Model training method, communication equipment and storage medium

Abstract

The invention provides a model training method, communication equipment and a storage medium. The model training method applied to a first communication node comprises the following steps: receiving an original data frame sent by a second communication node and used for neural network model training; and carrying out online training on the neural network model by adopting the original data frame.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of communication, in particular to a model training method, a communication device and a storage medium. BACKGROUND

[0002] At present, neural networks have been widely applied to various systems to replace traditional processing modules and have achieved good results. However, due to the black box characteristics of neural networks, their reliability and generalization ability have always been questioned. In the reasoning phase, if the neural network module encounters data that has not been seen in the training phase, its behavior cannot be estimated, and it may get completely illogical results. SUMMARY

[0003] Therefore, the embodiments of the present application provide a model training method, a communication device and a storage medium, which solve the technical problems of poor generalization ability and low reliability of the pre-trained model in the prior art.

[0004] The embodiments of the present application provide a model training method applied to a first communication node, comprising:

[0005] receiving an original data frame for neural network model training sent by a second communication node;

[0006] training the neural network model online by using the original data frame.

[0007] The embodiments of the present application provide a model training method applied to a second communication node, comprising:

[0008] sending an original data frame for neural network model training to a first communication node, so that the first communication node trains the neural network model online by using the original data frame.

[0009] The embodiments of the present application provide a model training device applied to a first communication node, comprising:

[0010] a receiver configured to receive an original data frame for neural network model training sent by a second communication node;

[0011] a training module configured to train the neural network model online by using the original data frame.

[0012] The embodiments of the present application provide a model training method applied to a second communication node, comprising:

[0013] a transmitter configured to send an original data frame for neural network model training to a first communication node, so that the first communication node trains the neural network model online by using the original data frame.

[0014] The embodiment of the present application provides a communication device, comprising a memory and one or more processors.

[0015] The memory is configured to store one or more programs.

[0016] When the one or more programs are executed by the one or more processors, the one or more processors implement the method in any of the above embodiments.

[0017] The embodiment of the present application provides a storage medium, which stores a computer program, and the computer program is executed by a processor to implement the method in any of the above embodiments. BRIEF DESCRIPTION OF DRAWINGS

[0018] Figure 1 is a flowchart of a model training method provided by the embodiment of the present application;

[0019] Figure 2 is an implementation schematic diagram of a pilot-based OFDM system intelligent receiver model provided by the embodiment of the present application;

[0020] Figure 3 is an implementation schematic diagram of a pilot-free OFDM system intelligent receiver model provided by the embodiment of the present application;

[0021] Figure 4 is an implementation schematic diagram of a SIP-based OFDM system intelligent receiver model provided by the embodiment of the present application;

[0022] Figure 5 is an implementation schematic diagram of data cache updating provided by the embodiment of the present application;

[0023] Figure 6 is an implementation schematic diagram of data enhancement by using a random phase method provided by the embodiment of the present application;

[0024] Figure 7 is an implementation schematic diagram of data enhancement by using a constellation symmetry method provided by the embodiment of the present application;

[0025] Figure 8 is another implementation schematic diagram of data enhancement by using a constellation symmetry method provided by the embodiment of the present application;

[0026] Figure 9 is an implementation schematic diagram of data enhancement by using a resource symmetry method provided by the embodiment of the present application;

[0027] Figure 10 is another implementation schematic diagram of data enhancement by using a resource symmetry method provided by the embodiment of the present application;

[0028] Figure 11 is an implementation schematic of a bandwidth segmentation method for data enhancement provided by an embodiment of the present application;

[0029] Figure 12 is an implementation flowchart of a cross-validation training algorithm provided by an embodiment of the present application;

[0030] Figure 13 is an implementation flowchart of another cross-validation training algorithm provided by an embodiment of the present application;

[0031] Figure 14 is a flowchart of another model training method provided by an embodiment of the present application;

[0032] Figure 15 is a structural block diagram of a model training device provided by an embodiment of the present application;

[0033] Figure 16 is a structural block diagram of another model training device provided by an embodiment of the present application;

[0034] Figure 17 is a structural schematic diagram of a communication device provided by an embodiment of the present application. DETAILED DESCRIPTION

[0035] Hereinafter, the embodiments of the present application will be described with reference to the accompanying drawings. The present application is described below in connection with the embodiment drawings, and the examples are only used to explain the present application, and are not used to limit the scope of the present application.

[0036] With the development and maturity of artificial intelligence technology, more and more technical fields have noticed the value of artificial intelligence technology, and have gradually applied it to the field to solve some problems that are difficult to solve by traditional means.

[0037] In the field of wireless communication, artificial intelligence technology has also achieved great success. In the 5G stage, equipment manufacturers and operators have tried to use artificial intelligence technology to improve link energy, optimize network deployment, and have achieved good results. At the beginning of the 6G discussion, many experts and scholars from operators and equipment manufacturers quickly reached a consensus that AI will play an important role in 6G standards, and decided to work together to promote the application of artificial intelligence technology in 6G standards.

[0038] Although everyone from the academic community to the industry community agrees that artificial intelligence technology is a disruptive technology that will have a revolutionary impact on future communication systems, they have also expressed concerns about the reliability and generalization ability of the technology.

[0039] Traditional communication system design is based on solid theoretical foundation in mathematics, electromagnetism, etc. The processing of each module in the system has a logical explanation. As long as the basic theory is not broken, it will always give a reasonable solution. Artificial intelligence technology, more specifically, neural network technology has a well-known black box feature. It maps input data to output data, and the internal mapping process cannot be explained by basic theory. This unexplainability means that when the input of the neural network is a data that has never appeared in the training data set, the output of the neural network is completely unpredictable, and the performance may drop sharply.

[0040] One way to solve this problem is to use a large enough data set to train the model that can cover all possible use cases, so that the model can experience all possible inputs in the training stage. This method has two obvious shortcomings. First, in a wireless communication system, the biggest uncertainty comes from the channel, and the complexity of the channel environment is large. In addition to the well-known multipath fading characteristics, there are various types of completely unpredictable interference, radio frequency nonlinearity and other influencing factors. It is almost impossible to collect training data that can cover all cases. Second, based on common sense, a model that can adapt to various application scenarios must be a relatively large model. The number of parameters of the model and the amount of calculation of the inference process will exceed the model that adapts to one or several scenarios, which will bring great challenges to the cost control of the product. Considering the above factors, this solution is not feasible.

[0041] Another way to solve this problem is to introduce an online training / optimization mechanism, use a model with good generalization ability (called a pre-trained model), and continuously train the model during the interaction between the two communication parties, so that the model gradually matches the current channel environment. This method is based on a simple fact that although the channel environment is complex and variable, its statistical characteristics, such as time domain, frequency domain correlation, and signal-to-noise ratio (SNR) of the channel, are relatively stable within a certain period of time. Therefore, as long as the model converges fast enough in the online training process, the model can adapt to the change of the channel, and its generalization ability and reliability are guaranteed.

[0042] Current research on online training / optimization focuses on the Internet field and is mainly used to predict user click behavior, so as to make content recommendation and advertisement placement more accurate. There is currently less research on online training in the field of wireless communication. The present application provides an online training / optimization scheme in a wireless communication system, which can effectively solve the problems of poor generalization ability and low reliability of the pre-trained model.

[0043] In an embodiment, Figure 1is a flowchart of a model training method provided by an embodiment of the present application. The embodiment is applied to the case of online training in a wireless communication system. The embodiment can be executed by a first communication node. In an example, the first communication node can be a terminal side, and correspondingly, the second communication node is a base station side; in an example, the first communication node can be a base station side, and correspondingly, the second communication node is a terminal side. As shown in Figure 1 the embodiment includes S110-S120.

[0044] S110, receiving an original data frame for neural network model training sent by a second communication node.

[0045] In an example, the first communication node is a terminal side, and the second communication node is a base station side. The second communication node can send the original data frame for neural network model training to multiple first communication nodes in its own coverage range in a broadcast manner. Correspondingly, all the first communication nodes in the coverage range of the second communication node receive the original data frame sent by the second communication node through their own antennas. In an example, the first communication node is a terminal side, and the second communication node is a base station side. The second communication node can send the original data frame for neural network model training to one first communication node in a unicast manner. In an example, in the process of transmitting the original data frame in a broadcast manner, the second communication node can actively issue the original data frame for neural network model training to the first communication node. In an example, in the process of transmitting the original data frame in a unicast manner, the first communication node sends a data frame request to the second communication node. After receiving the data frame request, the second communication node issues the original data frame for neural network model training to the first communication node corresponding to the data frame request.

[0046] S120, online training of the neural network model by using the original data frame.

[0047] The second communication node online trains the neural network model by using the original data frame, so that the trained neural network model can adapt to the current wireless communication environment, and the technical problems of low reliability and insufficient generalization ability of the neural network model are solved.

[0048] In an embodiment, the receiving frequency of the original data frame comprises: periodic; aperiodic. In an example, the original data frame can be received periodically for the first communication node; the original data frame can be transmitted periodically for the second communication node. In an example, the transmission period and the receiving period of the original data frame can be characterized by the number of training frames. For example, the transmission period and the receiving period of the original data frame can be 5 training frames. In an example, the first communication node can periodically transmit the original data frame for neural network model training to the second communication node, or aperiodically transmit the original data frame for neural network model training to the second communication node.

[0049] In an embodiment, the application range of the original data frame comprises: specific terminal; specific cell. In an example, the original data frame can be UE-specific, or cell-specific.

[0050] In an embodiment, the transmission mode of the original data frame comprises: the second communication node actively transmitting to the first communication node; the first communication node requesting the second communication node to transmit. In an example, the second communication node can actively transmit the original data frame to the first communication node. In an example, the first communication node can transmit the original data frame to the second communication node after the second communication node transmits a request to the first communication node. In an example, in the case that the second communication node actively transmits the original data frame to the first communication node, the original data frame can be cell-specific, i.e. all terminals in the cell can receive the original data frame. In an example, in the case that the second communication node transmits the original data frame to the first communication node after the first communication node transmits a request to the second communication node, the original data frame can be UE-specific, i.e. only the first communication node that transmits the request can receive the original data frame.

[0051] In an embodiment, the bandwidth of the original data frame is greater than or equal to the processing bandwidth of the neural network model. In an example, the processing bandwidth of the neural network model refers to the frequency domain bandwidth required for the data that the neural network model can process at a time; the bandwidth of the original data frame refers to the frequency domain bandwidth required for the original data frame. In an example, the frequency domain bandwidth can be characterized by the number of subcarriers.

[0052] In an embodiment, the neural network model comprises one of: a pilot-based orthogonal frequency division multiplexing system intelligent receiver model; a pilot-free orthogonal frequency division multiplexing system intelligent receiver model; a session initiation protocol-based orthogonal frequency division multiplexing system intelligent receiver model.

[0053] In an example, the neural network model can be a pilot-based Orthogonal Frequency Division Multiplexing (OFDM) system intelligent receiver model, Figure 2 is an implementation schematic diagram of a pilot-based OFDM system intelligent receiver model provided by an embodiment of the present application. As shown in the figure, Figure 2 In the pilot-based OFDM system intelligent receiver model, the OFDM demodulation signal and the reference signal are taken as the input of the intelligent receiver model, and the LLR information is taken as the output of the intelligent receiver model. The label during model training is the bit stream, and the loss function is the BCE. The function is equivalent to the combination of the channel estimation, equalization and demodulation module functions in the traditional link.

[0054] In an example, the neural network model can be a pilot-free OFDM system intelligent receiver model, Figure 3 is an implementation schematic diagram of a pilot-free OFDM system intelligent receiver model provided by an embodiment of the present application. As shown in the figure, Figure 3 The pilot-free OFDM system uses a special constellation diagram (the constellation diagram with modulation order 4 is taken as an example in the figure) to modulate data. Specifically, the constellation has a unique rotation, that is, unless the rotation angle is an integer multiple of 2π, the rotated constellation diagram and the original constellation diagram will not coincide. In the pilot-free OFDM system intelligent receiver model, the OFDM demodulation signal is taken as the input of the intelligent receiver model, and the LLR information is taken as the output of the intelligent receiver model. The label during model training is the bit stream, and the loss function is the BCE.

[0055] In an example, the neural network model can be a Session Initiation Protocol (SIP) based OFDM system intelligent receiver model, Figure 4 is an implementation schematic diagram of a SIP-based OFDM system intelligent receiver model provided by an embodiment of the present application. As shown in the figure, Figure 4 The SIP-based OFDM system can send data on all resources, and also send reference signals on all resources, without distinguishing independent data resources and reference signal resources. The data resource network and the reference signal resource grid are multiplied by different power weighting factors and then superimposed together, but the sum of the average powers of the data and the reference signal needs to be equal to 1. In the figure, A is the power weighting factor of the reference signal, which is a two-dimensional matrix and can be simply referred to as the weighting factor matrix. The dimension of the weighting factor matrix A is the same as the current resource network (the number of subcarriers x the number of symbols), and each element in A is a real number greater than 0 and less than 1, representing the power of the reference signal on the RE resource.

[0056] In the SIP-based OFDM system intelligent receiver model, the OFDM demodulation signal and the reference signal resource grid are taken as the input of the intelligent receiver model, the LLR information is taken as the output of the intelligent receiver model, the label during model training is the bit stream, and the loss function is the BCE. In an example, in the SIP-based OFDM system intelligent receiver model, a two-dimensional matrix composed of power weighting factors of the reference signal can be taken as the input of the intelligent receiver model.

[0057] In an example, the intelligent receiver model can be referred to as an intelligent receiver network, and can also be referred to as an intelligent receiver for short.

[0058] In an embodiment, the neural network model is trained online by using the original data frames, including:

[0059] The original data frames are subjected to data cache updating and data enhancement to obtain target data frames;

[0060] The target data frames are subjected to data grouping according to a data grouping mode to obtain two groups of target data;

[0061] The neural network model is trained online based on a cross-validation training algorithm and the two groups of target data. In an example, the online training of the neural network model by using the original data frames includes the following four steps: data cache updating, data enhancement, data grouping, and training of the neural network model by using the cross-validation training algorithm.

[0062] In an embodiment, the implementation process of data cache updating includes: pre-configuring a cache space with a target cache space size; storing a newly received original data frame in the cache space, and deleting the original data frame with the earliest time in the cache space from the cache space. In an example, the target cache space size can be represented by the number of training frames. Illustratively, the target cache space size can be 3 training frames, that is, at most 3 training frames can be accommodated in the cache space with the target cache space size. In an example, the management of the cache space can be based on the FIFO strategy, that is, each time a new original training frame is received by the first communication node, the new original data frame is added to the cache space, and the original data frame with the earliest time in the cache space is deleted from the cache space.

[0063] Figure 5 FIG. 1 is an implementation schematic diagram of data cache updating provided by an embodiment of the present application. Data cache updating refers to: a cache space with a target cache space size (denoted as save_length) is set, when a new training frame (i.e., a new original data frame) arrives, the training frame data is added to the cache space, and the data with the earliest time in the cache space is deleted. For example, Figure 5As shown, the target buffer size of the buffer space is set to 3 training frames, and the training frames are periodically transmitted, and the transmission period is 5 training frames.

[0064] Different from the conventional offline training mode, the online training has the following characteristics: 1. The training samples are limited; 2. The training samples arrive one after another in time; 3. The device for online training usually has limited storage space and computing resources; 4. Online training has certain real-time requirements compared with offline training.

[0065] In an embodiment of the present application, a data buffer updating mechanism is proposed, in which the save_length parameter is used to control the size of the buffer space. The most ideal case is to use the training frame data to train the neural network model when each training frame arrives, and then discard it after use. In this way, the consumption of storage space and computing resources is minimized, but the problem is that the value of the data is not fully utilized, and the less data used in training, the greater the risk of network overfitting. However, save_length cannot be set too large without limitation. On the one hand, the device for online training is limited by cost, and the storage space and computing resources are limited. On the other hand, in some environments, the channel statistical characteristics may change quickly, and the historical buffered training data may not conform to the current channel characteristics, and may not help or even hinder the training of the neural network model. By controlling the size of save_length, old data that does not conform to the current channel characteristics can be discarded in time to ensure the effectiveness of online training.

[0066] The management of the buffer space is based on the FIFO strategy, that is, each time new training data arrives, the new data is added to the buffer space, and the earliest data in time is deleted from the buffer space.

[0067] In an embodiment, the data enhancement method includes at least one of the following: random phase method; constellation symmetry method; resource symmetry method; bandwidth segmentation method. In an example, the number of samples for online training is much smaller than that for offline training, and the purpose of data enhancement is to derive more data from the original data frame (original training data) for model training. For example, in the CV field, the method of rotating pictures is often used to generate new pictures to enhance the network's ability to recognize pictures. In wireless communication, the key point of data enhancement is that the new data derived has the same statistical characteristics as the original data, that is, the same time-frequency domain correlation, the same SNR, etc.

[0068] In one embodiment, the implementation method of data enhancement using the random phase method includes: performing a multiplication operation of a random phase factor based on the original data frame and the signal transmission and reception relationship on each resource element (RE) to obtain a new signal transmission and reception relationship; and obtaining a corresponding target data frame based on the new signal transmission and reception relationship.

[0069] In one example, the random phase method is a data enhancement method that leverages knowledge from the communications field. Assuming the training frame data bit is b, the modulated transmitted data is x, the channel is h, the noise is n, and the received signal is y, then for an OFDM system, the signal transmission and reception relationship on each RE can be: y = hx + n;

[0070] Multiply both sides of the signal transmission and reception relationship by the random phase factor e jω , we get the new signal receiving and sending relationship: ye jω =he jω ·x+ne jω , the new signal receiving and sending relationship can be equivalent to: new =h new x+n new , where the random phase factor is combined with the original channel h to generate a new channel h new , h new Compared with the original channel h, there is only a phase factor difference, and the statistical characteristics are completely unchanged. new Compared with the original noise n, only the phase changes, and the SNR does not change. The new signal transmission and reception relationship y can be used new =h new x+n new Generate target data frame as new training sample.

[0071] There are two important points to note about the random phase method. First, the labels b of the training samples generated by the random phase method are consistent with the original data, because x in the equation does not change. Second, the number of training samples generated by the random phase method is not necessarily better. The more samples there are, the higher the correlation between them, which does not substantially help model training. In this embodiment of the application, the training samples can be generated by uniformly sampling the 2π phase space N equally, with the value of N determined based on the actual training results. Figure 6 This is a schematic diagram of implementing data enhancement using a random phase method provided in an embodiment of the present application. Figure 6 As shown, when N=8, the effect of random phase method is as follows Figure 6 shown.

[0072] In one example, the constellation symmetry method is a data augmentation method designed by utilizing the symmetry characteristics of the constellation. Specifically, the symmetry of the constellation includes but is not limited to:

[0073] Rotational symmetry, that is, the original constellation diagram completely overlaps with the original constellation diagram in position after being rotated by a certain angle (0<ω<2π), which means that the rotated data is still legal and interpretable data;

[0074] Axial symmetry: when the original constellation diagram is flipped with a straight line as the axis, the flipped constellation diagram and the original constellation diagram completely overlap in position, which means that the flipped data is still legal and interpretable.

[0075] These two methods are described in detail. Figure 7 This is a schematic diagram of an implementation of data enhancement using a constellation symmetry method provided by an embodiment of the present application. Regarding rotational symmetry, taking the 16QAM constellation as an example, the constellation has rotational symmetry of π / 2, π, and 3π / 2, that is, when the constellation is rotated by π / 2, π, and 3π / 2, the positions of the constellation points completely coincide with the original constellation. Figure 7 shown.

[0076] In one embodiment, the implementation of data enhancement using the constellation symmetry method includes: performing a multiplication operation of a rotational symmetry factor based on the original data frame and the signal transmission and reception relationship of each resource unit to obtain a new signal transmission and reception relationship; and obtaining a corresponding target data frame based on the new signal transmission and reception relationship. Assuming that the training frame data bit is b, the modulated transmitted data is x, the channel is h, the noise is n, and the received signal is y, then for the OFDM system, the signal transmission and reception relationship of each RE can be expressed as the following equation: y = hx + n; multiplying the signal transmission and reception relationship by the rotational symmetry factor e jπ / 2 , e jπ and e j3π / 2 , we get the new signal transmission and reception relationship:

[0077]

[0078] As shown above, the rotational symmetry factor can be combined with the original x to generate new data x new Compared with the original signal transmission and reception relationship, h has not changed, x new Still legal data, n new Compared with the original n, only the phase changes, and the SNR does not change. Now use the new signal transmission and reception relationship y new =h·x new +n new The target data frame can be generated as a new training sample.

[0079] Based on the pilot-free OFDM system intelligent receiver model, the pilot-free transmission must use constellation with rotation uniqueness to modulate data, so the constellation does not have rotational symmetry, and the method cannot be used for data enhancement.

[0080] Figure 8 is another implementation schematic diagram of the embodiment provided by the application for data enhancement using the constellation symmetry method. For the axial symmetry, taking the 16QAM constellation as an example, the constellation has symmetry about the x-axis, that is, when the constellation is flipped around the x-axis (that is, the conjugate of the data is taken), the positions of the constellation points are completely coincident with the original constellation, as shown in Figure 8

[0081] In an embodiment, the implementation of the data enhancement using the constellation symmetry method comprises: based on the original data frame and the signal transmission relationship on each resource unit, the conjugate value is taken to obtain a new signal transmission relationship; and based on the new signal transmission relationship, the corresponding target data frame is obtained. Assuming that the training frame data bit is b, the transmitted data after modulation is x, the channel is h, the noise is n, and the received signal is y, then for the OFDM system, the signal transmission relationship on each RE can be expressed as the following equation: t = hx + n; the conjugate value is taken on both sides of the signal transmission relationship to obtain a new signal transmission relationship: y = h x + n; as shown above, the new signal transmission relationship is generated by taking the conjugate, compared with the original signal transmission relationship, the statistical characteristics of h are not changed compared with the original h, x is still a legal data, and the SNR of n is not changed compared with the original n. Now, the equation y = hx + n can be used to generate the target data frame as a new training sample. * * * * As shown above, the new signal transmission relationship is generated by taking the conjugate, compared with the original signal transmission relationship, the statistical characteristics of h are not changed compared with the original h, x is still a legal data, and the SNR of n is not changed compared with the original n. Now, the equation y = hx + n can be used to generate the target data frame as a new training sample. * * * * * * *

[0082] In an embodiment, the implementation of the data enhancement using the resource symmetry method comprises: reversing the order of the original data frame on the orthogonal frequency division multiplexing resource grid to generate the corresponding target data frame. The resource symmetry method is a data enhancement algorithm designed by using the symmetry of the OFDM resource, and compared with the data enhancement using the random phase method and the constellation symmetry method, the resource symmetry method is a method for data enhancement from a more macroscopic perspective. Specifically, the order of the original data frame on the OFDM resource grid is reversed to generate a new data frame as a target data frame.

[0083] ​​​​​​​​​​​In an embodiment, the implementation of reversing order includes: frequency domain mirroring; time domain mirroring; time-frequency domain mirroring. In an example, the frequency domain mirroring is reversing order of subcarriers; the time domain mirroring is reversing order of OFDM symbols; the time-frequency domain mirroring is reversing order in both frequency domain and time domain.

[0084] In an embodiment, for a pilot-based OFDM system, the original data frame is reversed in order on the OFDM resource grid to generate a corresponding target data frame, including one of the following:

[0085] If the pilot pattern has time domain symmetry, the original data frame is reversed in order on the OFDM resource grid using time domain mirroring to generate a corresponding target data frame;

[0086] If the pilot pattern has frequency domain symmetry, the original data frame is reversed in order on the OFDM resource grid using frequency domain mirroring to generate a corresponding target data frame.

[0087] Figure 9 is an implementation schematic diagram of a resource symmetry method for data enhancement provided by an embodiment of the present application. For a pilot-based OFDM system, the principle is as shown in Figure 9 p1, p2, p3 and p4 in the original resource diagram are frequency domain resources, and s1, s2,..., s31 and s32 in the original resource diagram are time domain resources. A new resource grid can be used to generate a new training sample. It should be noted that the resource symmetry method requires that the resource grid has certain symmetry characteristics, more specifically, the pilot pattern has symmetry. In the case that the pilot pattern has time domain symmetry, time domain mirroring can be used to generate a target data frame as a new training sample; in the case that the pilot pattern has frequency domain symmetry, frequency domain mirroring can be used to generate a target data frame as a new training sample.

[0088] In an embodiment, for a pilot-based OFDM system, the original data frame is reversed in order on the OFDM resource grid to generate a corresponding target data frame, including one of the following:

[0089] If the pilot pattern has time domain symmetry, the original data frame is reversed in order on the OFDM resource grid using time domain mirroring to generate a corresponding target data frame;

[0090] If the pilot pattern has frequency domain symmetry, the original data frame is reversed in order on the OFDM resource grid using frequency domain mirroring to generate a corresponding target data frame.

[0091] Figure 10is another implementation schematic diagram of data enhancement using resource symmetry method provided by the embodiment of the present application. For the SIP-based OFDM system, the weighting factor matrix (such as A matrix) is required to have certain symmetry, as shown in Figure 10 In the case that the A matrix has time domain symmetry, the target data frame can be generated by using time domain mirroring as a new training sample; in the case that the A matrix has frequency domain symmetry, the target data frame can be generated by using frequency domain mirroring as a new training sample.

[0092] For the pilotless OFDM system, the resource grid naturally has time domain and frequency domain symmetry, and the mirroring resource grid can be directly generated by using the method to generate the target data frame as a new training sample.

[0093] In an embodiment, the implementation manner of data enhancement using bandwidth segmentation method includes: performing resource sliding segmentation based on the transmission bandwidth of the original data frame and the processing bandwidth of the neural network model to obtain a corresponding resource grid; and generating a corresponding target data frame based on each resource grid. In an example, the bandwidth segmentation method is a data enhancement method for the case that the training frame bandwidth is greater than the AI receiver bandwidth, and is also a macroscopic data enhancement method. Assuming that the training frame bandwidth is M resource blocks (RBs), the AI receiver bandwidth is B (the receiver bandwidth refers to the bandwidth size of data sent into the receiver at a time), and M > B, the resource sliding segmentation is performed on the bandwidth M in units of B, and the target data frame is generated by using the segmented resource grid as a new training sample. Figure 11 is an implementation schematic diagram of data enhancement using bandwidth segmentation method provided by the embodiment of the present application. As shown in Figure 11 The training frame bandwidth is 8 RBs, and the AI receiver bandwidth is 3 RBs, so the resource sliding segmentation is performed on RB0-RB7 to obtain 6 resource grids, which are block0-block5. Figure 11 In the embodiment, M = 8, B = 3, and 6 resource grids conforming to the AI receiver bandwidth are segmented, which can be used to generate training samples.

[0094] In an example, the random phase method, the constellation symmetry method, the resource symmetry method and the bandwidth segmentation method can be independent of each other or combined with each other to generate more training samples.

[0095] In an embodiment, the data grouping manner includes one of the following: randomly grouping into two groups; grouping into two groups according to the receiving order of the original data frames; and grouping by using different data enhancement methods. The data grouping is to divide the enhanced training samples into two groups according to a certain method, and the principle of grouping is that the lower the correlation of the two groups of data, the better. The grouping method includes: randomly grouping into two groups; grouping into two groups according to the order of arrival of the training frames; and grouping into two groups according to whether a certain data enhancement algorithm is used, for example, data obtained by using an axis symmetry enhancement algorithm can be taken as one group, and data obtained without using the axis symmetry enhancement algorithm can be taken as another group.

[0096] In an embodiment, the online training of the neural network model based on the cross-validation training algorithm and the two groups of target data includes: dividing a first group of target data in the two groups of target data into a training group, and dividing a second group of target data into a validation group; verifying the neural network model using the target data in the validation group to obtain a corresponding initial model performance; iteratively training the neural network model using the target data in the training group, and verifying the current model performance of the neural network model updated once by iteration using the target data in the validation group; and performing online training based on a comparison result between the initial model performance and the current model performance.

[0097] In an embodiment, the online training based on the comparison result between the initial model performance and the current model performance includes:

[0098] If the current model performance is greater than the initial model performance, the online training of the neural network model is stopped.

[0099] If the current model performance is less than or equal to the initial model performance, the online training of the neural network model is continued until the current iteration number reaches a preset iteration number threshold.

[0100] In an embodiment, the online training based on the comparison result between the initial model performance and the current model performance includes:

[0101] If the current model performance is greater than the initial model performance, the current learning rate and the penalty factor of the neural network model are multiplied, and the next iteration training is performed until the current learning rate is less than a preset learning rate threshold.

[0102] In an example, the implementation process of the cross-training algorithm includes:

[0103] Step 1, defining two groups of target data as training group A and validation group B respectively, wherein the training group is used to perform training of the model and update the parameters, and the validation group is used to verify the performance of the model;

[0104] Step 2, calculating the initial model performance using the validation group data, denoted as p;

[0105] Step 3, assuming the maximum number of iterations is K, the number of iterations is increased by 1 each time the entire data of the training set is used to train, and the current model performance is verified using the validation set data, denoted as p new ;

[0106] Step 4, determine the size relationship between p and p new , if p new is greater than p, it means that the model may have been over-fitted, and the model training is stopped; if p new is less than p, it means that the performance of the model is normally improved, p = p new , and the next iteration is continued.

[0107] Step 5, for the processing of d, another parallel solution is to determine the size relationship between p and p new , if p new is greater than p, it means that the model may have been over-fitted, but at this time the training is not stopped, only the current learning rate is multiplied by a penalty factor a less than 1, and the next iteration is continued. When the learning rate is less than a predetermined threshold (i.e., a predetermined learning rate threshold), the training is stopped.

[0108] Step 6, if the current number of iterations has reached the predetermined maximum value, the model training is stopped.

[0109] Step 7, optionally, the training set data and the validation set data can be exchanged, i.e., the original validation set data is used as the training set, and the original validation set is used as the training set again to perform training.

[0110] Figure 12 is an implementation flowchart of a cross-validation training algorithm provided by the embodiments of the present application. As shown in the figure, the implementation process of the cross-validation training algorithm includes: Figure 12

[0111] Step 1, initialization: set the maximum number of iterations K, the training iteration number k = 0, and calculate the initial model performance p of the validation set.

[0112] Step 2, use the training set data to perform model training and parameter updating.

[0113] Step 3, calculate the performance p new of the validation set data.

[0114] Step 4, determine whether p is less than p new , if yes, execute step 5; if no, execute step 7.

[0115] Step 5, determine whether k is less than K, if yes, execute step 6, if no, execute step 7.

[0116] Step 6, k = k + 1, p = p​new .

[0117] Step 7, training stops.

[0118] Figure 13 is another implementation flowchart of the cross-validation training algorithm provided by the embodiment of the present application. As shown in Figure 13 , the implementation process of the cross-validation training algorithm includes:

[0119] Step 1, initialization: set the maximum number of iterations K, the number of training iterations k = 0, calculate the initial model performance p of the validation group data, the learning rate μ, the learning rate penalty factor β, and the iteration stop learning rate threshold μ_t.

[0120] Step 2, perform model training and parameter update using the training group data.

[0121] Step 3, calculate the performance p of the validation group data new .

[0122] Step 4, determine whether p is less than p new , if yes, execute step 5; if no, execute step 6.

[0123] Step 5, determine whether k is less than K, if yes, execute step 8, if no, execute step 9.

[0124] Step 6, μ = μ * β.

[0125] Step 7, determine whether μ is greater than μ_t, if yes, execute step 5; if no, execute step 9.

[0126] Step 8, k = k + 1, p = p new .

[0127] Step 9, training stops.

[0128] One of the main features of online training is that the number of training samples is very limited. The number of samples in offline training is usually in the order of hundreds of thousands or millions. During the training process, the accuracy of the model can be focused on, but there is a risk of overfitting. However, the number of samples in offline training is very small. Even if various enhancement algorithms are used to expand the samples, the sample size is still very limited compared to offline training. During training, overfitting is likely to occur.

[0129] In actual offline training, too few iterations during model training can result in underfitting, and too many iterations can result in overfitting. How to set a proper number of iterations is a difficult problem, and the best number of iterations for different scenarios is not necessarily the same. It is best to have an adaptive method to control the number of iterations. In the training of neural networks, an early stopping iteration idea can avoid overfitting. The target data can be divided into a training group and a validation group. If the performance of the validation group is found to be declining, it means that there is a high risk of overfitting, and training needs to be stopped immediately or the current learning rate needs to be reduced, which effectively reduces the risk of overfitting while ensuring the effectiveness of model training, especially in the case of a large number of training group and validation group samples.

[0130] In an embodiment, the model training method applied to the first communication node further includes: dividing the first group of target data into a validation group and the second group of target data into a training group. The data of the training group and the data of the validation group can be exchanged, and the original validation group data is used as the training group, and the original validation group is used as the training group to perform training again. In this way, the value of the original validation group data is fully exploited, and online training can converge faster.

[0131] In an embodiment, Figure 14 is a flowchart of another model training method provided by the embodiments of the present application. The embodiment is applied to the case of online training in a wireless communication system. The embodiment can be executed by the second communication node. In an example, the first communication node can be the terminal side, and correspondingly, the second communication node is the base station side; in an example, the first communication node can be the base station side, and correspondingly, the second communication node is the terminal side. As shown in Figure 14 the embodiment includes S210.

[0132] S210, sending an original data frame for neural network model training to the first communication node, so that the first communication node uses the original data frame to perform online training on the neural network model.

[0133] In an embodiment, the receiving frequency of the original data frame includes: periodic; aperiodic.

[0134] In an embodiment, the application range of the original data frame includes: a specific terminal; a specific cell.

[0135] In an embodiment, the sending mode of the original data frame includes: the second communication node actively sends to the first communication node; the first communication node requests the second communication node to send.

[0136] In an embodiment, the bandwidth of the original data frame is greater than or equal to the processing bandwidth of the neural network model.

[0137] In an embodiment, the neural network model comprises one of: a pilot-based orthogonal frequency division multiplexing system intelligent receiver model; a pilot-free orthogonal frequency division multiplexing system intelligent receiver model; and a session initiation protocol-based orthogonal frequency division multiplexing system intelligent receiver model.

[0138] In an embodiment, the first communication node performs online training of the neural network model using the original data frames, comprising:

[0139] The first communication node performs data cache updating and data enhancement on the original data frames to obtain target data frames;

[0140] The first communication node performs data grouping on the target data frames according to a data grouping manner to obtain two groups of target data.

[0141] The first communication node performs online training of the neural network model based on a cross-validation training algorithm and the two groups of target data.

[0142] In an embodiment, the implementation process of data cache updating comprises:

[0143] A cache space with a target cache space size is pre-configured;

[0144] The received new original data frames are stored in the cache space, and the original data frame with the earliest cache time in the cache space is deleted from the cache space.

[0145] In an embodiment, the data enhancement method comprises at least one of: a random phase method; a constellation symmetry method; a resource symmetry method; and a bandwidth segmentation method.

[0146] In an embodiment, the implementation of the random phase method for data enhancement comprises:

[0147] A random phase factor is multiplied based on the original data frame and the signal transmission relationship on each resource unit to obtain a new signal transmission relationship;

[0148] A corresponding target data frame is obtained based on the new signal transmission relationship.

[0149] In an embodiment, the implementation of the constellation symmetry method for data enhancement comprises:

[0150] A rotation symmetry factor is multiplied based on the original data frame and the signal transmission relationship on each resource unit to obtain a new signal transmission relationship;

[0151] A corresponding target data frame is obtained based on the new signal transmission relationship.

[0152] In an embodiment, the implementation of the constellation symmetry method for data enhancement comprises:

[0153] Conjugate values are obtained based on the original data frame and the signal-transmitting relationship on each resource unit, and a new signal-transmitting relationship is obtained.

[0154] A corresponding target data frame is obtained based on the new signal-transmitting relationship.

[0155] In an embodiment, the implementation mode of data enhancement by using the resource symmetry method includes:

[0156] The original data frame is reversed in order on the orthogonal frequency division multiplexing resource grid to generate a corresponding target data frame.

[0157] In an embodiment, the implementation mode of reversing the order includes: frequency domain mirroring; time domain mirroring; time-frequency domain mirroring.

[0158] In an embodiment, for a pilot-based orthogonal frequency division multiplexing system, the original data frame is reversed in order on the orthogonal frequency division multiplexing resource grid to generate a corresponding target data frame, including one of the following:

[0159] If the pilot pattern has time domain symmetry, the original data frame is reversed in order on the orthogonal frequency division multiplexing resource grid by using time domain mirroring to generate a corresponding target data frame;

[0160] If the pilot pattern has frequency domain symmetry, the original data frame is reversed in order on the orthogonal frequency division multiplexing resource grid by using frequency domain mirroring to generate a corresponding target data frame.

[0161] In an embodiment, for a session initiation protocol-based orthogonal frequency division multiplexing system, the original data frame is reversed in order on the orthogonal frequency division multiplexing resource grid to generate a corresponding target data frame, including one of the following:

[0162] If the weighting factor matrix has time domain symmetry, the original data frame is reversed in order on the orthogonal frequency division multiplexing resource grid by using time domain mirroring to generate a corresponding target data frame;

[0163] If the weighting factor matrix has frequency domain symmetry, the original data frame is reversed in order on the orthogonal frequency division multiplexing resource grid by using frequency domain mirroring to generate a corresponding target data frame.

[0164] In an embodiment, the implementation mode of data enhancement by using the bandwidth segmentation method includes:

[0165] The resource grid is obtained by performing resource sliding segmentation based on the transmission bandwidth of the original data frame and the processing bandwidth of the neural network model;

[0166] A corresponding target data frame is generated based on each resource grid.

[0167] In an embodiment, the data grouping manner comprises one of the following: grouping randomly into two groups; grouping into two groups according to the receiving order of the original data frames; and grouping by using different data enhancement methods.

[0168] In an embodiment, the first communication node performs online training on the neural network model based on the cross-validation training algorithm and the two groups of target data, comprising:

[0169] The first communication node divides a first group of target data in the two groups of target data into a training group, and divides a second group of target data into a validation group;

[0170] The first communication node verifies the neural network model by using the target data in the validation group to obtain a corresponding initial model performance;

[0171] The first communication node iteratively trains the neural network model by using the target data in the training group, and verifies the current model performance of the neural network model updated once by iteration by using the target data in the validation group;

[0172] The first communication node performs online training based on a comparison result between the initial model performance and the current model performance.

[0173] In an embodiment, the first communication node performs online training based on a comparison result between the initial model performance and the current model performance, comprising:

[0174] If the current model performance is greater than the initial model performance, the online training of the neural network model is stopped;

[0175] If the current model performance is less than or equal to the initial model performance, the online training of the neural network model is continued until the current iteration number reaches a preset iteration number threshold.

[0176] In an embodiment, the first communication node performs online training based on a comparison result between the initial model performance and the current model performance, comprising:

[0177] If the current model performance is greater than the initial model performance, the current learning rate and the penalty factor of the neural network model are multiplied, and the next iteration training is performed until the current learning rate is less than a preset learning rate threshold.

[0178] In an embodiment, Figure 15 is a structural block diagram of a model training device provided by an embodiment of the present application. The embodiment is applied to a first communication node. As shown in Figure 15 the model training device in the embodiment comprises a receiver 310 and a training module 320.

[0179] The receiver 310 is configured to receive an original data frame sent by a second communication node for neural network model training;

[0180] The training module 320 is configured to perform online training of the neural network model by using the original data frames.

[0181] In an embodiment, the receiving frequency of the original data frames comprises: periodicity; non-periodicity.

[0182] In an embodiment, the application range of the original data frames comprises: a specific terminal; a specific cell.

[0183] In an embodiment, the sending mode of the original data frames comprises: the second communication node actively sending to the first communication node; the first communication node requesting the second communication node to send.

[0184] In an embodiment, the bandwidth of the original data frames is greater than or equal to the processing bandwidth of the neural network model.

[0185] In an embodiment, the neural network model comprises one of: a pilot-based orthogonal frequency division multiplexing system intelligent receiver model; a pilot-free orthogonal frequency division multiplexing system intelligent receiver model; a session initiation protocol-based orthogonal frequency division multiplexing system intelligent receiver model.

[0186] In an embodiment, the online training of the neural network model by using the original data frames comprises:

[0187] Data cache updating and data enhancement are performed on the original data frames to obtain target data frames;

[0188] The target data frames are grouped according to a data grouping manner to obtain two groups of target data.

[0189] The neural network model is trained online based on a cross-validation training algorithm and the two groups of target data.

[0190] In an embodiment, the implementation process of data cache updating comprises:

[0191] A cache space with a target cache space size is pre-configured;

[0192] The received new original data frames are stored in the cache space, and the original data frames cached in the cache space for the earliest time are deleted from the cache space.

[0193] In an embodiment, the data enhancement method comprises at least one of: a random phase method; a constellation symmetry method; a resource symmetry method; a bandwidth segmentation method.

[0194] In an embodiment, the implementation of the random phase method for data enhancement comprises:

[0195] Based on the original data frame and the signal-transmitting relationship on each resource unit, a multiplication operation of a random phase factor is performed to obtain a new signal-transmitting relationship;

[0196] Based on the new signal-transmitting relationship, a corresponding target data frame is obtained.

[0197] In an embodiment, an implementation manner of data enhancement by using the constellation symmetry method includes:

[0198] Based on the original data frame and the signal-transmitting relationship on each resource unit, a multiplication operation of a rotation symmetry factor is performed to obtain a new signal-transmitting relationship;

[0199] Based on the new signal-transmitting relationship, a corresponding target data frame is obtained.

[0200] In an embodiment, an implementation manner of data enhancement by using the constellation symmetry method includes:

[0201] Based on the original data frame and the signal-transmitting relationship on each resource unit, a conjugate value is obtained to obtain a new signal-transmitting relationship;

[0202] Based on the new signal-transmitting relationship, a corresponding target data frame is obtained.

[0203] In an embodiment, an implementation manner of data enhancement by using the resource symmetry method includes:

[0204] The original data frame is reversed in order on an orthogonal frequency division multiplexing resource grid to generate a corresponding target data frame.

[0205] In an embodiment, the implementation manner of reversing the order includes: frequency domain mirroring; time domain mirroring; time-frequency domain mirroring.

[0206] In an embodiment, for a pilot-based orthogonal frequency division multiplexing system, the original data frame is reversed in order on an orthogonal frequency division multiplexing resource grid to generate a corresponding target data frame, including one of:

[0207] If the pilot pattern has time domain symmetry, time domain mirroring is used to reverse the order of the original data frame on the orthogonal frequency division multiplexing resource grid to generate a corresponding target data frame;

[0208] If the pilot pattern has frequency domain symmetry, frequency domain mirroring is used to reverse the order of the original data frame on the orthogonal frequency division multiplexing resource grid to generate a corresponding target data frame.

[0209] In an embodiment, for a session initiation protocol-based orthogonal frequency division multiplexing system, the original data frame is reversed in order on an orthogonal frequency division multiplexing resource grid to generate a corresponding target data frame, including one of:

[0210] If the weighting factor matrix has time domain symmetry, the original data frame is reversed in order on the orthogonal frequency division multiplexing resource grid by time domain mirroring to generate the corresponding target data frame;

[0211] If the weighting factor matrix has frequency domain symmetry, the original data frame is reversed in order on the orthogonal frequency division multiplexing resource grid by frequency domain mirroring to generate the corresponding target data frame.

[0212] In an embodiment, the implementation mode of data enhancement by bandwidth segmentation method includes:

[0213] The resource sliding segmentation is performed based on the transmission bandwidth of the original data frame and the processing bandwidth of the neural network model to obtain the corresponding resource grid;

[0214] The corresponding target data frame is generated based on each resource grid.

[0215] In an embodiment, the data grouping mode includes one of the following: randomly divided into two groups; divided into odd and even two groups according to the receiving order of the original data frame; and grouped by using different data enhancement methods.

[0216] In an embodiment, the online training of the neural network model based on the cross-validation training algorithm and the two groups of target data includes:

[0217] The first group of target data in the two groups of target data is divided into a training group, and the second group of target data is divided into a validation group;

[0218] The target data in the validation group is used to verify the neural network model to obtain the corresponding initial model performance;

[0219] The target data in the training group is used to iteratively train the neural network model, and the target data in the validation group is used to verify the current model performance of the neural network model updated once by iteration;

[0220] The online training is performed based on the comparison result between the initial model performance and the current model performance.

[0221] In an embodiment, the online training based on the comparison result between the initial model performance and the current model performance includes:

[0222] If the current model performance is greater than the initial model performance, the online training of the neural network model is stopped;

[0223] If the current model performance is less than or equal to the initial model performance, the online training of the neural network model is continued until the current iteration number reaches the preset iteration number threshold.

[0224] In an embodiment, the online training based on the comparison result between the initial model performance and the current model performance includes:

[0225] If the current model performance is greater than the initial model performance, the current learning rate and the penalty factor of the neural network model are multiplied, and the next iteration training is performed until the current learning rate is less than the preset learning rate threshold.

[0226] In an embodiment, the model training apparatus applied to the first communication node further includes:

[0227] The division module is configured to divide the first group of target data into a verification group and divide the second group of target data into a training group.

[0228] The model training apparatus provided in the embodiment is configured to achieve Figure 1 The model training method applied to the first communication node in the embodiment, and the model training apparatus provided in the embodiment achieve similar principles and technical effects, which will not be described here.

[0229] In an embodiment, Figure 16 is a structural block diagram of another model training apparatus provided in the embodiment. The embodiment is applied to a second communication node. As shown in Figure 16 The model training apparatus in the embodiment includes a transmitter 410.

[0230] The transmitter 410 is configured to send an original data frame for neural network model training to the first communication node, so that the first communication node performs online training on the neural network model by using the original data frame.

[0231] In an embodiment, the receiving frequency of the original data frame includes: periodicity; non-periodicity.

[0232] In an embodiment, the application range of the original data frame includes: a specific terminal; a specific cell.

[0233] In an embodiment, the sending mode of the original data frame includes: the second communication node actively sends to the first communication node; the first communication node requests the second communication node to send.

[0234] In an embodiment, the bandwidth of the original data frame is greater than or equal to the processing bandwidth of the neural network model.

[0235] In an embodiment, the neural network model includes one of the following: a pilot-based orthogonal frequency division multiplexing system intelligent receiver model; a pilot-free orthogonal frequency division multiplexing system intelligent receiver model; a session initiation protocol-based orthogonal frequency division multiplexing system intelligent receiver model.

[0236] In an embodiment, the first communication node performs online training on the neural network model by using the original data frame, including:

[0237] The first communication node performs data cache updating and data enhancement on the original data frame to obtain a target data frame;

[0238] The first communication node performs data grouping on the target data frame according to a data grouping mode to obtain two groups of target data.

[0239] The first communication node performs online training on the neural network model based on the cross-validation training algorithm and the two groups of target data.

[0240] In an embodiment, the implementation process of data cache updating includes:

[0241] A cache space with a target cache space size is pre-configured;

[0242] The received new original data frame is stored in the cache space, and the original data frame with the earliest cache time in the cache space is deleted from the cache space.

[0243] In an embodiment, the data enhancement method includes at least one of the following: random phase method; constellation symmetry method; resource symmetry method; bandwidth segmentation method.

[0244] In an embodiment, the implementation of data enhancement using the random phase method includes:

[0245] Based on the original data frame and the signal transmission relationship on each resource unit, a multiplication operation of a random phase factor is performed to obtain a new signal transmission relationship;

[0246] Based on the new signal transmission relationship, a corresponding target data frame is obtained.

[0247] In an embodiment, the implementation of data enhancement using the constellation symmetry method includes:

[0248] Based on the original data frame and the signal transmission relationship on each resource unit, a multiplication operation of a rotational symmetry factor is performed to obtain a new signal transmission relationship;

[0249] Based on the new signal transmission relationship, a corresponding target data frame is obtained.

[0250] In an embodiment, the implementation of data enhancement using the constellation symmetry method includes:

[0251] Based on the original data frame and the signal transmission relationship on each resource unit, a conjugate value is obtained to obtain a new signal transmission relationship;

[0252] Based on the new signal transmission relationship, a corresponding target data frame is obtained.

[0253] In an embodiment, the implementation of data enhancement using the resource symmetry method includes:

[0254] The original data frame is reversed in order on the orthogonal frequency division multiplexing resource grid to generate a corresponding target data frame.

[0255] In an embodiment, the implementation of reversing the order includes: frequency domain mirroring; time domain mirroring; time-frequency domain mirroring.

[0256] In an embodiment, for a pilot-based orthogonal frequency division multiplexing system, reversing the order of the original data frame on the orthogonal frequency division multiplexing resource grid to generate a corresponding target data frame includes one of the following:

[0257] If the pilot pattern has time domain symmetry, time domain mirroring is used to reverse the order of the original data frame on the orthogonal frequency division multiplexing resource grid to generate a corresponding target data frame;

[0258] If the pilot pattern has frequency domain symmetry, frequency domain mirroring is used to reverse the order of the original data frame on the orthogonal frequency division multiplexing resource grid to generate a corresponding target data frame.

[0259] In an embodiment, for a session initiation protocol-based orthogonal frequency division multiplexing system, reversing the order of the original data frame on the orthogonal frequency division multiplexing resource grid to generate a corresponding target data frame includes one of the following:

[0260] If the weighting factor matrix has time domain symmetry, time domain mirroring is used to reverse the order of the original data frame on the orthogonal frequency division multiplexing resource grid to generate a corresponding target data frame;

[0261] If the weighting factor matrix has frequency domain symmetry, frequency domain mirroring is used to reverse the order of the original data frame on the orthogonal frequency division multiplexing resource grid to generate a corresponding target data frame.

[0262] In an embodiment, the implementation of data enhancement using bandwidth segmentation includes:

[0263] Based on the transmission bandwidth of the original data frame and the processing bandwidth of the neural network model, resource sliding segmentation is performed to obtain a corresponding resource grid;

[0264] Based on each resource grid, a corresponding target data frame is generated.

[0265] In an embodiment, the data grouping method includes one of the following: randomly divided into two groups; divided into odd and even groups according to the receiving order of the original data frame; and grouped using different data enhancement methods.

[0266] In an embodiment, the first communication node performs online training of the neural network model based on the cross-validation training algorithm and the two groups of target data, including:

[0267] The first communication node divides the first group of target data of the two groups of target data into a training group, and divides the second group of target data into a verification group;

[0268] The first communication node verifies the neural network model using the target data in the verification group to obtain the corresponding initial model performance;

[0269] The first communication node uses the target data in the training group to iteratively train the neural network model, and uses the target data in the validation group to validate the current model performance of the iteratively updated neural network model;

[0270] The first communication node performs online training based on a comparison result between the initial model performance and the current model performance.

[0271] In one embodiment, the first communication node performs online training based on a comparison result between the initial model performance and the current model performance, including:

[0272] If the current model performance is greater than the initial model performance, stop the online training of the neural network model;

[0273] If the current model performance is less than or equal to the initial model performance, the online training of the neural network model continues until the current number of iterations reaches the preset iteration number threshold.

[0274] In one embodiment, the first communication node performs online training based on a comparison result between the initial model performance and the current model performance, including:

[0275] If the current model performance is greater than the initial model performance, the current learning rate and penalty factor of the neural network model are multiplied, and the next iterative training is performed until the current learning rate is less than the preset learning rate threshold.

[0276] The model training device provided in this embodiment is configured to implement Figure 14 The model training method applied to the second communication node in the illustrated embodiment and the model training device provided in this embodiment are similar in implementation principle and technical effect, and will not be repeated here.

[0277] In one embodiment, Figure 17 This is a schematic diagram of the structure of a communication device provided in an embodiment of the present application. Figure 17 As shown, the device provided by this application includes: a processor 510, a memory 520 and a communication module 530. The number of processors 510 in the device can be one or more. Figure 17 In the example, a processor 510 is used. The number of memories 520 in the device can be one or more. Figure 17The memory 520 is taken as an example. The processor 510, the memory 520 and the communication module 530 of the device can be connected through a bus or other means, Figure 17 The device is taken as an example connected through a bus. In this embodiment, the device can be a first communication node or a second communication node.

[0278] The memory 520 as a computer readable storage medium can be configured to store software programs, computer executable programs and modules, such as program instructions / modules of the device of any embodiment of the present application (for example, the receiver 310 and the training module 320 in the model training apparatus applied to the first communication node). The memory 520 can include a program storage area and a data storage area, wherein the program storage area can store an operating system, an application required by at least one function; the data storage area can store data created according to the use of the device, etc. In addition, the memory 520 can include a high-speed random access memory, and can also include a non-volatile memory, such as at least one magnetic disk storage device, a flash memory device, or other non-volatile solid-state memory device. In some examples, the memory 520 can further include a memory remotely arranged with respect to the processor 510, which can be connected to the device through a network. Examples of the above network include but are not limited to the Internet, an intranet, a local area network, a mobile communication network and a combination thereof.

[0279] In the case of the communication device being the first communication node, the above-provided device can be configured to perform the model training method applied to the first communication node provided by any embodiment, and has the corresponding functions and effects.

[0280] In the case of the communication device being the second communication node, the above-provided device can be configured to perform the model training method applied to the second communication node provided by any embodiment, and has the corresponding functions and effects.

[0281] The embodiment of the present application also provides a storage medium containing computer executable instructions, which are used to execute a model training method applied to a first communication node when executed by a computer processor. The method comprises: receiving original data frames for neural network model training sent by a second communication node; and performing online training on the neural network model by using the original data frames.

[0282] The embodiment of the present application also provides a storage medium containing computer executable instructions, which are used to execute a model training method applied to a second communication node when executed by a computer processor. The method comprises: sending original data frames for neural network model training to a first communication node, so that the first communication node performs online training on the neural network model by using the original data frames.

[0283] Those skilled in the art will appreciate that the term user equipment encompasses any suitable type of wireless user equipment, such as a mobile phone, a portable data processing apparatus, a portable web browser, or a vehicle-mounted mobile station.

[0284] Generally, the various embodiments of the application can be implemented in hardware or special purpose circuits, software, logic or any combination thereof. For example, some aspects can be implemented in hardware, while other aspects can be implemented in

[0285] Embodiments of the application can be implemented by a data processor of a mobile device executing computer program instructions, for example in processor entities, or by hardware, or by a combination of software and hardware. Computer program instructions can be assembly instructions, Instruction Set Architecture (ISA) instructions, machine instructions, machine dependent instructions, microcode, firmware instructions, state setting data, or source code or object code written in any combination of one or more programming languages.

[0286] Any flowchart of the figures herein can represent program steps, or can represent interconnected logic circuits, modules, and functions, or can represent a combination of program steps and logic circuits, modules, and functions. The computer program can be stored on a memory. The memory can be of any type suitable to the local technical environment and can be implemented using any suitable data storage technology, such as, but not limited to, random access memory (RAM), read-only memory (ROM), optical storage devices, and systems, digital versatile disc (DVD) or compact disc (CD) systems, etc. The computer readable medium can comprise non-transitory storage media. The data processor can be of any type suitable to the local technical environment, and can include, but is not limited to, a general purpose computer, a special purpose computer, a microprocessor, a digital signal processor (DSP), an application specific integrated circuit (ASIC), a field programmable gate array (FPGA), and a processor based on multi-core processor architecture.

[0287] The embodiments of the application further provide a computer program product, comprising a computer program which, when executed by a processor, can implement the model training method provided by any of the embodiments of the application.

[0288] Computer program products can be written in any one of a number of programming languages, including object oriented languages such as Java, Smalltalk, C++, or conventional procedural programming languages, such as the "C" programming language or similar programming languages. The program code can execute entirely on the user's computer, partly on the user's computer, as a stand-alone software package, partly on the user's computer and partly on a remote computer or entirely on the remote computer or server. In the latter scenario, the remote computer can be connected to the user's computer through any type of network, including a local area network (LAN) or a wide area network (WAN), or the connection can be made to an external computer (for example, through the Internet using an Internet Service Provider).

[0289] The preferred embodiments of the application are described above with the understanding that these are only the preferred embodiments of the application and are not to be used to limit the application. The application can be changed and modified in various ways by those skilled in the art. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the application shall be included in the scope of the application.

Claims

1. A model training method, characterized in that: Applied to a first communication node, comprising: Receiving an original data frame for neural network model training sent by the second communication node; The neural network model is trained online using the original data frame.

2. The method according to claim 1, characterized in that The receiving frequency of the original data frame includes: periodicity and non-periodicity.

3. The method according to claim 1, characterized in that The application scope of the original data frame includes: a specific terminal; a specific cell.

4. The method according to claim 1, wherein The sending manner of the original data frame includes: the second communication node actively sending to the first communication node; the first communication node requests the second communication node to send.

5. The method according to claim 1, wherein The bandwidth of the original data frame is greater than or equal to the processing bandwidth of the neural network model.

6. The method according to claim 1, characterized in that The neural network model includes one of the following: an orthogonal frequency division multiplexing system intelligent receiver model based on a pilot; an orthogonal frequency division multiplexing system intelligent receiver model based on a pilot-free orthogonal frequency division multiplexing system intelligent receiver model; and an orthogonal frequency division multiplexing system intelligent receiver model based on a session initiation protocol.

7. The method according to claim 1, characterized in that The online training of the neural network model using the original data frame includes: Performing data cache update and data enhancement on the original data frame to obtain a target data frame; Grouping the target data frame according to a data grouping method to obtain two groups of target data; The neural network model is trained online based on a cross-validation training algorithm and the two sets of target data.

8. The method according to claim 7, characterized in that The implementation process of the data cache update includes: Pre-configure a cache space with the target cache size; The received new original data frame is stored in the cache space, and the original data frame with the earliest cache time in the cache space is deleted from the cache space.

9. The method according to claim 7, characterized in that The data enhancement method includes at least one of the following: a random phase method; a constellation diagram symmetry method; a resource symmetry method; and a bandwidth segmentation method.

10. The method according to claim 9, characterized in that The implementation of data enhancement using the random phase method includes: Performing a multiplication operation of a random phase factor based on the original data frame and the signal transmission and reception relationship on each resource unit to obtain a new signal transmission and reception relationship; A corresponding target data frame is obtained based on the new signal transmission and reception relationship.

11. The method according to claim 9, characterized in that The implementation of data enhancement using the constellation diagram symmetry method includes: Performing a multiplication operation of a rotational symmetry factor based on the original data frame and the signal transmission and reception relationship formula on each resource unit to obtain a new signal transmission and reception relationship formula; A corresponding target data frame is obtained based on the new signal transmission and reception relationship.

12. The method according to claim 9, characterized in that The implementation of data enhancement using the constellation diagram symmetry method includes: Perform conjugate value selection based on the original data frame and the signal receiving and sending relationship formula on each resource unit to obtain a new signal receiving and sending relationship formula; A corresponding target data frame is obtained based on the new signal transmission and reception relationship.

13. The method according to claim 9, characterized in that The implementation of data enhancement using the resource symmetry method includes: The original data frames are reversed in order on an orthogonal frequency division multiplexing resource grid to generate corresponding target data frames.

14. The method according to claim 13, characterized in that The implementation methods of the reverse order include: frequency domain mirroring; time domain mirroring; time-frequency domain mirroring.

15. The method according to claim 13, characterized in that For a pilot-based OFDM system, reversing the order of the original data frames on an OFDM resource grid to generate corresponding target data frames includes one of the following: If the pilot pattern has time domain symmetry, using time domain mirroring to reverse the order of the original data frame on the orthogonal frequency division multiplexing resource grid to generate a corresponding target data frame; If the pilot pattern has frequency domain symmetry, frequency domain mirroring is used to reverse the order of the original data frames on the orthogonal frequency division multiplexing resource grid to generate corresponding target data frames.

16. The method according to claim 13, characterized in that For an orthogonal frequency division multiplexing system based on the session initiation protocol, reversing the order of the original data frames on the orthogonal frequency division multiplexing resource grid to generate corresponding target data frames includes one of the following: If the weighting factor matrix has time domain symmetry, the original data frames are reversed on an orthogonal frequency division multiplexing resource grid using time domain mirroring to generate corresponding target data frames; If the weighting factor matrix has frequency domain symmetry, frequency domain mirroring is used to reverse the order of the original data frames on the orthogonal frequency division multiplexing resource grid to generate corresponding target data frames.

17. The method according to claim 9, characterized in that The implementation of data enhancement using the bandwidth splitting method includes: Perform resource sliding segmentation based on the transmission bandwidth of the original data frame and the processing bandwidth of the neural network model to obtain corresponding resource grids; A corresponding target data frame is generated based on each resource grid.

18. The method according to claim 7, characterized in that The data grouping method includes one of the following: randomly dividing into two groups; dividing into two odd and even groups according to the receiving order of the original data frames; and grouping using different data enhancement methods.

19. The method according to claim 7, characterized in that The online training of the neural network model based on the cross-validation training algorithm and the two sets of target data includes: Dividing the first set of target data of the two sets of target data into a training set, and dividing the second set of target data into a validation set; Validating the neural network model using the target data in the validation group to obtain corresponding initial model performance; Iteratively training the neural network model using the target data in the training group, and verifying the current model performance of the iteratively updated neural network model using the target data in the validation group; Online training is performed based on a comparison result between the initial model performance and the current model performance.

20. The method according to claim 19, characterized in that The performing online training based on the comparison result between the initial model performance and the current model performance includes: If the current model performance is greater than the initial model performance, stopping the online training of the neural network model; If the current model performance is less than or equal to the initial model performance, the online training of the neural network model is continued until the current number of iterations reaches a preset iteration number threshold.

21. The method according to claim 19, wherein The performing online training based on the comparison result between the initial model performance and the current model performance includes: If the current model performance is greater than the initial model performance, the current learning rate of the neural network model and the penalty factor are multiplied, and the next iterative training is performed until the current learning rate is less than a preset learning rate threshold.

22. The method according to any one of claims 19 to 21, characterized in that The method further comprises: The first set of target data of the two sets of target data is divided into a validation set, and the second set of target data is divided into a training set.

23. A model training method, characterized in that: Applied to a second communication node, comprising: Sending original data frames for neural network model training to the first communication node, so that the first communication node uses the original data frames to perform online training on the neural network model.

24. A communication device, characterized in that: include: memory, and one or more processors; The memory is configured to store one or more programs; When the one or more programs are executed by the one or more processors, the one or more processors implement the method according to any one of claims 1 to 22 or 23.

25. A storage medium, characterized in that The storage medium stores a computer program, and when the computer program is executed by a processor, the method according to any one of claims 1 to 22 or 23 is implemented.

Images