning and a special network structure, enabling efficient and scalable deep models. A Deep Residual Network consists of: an initial convolutional layer, a group of residual blocks, full average pooling, and fully connected layers. Each group of residual blocks contains multiple residual blocks, which are the basic building blocks of ResNet. Through these residual blocks, ResNet effectively solves the vanishing gradient problem and allows for the training of extremely deep networks.

[0033] Example 1

[0034] The base station and the drone mainly use line-of-sight communication. However, there are many high-rise residential areas in densely populated urban areas, and the airspace stations are at different heights. Therefore, in some scenarios, the base station and the drone will use non-line-of-sight communication. Figure 1 , Figure 2 The graphs show the gain of communication between the base station and the drone using different beams in line-of-sight (LAS) and non-LAS scenarios, respectively. Lighter colors represent higher communication speeds. Therefore, it's easy to see that... Figure 1 In the line-of-sight scenario shown, the light-colored area is relatively small, and there is only one beam pair with the highest gain. Although other beam pairs near this beam pair can also communicate, the communication rates differ significantly. Figure 2 In the non-line-of-sight scene shown, there are relatively more light-colored areas and multiple beam pairs with similar communication rates, meaning there are multiple high-rate alternative beams.

[0035] Therefore, in low-altitude communication, there are many candidate beams for communication between the base station and the UAV, and the beam search range is large. However, traditional beam switching methods are difficult to quickly find the optimal beam pair for switching within a large search range, which leads to lag or even interruption in the communication process between the base station and the UAV.

[0036] To address the aforementioned problems, embodiments of this application provide an optimal switching beam pair search method to solve the problem of low beam search efficiency caused by a large beam search range. It should be noted that the steps shown in the flowcharts of the accompanying drawings can be executed in a computer system such as a set of computer-executable instructions, and although a logical order is shown in the flowcharts, in some cases, the steps shown or described may be executed in a different order than that shown here.

[0037] Figure 3 This is a flowchart illustrating an optimal switching beam pair search method according to an embodiment of this application, as shown below. Figure 3 As shown, the method includes the following steps:

[0038] Step S302: Obtain target beam pair information map of multiple beam pairs between the base station and the target UAV within the coverage area of ​​the base station at the first moment.

[0039] Step S304: Analyze the target beam pair information map using the pre-trained search range prediction model to obtain the corresponding optimal target beam pair search range. The search range prediction model is constructed based on an improved deep residual network, and the improved deep residual network includes at least: a shallow feature extraction module, a deep feature extraction module based on a residual nesting structure, and a prediction output module.

[0040] Step S306: Use a heuristic algorithm to determine the optimal switching beam pair between the base station and the UAV at the second time point from the target optimal beam pair search range, where the second time point is the time point following the first time point.

[0041] Based on the scheme defined in steps S302 to S306 above, it can be seen that in this embodiment, the search range prediction model constructed using the improved deep residual network is used to analyze the target beam pair information map, thereby obtaining the search range of the optimal target beam pair where the optimal switching beam pair between the search base station and the UAV is located. This significantly reduces the search space and improves search efficiency. Subsequently, a heuristic algorithm is used to further refine the optimal switching beam pair within the predicted search range, ensuring continuous high-quality communication and low latency. This achieves the technical effect of efficient and stable communication, solving the challenges faced by traditional beam switching methods in high-speed UAV movement. Especially in dense urban non-line-of-sight communication environments, it can significantly improve communication rate and stability, meeting the timeliness requirements of emergency response and resource scheduling.

[0042] The following section explains each step of the optimal switching beam pair search method in conjunction with a specific implementation process.

[0043] As an optional implementation, in the technical solution provided in step S302 above, the target beam pair information map can be obtained according to the following steps:

[0044] Step 1: Obtain the communication quality parameters of multiple beam pairs between the base station and the target UAV at the first moment.

[0045] Among them, communication quality parameters include at least one of the following: spectral efficiency, gain information, signal-to-dryness ratio, etc.

[0046] Step 2: Obtain the first communication status information of the base station at the first moment and the second communication status information of the target UAV at the first moment.

[0047] The first communication status information includes at least the following: the first beam status information for each of the base station's multiple beams, such as beam azimuth (determined by mechanical and electronic azimuth), beam downtilt (determined by mechanical and electronic downtilt), beam width, beam height, beam gain, and beam activation time. In addition, the first communication status information also includes status information such as base station altitude. These pieces of information collectively describe the spatial pointing, energy characteristics, and lifetime of the base station beams.

[0048] The second communication status information includes at least the second beam status information of each of the target UAV's multiple beams, such as beam azimuth and beam downtilt angles. In addition, the second communication status information also includes the target UAV's latitude and longitude coordinates, three-dimensional velocity components, three-dimensional acceleration components, pitch angle, roll angle, dwell time of the target UAV at the base station, and dwell time of the target UAV under each of the base station's transmitted beams. This information reflects the target UAV's precise position, motion state, and the continuity of its communication with the base station.

[0049] Step 3: For each beam pair, the first beam state information, the second beam state information, the communication quality parameters of the beam pair, the base station state information, and the UAV state information of each of the two beams in the beam pair are taken as multiple features of the beam pair. The feature vector of the beam pair is composed of multiple features and the feature vector is standardized.

[0050] The purpose of standardization is to eliminate the influence of feature scale, ensuring that different features are on the same order of magnitude. Standardization typically includes mean removal and scaling: subtracting the mean from each feature and dividing by its standard deviation to ensure that all features have a mean of 0 and a variance of 1. This ensures that even if features originally had different value ranges, they can participate fairly in the subsequent machine learning process.

[0051] Step 4: Arrange the multiple beams of the base station and the multiple beams of the target UAV in ascending order of beam number, as multiple matrix rows and multiple matrix columns, and use the standardized feature vectors of each beam pair as matrix elements to construct the target beam pair information map.

[0052] This can be understood as follows: the beam numbers of the base station beams and the target UAV beams are arranged in ascending order and used as the horizontal and vertical coordinates to construct a two-dimensional grid. The standardized feature vectors are then mapped onto the corresponding grid points as the information of the grid points, thereby obtaining the target beam pair information map.

[0053] The target beam pair information map mentioned above can be in the following form:

[0054] ,

[0055] In the formula, Indicates the number of beams in the base station. Indicates the number of beams of the target drone. Indicates the first The base station beam and the first The feature vector (also known as "beam pair information block") of a beam pair composed of UAV beams.

[0056] Furthermore, the obtained target beam pair information map is input into a pre-trained search range prediction model for analysis to obtain the corresponding optimal target beam pair search range.

[0057] As an optional implementation, the training process of the above-mentioned search range prediction model may include:

[0058] Step 1: Construct a deep learning model that includes a shallow feature extraction module, a deep feature extraction module, and a prediction output module;

[0059] Step 2: Obtain the training sample set and the sample label set. The training sample set includes multiple beam pair information maps between the base station and the UAV as training samples. The sample label set includes the true optimal beam pair search range corresponding to each beam pair information map as the sample label.

[0060] Step 3: Iteratively train the deep learning model using the training sample set and sample label set to obtain the search range prediction model.

[0061] In the search range prediction model, the shallow feature extraction module typically includes several convolutional layers (e.g., the shallow feature extraction module includes a convolutional layer with a kernel size of 7*7 and a stride of 2) to extract low-level and local features of the beampair information map.

[0062] Deep feature extraction modules typically consist of multiple residual blocks connected by short skip connections. Each residual block includes multiple convolutional layers, batch normalization layers, and activation function layers (such as ReLU) connected by residuals, used to effectively extract high-level, abstract features from low-level image features. The output of each residual block is the "residual" of the input signal added to the convolutionally processed signal. This avoids the gradient vanishing and degradation problems that occur with increasing network depth, allowing the model to learn more complex feature representations.

[0063] The prediction output module predicts the search range of the optimal handover beam pair between the base station and the UAV based on the high-level and low-frequency features of the image output from the two modules mentioned above. This module consists of two fully connected layers and a Dropout layer, connected in the following order: first fully connected layer, Dropout layer, and second fully connected layer. The first fully connected layer converts the high-level image features into a low-dimensional prediction output, while the second fully connected layer converts the features output from the Dropout layer into the prediction output, thus obtaining the prediction search range.

[0064] In the specific training process, the deep learning model can be iteratively trained using the training sample set and sample label set to obtain the search range prediction model: For each training batch in the iterative training process, each training sample of the training batch is input into the deep learning model to obtain the search range of each predicted optimal beam pair output by the deep learning model. The target loss function is constructed using the search range of the predicted optimal beam pair and the corresponding sample label, and the model parameters of the deep learning model are adjusted according to the target loss function until the model parameters converge to obtain the search range prediction model.

[0065] The aforementioned objective loss function includes: angle loss function, search radius loss function, and physical constraint loss function. Therefore, the expression for the aforementioned objective loss function is: In the formula, Represents the angle loss function. This represents the search radius loss function. Represents the physical constraint loss function. , , These represent the adjustment coefficients for the angle loss function, search radius loss function, and physical constraint loss function, respectively. Where:

[0066] The above angle loss function The expression is:

[0067] ,

[0068] In the formula, N represents the total number of training samples in the training batch. This represents the true beam azimuth angle of the i-th training sample. This represents the predicted beam azimuth angle of the i-th training sample. This represents the true beam elevation angle of the i-th training sample. This represents the predicted beam elevation angle of the i-th training sample. This indicates the preset angle threshold. These represent the weighting coefficients.

[0069] Search radius loss function An asymmetric penalty function is used to encourage predicted radii to be slightly larger than the true value, thus avoiding missed detections. Therefore, a search radius loss function is employed. The expression can be written as:

[0070] ,

[0071] In the formula, N represents the total number of training samples in the training batch. This represents the true search radius of the i-th training sample. This represents the predicted search radius of the i-th training sample. This indicates the preset maximum allowed search range. This represents the adjustment coefficient.

[0072] In addition, physical constraint loss function The beam direction gradient constraint function and energy coverage constraint function are introduced, therefore, By introducing two physical constraints, prior knowledge such as electromagnetic wave propagation laws and antenna characteristics is embedded into the neural network, improving the model's rationality and generalization ability. Specifically:

[0073] Regarding the beam direction gradient constraint loss function It can directly target the base station beam pattern function. The derivative (gradient) of the base station beam pattern function describes the intensity distribution of the base station's transmitted signal in different spatial directions, while its derivative (gradient) reflects the rate at which this intensity distribution changes with direction. Therefore, this embodiment of the application, by differentiating the base station's beam pattern function in a certain direction (such as azimuth and elevation), can obtain the rate of change of beam intensity in that direction, allowing the setting of a reasonable gradient threshold as a beam direction constraint. Subsequently, if the gradient value of a beam pair exceeds the set threshold, it means that the directionality of that beam pair changes too rapidly, thus hindering stable communication; conversely, if the gradient value of a beam pair is small, it indicates that the beam direction change of that beam pair is gradual, resulting in better communication stability.

[0074] Regarding the energy cover constraint loss function The goal is to ensure that the predicted search radius contains sufficient energy to avoid missing beams. Therefore, the energy coverage of the predicted optimal beam corresponding to the training samples within the search range is less than a preset energy threshold. At that time, the energy coverage constraint loss function The value increases; while the energy coverage of the predicted optimal beam pair search range corresponding to the training samples is not less than the preset energy threshold. At that time, the energy coverage constraint loss function The value decreases. Among them, the search range of the predicted optimal beam pair corresponding to the training samples. The expression for energy coverage can be written as:

[0075] .

[0076] Therefore, this embodiment utilizes the search range prediction model trained above to perform multi-level, multi-view feature extraction and analysis on the target beam information map, accurately predicting the search range of the optimal target beam pair for beam switching at the next moment. This prediction process combines the advantages of image recognition and the powerful functions of deep learning, enabling efficient screening and prediction of the search range in complex environments. It significantly reduces the need to traverse all possible beam pairs, thereby greatly reducing the search time for the optimal switching beam pair between the search base station and the UAV, and improving the stability and efficiency of communication.

[0077] Finally, since the search range of the target optimal beam pair predicted by the model is determined by the center beam pair and the search radius, and the search range of the target optimal beam pair includes multiple candidate beam pairs, such as... Figure 4 As shown. Therefore, embodiments of this application can utilize heuristic algorithms to iteratively search for the optimal switching beam pair between the base station and the UAV at the second time step within the target optimal beam pair search range. The heuristic algorithms include: particle swarm optimization algorithm and simulated annealing algorithm.

[0078] The above search process will be explained below using the simulated annealing algorithm as an example.

[0079] (a) Initialization: Setting the temperature and cooling coefficient , where the initial temperature The initial temperature should be set high enough to accept more inferior solutions in the early stages of the search, increasing the randomness of the search. The initial temperature setting usually depends on the size and complexity of the problem.

[0080] (ii) In the first round of search, the center beam pair (i.e. Figure 4The cross points within the target optimal beam pair are used as the current optimal beam pair in the current search round. Multiple first neighboring beam pairs adjacent to the current optimal beam pair are determined within the search range of the target optimal beam pair, and the rate of change of spectral efficiency of each first neighboring beam pair compared to the current optimal beam pair is calculated according to the following formula:

[0081] ,

[0082] In the formula, Indicates the current optimal beam pair. Indicates the first neighboring beam pair, and .

[0083] When the rate of change of the spectral efficiency of the first target neighboring beam pair within multiple first neighboring beams compared to the current optimal beam pair exceeds a preset first threshold, the transfer probability of the first target neighboring beam can be calculated according to the following formula:

[0084] ,

[0085] In the formula, Indicates the predicted search radius. Indicates the maximum allowed search radius. This represents the adjustment coefficient.

[0086] When the transfer probability of the beam near the first target is higher than a preset second threshold, the beam near the first target is selected as the optimal beam pair in the next round of search, and the temperature is updated according to the following formula: .

[0087] (iii) In the second round and each subsequent search round, the neighboring beam pair switched in the previous search round is taken as the current optimal beam pair in the current search round. Within the search range of the target optimal beam pair, multiple second neighboring beam pairs adjacent to the current optimal beam pair are re-determined, and the spectral efficiency change rate of each second neighboring beam pair compared to the current optimal beam pair is determined.

[0088] When the rate of change of the spectral efficiency of the second target neighboring beam pair compared to the current optimal beam pair is higher than a first threshold, the transfer probability of the second target neighboring beam is calculated according to the following formula:

[0089] ,

[0090] In the formula, This represents the rate of change of the spectral efficiency of the beam adjacent to the second target. This represents the rate of change of the spectral efficiency of the current optimal beam pair in the current search round. Let represent the temperature in the k-th round, and , K represents the maximum number of iterations.

[0091] When the transfer probability of the second target's neighboring beam is higher than the second threshold, the second target's neighboring beam is selected as the optimal beam pair for the next search round, and the temperature of the current search round is updated according to the following formula: .

[0092] (iv) Repeat step (iii) above, and use the optimal beam pair of the target search wheel with a temperature lower than the preset third threshold (e.g., close to 0, i.e., the algorithm stops accepting inferior solutions) as the optimal switching beam pair between the base station and the UAV at the second moment.

[0093] Therefore, the embodiments of this application, through the aforementioned heuristic search method, can dynamically adjust the search strategy in a constantly changing communication environment, ensuring that the focus is always on the most promising beam pairs, rather than blindly searching the entire beam space. This search scheme can adapt to the communication needs of different scenarios while maintaining high efficiency.

[0094] Compared to traditional traversal search methods, the heuristic search method described above can quickly locate the optimal switching beam pair while significantly reducing the number of searches, thereby significantly reducing communication system latency and improving communication efficiency. For example, if a base station has 128 antennas and 2 RF links, using a fully connected structure; and a drone within the base station's coverage area has 16 antennas and 2 links, also using a fully connected structure, then the comparison of the number of searches required to find the optimal switching beam pair between the base station and the drone using traditional traversal search methods and heuristic search algorithms respectively is as follows: Figure 5 As shown, it is easy to see that the number of searches required to obtain the optimal switching beam pair using the traditional traversal search method is 4,194,304, while the number of searches required to obtain the optimal switching beam pair using the heuristic search algorithm provided in this application embodiment is 32. Therefore, the complexity of the heuristic search algorithm provided in this application embodiment is reduced by 99.999%.

[0095] Example 2

[0096] According to an embodiment of this application, an optimal switching beampair search device is also provided for implementing the optimal switching beampair search method in Embodiment 1, such as... Figure 6 As shown, the optimal switching beam pair search device includes at least: an acquisition module 62, a prediction module 64, and a search module 66, wherein:

[0097] The acquisition module 62 is used to acquire a target beam pair information map of multiple beam pairs between the base station and the target UAV within the coverage area of ​​the base station at the first moment.

[0098] The prediction module 64 is used to analyze the target beam pair information map using a pre-trained search range prediction model to obtain the corresponding target optimal beam pair search range. The search range prediction model is constructed based on an improved deep residual network, and the improved deep residual network includes at least: a shallow feature extraction module, a deep feature extraction module based on a residual nesting structure, and a prediction output module.

[0099] Search module 66 is used to determine the optimal switching beam pair between the base station and the UAV at the second time point from the target optimal beam pair search range using a heuristic algorithm, wherein the second time point is the time point following the first time point.

[0100] The following section describes the functions of each module of the optimal switching beam search device in conjunction with the specific implementation process.

[0101] As an optional implementation, the acquisition module 62 may acquire the target beam pair information map by following the steps below:

[0102] Step 1: Obtain the communication quality parameters of multiple beam pairs between the base station and the target UAV at the first moment.

[0103] Among them, communication quality parameters include at least one of the following: spectral efficiency, gain information, signal-to-dryness ratio, etc.

[0104] Step 2: Obtain the first communication status information of the base station at the first moment and the second communication status information of the target UAV at the first moment.

[0105] The first communication status information includes at least the following: the first beam status information for each of the base station's multiple beams, such as beam azimuth (determined by mechanical and electronic azimuth), beam downtilt (determined by mechanical and electronic downtilt), beam width, beam height, beam gain, and beam activation time. In addition, the first communication status information also includes status information such as base station altitude. These pieces of information collectively describe the spatial pointing, energy characteristics, and lifetime of the base station beams.

[0106] The second communication status information includes at least the second beam status information of each of the target UAV's multiple beams, such as beam azimuth and beam downtilt angles. In addition, the second communication status information also includes the target UAV's latitude and longitude coordinates, three-dimensional velocity components, three-dimensional acceleration components, pitch angle, roll angle, dwell time of the target UAV at the base station, and dwell time of the target UAV under each of the base station's transmitted beams. This information reflects the target UAV's precise position, motion state, and the continuity of its communication with the base station.

[0107] Step 3: For each beam pair, the first beam state information, the second beam state information, the communication quality parameters of the beam pair, the base station state information, and the UAV state information of each of the two beams in the beam pair are taken as multiple features of the beam pair. The feature vector of the beam pair is composed of multiple features and the feature vector is standardized.

[0108] Step 4: Arrange the multiple beams of the base station and the multiple beams of the target UAV in ascending order of beam number, as multiple matrix rows and multiple matrix columns, and use the standardized feature vectors of each beam pair as matrix elements to construct the target beam pair information map.

[0109] Furthermore, the prediction module 64 can use a pre-trained search range prediction model to analyze the target beam pair information map and obtain the corresponding optimal target beam pair search range.

[0110] Optionally, the training process of the above search range prediction model may include:

[0111] Step 1: Construct a deep learning model that includes a shallow feature extraction module, a deep feature extraction module, and a prediction output module;

[0112] Step 2: Obtain the training sample set and the sample label set. The training sample set includes multiple beam pair information maps between the base station and the UAV as training samples. The sample label set includes the true optimal beam pair search range corresponding to each beam pair information map as the sample label.

[0113] Step 3: Iteratively train the deep learning model using the training sample set and sample label set to obtain the search range prediction model.

[0114] In the search range prediction model, the shallow feature extraction module typically includes several convolutional layers to extract low-level and local features from the beampair information map. The deep feature extraction module typically includes multiple residual blocks connected by short skips. Each residual block contains multiple convolutional layers, batch normalization layers, and activation function layers (such as ReLU) connected by residuals, used to effectively extract high-level, abstract features based on low-level image features. The output of each residual block is the "residual" of the input signal added to the convolutionally processed signal. This avoids the gradient vanishing and degradation problems that occur with increasing network depth, allowing the model to learn more complex feature representations. The prediction output module predicts the search range of the optimal switching beampair between the base station and the UAV based on the high-level and low-frequency image features output by the two modules. This module consists of two fully connected layers and a Dropout layer, with the connection relationship being the first fully connected layer, the Dropout layer, and the second fully connected layer. The first fully connected layer is used to convert high-level image features into low-dimensional prediction outputs, while the second fully connected layer converts the features output by the Dropout layer into prediction outputs to obtain the prediction search range.

[0115] In the specific training process, the deep learning model can be iteratively trained using the training sample set and sample label set to obtain the search range prediction model: For each training batch in the iterative training process, each training sample of the training batch is input into the deep learning model to obtain the search range of each predicted optimal beam pair output by the deep learning model. The target loss function is constructed using the search range of the predicted optimal beam pair and the corresponding sample label, and the model parameters of the deep learning model are adjusted according to the target loss function until the model parameters converge to obtain the search range prediction model.

[0116] The aforementioned objective loss function includes: angle loss function, search radius loss function, and physical constraint loss function. Therefore, the expression for the aforementioned objective loss function is: In the formula, Represents the angle loss function. This represents the search radius loss function. Represents the physical constraint loss function. , , These represent the adjustment coefficients for the angle loss function, search radius loss function, and physical constraint loss function, respectively. Where:

[0117] The above angle loss function The expression is:

[0118] ,

[0119] In the formula, N represents the total number of training samples in the training batch. This represents the true beam azimuth angle of the i-th training sample. This represents the predicted beam azimuth angle of the i-th training sample. This represents the true beam elevation angle of the i-th training sample. This represents the predicted beam elevation angle of the i-th training sample. This indicates the preset angle threshold. These represent the weighting coefficients.

[0120] Search radius loss function An asymmetric penalty function is used to encourage predicted radii to be slightly larger than the true value, thus avoiding missed detections. Therefore, a search radius loss function is employed. The expression can be written as:

[0121] ,

[0122] In the formula, N represents the total number of training samples in the training batch. This represents the true search radius of the i-th training sample. This represents the predicted search radius of the i-th training sample. This indicates the preset maximum allowed search range. This represents the adjustment coefficient.

[0123] In addition, physical constraint loss function The beam direction gradient constraint function and energy coverage constraint function are introduced, therefore, By introducing two physical constraints, prior knowledge such as electromagnetic wave propagation laws and antenna characteristics is embedded into the neural network, improving the model's rationality and generalization ability. Specifically:

[0124] Regarding the beam direction gradient constraint loss function It can directly target the beam pattern function of the base station. The derivative (gradient) of the base station beam pattern function describes the intensity distribution of the base station's transmitted signal in different spatial directions, while its derivative (gradient) reflects the rate at which this intensity distribution changes with direction. Therefore, this embodiment of the application, by differentiating the base station's beam pattern function in a certain direction (such as azimuth and elevation), can obtain the rate of change of beam intensity in that direction, allowing the setting of a reasonable gradient threshold as a beam direction constraint. Subsequently, if the gradient value of a beam pair exceeds the set threshold, it means that the directionality of that beam pair changes too rapidly, thus hindering stable communication; conversely, if the gradient value of a beam pair is small, it indicates that the beam direction change of that beam pair is gradual, resulting in better communication stability.

[0125] Regarding the energy cover constraint loss function The goal is to ensure that the predicted search radius contains sufficient energy to avoid missing beams. Therefore, the energy coverage of the predicted optimal beam corresponding to the training samples within the search range is less than a preset energy threshold. At that time, the energy coverage constraint loss function The value increases; while the energy coverage of the predicted optimal beam pair search range corresponding to the training samples is not less than the preset energy threshold. At that time, the energy coverage constraint loss function The value decreases. Among them, the search range of the predicted optimal beam pair corresponding to the training samples. The expression for energy coverage can be written as:

[0126] .

[0127] Finally, since the search range of the target optimal beam pair predicted by the model is determined by the center beam pair and the search radius, and the search range includes multiple candidate beam pairs, the search module 66 can use heuristic algorithms to iteratively search for the optimal switching beam pair between the base station and the UAV at the second time step within the target optimal beam pair search range. The heuristic algorithms include: particle swarm optimization algorithm and simulated annealing algorithm.

[0128] The above search process will be explained below using the simulated annealing algorithm as an example.

[0129] (a) Initialization: Setting the temperature and cooling coefficient , where the initial temperature The initial temperature should be set high enough to accept more inferior solutions in the early stages of the search, increasing the randomness of the search. The initial temperature setting usually depends on the size and complexity of the problem.

[0130] (ii) In the first round of search, the center beam pair (i.e. Figure 4 The cross points within the target optimal beam pair are used as the current optimal beam pair in the current search round. Multiple first neighboring beam pairs adjacent to the current optimal beam pair are determined within the search range of the target optimal beam pair, and the rate of change of spectral efficiency of each first neighboring beam pair compared to the current optimal beam pair is calculated according to the following formula:

[0131] ,

[0132] In the formula, Indicates the current optimal beam pair. Indicates the first neighboring beam pair, and .

[0133] When the rate of change of the spectral efficiency of the first target neighboring beam pair within multiple first neighboring beams compared to the current optimal beam pair exceeds a preset first threshold, the transfer probability of the first target neighboring beam can be calculated according to the following formula:

[0134] ,

[0135] In the formula, Indicates the predicted search radius. Indicates the maximum allowed search radius. This represents the adjustment coefficient.

[0136] When the transfer probability of the beam near the first target is higher than a preset second threshold, the beam near the first target is selected as the optimal beam pair in the next round of search, and the temperature is updated according to the following formula: .

[0137] (iii) In the second round and each subsequent search round, the neighboring beam pair switched in the previous search round is taken as the current optimal beam pair in the current search round. Within the search range of the target optimal beam pair, multiple second neighboring beam pairs adjacent to the current optimal beam pair are re-determined, and the spectral efficiency change rate of each second neighboring beam pair compared to the current optimal beam pair is determined.

[0138] When the rate of change of the spectral efficiency of the second target neighboring beam pair compared to the current optimal beam pair is higher than a first threshold, the transfer probability of the second target neighboring beam is calculated according to the following formula:

[0139] ,

[0140] In the formula, This represents the rate of change of the spectral efficiency of the beam adjacent to the second target. This represents the rate of change of the spectral efficiency of the current optimal beam pair in the current search round. Let represent the temperature in the k-th round, and , K represents the maximum number of iterations.

[0141] When the transfer probability of the second target's neighboring beam is higher than the second threshold, the second target's neighboring beam is selected as the optimal beam pair for the next search round, and the temperature of the current search round is updated according to the following formula: .

[0142] (iv) Repeat step (iii) above, and use the optimal beam pair of the target search wheel with a temperature lower than the preset third threshold (e.g., close to 0, i.e., the algorithm stops accepting inferior solutions) as the optimal switching beam pair between the base station and the UAV at the second moment.

[0143] Therefore, the search module 66, through the aforementioned heuristic search method, can dynamically adjust its search strategy in a constantly changing communication environment, ensuring that it always focuses on the most promising beam pairs, rather than blindly searching the entire beam space. This search scheme can adapt to the communication needs of different scenarios while maintaining high efficiency.

[0144] It should be noted that each module in the optimal switching beam pair search device in this application embodiment corresponds one-to-one with each implementation step of the optimal switching beam pair search method in embodiment 1. Since embodiment 1 has been described in detail, some details not shown in this embodiment can be referred to embodiment 1, and will not be elaborated further here.

[0145] Example 3

[0146] According to an embodiment of this application, a computer program product is also provided, which includes a computer program, wherein when the computer program is executed by a processor, it implements the optimal switching beam pair search method in embodiment 1.

[0147] According to an embodiment of this application, a non-volatile storage medium is also provided, which includes a stored computer program, wherein the device containing the non-volatile storage medium executes the optimal switching beam pair search method in Embodiment 1 by running the computer program.

[0148] According to an embodiment of this application, a processor is also provided for running a computer program, wherein the computer program executes the optimal switching beam pair search method in embodiment 1 during runtime.

[0149] According to an embodiment of this application, an electronic device is also provided, comprising: a memory and a processor, wherein the memory stores a computer program, and the processor is configured to execute the optimal switching beam pair search method of embodiment 1 through the computer program.

[0150] Specifically, the computer program executes the following steps during runtime: acquiring a target beam pair information map of multiple beam pairs between the base station and the target UAV within the base station's coverage area at a first time; analyzing the target beam pair information map using a pre-trained search range prediction model to obtain the corresponding optimal target beam pair search range, wherein the search range prediction model is constructed based on an improved deep residual network, and the improved deep residual network includes at least: a shallow feature extraction module, a deep feature extraction module based on a residual nested structure, and a prediction output module; and determining the optimal switching beam pair between the base station and the UAV at a second time from the optimal target beam pair search range using a heuristic algorithm, wherein the second time is the time following the first time.

[0151] As an alternative implementation, the above-mentioned electronic device may exist in the form of a mobile terminal, a computer terminal, or a similar computing device. Figure 7 A hardware block diagram of an electronic device for implementing an optimal switched beam pair search method is shown. Figure 7 As shown, the electronic device 70 may include one or more (shown as 702a, 702b, ..., 702n) processors 702 (processors 702 may include, but are not limited to, microprocessors such as MCUs or programmable logic devices such as FPGAs), a memory 704 for storing data, and a transmission device 706 for communication functions. In addition, it may also include: a display, an input / output interface (I / O interface), a universal serial bus (USB) port (which may be included as one of the ports of a BUS bus), a network interface, a power supply, and / or a camera. Those skilled in the art will understand that... Figure 7 The structure shown is for illustrative purposes only and does not limit the structure of the aforementioned electronic device. For example, electronic device 70 may also include components that are more... Figure 7 The more or fewer components shown, or having the same Figure 7 The different configurations shown.

[0152] It should be noted that the aforementioned one or more processors 702 and / or other data processing circuits are generally referred to herein as "data processing circuits". These data processing circuits may be embodied, in whole or in part, in software, hardware, firmware, or any other combination thereof. Furthermore, the data processing circuits may be a single, independent processing module, or may be integrated, in whole or in part, into any other element of the electronic device 70. As involved in the embodiments of this application, the data processing circuits serve as a processor control mechanism (e.g., selection of a variable resistor termination path connected to an interface).

[0153] The memory 704 can be used to store software programs and modules of application software, such as the program instructions / data storage device corresponding to the optimal switching beam pair search method in this embodiment. The processor 702 executes various functional applications and data processing by running the software programs and modules stored in the memory 704, thereby implementing the aforementioned vulnerability detection method for the application. The memory 704 may include high-speed random access memory, and may also include non-volatile memory, such as one or more magnetic storage devices, flash memory, or other non-volatile solid-state memory. In some instances, the memory 704 may further include memory remotely located relative to the processor 702, and these remote memories can be connected to the electronic device 70 via a network. Examples of such networks include, but are not limited to, the Internet, corporate intranets, local area networks, mobile communication networks, and combinations thereof.

[0154] The transmission device 706 is used to receive or send data via a network. Specific examples of the network described above may include a wireless network provided by the communication provider of the electronic device 70. In one example, the transmission device 706 includes a Network Interface Controller (NIC), which can connect to other network devices via a base station to communicate with the Internet. In another example, the transmission device 706 may be a Radio Frequency (RF) module for wireless communication with the Internet.

[0155] The display can be, for example, a touchscreen liquid crystal display (LCD), which allows the user to interact with the user interface of the electronic device 70.

[0156] The sequence numbers of the above embodiments are for descriptive purposes only and do not represent the superiority or inferiority of the embodiments.

[0157] In the above embodiments of this application, the descriptions of each embodiment have different focuses. For parts not described in detail in a certain embodiment, please refer to the relevant descriptions of other embodiments.

[0158] In the several embodiments provided in this application, it should be understood that the disclosed technical content can be implemented in other ways. The device embodiments described above are merely illustrative; for example, the division of units can be a logical functional division, and in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the displayed or discussed mutual couplings, direct couplings, or communication connections may be through some interfaces; indirect couplings or communication connections between units or modules may be electrical or other forms.

[0159] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple units. Some or all of the units can be selected to achieve the purpose of this embodiment according to actual needs.

[0160] Furthermore, the functional units in the various embodiments of this application can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The integrated unit can be implemented in hardware or as a software functional unit.

[0161] If the integrated unit is implemented as a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, or all or part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods of the various embodiments of this application. The aforementioned storage medium includes various media capable of storing program code, such as a USB flash drive, read-only memory (ROM), random access memory (RAM), portable hard drive, magnetic disk, or optical disk.

[0162] The above are merely preferred embodiments of this application. It should be noted that those skilled in the art can make various improvements and modifications without departing from the principles of this application, and these improvements and modifications should also be considered within the scope of protection of this application.

Claims

1. An optimal switching beam pair search method, characterized in that, include: Obtain a target beam pair information map at a first moment between a base station and a target UAV within the coverage area of ​​the base station. The target beam pair information map is analyzed using a pre-trained search range prediction model to obtain the corresponding target optimal beam pair search range. The search range prediction model is constructed based on an improved deep residual network, and the improved deep residual network includes at least: a shallow feature extraction module, a deep feature extraction module based on a residual nesting structure, and a prediction output module. The optimal switching beam pair between the base station and the UAV at the second time point is determined from the search range of the target optimal beam pair using a heuristic algorithm, wherein the second time point is the time point following the first time point; The target loss function of the search range prediction model includes: an angle loss function, a search radius loss function, and a physical constraint loss function, and the expression of the target loss function is: ， In the formula, This represents the angle loss function. This represents the search radius loss function. This represents the physical constraint loss function. , , Let represent the adjustment coefficients of the angle loss function, the search radius loss function, and the physical constraint loss function, respectively, where: The angle loss function The expression is: ， In the formula, N represents the total number of training samples in the training batch. This represents the true beam azimuth angle of the i-th training sample. This represents the predicted beam azimuth angle of the i-th training sample. This represents the true beam elevation angle of the i-th training sample. This represents the predicted beam elevation angle of the i-th training sample. This indicates the preset angle threshold. These represent the weighting coefficients; The search radius loss function The expression is: ， In the formula, N represents the total number of training samples in the training batch. This represents the true search radius of the i-th training sample. This represents the predicted search radius of the i-th training sample. This indicates the preset maximum allowed search radius. Indicates the adjustment coefficient; The physical constraint loss function The expression is: ， In the formula, Let represent the beam directional gradient constraint loss function, and the beam directional gradient constraint loss function It is obtained by differentiating the base station beam pattern function. This represents the energy coverage constraint loss function, and when the energy coverage of the predicted optimal beam pair search range corresponding to the training sample is less than a preset energy threshold, the energy coverage constraint loss function is increased. The value of is determined by reducing the energy coverage constraint loss function when the energy coverage of the predicted optimal beam pair search range corresponding to the training sample is not less than the energy threshold. The value of .

2. The method according to claim 1, characterized in that, Acquire a target beam pair information map at a first moment between a base station and a target UAV within the coverage area of ​​the base station, including: The communication quality parameters of multiple beam pairs between the base station and the target UAV at a first moment are obtained, wherein the communication quality parameters include at least one of the following: spectral efficiency and signal-to-dryness ratio; The system acquires first communication status information of the base station at a first moment and second communication status information of the target UAV at a first moment. The first communication status information includes at least: first beam status information of each of the multiple beams of the base station and base station status information. The second communication status information includes at least: second beam status information of each of the multiple beams of the target UAV and UAV status information. The first beam status information and the second beam status information include at least: beam azimuth angle, beam pitch angle, beamwidth, and beam gain. For each beam pair, the first beam state information, the second beam state information, the communication quality parameters of the beam pair, the base station state information, and the UAV state information of each of the two beams in the beam pair are used as multiple features of the beam pair. The multiple features are used to form the feature vector of the beam pair, and the feature vector is standardized. The multiple beams of the base station and the multiple beams of the target UAV are arranged in ascending order of beam number as multiple matrix rows and multiple matrix columns. The standardized feature vectors of each beam pair are used as matrix elements to construct the target beam pair information map.

3. The method according to claim 1, characterized in that, The training process of the search range prediction model includes: Construct a deep learning model including the shallow feature extraction module, the deep feature extraction module, and the prediction output module; Obtain a training sample set and a sample label set, wherein the training sample set includes multiple beam pair information maps between the base station and the UAV as training samples, and the sample label set includes the true optimal beam pair search range corresponding to each beam pair information map as a sample label; The deep learning model is iteratively trained using the training sample set and the sample label set to obtain the search range prediction model.

4. The method according to claim 3, characterized in that, The shallow feature extraction module includes at least one convolutional layer, and the shallow feature extraction module is used to extract low-level image features of the beampair information map; The deep feature extraction module includes multiple residual blocks connected by short skips. Each residual block includes multiple convolutional layers, batch normalization layers, and activation function layers connected by residuals. The deep feature extraction module is used to extract high-level features of the image based on the low-level features of the image. The prediction output module includes two fully connected layers and a Dropout layer, and is used to convert the high-level features of the image into a prediction search range.

5. The method according to claim 3, characterized in that, The deep learning model is iteratively trained using the training sample set and the sample label set to obtain the search range prediction model, including: For each training batch in the iterative training process, each training sample of the training batch is input into the deep learning model to obtain the search range of each predicted optimal beam pair output by the deep learning model. The target loss function is constructed using the search range of the predicted optimal beam pair and the corresponding sample labels, and the model parameters of the deep learning model are adjusted according to the target loss function until the model parameters converge to obtain the search range prediction model.

6. The method according to claim 1, characterized in that, The target optimal beam pair search range is determined by the center beam pair and the search radius. The determination of the optimal handover beam pair between the base station and the UAV at the second moment from the target optimal beam pair search range using a heuristic algorithm includes: Initial temperature and cooling coefficient ; In the first round of search, the center beam pair is taken as the current optimal beam pair in the current search round. Multiple first neighboring beam pairs adjacent to the current optimal beam pair are determined within the search range of the target optimal beam pair, and the spectral efficiency change rate of each first neighboring beam pair compared to the current optimal beam pair is calculated according to the following formula: ， In the formula, This indicates the current optimal beam pair. Indicates the first neighboring beam pair, and When the rate of change of the spectral efficiency of the first target neighboring beam pair compared to the current optimal beam pair is higher than a preset first threshold, the transfer probability of the first target neighboring beam is calculated according to the following formula: ， In the formula, This indicates the predicted search radius. Indicates the maximum allowed search radius. This represents the adjustment coefficient; when the transfer probability of the first target's neighboring beam is higher than a preset second threshold, the first target's neighboring beam is taken as the optimal beam pair in the next search round, and the temperature of the current search round is updated according to the following formula: ; In the second and subsequent search rounds, the neighboring beam pairs switched in the previous search round are taken as the current optimal beam pairs in the current search round. Multiple second neighboring beam pairs adjacent to the current optimal beam pair are re-determined within the search range of the target optimal beam pair, and the rate of change of spectral efficiency of each second neighboring beam pair compared to the current optimal beam pair is determined. When the rate of change of spectral efficiency of the second target neighboring beam pair compared to the current optimal beam pair is higher than the first threshold, the transition probability of the second target neighboring beam is calculated according to the following formula: ， In the formula, This represents the rate of change of the spectral efficiency of the beam adjacent to the second target. This represents the rate of change of the spectral efficiency of the current optimal beam pair in the current search round. Let represent the temperature in the k-th round, and , K represents the maximum number of iterations for the search. When the transfer probability of the second target neighboring beam is higher than the second threshold, the second target neighboring beam is taken as the optimal beam pair in the next round of the search process, and the temperature of the current search wheel is updated. The above steps are repeated, and the optimal beam pair of the target search wheel with a temperature lower than the preset third threshold is taken as the optimal switching beam pair between the base station and the UAV at the second moment.

7. An optimal switching beam pair search device, characterized in that, include: The acquisition module is used to acquire a target beam pair information map of multiple beam pairs between the base station and the target UAV within the coverage area of ​​the base station at a first moment. The prediction module is used to analyze the target beam pair information map using a pre-trained search range prediction model to obtain the corresponding target optimal beam pair search range. The search range prediction model is constructed based on an improved deep residual network, and the improved deep residual network includes at least: a shallow feature extraction module, a deep feature extraction module based on a residual nesting structure, and a prediction output module. The search module is used to determine the optimal switching beam pair between the base station and the UAV at a second time point from the search range of the target optimal beam pair using a heuristic algorithm, wherein the second time point is the time point following the first time point; The target loss function of the search range prediction model includes: an angle loss function, a search radius loss function, and a physical constraint loss function, and the expression of the target loss function is: ， In the formula, This represents the angle loss function. This represents the search radius loss function. This represents the physical constraint loss function. , , Let represent the adjustment coefficients of the angle loss function, the search radius loss function, and the physical constraint loss function, respectively, where: The angle loss function The expression is: ， In the formula, N represents the total number of training samples in the training batch. This represents the true beam azimuth angle of the i-th training sample. This represents the predicted beam azimuth angle of the i-th training sample. This represents the true beam elevation angle of the i-th training sample. This represents the predicted beam elevation angle of the i-th training sample. This indicates the preset angle threshold. These represent the weighting coefficients; The search radius loss function The expression is: ， In the formula, N represents the total number of training samples in the training batch. This represents the true search radius of the i-th training sample. This represents the predicted search radius of the i-th training sample. This indicates the preset maximum allowed search radius. Indicates the adjustment coefficient; The physical constraint loss function The expression is: ， In the formula, Let represent the beam directional gradient constraint loss function, and the beam directional gradient constraint loss function It is obtained by differentiating the base station beam pattern function. This represents the energy coverage constraint loss function, and when the energy coverage of the predicted optimal beam pair search range corresponding to the training sample is less than a preset energy threshold, the energy coverage constraint loss function is increased. The value of is determined by reducing the energy coverage constraint loss function when the energy coverage of the predicted optimal beam pair search range corresponding to the training sample is not less than the energy threshold. The value of .

8. A computer program product, characterized in that, include: A computer program, wherein when executed by a processor, the computer program implements the optimal switching beam pair search method according to any one of claims 1 to 6.

9. An electronic device, characterized in that, include: A memory and a processor, wherein the memory stores a computer program, and the processor is configured to execute the optimal switching beam pair search method according to any one of claims 1 to 6 through the computer program.

Images