Beam selection methods, non-volatile storage media, and computer equipment

Abstract

This invention discloses a beam selection method, a non-volatile storage medium, and a computer device. The method includes: obtaining the beam index of the selected beam at a previous reference time, and obtaining the index difference corresponding to the selected beam at the previous reference time, wherein the index difference corresponding to the selected beam at the previous reference time is the index difference between the selected beam at each previous reference time and the selected beam at its previous time, and the previous reference time is earlier than the current time; selecting multiple candidate beams from all beams based on the beam index and the index difference corresponding to the selected beam at the previous reference time; and selecting a target beam from the multiple candidate beams based on the measurement results of the communication capabilities of the multiple candidate beams. This invention solves the technical problem of slow beam transmission path search in related beam tracking algorithms.

Description

Technical Field

[0001] This invention relates to the field of communication technology, and more specifically, to a beam selection method, a non-volatile storage medium, and a computer device. Background Technology

[0002] With the continuous evolution of communication technology, the communication frequency bands of mobile communication networks have gradually shifted from the traditional low-frequency bands below 6GHz to high-frequency bands such as millimeter waves and terahertz. The scale of antenna arrays has also gradually increased from a few elements to hundreds or even thousands of elements. Large-scale antenna arrays can improve signal gain through beamforming technology, effectively improving the coverage and communication quality of high-frequency communication networks. However, the beamwidth of antenna arrays operating in the high-frequency band is often narrow, requiring efficient beam tracking algorithms to ensure that the beams of the transceiver terminals are aligned to avoid communication interruptions.

[0003] When a terminal moves, the optimal beam of the terminal device may change. Therefore, it is necessary to track the optimal beam of the terminal device and switch to the optimal beam for communication to avoid degradation of communication quality. In related technologies, in order to find the beam with the best communication quality between the terminal and the antenna array, it is necessary to measure the communication quality of all beams and then select the optimal beam. This process is slow and wasteful of resources.

[0004] There is currently no effective solution to the above problems. Summary of the Invention

[0005] This invention provides a beam selection method, a non-volatile storage medium, and a computer device to at least solve the technical problem of slow beam transmission path search in related beam tracking algorithms.

[0006] According to one aspect of the present invention, a beam selection method is provided, comprising: obtaining a beam index of a selected beam at a previous reference time, and obtaining an index difference corresponding to the selected beam at the previous reference time, wherein the index difference corresponding to the selected beam at the previous reference time is the index difference between the selected beam at the previous reference time and the selected beam at the previous time, the previous reference time being earlier than the current time, the beam index of the selected beam at the previous reference time being used to estimate the search value of all configurable beams of the antenna array in a multipath hopping situation, and the index difference corresponding to the selected beam at the previous reference time being used to estimate the search value of all beams in a single-path continuous movement situation; selecting a plurality of candidate beams from the all beams based on the beam index of the selected beam at the previous reference time and the index difference corresponding to the selected beam at the previous reference time; and selecting a target beam from the plurality of candidate beams based on a measurement result of the communication capability of the plurality of candidate beams, wherein the target beam is used for communication at the current time.

[0007] Optionally, obtaining the beam index of the selected beam of the terminal at the preceding reference time and obtaining the index difference corresponding to the selected beam at the preceding reference time includes: when the preceding reference time includes at least one first time, obtaining the first beam selected by the terminal at the first time and the second beam selected by the terminal at a second time, wherein the second time corresponds one-to-one with the first time and the second time is the previous time of the corresponding first time; determining the index difference corresponding to each of the first times included in the preceding reference time according to the first index of the first beam and the second index of the second beam, wherein the index difference corresponding to each of the first times is the index difference between the first beam selected at the first time and the corresponding second beam.

[0008] Optionally, selecting multiple candidate beams from all configurable beams of the antenna array based on the beam index of the selected beam at the preceding reference time and the index difference corresponding to the selected beam at the preceding reference time includes: inputting the index difference corresponding to the selected beam at the preceding reference time into a single-path differential neural network to obtain the single-path search value of each of the beams, wherein the single-path differential neural network is pre-trained based on single-path differential training samples; inputting the beam index of the selected beam at the preceding reference time into a multipath hopping neural network to obtain the multipath hopping search value of each of the beams, wherein the multipath hopping neural network is pre-trained based on multipath hopping training samples; determining the beam search value of each of the beams based on the single-path search value and the multipath hopping search value; and selecting the multiple candidate beams from all the beams based on the beam search value of each of the beams.

[0009] Optionally, determining the beam search value of each of the beams based on the single-path search value and the multipath hop search value includes: determining the weight coefficients corresponding to the beam indices of the selected beams at the previous time step of the multiple current time steps based on the pre-trained target correspondence, wherein the beam indices and weight coefficients in the target correspondence are in one-to-one correspondence, and the weight coefficients represent the ratio between the single-path search value and the multipath hop search value; and combining the single-path search value and the multipath hop search value of each of the beams based on the weight coefficients to form the beam search value of each of the beams.

[0010] Optionally, the target correspondence is obtained by updating it in the following way: acquiring historical samples corresponding to historical moments, wherein each historical sample includes a historical index difference at the corresponding historical moment, a beam index of a historical selected beam, and a measurement result of the historical selected beam, wherein the historical index difference is the index difference between the historical selected beam selected at the previous moment and the selected beam selected at the two previous moments of the historical moment for each historical sample; determining the initial weight coefficients corresponding to each historical sample according to the initial correspondence, wherein the initial correspondence is the correspondence before updating the target correspondence; determining the historical beam search value of each historical selected beam for each historical sample according to the single-path differential neural network, the multipath jump neural network, and the initial weight coefficients; updating the initial weight coefficients according to the historical beam search value and the measurement result of the historical selected beam to obtain the weight coefficients corresponding to each historical selected beam; updating the initial correspondence according to the beam index and the weight coefficients of each historical selected beam to obtain the target correspondence.

[0011] Optionally, determining the historical beam search value of each historical selected beam of the historical samples based on the single-path difference neural network and the multipath jump neural network includes: inputting the historical index difference of each historical sample into the single-path difference neural network to obtain the sample single-path search value of each historical selected beam of the historical samples; inputting the beam index of each historical selected beam of the historical samples into the multipath jump neural network to obtain the sample multipath jump search value of each historical selected beam of the historical samples; and determining the historical beam search value of each historical selected beam of the historical samples based on the sample single-path search value, the sample multipath jump search value, and the initial weight coefficient.

[0012] Optionally, updating the initial weight coefficients based on the historical beam search values ​​to obtain the weight coefficients corresponding to each of the historical selected beams includes: constructing a quasi-target value vector for each of the historical samples' historical selected beams, wherein the quasi-target value vector is positively correlated with the measurement results of the historical selected beams; determining a target value vector for each of the historical samples' historical selected beams based on the quasi-target value vectors and the historical beam search values; determining a first value vector difference based on the sample's single-path search value and the sample's multi-path jump search value; determining a second value vector difference based on the historical beam search values ​​and the target value vectors; and updating the initial weight coefficients based on the first value vector difference and the second value vector difference to obtain the weight coefficients corresponding to each of the historical selected beams.

[0013] Optionally, before selecting the target beam from the plurality of candidate beams, the method further includes: performing a communication capability measurement on the plurality of candidate beams to obtain the measurement result.

[0014] According to another aspect of the present invention, a beam selection device is also provided, comprising: an acquisition module, configured to acquire the beam index of a selected beam at a previous reference time, and to acquire the index difference corresponding to the selected beam at the previous reference time, wherein the index difference corresponding to the selected beam at the previous reference time is the index difference between the selected beam at the previous reference time and the selected beam at the previous time, the previous reference time being earlier than the current time, the beam index of the selected beam at the previous reference time being used to estimate the search value of all configurable beams of the antenna array in a multipath hopping situation, and the index difference corresponding to the selected beam at the previous reference time being used to estimate the search value of all beams in a single-path continuous movement situation; a first selection module, configured to select a plurality of candidate beams from the all beams based on the beam index of the selected beam at the previous reference time and the index difference corresponding to the selected beam at the previous reference time; and a second selection module, configured to select a target beam from the plurality of candidate beams based on the measurement results of the communication capabilities of the plurality of candidate beams, wherein the target beam is used for communication at the current time.

[0015] According to another aspect of the present invention, a non-volatile storage medium is also provided, the non-volatile storage medium including a stored program, wherein, when the program is executed, the device where the non-volatile storage medium is located is controlled to perform any of the beam selection methods described above.

[0016] According to another aspect of the present invention, a computer device is also provided, the computer device including a memory and a processor, the memory being used to store a program, and the processor being used to run the program stored in the memory, wherein the program, when running, executes any of the beam selection methods described above.

[0017] In this embodiment of the invention, by utilizing the beam index and index difference of the beam selected by the terminal at a previous reference time before the current time, multiple candidate beams that are more likely to include the optimal beam at the current time are first selected from all beams. Then, the target beam with the best communication quality at the current time is obtained by further filtering from the multiple candidate beams. This achieves the purpose of quickly filtering out the beam channel with the best communication quality at the current time, thereby realizing the technical effect of speeding up the search for the best beam transmission path at the current time, and thus solving the technical problem of slow beam transmission path search in related beam tracking algorithms. Attached Figure Description

[0018] The accompanying drawings, which are included to provide a further understanding of the invention and form part of this application, illustrate exemplary embodiments of the invention and, together with their description, serve to explain the invention and do not constitute an undue limitation thereof. In the drawings:

[0019] Figure 1 A hardware block diagram of a computer terminal for implementing a beam selection method is shown.

[0020] Figure 2 This is a schematic flowchart of the beam selection method provided according to an embodiment of the present invention;

[0021] Figure 3 This is a schematic diagram of the beam selection network provided in an optional embodiment of the present invention;

[0022] Figure 4 This is a schematic diagram illustrating the tracking effect of a beam selection method provided according to an optional embodiment of the present invention;

[0023] Figure 5 This is a schematic diagram illustrating the tracking effect of beam tracking algorithms related to this technology;

[0024] Figure 6 This is a signal-to-noise ratio diagram of a beam selection method provided according to an optional embodiment of the present invention;

[0025] Figure 7 This is a structural block diagram of a beam selection device provided according to an embodiment of the present invention. Detailed Implementation

[0026] To enable those skilled in the art to better understand the present invention, the technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings of the embodiments of the present invention. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort should fall within the scope of protection of the present invention.

[0027] It should be noted that the terms "first," "second," etc., in the specification, claims, and accompanying drawings of this invention 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 the invention 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 a 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.

[0028] First, some nouns or terms that appear in the description of the embodiments of this application shall be interpreted as follows:

[0029] A beam is a method of concentrating transmitted electromagnetic waves in one direction to form a stream of energy. By controlling the direction and angle of electromagnetic wave transmission or reception, beams can be used to achieve directional transmission and reception of signals, improving the efficiency and accuracy of communication systems. For example, beam technology is commonly used in radar, satellite communications, and other fields for target detection and data transmission.

[0030] In a multiple-input multiple-output (MIMO) communication system, a beam index is a number used to identify a set of independent beam signals transmitted from a transmit antenna array. These independent beam signals can be used to achieve directional transmission or beamforming by adjusting the phase and amplitude on each transmit antenna.

[0031] According to an embodiment of the present invention, a method embodiment for selecting a beam is provided. It should be noted that the steps shown in the flowchart in the accompanying drawings can be executed in a computer system such as a set of computer-executable instructions. Furthermore, although a logical order is shown in the flowchart, in some cases the steps shown or described may be executed in a different order than that shown here.

[0032] The methods and embodiments provided in this application can be executed on mobile terminals, computer terminals, or similar computing devices. Figure 1 A hardware block diagram of a computer terminal for implementing a beam selection method is shown. Figure 1As shown, the computer terminal 10 may include one or more processors (shown as processors 102a, 102b, ..., 102n in the figure) (the processor may include, but is not limited to, a microprocessor MCU or a programmable logic device FPGA, etc.) and a memory 104 for storing data. 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 1 The structure shown is for illustrative purposes only and does not limit the structure of the aforementioned electronic device. For example, computer terminal 10 may also include... Figure 1 The more or fewer components shown, or having the same Figure 1 The different configurations shown.

[0033] It should be noted that the aforementioned one or more processors and / or other data processing circuits are generally referred to herein as "data processing circuits". These data processing circuits may be implemented wholly or partially as software, hardware, firmware, or any other combination thereof. Furthermore, the data processing circuits may be a single, independent processing module, or may be wholly or partially integrated into any other element in the computer terminal 10. As involved in the embodiments of this application, the data processing circuits serve as processor control (e.g., selection of a variable resistor termination path connected to an interface).

[0034] The memory 104 can be used to store software programs and modules of application software, such as the program instructions / data storage device corresponding to the beam selection method in the embodiments of the present invention. The processor executes various functional applications and data processing by running the software programs and modules stored in the memory 104, thereby implementing the beam selection method of the application program described above. The memory 104 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 104 may further include memory remotely located relative to the processor, and these remote memories can be connected to the computer terminal 10 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.

[0035] The display may be, for example, a touchscreen liquid crystal display (LCD) that allows the user to interact with the user interface of the computer terminal 10.

[0036] In related technologies, to find the beam with the best communication quality between the terminal and the antenna array, it is usually necessary to measure the communication quality of all beams that the terminal can currently receive, and then select the optimal beam. This process is slow and wasteful of resources. Alternatively, existing beam tracking methods often do not consider the impact of multipath effects on beam tracking in real-world environments. Especially during terminal movement, due to changes in the positional relationship between the transmitting and receiving ends and other objects in the environment, the transmission path pointed to by the beam may be blocked, or a new transmission path with better communication quality may appear. In this case, if the beam tracking algorithm cannot switch to the better transmission path in time, it will cause a decrease in communication quality.

[0037] To address the problem of how a terminal can quickly select a communication beam, this application provides a beam selection method. Figure 2 This is a flowchart illustrating the beam selection method provided in an embodiment of the present invention, as shown below. Figure 2 As shown, the method includes the following steps:

[0038] Step S202: Obtain the beam index of the selected beam at the previous reference time and obtain the index difference corresponding to the selected beam at the previous reference time. The index difference corresponding to the selected beam at the previous reference time is the index difference between the selected beam at the previous reference time and the selected beam at the previous time. The previous reference time is earlier than the current time. The beam index of the selected beam at the previous reference time is used to estimate the search value of all configurable beams of the antenna array in the case of multipath hopping. The index difference corresponding to the selected beam at the previous reference time is used to estimate the search value of all beams in the case of single-path continuous movement.

[0039] In this step, the terminal can be a portable mobile device, such as a smartphone or tablet, or other types of terminal devices that require beamforming for communication. When the terminal's location changes, the beam it receives will also change, and the communication quality of different beams for that terminal may also change. For example, the beam with the best communication quality for the terminal at the previous moment may, at the current moment, experience a decrease in communication quality due to obstacles blocking the connection between the terminal and the antenna array caused by changes in the terminal's location.

[0040] In this embodiment, the beam selected by the terminal at a certain moment is the beam selected by the terminal for communication at that moment; that is, the selected beam is the target beam used by the terminal for communication at that moment. The selected beam is obtained by filtering from all configurable beams of the antenna array. All configurable beams of the antenna array can refer to all beams of the transmitting antenna array that can be configured and used to transmit signals. The terminal can pre-store the relevant information of all beams in its own device before executing this beam selection method. The beam index of each beam is the signal number assigned by the antenna array to all beams, and the beam index of each configurable beam of the antenna array does not change with the location of the terminal.

[0041] It should be noted that the preceding reference time is a time prior to the current time. Preferably, the preceding reference time can be several times with equal time intervals prior to the current time. The selected beam at the preceding reference time is the beam selected by the terminal at that preceding reference time for communication with the antenna array. This beam is usually the beam with the best communication quality between the terminal and the antenna array at the preceding reference time. The method by which the terminal determines the selected beam at the preceding reference time can be the same as the beam selection method provided in this application for selecting the target beam at the current time. The time intervals between preceding reference times and between the preceding reference time and the current time can be equal. The specific value of the time interval can depend on the time step of the beam selection algorithm, that is, how often the terminal executes the beam selection algorithm.

[0042] The communication quality of different beams can be assessed using commonly used channel quality evaluation methods in the industry. For example, after measuring different beams, the beam with less energy attenuation can be considered as the beam with better communication quality, or the beam with a higher signal-to-noise ratio can be considered as the beam with better communication quality. Alternatively, other beam quality evaluation methods can be used, without specific limitations here.

[0043] As an optional embodiment, obtaining the beam index of the selected beam at the preceding reference time and the index difference corresponding to the selected beam at the preceding reference time can be done in the following way: when the preceding reference time includes at least one first time, obtain the first beam selected by the terminal at the first time and the second beam selected by the terminal at the second time, wherein the second time corresponds one-to-one with the first time and the second time is the previous time of the corresponding first time; determine the index difference corresponding to each of the first times included in the preceding reference time according to the first index of the first beam and the second index of the second beam, wherein the index difference corresponding to each of the first times is the index difference between the first beam selected at the first time and the corresponding second beam.

[0044] It should be noted that the first beam can be the beam selected by the terminal from all beams at the first moment for communicating with the antenna array, and the second beam can be the beam selected by the terminal from all beams at the second moment for communicating with the antenna array. Optionally, the first beam can be the beam with the best communication quality selected by the terminal at the first moment, and the second beam can be the beam with the best communication quality selected by the terminal at the second moment. The way the terminal selects the first beam at the first moment and the second beam at the second moment can be the same as the beam selection method for selecting the target beam at the current moment provided in this application. The number of first beams and second beams is the same and they correspond one-to-one. Each second beam is the selected beam of a first beam at the previous moment. Therefore, the first beams and second beams are exactly staggered, and the index difference corresponding to each first beam can be determined.

[0045] Optionally, the index difference corresponding to each first beam can be determined as follows: Based on the first index of the first beam and the second index of the second beam, the index difference between the first beam and its corresponding second beam is determined. The first index and the second index are the beam indices of the first beam and the second beam, respectively. Optionally, the first index and the second index in this step can be expressed as (x, y), where x represents the horizontal coordinate of the beam index and y represents the vertical coordinate of the beam index. The antenna array can number the beams based on the positional relationship between the transmitting antennas of each beam, obtaining a beam index in the form of (x, y). Therefore, the first index of the first beam can be recorded as (x1, y1), and the second index of the second beam can be recorded as (x2, y2). The index difference between the first beam and the second beam can then be determined based on the difference between the horizontal and vertical coordinates of the beam indices. For example, the index difference between the first beam and the second beam can be denoted as (x1-x2, y1-y2), and the index difference can be represented as a vector.

[0046] The principle behind using the index difference between the first beam and the second beam to predict the target beam is that this index difference can characterize the terminal's motion direction to some extent. For example, the index difference between the first beam and the second beam can be roughly considered as a vector characterizing the terminal's motion direction and distance during the time interval from the second moment to the first moment. The possibility of a significant change in the terminal's motion state between the second moment and the first moment, and between the first moment and the current moment, is very small. Therefore, it can be assumed that the index difference between the first and second indices is likely close to the index difference between the beam index of the target beam with the best communication quality at the current moment and the first index. This allows for direct prediction of the target beam's beam index, or it can narrow down the possible range of searching for the target beam among all beams at the current moment. For example, if there are 100 beams, and the terminal measures the communication quality of each of the 100 beams and selects the beam with the best communication quality as the target beam, then 100 measurement actions are required. This process is time-consuming and consumes the terminal's measurement resources. Using the beam selection method described above, the beam search range at the current moment can be estimated based on the first index of the first beam and the index difference between the first beam and the second index. The beam search range can include several beams from all beams, for example, 10 beams. The index difference of each of the 10 beams relative to the first beam is the same as or close to the index difference between the first beam and the second beam. This means the target beam with the best communication quality at the current moment is most likely to appear within this beam search range. The index difference of each of the 10 beams relative to the first beam is the difference between the beam index of each of the 10 beams and the beam index of the first beam. The terminal can then measure the communication quality of only several beams within this beam search range to obtain the beam with the best communication quality among those beams, and use this beam as the terminal's target beam at the current moment. This process significantly reduces the time and resource consumption of searching for a target beam among all beams, thus improving the efficiency of searching for the beam with the best communication quality at the current moment.

[0047] Step S204: Select multiple candidate beams from all beams based on the beam index of the selected beam at the previous reference time and the index difference corresponding to the selected beam at the previous reference time.

[0048] As an optional embodiment, multiple candidate beams at the current time can be selected from all beams in the following manner: The index difference corresponding to the selected beam at the previous reference time is input into a single-path difference neural network to obtain the single-path search value of each beam, wherein the single-path difference neural network is pre-trained based on single-path difference training samples; the beam index of the selected beam at the previous reference time is input into a multipath jump neural network to obtain the multipath jump search value of each beam, wherein the multipath jump neural network is pre-trained based on multipath jump training samples; the beam search value of each beam is determined based on the single-path search value and the multipath jump search value; and multiple candidate beams are selected from all beams based on the beam search values ​​of each beam.

[0049] Optionally, the single-path difference neural network and the multipath hopping neural network can be in parallel. The input to the single-path difference neural network can be the index difference between the first beam and the second beam, which can be denoted as the beam index difference vector Δs. The network output is the single-path search value v1 for all beams under continuous single-path movement. The input to the multipath hopping neural network can be the beam index s of the first beam, and the network output is the multipath hopping search value v2 for all beams under multipath hopping. The structures of the single-path difference neural network and the multipath hopping neural network can employ fully connected networks, locally connected networks, recurrent neural networks, convolutional neural networks, or other variations of the above network structures.

[0050] By paralleling a single-path differential neural network and a multipath hopping neural network, the beam search value of each beam is divided into two relatively independent parts: the single-path search value under the condition of continuous movement along a single transmission path, and the multipath hopping search value under the condition of hopping between multiple transmission paths. The single-path search value is the search value of the beam as a single-path beam if no multipath hopping occurs; the multipath hopping search value is the search value of the beam as a multipath hopping beam if no multipath hopping occurs.

[0051] Continuous movement along a single transmission path mainly occurs in open, unobstructed environments, such as outdoor plazas and stadiums. In this case, the main factors determining the change in beam index of the beam with the best communication quality at different times are the speed and direction of the terminal's movement. Therefore, relevant information can be extracted from the index difference Δs between the first beam and the second beam to determine which beams have statistically higher single-path search value.

[0052] Jumping between multiple transmission paths mainly occurs in environments with many obstructions and reflectors, such as offices and workshops. When the line-of-sight path between the base station and the terminal is blocked, a new communication link can be established by jumping to a reflective path on a wall or other object surface. Clearly, jumping between multiple transmission paths depends primarily on the relative positions of various obstructions, reflective surfaces, and the communication terminal in the environment. Assuming that the positions of objects in the communication environment other than the user terminal remain constant over a period of time, the current location of the terminal can be used to statistically determine whether the current path is likely to be blocked or whether a new, better transmission path is possible. In this case, the speed and direction of the terminal's movement are not the primary influencing factors. The terminal's position can be represented by the first index s of the first beam. Optionally, to simplify the network structure and accelerate computation, the multipath hopping neural network can use the first index s of the first beam as input to estimate, statistically speaking, the search value of each beam as a multipath hopping beam at the current location of the user terminal.

[0053] It should be noted that, for the terminal, it cannot determine whether each beam in all the beams it receives is a single-path beam or a multipath hopping beam. Therefore, the beam search value of any beam can be synthesized from the single-path search value and the multipath search value of the beam to comprehensively characterize the possibility that the beam is the target beam at the current moment.

[0054] As an optional implementation, when determining the beam search value of each beam based on the single-path search value and the multipath hop search value, the following method can be adopted: Based on the target correspondence obtained through pre-training, determine the weight coefficient corresponding to the beam index of the selected beam at the previous time step, wherein the beam index and weight coefficient in the target correspondence are in one-to-one correspondence, and the weight coefficient represents the ratio between the single-path search value and the multipath hop search value; based on the weight coefficient, combine the single-path search value and the multipath hop search value of each beam to form the beam search value of each beam.

[0055] Optionally, the pre-trained target correspondence can be a beam weight mapping table, which stores the mapping relationship between different beam indices and weight coefficients. The single-path search value v1 and multi-path hop search value v2 of any beam can form a combined value matrix [v1 v2] of the search values ​​of all beams, and the weight coefficient α corresponding to each beam is determined based on its respective beam index. s Therefore, the beam search value v of any all beams can be synthesized based on this. Optionally, the synthesis formula can be: v = α s v1+(1-α s v2, where 0 < α s <1.

[0056] As an optional implementation, a value hybrid network can be constructed after the single-path difference neural network and the multipath jump neural network to synthesize the beam search value of each beam in the entire beam set. The input to the value hybrid network can be the single-path search value and the multipath jump search value of all beams, in the form of a combined value matrix [v1 v2], and the output is the beam search value v of all beams. The value hybrid network can consist of a single convolutional layer with a kernel size of 1×2 and weights of [α...]. s 1-α s ], where 0 < α s <1.

[0057] The value hybrid network weights and sums the search values ​​of each beam under continuous movement along a single transmission path and under multiple transmission path hopping scenarios to give the beam search value. It's worth noting that hopping between multiple transmission paths generally only occurs at specific locations, such as the edge of an obstruction. At other locations in space, transmission paths generally do not hop abruptly; instead, they exhibit continuous changes along a single transmission path, such as the continuous movement of the line-of-sight path as the terminal moves in an open environment. To distinguish between these two scenarios, when the value hybrid network sums the single-path search value and the multi-path hopping search value, it can access the respective weight coefficients α of each value through the beam index s and the beam weight mapping table. s 1-α s The weighting coefficients are related to the user's location, and the beam index s can be considered to reflect the terminal's location. For example, when the terminal is located at the edge of an obstruction, a larger weighting coefficient needs to be assigned to the multipath hopping neural network, i.e., a smaller α is used. s In other locations, a larger weight coefficient can be given to the single-path difference neural network, that is, a larger α can be used. s To achieve the above objective, the weights α of the convolutional kernels in the value hybrid network... s Instead of being a fixed value, it is obtained from the beam weighting map table, which can have a total of N key-value pairs, represented as {s n →α s,n}, n = 1, 2, ..., N. Where s n α represents the beam index of the nth beam. s,n This refers to the weighting coefficient corresponding to the beam index of the nth beam. When calculating the beam search value of a beam, the weighting coefficient α can be obtained from the beam weight mapping table based on the beam index s. s The value is then passed through the convolution kernel [α]. s 1-α s The beam search value of the beam is calculated.

[0058] As an optional embodiment, the above-mentioned target correspondence can be obtained by updating it in the following way: Obtain historical samples corresponding to each historical moment, wherein each historical sample includes the historical index difference at the corresponding historical moment, the beam index of the historical selected beam, and the measurement result of the historical selected beam. The historical index difference is the index difference between the historical selected beam selected at the previous moment and the selected beam selected at the two moments before the historical moment for each historical sample. Based on the initial correspondence, determine the initial weight coefficients corresponding to each historical sample, wherein the initial correspondence is the correspondence before updating the target correspondence. Based on the single-path difference neural network, the multipath jump neural network, and the initial weight coefficients, determine the historical beam search value of the historical selected beam for each historical sample. Update the initial weight coefficients based on the historical beam search value and the measurement result of the historical selected beam to obtain the weight coefficients corresponding to each historical selected beam. Update the initial correspondence based on the beam index and the weight coefficients of each historical selected beam to obtain the target correspondence.

[0059] Based on this optional embodiment, continuous updates to the target correspondence can be achieved, enabling the updated target correspondence to better reflect the characteristics of the scenario in which the terminal is located. Optionally, the initial correspondence can be the target correspondence at the previous moment, and the target correspondence at the current moment can be the initial correspondence at the next moment. The update of the correspondence can be a continuous iterative process. For the moment following the current moment, the current moment is also one of the historical moments. Therefore, the historical sample corresponding to the current moment can be determined based on the index difference at the current moment, the beam index of the target beam, and the measurement results of the target beam. Optionally, for the current moment, the historical moment can include a previous reference moment, or it can include a first moment and a second moment.

[0060] It should be noted that the historical selected beam included in the historical samples can be the beam with the best terminal communication quality at the corresponding historical moment, and the historical index difference included in the historical samples can be the index difference between the beam with the best communication quality at the previous moment and the beam with the best communication quality at the two moments before the corresponding historical moment. The determination method of the historical index difference and the historical selected beam can adopt the method of determining the index difference and the target beam in the beam selection method proposed in this application.

[0061] As an optional embodiment, the historical beam search value of each historical sample can be determined based on the single-path difference neural network and the multipath jump neural network by the following steps: inputting the historical index difference of each historical sample into the single-path difference neural network to obtain the sample single-path search value of each historical sample; inputting the beam index of the selected historical beam of each historical sample into the multipath jump neural network to obtain the sample multipath jump search value of each historical sample; and determining the historical beam search value based on the sample single-path search value, the sample multipath jump search value, and the initial weight coefficient.

[0062] As an optional embodiment, when updating the initial weight coefficients, the initial weight coefficients can be updated based on the historical beam search values ​​to obtain the weight coefficients corresponding to each historical selected beam. This process may include the following steps: constructing quasi-target value vectors for each historical sample, wherein the quasi-target value vectors are positively correlated with the measurement results of the historical selected beams; determining the target value vectors for each historical sample based on the quasi-target value vectors and the historical beam search values; determining a first value vector difference based on the sample single-path search value and the sample multi-path jump search value; determining a second value vector difference based on the historical beam search values ​​and the target value vectors; and updating the initial weight coefficients based on the first value vector difference and the second value vector difference to obtain the weight coefficients corresponding to each historical selected beam.

[0063] Given M historical samples, quasi-target value vectors can be constructed for each of the M historical samples. Optionally, the M quasi-target value vectors can satisfy the following condition: if the measurement result p of beam i i Measurement results p that are superior to those of beam j j ,So If beam k is not measured, then The historical beam search value can be represented as v', and then it can be determined based on the quasi-target value vector. Construct the target value vector v for each historical sample based on the historical beam search value v'. tar ,For example, Where β is a constant factor, 0 ≤ β ≤ 1. Further, the single-path search value of a sample can be denoted as v'1, and the multi-path jump search value can be denoted as v'2. Then, the difference in the first value vector is v'1 - v'2, and the difference in the second value vector can be v' - v'2. tar The initial weight coefficients can be updated based on the difference between the first and second value vectors using the following formula: α s =min(max(α') s -η(v'1-v'2) T (v'-v tar ),0),1), where η is a constant factor, α's α is the initial weighting coefficient in this formula. s Select the corresponding weighting coefficients for each beam in the history.

[0064] Step S206: Based on the measurement results of the communication capabilities of multiple candidate beams, a target beam is selected from the multiple candidate beams. The target beam is used for communication at the current moment. It can be considered that the beam with the best communication quality that the terminal can use at the current moment is the target beam. Therefore, after the target beam is selected, the terminal can use the target beam for communication at the current moment.

[0065] As an optional embodiment, before selecting the target beam from multiple candidate beams, the following steps may be included: measuring the communication capabilities of the multiple candidate beams and obtaining the measurement results.

[0066] In this optional embodiment, the number of the top-ranked beams can be determined according to a predetermined threshold. For example, when the total number of beams is 100, the top 10 beams can be selected to obtain 10 candidate beams. Then, the communication capability of the 10 candidate beams can be measured. This can solve the technical problem of wasting time and measurement resources caused by measuring all beams in order to select the target beam.

[0067] In the above steps, by using the beam index and index difference of the beam selected by the terminal at the previous reference time before the current time, multiple candidate beams that are more likely to include the optimal beam at the current time are first selected from all beams. Then, the target beam with the best communication quality at the current time is obtained by further filtering from the multiple candidate beams. This achieves the purpose of quickly filtering out the beam channel with the best communication quality at the current time, thereby realizing the technical effect of speeding up the search for the best beam transmission path at the current time. This solves the technical problem of slow beam transmission path search in related beam tracking algorithms.

[0068] The above embodiments and optional embodiments provide an intelligent beam selection method suitable for multipath environments. The method has the ability to learn autonomously and explore the communication environment, effectively meeting the beam switching requirements in multipath environments and ensuring the robustness of the communication system and high-speed data transmission.

[0069] Optionally, the single-path difference neural network and the multi-path jump neural network can be continuously trained and iterated during the execution of the above beam selection method. For example, they can be trained based on the target value vectors v of multiple historical samples. tar In addition to the corresponding data of multiple historical samples, training samples are constructed for both the single-path difference neural network and the multi-path jump neural network. The single-path difference training samples can be {Δs', v}. 1,tarThe training samples for multipath jumps can be {s', v}. 2,tar}, where Δs' is the historical index difference of the historical samples, s' is the beam index of the historical selected beam for each historical sample, and v 1,tar =α h v tar v 2,tar =(1-α) h )v tar α h These are the weighting coefficients corresponding to historical samples.

[0070] Furthermore, the single-path difference training samples {Δs',v} corresponding to M historical samples can be used. 1,tar For training a single-path difference neural network, the loss function can be the mean square error function. Or other loss functions such as cross-entropy.

[0071] Furthermore, using the obtained M multipath jump training samples {s', v 2,tar For training a multipath hopping neural network, the loss function can be the mean squared error function. Or other loss functions such as cross-entropy.

[0072] Based on the above beam selection method, the present invention provides the following optional embodiments: Figure 3 This is a schematic diagram of the beam selection network provided in an optional embodiment of the present invention, as shown below. Figure 3 As shown, the beam selection network consists of three parts: a parallel single-path differential neural network and a multipath hopping neural network, as well as a value mixing network. Based on this beam selection network, the following steps can be achieved:

[0073] Step S1: Construct a single-path differential neural network. The number of input nodes can be determined by the length of the beam index differential sequence, and the number of output nodes is M, satisfying M≤N, where N is the total number of beams at the current time. The network structure of the single-path differential neural network can adopt different forms such as fully connected, locally connected, LSTM, and convolutional networks.

[0074] A multipath hopping neural network can be constructed. The network can have two input nodes, which are the beam index components in the vertical and horizontal directions, respectively. The number of output nodes is K, where K ≤ N. The network structure of the multipath hopping neural network can take different forms, such as fully connected, locally connected, LSTM, and convolutional networks.

[0075] A value hybrid network is constructed, with 2N input nodes and N output nodes, and includes convolutional layers. The output layers of the single-path difference neural network and the multipath jump neural network are zero-padded to form N×1 dimensional vectors, which serve as inputs to the value hybrid network. The convolutional kernels are of size 2, and each kernel takes the i-th element of the N×1 dimensional vector output by the single-path difference neural network and the multipath jump neural network as input. The weights of the convolutional kernels are obtained from the beam weight mapping table based on the beam index.

[0076] Step S2: The communication terminal inputs the beam index and index difference of the selected beam at the previous reference time before the current time into the beam selection network. The beam selection network outputs the beam search values ​​of all N beams, and then selects several beams with high beam search values ​​for search measurement. Based on the measurement results, the target beam with the best communication capability is selected for data transmission, and the measurement result of the target beam is saved to the sample pool. The measurement result of the target beam is P. A new sample {Δs,s,P} can be constructed based on the relevant data of the target beam, and this sample {Δs,s,P} is saved to the sample pool for training the single-path difference neural network, the multipath jump neural network, and the beam weight mapping table.

[0077] Optionally, the communication terminal can obtain the search value of M beams out of N beams under the condition of continuous single-path variation through a single-path differential neural network. The search values ​​of the remaining NM beams are padded with zeros to obtain the complete N×1 dimensional value vector v1. M beams are selected to speed up the beam search. M can also be equal to N, that is, to determine the beam search value of each of the N beams.

[0078] The communication terminal obtains the search value of K beams under multipath hopping conditions through a multipath hopping neural network. The search values ​​of the remaining NK beams are padded with zeros to obtain the complete N×1 dimensional value vector v2. K beams are selected to speed up the beam search. K can also be equal to N, that is, to determine the beam search value of each beam among all N beams.

[0079] The N×2 dimensional combined value matrix [v1 v2] is input into the second-level value mixing network, and the weights α are obtained from the beam weight mapping table according to the absolute index s of the beam. s The value, using the convolution kernel [α] s 1-α s Adding each row of the combined value matrix together yields an N×1 dimensional total value vector v = α. s v1+(1-α s v2, obtain the beam search value of each of the N beams.

[0080] Step S3: The communication terminal can take several samples from the sample pool to train and update the network weights of the beam selection network.

[0081] In one specific embodiment, the simulation results of tracking the base station's transmit beam using the beam selection algorithm proposed in this invention are as follows: Figure 4 As shown, Figure 5 This is a schematic diagram illustrating the tracking effect of a beam tracking algorithm related to this technology. The beam tracking period is 1 second, the user terminal's speed is 1 m / s, the base station is suspended at a height of 3 m, and the user moves randomly on a plane at a height of 0 m. Multiple transmission paths exist in the environment, requiring switching between these paths at specific locations. Simulation results show that... Figure 4 Within the three horizontal coordinate ranges of 0-50, 50-100, and 300-350, the beam selection method of this invention, compared to other comparative methods, can effectively support multipath switching of the beam, ensure the continuity of data communication, and achieve better tracking performance. Furthermore... Figure 5 Within the three horizontal coordinate ranges of 0–50, 50–100, and 300–350, the tracking beam index showed a significant drift from the ideal beam index, resulting in poor tracking performance.

[0082] In another specific embodiment, the simulation results of tracking the base station's transmit beam using the algorithm of the present invention are as follows: Figure 6 As shown in the figure. The simulation conditions are the same as in the previous embodiment. Simulation results show that, under different signal-to-noise ratio conditions, the algorithm described in this invention can significantly reduce beam tracking error compared to other comparative algorithms, thereby effectively improving communication quality.

[0083] It should be noted that, for the sake of simplicity, the foregoing method embodiments are all described as a series of actions. However, those skilled in the art should understand that the present invention is not limited to the described order of actions, because according to the present invention, some steps can be performed in other orders or simultaneously. Furthermore, those skilled in the art should also understand that the embodiments described in the specification are preferred embodiments, and the actions and modules involved are not necessarily essential to the present invention.

[0084] Through the above description of the embodiments, those skilled in the art can clearly understand that the beam selection method according to the above embodiments can be implemented by means of software plus necessary general-purpose hardware platform. Of course, it can also be implemented by hardware, but in many cases the former is a better implementation method. Based on this understanding, the technical solution of the present invention, or the part that contributes to the prior art, can be embodied in the form of a software product. This computer software product is stored in a storage medium (such as ROM / RAM, magnetic disk, optical disk) and includes several instructions to cause a terminal device (which may be a mobile phone, computer, server, or network device, etc.) to execute the methods of the various embodiments of the present invention.

[0085] According to embodiments of the present invention, a beam selection device for implementing the above-described beam selection method is also provided. Figure 7 This is a structural block diagram of a beam selection device provided according to an embodiment of the present invention, such as... Figure 7 As shown, the beam selection device includes: an acquisition module 72, a first selection module 74, and a second selection module 76. The beam selection device will be described below.

[0086] The acquisition module 72 is used to acquire the beam index of the selected beam at the previous reference time and the index difference corresponding to the selected beam at the previous reference time. The index difference corresponding to the selected beam at the previous reference time is the index difference between the selected beam at the previous reference time and the selected beam at the previous time. The previous reference time is earlier than the current time. The beam index of the selected beam at the previous reference time is used to estimate the search value of all configurable beams of the antenna array in the case of multipath hopping. The index difference corresponding to the selected beam at the previous reference time is used to estimate the search value of all beams in the case of single-path continuous movement.

[0087] The first selection module 74 is connected to the acquisition module 72 and is used to select multiple candidate beams from all beams based on the beam index of the selected beam at the previous reference time and the index difference corresponding to the selected beam at the previous reference time.

[0088] The second selection module 76, connected to the first selection module 74, is used to select a target beam from multiple candidate beams based on the measurement results of the communication capabilities of multiple candidate beams, wherein the target beam is used for communication at the current moment.

[0089] It should be noted that the aforementioned acquisition module 72, first selection module 74, and second selection module 76 correspond to steps S202 to S206 in the embodiments. Multiple modules and their corresponding steps implement the same instances and application scenarios, but are not limited to the content disclosed in the above embodiments. It should also be noted that the aforementioned modules, as part of the device, can run on the computer terminal 10 provided in the embodiments.

[0090] Embodiments of the present invention may provide a computer device. Optionally, in this embodiment, the computer device may be located in at least one of a plurality of network devices in a computer network. The computer device includes a memory and a processor.

[0091] The memory can be used to store software programs and modules, such as the program instructions / modules corresponding to the beam selection method and apparatus in this embodiment of the invention. The processor executes various functional applications and data processing by running the software programs and modules stored in the memory, thereby realizing the aforementioned beam selection method. The memory 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 may further include memory remotely located relative to the processor, and these remote memories can be connected to a computer terminal 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.

[0092] The processor can invoke information and application programs stored in memory via the transmission device to perform the following steps: obtaining the beam index of the selected beam at the previous reference time and obtaining the index difference corresponding to the selected beam at the previous reference time, wherein the index difference corresponding to the selected beam at the previous reference time is the index difference between the selected beam at the previous reference time and the selected beam at the previous time, the previous reference time is earlier than the current time, the beam index of the selected beam at the previous reference time is used to estimate the search value of all configurable beams of the antenna array in the case of multipath hopping, and the index difference corresponding to the selected beam at the previous reference time is used to estimate the search value of all beams in the case of single-path continuous movement; selecting multiple candidate beams from all beams based on the beam index of the selected beam at the previous reference time and the index difference corresponding to the selected beam at the previous reference time; selecting a target beam from multiple candidate beams based on the measurement results of the communication capabilities of the multiple candidate beams, wherein the target beam is used for communication at the current time.

[0093] Optionally, the processor may also execute program code for the following steps: obtaining the beam index of the selected beam at the preceding reference time and obtaining the index difference corresponding to the selected beam at the preceding reference time, including: when the preceding reference time includes at least one first time, obtaining the first beam selected by the terminal at the first time and the second beam selected by the terminal at the second time, wherein the second time corresponds one-to-one with the first time and the second time is the previous time of the corresponding first time; determining the index difference corresponding to each of the first times included in the preceding reference time according to the first index of the first beam and the second index of the second beam, wherein the index difference corresponding to each of the first times is the index difference between the first beam selected at the first time and the corresponding second beam.

[0094] Optionally, the processor may also execute program code for the following steps: selecting multiple candidate beams from all configurable beams of the antenna array based on the beam index of the selected beam at the previous reference time and the index difference corresponding to the selected beam at the previous reference time, including: inputting the index difference corresponding to the selected beam at the previous reference time into a single-path differential neural network to obtain the single-path search value of each beam, wherein the single-path differential neural network is pre-trained based on single-path differential training samples; inputting the beam index of the selected beam at the previous reference time into a multipath hopping neural network to obtain the multipath hopping search value of each beam, wherein the multipath hopping neural network is pre-trained based on multipath hopping training samples; determining the beam search value of each beam based on the single-path search value and the multipath hopping search value; and selecting multiple candidate beams from all beams based on the beam search value of each beam.

[0095] Optionally, the processor may also execute program code for the following steps: determining the beam search value of each beam based on the single-path search value and the multipath hop search value, including: determining the weight coefficients corresponding to the beam indices of the selected beams at the previous time step for multiple current time steps based on the target correspondence obtained through pre-training, wherein the beam indices and weight coefficients in the target correspondence are in one-to-one correspondence, and the weight coefficients represent the ratio between the single-path search value and the multipath hop search value; and combining the single-path search value and the multipath hop search value of each beam into the beam search value of each beam based on the weight coefficients.

[0096] Optionally, the processor may also execute program code with the following steps: the target correspondence is obtained by updating it as follows: acquiring historical samples corresponding to historical moments, wherein the historical samples include the historical index difference at the corresponding historical moment, the beam index of the historical selected beam, and the measurement result of the historical selected beam, and the historical index difference is the index difference between the historical selected beam selected at the previous moment and the selected beam selected at the two moments before the historical moment corresponding to each historical sample; determining the initial weight coefficients corresponding to each historical sample according to the initial correspondence, wherein the initial correspondence is the correspondence before updating the target correspondence; determining the historical beam search value of the historical selected beam of each historical sample according to the single-path difference neural network, the multipath jump neural network, and the initial weight coefficients; updating the initial weight coefficients according to the historical beam search value and the measurement result of the historical selected beam to obtain the weight coefficients corresponding to each historical selected beam; updating the initial correspondence according to the beam index and the weight coefficients of each historical selected beam to obtain the target correspondence.

[0097] Optionally, the processor may also execute program code for the following steps: determining the historical beam search value of each historical selected beam for each historical sample based on the single-path difference neural network and the multipath jump neural network, including: inputting the historical index difference of each historical sample into the single-path difference neural network to obtain the sample single-path search value of each historical selected beam for each historical sample; inputting the beam index of each historical selected beam for each historical sample into the multipath jump neural network to obtain the sample multipath jump search value of each historical selected beam for each historical sample; and determining the historical beam search value of each historical selected beam for each historical sample based on the sample single-path search value, the sample multipath jump search value, and the initial weight coefficient.

[0098] Optionally, the processor may also execute program code for the following steps: updating the initial weight coefficients based on historical beam search values ​​to obtain the weight coefficients corresponding to each historical selected beam, including: constructing a quasi-target value vector for each historical selected beam of each historical sample, wherein the quasi-target value vector is positively correlated with the measurement results of the historical selected beam; determining the target value vector for each historical selected beam of each historical sample based on the quasi-target value vector and the historical beam search values; determining a first value vector difference based on the sample single-path search value and the sample multi-path jump search value; determining a second value vector difference based on the historical beam search values ​​and the target value vector; and updating the initial weight coefficients based on the first value vector difference and the second value vector difference to obtain the weight coefficients corresponding to each historical selected beam.

[0099] Optionally, the processor may also execute program code that includes the following steps: before selecting the target beam from multiple candidate beams, it further includes: measuring the communication capabilities of the multiple candidate beams and obtaining the measurement results.

[0100] Those skilled in the art will understand that all or part of the steps in the various methods of the above embodiments can be implemented by a program instructing the hardware related to the terminal device. The program can be stored in a non-volatile storage medium, which may include: flash drive, read-only memory (ROM), random access memory (RAM), magnetic disk or optical disk, etc.

[0101] Embodiments of the present invention also provide a non-volatile storage medium. Optionally, in this embodiment, the aforementioned non-volatile storage medium can be used to store the program code executed by the beam selection method provided in the above embodiments.

[0102] Optionally, in this embodiment, the non-volatile storage medium may be located in any computer terminal in a group of computer terminals in a computer network, or in any mobile terminal in a group of mobile terminals.

[0103] Optionally, in this embodiment, the non-volatile storage medium is configured to store program code for performing the following steps: obtaining the beam index of the selected beam at the previous reference time and obtaining the index difference corresponding to the selected beam at the previous reference time, wherein the index difference corresponding to the selected beam at the previous reference time is the index difference between the selected beam at the previous reference time and the selected beam at the previous time, the previous reference time is earlier than the current time, the beam index of the selected beam at the previous reference time is used to estimate the search value of all configurable beams of the antenna array in the case of multipath hopping, and the index difference corresponding to the selected beam at the previous reference time is used to estimate the search value of all beams in the case of single-path continuous movement; selecting multiple candidate beams from all beams based on the beam index of the selected beam at the previous reference time and the index difference corresponding to the selected beam at the previous reference time; selecting a target beam from multiple candidate beams based on the measurement results of the communication capabilities of the multiple candidate beams, wherein the target beam is used for communication at the current time.

[0104] Optionally, in this embodiment, the non-volatile storage medium is configured to store program code for performing the following steps: obtaining the beam index of the selected beam of the terminal at the preceding reference time, and obtaining the index difference corresponding to the selected beam at the preceding reference time, including: when the preceding reference time includes at least one first time, obtaining the first beam selected by the terminal at the first time and the second beam selected by the terminal at the second time, wherein the second time corresponds one-to-one with the first time, and the second time is the previous time of the corresponding first time; determining the index difference corresponding to each of the first times included in the preceding reference time according to the first index of the first beam and the second index of the second beam, wherein the index difference corresponding to each of the first times is the index difference between the first beam selected at the first time and the corresponding second beam.

[0105] Optionally, in this embodiment, the non-volatile storage medium is configured to store program code for performing the following steps: selecting multiple candidate beams from all configurable beams of the antenna array based on the beam index of the selected beam at the previous reference time and the index difference corresponding to the selected beam at the previous reference time, including: inputting the index difference corresponding to the selected beam at the previous reference time into a single-path differential neural network to obtain the single-path search value of each beam, wherein the single-path differential neural network is pre-trained based on single-path differential training samples; inputting the beam index of the selected beam at the previous reference time into a multipath hopping neural network to obtain the multipath hopping search value of each beam, wherein the multipath hopping neural network is pre-trained based on multipath hopping training samples; determining the beam search value of each beam based on the single-path search value and the multipath hopping search value; and selecting multiple candidate beams from all beams based on the beam search value of each beam.

[0106] Optionally, in this embodiment, the non-volatile storage medium is configured to store program code for performing the following steps: determining the beam search value of each beam based on the single-path search value and the multipath hop search value, including: determining the weight coefficients corresponding to the beam indices of the selected beams at the previous time step for multiple current time steps based on the target correspondence obtained through pre-training, wherein the beam indices and weight coefficients in the target correspondence are in one-to-one correspondence, and the weight coefficients characterize the ratio between the single-path search value and the multipath hop search value; and combining the single-path search value and the multipath hop search value of each beam into the beam search value of each beam based on the weight coefficients.

[0107] Optionally, in this embodiment, the non-volatile storage medium is configured to store program code for performing the following steps: the target correspondence is obtained after being updated in the following manner: acquiring historical samples corresponding to historical moments, wherein the historical samples respectively include the historical index difference at the corresponding historical moment, the beam index of the historical selected beam, and the measurement result of the historical selected beam, the historical index difference being the index difference between the historical selected beam selected at the previous moment and the selected beam selected at the two moments before the historical moment corresponding to each historical sample; determining the initial weight coefficients corresponding to each historical sample according to the initial correspondence, wherein the initial correspondence is the correspondence before updating the target correspondence; determining the historical beam search value of the historical selected beam of each historical sample according to the single-path difference neural network, the multipath jump neural network, and the initial weight coefficients; updating the initial weight coefficients according to the historical beam search value and the measurement result of the historical selected beam to obtain the weight coefficients corresponding to each historical selected beam; updating the initial correspondence according to the beam index and the weight coefficient of each historical selected beam to obtain the target correspondence.

[0108] Optionally, in this embodiment, the non-volatile storage medium is configured to store program code for performing the following steps: determining the historical beam search value of each historical selected beam of each historical sample based on a single-path difference neural network and a multipath jump neural network, including: inputting the historical index difference of each historical sample into the single-path difference neural network to obtain the sample single-path search value of each historical selected beam of each historical sample; inputting the beam index of each historical selected beam of each historical sample into the multipath jump neural network to obtain the sample multipath jump search value of each historical selected beam of each historical sample; and determining the historical beam search value of each historical selected beam of each historical sample based on the sample single-path search value, the sample multipath jump search value, and the initial weight coefficient.

[0109] Optionally, in this embodiment, the non-volatile storage medium is configured to store program code for performing the following steps: updating the initial weight coefficients based on the historical beam search value to obtain the weight coefficients corresponding to each of the historical selected beams, including: constructing a quasi-target value vector for each of the historical selected beams of the historical samples, wherein the quasi-target value vector is positively correlated with the measurement results of the historical selected beams; determining the target value vector for each of the historical selected beams of the historical samples based on the quasi-target value vector and the historical beam search value; determining a first value vector difference based on the sample single-path search value and the sample multi-path jump search value; determining a second value vector difference based on the historical beam search value and the target value vector; and updating the initial weight coefficients based on the first value vector difference and the second value vector difference to obtain the weight coefficients corresponding to each of the historical selected beams.

[0110] Optionally, in this embodiment, the non-volatile storage medium is configured to store program code for performing the following steps: before selecting a target beam from multiple candidate beams, the method further includes: measuring the communication capabilities of the multiple candidate beams to obtain measurement results.

[0111] The sequence numbers of the above embodiments of the present invention are for descriptive purposes only and do not represent the superiority or inferiority of the embodiments.

[0112] In the above embodiments of the present invention, 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.

[0113] 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.

[0114] 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.

[0115] Furthermore, the functional units in the various embodiments of the present invention 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.

[0116] 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 non-volatile storage medium. Based on this understanding, the technical solution of the present invention, 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 the present invention. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, read-only memory (ROM), random access memory (RAM), portable hard drives, magnetic disks, or optical disks.

[0117] The above description is only a preferred embodiment of the present invention. It should be noted that for those skilled in the art, several improvements and modifications can be made without departing from the principle of the present invention, and these improvements and modifications should also be considered within the scope of protection of the present invention.

Claims

1. A method of beam selection, the method comprising: include: The method obtains the beam index of the selected beam at the previous reference time and the index difference corresponding to the selected beam at the previous reference time. The index difference corresponding to the selected beam at the previous reference time is the index difference between the selected beam at the previous reference time and the selected beam at the previous time. The previous reference time is earlier than the current time. The beam index of the selected beam at the previous reference time is used to estimate the search value of all configurable beams of the antenna array in the case of multipath hopping. The index difference corresponding to the selected beam at the previous reference time is used to estimate the search value of all beams in the case of single-path continuous movement. Based on the beam index of the selected beam at the preceding reference time and the index difference corresponding to the selected beam at the preceding reference time, multiple candidate beams are selected from all beams. Based on the measurement results of the communication capabilities of the plurality of candidate beams, a target beam is selected from the plurality of candidate beams, wherein the target beam is used for communication at the current time.

2. The method according to claim 1, characterized in that, The step of obtaining the beam index of the selected beam of the terminal at the preceding reference time, and obtaining the index difference corresponding to the selected beam at the preceding reference time, includes: When the preceding reference time includes at least one first time, the first beam selected by the terminal at the first time and the second beam selected by the terminal at the second time are obtained, wherein the second time corresponds one-to-one with the first time and the second time is the previous time of the corresponding first time. Based on the first index of the first beam and the second index of the second beam, the index difference corresponding to each of the first times included in the preceding reference time is determined, wherein the index difference corresponding to each of the first times is the index difference between the selected first beam and the corresponding second beam at the first time.

3. The method according to claim 1, characterized in that, Based on the beam index of the selected beam at the preceding reference time and the index difference corresponding to the selected beam at the preceding reference time, multiple candidate beams are selected from all configurable beams of the antenna array, including: The index difference corresponding to the selected beam at the previous reference time is input into the single-path difference neural network to obtain the single-path search value of each of the beams. The single-path difference neural network is pre-trained based on the single-path difference training samples. The beam index of the selected beam at the previous reference time is input into the multipath hopping neural network to obtain the multipath hopping search value of each of the beams. The multipath hopping neural network is pre-trained based on multipath hopping training samples. The beam search value of each of the beams is determined based on the single-path search value and the multipath hop search value. Based on the beam search value of each of the beams, the plurality of candidate beams are selected from the total number of beams.

4. The method according to claim 3, characterized in that, Based on the single-path search value and the multipath hop search value, the beam search value of each of the beams is determined, including: Based on the target correspondence obtained through pre-training, the weight coefficient corresponding to the beam index of the selected beam in the previous time step is determined. In the target correspondence, the beam index and the weight coefficient are in one-to-one correspondence, and the weight coefficient represents the ratio between the single-path search value and the multi-path jump search value. Based on the weighting coefficients, the single-path search value and multipath hop search value of each of the beams are combined into the beam search value of each of the beams.

5. The method according to claim 4, characterized in that, The target correspondence is obtained by updating it in the following way: Obtain historical samples corresponding to each historical moment, wherein each historical sample includes the historical index difference at the corresponding historical moment, the beam index of the historical selected beam, and the measurement result of the historical selected beam. The historical index difference is the index difference between the historical selected beam selected at the previous historical moment and the selected beam selected at the two previous historical moments for each historical sample. Based on the initial correspondence, the initial weight coefficients corresponding to each of the historical samples are determined, wherein the initial correspondence is the correspondence before the target correspondence is updated; Based on the single-path differential neural network, the multipath jump neural network, and the initial weight coefficients, the historical beam search value of each historical sample for its historical selected beam is determined. The initial weight coefficients are updated based on the historical beam search value and the measurement results of the historical selected beams to obtain the weight coefficients corresponding to each of the historical selected beams. The initial correspondence is updated based on the beam index and weight coefficient of each of the historically selected beams to obtain the target correspondence.

6. The method according to claim 5, characterized in that, Based on the single-path differential neural network and the multi-path jump neural network, the historical beam search value of each historical sample for its respective historical selected beam is determined, including: The historical index difference of each of the historical samples is input into the single-path difference neural network to obtain the sample single-path search value of each of the historical selected beams of the historical samples. The beam index of the historical selected beam of each historical sample is input into the multipath jump neural network to obtain the sample multipath jump search value of the historical selected beam of each historical sample. Based on the sample single-path search value, the sample multipath jump search value, and the initial weight coefficient, the historical beam search value of each historical sample for its selected historical beam is determined.

7. The method according to claim 6, characterized in that, The initial weight coefficients are updated based on the historical beam search value to obtain the weight coefficients corresponding to each of the historical selected beams, including: Construct quasi-target value vectors for each of the historical selected beams of the historical samples, wherein the quasi-target value vectors are positively correlated with the measurement results of the historical selected beams; Based on the quasi-target value vector and the historical beam search value, the target value vector of the historical selected beam for each of the historical samples is determined; The first value vector difference is determined based on the single-path search value and the multi-path jump search value of the sample. The second value vector difference is determined based on the historical beam search value and the target value vector; The initial weight coefficients are updated based on the first value vector difference and the second value vector difference to obtain the weight coefficients corresponding to each of the historically selected beams.

8. The method according to any one of claims 1 to 7, characterized in that, Before selecting the target beam from the plurality of candidate beams, the process further includes: The communication capability of the multiple candidate beams is measured to obtain the measurement results.

9. A non-volatile storage medium, characterized in that, The non-volatile storage medium includes a stored program, wherein, when the program is executed, it controls the device containing the non-volatile storage medium to perform the beam selection method according to any one of claims 1 to 8.

10. A computer device, characterized in that, The computer device includes a memory and a processor, the memory being used to store a program, and the processor being used to run the program stored in the memory, wherein the program, when running, executes the beam selection method according to any one of claims 1 to 8.

Images