Beam prediction method, electronic equipment and computer readable storage medium

Abstract

The invention discloses a beam prediction method, electronic equipment and a computer readable storage medium, and relates to the technical field of wireless communication, and the beam prediction method comprises the steps: collecting a receiving beam and the intensity of a preset proportion of a transmitting beam corresponding to the receiving beam, and obtaining a beam pair measurement result; the beam pair measurement result and the current environment information are input into a preset target beam prediction model, a target beam pair prediction result is obtained, the target beam pair prediction result at least comprises the intensity of all transmitting beams corresponding to the receiving beam, and the intensity of all transmitting beams corresponding to the receiving beam is obtained. The target beam prediction model is obtained by training multiple groups of receiving beam intensity data and corresponding transmitting beam intensity data in multiple transmission environments. According to the invention, the pre-trained target beam prediction model is adopted, the intensity of all the transmitting beams corresponding to the receiving beam is predicted according to the intensity of the partial proportion transmitting beams corresponding to the receiving beam, and the measurement overhead of beam scanning is saved.

Description

Technical Field

[0001] This application relates to the field of wireless communication technologies, and in particular, to a beam prediction method, an electronic device, and a computer-readable storage medium. Background Art

[0002] In the field of wireless communication technologies, the wireless Massive (large-scale)-MIMO (Multiple Input Multiple Output) technology can significantly improve the capacity of communication systems. In high-frequency scenarios, the formation and selection of intelligent beams are the key to further improving the capacity of communication systems. Accurate MIMO beam selection is based on beam measurement. Traditional beam measurement schemes need to comprehensively scan all beams in the communication system to obtain the intensity measurement results of all beams, and then select the beam with the strongest intensity based on the measurement results. However, when the number of beams in the communication system is large, comprehensively scanning all beams requires huge measurement overhead and brings high communication costs. Summary of the Invention

[0003] The main purpose of this application is to provide a beam prediction method, an electronic device, and a computer-readable storage medium, aiming to solve the technical problem of excessive measurement overhead in beam measurement for MIMO communication.

[0004] To achieve the above object, this application provides a beam prediction method, which includes: Collect the intensities of a received beam and a preset proportion of transmitted beams corresponding to the received beam to obtain beam pair measurement results; Input the beam pair measurement results and current environmental information into a preset target beam prediction model to obtain a target beam pair prediction result, where the target beam pair prediction result at least includes the intensities of all transmitted beams corresponding to the received beam, and the target beam prediction model is trained by multiple groups of received beam intensity data and corresponding transmitted beam intensity data under various transmission environments.

[0005] Optionally, the target beam prediction model at least includes a beam preprocessing module, an environmental preprocessing module, and a final output module. The step of inputting the beam pair measurement results and current environmental information into a preset target beam prediction model to obtain a target beam pair prediction result includes: Preprocess the beam pair measurement results through the beam preprocessing module and output a beam processing result; Preprocess the current environmental information through the environmental preprocessing module and output a first environmental processing result; Determine the target beam pair prediction result according to the beam processing result, the first environmental processing result, and the final output module.

[0006] Optionally, the beam preprocessing module includes a first beam preprocessing module and a second beam preprocessing module, the beam processing result includes a first beam processing result and a second beam processing result, and the step of preprocessing the beam pair measurement result through the beam preprocessing module and outputting the beam processing result includes: Input the beam pair measurement result into the first beam preprocessing module, preprocess the beam pair measurement result through the first beam preprocessing module, and output the first beam processing result; Divide the beam pair measurement result into a first preset number of sub-beam pair measurement results, and input each of the sub-beam pair measurement results into the second beam preprocessing module respectively to output the second beam processing result.

[0007] Optionally, the target beam prediction model further includes a first subsequent processing module and a second subsequent processing module, and the step of determining the target beam pair prediction result according to the beam processing result, the first environment processing result, and the final output module includes: Determine the first beam pair prediction result according to the first beam processing result, the first environment processing result, a preset matrix, and the first subsequent processing module; Determine the second beam pair prediction result according to the second beam processing result, the first environment processing result, and the second subsequent processing module; Input the first beam pair prediction result and the second beam pair prediction result into the final output module to output the target beam pair prediction result.

[0008] Optionally, the step of determining the first beam pair prediction result according to the first beam processing result, the first environment processing result, a preset matrix, and the first subsequent processing module includes: Concatenate the first beam processing result multiplied by 2 with the preset matrix to obtain a first concatenated result; or Concatenate the first beam processing result with the first environment processing result to obtain a first concatenated result; Input the first concatenated result into the first subsequent processing module to obtain the first beam pair prediction result.

[0009] Optionally, the target beam prediction model further includes a simple environment processing module, and the step of determining the second beam pair prediction result according to the second beam processing result, the first environment processing result, and the second subsequent processing module includes: Input the first environment processing result into the simple environment processing module to obtain a second environment processing result; Splice the second beam processing result and the second environment processing result to obtain a second splicing result; Input the second splicing result into the second subsequent processing module to obtain a second beam pair prediction result.

[0010] Optionally, before the step of inputting the beam pair measurement result and the current environment information into a preset target beam prediction model to obtain a target beam pair prediction result, the method further includes: In a variety of different transmission environments, based on a preset antenna order set and a preset carrier frequency, collect received beam intensities, corresponding transmitted beam intensities, and environment information to obtain a training data set, where the antenna order set includes multiple different antenna order combinations; Train a preset initial beam prediction model with the training data set to obtain a target beam prediction weight and a target environment migration weight corresponding to the initial beam prediction model; Update the model parameters of the initial beam prediction model according to the beam prediction weight and the environment migration weight to obtain a target beam prediction model.

[0011] Optionally, the step of training a preset initial beam prediction model with the training data set to obtain a target beam prediction weight and a target environment migration weight corresponding to the initial beam prediction model includes: Select multiple groups of input beam pair intensity data from the training data set, where each group of beam pair intensity data includes a first preset number of received beam intensity data and a corresponding second preset number of transmitted beam intensity data; Input each beam pair intensity data and the corresponding environment information into the initial beam prediction model, and perform prediction through the initial beam prediction model to output a third beam pair prediction result, a fourth beam pair prediction result, and a final beam pair prediction result; According to the third beam pair prediction result, the fourth beam pair prediction result, the final beam pair prediction result, and the beam pair intensity data, calculate the function losses corresponding to the third beam pair prediction result, the fourth beam pair prediction result, and the final beam pair prediction result respectively; Based on the function losses corresponding to the third beam pair prediction result, the fourth beam pair prediction result, and the final beam pair prediction result, iteratively optimize the preset beam prediction weight and the preset environment migration weight in the initial beam prediction model to obtain the target beam prediction weight and the target environment migration weight.

[0012] This application also provides a beam prediction device, and the beam prediction device includes: A beam pair measurement module, configured to collect the intensities of a received beam and a preset proportion of transmitted beams corresponding to the received beam, and obtain a beam pair measurement result; A beam pair prediction module, configured to input the beam pair measurement result and current environmental information into a preset target beam prediction model to obtain a target beam pair prediction result, where the target beam pair prediction result at least includes the intensities of all transmitted beams corresponding to the received beam, and the target beam prediction model is trained by multiple groups of received beam intensity data and corresponding transmitted beam intensity data in multiple transmission environments.

[0013] This application also provides an electronic device, which is a physical device. The electronic device includes: a memory, a processor, and a program of the beam prediction method stored on the memory and executable on the processor. When the program of the beam prediction method is executed by the processor, the steps of the beam prediction method as described above can be implemented.

[0014] This application also provides a computer-readable storage medium, on which a program for implementing the beam prediction method is stored. When the program of the beam prediction method is executed by a processor, the steps of the beam prediction method as described above are implemented.

[0015] This application also provides a computer program product, including a computer program. When the computer program is executed by a processor, the steps of the beam prediction method as described above are implemented.

[0016] This application provides a beam prediction method, an electronic device, and a computer-readable storage medium. First, the intensities of a received beam and a preset proportion of transmitted beams corresponding to the received beam are collected to obtain a beam pair measurement result, and then the beam pair measurement result and current environmental information are input into a preset target beam prediction model to obtain a target beam pair prediction result, where the target beam pair prediction result at least includes the intensities of all transmitted beams corresponding to the received beam, and the target beam prediction model is trained by multiple groups of received beam intensity data and corresponding transmitted beam intensity data in multiple transmission environments. The technical solution of this application uses a pre-trained target beam prediction model to predict the intensities of all transmitted beams corresponding to a received beam based on the intensities of a partial proportion of transmitted beams corresponding to the received beam, providing an accurate data basis for intelligent beam selection, achieving accurate and low-overhead intelligent beam prediction, saving the measurement overhead of beam scanning, reducing communication costs, and solving the technical problem of excessive measurement overhead in beam measurement of MIMO communication.

[0017] In addition, since the beam prediction method of the present application introduces the current environmental information and the target beam prediction model trained with multiple groups of received beam intensity data and corresponding transmitted beam intensity data under various transmission environments, the beam prediction method has a certain environmental migration ability and can adapt to various different transmission environments. BRIEF DESCRIPTION OF THE DRAWINGS

[0018] The accompanying drawings herein are incorporated into and constitute a part of this specification, showing embodiments consistent with the present application, and are used together with the specification to explain the principles of the present application.

[0019] To more clearly illustrate the technical solutions in the embodiments of the present application or in the prior art, the following will briefly introduce the drawings required for use in the description of the embodiments or the prior art. Obviously, for those of ordinary skill in the art, other drawings can be obtained based on these drawings without creative efforts.

[0020] Figure 1 Schematic flowchart of the first embodiment of the beam prediction method of the present application; Figure 2 Schematic diagram of the number of antenna elements of 64TRX (16H4V) involved in the beam prediction method of the present application; Figure 3 Schematic diagram for comparing two receiving poles of 64 out of 4*64 of a certain sample in the beam prediction training set involved in the beam prediction method of the present application; Figure 4 Schematic diagram of the 64-beam distribution of 16H4V involved in the beam prediction method of the present application; Figure 5 Schematic diagram for comparing one receiving pole of 128 out of 4*128 of a certain sample in the beam prediction training set involved in the beam prediction method of the present application; Figure 6 Schematic diagram of the 128-beam distribution of 16H4V involved in the beam prediction method of the present application; Figure 7 Schematic diagram of the structure and data flow of the target beam prediction model in the first embodiment of the beam prediction method of the present application; Figure 8 Brief schematic diagram of the backbone prediction model in the first embodiment of the beam prediction method of the present application; Figure 9 Schematic diagram of the training of each prediction weight in the second embodiment of the beam prediction method of the present application; Figure 10 Schematic diagram of the composition structure of the beam prediction device in the embodiment of the present application; Figure 11 Schematic diagram of the device structure of the hardware operating environment involved in the beam prediction method in the embodiment of the present application.

[0021] The realization of the purpose, functional features and advantages of this application will be further described in conjunction with embodiments with reference to the accompanying drawings. Specific Embodiments

[0022] To make the above objects, features, and advantages of this application more obvious and understandable, the technical solutions in the embodiments of this application will be clearly and completely described below in conjunction with the accompanying drawings in the embodiments of this application. Obviously, the described embodiments are only a part of the embodiments of this application, rather than all the embodiments. All other embodiments obtained by those of ordinary skill in the art based on the embodiments in this application without creative efforts shall fall within the scope of protection of this application.

[0023] Embodiment 1 Nowadays, wireless Massive-MIMO technology has significantly improved the capacity of communication systems. In high-frequency scenarios, more accurate and lower-overhead intelligent beamforming is the key to achieving a greater increase in the capacity of future new-generation communication systems. Among them, the correct MIMO beam selection is based on the measurement of quasi-beam intensity. However, as the number of beams increases, the overhead of globally scanning all beams and measuring them is extremely large, which is unacceptable in the actual application process of communication systems. The idea of the technical solution of the embodiment of this application is to first perform a sparse beam scan measurement of a certain proportion, and predict other unmeasured beams based on the measurement results, and finally select the beam with the highest intensity. And how to obtain the most accurate beam prediction result with a given measurement overhead through deep learning neural network technology is an important requirement.

[0024] Based on the above requirements, the embodiment of this application provides a beam prediction method. In the first embodiment of the beam prediction method of this application, with reference to Figure 1 , the beam prediction method includes: Step S10, collect the intensity of the received beam and a preset proportion of the transmitted beams corresponding to the received beam to obtain the beam pair measurement result; In the embodiments of the present application, it should be noted that step S10 is executed in a communication system at a certain carrier frequency. In this communication system, it may include the 64x4 beam pair intensity corresponding to the carrier frequency f1, which is a set of 64x4 beam pairs composed of 64 transmitting beams and 4 receiving beams, or the 128x4 beam pair intensity corresponding to the carrier frequency f2, which is a set of 128x4 beam pairs composed of 128 transmitting beams and 4 receiving beams. In the embodiments of the present application, the number of receiving beams can be 4, and the intensity of a certain preset proportion of the transmitting beams corresponding to each receiving beam is collected respectively. Among them, the preset proportion can be set according to user needs. The lower the preset proportion, the fewer the transmitting beams to be measured, and the greater the saved overhead, but it may lead to a decrease in accuracy. If the preset proportion is too high, it will lead to an increase in measurement overhead.

[0025] As a preference, the preset proportion is 1 / 8. For the communication system with the carrier frequency f1, consider a set of 64x4 beam pairs composed of 64 transmitting beams and 4 receiving beams. For each receiving beam, scan 8 of the transmitting beams to obtain the measurement results of 8x4 beam pairs, and then use the measurement results of these 8x4 beam pairs as inputs for beam pair prediction. Among them, each beam pair measurement result includes the receiving beam intensity and the intensity of a corresponding transmitting beam. Through experiments, in this case, the measurement overhead is relatively small, and the prediction accuracy can also reach more than 90%.

[0026] Step S20: Input the beam pair measurement results and the current environmental information into a preset target beam prediction model to obtain a target beam pair prediction result, where the target beam pair prediction result at least includes the intensities of all the transmitting beams corresponding to the receiving beam, and the target beam prediction model is trained by multiple groups of receiving beam intensity data and corresponding transmitting beam intensity data in various transmission environments.

[0027] It should be noted that in the embodiments of the present application, in addition to collecting the transmitting beam intensity, it is also necessary to collect the transmission environment picture of the current communication environment (including building layout, building height, etc.), the base station location, and the location of the terminal reporting the signal strength information as the current environmental information. The beam prediction method of the embodiments of the present application introduces environmental information as an input for beam prediction, so as to achieve environmental migration of beam prediction, enabling the beam prediction method to be applied to different transmission environments. Correspondingly, the above-mentioned target beam prediction model is also trained by the environmental information and beam pair measurement results in different transmission environments.

[0028] In addition, the target beam prediction model is trained by using environmental information and beam pair measurement results in a communication system with the same carrier frequency in different transmission environments. That is to say, the target beam prediction model is adapted to beam prediction under one carrier frequency. For example, it can predict a set of 64x4 beam pairs composed of 64 transmit beams and 4 receive beams corresponding to the carrier frequency f1. If beam prediction under other carrier frequencies needs to be implemented, additional training of the beam prediction model under other carrier frequencies is required, thereby enabling carrier frequency migration of beam prediction.

[0029] In step S20, after inputting the beam pair measurement results and the current environmental information into the preset target beam prediction model, the target beam prediction model can process and predict the input beam pair measurement results and the current environmental information to finally output a target beam pair prediction result composed of the intensities of all transmit beams corresponding to multiple receive beams and the intensities of the receive beams. For example, if the input is the intensities of 4 receive beams and 8 transmit beams corresponding to each of the receive beams, the output is a set of beam pairs composed of 4*64 receive beams and transmit beams.

[0030] Exemplarily, after outputting the target beam pair measurement results, K strongest beam pairs can be determined according to the preset selection quantity, where K = {1, 3, 5}.

[0031] In the embodiments of the present application, the reason why the target beam prediction model pre-trained by technologies such as a neural network model structure and a machine learning algorithm can predict the intensities of all beam pairs based on the beam pair measurement results and environmental information composed of a certain proportion of transmit beam intensities and receive beam intensities is that there is a certain regularity between the intensity values of each beam pair and the intensity values of other beam pairs, and there is also a certain correlation between the environmental information and the final beam pair intensity values. The specific principle is as follows.

[0032] In the field of wireless communication technology, the coverage ability of a radio frequency module is usually represented by the number of TRXs (Transmitter, transmitters) or mHnV (antenna array number). As Figure 2 shown, the number of antenna elements of 64TRX(16H4V) is 192, and 64TRX has 32 grids. Each grid contains two polarization directions of +45° / -45° for 1 drive 3 physical antennas, a total of 64 physical antennas. The 16 physical antennas in the horizontal direction can process the signals of 16 TRXs in each row, simply referred to as 16H. After the energy is concentrated, the horizontal coverage can be determined together. The 4 physical antennas in the vertical direction can process the signals of 4 TRXs in one polarization in each column, simply referred to as 4V. After the energy is concentrated, the vertical coverage can be determined together.

[0033] Analyzing the training set samples one by one reveals that, as Figure 3 , the 64 transmit beams on each receive beam exhibit a 16H4V pattern. (For task2, it is 16H8V, where task2 refers to carrier frequency migration). Among them, the data structure of the training set samples is <intensity of 64x4 beam pairs, transmission environment picture, base station (BS) location, terminal location>, and the transmission environment picture shows the layout of buildings in the scenario where the data is generated. There may or may not be building blockages between the BS and the terminal, so these transmission environments may be NLOS or LOS environments. The test dataset contains 1000 test samples, and its data structure is the same as that of the training samples. The beam prediction accuracy is explored for three different values of K (1, 3, 5).

[0034] By observing the 4*64 EDA (Exploratory Data Analysis) of the data, it is found that the data on the actual terminal receiving poles is close for 0 / 2 and close for 1 / 3. It can be basically determined that 0 / 2 is orthogonal and 1 / 3 is orthogonal. The orthogonal antennas are basically in a state where the data features are close, providing a theoretical basis for beam prediction based on the measurement results of partial beam pairs.

[0035] Scattering the strongest beam distribution of the sample points on the map situation, as Figure 4 shown. According to the 64-beam distribution of 16H4V (the horizontal and vertical coordinates are positions), the numbers above the distribution points are the positions in 4V, and the numbers below are the numbers in 16H. It can be seen that the 4V position is strongly related to the distance between the terminal and the BS and the presence or absence of blockages, and the 16H position is strongly related to the azimuth angle between the terminal and the BS. This provides a theoretical basis for beam prediction based on environmental information.

[0036] The above content explains the prediction of 64*4 beam pairs when the carrier frequency is f1. Next, the prediction principle of 128*4 beam pairs for the carrier frequency f2 is elaborated.

[0037] As Figure 5 shown, the 128 transmit beams on the receive beam of each carrier frequency migration sample exhibit a 16H8V pattern; as Figure 6 shown, scattering the strongest beam distribution of the sample points on the map situation (the horizontal and vertical coordinates are positions), as Figure 5 shown. According to the 128-beam distribution of 16H8V, the numbers in the upper right of the distribution points are the positions in 8V, and the numbers below are the numbers in 16H. It can be seen that the 8V position is strongly related to the distance between the terminal and the BS and the presence or absence of blockages, and the 16H position is strongly related to the azimuth angle between the terminal and the BS.

[0038] Compare the data characteristics of beam prediction / environment migration in the previous data EDA part. Consider compressing the data of carrier frequency migration on the horizontal axis of the transmitting antenna by the mean into 4*16 for training to predict 4 * 16 * 8, and then using the generative model to compress the data of beam prediction and environment migration on the horizontal axis of the transmitting antenna by the mean into 4*16 and combining 19 environmental information to predict 4* 16 * 8.

[0039] After analysis, by introducing the data of beam prediction / environment migration, the training data of carrier frequency migration is expanded from the original 1000 to 12000, and there is an improvement of about 10 percentage points in the actual effect.

[0040] In another feasible embodiment, the target beam prediction model at least includes a beam preprocessing module, an environment preprocessing module, and a final output module. In step S20, the step of inputting the beam pair measurement result and the current environmental information into a preset target beam prediction model to obtain the target beam pair prediction result may include: Step S21, preprocess the beam pair measurement result through the beam preprocessing module, and output a beam processing result; Step S22, preprocess the current environmental information through the environment preprocessing module, and output a first environment processing result; Step S23, determine the target beam pair prediction result according to the beam processing result, the first environment processing result, and the final output module.

[0041] In the embodiment of the present application, the composition structure characteristics of the target beam prediction model are disclosed. In the target beam prediction model, it at least includes a beam preprocessing module, an environment preprocessing module, and a final output module. Each of the modules is a MixerBlock (mixer module) module. The target beam prediction model can be understood as a backbone prediction model obtained by combining several MixerBlock modules. The backbone prediction model can analyze the environmental information and beam data, take 8x4 beam pair measurement results and 19 environmental information as input data, and obtain 64*4 (128*4) predicted beam pair prediction results. Among them, the 19 environmental information at least includes information such as the longitude and latitude of the base station / terminal, the polar coordinates of the terminal relative to the base station, the length and width of the environmental picture, and the blocking information between the base station and the terminal. Specifically, refer to Figure 7 , in the backbone prediction model, preprocessing is respectively performed through the beam preprocessing module and the environment preprocessing module. Among them, the beam preprocessing module includes beam_blocks (beam module) and simple_beam_blocks (simple beam module), and the environment preprocessing module is env_blocks. In addition, Figure 7The figure shows a prediction model for 64*4 beam pairs at a carrier frequency of f1. When the corresponding carrier frequency is f2, compared with the Figure 7 model structure in it, the main replacement is that the input of 8x4 beam pair measurement results becomes the mean compression results of 16x4 beam pairs, and the output result is replaced from the prediction results of 64 beam pairs * 4 to the prediction results of 128 beam pairs, and the others are the same.

[0042] It should be noted that after the input beam preprocessing module and the current environmental information are processed by beam_blocks, simple_beam_blocks, and env_blocks and the processing results are output, through steps such as reprocessing and splicing, finally, the first beam pair prediction result corresponding to the beam pair measurement result and the second beam pair prediction result corresponding to the current environmental information are obtained and input into the final output module (final_mlp), and the target beam pair prediction result is output through the final output module. Here, "mlp" refers to the Multi-Layer Perceptron.

[0043] Exemplarily, the above backbone prediction model diagram is as Figure 8 shown. Taking the measurement results of 8x4 beam pairs and 19 environmental information as input data, a model obtained by combining several MixerBlock modules is used, and finally the predicted measurement results of 64*4 (128*4) beam pairs are obtained.

[0044] Furthermore, the beam preprocessing module includes a first beam preprocessing module and a second beam preprocessing module, and the beam processing results include a first beam processing result and a second beam processing result. In step S21, the step of preprocessing the beam pair measurement results through the beam preprocessing module and outputting the beam processing results may include: Step S211: Input the beam pair measurement results into the first beam preprocessing module, and preprocess the beam pair measurement results through the first beam preprocessing module to output the first beam processing result; Step S212: Divide the beam pair measurement results into a first preset number of sub-beam pair measurement results, and input each of the sub-beam pair measurement results into the second beam preprocessing module respectively to output the second beam processing result.

[0045] Refer to Figure 7, the beam preprocessing module includes a first beam preprocessing module (beam_blocks) and a second beam preprocessing module (simple_beam_blocks), and the first preset quantity is the number of receiving beams (4). In the embodiment of the present application, when inputting the measurement results of 8x4 beam pairs, two different processing schemes are performed simultaneously. The first case is to directly input into a beam_blocks module, and the second case is to split them into 4 groups of sub-beam pair measurement results. Among them, each group of sub-beam pair measurement results includes the intensity of one receiving beam and the intensities of the corresponding 8 transmitting beams, and then they are sent into the same simple_beam_blocks in the original order. The original order refers to the receiving order among the transmitting beams corresponding to the receiving beam.

[0046] In the technical solution of the embodiment of the present application, by preprocessing the measured beam pair measurement results in two ways respectively, it takes into account both the overall characteristics and the individual characteristics of each receiving beam, enriches the considered feature dimensions, and makes the predicted result of the output target beam pair more stable and accurate.

[0047] Furthermore, the target beam prediction model further includes a first subsequent processing module and a second subsequent processing module. In step S23, the step of determining the predicted result of the target beam pair according to the beam processing result, the first environment processing result, and the final output module may include: Step S231, determining a first beam pair prediction result according to the first beam processing result, the first environment processing result, a preset matrix, and the first subsequent processing module; Step S232, determining a second beam pair prediction result according to the second beam processing result, the first environment processing result, and the second subsequent processing module; Step S233, inputting the first beam pair prediction result and the second beam pair prediction result into the final output module to output the predicted result of the target beam pair.

[0048] In the embodiment of the present application, it specifically discloses a method of, after obtaining the first beam processing result, the second beam processing result, and the first environment processing result, combining the first beam processing result and the second beam processing result with the environment processing result respectively to obtain two different beam pair prediction results and outputting the predicted result of the target beam pair according to the two beam pair prediction results.

[0049] Refer to Figure 7, the first subsequent processing module is last_blocks, which is used to perform subsequent processing on the splicing result of the first beam processing result and the first environment processing result, or the splicing result with a preset matrix; the second subsequent processing module is last_blocks_simple, which is used to perform subsequent processing on the subsequent result of the second beam processing result and the first environment processing result. It should be noted that the preset matrix can be a matrix of all 0s, which is used to be spliced with the first beam processing result alone in some processing processes to avoid the excessive influence of environmental information.

[0050] In the technical solution of the embodiment of the present application, after obtaining the first beam pair prediction result and the second beam pair prediction result, both the first beam pair prediction result and the second beam pair prediction result are input into the final_mlp module, and the final prediction result is generated through the final_mlp module, and the influence weights of the first beam pair prediction result and the second beam pair prediction result on the final target beam pair prediction result are measured respectively, so as to output an accurate target beam pair prediction result.

[0051] As a preference, in the final_mlp module, the weights of the first beam pair prediction result and the second beam pair prediction result are both 0.5, that is, the influence degrees of the first beam pair prediction result and the second beam pair prediction result on the finally obtained target beam pair prediction result are the same. In this way, a higher-precision output result can be obtained.

[0052] The input model processing process of the above two 8x4 beam pair measurement result processing schemes can be regarded as applying a fusion method inside the model, that is, a model fusion is performed on the overall predicted 64*4 beam pairs and the 64 beam pairs predicted on 4 receiving poles respectively, which enhances the stability and accuracy of the model output result.

[0053] Further, in step S231, the step of determining the first beam pair prediction result according to the first beam processing result, the first environment processing result, the preset matrix and the first subsequent processing module may include: Step A10, splicing the first beam processing result multiplied by 2 with the preset matrix to obtain a first splicing result; or Step A20, splicing the first beam processing result and the first environment processing result to obtain a first splicing result; Step A30, inputting the first splicing result into the first subsequent processing module to obtain a first beam pair prediction result.

[0054] In an embodiment of the present application, a method for processing selective drop (loss) information is provided. The first beam processing result can be concatenated with a preset matrix or a first environmental processing result according to a preset probability to obtain a first concatenated result that does not contain environmental information or contains environmental information. During the execution of the above steps, either step A10 to step A30 are executed, or step A20 to step A30 are executed. In one execution, only one of the options is selected.

[0055] This is because, during the process of predicting beam pairs, it is necessary to avoid overfitting the final beam pair prediction result with the environmental processing information to a certain extent. Therefore, a certain proportion of the first environmental processing results need to be discarded during the actual prediction process (including the training process of the model). The preset matrix can be a matrix of all zeros. When the first beam processing result is concatenated with the matrix of all zeros, the obtained first concatenated result only includes the prediction information based on the beam measurement results and does not include environmental information. Additionally, it should be noted that before concatenating with the matrix of all zeros, it needs to be multiplied by 2 to make the size of the obtained concatenated matrix consistent with the size of the first concatenated result obtained by concatenating the first beam processing result and the first environmental processing result. When the first beam processing result is concatenated with the matrix of all zeros, the first environmental processing result is discarded.

[0056] As a preference, through experiments, it is obtained that when the probabilities corresponding to executing step A10 and executing step A20 are both 50%, the accuracy of the final output target beam pair prediction result is relatively the highest.

[0057] In another feasible embodiment, the target beam prediction model further includes a simple environmental processing module. In step S232, the step of determining the second beam pair prediction result according to the second beam processing result, the first environmental processing result, and the second subsequent processing module may include: Step B10, inputting the first environmental processing result into the simple environmental processing module to obtain a second environmental processing result; Step B20, concatenating the second beam processing result and the second environmental processing result to obtain a second concatenated result; Step B30, inputting the second concatenated result into the second subsequent processing module to obtain a second beam pair prediction result.

[0058] Refer to Figure 7, the simple environment processing module is env_blocks_simple, which is used to further process the first environment processing result output by the environment and processing module so that the obtained second environment processing result corresponds to the second beam processing result in form for convenient CAT splicing. Since the second beam processing result is obtained by processing the measurement results of 4 separate groups of 8 beam pairs after splitting, finally, the result of the CAT splicing operation (the second splicing result) is output to the last_blocks_simple (the second subsequent processing module) module for processing to obtain the measurement results of 4 groups of 64 beam pairs, that is, a 64*4 beam pair prediction result.

[0059] In addition, it should be noted that the second beam pair prediction result is used to be jointly input into the final_mlp module together with the first beam pair prediction result for the final_mlp module to output an accurate target beam pair prediction result.

[0060] The embodiment of the present application provides a beam prediction method. First, the intensities of the received beams and the preset proportion of the transmitted beams corresponding to the received beams are collected to obtain the beam pair measurement results. Then, the beam pair measurement results and the current environment information are input into a preset target beam prediction model to obtain the target beam pair prediction result. Among them, the target beam pair prediction result at least includes the intensities of all the transmitted beams corresponding to the received beams, and the target beam prediction model is trained by multiple groups of received beam intensity data and corresponding transmitted beam intensity data under various transmission environments. The technical solution of the embodiment of the present application adopts a pre-trained target beam prediction model to predict the intensities of all the transmitted beams corresponding to the received beams according to the intensities of a partial proportion of the transmitted beams corresponding to the received beams, provides an accurate data basis for intelligent beam selection, realizes accurate and low-overhead intelligent beam prediction, saves the measurement overhead of beam scanning, reduces the communication cost, and solves the technical problem of excessive measurement overhead in beam measurement of MIMO communication.

[0061] In addition, since the current environment information and the target beam prediction model trained by multiple groups of received beam intensity data and corresponding transmitted beam intensity data under various transmission environments are introduced in the beam prediction method of the embodiment of the present application, the beam prediction method has a certain environment migration ability and can adapt to various different transmission environments.

[0062] Embodiment 2 Based on the first embodiment of the present application, in another embodiment of the present application, the same or similar content as that in the above-mentioned first embodiment can be referred to the above introduction and will not be repeated hereinafter. On this basis, a method for training a target beam prediction model is further provided. Before the step of inputting the beam pair measurement result and the current environment information into a preset target beam prediction model to obtain a target beam pair prediction result, the method may further include: Step C10, in a variety of different transmission environments, based on a preset antenna order set and a preset carrier frequency, collect received beam intensities, corresponding transmitted beam intensities, and environment information to obtain a training data set, where the antenna order set includes multiple different antenna order combinations; Step C20, train a preset initial beam prediction model through the training data set to obtain a target beam prediction weight and a target environment migration weight corresponding to the initial beam prediction model; Step C30, update the model parameters of the initial beam prediction model according to the beam prediction weight and the environment migration weight to obtain a target beam prediction model.

[0063] In the embodiment of the present application, multiple Massive-MIMO communications can be performed respectively based on a preset carrier frequency in different multiple transmission environments, and during the communication process, a comprehensive scan and collection of received beam intensities and corresponding transmitted beam intensities of all are performed to obtain intensity data of complete received beams and all transmitted beams corresponding to the received beams, and the current environment information is collected. The beam pair intensity data and the environment information at that time of each communication are used as a group to construct a training data set composed of multiple groups of beam pair intensity data in different multiple transmission environments.

[0064] It should be noted that when collecting transmitted beam intensities through an antenna, the data of the receiving antenna can be randomly disordered according to a preset antenna order combination, so that the obtained beam pair intensity data is disordered, achieving the effect of data enhancement to improve the generalization ability of the model. It can be understood that since there are situations where the data of 0 / 2 and 1 / 3 are close at the terminal reception pole, the data of the receiving antenna is randomly disordered during the model training process to obtain training set data with stronger generalization ability for the model.

[0065] Exemplarily, in each iteration process, the following antenna order set is cycled in a period of 8: non-disordered (i.e., normal order), [2,1,0,3], [0,3,2,1], [2,3,0,1], [1,0,3,2], [3,0,1,2], [1,2,3,0], [3,2,1,0].

[0066] In addition, before training the initial beam prediction model, first design the structure of the initial beam prediction model, specifically including each MixerBlock module in the beam prediction model. The model structure described in the previous application embodiment can be referred to and will not be elaborated here. Then, initialize each model parameter of the initial beam prediction model. The model parameters may include beam prediction weights (task1), carrier frequency migration prediction weights (task2), and environment migration prediction weights (task3), etc. These prediction weights together constitute all the model parameters of the initial beam prediction model. The beam prediction weights, carrier frequency migration prediction weights, and environment migration prediction weights respectively reflect the influence of beam measurement results, environment information, and carrier frequency on the final beam pair prediction result.

[0067] Refer to Figure 9 , when training the initial beam prediction model (i.e., the backbone prediction model), input each group of beam pair intensity data and environment information in the training dataset into the initial beam prediction model, and iteratively optimize the model parameters (task1 and task2) in the initial beam prediction model to obtain the final target beam prediction weights and target environment migration prediction weights. For carrier frequency migration (task2), first generate training data using beam prediction (task1) and environment migration (task3) data, and then train on the basis of the generated training data to obtain the carrier frequency migration prediction weights. In addition, it should be noted that the target beam prediction model trained based on the training dataset corresponding to the beam pair intensity data and environment information in different transmission environments under the same carrier frequency is only applicable to beam prediction in the communication system under this carrier frequency. If beam prediction is required at other carrier frequencies, a beam prediction model trained according to the training dataset corresponding to other carrier frequencies is needed.

[0068] After the target beam prediction weights and target environment migration prediction weights are determined, the corresponding model parameters in the initial beam prediction model can be replaced with the target beam prediction weights and target environment migration prediction weights, and the updated initial beam prediction model is the target beam prediction model.

[0069] Further, the step of training the preset initial beam prediction model through the training dataset to obtain the target beam prediction weights and target environment migration weights corresponding to the initial beam prediction model includes: Step C21, screen out multiple groups of input beam pair intensity data from the training dataset, where each group of beam pair intensity data includes a first preset number of received beam intensity data and corresponding second preset number of transmitted beam intensity data; Step C22: Input the beam pair intensity data and the corresponding environmental information into the initial beam prediction model, and perform prediction through the initial beam prediction model to output the prediction results of the third beam pair, the prediction results of the fourth beam pair, and the final beam pair prediction results; Step C23: Calculate the function losses corresponding to the prediction results of the third beam pair, the prediction results of the fourth beam pair, and the final beam pair prediction results respectively according to the prediction results of the third beam pair, the prediction results of the fourth beam pair, the final beam pair prediction results, and the beam pair intensity data; Step C24: Iteratively optimize the preset beam prediction weights and the preset environmental migration weights in the initial beam prediction model based on the function losses corresponding to the prediction results of the third beam pair, the prediction results of the fourth beam pair, and the final beam pair prediction results respectively to obtain the target beam prediction weights and the target environmental migration weights.

[0070] In the technical solution of the embodiment of the present application, after obtaining the training data set, screening is required, that is, multiple groups of beam pair intensity data are screened out. Among them, each group of beam pair intensity data includes the first preset number of received beam intensity data and the corresponding second preset number of transmitted beam intensity data. Among them, the first preset number is 4, that is, the number of received beams, and the second preset number is the number of a preset proportion of all transmitted beams corresponding to the received beams. When the preset proportion is 1 / 8, the second preset number is 8. Each group of beam pair intensity data is used as input data during model training to simulate the measurement results of some beam pairs in sparse scanning in actual applications.

[0071] After these groups of beam pair intensity data and the corresponding environmental information are input into the initial beam prediction model, the initial beam prediction model will output the corresponding prediction results. At this time, the complete beam pair intensity data (4*64 beam pair intensity data) corresponding to each group of beam pair intensity data in the training set data can be used as the true label to measure the accuracy of the prediction results and how much the function loss of the model is. After calculating the function loss, the model parameters can be iteratively optimized according to the change trend of the function loss, and whether to stop training can be determined according to indicators such as the convergence of the function loss or the prediction accuracy.

[0072] It should be noted that with reference to Figure 7, in the embodiments of the present application, the third beam prediction result and the fourth beam prediction result are the output results of the last_blocks_simple module and the last_blocks module in the initial beam prediction model respectively, and the final beam prediction result is the output result of the final_mlp model. In the embodiments of the present application, through the design of the multi-loss function structure, the function losses of multiple prediction results at different levels can be obtained, enriching the dimension of the function loss measurement, and can enhance the stability and prediction accuracy of the model. Among them, the calculation of the loss function can adopt various mature loss function calculation methods in the prior art, which will not be elaborated here. In addition, the output structure weight of the final_mlp module can be used as the loss function, and the training objective is to make the weights of the two output results of the final_mlp as close to 0.5 as possible.

[0073] In addition, it should be noted that in the process of training the initial beam prediction model through the function losses of three dimensions, one or more of the training enhancement means such as EMA (Exponential Moving Average), cosine learning rate decay, and model fusion can be used to train the initial beam prediction model, thereby establishing a relatively accurate intelligent beam prediction model. Among them, the model fusion method includes dividing and training the intensity data of a preset number (such as 10 groups) of random beam pairs, and then performing mean fusion on these prediction results, so as to ensure that the size of all fusion models does not exceed 200M and reduce the size of the model.

[0074] The embodiments of the present application provide a training method for a beam prediction model, which mainly trains an initial beam prediction model through beam pair intensity data and environmental information in a variety of transmission environments, obtains a target beam prediction model with environmental migration ability, and can adapt to a variety of different transmission environments. The generalization ability, stability and prediction accuracy of the target beam prediction model are enhanced through receiving antenna disordered data, multi-dimensional model parameters, multi-loss function structure and multi-training means, and a target beam prediction model with better performance is obtained. After practical tests, the average accuracy of the beam prediction model under beam prediction, carrier frequency migration and environmental migration reaches more than 0.93.

[0075] Embodiment 3 The embodiments of the present application further provide a beam prediction device, which is applied to an electronic device. Referring to Figure 10 , the beam prediction device includes: A beam pair measurement module 10, configured to collect the intensity of a received beam and a preset proportion of transmitted beams corresponding to the received beam, and obtain a beam pair measurement result; The beam pair prediction module 20 is configured to input the beam pair measurement result and the current environment information into a preset target beam prediction model to obtain a target beam pair prediction result, where the target beam pair prediction result at least includes the intensities of all the transmit beams corresponding to the receive beam, and the target beam prediction model is trained by multiple groups of receive beam intensity data and corresponding transmit beam intensity data under various transmission environments.

[0076] Optionally, the target beam prediction model at least includes a beam preprocessing module, an environment preprocessing module, and a final output module. The beam pair prediction module 20 is further configured to: Preprocess the beam pair measurement result through the beam preprocessing module and output a beam processing result; Preprocess the current environment information through the environment preprocessing module and output a first environment processing result; Determine the target beam pair prediction result according to the beam processing result, the first environment processing result, and the final output module.

[0077] Optionally, the beam preprocessing module includes a first beam preprocessing module and a second beam preprocessing module. The beam processing result includes a first beam processing result and a second beam processing result. The beam pair prediction module 20 is further configured to: Input the beam pair measurement result into the first beam preprocessing module, preprocess the beam pair measurement result through the first beam preprocessing module, and output a first beam processing result; Divide the beam pair measurement result into a first preset number of sub-beam pair measurement results, and input each of the sub-beam pair measurement results into the second beam preprocessing module respectively to output a second beam processing result.

[0078] Optionally, the target beam prediction model further includes a first subsequent processing module and a second subsequent processing module. The beam pair prediction module 20 is further configured to: Determine a first beam pair prediction result according to the first beam processing result, the first environment processing result, a preset matrix, and the first subsequent processing module; Determine a second beam pair prediction result according to the second beam processing result, the first environment processing result, and the second subsequent processing module; Input the first beam pair prediction result and the second beam pair prediction result into the final output module to output the target beam pair prediction result.

[0079] Optionally, the beam pair prediction module 20 is further configured to: Multiply the first beam processing result by 2 and splice it with a preset matrix to obtain a first splicing result; or Stitch the first beam processing result and the first environment processing result to obtain a first stitching result; Input the first stitching result into the first subsequent processing module to obtain a first beam pair prediction result.

[0080] Optionally, the target beam prediction model further includes a simple environment processing module, and the beam pair prediction module 20 is further configured to: Input the first environment processing result into the simple environment processing module to obtain a second environment processing result; Stitch the second beam processing result and the second environment processing result to obtain a second stitching result; Input the second stitching result into the second subsequent processing module to obtain a second beam pair prediction result.

[0081] Optionally, the beam prediction device further includes a model training module, and the model training module is further configured to: In a variety of different transmission environments, based on a preset antenna order set and a preset carrier frequency, collect received beam intensities, corresponding transmitted beam intensities, and environmental information to obtain a training data set, where the antenna order set includes multiple different antenna order combinations; Train a preset initial beam prediction model through the training data set to obtain a target beam prediction weight and a target environment migration weight corresponding to the initial beam prediction model; Update the model parameters of the initial beam prediction model according to the beam prediction weight and the environment migration weight to obtain a target beam prediction model.

[0082] Optionally, the model training module is further configured to: Select multiple groups of input beam pair intensity data from the training data set, where each group of beam pair intensity data includes a first preset number of received beam intensity data and corresponding second preset number of transmitted beam intensity data; Input each group of beam pair intensity data and corresponding environmental information into the initial beam prediction model, and perform prediction through the initial beam prediction model to output a third beam pair prediction result, a fourth beam pair prediction result, and a final beam pair prediction result; According to the third beam pair prediction result, the fourth beam pair prediction result, the final beam pair prediction result, and the beam pair intensity data, calculate the function losses corresponding to the third beam pair prediction result, the fourth beam pair prediction result, and the final beam pair prediction result respectively; Based on the function losses corresponding to the prediction results of the third beam pair, the fourth beam pair, and the final beam pair respectively, iteratively optimize the preset beam prediction weights and the preset environment migration weights in the initial beam prediction model to obtain the target beam prediction weights and the target environment migration weights.

[0083] The beam prediction device provided in this application adopts the beam prediction method in the above embodiment, and solves the technical problem of excessive measurement overhead in beam measurement of MIMO communication. Compared with the prior art, the beneficial effects of the beam prediction device provided in the embodiments of this application are the same as those of the beam prediction method provided in the above embodiment, and other technical features in this beam prediction device are the same as the features disclosed in the method of the previous embodiment, which will not be elaborated here.

[0084] Embodiment 4 The embodiments of this application provide an electronic device, which includes: at least one processor; and a memory communicatively linked to the at least one processor; wherein, the memory stores instructions executable by the at least one processor, and the instructions are executed by the at least one processor so that the at least one processor can execute the beam prediction method in Embodiment 1 above.

[0085] Next, refer to Figure 11 , which shows a schematic structural diagram of an electronic device suitable for implementing the embodiments of the present disclosure. The electronic devices in the embodiments of the present disclosure may include, but are not limited to, mobile terminals such as mobile phones, laptop computers, digital broadcast receivers, PDAs (personal digital assistants), PADs (tablet computers), PMPs (Portable Media Players), vehicle terminals (such as vehicle navigation terminals), etc., and fixed terminals such as digital TVs, desktop computers, etc. Figure 11 The electronic device shown is only an example and should not impose any limitation on the functions and usage scope of the embodiments of the present disclosure.

[0086] As Figure 11As shown, the electronic device may include a processing system 1001 (such as a central processing unit, a graphics processing unit, etc.), which may perform various appropriate actions and processes according to a program stored in a read-only memory 1002 (ROM, read only memory) or a program loaded from a storage system 1003 into a random access memory 1004 (RAM, random access memory). In the RAM 1004, various programs and data required for the operation of the electronic device are also stored. The processing system 1004, the ROM 1002, and the RAM 1004 are connected to each other through a bus 1005. An input / output (I / O) interface 1006 is also linked to the bus 1005.

[0087] Generally, the following systems may be linked to the I / O interface 1006: an input system 1007 including, for example, a touch screen, a touchpad, a keyboard, a mouse, an image sensor, a microphone, an accelerometer, a gyroscope, etc.; an output system 1008 including, for example, a liquid crystal display (LCD, liquid crystal display), a speaker, a vibrator, etc.; a storage system 1003 including, for example, a magnetic tape, a hard disk, etc.; and a communication system 1009. The communication system 1009 may allow the electronic device to communicate with other devices wirelessly or wiredly to exchange data. Although the figure shows an electronic device with various systems, it should be understood that it is not required to implement or have all the shown systems. More or fewer systems may be alternatively implemented or had.

[0088] In particular, according to an embodiment of the present disclosure, the process described above with reference to the flowchart may be implemented as a computer software program. For example, an embodiment of the present disclosure includes a computer program product, which includes a computer program carried on a computer-readable medium, and the computer program includes program codes for performing the method shown in the flowchart. In such an embodiment, the computer program may be downloaded and installed from a network through the communication system, or installed from the storage system, or installed from the ROM. When the computer program is executed by the processing system, the above functions defined in the method of the embodiment of the present disclosure are performed.

[0089] The electronic device provided in the present application adopts the beam prediction method in the above embodiment, and solves the technical problem of excessive measurement overhead in beam measurement of MIMO communication. Compared with the prior art, the beneficial effects of the electronic device provided in the embodiment of the present application are the same as those of the beam prediction method provided in the above Embodiment 1, and other technical features in the electronic device are the same as those disclosed in the method of the previous embodiment, and will not be elaborated here.

[0090] It should be understood that the various parts of the present disclosure can be implemented by hardware, software, firmware, or a combination thereof. In the description of the above embodiments, specific features, structures, materials, or characteristics can be combined in a suitable manner in any one or more embodiments or examples.

[0091] As described above, the above are only specific embodiments of the present application, but the protection scope of the present application is not limited thereto. Any person skilled in the art within the technical scope disclosed by the present application can easily think of changes or substitutions, which should all be covered within the protection scope of the present application. Therefore, the protection scope of the present application should be subject to the protection scope of the claims.

[0092] Embodiment Five This embodiment provides a computer-readable storage medium having computer-readable program instructions stored thereon, and the computer-readable program instructions are used to execute the beam prediction method in the above-mentioned Embodiment One.

[0093] The computer-readable storage medium provided by the embodiments of the present application can be, for example, a USB flash drive, but is not limited to electrical, magnetic, optical, electromagnetic, infrared, or semiconductor systems, devices, or any combination of the above. More specific examples of the computer-readable storage medium may include, but are not limited to: electrical links having one or more wires, portable computer disks, hard disks, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM, Erasable Programmable Read-Only Memory or flash memory), optical fibers, portable compact disk read-only memory (CD-ROM), optical storage devices, magnetic storage devices, or any suitable combination of the above. In this embodiment, the computer-readable storage medium can be any tangible medium that contains or stores a program, and the program can be used by or in combination with an instruction execution system, device, or device. The program code contained on the computer-readable storage medium can be transmitted by any appropriate medium, including but not limited to: wires, optical cables, RF (radio frequency), etc., or any suitable combination of the above.

[0094] The above computer-readable storage medium can be included in an electronic device; it can also exist separately without being assembled into the electronic device.

[0095] The above computer-readable storage medium carries one or more programs, which, when executed by an electronic device, cause the electronic device to: collect the intensity of a received beam and a preset proportion of transmitted beams corresponding to the received beam, and obtain a beam pair measurement result; input the beam pair measurement result and current environmental information into a preset target beam prediction model to obtain a target beam pair prediction result, where the target beam pair prediction result at least includes the intensity of all transmitted beams corresponding to the received beam, and the target beam prediction model is trained with multiple groups of received beam intensity data and corresponding transmitted beam intensity data under various transmission environments.

[0096] Computer program code for performing the operations of the present disclosure may be written in one or more programming languages or combinations thereof. The programming languages include object-oriented programming languages such as Java, Smalltalk, C++, and also include conventional procedural programming languages such as the "C" language or similar programming languages. The program code may execute entirely on the user's computer, partly on the user's computer, as a stand-alone software package, partly on the user's computer and partly on a remote computer, or entirely on the remote computer or server. In the case of a remote computer, the remote computer may be linked to the user's computer through any type of network, including a local area network (LAN) or a wide area network (WAN), or may be linked to an external computer (e.g., through the Internet using an Internet service provider).

[0097] The flowcharts and block diagrams in the accompanying drawings illustrate the possible architectures, functions, and operations of systems, methods, and computer program products according to various embodiments of the present application. In this regard, each block in the flowchart or block diagram may represent a module, a program segment, or a part of code that contains one or more executable instructions for implementing a specified logical function. It should also be noted that in some alternative implementations, the functions marked in the blocks may occur in a different order than marked in the accompanying drawings. For example, two consecutive blocks shown may actually be executed substantially in parallel, and they may sometimes be executed in the reverse order, depending on the functions involved. It should also be noted that each block in the block diagram and / or flowchart, and the combinations of blocks in the block diagram and / or flowchart, may be implemented by a dedicated hardware-based system for performing the specified functions or operations, or may be implemented by a combination of dedicated hardware and computer instructions.

[0098] The modules involved in the embodiments of the present disclosure can be implemented in software or in hardware. In some cases, the name of the module does not constitute a limitation on the unit itself.

[0099] The computer-readable storage medium provided by the present application stores computer-readable program instructions for executing the above beam prediction method, and solves the technical problem of excessive measurement overhead in beam measurement of MIMO communication. Compared with the prior art, the beneficial effects of the computer-readable storage medium provided by the embodiments of the present application are the same as those of the beam prediction method provided by the above embodiments, and will not be elaborated here.

[0100] Embodiment Six The present application further provides a computer program product, including a computer program, and the steps of the above beam prediction method are implemented when the computer program is executed by a processor.

[0101] The computer program product provided by the present application solves the technical problem of excessive measurement overhead in beam measurement of MIMO communication. Compared with the prior art, the beneficial effects of the computer program product provided by the embodiments of the present application are the same as those of the beam prediction method provided by the above embodiments, and will not be elaborated here.

[0102] The above are only the preferred embodiments of the present application, and do not limit the patent scope of the present application. Any equivalent structure or equivalent process transformation made by using the specification and drawings of the present application, or directly or indirectly applied in other related technical fields, shall be equally included in the patent scope of the present application.

Claims

1. A beam prediction method, characterized in that, the beam prediction method includes: collecting the intensities of a received beam and a preset proportion of transmitted beams corresponding to the received beam to obtain beam pair measurement results; inputting the beam pair measurement results and current environment information into a preset target beam prediction model to obtain a target beam pair prediction result, wherein the target beam pair prediction result at least includes the intensities of all transmitted beams corresponding to the received beam, and the target beam prediction model is trained by multiple groups of received beam intensity data and corresponding transmitted beam intensity data under various transmission environments.

2. The beam prediction method according to claim 1, characterized in that, the target beam prediction model at least includes a beam preprocessing module, an environment preprocessing module, and a final output module, and the step of inputting the beam pair measurement results and current environment information into the preset target beam prediction model to obtain a target beam pair prediction result includes: preprocessing the beam pair measurement results through the beam preprocessing module to output a beam processing result; preprocessing the current environment information through the environment preprocessing module to output a first environment processing result; determining the target beam pair prediction result according to the beam processing result, the first environment processing result, and the final output module.

3. The beam prediction method according to claim 2, characterized in that, the beam preprocessing module includes a first beam preprocessing module and a second beam preprocessing module, the beam processing result includes a first beam processing result and a second beam processing result, and the step of preprocessing the beam pair measurement results through the beam preprocessing module to output a beam processing result includes: inputting the beam pair measurement results into the first beam preprocessing module, and preprocessing the beam pair measurement results through the first beam preprocessing module to output a first beam processing result; dividing the beam pair measurement results into a first preset number of sub-beam pair measurement results, and respectively inputting each sub-beam pair measurement result into the second beam preprocessing module to output a second beam processing result.

4. The beam prediction method according to claim 3, characterized in that, the target beam prediction model further includes a first subsequent processing module and a second subsequent processing module, and the step of determining the target beam pair prediction result according to the beam processing result, the first environment processing result, and the final output module includes: determining a first beam pair prediction result according to the first beam processing result, the first environment processing result, a preset matrix, and the first subsequent processing module; determining a second beam pair prediction result according to the second beam processing result, the first environment processing result, and the second subsequent processing module; inputting the first beam pair prediction result and the second beam pair prediction result into the final output module to output a target beam pair prediction result.

5. The beam prediction method according to claim 4, characterized in that, The step of determining the first beam pair prediction result according to the first beam processing result, the first environment processing result, the preset matrix, and the first subsequent processing module includes: Concatenating the first beam processing result multiplied by 2 with the preset matrix to obtain a first concatenation result; or Concatenating the first beam processing result with the first environment processing result to obtain a first concatenation result; Inputting the first concatenation result into the first subsequent processing module to obtain a first beam pair prediction result.

6. The beam prediction method according to claim 4, wherein, the target beam prediction model further includes a simple environment processing module, and the step of determining the second beam pair prediction result according to the second beam processing result, the first environment processing result, and the second subsequent processing module includes: Inputting the first environment processing result into the simple environment processing module to obtain a second environment processing result; Concatenating the second beam processing result with the second environment processing result to obtain a second concatenation result; Inputting the second concatenation result into the second subsequent processing module to obtain a second beam pair prediction result.

7. The beam prediction method according to any one of claims 1-6, wherein, before the step of inputting the beam pair measurement result and the current environment information into a preset target beam prediction model to obtain a target beam pair prediction result, the method further includes: In a variety of different transmission environments, based on a preset antenna order set and a preset carrier frequency, collecting received beam intensities, corresponding transmitted beam intensities, and environment information to obtain a training data set, wherein the antenna order set includes multiple different antenna order combinations; Training a preset initial beam prediction model through the training data set to obtain a target beam prediction weight and a target environment migration weight corresponding to the initial beam prediction model; Updating the model parameters of the initial beam prediction model according to the beam prediction weight and the environment migration weight to obtain a target beam prediction model.

8. The beam prediction method according to claim 7, wherein, the step of training a preset initial beam prediction model through the training data set to obtain a target beam prediction weight and a target environment migration weight corresponding to the initial beam prediction model includes: Screening out multiple groups of input beam pair intensity data from the training data set, wherein each group of beam pair intensity data includes a first preset number of received beam intensity data and corresponding second preset number of transmitted beam intensity data; Inputting each beam pair intensity data and corresponding environment information into the initial beam prediction model, and performing prediction through the initial beam prediction model to output a third beam pair prediction result, a fourth beam pair prediction result, and a final beam pair prediction result; Calculating the function losses corresponding to the third beam pair prediction result, the fourth beam pair prediction result, and the final beam pair prediction result respectively according to the third beam pair prediction result, the fourth beam pair prediction result, the final beam pair prediction result, and the beam pair intensity data; Based on the function losses corresponding to the prediction results of the third beam pair, the fourth beam pair, and the final beam pair prediction results respectively, iteratively optimize the preset beam prediction weight and the preset environment migration weight in the initial beam prediction model to obtain the target beam prediction weight and the target environment migration weight.

9. An electronic device, characterized in that, the electronic device includes: at least one processor; and, a memory communicatively linked to the at least one processor; wherein, the memory stores instructions executable by the at least one processor, and the instructions are executed by the at least one processor to enable the at least one processor to execute the steps of the beam prediction method according to any one of claims 1 to 8.

10. A computer-readable storage medium, characterized in that, a program for implementing the beam prediction method is stored on the computer-readable storage medium, and the program for implementing the beam prediction method is executed by a processor to implement the steps of the beam prediction method according to any one of claims 1 to 8.

Images