Beam prediction method and device and storage medium
By acquiring beam pairs with low discrimination in intelligent beam prediction and inputting them into the target beam prediction model, the problem of poor convergence effect of neural network models is solved, and the accuracy of beam prediction is improved.
Patent Information
- Application Number
- CN202310558980.6
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2023-05-17
- Publication Date
- 2024-11-22
AI Technical Summary
In intelligent beam prediction, due to the low beam distinction of some groups in the training data set, the convergence effect of the neural network model is poor, affecting the accuracy of beam prediction.
By acquiring a plurality of first prediction beam pairs, wherein the first prediction beam pair has a correlation degree value between the RSRP received powers of different reference signals, the correlation degree value and RSRP are input into the target beam prediction model, and the correlation degree value and RSRP are obtained. The second prediction beam pairs the corresponding RSRP, and uses the mapping relationships learned by the target beam prediction model to perform intelligent beam prediction.
It effectively improves the accuracy of beam prediction, can intelligently predict beam pairs with unknown reference signal reception power values, and improves the beam's ability to identify reference signal reception power values in the data set.
Smart Images

Figure CN119012350A_ABST
Abstract
Description
Technical Field
[0001] The present application relates to the field of communication technology, and in particular to a beam prediction method, device and storage medium. Background Art
[0002] In the related art, the training of the intelligent beam prediction model used in intelligent beam prediction is usually to input all groups of data in the training data set into the neural network model, train the neural network model, and adjust the model parameters to obtain the trained intelligent beam prediction model.
[0003] In this way, since the discrimination of some group beam training data in the training data set is very low, it can be regarded as a kind of data fuzziness, which affects the recognition effect of the neural network in identifying the strongest beam, affects the convergence of the neural network model, and the convergence effect of the intelligent beam prediction model is not good, thus affecting the accuracy of beam prediction. Summary of the invention
[0004] The present application aims to solve one of the technical problems in the related art at least to some extent.
[0005] To this end, the purpose of the present application is to propose a beam prediction method, device and processor-readable storage medium.
[0006] The first aspect embodiment of the present application proposes a beam prediction method, including: obtaining multiple first prediction beam pairs, wherein the first prediction beam pairs have a first reference signal received power RSRP, and the correlation value between different first RSRPs is less than a correlation threshold; inputting the correlation value and the multiple first RSRPs into a target beam prediction model to obtain a second RSRP corresponding to a second prediction beam pair, wherein the target beam prediction model has learned the mapping relationship between the correlation value, the multiple first RSRPs, and the second RSRP.
[0007] The second aspect embodiment of the present application proposes a beam prediction device, including: an acquisition unit, used to acquire multiple first prediction beam pairs, wherein the first prediction beam pairs have a first reference signal received power RSRP, and the correlation value between different first RSRPs is less than a correlation threshold; a processing unit, used to input the correlation value and the multiple first RSRPs into a target beam prediction model to obtain a second RSRP corresponding to the second prediction beam pair, wherein the target beam prediction model has learned the mapping relationship between the correlation value, the multiple first RSRPs, and the second RSRP.
[0008] The third aspect embodiment of the present application proposes a beam prediction device, including: a memory, a transceiver, and a processor: a memory, used to store computer programs; a transceiver, used to send and receive data under the control of the processor; a processor, used to obtain multiple first predicted beam pairs, wherein the first predicted beam pairs have a first reference signal received power RSRP, and the correlation value between different first RSRPs is less than a correlation threshold; the correlation value and the multiple first RSRPs are input into a target beam prediction model to obtain a second RSRP corresponding to the second predicted beam pair, wherein the target beam prediction model has learned the mapping relationship between the correlation value, the multiple first RSRPs, and the second RSRP.
[0009] The processor-readable storage medium proposed in the fourth aspect embodiment of the present application stores a computer program, and the computer program is used to enable the processor to execute: the beam prediction method proposed in the first aspect embodiment of the present application.
[0010] The beam prediction method, device and processor-readable storage medium proposed in the present application obtain multiple first predicted beam pairs, wherein the first predicted beam pairs have a first reference signal received power RSRP, and the correlation value between different first RSRPs is less than a correlation threshold; the correlation value and the multiple first RSRPs are input into a target beam prediction model to obtain a second RSRP corresponding to the second predicted beam pair, wherein the target beam prediction model has learned the mapping relationship between the correlation value, the multiple first RSRPs, and the second RSRP, and can perform intelligent beam prediction processing on a beam pair data set for beam pairs with unknown reference signal received power values to obtain reference signal received power values of all beam pairs in the beam pair data set. Since the target beam prediction model used is trained based on data after directivity enhancement processing, the accuracy of beam prediction can be effectively improved.
[0011] Additional aspects and advantages of the present application will be given in part in the description below, and in part will become apparent from the description below, or will be learned through the practice of the present application. BRIEF DESCRIPTION OF THE DRAWINGS
[0012] The above and / or additional aspects and advantages of the present application will become apparent and easily understood from the following description of the embodiments in conjunction with the accompanying drawings, in which:
[0013] Figure 1 is a flowchart of a beam prediction method proposed in an embodiment of the present application;
[0014] Figure 2 is a schematic diagram of the structure of a communication system in an embodiment of the present application;
[0015] Figure 3is a flowchart of a beam prediction method proposed in another embodiment of the present application;
[0016] Figure 4 is a flowchart of a beam prediction method proposed in another embodiment of the present application;
[0017] Figure 5 It is a schematic diagram of the training process of the intelligent beam prediction model in the embodiment of the present application;
[0018] Figure 6 is a flowchart of a beam prediction method proposed in another embodiment of the present application;
[0019] Figure 7 is a schematic diagram of the structure of a beam prediction device proposed in another embodiment of the present application;
[0020] Figure 8 It is a schematic diagram of the structure of a beam prediction device proposed in another embodiment of the present application. DETAILED DESCRIPTION
[0021] In the embodiments of the present application, the term "and / or" describes the association relationship of the associated objects, indicating that three relationships may exist. For example, A and / or B may represent: A exists alone, A and B exist at the same time, and B exists alone. The character " / " generally indicates that the associated objects before and after are in an "or" relationship.
[0022] In the embodiments of the present application, the term "plurality" refers to two or more than two, and other quantifiers are similar.
[0023] The following will be combined with the drawings in the embodiments of the present application to clearly and completely describe the technical solutions in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of this application.
[0024] The present application is to solve the problem that in the related art, when performing intelligent beam prediction, all group data in the training data set are input into the neural network model to train the neural network model to obtain the intelligent beam prediction model. Since the discrimination degree of some group beam training data in the training data set is very low, there is data ambiguity, which affects the recognition effect of the neural network in identifying the strongest beam, affects the convergence of the neural network model, and the convergence effect of the intelligent beam prediction model is not good, which affects the technical problem of beam prediction accuracy. A beam prediction method is proposed, by obtaining multiple first predicted beam pairs, wherein the first predicted beam pairs have a first reference signal received power RSRP, and different first RSRPs The correlation value between the two beam pairs is less than the correlation threshold; the correlation value and the multiple first RSRPs are input into the target beam prediction model to obtain a second RSRP corresponding to the second predicted beam pair, wherein the target beam prediction model has learned the mapping relationship between the correlation value, the multiple first RSRPs, and the second RSRP, and can perform intelligent beam prediction processing on the beam pair data set of beam pairs with unknown reference signal received power values to obtain the reference signal received power values of all beam pairs in the beam pair data set. Since the target beam prediction model used is trained based on the data after the directivity enhancement processing, the beam prediction accuracy can be effectively improved.
[0025] The technical solution provided in the embodiment of the present application can be applicable to a variety of systems, especially 5G systems. For example, the applicable system can be a global system of mobile communication (GSM) system, a code division multiple access (CDMA) system, a wideband code division multiple access (WCDMA) general packet radio service (GPRS) system, a long term evolution (LTE) system, a LTE frequency division duplex (FDD) system, a LTE time division duplex (TDD) system, an advanced long term evolution (LTE-A) system, a universal mobile telecommunication system (UMTS), a world-wide interoperability for microwave access (WiMAX) system, a 5G new radio (NR) system, etc. These various systems include terminal equipment and network equipment. The system may also include a core network part, such as an evolved packet system (EPS), a 5G system (5GS), etc.
[0026] Figure 1 It is a flowchart of a beam prediction method proposed in an embodiment of the present application.
[0027] It should be noted that the execution subject of the beam prediction method of this embodiment is a beam prediction device, which can be implemented by software and / or hardware, and can be configured in a terminal device. The terminal device refers to a communication device that can be used to forward and receive digital signals, and the terminal device can send signals to a network device through a channel.
[0028] The terminal device involved in the embodiments of the present application may be a device that provides voice and / or data connectivity to a user, a handheld device with a wireless connection function, or other processing devices connected to a wireless modem, etc. In different systems, the names of the terminal devices may also be different.
[0029] For example, in a 5G system, a terminal device may be referred to as a user equipment (UE). A wireless terminal device may communicate with one or more core networks (CN) via a radio access network (RAN). The wireless terminal device may be a mobile terminal device, such as a mobile phone (or "cellular" phone) and a computer with a mobile terminal device, for example, a portable, pocket-sized, handheld, computer-built-in or vehicle-mounted mobile device, which exchanges language and / or data with a radio access network.
[0030] For example, Personal Communication Service (PCS) phones, cordless phones, Session Initiated Protocol (SIP) phones, Wireless Local Loop (WLL) stations, Personal Digital Assistants (PDA) and other devices. Wireless terminal devices may also be referred to as systems, subscriber units, subscriber stations, mobile stations, mobile stations, remote stations, access points, remote terminal devices, access terminal devices, user terminal devices, user agents, and user devices, which are not limited in the embodiments of the present application.
[0031] The network device can be used to convert received air frames to and from Internet Protocol (IP) packets, acting as a router between the wireless terminal device and the rest of the access network, which may include an Internet Protocol (IP) communications network. The network device can also coordinate the management of attributes of the air interface.
[0032] For example, the network device involved in the embodiments of the present application may be a network device (Base Transceiver Station, BTS) in the Global System for Mobile communications (Global System for Mobile communications, GSM) or Code Division Multiple Access (Code Division Multiple Access, CDMA), or a network device (NodeB) in Wide-band Code Division Multiple Access (WCDMA), or an evolutionary network device (evolutional Node B, eNB or e-NodeB) in the long term evolution (long term evolution, LTE) system, a 5G base station (gNB) in the 5G network architecture (next generation system), or a home evolved Node B (Home evolved Node B, HeNB), a relay node, a home base station (femto), a pico base station (pico), etc., which is not limited in the embodiments of the present application. In some network structures, the network device may include a centralized unit (CU) node and a distributed unit (DU) node, and the centralized unit and the distributed unit may also be arranged geographically separately.
[0033] Network devices and terminal devices can each use one or more antennas for multiple-input multiple-output (MIMO) transmission. MIMO transmission can be single-user MIMO (SU-MIMO) or multi-user MIMO (MU-MIMO). Depending on the form and number of antenna combinations, MIMO transmission can be 2D-MIMO, 3D-MIMO, FD-MIMO or massive-MIMO, or it can be diversity transmission, precoded transmission or beamforming transmission, etc.
[0034] like Figure 1 As shown, the beam prediction method includes:
[0035] S101: Acquire multiple first predicted beam pairs, wherein the first predicted beam pairs have a first reference signal received power RSRP, and correlation values between different first RSRPs are less than a correlation threshold.
[0036] Among them, the first predicted beam pair refers to a beam pair formed by the signal transmission between the user and the base station obtained in advance. The first predicted beam pair has a corresponding first reference signal receiving power value (Reference Signal Receiving Power, RSRP), that is, the first RSRP is a known value and can be obtained by actual measurement.
[0037] Among them, the first reference signal received power RSRP refers to the reference signal received power value of the first predicted beam pair. The first RSRP can be obtained by sparse beam scanning measurement of the beam pair formed between the base station sending the beam and the user receiving the beam. The correlation value between different first RSRPs is less than the correlation threshold.
[0038] The correlation value refers to data information used for distinguishing characteristics between different first RSRPs, and the correlation value can be used to describe the degree of data proximity of the strongest part of the first RSRPs among multiple first RSRPs.
[0039] Among them, the correlation degree threshold refers to a threshold value that can be used to divide the discrimination between different first RSRPs. When the correlation degree value is less than the correlation degree threshold, the discrimination between different first RSRPs is relatively high, and the data discrimination requirement of the beam prediction model is met. When the correlation degree value is greater than or equal to the correlation degree threshold, the discrimination between different first RSRPs is relatively low.
[0040] In the embodiment of the present application, when obtaining multiple first predicted beam pairs, the transmission beam capability of the base station and the receiving beam capability of the user in the communication system can be first obtained. Assuming that the transmission beam capability of the base station is M and the receiving beam capability of the user is N, there are MN possibilities for beam matching between the base station and the user, that is, there may be MN beam pairs to obtain multiple predicted beam pairs. For example, Figure 2 As shown, Figure 2 It is a structural diagram of the communication system in an embodiment of the present application. Sparse beam scanning measurement can be performed on the beam pair formed between the base station that sends the beam and the user that receives the beam to obtain the reference signal received power of some predicted beam pairs. The partial predicted beam pair is used as the first predicted beam pair, and the measured partial reference signal received power is used as the first reference signal received power RSRP.
[0041] S102: Input the correlation value and the multiple first RSRPs into a target beam prediction model to obtain a second RSRP corresponding to a second predicted beam pair, wherein the target beam prediction model has learned a mapping relationship between the correlation value, the multiple first RSRPs, and the second RSRP.
[0042] Among them, the second prediction beam pair refers to the prediction beam pair other than the first prediction beam pair in the full number of beam pairs formed between the base station and the user, the second prediction beam pair refers to the prediction beam pair on which the reference signal received power prediction processing is to be performed among multiple prediction beam pairs, and the RSRP of the second prediction beam pair is an unknown quantity.
[0043] The second RSRP refers to the RSRP prediction value corresponding to the second predicted beam pair obtained after performing reference signal received power prediction processing on the second predicted beam pair.
[0044] Among them, the target beam prediction model refers to a prediction model obtained by training the neural network model based on the beam training data set and adjusting the parameters, and is used to predict the reference signal received power of the beam pair. The target beam prediction model has learned the mapping relationship between the correlation degree value, multiple first RSRPs, and the second RSRP. The target beam prediction model has the ability to predict the unknown second RSRP based on the known first RSRP.
[0045] After obtaining the multiple first predicted beam pairs as described above, the embodiment of the present application may input the correlation value and the multiple first RSRPs into the target beam prediction model to obtain the second RSRP corresponding to the second predicted beam pair.
[0046] In an embodiment of the present application, when the correlation value and multiple first RSRPs are input into the target beam prediction model to obtain the second RSRP corresponding to the second predicted beam pair, an initial beam prediction model for training can be first built, and a beam training data set can be obtained to train the initial beam prediction model. The beam training data set can be divided and processed to divide the beam training data set into a normal beam training data set and a beam training data set with data ambiguity, wherein the correlation value between different RSRPs of the normal beam training data set is less than the correlation threshold, and the correlation value between different RSRPs of the beam training data set with data ambiguity is greater than or equal to the correlation threshold, and the partial data set with data ambiguity is divided and processed. According to the enhanced processing, an enhanced beam training data set is obtained, so that the correlation value between different RSRPs of the enhanced beam training data set is less than the correlation threshold, and then the initial beam prediction model is trained based on the ordinary beam training data set and the enhanced beam training data set, respectively, and the RSRP of some beam pairs in the data set are input into the initial beam prediction model, and the initial beam prediction model is used to predict the RSRP of other beam pairs in the data set, and then the predicted RSRP of other beam pairs is compared with their known RSRP to obtain the loss value, and the model parameters are adjusted according to the loss value, and multiple iterative training is performed until the loss value of the model meets the preset loss value, and it is determined that the model meets the convergence conditions, and the target beam prediction model is obtained by convergence.
[0047] In the embodiment of the present application, the target beam prediction model has the ability to predict the reference signal receiving power of the beam pair. The specific training process of the target beam prediction model can be referred to the following embodiment.
[0048] After the target beam prediction model is obtained through the above training, in the embodiment of the present application, multiple first RSRPs can be input into the target beam prediction model, and the correlation values between the multiple first RSRPs satisfy the correlation threshold. The first RSRP is processed based on the target beam prediction model to obtain the predicted RSRP of the second predicted beam pair of the unknown reference signal received power output by the target beam prediction model, and the obtained predicted RSRP is used as the second RSRP, so as to obtain the RSRP of all the multiple predicted beam pairs between the user and the base station.
[0049] In an embodiment of the present application, after processing the first RSRP based on the target beam prediction model to obtain the second RSRP of the second predicted beam pair, the reference signal received power of all beam pairs between all users and the base station is obtained, and then the strongest beam pair can be screened from the multiple beam pairs to achieve the selection of the optimal beam pair, and the beam pair corresponding to the maximum value of the reference signal received power can be selected as the optimal beam pair.
[0050] In this embodiment, by obtaining multiple first predicted beam pairs, wherein the first predicted beam pairs have a first reference signal received power RSRP, and the correlation value between different first RSRPs is less than a correlation threshold, the correlation value and the multiple first RSRPs are input into a target beam prediction model to obtain a second RSRP corresponding to a second predicted beam pair, wherein the target beam prediction model has learned the mapping relationship between the correlation value, the multiple first RSRPs, and the second RSRP, and can perform intelligent beam prediction processing on a beam pair data set for beam pairs with unknown reference signal received power values to obtain reference signal received power values of all beam pairs in the beam pair data set. Since the target beam prediction model used is trained based on data after directivity enhancement processing, the accuracy of beam prediction can be effectively improved.
[0051] Figure 3 It is a flowchart of a beam prediction method proposed in another embodiment of the present application.
[0052] like Figure 3 As shown, the beam prediction method includes:
[0053] S301: Acquire multiple sample beam pairs, wherein each sample beam pair has a third RSRP.
[0054] The sample beam pair refers to the beam pair data used for beam prediction model training. Multiple sample beam pairs constitute a sample beam pair data set. Each sample beam pair has a third RSRP, that is, the reference signal received power value of each sample beam pair is known.
[0055] The third RSRP refers to the reference signal received power value of the sample beam pair.
[0056] In an embodiment of the present application, a beam pair data set for model training can be obtained, and the reference signal received power of the beam pair used for model training is known. The beam pair used for model training is used as a sample beam pair, and its corresponding reference signal received power is used as the RSRP of the corresponding sample beam pair.
[0057] S302: Divide the multiple sample beam pairs into first sample beam pairs and second sample beam pairs according to the multiple third RSRPs, wherein a first correlation value between the third RSRPs of different first sample beam pairs is less than a correlation threshold, and a second correlation value between the third RSRPs of different second sample beam pairs is greater than or equal to the correlation threshold.
[0058] Among them, the first sample beam pair refers to a beam pair data set without data ambiguity obtained by dividing from multiple sample beam pairs, and the absence of data ambiguity means that among the third RSRPs of the sample beam pairs in the first sample beam pairs, the values of the strongest third RSRPs are not close, and have greater distinction.
[0059] The first correlation degree value refers to data information used to describe the distinction between different third RSRPs of the first sample beam pair, and the first correlation degree value is less than the correlation degree threshold.
[0060] The second sample beam pair refers to a beam pair data set with data ambiguity obtained by dividing from multiple sample beam pairs, and the existence of data ambiguity means that among the third RSRPs of the sample beam pairs in the second sample beam pairs, the values of the strongest third RSRPs are relatively close, and the values of the strongest third RSRPs are less distinguishable.
[0061] The second correlation degree value refers to data information used to describe the distinction between different third RSRPs of the second sample beam pair, and the second correlation degree value is greater than or equal to the correlation degree threshold.
[0062] In the embodiment of the present application, after acquiring the multiple sample beam pairs as described above, the multiple sample beam pairs may be divided into first sample beam pairs and second sample beam pairs according to the multiple third RSRPs.
[0063] In the embodiment of the present application, when a plurality of sample beam pairs are divided into a first sample beam pair and a second sample beam pair according to a plurality of third RSRPs, the plurality of sample beam pairs may be grouped to obtain a plurality of groups of sample beam pairs, and a judgment is made as to whether data ambiguity exists in each group of sample beam pairs according to the third RSRP corresponding to each group of sample beam pairs, that is, a judgment is made as to whether the values of the strongest third RSRPs in each group of sample beam pairs are relatively close, and a correlation value between the third RSRPs of each group of sample beam pairs is obtained to obtain the discrimination degree of the strongest third RSRPs. If the correlation value is less than a correlation threshold, it indicates that the discrimination degree between the third RSRPs of the group of sample beam pairs is high, and the group of sample beam pairs is divided into a first sample beam pair. If the correlation value is greater than or equal to the correlation threshold, it indicates that the discrimination degree between the third RSRPs of the group of sample beam pairs is low, and the group of sample beam pairs is divided into a second sample beam pair, so as to realize that the plurality of sample beam pairs are divided into a first sample beam pair and a second sample beam pair according to the plurality of third RSRPs.
[0064] Optionally, in some embodiments, based on multiple third RSRPs, multiple sample beam pairs are divided into first sample beam pairs and second sample beam pairs. The multiple sample beam pairs can be grouped to obtain multiple sample beam pair groups, wherein each sample beam pair group contains at least two sample beam pairs, and a reference RSRP of the sample beam pair group is obtained, wherein the reference RSRP is determined by the third RSRP of at least two sample beam pairs in the sample beam pair group. Based on the multiple reference RSRPs and the first receiving power threshold, the multiple sample beam pairs are divided into first sample beam pairs and second sample beam pairs, so that data ambiguity judgment can be made by correlation of the RSRP values of the part with the largest value among all beam pairs. Based on the data ambiguity judgment result, the training data set is divided into a normal data set and a fuzzy data set, and the directionality enhancement processing is performed on some optimal beams in the fuzzy data set to obtain two sets of data sets after the enhancement processing, and the initial beam prediction model is trained to obtain a better target beam prediction model for beam prediction.
[0065] The sample beam pair group refers to a sample beam pair group obtained by grouping a plurality of sample beam pairs, and each sample beam pair group includes at least two sample beam pairs.
[0066] The first receiving power threshold refers to a pre-set power threshold value for dividing the plurality of sample beam pairs into groups.
[0067] Among them, the reference RSRP refers to the reference data of each sample beam pair group determined to determine the sample data set to which the sample beam pair group belongs. According to the reference RSRP and the first receiving power threshold, multiple sample beam pair groups can be divided into a first sample beam pair and a second sample beam pair.
[0068] In an embodiment of the present application, when a plurality of sample beam pairs are divided into a first sample beam pair and a second sample beam pair according to a plurality of third RSRPs, the plurality of sample beam pairs may be grouped and processed to obtain a sample beam pair group, each sample beam pair group including at least two sample beam pairs, and then a reference RSRP corresponding to each sample beam pair group may be obtained, and a portion of the third RSRPs with the largest value may be selected from the third RSRPs of each sample beam pair group, and an average value of the portion of the third RSRPs may be determined as the reference RSRP, and an RSRP threshold value for processing the reference RSRP to perform judgment processing on a sample beam pair data set to which the sample beam pair group belongs may be set as a first receiving power threshold, and then, according to the plurality of reference RSRPs and the first receiving power threshold, the plurality of sample beam pairs may be divided into a first sample beam pair and a second sample beam pair, and the sample beam pairs in the sample beam pair group whose reference RSRP is less than or equal to the first receiving power threshold may be divided into a first sample beam pair, and the sample beam pairs in the sample beam pair group whose reference RSRP is greater than the first receiving power threshold may be divided into a second sample beam pair.
[0069] S303: Train an initial beam prediction model based on the third RSRP of the first sample beam pair, the first correlation value, the third RSRP of the second sample beam pair, and the second correlation value to obtain a target beam prediction model.
[0070] Among them, the initial beam prediction model refers to a neural network model used to train it to obtain a target beam prediction model. The initial beam prediction model can be, for example, a residual neural network model (Residual Network, ResNet), or it can also be any other neural network model that can be trained to have a beam prediction function, without limitation.
[0071] In the embodiment of the present application, after dividing the multiple sample beam pairs into the first sample beam pair and the second sample beam pair according to the multiple third RSRPs, the initial beam prediction model can be trained based on the third RSRP of the first sample beam pair, the first correlation value, the third RSRP of the second sample beam pair, and the second correlation value to obtain the target beam prediction model.
[0072] In an embodiment of the present application, when the initial beam prediction model is trained based on the third RSRP of the first sample beam pair, the first correlation value, the third RSRP of the second sample beam pair, and the second correlation value to obtain the target beam prediction model, the initial beam prediction model can be first trained based on the third RSRP value of the first sample beam pair, the first correlation value is less than the correlation threshold, and part of the third RSRP of part of the first sample beam pairs is input into the initial beam prediction model, and the RSRP values of all first sample beam pairs are predicted based on the initial beam prediction model using part of the third RSRP, and the predicted RSRP value is compared with the original RSRP value to obtain the loss value of the initial beam prediction model, and the model parameters are adjusted, and multiple iterative training is performed, and the model parameters are adjusted multiple times until the model converges, and then the third RSRP of the second sample beam pair is enhanced, and the same parameter adjustment method is used based on the second sample beam pair after the enhanced processing, and the model after the first sample beam pair training convergence is further trained, and the model parameters are adjusted through iterative training until the model converges, and the converged prediction model is used as the target beam prediction model.
[0073] S304: Acquire multiple first predicted beam pairs, wherein the first predicted beam pairs have a first reference signal received power RSRP, and correlation values between different first RSRPs are less than a correlation threshold.
[0074] S305: Input the correlation value and the multiple first RSRPs into the target beam prediction model to obtain a second RSRP corresponding to the second predicted beam pair, wherein the target beam prediction model has learned the mapping relationship between the correlation value, the multiple first RSRPs, and the second RSRP.
[0075] For the detailed description of S304 and S305 , please refer to the above embodiments, which will not be repeated here.
[0076] In this embodiment, by obtaining multiple first predicted beam pairs, wherein the first predicted beam pairs have a first reference signal received power RSRP, and the correlation value between different first RSRPs is less than a correlation threshold, the correlation value and the multiple first RSRPs are input into a target beam prediction model to obtain a second RSRP corresponding to a second predicted beam pair, wherein the target beam prediction model has learned the mapping relationship between the correlation value, the multiple first RSRPs, and the second RSRP, and can perform intelligent beam prediction processing on a beam pair data set for beam pairs with unknown reference signal received power values to obtain reference signal received power values for all beam pairs in the beam pair data set. Since the target beam prediction model used is trained based on data after directivity enhancement processing, the beam prediction accuracy can be effectively improved. By The sample beam pairs are grouped to obtain multiple sample beam pair groups, wherein each sample beam pair group includes at least two sample beam pairs, and a reference RSRP of the sample beam pair group is obtained, wherein the reference RSRP is determined by the third RSRP of at least two sample beam pairs in the sample beam pair group; the multiple sample beam pairs are divided into a first sample beam pair and a second sample beam pair according to the multiple reference RSRPs and a first receiving power threshold, so that data ambiguity judgment can be made by correlation of some RSRP values with the largest value in all beam pairs; based on the data ambiguity judgment result, the training data set is divided into a normal data set and a fuzzy data set; directivity enhancement processing is performed on some optimal beams in the fuzzy data set, so as to obtain two groups of data sets after the enhancement processing, and an initial beam prediction model is trained to obtain a better target beam prediction model for beam prediction.
[0077] Figure 4 It is a flowchart of a beam prediction method proposed in another embodiment of the present application.
[0078] like Figure 4 As shown, the beam prediction method includes:
[0079] S401: Acquire multiple sample beam pairs, wherein each sample beam pair has a third RSRP.
[0080] S402: Grouping a plurality of sample beam pairs to obtain a plurality of sample beam pair groups, wherein each sample beam pair group includes at least two sample beam pairs.
[0081] For the detailed description of S401 and S402 , please refer to the above embodiments, which will not be repeated here.
[0082] S403: Determine an average value of third RSRPs of at least two sample beam pairs in the sample beam pair group, wherein the average value is used as a reference RSRP.
[0083] In an embodiment of the present application, when obtaining the reference RSRP of the sample beam pair group, the third RSRP of the sample beam pair group can be normalized, and then the normalized third RSRP can be sorted to determine at least two third RSRPs with the largest values from the third RSRPs of the sorted sample beam pair group, and the average value of the third RSRPs of at least two sample beam pairs is taken, and the average value is used as the reference RSRP of the sample beam pair group.
[0084] For example, when a plurality of sample beam pairs are divided into a first sample beam pair and a second sample beam pair, a plurality of sample beam pairs for model training are first obtained. The plurality of sample beam pairs can be obtained using the entire training data set matrix L Train ×MN represents, where M represents the base station's transmit beam capability, N represents the user's receive beam capability, and for multiple sample beam pairs L Train ×MN are grouped, and then feature recognition and division processing is performed on the grouped multiple sample beam pairs to obtain L Train_R The sample beam pair in the group is taken as the first sample beam pair, and L is obtained. Train_F The sample beam pair in the sample beam pair group is used as the second sample beam pair. When the multiple sample beam pair groups after grouping are subjected to feature recognition and division processing, at least two third RSRPs can be determined from the third RSRPs of the sample beam pair group. The maximum value normalization can be performed on each sample beam pair group. The specific processing flow is to first select the strongest reference signal received power RSRP from the third RSRP of the sample beam pair group. max Then divide all the third RSRP by RSRP max , to select the K third RSRPs with the largest third RSRP values from the group of sample beam pairs, usually the reference value of K is 2, 3, 4, and the average RSRP of the strongest K third RSRPs is calculated mean , and the average power RSRP mean As a reference RSRP.
[0085] S404: Divide a plurality of sample beam pairs into first sample beam pairs and second sample beam pairs according to a plurality of reference RSRPs and a first receiving power threshold.
[0086] Optionally, in some embodiments, multiple sample beam pairs are divided into first sample beam pairs and second sample beam pairs based on multiple reference RSRPs and a first receiving power threshold. If the reference RSRP is less than or equal to the first receiving power threshold, the sample beam pair in the sample beam pair group corresponding to the reference RSRP is used as the first sample beam pair. If the reference RSRP is greater than the first receiving power threshold, the sample beam pair in the sample beam pair group corresponding to the reference RSRP is used as the second sample beam pair.
[0087] In an embodiment of the present application, when multiple sample beam pairs are divided into first sample beam pairs and second sample beam pairs according to multiple reference RSRPs and a first receiving power threshold, multiple reference RSRPs can be numerically compared with the first receiving power threshold. If the reference RSRP is less than or equal to the first receiving power threshold, the sample beam pair in the sample beam pair group corresponding to the reference RSRP is used as the first sample beam pair. If the reference RSRP is greater than the first receiving power threshold, the sample beam pair in the sample beam pair group corresponding to the reference RSRP is used as the second sample beam pair.
[0088] For example, for n groups of sample beam pairs obtained after grouping a plurality of prediction beams, a cyclic calculation grouping process is performed, and the number of initialization cycles n=1 is set, and the number of sample beam pairs whose sample beam pairs belong to the first sample beam pair is L Train_R = 0, the number of sample beam pair groups whose sample beam pairs belong to the second sample beam pair L Train_F = 0, respectively determine the reference RSRP of n sample beam pairs mean Afterwards, the reference RSRP of each sample beam pair group can be mean and the first receiving power threshold RSRP Th For comparison, if we refer to RSRP mean Greater than the first receiving power threshold RSRP Th , then the sample beam pairs in the nth group of sample beam pairs are classified as the second sample beam pairs L Train_F , let L Train_F =L Train_F +1, if referring to RSRP mean Less than or equal to the first receiving power threshold RSRP Th , classify the sample beam pairs in the nth sample beam pair group as the first sample beam pair, and then let L Train_R =L Train_R +1, where the first receiving power threshold RSRP can usually be set Th The reference value of is 0.95. After processing each sample beam pair, the number of cycles is n = n + 1. When n is less than L Train When , the sample beam pair groups that have not been grouped are continued to be processed; otherwise, the loop is terminated to complete the grouping processing of multiple sample beam pair groups.
[0089] S405: Train an initial beam prediction model based on the third RSRP and the first correlation value of the first sample beam pair to obtain a first beam prediction model.
[0090] The first beam prediction model refers to a beam prediction model obtained by training an initial beam prediction model until convergence based on the third RSRP and the first correlation value of the first sample beam pair.
[0091] In an embodiment of the present application, when the initial beam prediction model is trained based on the third RSRP and the first correlation value of the first sample beam pair to obtain the first beam prediction model, part of the third RSRP of the third RSRP of the first sample beam pair can be input into the initial beam prediction model, and the first correlation value corresponding to the third RSRP of the first sample beam pair is less than the correlation threshold. The initial beam prediction model is used to predict the reference signal received power of all first sample beam pairs as the predicted RSRP, and the predicted RSRP of each first sample beam pair is compared with the corresponding third RSRP for loss value. The model parameters are adjusted according to the loss value, and multiple iterative prediction trainings are performed until the model converges. The converged prediction model at this time is used as the first beam prediction model.
[0092] Optionally, in some embodiments, the number of first sample beam pairs is multiple, and the multiple first sample beam pairs are composed of a first reference sample beam pair and a first predicted sample beam pair, wherein the initial beam prediction model is trained based on the third RSRP and the first correlation degree value of the first sample beam pair to obtain the first beam prediction model, and the third RSRP and the first correlation degree value of the first reference sample beam pair can be input into the initial beam prediction model to obtain the predicted RSRP of the first predicted sample beam pair output by the initial beam prediction model, and if the predicted RSRP of the first predicted sample beam pair and the third RSRP of the first predicted sample beam pair meet the convergence condition, the trained initial beam prediction model is used as the first beam prediction model.
[0093] The first reference sample beam pair refers to a portion of the reference sample beam pairs corresponding to a portion of the third RSRP values in the first sample beam pair that are used as inputs of the initial sample model.
[0094] The first predicted sample beam pair refers to a portion of the reference sample beam pairs in the first sample beam pair whose corresponding third RSRP values are to be predicted, and the first reference sample beam pair and the first predicted sample beam pair together constitute the first sample beam pair.
[0095] In an embodiment of the present application, when the initial beam prediction model is trained based on the third RSRP of the first sample beam pair to obtain the first beam prediction model, the third RSRP of the first reference sample beam pair and the first correlation value can be input into the initial beam prediction model to obtain the predicted RSRP of the first predicted sample beam pair output by the initial beam prediction model, and the predicted RSRP of the first predicted sample beam pair is compared with the third RSRP corresponding to the first predicted sample beam pair to obtain a predicted loss value, and the parameters of the initial beam prediction model are adjusted according to the predicted loss value, and then the input training process is repeated multiple times until the loss value between the predicted RSRP of the first predicted sample beam pair and the third RSRP of the first predicted sample beam pair meets the preset loss value, indicating that the convergence condition between the predicted RSRP of the first predicted sample beam pair and the third RSRP of the first predicted sample beam pair is met. At this time, it is determined that the trained initial beam prediction model converges, and the trained initial beam prediction model is used as the first beam prediction model.
[0096] S406: Perform enhancement processing on the third RSRP of the second sample beam pair to obtain enhanced second sample beam pairs, wherein a third correlation value between the third RSRPs of different enhanced second sample beam pairs is less than a correlation threshold.
[0097] The third correlation degree value refers to data information used to describe the degree of differentiation between the third RSRPs of the enhanced second sample beam pairs.
[0098] In an embodiment of the present application, when the third RSRP of the second sample beam pair is enhanced to obtain the enhanced second sample beam pair, the largest part of the third RSRP in the second sample beam pair can be enhanced and amplified to obtain the enhanced second sample beam pair.
[0099] Optionally, in some embodiments, the third RSRP of the second sample beam pair is enhanced to obtain the enhanced second sample beam pair, and the enhancement processing parameters corresponding to the second sample beam pair can be obtained. The third RSRP of the second sample beam pair is processed according to the enhancement processing parameters to obtain the enhanced second sample beam pair, wherein the enhancement processing parameters include: one or more amplification processing parameters, wherein the amplification processing parameters are used to amplify the RSRP, and the value of the amplification processing parameters is greater than 1. When the enhancement processing parameters include multiple amplification processing parameters, the values of the multiple amplification processing parameters are different, so that the training can be completed. The enhanced processing of the fuzzy data in the beam data set highlights the strongest beam, so that when the network training is performed based on the enhanced beam data set, the interference of fuzzy data features is avoided, and the model training effect is effectively improved, so that the enhanced processing parameters can be pre-set to amplify some of the second sample beam pairs, so that the data distinction between the third RSRPs of the enhanced second sample beam pairs is higher, and the target prediction model obtained by training the first beam prediction model based on the third RSRP of the enhanced second sample beam pairs has better convergence and has a more superior beam prediction ability, thereby effectively improving the accuracy of beam prediction.
[0100] Among them, the enhanced processing parameters include: one or more amplification processing parameters, the amplification processing parameters are used to amplify the RSRP, the value of the amplification processing parameters is greater than 1, so that when the amplification processing parameters are multiplied by the third RSRP, the value of the third RSRP is amplified, and when the enhanced processing parameters include multiple amplification processing parameters, the values of the multiple amplification processing parameters are different.
[0101] In an embodiment of the present application, when the third RSRP of the second sample beam pair is enhanced to obtain the enhanced second sample beam pair, one or more amplification processing parameters corresponding to the second sample beam pair can be obtained as enhancement processing parameters, and the third RSRP with the largest value of the second sample beam pair is amplified based on the enhancement processing parameters. The value of the enhancement processing parameter can be adaptively and freely set according to the actual application, and the third RSRP with the largest value is multiplied by the corresponding enhancement processing parameter to obtain the enhanced second sample beam pair.
[0102] S407: Train the first beam prediction model based on the enhanced third RSRP and the third correlation value of the second sample beam pair to obtain a target beam prediction model.
[0103] In the embodiment of the present application, after the third RSRP of the second sample beam pair is enhanced to obtain the enhanced second sample beam pair, the first beam prediction model can be trained based on the third RSRP and the third correlation value of the enhanced second sample beam pair to obtain the target beam prediction model.
[0104] In an embodiment of the present application, when the first beam prediction model is trained based on the third RSRP of the enhanced second sample beam pair to obtain the target beam prediction model, part of the third RSRP of the enhanced second sample beam pair can be input into the initial beam prediction model, and the first beam prediction model is used to predict the reference signal received power of all enhanced second sample beam pairs as the predicted RSRP, and the predicted RSRP of each enhanced second sample beam pair is compared with the corresponding third RSRP to obtain the loss value of the model, and then the model parameters are adjusted according to the loss value of the model, and the training is iterated multiple times until the model converges, and the first beam prediction model that converges at this time is used as the target beam prediction model.
[0105] Optionally, in some embodiments, the enhanced second sample beam pair is composed of a second reference sample beam pair and a second predicted sample beam pair, wherein the first beam prediction model is trained based on the third RSRP and the third correlation value of the enhanced second sample beam pair to obtain a target beam prediction model, and the third RSRP and the third correlation value of the second reference sample beam pair can be input into the first beam prediction model to obtain the predicted RSRP of the second predicted sample beam pair output by the first beam prediction model; if the predicted RSRP of the second predicted sample beam pair and the third RSRP of the second predicted sample beam pair meet the convergence condition, the trained first beam prediction model is used as the target beam prediction model.
[0106] The second reference sample beam pair refers to a portion of the reference sample beam pairs corresponding to a portion of the third RSRP values in the second sample beam pair that are used as inputs of the initial sample model.
[0107] The second predicted sample beam pair refers to a portion of the reference sample beam pair in the second sample beam pair whose corresponding third RSRP value is to be predicted, and the second reference sample beam pair and the second predicted sample beam pair together constitute the first sample beam pair.
[0108] In an embodiment of the present application, when the first beam prediction model is trained based on the third RSRP and the third correlation value of the enhanced second sample beam pair to obtain the target beam prediction model, the third RSRP of the second reference sample beam pair can be input into the first beam prediction model to obtain the predicted RSRP of the second prediction sample beam pair output by the first beam prediction model, and the predicted RSRP of the second prediction sample beam pair is compared with the third RSRP of the second prediction sample beam pair to obtain the predicted loss value of the model, and the parameters of the first beam prediction model are adjusted according to the predicted loss value, and then the input training process is repeated multiple times until the loss value between the predicted RSRP of the second prediction sample beam pair and the third RSRP corresponding to the second prediction sample beam meets the preset loss value, indicating that the predicted RSRP of the second prediction sample beam pair and the third RSRP of its second prediction sample beam pair meet the convergence condition, and at this time, it is determined that the trained first beam prediction model converges, and the trained first beam prediction model is used as the target beam prediction model.
[0109] For example, Figure 5 As shown, Figure 5 It is a training flow diagram of the intelligent beam prediction model in the embodiment of the present application. When training the target beam prediction model, the initial beam prediction model can be obtained first. Taking the training of the residual neural network model ResNet as the initialization beam prediction model as an example, a large amount of training data must first be collected or captured from the platform. The training data is usually the reference signal RSRP value of the fractional beam measured by the user end. For example, the total amount of data is L, and L should usually be a large value. Before the network training starts, all L groups of data are first divided into training data sets and test data sets, which are used for parameter training of the network model and for performance testing of the network. The full amount of training data sets can be used with the training data set matrix L Train ×MN indicates.
[0110] In the embodiment of the present application, based on the training data set L Train When training the target beam prediction model, firstly, the matrix L of the entire training data set is Train ×MN are grouped to obtain multiple sample beam pairs, and feature recognition and classification processing is performed on the multiple sample beam pairs to obtain a common data set L Train_R ×MN and fuzzy dataset L Train_F ×MN, the sample beam pair in the common data set is taken as the first sample beam pair, the sample beam pair in the fuzzy data set is taken as the second sample beam pair, the second sample beam pair in the fuzzy data set is enhanced to highlight the strongest beam, and the enhanced data set is obtained. The sample beam pair in the enhanced data set is taken as the enhanced second sample beam pair, and then the common data set L can be used. Train_R×MN and the enhanced fuzzy dataset L Train_F ×MN groups are used to train the initial beam prediction model. The residual neural network model ResNet is used as the initial beam prediction model for parameter training in two steps. First, L Train_R The first sample beam pair is formed, the ResNet model is trained, and the model parameters are iteratively updated until the model converges. Then, L Train_F The second sample beam pair after group enhancement is further trained and updated on the basis of the ResNet model parameters obtained by training the first sample beam pair until convergence, until the model converges, and the converged model is used as the target beam prediction model, wherein the input of the ResNet model is part of the MN RSRP data, for example, there are I (I<MN) input data, and the label output of the model is all MN RSRP data, which essentially reconstructs the complete RSRP data distribution based on a small amount of data.
[0111] In the embodiment of the present application, an artificial intelligence-based beam prediction method is adopted, which can predict the optimal beam pair based on a small number of beam pair measurements, thereby greatly reducing the measurement resource overhead of the base station.
[0112] S408: Acquire multiple predicted beam pairs, wherein a first predicted beam pair among the multiple predicted beam pairs has a first reference signal received power RSRP.
[0113] S409: Process the first RSRP based on the target beam prediction model to obtain a second RSRP of a second predicted beam pair, wherein the second predicted beam pair is a predicted beam pair among the multiple predicted beam pairs except the first predicted beam pair.
[0114] For the detailed description of S408 and S409 , please refer to the above embodiments, which will not be repeated here.
[0115] In this embodiment, a plurality of first predicted beam pairs are obtained, wherein the first predicted beam pairs have a first reference signal received power RSRP, and the correlation value between different first RSRPs is less than a correlation threshold, and the correlation value and the plurality of first RSRPs are input into a target beam prediction model to obtain a second RSRP corresponding to a second predicted beam pair, wherein the target beam prediction model has learned a mapping relationship between the correlation value, the plurality of first RSRPs, and the second RSRP, and can perform intelligent beam prediction processing on a beam pair data set for beam pairs with unknown reference signal received power values to obtain reference signal received power values for all beam pairs in the beam pair data set. Since the target beam prediction model used is trained based on data after directivity enhancement processing, The beam prediction accuracy can be effectively improved by determining the enhanced processing parameters corresponding to the second sample beam pair, wherein the enhanced processing parameters include: one or more amplification processing parameters, wherein the amplification processing parameters are used to amplify RSRP, and the third RSRP of the second sample beam pair is processed according to the enhanced processing parameters to obtain an enhanced second sample beam pair, so that the enhanced processing parameters can be pre-set to amplify some of the second sample beam pairs, so that the data distinction between the third RSRPs of the enhanced second sample beam pairs is higher, and the target prediction model obtained by training the first beam prediction model based on the third RSRP of the enhanced second sample beam pair has better convergence and has a more superior beam prediction capability, thereby effectively improving the accuracy of beam prediction.
[0116] Figure 6 It is a flowchart of a beam prediction method proposed in another embodiment of the present application.
[0117] like Figure 6 As shown, the beam prediction method includes:
[0118] S601: Acquire multiple sample beam pairs, wherein each sample beam pair has a third RSRP.
[0119] S602: Divide a plurality of sample beam pairs into first sample beam pairs and second sample beam pairs according to a plurality of third RSRPs.
[0120] S603: Train an initial beam prediction model based on the third RSRP and the first correlation value of the first sample beam pair to obtain a first beam prediction model.
[0121] For the detailed description of S601 to S603 , please refer to the above embodiment, which will not be repeated here.
[0122] S604: Obtain enhancement processing parameters corresponding to the second sample beam pair.
[0123] In an embodiment of the present application, when obtaining the enhanced processing parameters corresponding to the second sample beam pair, it is possible to first determine the number of third RSRPs with the largest value selected when determining the reference RSRP of the second sample beam pair, determine an enhanced processing parameter that is one less than the number of the third RSRP with the largest value, and set the data amplification parameters as the enhanced processing parameters from large to small according to the value of the third RSRP with the largest value.
[0124] For example, let the number of the third RSRP with the largest value be K. When K=2, L Train_F The strongest beam RSRP value of each group in the data set is amplified by P1 times. When K=3, L Train_F The strongest beam RSRP value of each group in the data set is amplified by P1 times, and the second strongest beam RSRP value is amplified by P2 times. When K = 4, L Train_F The strongest beam RSRP value of each group in the data set is amplified by P1 times, the second strongest beam RSRP value is amplified by P2 times, and the third strongest beam RSRP value is amplified by P3 times. The P1, P2, and P3 values can be freely configured. Reference values are such as P1=1.5, P2=1.3, and P3=1.1. After the above data processing, the second sample beam pair of the original fuzzy data set can be processed into a style similar to the first sample beam pair of the ordinary data set. The updated data set is the enhanced second sample beam pair, and its data volume does not change.
[0125] S605: Determine a second sample beam pair to be processed from a plurality of second sample beam pairs.
[0126] The second sample beam pair to be processed refers to a second sample beam pair to be amplified and enhanced among multiple second sample beam pairs.
[0127] In an embodiment of the present application, when determining a second sample beam pair to be processed from multiple second sample beam pairs, the third RSRPs of the second sample beam pairs can be sorted from large to small, and the number of values can be set in advance. From the sorted second sample beam pairs, the second sample beam pairs corresponding to the third RSRPs with the largest number of values of 2, 3 or 4 are taken out as the second sample beam pairs to be processed.
[0128] Optionally, in some embodiments, a second sample beam pair to be processed is determined from a plurality of second sample beam pairs, and a third RSRP greater than a second receiving power threshold can be selected from a plurality of third RSRPs corresponding to the plurality of second sample beam pairs, and the second sample beam pair corresponding to the third RSRP greater than the second receiving power threshold is used as the second sample beam pair to be processed, so that the reference signal receiving power to be amplified can be selected from the plurality of second sample beam pairs for enhanced processing, thereby making the distinction between the plurality of second sample beam pairs after the enhanced processing higher.
[0129] The second receiving power threshold refers to a preset threshold value for determining a second sample beam pair to be processed from a plurality of second sample beam pairs. The second receiving power threshold can be adaptively set according to the number of second sample beam pairs to be processed to be selected.
[0130] In an embodiment of the present application, when determining a second sample beam pair to be processed from a plurality of second sample beam pairs, a plurality of third RSRPs corresponding to the plurality of second sample beam pairs can be sorted, and then a second receiving power threshold is set according to the number of second sample beam pairs to be processed that need to be selected, and then a third RSRP greater than the second receiving power threshold is selected from the plurality of third RSRPs corresponding to the plurality of second sample beam pairs, and the second sample beam pair corresponding to the third RSRP greater than the second receiving power threshold is used as the second sample beam pair to be processed.
[0131] S606: Process the third RSRP of the second sample beam pair to be processed according to the enhanced processing parameter to obtain an enhanced second sample beam pair.
[0132] In the embodiment of the present application, after determining the enhanced processing parameters corresponding to the second sample beam pair and determining the second sample beam pair to be processed from multiple second sample beam pairs, the third RSRP of the second sample beam pair to be processed can be processed according to the enhanced processing parameters to obtain the enhanced second sample beam pair.
[0133] In an embodiment of the present application, when the third RSRP of the second sample beam pair to be processed is processed according to the enhanced processing parameters to obtain the enhanced second sample beam pair, the second sample beam pair to be processed can be multiplied by the corresponding enhanced processing parameters to obtain the third RSRP after the enhanced processing, thereby obtaining the enhanced second sample beam pair.
[0134] For example, assume that the number of the third RSRP with the largest value in the second sample beam pair is K. When K=2, an enhanced processing parameter P1 is set, and the strongest beam RSRP value among the third RSRPs of the second sample beam pair is multiplied by the enhanced processing parameter P1 to amplify the strongest beam RSRP by P1 times. When K=3, the second sample beam pair corresponding to the strongest beam RSRP value and the second strongest beam RSRP value is used as the second sample beam pair to be processed, and the enhanced processing parameters P1 and P2 are set. The strongest beam RSRP value is multiplied by the enhanced processing parameter P1 to amplify the strongest beam RSRP value by P1 times, and the second strongest beam RSRP value is multiplied by the enhanced processing parameter P2 to amplify the second strongest beam RSRP value by P2 times. When K=4, the strongest beam RSRP and the second strongest beam RSRP values are multiplied by P1 and P2. The second sample beam pair corresponding to the third strongest beam RSRP value is used as the sample beam pair to be processed, the strongest beam RSRP value is multiplied by the enhanced processing parameter P1 to amplify the strongest beam RSRP value by P1 times, the second strongest beam RSRP value is multiplied by the enhanced processing parameter P2 to amplify the second strongest beam RSRP value by P2 times, and the third strongest beam RSRP value is multiplied by the enhanced processing parameter P3 to amplify the third strongest beam RSRP value by P3 times, wherein P1, P2, and P3 values can be freely configured, and reference values such as P1=1.5, P2=1.3, and P3=1.1. After the above data processing, the original fuzzy data set can be processed into a style similar to that of an ordinary data set, and the sample beam pair in the enhanced data set can be used as the enhanced second sample beam pair, and the data amount does not change.
[0135] S607: Train the first beam prediction model based on the enhanced third RSRP and the third correlation value of the second sample beam pair to obtain a target beam prediction model.
[0136] In an embodiment of the present application, an intelligent beam prediction method is proposed for grouping and strengthening training data based on beam RSRP value characteristics. Data ambiguity judgment is made by judging the correlation of the first K RSRP values in all beam pairs. Based on the data ambiguity judgment, the training data set is divided into a normal data set and a fuzzy data set, and the fuzzy data set is strengthened with the optimal beam. Then, the neural network is trained based on the two groups of data sets to obtain a better network model for beam prediction, so that the differentiation of the training data set can be identified, and data separation is performed on data with data ambiguity and normal data. The model is first trained with normal data to initially obtain model parameters, and the strongest beam is strengthened for the fuzzy data to improve the directional directivity of the data. Finally, the preliminarily converged beam prediction model is updated and trained based on the strengthened data set, effectively improving the accuracy of beam prediction by the beam prediction model after the final training convergence.
[0137] S608: Acquire multiple first predicted beam pairs, wherein the first predicted beam pairs have a first reference signal received power RSRP, and correlation values between different first RSRPs are less than a correlation threshold.
[0138] S609: Input the correlation value and the multiple first RSRPs into the target beam prediction model to obtain a second RSRP corresponding to the second predicted beam pair, wherein the target beam prediction model has learned the mapping relationship between the correlation value, the multiple first RSRPs, and the second RSRP.
[0139] For the detailed description of S607 to S609 , please refer to the above embodiment, which will not be repeated here.
[0140] In this embodiment, by obtaining multiple first predicted beam pairs, wherein the first predicted beam pairs have a first reference signal received power RSRP, and the correlation value between different first RSRPs is less than a correlation threshold, the correlation value and the multiple first RSRPs are input into a target beam prediction model to obtain a second RSRP corresponding to a second predicted beam pair, wherein the target beam prediction model has learned a mapping relationship between the correlation value, the multiple first RSRPs, and the second RSRP, and can perform intelligent beam prediction processing on a beam pair data set for beam pairs with unknown reference signal received power values to obtain the correlation value of all beam pairs in the beam pair data set. The reference signal received power value can be effectively improved by using a target beam prediction model that is trained based on data after directivity enhancement processing. A third RSRP greater than the second received power threshold is selected from multiple third RSRPs corresponding to multiple second sample beam pairs, and the second sample beam pair corresponding to the third RSRP greater than the second received power threshold is used as the second sample beam pair to be processed. Therefore, the reference signal received power that needs to be amplified can be selected from the multiple second sample beam pairs for enhanced processing, thereby making the distinction between the multiple second sample beam pairs after the enhanced processing higher.
[0141] Figure 7 It is a structural diagram of a beam prediction device proposed in one embodiment of the present application.
[0142] like Figure 7 As shown, the beam prediction device 70 includes:
[0143] An acquisition unit 701 is configured to acquire a plurality of first predicted beam pairs, wherein the first predicted beam pairs have a first reference signal received power RSRP, and correlation values between different first RSRPs are less than a correlation threshold;
[0144] The processing unit 702 is used to input the correlation value and the multiple first RSRPs into the target beam prediction model to obtain a second RSRP corresponding to the second predicted beam pair, wherein the target beam prediction model has learned the mapping relationship between the correlation value, the multiple first RSRPs, and the second RSRP.
[0145] In some embodiments of the present application, the target beam prediction model in the processing unit 702 is trained based on the following method:
[0146] Acquire a plurality of sample beam pairs, wherein each sample beam pair has a third RSRP;
[0147] According to the plurality of third RSRPs, the plurality of sample beam pairs are divided into first sample beam pairs and second sample beam pairs, wherein a first correlation value between the third RSRPs of different first sample beam pairs is less than a correlation threshold, and a second correlation value between the third RSRPs of different second sample beam pairs is greater than or equal to the correlation threshold; and
[0148] An initial beam prediction model is trained based on the third RSRP of the first sample beam pair, the first correlation value, the third RSRP of the second sample beam pair, and the second correlation value to obtain a target beam prediction model.
[0149] In some embodiments of the present application, the multiple sample beam pairs are divided into first sample beam pairs and second sample beam pairs according to multiple third RSRPs, including:
[0150] Grouping the plurality of sample beam pairs to obtain a plurality of sample beam pair groups, wherein each sample beam pair group includes at least two sample beam pairs;
[0151] Acquire a reference RSRP of the sample beam pair group, wherein the reference RSRP is determined by a third RSRP of at least two sample beam pairs in the sample beam pair group;
[0152] According to a plurality of reference RSRPs and a first received power threshold, a plurality of sample beam pairs are divided into first sample beam pairs and second sample beam pairs.
[0153] In some embodiments of the present application, the method of dividing a plurality of sample beam pairs into a first sample beam pair and a second sample beam pair according to a plurality of reference RSRPs and a first received power threshold comprises:
[0154] If the reference RSRP is less than or equal to the first received power threshold, taking a sample beam pair in the sample beam pair group corresponding to the reference RSRP as a first sample beam pair;
[0155] If the reference RSRP is greater than the first received power threshold, a sample beam pair in the sample beam pair group corresponding to the reference RSRP is used as a second sample beam pair.
[0156] In some embodiments of the present application, obtaining a reference RSRP of a sample beam pair group includes:
[0157] An average value of third RSRPs of at least two sample beam pairs in the sample beam pair group is determined, wherein the average value is taken as a reference RSRP.
[0158] In some embodiments of the present application, the initial beam prediction model is trained based on the third RSRP of the first sample beam pair, the first correlation value, the third RSRP of the second sample beam pair, and the second correlation value to obtain the target beam prediction model, including:
[0159] Training an initial beam prediction model based on a third RSRP and a first correlation value of the first sample beam pair to obtain a first beam prediction model;
[0160] Performing enhancement processing on the third RSRP of the second sample beam pair to obtain an enhanced second sample beam pair, wherein a third correlation value between the third RSRPs of different enhanced second sample beam pairs is less than a correlation threshold;
[0161] The first beam prediction model is trained based on the third RSRP and the third correlation value of the enhanced second sample beam pair to obtain a target beam prediction model.
[0162] In some embodiments of the present application, the number of the first sample beam pairs is multiple, and the multiple first sample beam pairs are composed of the first reference sample beam pair and the first predicted sample beam pair;
[0163] The initial beam prediction model is trained based on the third RSRP and the first correlation value of the first sample beam pair to obtain the first beam prediction model, including:
[0164] Inputting the third RSRP and the first correlation value of the first reference sample beam pair into the initial beam prediction model to obtain the predicted RSRP of the first predicted sample beam pair output by the initial beam prediction model;
[0165] If the predicted RSRP of the first prediction sample beam pair and the third RSRP of the first prediction sample beam pair meet the convergence condition, the trained initial beam prediction model is used as the first beam prediction model.
[0166] In some embodiments of the present application, the number of the enhanced second sample beam pairs is multiple, and the enhanced second sample beam pairs are composed of the second reference sample beam pair and the second predicted sample beam pair;
[0167] The first beam prediction model is trained based on the third RSRP and the third correlation value of the enhanced second sample beam pair to obtain the target beam prediction model, including:
[0168] Inputting the third RSRP and the third correlation value of the second reference sample beam pair into the first beam prediction model to obtain the predicted RSRP of the second predicted sample beam pair output by the first beam prediction model;
[0169] If the predicted RSRP of the second prediction sample beam pair and the third RSRP of the second prediction sample beam pair meet the convergence condition, the trained first beam prediction model is used as the target beam prediction model.
[0170] In some embodiments of the present application, the third RSRP of the second sample beam pair is enhanced to obtain the enhanced second sample beam pair, including:
[0171] Acquire enhancement processing parameters corresponding to the second sample beam pair;
[0172] Processing the third RSRP of the second sample beam pair according to the enhanced processing parameter to obtain an enhanced second sample beam pair;
[0173] The enhanced processing parameters include: one or more amplification processing parameters, wherein the amplification processing parameters are used to amplify RSRP, and the value of the amplification processing parameters is greater than 1. When the enhanced processing parameters include multiple amplification processing parameters, the values of the multiple amplification processing parameters are different.
[0174] In some embodiments of the present application, the number of the second sample beam pairs is multiple; the training method of the target beam prediction model also includes:
[0175] Determine a second sample beam pair to be processed from a plurality of second sample beam pairs;
[0176] The step of processing the third RSRP of the second sample beam pair according to the enhanced processing parameter to obtain an enhanced second sample beam pair includes:
[0177] The third RSRP of the second sample beam pair to be processed is processed according to the enhanced processing parameter to obtain an enhanced second sample beam pair.
[0178] In some embodiments of the present application, determining a second sample beam pair to be processed from a plurality of second sample beam pairs includes:
[0179] Selecting a third RSRP greater than a second received power threshold from a plurality of third RSRPs respectively corresponding to a plurality of second sample beam pairs;
[0180] A second sample beam pair corresponding to a third RSRP greater than the second receiving power threshold is used as a second sample beam pair to be processed.
[0181] It should be noted that the division of units in the embodiments of the present application is schematic and is only a logical function division. There may be other division methods in actual implementation. In addition, each functional unit in each embodiment of the present application may be integrated into a processing unit, or each unit may exist physically separately, or two or more units may be integrated into one unit. The above-mentioned integrated unit may be implemented in the form of hardware or in the form of software functional units.
[0182] If the integrated unit is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a processor-readable storage medium.
[0183] Based on this understanding, the technical solution of the present application, 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, which is stored in a storage medium and includes several instructions for enabling a computer device (which can be a personal computer, server, or network device, etc.) or a processor to execute all or part of the steps of the various embodiments of the present application. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (ROM), random access memory (RAM), disk or optical disk, etc., various media that can store program codes.
[0184] It should be noted here that the above-mentioned device provided in the embodiment of the present application can implement all the method steps implemented in the above-mentioned method embodiment, and can achieve the same technical effect. The parts and beneficial effects of this embodiment that are the same as those in the method embodiment will not be described in detail here.
[0185] In this embodiment, by obtaining multiple first predicted beam pairs, wherein the first predicted beam pairs have a first reference signal received power RSRP, and the correlation value between different first RSRPs is less than a correlation threshold, the correlation value and the multiple first RSRPs are input into a target beam prediction model to obtain a second RSRP corresponding to a second predicted beam pair, wherein the target beam prediction model has learned the mapping relationship between the correlation value, the multiple first RSRPs, and the second RSRP, and can perform intelligent beam prediction processing on a beam pair data set for beam pairs with unknown reference signal received power values to obtain reference signal received power values of all beam pairs in the beam pair data set. Since the target beam prediction model used is trained based on data after directivity enhancement processing, the accuracy of beam prediction can be effectively improved.
[0186] Figure 8 It is a structural diagram of a beam prediction device proposed in another embodiment of the present application.
[0187] See also Figure 8 The beam prediction device 80 includes a memory 801, a transceiver 802 and a processor 803: the memory 801 is used to store a computer program; the transceiver 802 is used to send and receive data under the control of the processor 803; the processor 803 is used to read the computer program in the memory 801 and perform the following operations:
[0188] Acquire multiple first predicted beam pairs, wherein the first predicted beam pairs have first reference signal received powers RSRP, and correlation values between different first RSRPs are less than a correlation threshold;
[0189] The correlation degree value and the multiple first RSRPs are input into a target beam prediction model to obtain a second RSRP corresponding to a second predicted beam pair, wherein the target beam prediction model has learned a mapping relationship between the correlation degree value, the multiple first RSRPs, and the second RSRP.
[0190] Among them, Figure 8In the embodiment, the bus architecture may include any number of interconnected buses and bridges, specifically one or more processors represented by processor 803 and various circuits of memory represented by memory 801 are linked together. The bus architecture may also link together various other circuits such as peripheral devices, voltage regulators, and power management circuits, which are well known in the art and are therefore not further described herein. The bus interface provides an interface. The transceiver 802 may be a plurality of components, namely, a transmitter and a receiver, providing a unit for communicating with various other devices on a transmission medium, which may include wireless channels, wired channels, optical cables, and other transmission media. For different user devices, the beam prediction device 80 may also include a user interface 804, which may be an interface capable of externally and internally connecting required devices, and the connected devices include but are not limited to a keypad, a display, a speaker, a microphone, a joystick, and the like.
[0191] The processor 803 is responsible for managing the bus architecture and general processing, and the memory 801 can store data used by the processor 803 when performing operations.
[0192] Optionally, the processor 803 may be a CPU (central processing unit), an ASIC (Application Specific Integrated Circuit), an FPGA (Field-Programmable Gate Array) or a CPLD (Complex Programmable Logic Device), and the processor may also adopt a multi-core architecture.
[0193] The processor calls the computer program stored in the memory to execute any method provided in the embodiments of the present application according to the obtained executable instructions. The processor and the memory can also be arranged physically separately.
[0194] In some embodiments of the present application, the target beam prediction model is trained based on the following method:
[0195] Acquire a plurality of sample beam pairs, wherein each of the sample beam pairs has a third RSRP;
[0196] According to the plurality of the third RSRPs, the plurality of sample beam pairs are divided into first sample beam pairs and second sample beam pairs, wherein a first correlation value between the third RSRPs of different first sample beam pairs is less than the correlation threshold, and a second correlation value between the third RSRPs of different second sample beam pairs is greater than or equal to the correlation threshold; and
[0197] An initial beam prediction model is trained based on the third RSRP of the first sample beam pair, the first correlation value, the third RSRP of the second sample beam pair, and the second correlation value to obtain the target beam prediction model.
[0198] In some embodiments of the present application, the processor 803 is further configured to:
[0199] Grouping the plurality of sample beam pairs to obtain a plurality of sample beam pair groups, wherein each of the sample beam pair groups includes at least two of the sample beam pairs;
[0200] Acquire a reference RSRP of the sample beam pair group, wherein the reference RSRP is determined by a third RSRP of at least two of the sample beam pairs in the sample beam pair group;
[0201] The plurality of sample beam pairs are divided into first sample beam pairs and second sample beam pairs according to the plurality of reference RSRPs and a first received power threshold.
[0202] In some embodiments of the present application, the processor 803 is further configured to:
[0203] If the reference RSRP is less than or equal to the first received power threshold, taking a sample beam pair in the sample beam pair group corresponding to the reference RSRP as the first sample beam pair;
[0204] If the reference RSRP is greater than the first received power threshold, a sample beam pair in the sample beam pair group corresponding to the reference RSRP is used as the second sample beam pair.
[0205] In some embodiments of the present application, the processor 803 is further configured to:
[0206] An average value of third RSRPs of at least two of the sample beam pairs in the sample beam pair group is determined, wherein the average value is used as the reference RSRP.
[0207] In some embodiments of the present application, the processor 803 is further configured to:
[0208] Training the initial beam prediction model based on the third RSRP of the first sample beam pair and the first correlation value to obtain a first beam prediction model;
[0209] Performing enhancement processing on the third RSRP of the second sample beam pair to obtain an enhanced second sample beam pair, wherein a third correlation value between the third RSRPs of different enhanced second sample beam pairs is less than the correlation threshold;
[0210] The first beam prediction model is trained based on the third RSRP of the enhanced second sample beam pair and the third correlation value to obtain the target beam prediction model.
[0211] In some embodiments of the present application, the number of the first sample beam pairs is multiple, and the multiple first sample beam pairs are composed of the first reference sample beam pair and the first predicted sample beam pair; wherein the processor 803 is further configured to:
[0212] Inputting the third RSRP of the first reference sample beam pair and the first correlation value into the initial beam prediction model to obtain a predicted RSRP of the first predicted sample beam pair output by the initial beam prediction model;
[0213] If the predicted RSRP of the first prediction sample beam pair and the third RSRP of the first prediction sample beam pair meet a convergence condition, the trained initial beam prediction model is used as the first beam prediction model.
[0214] In some embodiments of the present application, the number of the enhanced second sample beam pairs is multiple, and the enhanced second sample beam pairs are composed of a second reference sample beam pair and a second predicted sample beam pair; wherein the processor 803 is further configured to:
[0215] Inputting the third RSRP of the second reference sample beam pair and the third correlation value into the first beam prediction model to obtain the predicted RSRP of the second predicted sample beam pair output by the first beam prediction model;
[0216] If the predicted RSRP of the second prediction sample beam pair and the third RSRP of the second prediction sample beam pair meet the convergence condition, the trained first beam prediction model is used as the target beam prediction model.
[0217] In some embodiments of the present application, the processor 803 is further configured to:
[0218] Acquire enhancement processing parameters corresponding to the second sample beam pair;
[0219] Processing the third RSRP of the second sample beam pair according to the enhanced processing parameter to obtain the enhanced second sample beam pair;
[0220] The enhanced processing parameters include: one or more amplification processing parameters, wherein the amplification processing parameters are used to amplify RSRP, and the value of the amplification processing parameters is greater than 1. When the enhanced processing parameters include multiple amplification processing parameters, the values of the multiple amplification processing parameters are different.
[0221] In some embodiments of the present application, the number of the second sample beam pairs is multiple; wherein the processor 803 is further configured to:
[0222] Determine a second sample beam pair to be processed from a plurality of second sample beam pairs;
[0223] The step of processing the third RSRP of the second sample beam pair according to the enhanced processing parameter to obtain an enhanced second sample beam pair includes:
[0224] The third RSRP of the second sample beam pair to be processed is processed according to the enhanced processing parameter to obtain an enhanced second sample beam pair.
[0225] In some embodiments of the present application, the processor 803 is further configured to:
[0226] Selecting a third RSRP greater than a second received power threshold from a plurality of third RSRPs respectively corresponding to a plurality of second sample beam pairs;
[0227] A second sample beam pair corresponding to a third RSRP greater than the second receiving power threshold is used as a second sample beam pair to be processed.
[0228] It should be noted here that the above-mentioned device provided in the embodiment of the present application can implement all the method steps implemented in the above-mentioned method embodiment, and can achieve the same technical effect. The parts and beneficial effects of this embodiment that are the same as those in the method embodiment will not be described in detail here.
[0229] In order to implement the above embodiment, the embodiment of the present application proposes a processor-readable storage medium, wherein the processor-readable storage medium stores a computer program, and the computer program is used to enable the processor to execute a beam prediction method.
[0230] Those skilled in the art will appreciate that the embodiments of the present application may be provided as methods, systems, or computer program products. Therefore, the present application may adopt the form of a complete hardware embodiment, a complete software embodiment, or an embodiment in combination with software and hardware. Moreover, the present application may adopt the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage and optical storage, etc.) that contain computer-usable program code.
[0231] The present application is described with reference to the flowcharts and / or block diagrams of the methods, devices (systems), and computer program products according to the embodiments of the present application. It should be understood that each process and / or box in the flowchart and / or block diagram, as well as the combination of the processes and / or boxes in the flowchart and / or block diagram, can be implemented by computer executable instructions. These computer executable instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to generate a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowchart and / or block diagram. Figure 1 A process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.
[0232] These processor executable instructions may also be stored in a processor readable memory that can direct a computer or other programmable data processing device to operate in a specific manner, so that the instructions stored in the processor readable memory produce an article of manufacture including an instruction device, which implements the process Figure 1 A process or multiple processes and / or boxes Figure 1 A function specified in one or more boxes.
[0233] These processor-executable instructions may also be loaded onto a computer or other programmable data processing device so that a series of operational steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing instructions for executing on the computer or other programmable device to implement the process. Figure 1 A process or multiple processes and / or boxes Figure 1 The steps for the functions specified in one or more boxes.
[0234] Obviously, those skilled in the art can make various changes and modifications to the present application without departing from the spirit and scope of the present application. Thus, if these modifications and variations of the present application fall within the scope of the claims of the present application and their equivalents, the present application is also intended to include these modifications and variations.
[0235] It should be noted here that the above-mentioned device provided in the embodiment of the present application can implement all the method steps implemented in the above-mentioned method embodiment, and can achieve the same technical effect. The parts and beneficial effects of this embodiment that are the same as those in the method embodiment will not be described in detail here.
[0236] In order to implement the above embodiment, the embodiment of the present application proposes a processor-readable storage medium, wherein the processor-readable storage medium stores a computer program, and the computer program is used to enable the processor to execute a beam prediction method.
[0237] Those skilled in the art will appreciate that the embodiments of the present application may be provided as methods, systems, or computer program products. Therefore, the present application may adopt the form of a complete hardware embodiment, a complete software embodiment, or an embodiment in combination with software and hardware. Moreover, the present application may adopt the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage and optical storage, etc.) that contain computer-usable program code.
[0238] The present application is described with reference to the flowcharts and / or block diagrams of the methods, devices (systems), and computer program products according to the embodiments of the present application. It should be understood that each process and / or box in the flowchart and / or block diagram, as well as the combination of the processes and / or boxes in the flowchart and / or block diagram, can be implemented by computer executable instructions. These computer executable instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to generate a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowchart and / or block diagram. Figure 1 A process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.
[0239] These processor executable instructions may also be stored in a processor readable memory that can direct a computer or other programmable data processing device to operate in a specific manner, so that the instructions stored in the processor readable memory produce an article of manufacture including an instruction device, which implements the process Figure 1 A process or multiple processes and / or boxes Figure 1 A function specified in one or more boxes.
[0240] These processor-executable instructions may also be loaded onto a computer or other programmable data processing device so that a series of operational steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing instructions for executing on the computer or other programmable device to implement the process. Figure 1 A process or multiple processes and / or boxes Figure 1 The steps for the functions specified in one or more boxes.
[0241] Obviously, those skilled in the art can make various changes and modifications to the present application without departing from the spirit and scope of the present application. Thus, if these modifications and variations of the present application fall within the scope of the claims of the present application and their equivalents, the present application is also intended to include these modifications and variations.
[0242] It should be noted that, in the description of this application, the terms "first", "second", etc. are only used for descriptive purposes and cannot be understood as indicating or implying relative importance. In addition, in the description of this application, unless otherwise specified, the meaning of "plurality" is two or more.
[0243] Any process or method description in a flowchart or otherwise described herein may be understood to represent a module, segment or portion of code that includes one or more executable instructions for implementing the steps of a specific logical function or process, and the scope of the preferred embodiments of the present application includes alternative implementations in which functions may not be performed in the order shown or discussed, including performing functions in a substantially simultaneous manner or in the reverse order depending on the functions involved, which should be understood by technicians in the technical field to which the embodiments of the present application belong.
[0244] It should be understood that the various parts of the present application can be implemented by hardware, software, firmware or a combination thereof. In the above-mentioned embodiments, multiple steps or methods can be implemented by software or firmware stored in a memory and executed by a suitable instruction execution system. For example, if implemented by hardware, as in another embodiment, it can be implemented by any one of the following technologies known in the art or their combination: a discrete logic circuit having a logic gate circuit for implementing a logic function for a data signal, a dedicated integrated circuit having a suitable combination of logic gate circuits, a programmable gate array (PGA), a field programmable gate array (FPGA), etc.
[0245] A person skilled in the art may understand that all or part of the steps in the method for implementing the above-mentioned embodiment may be completed by instructing related hardware through a program, and the program may be stored in a computer-readable storage medium, which, when executed, includes one or a combination of the steps of the method embodiment.
[0246] In addition, each functional unit in each embodiment of the present application may be integrated into a processing module, or each unit may exist physically separately, or two or more units may be integrated into one module. The above-mentioned integrated module may be implemented in the form of hardware or in the form of a software functional module. If the integrated module is implemented in the form of a software functional module and sold or used as an independent product, it may also be stored in a computer-readable storage medium.
[0247] The storage medium mentioned above can be a read-only memory, a magnetic disk or an optical disk, etc.
[0248] In the description of this specification, the description with reference to the terms "one embodiment", "some embodiments", "example", "specific example", or "some examples" means that the specific features, structures, materials or characteristics described in conjunction with the embodiment or example are included in at least one embodiment or example of the present application. In this specification, the schematic representation of the above terms does not necessarily refer to the same embodiment or example. Moreover, the specific features, structures, materials or characteristics described may be combined in any one or more embodiments or examples in a suitable manner.
[0249] Although the embodiments of the present application have been shown and described above, it can be understood that the above embodiments are exemplary and cannot be understood as limitations on the present application. Ordinary technicians in this field can change, modify, replace and modify the above embodiments within the scope of the present application.
Claims
1. A beam prediction method, characterized in that: The method comprises: Acquire multiple first predicted beam pairs, wherein the first predicted beam pairs have first reference signal received powers RSRP, and correlation values between different first RSRPs are less than a correlation threshold; The correlation degree value and the multiple first RSRPs are input into a target beam prediction model to obtain a second RSRP corresponding to a second predicted beam pair, wherein the target beam prediction model has learned a mapping relationship between the correlation degree value, the multiple first RSRPs, and the second RSRP.
2. The method according to claim 1, characterized in that The target beam prediction model is trained based on the following method: Acquire a plurality of sample beam pairs, wherein each of the sample beam pairs has a third RSRP; According to the plurality of third RSRPs, the plurality of sample beam pairs are divided into first sample beam pairs and second sample beam pairs, wherein a first correlation value between the third RSRPs of different first sample beam pairs is less than the correlation threshold, and a second correlation value between the third RSRPs of different second sample beam pairs is greater than or equal to the correlation threshold; An initial beam prediction model is trained based on the third RSRP of the first sample beam pair, the first correlation value, the third RSRP of the second sample beam pair, and the second correlation value to obtain the target beam prediction model.
3. The method according to claim 2, characterized in that The step of dividing the plurality of sample beam pairs into first sample beam pairs and second sample beam pairs according to the plurality of third RSRPs comprises: Grouping the plurality of sample beam pairs to obtain a plurality of sample beam pair groups, wherein each of the sample beam pair groups includes at least two of the sample beam pairs; Acquire a reference RSRP of the sample beam pair group, wherein the reference RSRP is determined by a third RSRP of at least two of the sample beam pairs in the sample beam pair group; The plurality of sample beam pairs are divided into first sample beam pairs and second sample beam pairs according to the plurality of reference RSRPs and a first received power threshold.
4. The method according to claim 3, characterized in that The dividing the plurality of sample beam pairs into first sample beam pairs and second sample beam pairs according to the plurality of reference RSRPs and the first received power threshold comprises: If the reference RSRP is less than or equal to the first received power threshold, taking a sample beam pair in the sample beam pair group corresponding to the reference RSRP as the first sample beam pair; If the reference RSRP is greater than the first received power threshold, a sample beam pair in the sample beam pair group corresponding to the reference RSRP is used as the second sample beam pair.
5. The method according to claim 3, characterized in that The obtaining a reference RSRP of the sample beam pair group includes: An average value of third RSRPs of at least two of the sample beam pairs in the sample beam pair group is determined, wherein the average value is used as the reference RSRP.
6. The method according to claim 2, characterized in that The training of the initial beam prediction model based on the third RSRP of the first sample beam pair, the first correlation value, the third RSRP of the second sample beam pair, and the second correlation value to obtain the target beam prediction model includes: Training the initial beam prediction model based on the third RSRP of the first sample beam pair and the first correlation value to obtain a first beam prediction model; Performing enhancement processing on the third RSRP of the second sample beam pair to obtain an enhanced second sample beam pair, wherein a third correlation value between the third RSRPs of different enhanced second sample beam pairs is less than the correlation threshold; The first beam prediction model is trained based on the third RSRP of the enhanced second sample beam pair and the third correlation value to obtain the target beam prediction model.
7. The method according to claim 6, characterized in that The number of the first sample beam pairs is multiple, and the multiple first sample beam pairs are composed of the first reference sample beam pairs and the first prediction sample beam pairs; The step of training the initial beam prediction model based on the third RSRP of the first sample beam pair and the first correlation value to obtain the first beam prediction model includes: Inputting the third RSRP of the first reference sample beam pair and the first correlation value into the initial beam prediction model to obtain a predicted RSRP of the first predicted sample beam pair output by the initial beam prediction model; If the predicted RSRP of the first prediction sample beam pair and the third RSRP of the first prediction sample beam pair meet a convergence condition, the trained initial beam prediction model is used as the first beam prediction model.
8. The method according to claim 6, characterized in that The number of the enhanced second sample beam pairs is multiple, and the enhanced second sample beam pairs are composed of a second reference sample beam pair and a second predicted sample beam pair; The step of training the first beam prediction model based on the third RSRP of the enhanced second sample beam pair and the third correlation value to obtain the target beam prediction model includes: Inputting the third RSRP of the second reference sample beam pair and the third correlation value into the first beam prediction model to obtain the predicted RSRP of the second predicted sample beam pair output by the first beam prediction model; If the predicted RSRP of the second prediction sample beam pair and the third RSRP of the second prediction sample beam pair meet the convergence condition, the trained first beam prediction model is used as the target beam prediction model.
9. The method according to claim 6, characterized in that The step of enhancing the third RSRP of the second sample beam pair to obtain an enhanced second sample beam pair includes: Acquire enhancement processing parameters corresponding to the second sample beam pair; Processing the third RSRP of the second sample beam pair according to the enhanced processing parameter to obtain the enhanced second sample beam pair; The enhanced processing parameters include: one or more amplification processing parameters, wherein the amplification processing parameters are used to amplify RSRP, and the value of the amplification processing parameters is greater than 1. When the enhanced processing parameters include multiple amplification processing parameters, the values of the multiple amplification processing parameters are different.
10. The method according to claim 9, characterized in that The number of the second sample beam pairs is multiple; the method further includes: Determine a second sample beam pair to be processed from a plurality of the second sample beam pairs; The step of processing the third RSRP of the second sample beam pair according to the enhanced processing parameter to obtain the enhanced second sample beam pair includes: The third RSRP of the second sample beam pair to be processed is processed according to the enhanced processing parameter to obtain the enhanced second sample beam pair.
11. The method according to claim 10, characterized in that The step of determining a second sample beam pair to be processed from the plurality of second sample beam pairs comprises: Selecting a third RSRP greater than a second received power threshold from a plurality of third RSRPs respectively corresponding to a plurality of the second sample beam pairs; The second sample beam pair corresponding to the third RSRP greater than the second receiving power threshold is used as the second sample beam pair to be processed.
12. A beam prediction device, characterized in that: The device comprises: an acquisition unit, configured to acquire a plurality of first predicted beam pairs, wherein the first predicted beam pairs have a first reference signal received power RSRP, and correlation values between different first RSRPs are less than a correlation threshold; A processing unit is used to input the correlation value and multiple first RSRPs into a target beam prediction model to obtain a second RSRP corresponding to a second predicted beam pair, wherein the target beam prediction model has learned a mapping relationship between the correlation value, multiple first RSRPs, and the second RSRP.
13. A beam prediction device, characterized in that: The invention comprises a memory, a transceiver and a processor: the memory is used to store a computer program; the transceiver is used to send and receive data under the control of the processor; the processor is used to read the computer program in the memory and perform the following operations: Acquire multiple first predicted beam pairs, wherein the first predicted beam pairs have first reference signal received powers RSRP, and correlation values between different first RSRPs are less than a correlation threshold; The correlation degree value and the multiple first RSRPs are input into a target beam prediction model to obtain a second RSRP corresponding to a second predicted beam pair, wherein the target beam prediction model has learned a mapping relationship between the correlation degree value, the multiple first RSRPs, and the second RSRP.
14. The device according to claim 13, characterized in that The processor is further configured to: The target beam prediction model is trained based on the following method: Acquire a plurality of sample beam pairs, wherein each of the sample beam pairs has a third RSRP; According to the plurality of the third RSRPs, the plurality of sample beam pairs are divided into first sample beam pairs and second sample beam pairs, wherein a first correlation value between the third RSRPs of different first sample beam pairs is less than the correlation threshold, and a second correlation value between the third RSRPs of different second sample beam pairs is greater than or equal to the correlation threshold; and An initial beam prediction model is trained based on the third RSRP of the first sample beam pair, the first correlation value, the third RSRP of the second sample beam pair, and the second correlation value to obtain the target beam prediction model.
15. The device according to claim 14, characterized in that The processor is further configured to: Grouping the plurality of sample beam pairs to obtain a plurality of sample beam pair groups, wherein each of the sample beam pair groups includes at least two of the sample beam pairs; Acquire a reference RSRP of the sample beam pair group, wherein the reference RSRP is determined by a third RSRP of at least two of the sample beam pairs in the sample beam pair group; The plurality of sample beam pairs are divided into first sample beam pairs and second sample beam pairs according to the plurality of reference RSRPs and a first received power threshold.
16. The device according to claim 15, characterized in that The processor is further configured to: If the reference RSRP is less than or equal to the first received power threshold, taking a sample beam pair in the sample beam pair group corresponding to the reference RSRP as the first sample beam pair; If the reference RSRP is greater than the first received power threshold, a sample beam pair in the sample beam pair group corresponding to the reference RSRP is used as the second sample beam pair.
17. The device according to claim 15, characterized in that The processor is further configured to: An average value of third RSRPs of at least two of the sample beam pairs in the sample beam pair group is determined, wherein the average value is used as the reference RSRP.
18. The device according to claim 14, characterized in that The processor is further configured to: Training the initial beam prediction model based on the third RSRP of the first sample beam pair and the first correlation value to obtain a first beam prediction model; Performing enhancement processing on the third RSRP of the second sample beam pair to obtain an enhanced second sample beam pair, wherein a third correlation value between the third RSRPs of different enhanced second sample beam pairs is less than the correlation threshold; The first beam prediction model is trained based on the third RSRP of the enhanced second sample beam pair and the third correlation value to obtain the target beam prediction model.
19. The device according to claim 18, characterized in that The number of the first sample beam pairs is multiple, and the multiple first sample beam pairs are composed of the first reference sample beam pairs and the first prediction sample beam pairs; Wherein, the processor is further used for: Inputting the third RSRP of the first reference sample beam pair and the first correlation value into the initial beam prediction model to obtain a predicted RSRP of the first predicted sample beam pair output by the initial beam prediction model; If the predicted RSRP of the first prediction sample beam pair and the third RSRP of the first prediction sample beam pair meet a convergence condition, the trained initial beam prediction model is used as the first beam prediction model.
20. The device according to claim 18, characterized in that The number of the enhanced second sample beam pairs is multiple, and the enhanced second sample beam pairs are composed of a second reference sample beam pair and a second predicted sample beam pair; Wherein, the processor is further used for: Inputting the third RSRP of the second reference sample beam pair and the third correlation value into the first beam prediction model to obtain the predicted RSRP of the second predicted sample beam pair output by the first beam prediction model; If the predicted RSRP of the second prediction sample beam pair and the third RSRP of the second prediction sample beam pair meet the convergence condition, the trained first beam prediction model is used as the target beam prediction model.
21. The device according to claim 18, characterized in that The processor is further configured to: Acquire enhancement processing parameters corresponding to the second sample beam pair; Processing the third RSRP of the second sample beam pair according to the enhanced processing parameter to obtain the enhanced second sample beam pair; The enhanced processing parameters include: one or more amplification processing parameters, wherein the amplification processing parameters are used to amplify RSRP, and the value of the amplification processing parameters is greater than 1. When the enhanced processing parameters include multiple amplification processing parameters, the values of the multiple amplification processing parameters are different.
22. The device according to claim 21, characterized in that The number of the second sample beam pairs is multiple; wherein the processor is further used to: Determine a second sample beam pair to be processed from a plurality of the second sample beam pairs; The step of processing the third RSRP of the second sample beam pair according to the enhanced processing parameter to obtain the enhanced second sample beam pair includes: The third RSRP of the second sample beam pair to be processed is processed according to the enhanced processing parameter to obtain the enhanced second sample beam pair.
23. The device according to claim 22, characterized in that The processor is further configured to: Selecting a third RSRP greater than a second received power threshold from a plurality of third RSRPs respectively corresponding to a plurality of the second sample beam pairs; The second sample beam pair corresponding to the third RSRP greater than the second receiving power threshold is used as the second sample beam pair to be processed.
24. A processor-readable storage medium, characterized in that: The processor-readable storage medium stores a computer program, and the computer program is used to enable the processor to execute the beam prediction method according to any one of claims 1 to 11.
Citation Information
Cited By
Method for reconstructing wave beam characteristic parameters, method for determining wave beam, method for reconstructing object characteristic parameters and equipment
CN120357974A
Methods for reconstructing beam characteristic parameters, methods for determining beams, methods and equipment for reconstructing object characteristic parameters.
CN120357974B
Beam selection method and device, communication system, storage medium and chip system
CN121013090A
Beam selection methods, devices, communication systems, storage media, and chip systems
CN121013090B
