Transmission mode determination method and device, computer equipment and storage medium

Abstract

The invention provides a transmission mode determination method and device, computer equipment and a storage medium, and the method comprises the steps: obtaining channel feature data of a to-be-configured object, and determining an online reasoning network corresponding to the to-be-configured object; inputting the channel characteristic data into an online reasoning network to obtain a first transmission mode corresponding to the object to be configured; and performing data transmission on the object to be configured according to the first transmission mode. According to the method and the device, the first transmission mode does not need to be determined through a theoretical calculation mode, so that inaccurate calculation results caused by measurement errors are avoided, the accuracy of transmission mode determination can be improved, and the method and the device can be flexibly adapted to different scenes. And meanwhile, the first transmission mode does not need to be determined by a transmission mode selection algorithm based on heuristic adjustment, so that the heuristic overhead is saved, and the application range of the method and the device is expanded.

Description

Technical Field

[0001] The present disclosure relates to the field of communication technology, and in particular to a transmission mode determination method, apparatus, computer equipment, and storage medium. Background Art

[0002] In a wireless transmission system, different data transmission modes can be set for different users to improve data transmission efficiency. The data transmission mode refers to the beamforming scheme adopted by the base station for downlink users, including but not limited to BC (BroadCast) weight forming, PMI (Precoding Matrix Indicator) weight forming, and BF (BeamForming) weight forming.

[0003] In the related scheme for determining the data transmission mode, the communication channel corresponding to the user can be measured, and based on the measurement results, a data transmission mode with higher theoretical performance can be selected through theoretical calculation. However, this scheme does not take into account the inaccurate calculation results caused by measurement errors.

[0004] Alternatively, a transmission mode selection algorithm based on trial adjustment can be used. Specifically, a data transmission mode with higher performance can be determined by adjusting the data transmission mode in real time during the data transmission process. However, this solution has strong limitations. It is not effective for small packet users with small data transmission volume. It also causes additional trial overhead for large packet users with long data transmission time, thereby reducing the transmission performance of large packet users. This is because trial does not guarantee positive performance gain and may result in trial errors. Summary of the invention

[0005] The embodiments of the present disclosure at least provide a transmission mode determination method, apparatus, computer equipment, and storage medium.

[0006] In a first aspect, an embodiment of the present disclosure provides a transmission mode determination method, including:

[0007] Acquire channel characteristic data of the object to be configured, and determine an online inference network corresponding to the object to be configured, wherein the channel characteristic data is used to indicate characteristics of a communication channel between the object to be configured and a base station;

[0008] Inputting the channel characteristic data into an online inference network to obtain a first transmission mode corresponding to the object to be configured, where the first transmission mode is used to indicate a beamforming mode and a number of transmission streams corresponding to the object to be configured;

[0009] According to the first transmission mode, data is transmitted to the object to be configured.

[0010] In an optional implementation, the determining the online reasoning network corresponding to the object to be configured includes:

[0011] Constructing a training data set, and training the training network based on the training data set to obtain an offline training network;

[0012] According to a preset parameter transmission cycle, the network parameters of the offline training network are periodically acquired, and the network parameters are transmitted to the online inference network.

[0013] In an optional implementation, the constructing of the training data set includes:

[0014] Collecting historical channel characteristic data and transmission result data corresponding to the base station;

[0015] Determining a transmission success rate of the historical channel characteristic data based on the transmission result data;

[0016] Determine valid sample data in which the transmission success rate is within a convergence interval in the historical channel characteristic data, and determine the historical transmission performance corresponding to the valid sample data as a training label;

[0017] A training data set is determined based on the sample data and the corresponding training labels.

[0018] In an optional implementation manner, the collecting the historical channel characteristic data and transmission result data corresponding to the base station includes:

[0019] In the process of collecting the historical channel characteristic data, real-time statistics are collected between adjacent historical channel characteristic data;

[0020] Clear the historical channel characteristic data and the corresponding transmission result data whose collection interval exceeds the collection threshold in the storage space;

[0021] Based on the historical channel characteristic data and the corresponding transmission result data stored in the storage space, a step of determining sample data whose transmission success rate satisfies the convergence condition is performed.

[0022] In an optional implementation manner, the training of the to-be-trained network based on the training data set to obtain an offline training network includes:

[0023] Determine the number of samples in the training data set, and when the number of samples meets the dimensionality reduction condition, determine the data to be reduced in dimensionality in the sample data;

[0024] The data to be reduced in dimension is subjected to dimensionality reduction processing to obtain reduced-dimensionality data, and a network to be trained is trained based on the reduced-dimensionality data to obtain an offline training network.

[0025] In an optional implementation manner, the step of performing dimensionality reduction processing on the data to be dimensionality reduced to obtain the dimensionality reduced data includes:

[0026] When the data to be reduced in dimensionality includes multidimensional data for characterizing similar channel characteristics, based on the dimensional direction of the multidimensional data, performing accumulation processing on the multidimensional data to obtain aggregated data; and

[0027] When the data to be reduced in dimensionality include data to be classified, clustering processing is performed on the data to be classified to obtain classified data.

[0028] In an optional implementation, the online inference network includes a plurality of regression networks, wherein each regression network corresponds to at least one preset transmission mode;

[0029] The step of inputting the channel characteristic data into an online inference network to obtain a first transmission mode corresponding to the object to be configured includes:

[0030] The channel characteristic data is respectively transmitted to each of the regression networks to obtain a performance estimation value output by each of the regression networks, wherein the performance estimation value is used to indicate the transmission performance when data is transmitted to the object to be configured through the preset transmission mode;

[0031] The preset transmission mode corresponding to the regression network with the best estimation value is determined as the first transmission mode.

[0032] In an optional implementation, the online reasoning network includes a classification network;

[0033] The step of inputting the channel characteristic data into an online inference network to obtain a first transmission mode corresponding to the object to be configured includes:

[0034] Modifying the channel characteristic data based on a transmission stream number parameter corresponding to the classified network;

[0035] The modified channel characteristic data is input into the classification network to obtain a first transmission mode corresponding to the object to be configured.

[0036] In an optional implementation manner, the method further includes:

[0037] Before performing data transmission for the object to be configured according to the first transmission mode, acquiring historical configuration information, and storing the first transmission mode in the historical configuration information, wherein the historical configuration information includes the historical transmission mode configured for the object to be configured;

[0038] A second transmission mode in which the configuration frequency meets the frequency condition is determined in the historical configuration information, so as to perform data transmission on the object to be configured according to the second transmission mode.

[0039] In a second aspect, an embodiment of the present disclosure further provides a transmission mode determination device, including:

[0040] An acquisition unit, used to acquire channel characteristic data of an object to be configured, and determine an online inference network corresponding to the object to be configured, wherein the channel characteristic data is used to indicate characteristics of a communication channel between the object to be configured and a base station;

[0041] An online reasoning unit, used for inputting the channel characteristic data into an online reasoning network to obtain a first transmission mode corresponding to the object to be configured, wherein the first transmission mode is used for indicating a beamforming angle and a number of transmission streams corresponding to the object to be configured;

[0042] A data transmission unit is used to transmit data to the object to be configured according to the first transmission mode.

[0043] In a third aspect, an embodiment of the present disclosure further provides a computer device, comprising: a processor, a memory and a bus, wherein the memory stores machine-readable instructions executable by the processor, and when the computer device is running, the processor and the memory communicate via the bus, and when the machine-readable instructions are executed by the processor, the steps of the above-mentioned first aspect, or any possible implementation of the first aspect are performed.

[0044] In a fourth aspect, the present application further provides a computer storage medium storing computer executable instructions, wherein the computer executable instructions are used to execute the transmission mode determination method described in any one of the above items of the present application.

[0045] The above technical solution provided by the embodiment of the present application has the following advantages over the prior art: the method provided by the embodiment of the present application can first obtain the channel characteristic data of the object to be configured, and determine the online inference network corresponding to the object to be configured, wherein the channel characteristic data is used to indicate the characteristics of the communication channel between the object to be configured and the base station. Next, the channel characteristic data can be input into the online inference network to obtain the first transmission mode corresponding to the object to be configured, and the first transmission mode is used to indicate the beamforming mode and the number of transmission streams corresponding to the object to be configured. Then, according to the first transmission mode, the object to be configured can be transmitted with data, so that there is no need to determine the first transmission mode by theoretical calculation, thus avoiding the inaccurate calculation results caused by measurement errors, and can improve the accuracy of transmission mode determination, and flexibly adapt to different scenarios. At the same time, there is no need to determine the first transmission mode based on the transmission mode selection algorithm adjusted by trial and error, which saves the trial overhead and improves the scope of application of the present disclosure. BRIEF DESCRIPTION OF THE DRAWINGS

[0046] In order to more clearly illustrate the technical solutions of the embodiments of the present disclosure, the following is a brief introduction to the drawings required for use in the embodiments. The drawings herein are incorporated into the specification and constitute a part of the specification. These drawings illustrate embodiments consistent with the present disclosure and are used together with the specification to illustrate the technical solutions of the present disclosure. It should be understood that the following drawings only illustrate certain embodiments of the present disclosure and should not be regarded as limiting the scope. For ordinary technicians in this field, other relevant drawings can also be obtained based on these drawings without creative work.

[0047] Figure 1 A flow chart of a transmission mode determination method provided by an embodiment of the present disclosure is shown;

[0048] Figure 2 A schematic diagram showing a process of updating an online reasoning network provided by an embodiment of the present disclosure is shown;

[0049] Figure 3 An architecture diagram of a transmission mode determination system provided by an embodiment of the present disclosure is shown;

[0050] Figure 4 A schematic diagram of a transmission mode determination device provided by an embodiment of the present disclosure is shown;

[0051] Figure 5 A schematic diagram of a computer device provided by an embodiment of the present disclosure is shown. DETAILED DESCRIPTION

[0052] In order to make the purpose, technical scheme and advantages of the embodiments of the present disclosure clearer, the technical scheme in the embodiments of the present disclosure will be clearly and completely described below in conjunction with the drawings in the embodiments of the present disclosure. Obviously, the described embodiments are only part of the embodiments of the present disclosure, rather than all of the embodiments. The components of the embodiments of the present disclosure generally described and shown in the drawings here can be arranged and designed in various different configurations. Therefore, the following detailed description of the embodiments of the present disclosure provided in the drawings is not intended to limit the scope of the present disclosure for protection, but merely represents the selected embodiments of the present disclosure. Based on the embodiments of the present disclosure, all other embodiments obtained by those skilled in the art without making creative work belong to the scope of protection of the present disclosure.

[0053] It should be noted that similar reference numerals and letters denote similar items in the following drawings, and therefore, once an item is defined in one drawing, further definition and explanation thereof is not required in subsequent drawings.

[0054] The term "and / or" herein only describes an association relationship, indicating that three relationships may exist. For example, A and / or B may represent the following three situations: A exists alone, A and B exist at the same time, and B exists alone. In addition, the term "at least one" herein represents any combination of at least two of any one or more of a plurality of. For example, including at least one of A, B, and C may represent including any one or more elements selected from the set consisting of A, B, and C.

[0055] Research has found that in a wireless transmission system, different data transmission modes can be set for different users to improve data transmission efficiency. The data transmission mode refers to the beamforming scheme adopted by the base station for downlink users, including but not limited to BC (BroadCast) weight shaping, PMI (Precoding Matrix Indicator) weight shaping, and BF (BeamForming) weight shaping.

[0056] In the related scheme for determining the data transmission mode, the communication channel corresponding to the user can be measured, and based on the measurement results, a data transmission mode with higher theoretical performance can be selected through theoretical calculation. However, this scheme does not take into account the inaccurate calculation results caused by measurement errors.

[0057] Alternatively, a transmission mode selection algorithm based on trial adjustment can be used. Specifically, a data transmission mode with higher performance can be determined by adjusting the data transmission mode in real time during the data transmission process. However, this solution has strong limitations. It is not effective for small packet users with small data transmission volume. It also causes additional trial overhead for large packet users with long data transmission time, thereby reducing the transmission performance of large packet users. This is because trial does not guarantee positive performance gain and may result in trial errors.

[0058] Based on the above research, the present disclosure provides a transmission mode determination method, device, computer equipment and storage medium. In an embodiment of the present disclosure, the channel characteristic data of the object to be configured is obtained, and the online inference network corresponding to the object to be configured is determined, wherein the channel characteristic data is used to indicate the characteristics of the communication channel between the object to be configured and the base station. Next, the channel characteristic data can be input into the online inference network to obtain the first transmission mode corresponding to the object to be configured, and the first transmission mode is used to indicate the beamforming mode and the number of transmission streams corresponding to the object to be configured. Then, according to the first transmission mode, the object to be configured can be transmitted data, thereby eliminating the need to determine the first transmission mode by theoretical calculation, thus avoiding inaccurate calculation results caused by measurement errors, and being able to improve the accuracy of transmission mode determination and flexibly adapt to different scenarios. At the same time, there is no need to determine the first transmission mode based on a transmission mode selection algorithm adjusted by trial and error, which saves trial and error overhead and improves the scope of application of the present disclosure.

[0059] To facilitate understanding of this embodiment, a transmission mode determination method disclosed in an embodiment of the present disclosure is first introduced in detail. The execution subject of the transmission mode determination method provided in the embodiment of the present disclosure is generally a computer device with certain computing capabilities. In some possible implementations, the transmission mode determination method can be implemented by a processor calling a computer-readable instruction stored in a memory.

[0060] See also Figure 1 FIG. 1 is a flow chart of a transmission mode determination method provided by an embodiment of the present disclosure, the method comprising steps S101 to S105, wherein:

[0061] S101: Acquire channel characteristic data of an object to be configured, and determine an online inference network corresponding to the object to be configured, wherein the channel characteristic data is used to indicate characteristics of a communication channel between the object to be configured and a base station.

[0062] In the disclosed embodiment, the channel characteristic data may include: channel signal-to-interference-to-noise ratio, channel received power, channel condition number, channel measurement period, channel angle domain projection, user chip type, user moving speed, neighboring area interference, etc. It should be understood that the channel characteristic data is data related to the user's selection of data transmission mode.

[0063] Specifically, the above-mentioned channel signal-to-interference plus noise ratio may include: CQISINR (Channel Quality Indicator Signal-to-Interference plus Noise Ratio, channel quality indicator signal interference plus noise ratio) and SRSSINR (Signal-to-Interference plus Noise Ratio of Sounding Reference Signal, signal to interference plus noise ratio of detection reference signal). Among them, CQISINR refers to the CQISINR value calculated based on the number of data streams fed back by the user (hereinafter referred to as RI) and CQI, and a corresponding CQISINR value is maintained under different RIs. SRSSINR refers to the average of SRSSINR measured at different ports. The above-mentioned channel receiving power, i.e. RSRP (Reference Signal Receiving Power, reference signal receiving power), refers to the receiving power of the reference signal measured by the user terminal. The channel condition number refers to the ratio of the maximum singular value to the minimum singular value after the base station performs SVD (sigular value decomposition) on H to obtain several singular values ​​by performing uplink channel estimation based on the SRS (Sounding Reference Signal) link.

[0064] In addition, the channel angle domain projection (also known as the projection energy vector) refers to the base station performing uplink channel estimation based on the SRS link to obtain the channel vector h; calculating the projection energy value of h on several preset beam vectors; normalizing these values ​​to form a projection energy vector. Since the preset beam vectors are distributed in a two-dimensional angle domain (for example, vertical 4*horizontal 8=32 preset beam vectors), the projection energy vector can indicate the projection energy distribution of the user channel in the two-dimensional angle domain. The chip type refers to the corresponding index obtained by looking up the table according to the chip type of the user terminal. The chip type can be used to indicate the transmission mode that the user chip can support. The user movement speed refers to the user movement speed estimated based on the Doppler frequency deviation or other methods. Neighboring cell interference refers to the interference measurement value obtained based on certain neighboring cell interference measurement methods.

[0065] In the disclosed embodiment, the channel characteristic data of the object to be configured can be collected by the channel measurement module, and the specific data collection method is not described in detail here. After collecting the channel characteristic data, the online reasoning network corresponding to the object to be configured can be determined, wherein the online reasoning network is a deep learning network, and the corresponding transmission mode can be determined for the object to be configured based on the input channel characteristic data.

[0066] S103: Input the channel characteristic data into an online inference network to obtain a first transmission mode corresponding to the object to be configured, where the first transmission mode is used to indicate a beamforming mode and a number of transmission streams corresponding to the object to be configured.

[0067] S105: Perform data transmission on the object to be configured according to the first transmission mode.

[0068] In an embodiment of the present disclosure, multiple preset transmission modes can be set, and a first transmission mode corresponding to the object to be configured can be determined among these multiple transmission modes. The first transmission mode can be the transmission mode with the highest transmission performance when transmitting data with the object to be configured among the preset transmission modes.

[0069] Here, the above-mentioned preset transmission mode may include: BC weight, PMI weight, BF weight, wherein the BF weight also distinguishes different numbers of streams. The BC and PMI modes do not need to distinguish the number of streams, because the number of transmission streams in the BC mode is generally 1 stream, and the PMI mode generally uses the RI (Rank Indicator) reported by the UE as the number of transmission streams, and the base station cannot change the number of transmission streams in the PMI mode.

[0070] Specifically, BC weight refers to the inherent, pre-configured, downlink broadcast shaping weight with wide beam coverage. PMI weight refers to the user reporting PMI and RI based on the CSI-RS (Channel State Information-Reference Signal) link, and the base station obtains the pre-configured shaping weight based on PMI and RI lookup table. BF weight refers to the base station performing uplink channel estimation based on the SRS (Sounding Reference Signal) link, and then applying SVD (Singular Value Decomposition) or ZF (Zero-Forcing) or other shaping algorithms to the channel matrix based on the number of transmission streams to obtain a non-pre-configured shaping weight that matches the real-time channel state.

[0071] It should be understood that the preset transmission mode includes but is not limited to the above three methods. As long as the base station adopts a beamforming scheme for downlink users, it can be classified into the category of the preset transmission mode.

[0072] After determining the above-mentioned first transmission mode, the base station can transmit data to the object to be configured based on the first transmission mode, thereby improving the user's spectrum efficiency, transmission rate and user experience. It should be understood that in the process of data transmission, the channel characteristic data, the first transmission mode and the transmission result determination data corresponding to the object to be configured can be collected through the data collection, storage and screening module to determine and construct a training data set based on the channel characteristic data, the first transmission mode and the transmission result data, so as to train the above-mentioned online inference network.

[0073] It can be known from the above description that in an embodiment of the present disclosure, the channel characteristic data of the object to be configured is obtained, and the online inference network corresponding to the object to be configured is determined, wherein the channel characteristic data is used to indicate the characteristics of the communication channel between the object to be configured and the base station. Next, the channel characteristic data can be input into the online inference network to obtain the first transmission mode corresponding to the object to be configured, and the first transmission mode is used to indicate the beamforming mode and the number of transmission streams corresponding to the object to be configured. Then, according to the first transmission mode, data transmission can be performed on the object to be configured, thereby eliminating the need to determine the first transmission mode by theoretical calculation, thus avoiding inaccurate calculation results caused by measurement errors, improving the accuracy of transmission mode determination, and flexibly adapting to different scenarios. At the same time, there is no need to determine the first transmission mode based on a transmission mode selection algorithm adjusted by trial and error, thus saving trial overhead and improving the scope of application of the present disclosure.

[0074] In an optional implementation, the above step S101, determining the online reasoning network corresponding to the object to be configured, specifically includes the following process:

[0075] S1011: Construct a training data set, and train the network to be trained based on the training data set to obtain an offline training network.

[0076] S1012: According to a preset parameter transmission cycle, periodically obtain network parameters of the offline training network, and transmit the network parameters to the online inference network.

[0077] In the embodiment of the present disclosure, considering that real-time training of the online inference network will take up a certain amount of training time, an offline training network identical to the online inference network may be set to train the offline training network in real time based on the training data set, thereby updating the network parameters of the offline training network and transferring the updated network parameters to the online inference network.

[0078] Here, firstly, the training data can be collected by the above data collection, storage, and screening modules to construct a training data set. The specific method of constructing the training data set is as follows, which will not be described in detail here. It should be understood that the training data can be user data of users in the area corresponding to the above base station.

[0079] like Figure 2 The figure shows a schematic diagram of the process of updating the online inference network, wherein firstly, the data acquisition, storage, and screening modules can be started according to the configured period TrainPeriod, and the running time of the data acquisition, storage, and screening modules after each startup is DataProcessTime, wherein TrainPeriod and DataProcessTime are pre-configured parameters.

[0080] After the data collection, storage, and screening module collects the training data, a training data set can be determined based on the training data, and the training data set is sent to the offline training module. At this time, the offline training module starts to start and trains the offline training network based on the training data set. The running time of the offline training module is TrainTime.

[0081] It should be understood that TrainTime is a random quantity due to fluctuations in offline computing power and the size of the training data set. Specifically, the maximum possible value of TrainTime cannot be greater than TrainPeriod-DataProcessTime.

[0082] For example, when TrainTime is greater than TrainPeriod-DataProcessTime, the offline training module can be instructed to stop training, delete the training results and not pass the training results to the online reasoning module, thereby improving the reliability of the training results, wherein the training results include the above-mentioned network parameters.

[0083] After the offline training module is completed, the trained network parameters can be periodically transferred to the online reasoning module according to the preset parameter transmission cycle. Specifically, the online reasoning module uses the network parameters until the network parameters are updated in the next preset parameter transmission cycle. If the offline training module does not transfer the network parameters due to training failure in the next preset parameter transmission cycle, the online reasoning module continues to use the previous network parameters.

[0084] In the embodiment of the present disclosure, considering that real-time training of the online inference network will take up a certain amount of training time, an offline training network identical to the online inference network can be set to train the offline training network in real time based on the training data set, thereby updating the network parameters of the offline training network, and passing the updated network parameters to the online inference network, thereby not affecting the use of the online inference network.

[0085] In an optional implementation, the above step S1011, constructing a training data set, specifically includes the following process:

[0086] S11: Collect historical channel characteristic data and transmission result data corresponding to the base station.

[0087] S12: Determine the transmission success rate of the historical channel characteristic data based on the transmission result data.

[0088] S13: Determine valid sample data whose transmission success rate is within a convergence interval in the historical channel characteristic data, and determine the historical transmission performance corresponding to the valid sample data as a training label.

[0089] S14: Determine a training data set based on the sample data and the corresponding training labels.

[0090] In the disclosed embodiment, firstly, the training data corresponding to the base station can be collected through the above-mentioned data collection, storage and screening module. The training data includes historical channel characteristic data and transmission result data, and distinguishes the UEid corresponding to each historical channel characteristic data and transmission result data. Here, the UEid is used to identify the user's identity information.

[0091] After the above training data is collected, the training data can be stored in the corresponding UEid storage space. It should be understood that the storage space corresponding to each UEid includes multiple sub-areas, each sub-area is used to store the historical channel characteristic data and transmission result data corresponding to the corresponding transmission mode.

[0092] For example, as shown in Table 1, this is the storage space corresponding to the UEid, wherein the storage space includes the sub-area corresponding to the above-mentioned BC weight, the sub-area corresponding to the PMI weight, the sub-area corresponding to the BF weight (the number of transmission streams is 1, that is, LayerNum=1 in the table), the sub-area corresponding to the BF weight (the number of transmission streams is 2, that is, LayerNum=2 in the table), the sub-area corresponding to the BF weight (the number of transmission streams is 3, that is, LayerNum=3 in the table), and the sub-area corresponding to the BF weight (the number of transmission streams is 4, that is, LayerNum=4 in the table).

[0093]

[0094] Table 1 Storage space corresponding to UEid

[0095] It should be understood that after collecting the above training data, the data collection, storage and screening module can store the training data in a sub-area of ​​the corresponding storage space based on the UEid, transmission mode and number of transmission streams corresponding to the training data.

[0096] Specifically, based on the data in the above sub-areas, sample data [projected energy vector (i.e., channel angle domain projection in channel characteristic data), channel condition number, CQISINR, SRSSINR, RSRP, SRS period, CSI-RS period, chip type, user mobile speed, neighboring area interference, other user channel characteristics, transmission performance (e.g., spectrum efficiency SE)] can be constructed according to a preset sample template. Among them, the CQISINR under BC weight and PMI weight is the CQISINR value of RI=1, and the CQISINR under BF weight is the CQISINR value of RI=LayerNum. SE is obtained by looking up the table based on the user's MCS (Mymova Checkin System, modulation and coding strategy) and LayerNum. At the same time, the timestamp corresponding to the sample data and the transmission result data can also be stored, and the transmission result data includes Ack or Nack, wherein the Ack indicates that the data packet is correctly transmitted, and the Nack is used to indicate that the data packet is erroneously transmitted.

[0097] Next, based on the transmission result data, valid sample data in a convergence interval may be determined in the sample data. The convergence interval may be used to indicate a success rate range of data transmission.

[0098] In specific implementation, first configure a time window with a length of WindowLen, and the time threshold corresponding to the time window is TimeThr. Add the collected sample data to the time window one by one. When the number of sample data in the time window is equal to WindowLen, make the following judgment. If the maximum value of the timestamp interval between adjacent sample data is not greater than TimeThr, the transmission success rate corresponding to the sample data can be calculated based on the transmission result data, where the transmission success rate can be: (the number of sample data Acks in the time window) / (the total number of sample data in the time window). If the maximum value of the timestamp interval between adjacent sample data is greater than TimeThr, delete all samples in the time window, thereby filtering out sample data with longer time intervals.

[0099] Then, the success rate range [BlerLowThr, BlerHighThr] indicated by the convergence interval can be obtained, and the sample data with a transmission success rate within the success rate range can be determined as valid sample data, and the historical transmission performance corresponding to the valid sample data can be determined as a training label.

[0100] It should be understood that the above WindowLen, TimeThr, BlerLowThr, and BlerHighThr are all pre-configured parameters, and the specific configuration method can be adjusted according to the actual usage scenario, which is not limited in the present disclosure.

[0101] In the disclosed embodiments, it is considered that the transmission performance (e.g., spectral efficiency SE) of the sample data in the non-convergent case is not the actual transmission performance corresponding to the current channel conditions. If the non-real transmission performance data is retained, it will lead to model learning bias, which will lead to a decrease in the performance of the transmission mode adaptation algorithm of the offline training network. Therefore, when constructing a training data set, the convergence of the above sample data can be judged based on the pre-configured WindowLen, TimeThr, BlerLowThr, and BlerHighThr parameters through the data acquisition, storage, and screening modules, thereby improving the reliability of the sample data and ensuring the training effect of the offline training network through the training data set.

[0102] In an optional implementation, the above step S11, collecting the historical channel characteristic data and transmission result data corresponding to the base station, specifically includes the following process:

[0103] (1) In the process of collecting the historical channel characteristic data, real-time statistics are collected between adjacent historical channel characteristic data;

[0104] (2) clearing the historical channel characteristic data and the corresponding transmission result data in the storage space where the collection interval exceeds the collection threshold;

[0105] (3) Based on the historical channel characteristic data and the corresponding transmission result data stored in the storage space, a step of determining sample data whose transmission success rate satisfies the convergence condition is executed.

[0106] In the embodiment of the present disclosure, considering that the longer the collection interval of historical channel characteristic data is, the lower the reliability is, the historical channel characteristic data with a longer collection interval and the corresponding transmission result data in the storage space can be discarded.

[0107] In specific implementation, the collection intervals between adjacent historical channel characteristic data can be counted in real time, and the adjacent historical channel characteristic data whose collection interval exceeds the collection threshold can be determined to clear the adjacent historical channel characteristic data and the corresponding transmission result data, wherein the collection threshold can be the above-mentioned TimeThr.

[0108] In the disclosed embodiment, historical channel characteristic data with a long collection interval and corresponding transmission result data in the storage space can be discarded, thereby improving the reliability of the sample data in the storage space and ensuring the training effect of the offline training network through the training data set.

[0109] In an optional implementation, the above step S1011, training the to-be-trained network based on the training data set to obtain an offline training network, specifically includes the following process:

[0110] S21: Determine the number of samples in the training data set, and when the number of samples meets the dimensionality reduction condition, determine the data to be reduced in dimensionality in the sample data.

[0111] S22: Performing dimensionality reduction processing on the data to be reduced in dimension to obtain reduced-dimensionality data, and training the network to be trained based on the reduced-dimensionality data to obtain an offline training network.

[0112] In the embodiment of the present disclosure, the above-mentioned dimensionality reduction condition can be used to indicate the sample quantity interval [SampleNumThr1, SampleNumThr2], where SampleNumThr1 and SampleNumThr2 are both sample quantity threshold values, and SampleNumThr1≤SampleNumThr2. When the sample quantity is in the sample quantity interval indicated by the dimensionality reduction condition, it can be considered that the sample quantity meets the dimensionality reduction condition. When the sample quantity is not in the sample quantity interval indicated by the dimensionality reduction condition, it can be considered that the sample quantity meets the non-dimensionality reduction condition, and the dimensionality reduction data to be treated is not subjected to dimensionality reduction processing.

[0113] After determining that the number of samples meets the dimensionality reduction conditions, the historical channel characteristic data in the sample data, that is, the above-mentioned data to be reduced in dimension, can be obtained, and the data to be reduced in dimension can be processed by dimensionality reduction. Specifically, the historical channel characteristic data can be first classified, and different types of historical channel characteristic data can be processed by different dimensionality reduction methods to obtain reduced dimensionality data. The specific method for determining the reduced dimensionality data is described as follows and will not be described in detail here.

[0114] In the disclosed embodiment, it can be decided whether to perform dimensionality reduction processing on sample data based on dimensionality reduction conditions. Specifically, when the number of samples in the training data set is small, the sample data can be processed with dimensionality reduction, thereby avoiding overfitting of the offline training network, resulting in a decrease in the generalization ability of the learned model on other samples that have not appeared, and thus causing a decrease in overall system performance. In addition, when the number of samples in the training data set is large, the sample data may not be processed with dimensionality reduction, thereby avoiding information loss caused by dimensionality reduction processing, so that the offline training network can fully learn the accurate correlation between a large amount of historical channel feature data, and between historical channel feature data and transmission performance, wherein the accurate correlation helps to improve the accuracy of the offline training network, thereby improving the overall system performance.

[0115] In an optional implementation, the above step S22, performing dimensionality reduction processing on the data to be dimensionality reduced to obtain dimensionality reduced data, specifically includes the following process:

[0116] When the data to be reduced in dimensionality includes multidimensional data for characterizing similar channel characteristics, the multidimensional data is accumulated based on the dimensional direction of the multidimensional data to obtain aggregated data; and when the data to be reduced in dimensionality includes data to be classified, the data to be classified is clustered to obtain classified data.

[0117] In the embodiment of the present disclosure, the multidimensional data may be the channel angle domain projection, wherein the channel angle domain projection is divided into an angle domain vertical direction and an angle domain horizontal direction. Therefore, the channel angle domain projection may be first accumulated along the angle domain vertical direction.

[0118] For example, the above channel angle domain projection includes 8 horizontal * 4 vertical projection energy values. After accumulation along the vertical direction of the angle domain, 8 horizontal projection energy values ​​are obtained. Then the square sum of the 8 projection energy values ​​in the horizontal direction of the angle domain is calculated to obtain the energy aggregation measurement value in the horizontal direction of the angle domain.

[0119] Next, the angle domain vertical energy aggregation measurement value corresponding to the channel angle domain projection can be calculated in accordance with the above-mentioned method of determining the angle domain horizontal energy aggregation measurement value. The specific method of determining the angle domain vertical energy aggregation measurement value will not be repeated here. Then, the aggregation data corresponding to the above-mentioned channel angle domain projection can be determined based on the angle domain horizontal energy aggregation measurement value and the angle domain vertical energy aggregation measurement value.

[0120] In the embodiment of the present disclosure, the data to be classified may be the chip type. Specifically, the processing capability corresponding to the chip type may be queried, and the chip type may be clustered based on the chip processing capability to obtain classified data. In addition, the classification label of the classified data may be determined, and the chip index of the chip type in the classified data may be determined as the classification label to achieve data dimensionality reduction.

[0121] For example, it is determined through query that the chip types with similar processing capabilities are A1, A2, and A3. Then, the chip types A1, A2, and A3 can be clustered to obtain classified data, wherein the classification label corresponding to the classified data can be A.

[0122] In addition, the data to be reduced in dimension may also include data to be quantized, such as a channel signal to interference noise ratio. Specifically, the quantization process may be performed based on a value range of the channel signal to interference noise ratio. For example, if the value range of the channel signal to interference noise ratio is [B1, B40], then the channel signal to interference noise ratio corresponding to B1-B10 may be determined as number 1, ..., the channel signal to interference noise ratio corresponding to B31-B40 may be determined as number 4, thereby achieving data dimensionality reduction of the channel signal to interference noise ratio.

[0123] In the disclosed embodiment, when the number of training samples in the training data set is small, the sample data can be processed for dimensionality reduction, thereby avoiding overfitting of the offline training network, resulting in a decrease in the generalization ability of the learned model on other samples that have not appeared, and further resulting in a decrease in the overall system performance.

[0124] In an optional implementation, the online inference network includes a plurality of regression networks, wherein each regression network corresponds to at least one preset transmission mode, and the step S103, inputting the channel characteristic data into the online inference network to obtain the first transmission mode corresponding to the object to be configured, specifically includes the following process:

[0125] (1) respectively transmitting the channel characteristic data to each of the regression networks to obtain a performance estimation value output by each of the regression networks, wherein the performance estimation value is used to indicate the transmission performance when data is transmitted to the object to be configured through the preset transmission mode;

[0126] (2) Determine the preset transmission mode corresponding to the regression network with the best estimation value as the first transmission mode.

[0127] In the embodiment of the present disclosure, the above-mentioned regression network can be an MLP (Multilayer Perceptron) regression network. When the above-mentioned preset transmission mode includes BC weight, PMI weight and BF weight (the BF weight includes: BF&LayerNum=1, BF&LayerNum=2, BF&LayerNum=3, BF&LayerNum=4), 6 MLP regression networks can be set to correspond to the above-mentioned 6 preset transmission modes respectively.

[0128] Next, corresponding offline training networks can be set for the above-mentioned multiple regression networks respectively, and the training data set corresponding to each offline training network can be determined. The specific method of determining the above-mentioned training data set is as described above and will not be repeated here. When training the offline training network, the network parameters can use the last training results. If the offline training network has not been trained, the initial value of the network parameter can be set to a random value. After the offline training network is trained, the corresponding online inference network can be updated based on the network parameters of the offline training network.

[0129] After collecting the channel characteristic data of the object to be configured, the channel characteristic data can be input into each regression network in the online inference network respectively to obtain an estimated value of the transmission performance of the channel characteristic data under the preset transmission mode corresponding to each regression network, and the preset transmission mode corresponding to the regression network with the best estimated value can be determined as the first transmission mode.

[0130] It should be understood that before the channel characteristic data is input into the regression network, the CQISINR value in the channel characteristic data can be adjusted based on the number of transmission streams corresponding to the preset transmission mode corresponding to the regression network so that the CQISINR value is adapted to the transmission mode.

[0131] In the disclosed embodiment, the online inference network includes multiple regression networks, wherein each regression network corresponds to a preset transmission mode. After the input channel characteristic data is calculated by multiple regression networks, the estimated performance under multiple preset transmission modes can be obtained, thereby achieving reasonable performance estimation under different transmission models. By comparing the estimated performance under different transmission models, the preset transmission model with the best estimated performance is determined as the first transmission mode, thereby achieving the optimal transmission mode selection.

[0132] In an optional implementation, the online inference network includes a classification network, and the step S103, inputting the channel characteristic data into the online inference network to obtain the first transmission mode corresponding to the object to be configured, further includes the following process:

[0133] (1) modifying the channel characteristic data based on a transmission stream number parameter corresponding to the classified network;

[0134] (2) Inputting the modified channel characteristic data into the classification network to obtain a first transmission mode corresponding to the object to be configured.

[0135] In the disclosed embodiment, the classification network may be a multi-classification network, which may calculate the first transmission mode with the best estimated value of transmission performance based on the input channel characteristic data. For example, the multi-classification network may be an MLP classification network, an LGBM (Light Gradient Boosting Machine) classification network, etc.

[0136] It should be understood that before the channel characteristic data is input into the classification network, the channel characteristic data may be preprocessed first. Specifically, the CQISINR value in the channel characteristic data may be adaptively adjusted based on the above transmission stream number parameter. Here, the channel stream number parameter may be set to 1.

[0137] In the disclosed embodiment, when the multi-classification network is trained, the training can be performed by the trained multiple regression networks. Specifically, the sample data can be first input into each regression network to obtain the transmission performance when each preset transmission mode is used in a specific channel state. It should be understood that since the CQISINR values ​​corresponding to different preset transmission modes are different, before the sample data is input into the regression network, the CQISINR value therein can be adjusted to the CQISINR value under the number of transmission streams of the preset transmission mode corresponding to the regression network.

[0138] For example, if the CQISINR value in the above sample data is the CQISINR value when the number of transmission streams is 1, and the number of transmission streams corresponding to the regression network is 2, then the CQISINR value in the sample data can be modified to the CQISINR value when the number of transmission streams is 2.

[0139] Next, based on the prediction results output by the above multiple regression networks, a preset transmission mode with the best transmission performance can be selected as a training label for the current sample data to construct a training data set.

[0140] After determining the training data set, the CQISINR value in the sample data can be adaptively adjusted. As described above, the CQISINR value can be set to the CQISNR value when the number of transmission streams is 1. Then, the above multi-classification network can be trained by the adjusted training data set. For example, the multi-classification network is trained by a supervised learning method. The specific training method is based on what can be implemented and is not limited in the present disclosure.

[0141] In an embodiment of the present disclosure, the online inference network may include a multi-classification network, so that the first transmission mode corresponding to the input channel characteristic data can be directly calculated based on the multi-classification network, thereby reducing the calculation complexity.

[0142] In an optional implementation, before transmitting data to the object to be configured according to the first transmission mode, the above Figure 1 The corresponding embodiment also includes the following process:

[0143] (1) acquiring historical configuration information, and storing the first transmission mode in the historical configuration information, wherein the historical configuration information includes the historical transmission mode configured for the object to be configured;

[0144] (2) Determine in the historical configuration information a second transmission mode in which the configuration frequency satisfies the frequency condition, so as to perform data transmission to the object to be configured according to the second transmission mode.

[0145] In the embodiment of the present disclosure, the above historical configuration information may be a filter window of length FilterWindowLen=5, which may include the transmission mode configured most times for the object to be configured. In addition, the frequency condition may be used to indicate the transmission mode configured most times.

[0146] After the first transmission mode is input into the filter window, the transmission mode with the most configuration times can be re-counted based on the filter window, and the transmission mode can be determined as the second transmission mode. It should be understood that in the case where there are multiple transmission modes determined based on the filter window, a transmission mode can be randomly selected from the multiple transmission modes as the second transmission mode.

[0147] In the embodiment of the present disclosure, the first transmission mode can be filtered based on the filter window to obtain the second transmission mode, so as to make the transmission mode selection result smoother and prevent strong jitter. For example, if the first transmission mode is the above-mentioned BC weight, and the transmission mode with the most configuration times corresponding to the object to be configured is the above-mentioned BF weight, it will cause strong jitter in the final configuration result.

[0148] Those skilled in the art will appreciate that, in the above method of specific implementation, the order in which the steps are written does not imply a strict execution order and does not constitute any limitation on the implementation process. The specific execution order of the steps should be determined by their functions and possible internal logic.

[0149] Based on the same inventive concept, a transmission mode determination device corresponding to the transmission mode determination method is also provided in the embodiment of the present disclosure. Since the principle of solving the problem by the device in the embodiment of the present disclosure is similar to the above-mentioned transmission mode determination method in the embodiment of the present disclosure, the implementation of the device can refer to the implementation of the method, and the repeated parts will not be repeated.

[0150] See also Figure 3 As shown, it is an architecture diagram of the transmission mode determination system provided by the embodiment of the present disclosure, and the transmission mode determination system includes a real-time layer and a non-real-time layer, wherein the real-time layer is used to determine the first transmission mode or the second transmission mode corresponding to the object to be configured, and the non-real-time layer is used to train the offline training network to obtain network parameters. Specifically, the non-real-time layer includes a data storage module, a data screening module, and an offline training module; the real-time layer includes a data acquisition module, a channel measurement module, an online reasoning module, and a physical layer transmission module.

[0151] The data collection module is used to collect the above historical channel characteristic data and transmission result data.

[0152] The data storage module is used to store the above historical channel characteristic data and transmission result data.

[0153] A data screening module is used to determine the transmission success rate of the above-mentioned historical channel characteristic data based on the transmission result data, and to determine the sample data whose transmission success rate meets the convergence condition in the historical channel characteristic data, and to determine the historical transmission performance corresponding to the sample data as a training label, and to determine the training data set based on the sample data and the corresponding training label.

[0154] In the embodiment of the present disclosure, the specific process of data collection, data storage and data screening is as described in the embodiment corresponding to the above step S1011, and will not be repeated here.

[0155] The offline training module is used to obtain the above training data set to train the above network to be trained, obtain the offline training network, and obtain the network parameters of the offline training network.

[0156] In the embodiment of the present disclosure, the process of determining the offline training network is as described in the embodiment corresponding to the above step S1011, and will not be repeated here.

[0157] The channel measurement module is used to measure the channel characteristic data of the above-mentioned object to be configured.

[0158] The online inference module is used to update the online inference network according to the network parameters transmitted by the offline training module, and transmit the channel characteristic data to the online inference network to obtain the first transmission mode.

[0159] In the embodiment of the present disclosure, the manner of determining the first transmission mode is as described in the embodiment corresponding to the above step S103, and will not be repeated here.

[0160] The physical layer transmission module is used to transmit data to the object to be configured according to the first transmission mode.

[0161] In summary, in the embodiments of the present disclosure, first, the channel characteristic data of the object to be configured can be obtained, and the online inference network corresponding to the object to be configured can be determined, wherein the channel characteristic data is used to indicate the characteristics of the communication channel between the object to be configured and the base station. Next, the channel characteristic data can be input into the online inference network to obtain the first transmission mode corresponding to the object to be configured, and the first transmission mode is used to indicate the beamforming mode and the number of transmission streams corresponding to the object to be configured. Then, according to the first transmission mode, data transmission can be performed on the object to be configured, thereby eliminating the need to determine the first transmission mode by theoretical calculation, thus avoiding inaccurate calculation results caused by measurement errors, improving the accuracy of transmission mode determination, and flexibly adapting to different scenarios. At the same time, there is no need to determine the first transmission mode based on a transmission mode selection algorithm adjusted by trial and error, thus saving trial overhead and improving the scope of application of the present disclosure.

[0162] Based on the same inventive concept, a transmission mode determination device corresponding to the transmission mode determination method is also provided in the embodiment of the present disclosure. Since the principle of solving the problem by the device in the embodiment of the present disclosure is similar to the above-mentioned transmission mode determination method in the embodiment of the present disclosure, the implementation of the device can refer to the implementation of the method, and the repeated parts will not be repeated.

[0163] Reference Figure 4 FIG. 4 is a schematic diagram of a transmission mode determination device provided by an embodiment of the present disclosure, wherein the device comprises: an acquisition unit 41, an online reasoning unit 42, and a data transmission unit 43; wherein:

[0164] An acquisition unit 41 is used to acquire channel characteristic data of an object to be configured, and determine an online inference network corresponding to the object to be configured, wherein the channel characteristic data is used to indicate characteristics of a communication channel between the object to be configured and a base station;

[0165] An online reasoning unit 42, used for inputting the channel characteristic data into an online reasoning network to obtain a first transmission mode corresponding to the object to be configured, wherein the first transmission mode is used for indicating a beamforming angle and a number of transmission streams corresponding to the object to be configured;

[0166] The data transmission unit 43 is used to transmit data to the object to be configured according to the first transmission mode.

[0167] In an embodiment of the present disclosure, first, the channel characteristic data of the object to be configured can be obtained, and the online inference network corresponding to the object to be configured can be determined, wherein the channel characteristic data is used to indicate the characteristics of the communication channel between the object to be configured and the base station. Next, the channel characteristic data can be input into the online inference network to obtain the first transmission mode corresponding to the object to be configured, and the first transmission mode is used to indicate the beamforming mode and the number of transmission streams corresponding to the object to be configured. Then, according to the first transmission mode, data transmission can be performed on the object to be configured, thereby eliminating the need to determine the first transmission mode by theoretical calculation, thus avoiding inaccurate calculation results caused by measurement errors, improving the accuracy of transmission mode determination, and flexibly adapting to different scenarios. At the same time, there is no need to determine the first transmission mode based on a transmission mode selection algorithm adjusted by trial and error, thus saving trial overhead and improving the scope of application of the present disclosure.

[0168] In a possible implementation manner, the acquisition unit 41 is further configured to:

[0169] Constructing a training data set, and training the training network based on the training data set to obtain an offline training network;

[0170] According to a preset parameter transmission cycle, the network parameters of the offline training network are periodically obtained, and the network parameters are transmitted to the online inference network.

[0171] In a possible implementation manner, the acquisition unit 41 is further configured to:

[0172] Collecting historical channel characteristic data and transmission result data corresponding to the base station;

[0173] Determining a transmission success rate of the historical channel characteristic data based on the transmission result data;

[0174] Determine valid sample data in which the transmission success rate is within a convergence interval in the historical channel characteristic data, and determine the historical transmission performance corresponding to the valid sample data as a training label;

[0175] A training data set is determined based on the sample data and the corresponding training labels.

[0176] In a possible implementation manner, the acquisition unit 41 is further configured to:

[0177] In the process of collecting the historical channel characteristic data, real-time statistics are collected between adjacent historical channel characteristic data;

[0178] Clear the historical channel characteristic data and the corresponding transmission result data whose collection interval exceeds the collection threshold in the storage space;

[0179] Based on the historical channel characteristic data and the corresponding transmission result data stored in the storage space, a step of determining sample data whose transmission success rate satisfies the convergence condition is performed.

[0180] In a possible implementation manner, the acquisition unit 41 is further configured to:

[0181] Determine the number of samples in the training data set, and when the number of samples meets the dimensionality reduction condition, determine the data to be reduced in dimensionality in the sample data;

[0182] The data to be reduced in dimension is subjected to dimensionality reduction processing to obtain reduced-dimensionality data, and a network to be trained is trained based on the reduced-dimensionality data to obtain an offline training network.

[0183] In a possible implementation manner, the acquisition unit 41 is further configured to:

[0184] When the data to be reduced in dimensionality includes multidimensional data for characterizing similar channel characteristics, based on the dimensional direction of the multidimensional data, performing accumulation processing on the multidimensional data to obtain aggregated data; and

[0185] When the data to be reduced in dimensionality include data to be classified, clustering processing is performed on the data to be classified to obtain classified data.

[0186] In a possible implementation manner, the online reasoning network includes multiple regression networks, wherein each regression network corresponds to at least one preset transmission mode, and the online reasoning unit 42 is further used to:

[0187] The channel characteristic data is respectively transmitted to each of the regression networks to obtain a performance estimation value output by each of the regression networks, wherein the performance estimation value is used to indicate the transmission performance when data is transmitted to the object to be configured through the preset transmission mode;

[0188] The preset transmission mode corresponding to the regression network with the best estimation value is determined as the first transmission mode.

[0189] In a possible implementation manner, the online reasoning network includes a classification network, and the online reasoning unit 42 is further used for:

[0190] Modifying the channel characteristic data based on a transmission stream number parameter corresponding to the classified network;

[0191] The modified channel characteristic data is input into the classification network to obtain a first transmission mode corresponding to the object to be configured.

[0192] In a possible implementation manner, the device is also used for:

[0193] Before performing data transmission for the object to be configured according to the first transmission mode, acquiring historical configuration information, and storing the first transmission mode in the historical configuration information, wherein the historical configuration information includes the historical transmission mode configured for the object to be configured;

[0194] A second transmission mode in which the configuration frequency meets the frequency condition is determined in the historical configuration information, so as to perform data transmission on the object to be configured according to the second transmission mode.

[0195] For descriptions of the processing flow of each unit in the device and the interaction flow between each unit, reference may be made to the relevant descriptions in the above method embodiments, which will not be described in detail here.

[0196] Corresponds to Figure 1 The transmission mode determination method in the present disclosure also provides a computer device 500, such as Figure 5 FIG. 5 is a schematic diagram of the structure of a computer device 500 provided in an embodiment of the present disclosure, including:

[0197] Processor 51, memory 52, and bus 53; memory 52 is used to store execution instructions, including memory 521 and external memory 522; memory 521 here is also called internal memory, which is used to temporarily store operation data in processor 51 and data exchanged with external memory 522 such as hard disk. Processor 51 exchanges data with external memory 522 through memory 521. When the computer device 500 is running, the processor 51 communicates with the memory 52 through bus 53, so that the processor 51 executes the following instructions:

[0198] Acquire channel characteristic data of the object to be configured, and determine an online inference network corresponding to the object to be configured, wherein the channel characteristic data is used to indicate characteristics of a communication channel between the object to be configured and a base station;

[0199] Inputting the channel characteristic data into an online inference network to obtain a first transmission mode corresponding to the object to be configured, where the first transmission mode is used to indicate a beamforming mode and a number of transmission streams corresponding to the object to be configured;

[0200] According to the first transmission mode, data is transmitted to the object to be configured.

[0201] The embodiment of the present disclosure also provides a computer-readable storage medium, on which a computer program is stored, and when the computer program is executed by a processor, the steps of the transmission mode determination method described in the above method embodiment are executed. The storage medium can be a volatile or non-volatile computer-readable storage medium.

[0202] The embodiments of the present disclosure also provide a computer program product, which carries a program code. The instructions included in the program code can be used to execute the steps of the transmission mode determination method described in the above method embodiment. For details, please refer to the above method embodiment, which will not be repeated here.

[0203] The computer program product may be implemented in hardware, software or a combination thereof. In one optional embodiment, the computer program product is implemented as a computer storage medium. In another optional embodiment, the computer program product is implemented as a software product, such as a software development kit (SDK).

[0204] Those skilled in the art can clearly understand that, for the convenience and simplicity of description, the specific working process of the system and device described above can refer to the corresponding process in the aforementioned method embodiment, and will not be repeated here. In the several embodiments provided in the present disclosure, it should be understood that the disclosed system, device and method can be implemented in other ways. The device embodiments described above are merely schematic. For example, the division of the units is only a logical function division. There may be other division methods in actual implementation. For example, multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be through some communication interfaces, and the indirect coupling or communication connection of the device or unit can be electrical, mechanical or other forms.

[0205] The units described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed on multiple network units. Some or all of the units may be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0206] In addition, each functional unit in each embodiment of the present disclosure may be integrated into one processing unit, or each unit may exist physically separately, or two or more units may be integrated into one unit.

[0207] If the functions are implemented in the form of software functional units and sold or used as independent products, they can be stored in a non-volatile computer-readable storage medium that is executable by a processor. Based on this understanding, the technical solution of the present disclosure, or the part that contributes to the prior art or the 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 a computer device (which can be a personal computer, a server, or a network device, etc.) to perform all or part of the steps of the method described in each embodiment of the present disclosure. The aforementioned storage medium includes: various media that can store program codes, such as a USB flash drive, a mobile hard disk, a read-only memory (ROM), a random access memory (RAM), a magnetic disk or an optical disk.

[0208] Finally, it should be noted that the above-described embodiments are only specific implementation methods of the present disclosure, which are used to illustrate the technical solutions of the present disclosure, rather than to limit them. The protection scope of the present disclosure is not limited thereto. Although the present disclosure is described in detail with reference to the aforementioned embodiments, ordinary technicians in the field should understand that any technician familiar with the technical field can still modify the technical solutions recorded in the aforementioned embodiments within the technical scope disclosed in the present disclosure, or can easily think of changes, or make equivalent replacements for some of the technical features therein; and these modifications, changes or replacements do not make the essence of the corresponding technical solutions deviate from the spirit and scope of the technical solutions of the embodiments of the present disclosure, and should be included in the protection scope of the present disclosure. Therefore, the protection scope of the present disclosure shall be based on the protection scope of the claims.

Claims

1. A method for determining a transmission mode, characterized in that: include: Acquire channel characteristic data of the object to be configured, and determine an online inference network corresponding to the object to be configured, wherein the channel characteristic data is used to indicate characteristics of a communication channel between the object to be configured and a base station; Inputting the channel characteristic data into an online inference network to obtain a first transmission mode corresponding to the object to be configured, where the first transmission mode is used to indicate a beamforming mode and a number of transmission streams corresponding to the object to be configured; According to the first transmission mode, data is transmitted to the object to be configured.

2. The method according to claim 1, characterized in that The determining of the online reasoning network corresponding to the object to be configured includes: Constructing a training data set, and training the training network based on the training data set to obtain an offline training network; According to a preset parameter transmission cycle, the network parameters of the offline training network are periodically obtained, and the network parameters are transmitted to the online inference network.

3. The method according to claim 2, characterized in that The constructing of the training data set comprises: Collecting historical channel characteristic data and transmission result data corresponding to the base station; Determining a transmission success rate of the historical channel characteristic data based on the transmission result data; Determine valid sample data in which the transmission success rate is within a convergence interval in the historical channel characteristic data, and determine the historical transmission performance corresponding to the valid sample data as a training label; A training data set is determined based on the sample data and the corresponding training labels.

4. The method according to claim 3, characterized in that The collecting of historical channel characteristic data and transmission result data corresponding to the base station includes: In the process of collecting the historical channel characteristic data, real-time statistics are collected between adjacent historical channel characteristic data; Clear the historical channel characteristic data and the corresponding transmission result data whose collection interval exceeds the collection threshold in the storage space; Based on the historical channel characteristic data and the corresponding transmission result data stored in the storage space, a step of determining sample data whose transmission success rate satisfies the convergence condition is performed.

5. The method according to claim 2, characterized in that: The step of training the training network based on the training data set to obtain an offline training network includes: Determine the number of samples in the training data set, and when the number of samples meets the dimensionality reduction condition, determine the data to be reduced in dimensionality in the sample data; The data to be reduced in dimension is subjected to dimensionality reduction processing to obtain reduced-dimensionality data, and a network to be trained is trained based on the reduced-dimensionality data to obtain an offline training network.

6. The method according to claim 5, characterized in that The step of performing dimensionality reduction processing on the data to be dimensionality reduced to obtain dimensionality reduced data comprises: When the data to be reduced in dimensionality includes multidimensional data for characterizing similar channel characteristics, based on the dimensional direction of the multidimensional data, performing accumulation processing on the multidimensional data to obtain aggregated data; and When the data to be reduced in dimensionality include data to be classified, clustering processing is performed on the data to be classified to obtain classified data.

7. The method according to claim 1, characterized in that The online inference network includes a plurality of regression networks, wherein each regression network corresponds to at least one preset transmission mode; The step of inputting the channel characteristic data into an online inference network to obtain a first transmission mode corresponding to the object to be configured includes: The channel characteristic data is respectively transmitted to each of the regression networks to obtain a performance estimation value output by each of the regression networks, wherein the performance estimation value is used to indicate the transmission performance when data is transmitted to the object to be configured through the preset transmission mode; The preset transmission mode corresponding to the regression network with the best estimation value is determined as the first transmission mode.

8. The method according to claim 1, characterized in that The online reasoning network includes a classification network; The step of inputting the channel characteristic data into an online inference network to obtain a first transmission mode corresponding to the object to be configured includes: Modifying the channel characteristic data based on a transmission stream number parameter corresponding to the classified network; The modified channel characteristic data is input into the classification network to obtain a first transmission mode corresponding to the object to be configured.

9. The method according to claim 1, characterized in that: The method further comprises: Before performing data transmission for the object to be configured according to the first transmission mode, acquiring historical configuration information, and storing the first transmission mode in the historical configuration information, wherein the historical configuration information includes the historical transmission mode configured for the object to be configured; A second transmission mode in which the configuration frequency meets the frequency condition is determined in the historical configuration information, so as to perform data transmission on the object to be configured according to the second transmission mode.

10. A transmission mode determination device, characterized in that: include: An acquisition unit, used to acquire channel characteristic data of an object to be configured, and determine an online inference network corresponding to the object to be configured, wherein the channel characteristic data is used to indicate characteristics of a communication channel between the object to be configured and a base station; An online reasoning unit, used for inputting the channel characteristic data into an online reasoning network to obtain a first transmission mode corresponding to the object to be configured, wherein the first transmission mode is used for indicating a beamforming angle and a number of transmission streams corresponding to the object to be configured; A data transmission unit is used to transmit data to the object to be configured according to the first transmission mode.

11. A computer device, characterized in that: include: A processor, a memory and a bus, wherein the memory stores machine-readable instructions executable by the processor, and when the computer device is running, the processor and the memory communicate via the bus, and when the machine-readable instructions are executed by the processor, the steps of the transmission mode determination method as described in any one of claims 1 to 9 are performed.

12. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the steps of the transmission mode determination method according to any one of claims 1 to 9 are executed.

Images