Data transmission method and device

Abstract

The invention provides a data transmission method and device, relates to the field of communication, and is helpful for enabling equipment providing training data to delete part or all of the training data in time, and helpful for enabling the load of the equipment providing the training data to be relatively small. The method is applied to a first communication device, the first communication device can be data providing equipment for providing training data or a chip or a circuit in the data providing equipment, and the method comprises the following steps: receiving first indication information from a second communication device, the first indication information comprises information indicating deletion of a first part of tags in a plurality of tags in training data and / or storage of a second part of tags in the plurality of tags, the plurality of tags comprise tags corresponding to part or all of a plurality of features in the training data, and the training data is used for model training of the second communication device; the second communication device is a model training device for model training or a chip or a circuit in the model training device.

Description

Technical Field

[0001] This application relates to the field of communications, and particularly to data transmission methods and apparatus in the field of communications. Background Technology

[0002] Machine learning (ML) is an important method in the field of artificial intelligence (AI). Through machine learning, patterns in data can be learned, and new data can be predicted and classified based on these learned patterns. These learned patterns can be represented by a model. To improve model performance, such as increasing the accuracy of model predictions, electronic devices can train the model based on training data. For example, an electronic device training a model can obtain training data from another electronic device, which is a device providing the training data (can be called a data providing device), and this data providing device can collect and store the training data; the electronic device training the model can then train the model based on the received training data.

[0003] However, this approach may place a heavy load on the data providing device, potentially impacting other services running on that device. Summary of the Invention

[0004] This application provides a data transmission method and apparatus that can reduce the load on the data providing device, thereby helping the device to handle other services normally.

[0005] In a first aspect, a data transmission method is provided, the method comprising: receiving first instruction information, the first instruction information including information indicating the deletion of a first portion of labels among multiple labels in training data and / or the saving of a second portion of labels among multiple labels, the multiple labels including labels corresponding to some or all features among multiple features in the training data, the training data being used for model training.

[0006] In one possible implementation, the method is performed by a first communication device. The first communication device can be understood as a data providing device, i.e., a device that provides training data. The first communication device can be a terminal device, an access network device, or a core network device, and the first communication device can be the device itself or a component applied to the device (e.g., a chip, a chip system, a circuit, software and / or hardware module, etc.).

[0007] The first instruction information can be information sent from the second communication device to the first communication device. The second communication device can be understood as an electronic device for model training, such as a model training device. Furthermore, the training data used by the second communication device for model training can be obtained from the first communication device.

[0008] Training data can be understood as labeled data, meaning it may include some or all of the features from multiple features, as well as multiple labels. The second communication device can obtain all the training data from the first communication device and perform model training based on it. Alternatively, the second communication device can also obtain partial training data from the first communication device and perform model training based on that partial data.

[0009] In the data transmission method of this application, the second communication device can instruct the first communication device to delete a first portion of tags and / or save a second portion of tags, so that the first communication device can determine whether to delete the first portion of tags and / or whether to save the second portion of tags. If the first communication device determines that the first portion of tags can be deleted, it can delete the first portion of tags to release some of its memory resources. If the first communication device determines that the second portion of tags can be saved, it can also delete tags other than the second portion of tags to release some of its memory resources. Thus, compared to the first communication device continuously saving all tags, it can release some memory resources earlier, thereby improving the performance of the first communication device and enabling it to process other services normally.

[0010] Furthermore, the first communication device can also save tags other than the first set of tags or save the second set of tags based on the first instruction information, so that the second communication device can subsequently retrieve the tags saved by the first communication device. In this way, compared to the first communication device prematurely deleting all tags, it helps the second communication device to obtain enough tags from the first communication device in a timely manner for model training, thereby helping the second communication device to train the model in a timely manner.

[0011] In some embodiments of the first aspect, the first instruction information includes at least one of the following: information indicating a first part of the label; information indicating the deletion of the first part of the label; information indicating a second part of the label; or, information indicating the saving of the second part of the label.

[0012] The information indicating the first part of the label includes information indicating the index of each label in the first part of the label, or the information indicating the first part of the label includes a first bitmap used to indicate the first part of the label among multiple labels; and / or the information indicating the second part of the label includes information indicating the index of each label in the second part of the label, or the information indicating the second part of the label includes a second bitmap used to indicate the second part of the label among multiple labels.

[0013] This makes it easier for the first communication device to determine the first part of the tag and / or the second part of the tag based on the first instruction information, and to determine whether to delete the first part of the tag and / or save the second part of the tag.

[0014] In some embodiments of the first aspect, the first portion of the tag includes: all of the tags in a plurality of tags, or tags in a plurality of tags that the first communication device has sent to the second communication device, or tags in a plurality of tags that the first communication device has not sent to the second communication device; and / or, the second portion of the tag includes: all of the tags in a plurality of tags, or tags in a plurality of tags that the first communication device has sent to the second communication device, or tags in a plurality of tags that the first communication device has not sent to the second communication device.

[0015] It should be understood that, in the case where the first instruction information instructs the deletion of the first part of the label and the saving of the second part of the label, the first part of the label and the second part of the label do not include the same label.

[0016] This allows the first communication device to determine, based on the first instruction, whether to delete all tags, tags that have been reported, or tags that have not been reported from among the multiple tags; and / or to determine whether to save all tags, tags that have been reported, or tags that have not been reported from among the multiple tags.

[0017] In some possible implementations, the first set of labels may also be labels corresponding to features that the second communication device determines have low value for model training. For example, if the second communication device determines that labels 1 and 2 among multiple features have low value for model training, then the second communication device may instruct the first communication device to delete labels 1 and 2, etc.

[0018] In some possible implementations, the second part of the label may also be the label corresponding to a feature that the second communication device has high value for model training, or the second part of the label may be a label that the second communication device determines may need to acquire later. This facilitates the second communication device in subsequently acquiring the second part of the label from the first communication device to continue model training.

[0019] In some embodiments of the first aspect, the method further includes: receiving second indication information, the second indication information including information indicating a first duration or information indicating a first time period, the first duration being used to indicate the duration for which the first communication device stores the tag, and the first time period being used to indicate the time period for which the first communication device stores the tag.

[0020] This allows the first communication device to save tags within a first duration or time period based on the second instruction information, so that the first communication device can promptly send the tags saved by the first communication device to the second communication device within the first duration or time period, thereby helping the second communication device to train the model in a timely manner.

[0021] In some embodiments of the first aspect, the method further includes: sending third indication information, the third indication information including information indicating a second duration or information indicating a second time period, the second duration being used to indicate the duration for which the first communication device stores the tag, and the second time period being used to indicate the time period for which the first communication device stores the tag.

[0022] In this way, the second communication device can determine the second duration or second time period for which the first communication device can store the labels. This facilitates the second communication device acquiring enough labels to train the model within the second duration or second time period, helping the model training device to train the model in a timely manner. For example, when the second duration or second time period is short, the second communication device can acquire more labels from the first communication device at once. For instance, the third part of the labels acquired from the first communication device can include a larger number of labels, allowing the second communication device to train the model earlier based on more labels, thus reducing the number of times the second communication device needs to acquire labels from the first communication device.

[0023] In some embodiments of the first aspect, the method further includes: sending a third part tag, wherein the third part tag is some or all of a plurality of tags.

[0024] In this scenario, the first part of the label can be the same as the third part of the label, in which case the first instruction information can instruct the deletion of the reported label; alternatively, the first part of the label can also be different from the third part of the label. The second part of the label can be the same as the third part of the label, in which case the first instruction information can instruct the saving of the reported label; alternatively, the second part of the label can be a label other than the third part of the label, in which case the first instruction information can instruct the saving of the unreported label; alternatively, the second part of the label can also include other labels, such as multiple labels or some of the first part of the label.

[0025] This makes it easier for the second communication device to train a model based on the third part of the label and the features corresponding to the third part of the label.

[0026] In some embodiments of the first aspect, the method further includes: sending fourth indication information, the fourth indication information including information indicating a third duration or information indicating a third time period, the third duration being used to indicate the duration for which the first communication device stores the fourth part tag, the third time period being used to indicate the time period for which the first communication device stores the fourth part tag, and the fourth part tag being a tag other than the third part tag among a plurality of tags.

[0027] In this way, the second communication device can determine that the first communication device can save the third duration or third time period of the fourth part of the tags, so that the second communication device can obtain the tags in the fourth part of the tags from the first communication device within the third duration or third time period, which helps the model training device to train the model in a timely manner.

[0028] In some embodiments of the first aspect, the method further includes: sending information indicating a first data amount, the first data amount being the maximum data amount of the tag stored in the first communication device, and the third part of the tag being determined based on the first data amount.

[0029] Thus, when the initial data volume is large, meaning the first communication device can store a large number of tags, the number of tags included in the third part of the tags obtained by the second communication device from the first communication device can be smaller, so that the transmission of the third part of the tags occupies relatively less time-frequency domain resources. When the initial data volume is small, meaning the first communication device can store fewer tags, the number of tags included in the third part of the tags obtained by the second communication device from the first communication device can be larger, so that the first communication device can delete more tags, such as instructing the first communication device to delete the third part of the tags, thereby preventing the load on the first communication device from being too large and allowing the first communication device to handle other services normally.

[0030] In some embodiments of the first aspect, the method further includes sending a plurality of features before receiving the first indication information.

[0031] This makes it easier for the second communication device to determine high-value unlabeled data based on multiple features, namely the features corresponding to the third part of the label.

[0032] In one possible implementation, the multiple features may be determined based on a fifth indication information. The method may further include: receiving the fifth indication information, which indicates one or more of the following: the data type of the multiple features, the data type of the tags corresponding to the multiple features, the number of the multiple features, or the time period during which the first communication device collected the multiple features.

[0033] In this way, the first communication device can measure multiple features based on the fifth indication information, and can send multiple features to the second communication device based on the fifth indication information.

[0034] In some embodiments of the first aspect, the method further includes: receiving information for indicating feedback features and / or no-feedback labels.

[0035] In this way, the first communication device can send features to the second communication device without sending tags, so that tags corresponding to some of the features can be sent to the second communication device later. Compared with sending multiple features and tags corresponding to each of the multiple features at the same time, the data volume of tags corresponding to some of the features is smaller and the time and frequency domain resources occupied are relatively less.

[0036] In some embodiments of the first aspect, the method further includes receiving information indicating whether to save the tag.

[0037] In this way, the first communication device can determine whether to save the tag based on this information. If it is determined not to save the tag, the first communication device can delete the tag as early as possible, thereby freeing up more memory resources; if it is determined to save the tag, the tag can be saved so that it can be sent to the second communication transpose in a timely manner.

[0038] In some embodiments of the first aspect, the method further includes: determining, based on first indication information, whether to delete the first part of the label and / or whether to save the second part of the label.

[0039] This facilitates the first communication device to delete and / or save tags based on the instructions of the second communication device, helps the first communication device to delete the first part of the tags in a timely manner to free up memory resources, and / or helps the second communication device to save the second part of the tags so that the first communication device can subsequently send the tags in the second part of the tags to the second communication device in a timely manner.

[0040] Secondly, another data transmission method is provided, the method comprising: determining first instruction information, the first instruction information including information indicating the deletion of a first part of the labels in a plurality of labels in the training data and / or the saving of a second part of the labels in a plurality of labels, the plurality of labels including labels corresponding to some or all of the features in a plurality of features in the training data, the training data being used for model training; and sending the first instruction information.

[0041] In one possible implementation, the method is performed by a second communication device. The second communication device can be understood as a model training device, that is, a device that acquires training data from a data providing device and trains a model based on the training data. The second communication device can be a terminal device, an access network device, or a core network device, and can be the device itself or a component applied to the device (e.g., a chip, chip system, circuit, software and / or hardware module, etc.).

[0042] In some embodiments of the second aspect, the first instruction information includes at least one of the following: information indicating a first part of the label; information indicating the deletion of the first part of the label; information indicating a second part of the label; or, information indicating the saving of the second part of the label.

[0043] In some embodiments of the second aspect, the first part of the tag includes: all of the tags in a plurality of tags, or tags in a plurality of tags that the first communication device has sent to the second communication device, or tags in a plurality of tags that the first communication device has not sent to the second communication device; and / or, the second part of the tag includes: all of the tags in a plurality of tags, or tags in a plurality of tags that the first communication device has sent to the second communication device, or tags in a plurality of tags that the first communication device has not sent to the second communication device.

[0044] In some embodiments of the second aspect, the method further includes: sending second indication information, the second indication information including information indicating a first duration or information indicating a first time period, the first duration being used to indicate the duration for which the first communication device stores the tag, and the first time period being used to indicate the time period for which the first communication device stores the tag.

[0045] In some embodiments of the second aspect, the method further includes: receiving third indication information, the third indication information including information indicating a second duration or information indicating a second time period, the second duration being used to indicate the duration for which the first communication device stores the tag, and the second time period being used to indicate the time period for which the first communication device stores the tag.

[0046] In some embodiments of the second aspect, the method further includes: receiving a third part tag, wherein the third part tag is some or all of a plurality of tags.

[0047] In some embodiments of the second aspect, the method further includes: receiving fourth indication information, the fourth indication information including information indicating a third duration or information indicating a third time period, the third duration being used to indicate the duration for which the first communication device stores the fourth part of the tag, the third time period being used to indicate the time period for which the first communication device stores the fourth part of the tag, and the fourth part of the tag being a tag other than the third part of the tag among a plurality of tags.

[0048] In some embodiments of the second aspect, the method further includes: receiving information indicating a first data amount, the first data amount being the maximum data amount of a tag stored by the first communication device, and the third part of the tag being determined based on the first data amount.

[0049] In some embodiments of the second aspect, the method further includes receiving a plurality of features before receiving the first indication information.

[0050] In one possible implementation, the multiple features may be determined based on a fifth indication message. The method may further include: sending the fifth indication message, which indicates one or more of the following: the data type of the multiple features, the data type of the tags corresponding to the multiple features, the number of the multiple features, or the time period during which the first communication device collects the multiple features.

[0051] In some embodiments of the second aspect, the method further includes: sending information for indicating feedback features and / or no-feedback labels.

[0052] In some embodiments of the second aspect, the method further includes sending information indicating whether to save the tag.

[0053] Thirdly, a communication device is provided, which can be used in the first communication device of the first aspect or the second communication device of the second aspect. The communication device can be a terminal device or a network device, or it can be a device in the terminal device or network device (e.g., a chip, or a chip system, or a circuit, such as a circuit or chip in the terminal device that is responsible for communication functions (e.g., a modem chip, also known as a baseband chip, or a system-on-chip (SoC) chip or system-in-package (SIP) chip containing a modem core)), or a device that can be matched with the terminal device or network device, or it can be a logic module or software that can implement all or part of the functions of the terminal device or network device.

[0054] In one possible implementation, the communication device may include modules or units that perform the methods / operations / steps / actions described in the first or second aspect. These modules or units may be hardware circuits, software, or a combination of hardware circuits and software.

[0055] In one possible implementation, the communication device is used in the first aspect of the first communication device, which may include a transceiver unit. The transceiver unit is used to receive first instruction information, which includes instructions regarding deleting a first portion of labels from a plurality of labels in training data and / or saving a second portion of labels from a plurality of labels. The plurality of labels include labels corresponding to some or all of the features in a plurality of features in the training data, and the training data is used for model training.

[0056] Alternatively, the communication device may be used as a second communication device in the second aspect, and the communication device may include a processing unit and a transceiver unit. The processing unit is used to determine first instruction information, the first instruction information including information indicating the deletion of a first portion of labels from a plurality of labels in the training data and / or the saving of a second portion of labels from a plurality of labels, the plurality of labels including labels corresponding to some or all of the features in a plurality of features in the training data, the training data being used for model training; the transceiver unit is used to send the first instruction information.

[0057] Fourthly, this application provides another communication device, including a processor coupled to a memory, which can be used to execute instructions in the memory to implement the method in any of the possible implementations of the first or second aspect described above. Optionally, the communication device further includes a memory. Optionally, the communication device further includes a communication interface, to which the processor is coupled.

[0058] In one implementation, the communication device is a terminal device or a network device. When the communication device is a terminal device or a network device, the communication interface can be a transceiver or an input / output interface.

[0059] In another implementation, the communication device is a chip applicable to terminal devices or network devices. When the communication device is a chip applicable to terminal devices or network devices, the aforementioned communication interface can be an input / output interface.

[0060] Fifthly, a processor is provided, comprising: an input circuit, an output circuit, and a processing circuit. The processing circuit is configured to receive signals through the input circuit and transmit signals through the output circuit, causing the processor to execute the method in any possible implementation of the first or second aspect described above.

[0061] In the specific implementation process, the processor can be a chip, the input circuit can be an input pin, the output circuit can be an output pin, and the processing circuit can be a transistor, gate circuit, flip-flop, and various logic circuits. The input signal received by the input circuit can be received and input by, for example, but not limited to, a receiver, and the signal output by the output circuit can be output to, for example, but not limited to, a transmitter and transmitted by the transmitter. Furthermore, the input circuit and the output circuit can be the same circuit, which is used as the input circuit and the output circuit at different times. This application does not limit the specific implementation of the processor and various circuits.

[0062] In a sixth aspect, a communication device is provided, including a processor and a memory. The processor is configured to read instructions stored in the memory, receive signals via a receiver, and transmit signals via a transmitter to execute the method in any possible implementation of the first or second aspect described above.

[0063] Optionally, the processor may be one or more, and the memory may be one or more.

[0064] Optionally, the memory may be integrated with the processor, or the memory may be separated from the processor.

[0065] In the specific implementation process, the memory can be a non-transitory memory, such as read-only memory (ROM), which can be integrated with the processor on the same chip or set on different chips. This application does not limit the type of memory or the way the memory and processor are set.

[0066] It should be understood that the relevant data interaction process, such as sending instruction information, can be a process of outputting instruction information from the processor, and receiving capability information can be a process of the processor receiving input capability information. Specifically, the data processed and output can be sent to the transmitter, and the input data received by the processor can come from the receiver. Here, the transmitter and receiver can be collectively referred to as a transceiver.

[0067] The communication device in the sixth aspect above can be a chip. The processor can be implemented in hardware or software. When implemented in hardware, the processor can be a logic circuit, integrated circuit, etc. When implemented in software, the processor can be a general-purpose processor that reads software code stored in memory. The memory can be integrated into the processor or located outside the processor and exist independently.

[0068] In a seventh aspect, a computer program product is provided, the computer program product comprising: a computer program (also referred to as code or instructions), which, when the computer program is run, causes a computer to perform the method in any possible implementation of the first or second aspect described above.

[0069] Eighthly, a computer-readable storage medium is provided that stores a computer program (also referred to as code or instructions) that, when executed on a computer, causes the computer to perform the methods in any possible implementation of the first or second aspect described above. Attached Figure Description

[0070] Figure 1 This is a schematic diagram of a communication system to which the embodiments of this application apply;

[0071] Figure 2 This is a schematic diagram of a supervised learning process;

[0072] Figure 3 This is a schematic diagram of a model training process;

[0073] Figure 4 A schematic diagram illustrating a process for identifying high-value unlabeled data;

[0074] Figure 5 A flowchart illustrating a data transmission method provided in an embodiment of this application;

[0075] Figure 6 A flowchart illustrating another data transmission method provided in an embodiment of this application;

[0076] Figure 7 A schematic block diagram of a communication device provided in an embodiment of this application;

[0077] Figure 8 A schematic block diagram of another communication device provided in the embodiments of this application;

[0078] Figure 9 A schematic block diagram of another communication device provided in the embodiments of this application;

[0079] Figure 10 This is a schematic block diagram of an O-RAN architecture provided for an embodiment of this application. Detailed Implementation

[0080] The technical solutions in this application will now be described with reference to the accompanying drawings.

[0081] To facilitate understanding of the embodiments of this application, the following points are explained first:

[0082] First, in the embodiments of this application, the terms "first" and "second" are used to distinguish identical or similar items with essentially the same function and effect. For example, the first value and the second value are only used to distinguish different values ​​and do not limit their order. Those skilled in the art will understand that the terms "first" and "second" do not limit the quantity or execution order, and the terms "first" and "second" do not necessarily imply that they are different.

[0083] It should be noted that, in the embodiments of this application, the words "exemplarily" or "for example" are used to indicate examples, illustrations, or explanations. Any embodiment or design scheme described as "exemplarily" or "for example" in this application should not be construed as being more preferred or advantageous than other embodiments or design schemes. Specifically, the use of the words "exemplarily" or "for example" is intended to present the relevant concepts in a specific manner.

[0084] In the embodiments of the present application, "at least one" means one or more, and "multiple" means two or more. "And / or" describes the association relationship of associated objects and indicates that there can be three relationships. For example, A and / or B can represent: A exists alone, A and B exist simultaneously, and B exists alone, where A and B can be singular or plural. The character " / " generally indicates that the associated objects before and after are in an "or" relationship. "At least one (item)" or its similar expression refers to any combination of these items, including any combination of single item or plural items. For example, at least one (item) of a, b, or c can represent: a, b, c, a - b, a - c, b - c, or a - b - c, where a, b, and c can be single or multiple.

[0085] Second, in the embodiments of the present application, "send" and "receive" represent the direction of signal transmission. For example, "send information to the second device" can be understood as the destination of this information is the second device, which can include directly sending through the air interface, and also includes indirectly sending through the air interface by other units or modules. "Receive the configuration information from the charging" can be understood as the source of this configuration information is the second device, which can include directly receiving from the second device through the air interface, and can also include indirectly receiving from the second device through the air interface by other units or modules. "Send" can also be understood as "output" of the chip interface, and "receive" can also be understood as "input" of the chip interface.

[0086] In other words, sending and receiving can be carried out between devices, for example, between the second device and the first device; or can be carried out within a device, for example, sending or receiving between components, modules, chips, software modules or hardware modules within the device through a bus, trace or interface.

[0087] It can be understood that before the information is sent from the source end to the destination end, necessary processing may be performed, such as encoding, modulation, etc. After the destination end receives the information from the source end, corresponding processing can also be performed, such as decoding, demodulation, etc., so as to interpret the valid information from the source end. Similar expressions in the present application can be understood similarly and will not be elaborated here.

[0088] Third, for the convenience of understanding, multiple examples of information or indication information are provided in this article, such as the first indication information, the second indication information, etc. The sequence of transmission and the names of these information are all examples and should not constitute any limitation to the present application.

[0089] Fourth, in the embodiments of this application, "instruction" can include direct instruction and indirect instruction, as well as explicit instruction and implicit instruction. The information indicated by a certain piece of information (as described below, the instruction information) is called the information to be instructed. In the specific implementation process, there are many ways to indicate the information to be instructed, such as, but not limited to, directly indicating the information to be instructed, such as the information to be instructed itself or its index. It can also indirectly indicate the information to be instructed by indicating other information, where there is a correlation between the other information and the information to be instructed; or it can only indicate a part of the information to be instructed, while the other parts of the information to be indicated are known or pre-agreed upon. For example, the instruction of specific information can be achieved by using a pre-agreed (e.g., protocol predefined) arrangement of various pieces of information, thereby reducing the instruction overhead to a certain extent. This application does not limit the specific method of instruction.

[0090] It is understandable that, for the sender of the instruction information, the instruction information can be used to indicate the information to be indicated, and for the receiver of the instruction information, the instruction information can be used to determine the information to be indicated.

[0091] Fifth, the tables in the embodiments of this application are merely examples. The values ​​of the information in each table are only examples and can be configured to other values; this application is not limited thereto. The tables do not limit the scope of protection of this application. For example, appropriate modifications and adjustments can be made based on the tables described above, such as splitting, merging, etc. Furthermore, the parameter names shown in the headings of each table can also use other names understandable to the communication device, and the values ​​or representations of the parameters can also be other values ​​or representations understandable to the communication device. Moreover, in the implementation of the above tables, other data structures can also be used, such as arrays, queues, containers, stacks, linear lists, pointers, linked lists, trees, graphs, structures, classes, heaps, hash tables, or hash tables, etc.

[0092] Sixth, in the embodiments of this application, descriptions such as "when," "under the circumstances," "if," and "if" all refer to the fact that the device (e.g., network device or terminal device) will make corresponding processing under certain objective circumstances. They are not time limits, nor do they require the device (e.g., network device or terminal device) to have a judgment action when implementing it, nor do they mean that there are other limitations.

[0093] Seventh, the predefined terms in this application can be understood as: definition, pre-defined, storage, pre-storage, pre-negotiation, pre-configuration, solidification, or pre-firing.

[0094] Eighth, the term "storage" in this application can refer to storage in one or more memory devices. These memory devices can be separate installations or integrated into an encoder, decoder, processor, or communication device. Alternatively, some memory devices can be separately installed, while others can be integrated into the decoder, processor, or communication device. The type of memory can be any form of storage medium, and this application does not limit this.

[0095] The technical solutions of this application can be applied to various communication systems, such as: Long Term Evolution (LTE) systems, LTE Frequency Division Duplex (FDD) systems, LTE Time Division Duplex (TDD) systems, 5th Generation (5G) systems, or New Radio (NR) systems, and future communication systems.

[0096] The terminal equipment in this application embodiment can also be referred to as: user equipment (UE), mobile station (MS), mobile terminal (MT), access terminal, user unit, user station, mobile station, mobile station, remote station, remote terminal, mobile device, user terminal, terminal, wireless communication equipment, user agent, or user device, etc.

[0097] Terminal devices can be devices that provide voice / data connectivity to users, such as handheld devices with wireless connectivity, in-vehicle devices, etc. Currently, examples of terminal devices include: mobile phones, tablets, laptops, PDAs, mobile internet devices (MIDs), point-of-sale (POS) machines, customer-premises equipment (CPEs), light user equipment (UEs), reduced capability UEs (REDCAP UEs), wearable devices, virtual reality (VR) devices, augmented reality (AR) devices, wireless terminals in industrial control, wireless terminals in self-driving vehicles, wireless terminals in remote medical surgery, wireless terminals in smart grids, wireless terminals in transportation safety, wireless terminals in smart cities, wireless terminals in smart homes, cellular phones, cordless phones, session initiation protocol (SIP) phones, wireless local loop (WLL) stations, and personal digital assistants (PDAs). This application does not limit the scope to include devices such as personal assistants (PDAs), handheld devices with wireless communication capabilities, computing devices or other processing devices connected to a wireless modem, in-vehicle devices, wearable devices, terminal devices in 5G networks, or terminal devices in future evolved public land mobile networks (PLMNs).

[0098] By way of example and not limitation, in this application, the terminal device can be a terminal device in an Internet of Things (IoT) system. The Internet of Things is an important component of future information technology development. Its main technical characteristic is connecting objects to networks through communication technologies, thereby realizing an intelligent network of human-machine interconnection and object-to-object interconnection. Exemplarily, the terminal device in the embodiments of this application can be a wearable device. Wearable devices, also known as wearable smart devices, are a general term for devices that apply wearable technology to intelligently design and develop everyday wearables, such as glasses, gloves, watches, clothing, and shoes. Wearable devices are portable devices that can be worn directly on the body or integrated into a user's clothing or accessories. Wearable devices are not merely hardware devices; they can also achieve powerful functions through software support, data interaction, and cloud interaction. Broadly defined, wearable smart devices include those with comprehensive functions, large size, and the ability to achieve complete or partial functions without relying on a smartphone, such as smartwatches or smart glasses, as well as those focused on a specific application function and requiring the use of other devices such as smartphones, such as various smart bracelets and smart jewelry for vital sign monitoring.

[0099] By way of example and not limitation, in the embodiments of this application, the terminal device can also be a terminal device in machine-type communication (MTC). Furthermore, the terminal device can also be an on-board module, on-board component, on-board chip, or on-board unit, etc., built into a vehicle as one or more components or units. The vehicle can implement the methods provided in this application through the built-in on-board module, on-board component, on-board chip, or on-board unit, etc. Therefore, the embodiments of this application can also be applied to vehicle networking, such as vehicle-to-everything (V2X), long-term evolution-vehicle (LTE-V) technology, and vehicle-to-vehicle (V2V) technology.

[0100] The network equipment involved in this application may include access network equipment and core network equipment.

[0101] Access network equipment, also known as radio access network (RAN) equipment, is a device that communicates with terminal devices and has wireless transceiver capabilities. RAN equipment provides wireless communication services, allowing terminals to access the wireless network. RAN equipment can be a node in the radio access network, often referred to as a RAN node.

[0102] In one possible scenario, a RAN node can be a base station (BS), an evolved NodeB (eNodeB), a transmission reception point (TRP), a home evolved NodeB (or home Node B, HNB), a Wi-Fi access point (AP), a mobile switching center, a next-generation NodeB (gNB) in a 5G mobile communication system, a next-generation base station in a future mobile communication system, or a base station in a future mobile communication system. A RAN node can also be a device that performs base station functions in device-to-device (D2D) communication systems, vehicle-to-everything (V2X) communication systems, machine-to-machine (M2M) communication systems, and internet-to-things (IoT) communication systems. A RAN node can also be a RAN node in a non-terrestrial network (NTN), meaning that a RAN node can be deployed on a high-altitude platform or a satellite. RAN nodes can be macro base stations, micro base stations, indoor stations, relay nodes, donor nodes, etc., or radio controllers in cloud radio access network (CRAN) scenarios, or nodes in open radio access network (O-RAN or ORAN) scenarios. Optionally, RAN nodes can also be servers, wearable devices, vehicles, or in-vehicle equipment. For example, in V2X technology, RAN nodes can be roadside units (RSUs). Of course, RAN nodes can also be nodes in the core network.

[0103] In another possible scenario, multiple RAN nodes collaborate to assist the terminal in achieving wireless access, with different RAN nodes each implementing a portion of the base station's functions. For example, RAN nodes can be central units (CUs), distributed units (DUs), CU-control plane (CPs), CU-user plane (UPs), or radio units (RUs), etc. CUs and DUs can be set up separately or included in the same network element, such as a baseband unit (BBU). RUs can be included in radio frequency equipment or radio frequency units, such as remote radio units (RRUs), active antenna units (AAUs), or remote radio heads (RRHs).

[0104] In different systems, CU (or CU-CP and CU-UP), DU, or RU may have different names, but those skilled in the art will understand their meaning. For example, in the ORAN system, CU can also be called open CU (O-CU), DU can also be called open DU (O-DU), CU-CP can also be called open CU-CP (O-CU-CP), CU-UP can also be called open CU-UP (O-CU-UP), and RU can also be called open RU (O-RU).

[0105] Any one of the CU (or CU-CP, CU-UP), DU, and RU units can be implemented through software modules, hardware modules, or a combination of software and hardware modules. That is, the wireless access network device in this application can be a virtualized device, for example, implemented through general-purpose hardware and instantiated virtualization functions, or dedicated hardware and instantiated virtualization functions. The general-purpose hardware can be a server, such as a cloud server.

[0106] The core network equipment in this application embodiment can be a core network equipment in a 4G system, such as a mobile management entity (MME) or a serving gateway (SGW), or a core network equipment in a 5G system, such as an access and mobility management function (AMF) network element or a user plane function (UPF) network element. It can also be a core network equipment with other names, or it can be a core network equipment in a future communication system. This application embodiment does not limit this.

[0107] First, let's introduce some of the technical terms used in this application.

[0108] 1. Supervised learning

[0109] It is a machine learning method in which the model is trained based on training samples. Each training sample consists of input data and the corresponding correct output. The goal of the model is to learn the mapping relationship from input to output and build a model to express this mapping relationship so that it can make accurate predictions on new data.

[0110] 2. Features and Labels

[0111] Features can be used to describe the attributes or variables of data and are the inputs that models use for learning and prediction.

[0112] Labels are the output or result of each training sample in supervised learning; they are the values ​​that the model needs to predict. Labels are usually known and are used to guide the model's training process.

[0113] 3. Unlabeled data and labeled data

[0114] Unlabeled data refers to data that does not contain labels, and unlabeled data can also be understood as features.

[0115] Labeled data can include two parts: features and labels. The training data used to train the model can be labeled data.

[0116] 4. Data labeling

[0117] Labeling, also known as tagging, refers to the process of generating corresponding labels for unlabeled data (i.e., features).

[0118] 5. Simulation data and live network data

[0119] Simulation data refers to data generated by simulating a real system or process. Live network data refers to real data collected from an actual operating network or system, reflecting the system's actual state and behavior in a real environment. In other words, simulation data can be understood as simulated data, while live network data can be understood as real data.

[0120] To facilitate understanding of the embodiments of this application, firstly, in conjunction with Figure 1 The communication system applicable to the embodiments of this application will be described in detail.

[0121] Figure 1 This is a schematic diagram of the architecture of a communication system 1000 applicable to the methods provided in the embodiments of this application. For example... Figure 1 As shown, the communication system 1000 includes a wireless access network 100 and a core network 200. Optionally, the communication system 1000 may also include an Internet 300.

[0122] The wireless access network 100 may include at least one access network device (such as...) Figure 1 110a and 110b in the above), may also include at least one terminal device (such as Figure 1 (120a-120j in the middle).

[0123] Terminal devices can connect to access network devices wirelessly, while access network devices can connect to the core network wirelessly or via wired connections. Core network devices and access network devices can be independent, separate physical devices, or they can integrate the functions of core network devices and the logical functions of access network devices onto the same physical device. Alternatively, a single physical device can integrate some core network device functions and some access network device functions. Terminal devices and access network devices can connect to each other via wired or wireless connections.

[0124] It should be understood that the connection between devices can also be understood as the ability of devices to communicate with each other. For example, the connection between access network devices means that access network devices can communicate with each other.

[0125] Wireless access network devices and terminal devices, as well as access network devices and terminal devices, can communicate using licensed spectrum, unlicensed spectrum, or both simultaneously. Communication can occur using spectrum below 6 GHz, spectrum above 6 GHz, or both simultaneously. The embodiments of this application do not limit the spectrum resources used for wireless communication.

[0126] Among them, the wireless access network equipment can be a base station deployed in the air, such as a satellite base station 110a; or it can be a base station deployed indoors, such as a micro base station or an indoor station 110b.

[0127] The terminal device can be a terminal device deployed in the air, such as... Figure 1 The 120i can be a helicopter or drone; it can also be a terminal device deployed on the ground, such as... Figure 1 The following are examples: mobile phones 120a, 120e, 120f, 120j, vehicle 120b, computer 110b, and printer 120h.

[0128] Access network equipment and terminal equipment can be fixed or mobile. For example, access network equipment and terminal equipment can be deployed on land, including indoors or outdoors, handheld or vehicle-mounted; they can also be deployed on water; and they can also be deployed in the air on airplanes, balloons, and artificial satellites.

[0129] The roles of access network equipment and terminal equipment can be relative. For example, Figure 1 In this context, the helicopter or drone 100i can be configured as a mobile base station. For those 120j accessing the wireless access network 100 via 120i, 120i is a base station; however, for 110a, 120i is a terminal device, meaning that 110a and 120i communicate via a wireless air interface protocol. Alternatively, 110a and 120i can also communicate via an interface protocol between access network devices; in this case, 120i is also a base station relative to 110a. Therefore, both wireless access network devices and terminal devices can be collectively referred to as communication devices. Figure 1 The 110a, 110b, and 120a-120j in the text can be referred to as communication devices with their respective corresponding functions, such as communication devices with base station functions or communication devices with terminal functions.

[0130] It should be understood that Figure 1 This is just an illustration; the communication system may also include other devices, such as wireless repeaters and wireless backhaul devices. Figure 1 Not shown in the figure. This application does not limit this aspect in its embodiments.

[0131] The aforementioned communication devices, such as Figure 1Access network devices or terminal devices can be configured with multiple antennas. These multiple antennas may include at least one transmitting antenna for transmitting signals and at least one receiving antenna for receiving signals. Additionally, each communication device also includes a transmitter chain and a receiver chain, which, as will be understood by those skilled in the art, may include multiple components related to signal transmission and reception (e.g., processors, modulators, multiplexers, demodulators, demultiplexers, or antennas). Therefore, access network devices and terminal devices can communicate via multi-antenna technology.

[0132] Optionally, the communication system 1000 may also include other network entities such as a network controller and a mobility management entity, but the embodiments of this application are not limited thereto.

[0133] It should also be understood that the method provided in the embodiments of this application can be applied to a variety of communication systems, including 5G new radio (NR) systems. Communication system 1000 is only an example. This application does not limit the specific architecture of the applicable system, nor does it limit the number and form of various devices contained in each communication system.

[0134] The devices in the aforementioned communication system 1000 can perform machine learning (ML), an important branch of artificial intelligence (AI) technology that enables learning and prediction. Through machine learning, patterns in data can be learned, and new data can be predicted and classified based on these learned patterns.

[0135] Supervised learning is an important learning method in machine learning. Supervised learning learns the mapping relationship between input and output data and builds a model to express this mapping relationship. In supervised learning, the model trainer, i.e., the electronic device that trains the model, needs to train the model based on a large amount of data. The data used in the model training process can also be called training samples, training data, or sample data, etc.

[0136] like Figure 2 As shown, each data point in the training data consists of two parts: input data and output data. The input data can be called a feature, and the output data can be called a label. Therefore, a data point in the training data can be represented as <feature, label>.

[0137] It should be noted that, Figure 2The output data shown is not the actual output data of the model after the input data (i.e., features) is input, but rather the output data (i.e., labels) from the training data used during model training. The labels in the training data can be real or accurate data. The model trainer can adjust the parameters in the model based on the output data (i.e., labels) and the actual output data of the model until a model that meets the conditions is obtained, that is, a trained model.

[0138] It should be understood that, for ease of description, the data used by the model trainer to train the model will be referred to as training data in the following text. Training data can be labeled data, that is, training data includes features and labels, which will not be elaborated on further in the following text.

[0139] It should be noted that the model trainer can be understood as a device that needs to acquire training data and train the model based on that data. In some implementations, the model trainer can acquire training data from other devices. For ease of distinction, the model trainer will be referred to as the model training device, and the device providing the training data will be referred to as the data providing device. That is, the data providing device can measure the training data and can send the training data to the model training device so that the model training device can train the model based on the training data.

[0140] However, in some scenarios, the cost of obtaining labels from training data may be high, such as the following two scenarios.

[0141] Scenario 1: Data providing devices require significant computing power to collect labels. For example, the model training device needs to train a model for predicting the location of a terminal device. The data providing device can be the terminal device, and the data type of the labels in the training data is the location information of the terminal device (e.g., coordinates). To provide accurate labels to the model training device, the terminal device needs to measure its own location, resulting in high energy consumption.

[0142] Scenario 2: Tag transmission requires significant transmission resources. For example, the model training device needs to train a model related to transmission channel characteristic information (such as a channel matrix), such as a decompression model. The tag, however, represents the raw channel characteristic information. Since the tag data volume is typically large, transmitting the tag requires substantial transmission resources.

[0143] Therefore, to reduce costs, model training devices can train models using as little training data (labeled data) as possible, allowing data providers to label fewer features and thus reduce power consumption; and / or, data providers can transmit fewer labels to model training devices, minimizing transmission resource consumption. Active learning is one method that can reduce the number of labels required.

[0144] When labels are acquired through active learning, the data providing device may not label all features. For example, it can label high-value unlabeled data (i.e., features) while leaving low-value unlabeled data unlabeled, thereby reducing the number of features that need to be labeled and thus reducing the power consumption of the data providing device. And / or, the data providing device can send some labels to the model training device, thereby reducing the transmission resources consumed by label transmission.

[0145] Below, in conjunction with Figure 3 The method 300 for collecting data through active learning to train a model is explained. For example... Figure 3 As shown, method 300 includes the following steps:

[0146] S301, The model training device acquires the initial model.

[0147] The initial model can be, for example, a model trained on a small amount of labeled data. This small amount of labeled data can be collected by the model training device from other devices, or data measured by the model training device. When the initial model does not meet the requirements and training is necessary, the following steps can be performed.

[0148] S302. The model training device acquires a set of unlabeled data and identifies several high-value unlabeled data from it.

[0149] It should be understood that high-value unlabeled data can be understood as features that the model training equipment determines need to be labeled, and high-value unlabeled data is a portion of a set of unlabeled data acquired by the model training equipment. High-value unlabeled data can also be understood as data that the model training equipment determines has significant value for improving the performance of the model to be trained.

[0150] The model to be trained can be an initial model or a model obtained based on the initial model.

[0151] There are many methods for model training devices to identify high-value unlabeled data. For example, when a model is being trained to identify whether an animal in a picture is a cat or a dog, the features in the training data are the pictures, and the labels are the animal categories. Model training devices can use methods such as... Figure 4 The process shown identifies multiple high-value unlabeled data points from a set of unlabeled data (multiple images).

[0152] Taking any image from a set of unlabeled data as an example, such as Figure 4 As shown, the model training device can input this image into the model to be trained. The model analyzes and processes the image and outputs the confidence scores for the two animal categories. The model training device can normalize these confidence scores using a softmax function to obtain two normalized confidence scores. One confidence score represents the probability that the animal in the image is a cat; the other represents the probability that the animal in the image is a dog. If there is a significant difference between the two confidence scores, for example, a confidence score of 0.9 for cats and 0.1 for dogs, it indicates that the model can clearly distinguish the animal in the image. In other words, this image does not offer much value for improving the model's performance, and the model training device does not need to use this image and its corresponding label to train the model. Therefore, the data provider does not need to label the image, and the model training device can determine that the image is not high-value unlabeled data.

[0153] Conversely, if there is no significant difference between the two confidence levels—for example, a confidence level of 0.55 for cats and 0.45 for dogs—it indicates that the model to be trained cannot clearly distinguish the animals in the image. In other words, this image is valuable for improving the performance of the model, and the training equipment needs to use this image and its corresponding label to train the model. Therefore, the data provider needs to label the image, and the training equipment can then determine that the image is high-value unlabeled data.

[0154] By iterating through each data point in a set of unlabeled data in the manner described above, the model training device can identify high-value unlabeled data points from that set of unlabeled data.

[0155] S303, The model training device sends Request 1 to the data providing device, Request 1 being used to request labels corresponding to high-value unlabeled data. Correspondingly, the data providing device receives Request 1 from the model training device.

[0156] It should be understood that the labels corresponding to high-value unlabeled data can be understood as labels obtained by annotating high-value unlabeled data.

[0157] S304. The data providing device sends the labels corresponding to the high-value unlabeled data to the model training device. Correspondingly, the model training device receives the labels corresponding to the high-value unlabeled data from the data providing device.

[0158] It should be understood that prior to S304, the data providing device could measure the label corresponding to each data point in a set of unlabeled data, and could send the labels corresponding to high-value unlabeled data to the model training device based on request 1, thus consuming less transmission resources for label transmission. Alternatively, after S303 and before S304, the data providing device could label the high-value unlabeled data based on request 1, and the data providing device could choose not to label the remaining unlabeled data, thus reducing the power consumption of the data providing device.

[0159] In this way, the model training device can obtain training data, which is labeled data, and includes high-value unlabeled data and the labels corresponding to the high-value unlabeled data.

[0160] S305. The model training device uses training data to train the model to be trained, and obtains a new model.

[0161] Alternatively, if the new model does not meet the conditions, that is, the model to be trained is not trained, S302 to S305 can be executed repeatedly until the model training device obtains a model that meets the conditions.

[0162] It should be noted that when S302 to S305 are executed repeatedly, the new labels obtained by the model training device from the data providing device may be the same as or different from the labels corresponding to the previously obtained high-value unlabeled data. This application does not impose specific restrictions on this.

[0163] However, in some cases, in order for the model training device to obtain labels from the data provider device multiple times and in a timely manner, the data provider device will continuously store the collected labels. For data provider devices with small storage capacity, this may occupy a large amount of storage space, resulting in other data not being able to be stored, and thus affecting the data provider device's ability to process other business.

[0164] Alternatively, in some cases, the data providing device may delete collected labels due to insufficient storage space or high load. In this situation, if the model training device has not yet completed its model training and requests labels from the data providing device again, the data providing device may not be able to send the labels in a timely manner, thus affecting the model training progress. For example, in method 300, if the model training device has not completed its training and repeatedly executes S302 to S305 to request labels from the data providing device again, and the data providing device has already deleted the collected labels, it may be unable to provide the requested labels, affecting model training; or, the data providing device may need to remeasure the requested labels, resulting in a significant delay in sending labels from the data providing device to the model training device.

[0165] In other words, the data providing device cannot determine whether to delete labels, which may affect the normal training of the model by the model training device, or may affect the performance of the data providing device.

[0166] In view of this, this application provides a data transmission method in which a model training device can instruct a data providing device to delete a first part of the labels and / or save a second part of the labels via instruction information. Thus, the data providing device can determine whether to delete the first part of the labels and / or save the second part of the labels based on the instructions from the model training device. On the one hand, when the data providing device determines that the first part of the labels can be deleted, it can delete the first part of the labels to free up some memory resources; and / or, when it determines that the second part of the labels can be saved, it can also delete labels other than the second part of the labels to free up some memory resources. This helps the data providing device to delete the labels that the model training device indicates can be deleted in a timely manner, thereby freeing up some storage space on the data providing device, contributing to better performance of the data providing device, and enabling the data providing device to handle other tasks normally.

[0167] On the other hand, it enables the data providing device to save labels that the model training device has not instructed to be deleted, allowing the model training device to retrieve the saved labels from the data providing device in a timely manner before training the model, which helps the model training device to train the model efficiently.

[0168] Below, in conjunction with Figure 5 and Figure 6This application provides a detailed description of the data transmission method. The embodiments shown in this application illustrate the data transmission method provided by this application from the perspective of device interaction. The specific forms and quantities of the devices shown are merely examples and should not constitute any limitation on the implementation of the method provided by this application. Below, taking a model training device and a data providing device as the main implementers, the data transmission method of the embodiments of this application will be described in detail.

[0169] It should be understood that the data providing device can be referred to as the first communication device, and can be a terminal device, access network device, or core network device. The model training device can be referred to as the second communication device, and can also be a terminal device, access network device, or core network device. Furthermore, the model training device and the data providing device can be the device (terminal device, access network device, or core network device) itself, or a chip, chip system, or processor that supports the device in implementing data transmission methods, or a logic module or software that can implement all or part of the device; this application does not specifically limit this.

[0170] For example, the model training device can be a network management system, and the data providing device can be a base station, with the network management system collecting training data from the base station; the model training device can be a core network device, and the data providing device can be a terminal device, with the core network device collecting training data from the terminal device; the model training device can be a core network device, and the data providing device can be a base station, with the core network device collecting training data from the base station; the model training device is base station 1, and the data providing device is base station 2, with base station 1 collecting training data from base station 2; the model training device is a network element responsible for data collection in the access network device, and the data providing device is a terminal device, with the network element responsible for data collection in the access network device collecting training data from the terminal device; the model training device is a network element responsible for data collection in the access network device, and the data providing device is a base station, with the network element responsible for data collection in the access network device collecting training data from the base station; the model training device is an intelligent controller (RAN intelligent controller, RIC), and the data providing device is a terminal device, with the RIC collecting training data from the terminal device; or, the model training device is an RIC, and the data providing device is a base station, with the RIC collecting training data from the base station, etc.

[0171] Figure 5 This is a flowchart illustrating a data transmission method 500 provided in an embodiment of this application. Method 500 is applicable to a communication system 1000 and includes the following steps:

[0172] S501. The model training device acquires multiple features. For example, the data providing device sends multiple features to the model training device; correspondingly, the model training device receives multiple features from the data providing device.

[0173] Among them, many features are similar to a set of unlabeled data mentioned above. The model training device can identify high-value unlabeled data from multiple features, such as the features corresponding to the labels in the third part below, and thus obtain the labels corresponding to the high-value unlabeled data from the data providing device.

[0174] It is understandable that model training devices can train models using some or all of the training data. The features included in the training data can be some or all of multiple features, and the labels included in the training data are the labels corresponding to some or all of these features. These labels can be collected by the data providing device; that is, the data providing device can label some or all of the multiple features to obtain the labels corresponding to those features.

[0175] It should be understood that multiple features can also be referred to as a set of unlabeled data, multiple unlabeled data, feature set, or unlabeled data set, etc. This application does not specifically limit the name of multiple features.

[0176] S502, the model training device sends a first instruction to the data providing device. The first instruction includes instructions to delete a first portion of the labels from multiple labels in the training data and / or to save a second portion of the labels from multiple labels. The multiple labels include labels corresponding to some or all of the features in a set of features. The training data is used for model training. Correspondingly, the data providing device receives the first instruction from the model training device.

[0177] Here, "multiple tags" can refer to tags collected by the data providing device based on multiple features, and the multiple tags include tags corresponding to each feature among some or all of the multiple features. In one possible implementation, the data providing device can collect tags corresponding to each feature among the multiple features, and the multiple tags include tags corresponding to all features among the multiple features.

[0178] In another possible implementation, the data providing device can collect labels corresponding to some of the features among multiple features, and the multiple labels include the label corresponding to each feature among the partial features.

[0179] The first part of the label can be some or all of the multiple labels, and the second part of the label can also be some or all of the multiple labels, and the first indication information can be one of the following three cases.

[0180] Case 1: If the first instruction information includes information about deleting the first part of the tags, the data providing device can determine that some or all of the tags can be deleted. If the first part of the tags is a subset of multiple tags, deleting the first part of the tags can also be understood as saving the tags other than the first part of the tags among the multiple tags.

[0181] Case 2: If the first instruction information includes information about saving the second part of the tags, the data providing device can determine that some or all of the tags can be saved, and if the second part of the tags is a subset of multiple tags, saving the second part of the tags can also be understood as deleting the tags other than the second part of the tags from the multiple tags.

[0182] Scenario 3: If the first instruction includes instructions on deleting the first part of the label and instructions on saving the second part of the label, the first part of the label and the second part of the label do not contain the same label. That is, if the first part of the label includes a label, such as label x, the second part of the label does not include label x.

[0183] S503, the data providing device sends the third part of a plurality of labels to the model training device. Correspondingly, the model training device receives the third part of the labels from the data providing device.

[0184] The third part of the label can be some or all of the labels from a set of multiple labels. Multiple features include the features corresponding to each label in the third part of the label. The training data acquired by the model training device includes: the third part of the label and the features corresponding to each label in the third part of the label.

[0185] In this way, the model training device can train the model based on the acquired training data. For example, after S503, method 500 may include: the model training device training the model based on the third part of the labels and the features corresponding to each label in the third part of the labels.

[0186] The third part of the label can be understood as the label corresponding to high-value unlabeled data; that is, the feature corresponding to the third part of the label can be understood as high-value unlabeled data. Furthermore, the high-value unlabeled data can be determined by the model training device from multiple features. The third part of the label may satisfy the following relationship with the first or second part of the label.

[0187] The third part of the labels can be the same as the first part of the labels. In this case, the first part of the labels can be understood as the labels that have been fed back from the multiple labels, that is, the labels that the data providing device has sent to the model training device. Alternatively, the third part of the labels can also be partially different from the first part of the labels. In this case, the first part of the labels can include some or all of the labels that have been fed back and some or all of the labels that have not been fed back. Alternatively, the third part of the labels can also be completely different from the first part of the labels. In this case, the first part of the labels can also be understood as the labels that have not been fed back from the multiple labels, that is, the labels that the data providing device has not sent to the model training device.

[0188] Similarly, the third part of the label can be the same as the second part of the label. In this case, the second part of the label can be understood as the labels that have been fed back from the multiple labels, that is, the labels that the data providing device has sent to the model training device. Alternatively, the third part of the label can also be partially different from the second part of the label. In this case, the second part of the label can include some or all of the fed-back labels and some or all of the unfed labels from the multiple labels. Alternatively, the third part of the label can also be completely different from the second part of the label. In this case, the second part of the label can also be understood as the unfed labels from the multiple labels, that is, the labels that the data providing device has not sent to the model training device.

[0189] It should be understood that S502 can be executed before S503. In this case, the first part of the tags or the second part of the tags can refer to all tags, tags that have been responded to, or tags that have not been responded to, so that the data providing device can subsequently delete and / or save tags of the corresponding type based on the first instruction information. Alternatively, S502 can also be executed after S503. In this case, before the data providing device receives the first instruction information, the data providing device can save all tags first, and can delete and / or save tags of the corresponding type based on the first instruction information after receiving the first instruction information. This application does not specifically limit this.

[0190] It should be noted that in method 500, S501 and S503 are optional. For example, if the model training device can store multiple features, then method 500 may not include S501; and / or, if the event of the data providing device sending the third part of the label to the model training device may not be triggered, then method 500 may not include S503.

[0191] The data transmission method of this application allows the data providing device to determine whether to delete a first part of the labels and / or whether to save a second part of the labels based on instructions from the model training device. When it is determined that the first part of the labels can be deleted based on the first instruction information, the data providing device can delete the first part of the labels. When it is determined that the second part of the labels can be saved, and the second part of the labels is a subset of multiple labels, the data providing device can delete the labels other than the second part of the labels. This allows the data providing device to release some memory resources, which helps to improve the performance of the data providing device and allows it to handle other tasks normally.

[0192] In addition, if the data providing device determines to save the second part of the labels, it can save the second part of the labels; or, if it determines to delete the first part of the labels, and the first part of the labels is only a portion of multiple labels, it can save the labels other than the first part of the labels. This makes it easier for the subsequent model training device to obtain the saved labels from the data providing device, and the data providing device can provide the labels to the model training device in a timely manner, which helps the model training device to train the model in a timely manner.

[0193] For example, after S502, method 500 may further include: the data providing device determining, based on the first indication information, whether to delete the first part of the labels and / or whether to save the second part of the labels. If the data providing device determines to delete the first part of the labels, the first part of the labels can be deleted to release some memory resources; if the data providing device determines to save the second part of the labels, the second part of the labels can be saved so that the model training device can subsequently obtain some or all of the labels in the second part of the labels from the data providing device and continue training the model based on some or all of the labels in the second part of the labels.

[0194] It should be understood that the data providing device can delete the first part of the labels and / or save the second part of the labels after S503, so that the data providing device can determine whether to delete and / or save the labels after sending the labels to the model training device. This reduces the possibility that the data providing device has already deleted some or all of the labels in the third part of the labels before S503, allowing the data providing device to send the third part of the labels to the model training device in a timely manner.

[0195] Below, we will provide a detailed explanation of how the first instruction information in cases 1 to 3 above indicates the deletion of the first part of the label and / or the saving of the second part of the label.

[0196] Case 1: The first instruction information includes instructions regarding the deletion of a first portion of the labels from multiple labels in the training data.

[0197] The first instruction information may include information related to the information indicating the deletion of the first part of the label, so that the data providing device can determine whether to delete the first part of the label based on the first instruction information.

[0198] In one possible implementation, the first set of labels may include any of the following: all labels from a plurality of labels; or labels from a plurality of labels that the data providing device has sent to the model training device; or labels from a plurality of labels that the data providing device has not sent to the model training device.

[0199] It should be understood that "all labels" in multiple labels can refer to all labels collected by the data providing device, and the features corresponding to these labels belong to multiple features. Labels that the data providing device has sent to the model training device in multiple labels can also be called feedback data, feedback labels, labels received by the model training device, or labels acquired by the model training device, etc. Labels that the data providing device has not sent to the model training device in multiple labels can also be called non-feedback data, non-feedback labels, labels not received by the model training device, or labels not acquired by the model training device, etc. This application does not specifically limit the names of the labels that the data providing device has sent to the model training device in multiple labels or the names of the labels that the data providing device has not sent to the model training device in multiple labels.

[0200] When the first part of the label includes labels that the data providing device has already sent to the model training device, such as the first part of the label including a third part of the label, the first indication information indicating deletion of the first part of the label can be replaced, for example, by the first indication information indicating that the first part of the label has been successfully received, so that the data providing device can determine that the first part of the label can be deleted based on the first indication information. For example, when the first part of the label includes the same third part of the label, S502 can be executed after S503, and the first indication information can be, for example, an acknowledgment (ACK) message, and the ACK message indicates that the model training device has successfully received the third part of the label, so that the data providing device can determine that the third part of the label can be deleted based on the ACK message.

[0201] Alternatively, if the first part of the label includes labels that the data providing device did not send to the model training device among multiple labels, the first instruction information indicating the deletion of the first part of the label can be replaced, for example, by: the first instruction information indicating that the first part of the label was not successfully received, then the data providing device can determine that the first part of the label can be deleted based on the first instruction information.

[0202] In case 1, the first instruction information may include, for example, information indicating the first part of the label and / or information indicating the deletion of the first part of the label.

[0203] It should be understood that the first indication information may include information indicating the first part of the label, so that the data providing device can determine the first part of the label based on the information indicating the first part of the label; and the field used to carry the information indicating the first part of the label may be predefined as a field carrying information indicating the label to be deleted, so that the data providing device can determine to delete the first part of the label based on the information indicating the first part of the label.

[0204] Alternatively, the first indication information may include information indicating the deletion of a first portion of tags. Here, "first portion of tags" can refer broadly to tags that have received feedback, tags that have not received feedback, or all tags. The data providing device can determine whether to delete tags based on this information. Furthermore, the field carrying the information indicating the deletion of the first portion of tags can be predefined as a field indicating whether to delete tags that have received feedback, tags that have not received feedback, or all tags. In this case, the data providing device can determine whether to delete tags that have received feedback, tags that have not received feedback, or all tags based on this information. For example, the information indicating the deletion of the first portion of tags can be 1 bit, such as 0 or 1. The field carrying this information can be predefined as a field indicating the deletion of tags that have received feedback. The data providing device can determine whether to delete tags that have received feedback based on the 1 or 0 carried in this field; these tags are the first portion of tags.

[0205] Alternatively, the first instruction information may also include information indicating the first part of the label and information indicating the deletion of the first part of the label. In this case, the data providing device can determine the first part of the label based on the information indicating the first part of the label, and can determine the deletion of the first part of the label based on the information indicating the deletion of the first part of the label.

[0206] The information indicating the first part of the labels may include, for example, the index of each label in the first part of the labels. The index can be an integer, such as an integer greater than or equal to 0.

[0207] It should be noted that, in the embodiments of this application, the data types of features and labels can be determined based on the model to be trained, and the concepts of a feature and a label may also differ depending on the model to be trained. For example, when the model to be trained is used to locate a terminal device, a feature can be, for example, a channel impulse response (CIR), and a label can be location information, such as a coordinate (x, y). Alternatively, when the model to be trained is used to predict the complete channel matrix, that is, the model is used to predict the complete channel matrix based on known channel vectors, a feature can be M channel vectors, and a label can be a channel matrix, where M is a positive integer. Here, a channel vector is typically a series of numerical values, represented as a one-dimensional array or vector, containing channel gain and phase information, and the elements in the channel vector are typically complex numbers. The channel matrix is ​​typically a two-dimensional matrix, and the dimension of this two-dimensional matrix can be determined based on the number of transmitting antennas and the number of receiving antennas.

[0208] In possible implementation 1, the index of the first part of the tags can be some or all of the indices of multiple tags. For example, the data providing device can arrange multiple tags sequentially, and the indices of the arranged tags can increase or decrease sequentially from 0 or other positive integers. Then, the data providing device can determine the first part of the tags from the multiple tags based on first indication information. For example, assuming the indices of the multiple tags are integers that increase sequentially starting from 0, and the first indication information indicates 0, 7, and 9, then the first part of the tags can include the 1st tag, the 8th tag, and the 10th tag from the arranged tags.

[0209] In a possible implementation 2, the index of the first part of the tag can be part or all of the index of the third part of the tag. Exemplarily, the data providing device can arrange the third part of the tags sequentially, with the indices of the arranged third part of the tags increasing or decreasing sequentially from 0 or other positive integers. For example, assuming the indices of the third part of the tags are integers that increase sequentially starting from 0, and the first indication information indicates 1, 7, and 9, then the first part of the tags can include the second, eighth, and tenth tags in the arranged third part of the tags.

[0210] In possible implementation 3, the index of the first part of the label can also be understood as the index of the feature corresponding to the first part of the label. The index of the feature corresponding to the first part of the label is part or all of the indices of multiple features. For example, multiple features are arranged sequentially according to the order in which the data providing device sends them to the model training device. The indices of the multiple features after the arrangement can be integers that increase or decrease sequentially starting from 0 or other integers greater than 0, such as 0, 1, 2, etc. The information indicating the first part of the label can include part or all of the indices of the multiple features after the arrangement. Then, the first part of the label is the label corresponding to the feature indicated by the part or all of the indices. For example, if the indices of the multiple features are integers that increase starting from 0, and the information indicating the first part of the label indicates 0, 3, and 7, then the first part of the label includes the label corresponding to the 1st feature, the label corresponding to the 4th feature, and the label corresponding to the 8th feature in the multiple features after the arrangement.

[0211] The information indicating the first part of the label can also be, for example, a first bitmap.

[0212] In possible implementation 4, the number of bits in the first bitmap can be the number of tags included in the multiple tags, and each bit in the first bitmap indicates one tag among the multiple tags. The tags indicated by the bits in the first bitmap that are 1 (or 0) can be the first part of the tags among the multiple tags. For example, if the first bitmap is 11010001, then the first part of the tags can include the four tags indicated by the four "1" bits.

[0213] In a possible implementation 5, the number of bits in the first bitmap can be the number of tags included in the third part of the tag, and each bit in the first bitmap indicates one tag in the third part of the tag. The tags indicated by the bits in the first bitmap that are 1 (or 0) can be the first part of the tags in the third part of the tag. For example, if the first bitmap is 10010001, then the first part of the tag can include three tags indicated by three "1" bits.

[0214] In a possible implementation 6, the number of bits in the first bitmap can be the number of features included in a plurality of features, where each bit is used to indicate one feature among the plurality of features. The feature indicated by a bit in the first bitmap that is 1 (or 0) can be the feature corresponding to the first part label among the plurality of features. Therefore, the data providing device can determine the feature corresponding to the first part label based on the first bitmap, and thus determine the first part label. For example, if the first bitmap is 10010001, the feature corresponding to the first part label can include three features indicated by three "1"s. The data providing device can then determine the three labels corresponding to these three features, and these three labels are the first part label.

[0215] The message instructing the deletion of the first tag can be a Boolean type, for example. A true value indicates deletion of the first tag; a false value indicates no deletion. Alternatively, a false value indicates deletion; a true value indicates no deletion. The message instructing the deletion of the first tag can also be a 1-bit value. A 1 (or 0) value indicates deletion; a 0 (or 1) value indicates no deletion. Instructing no deletion can also be replaced with an instruction to save the first tag, etc.

[0216] Case 2: The first instruction information includes information about saving a second portion of the labels from the multiple labels in the training data.

[0217] The first instruction information may include information related to the instruction to save the second part of the tag, so that the data providing device can determine whether to save the second part of the tag based on the first instruction information.

[0218] In one possible implementation, the second set of labels may include any of the following: all the labels among a plurality of labels; or, labels among the plurality of labels that the data providing device has sent to the model training device; or, labels among the plurality of labels that the data providing device has not sent to the model training device.

[0219] It should be understood that the descriptions of all labels, labels sent to the model training device, and labels not sent to the model training device can be found in Case 1, and will not be repeated here.

[0220] When the second part of the labels includes labels that the data providing device did not send to the model training device, the first indication information indicating that the second part of the labels should be saved can be replaced, for example, by stating that the first indication information indicates that the second part of the labels was not successfully received, in which case the data providing device can determine that the second part of the labels can be saved based on the first indication information. Alternatively, when the second part of the labels includes labels that the data providing device has sent to the model training device, the first indication information indicating that the second part of the labels should be saved can be replaced, for example, by stating that the first indication information indicates that the second part of the labels was successfully received, in which case the data providing device can determine that the second part of the labels can be saved based on the first indication information.

[0221] In case 2, the first instruction information may include, for example, information indicating the second part of the label and / or information indicating the storage of the second part of the label.

[0222] It should be understood that the first indication information may include information indicating the second part of the label, so that the data providing device can determine the second part of the label based on the information indicating the second part of the label; and the field used to carry the information indicating the second part of the label may be predefined as a field carrying information indicating the label to be saved, so that the data providing device can determine to save the second part of the label based on the information indicating the second part of the label.

[0223] Alternatively, the first indication information may include information indicating the saving of the second part of the tags. Here, the second part of the tags can refer to tags that have received feedback, tags that have not received feedback, or all tags. The data providing device can determine whether to save tags based on this information. Furthermore, the field carrying the information indicating the saving of the second part of the tags can be predefined as a field indicating whether to save tags that have received feedback, tags that have not received feedback, or all tags. In this case, the data providing device can determine whether to save tags that have received feedback, tags that have not received feedback, or all tags based on this information. For example, the information indicating the saving of the second part of the tags can be 1 bit, such as 0 or 1. The field carrying this information can be predefined as a field indicating the deletion of tags that have not received feedback. The data providing device can determine whether to save tags that have not received feedback based on the 1 or 0 carried in this field; these tags are the second part of the tags.

[0224] Alternatively, the first instruction information may also include information indicating the second part of the tag and information indicating the saving of the second part of the tag. In this case, the data providing device can determine the second part of the tag based on the information indicating the second part of the tag, and can determine the saving of the second part of the tag based on the information indicating the saving of the second part of the tag.

[0225] Similar to the information indicating the first part of the label, the information indicating the second part of the label can also be an index or a bitmap.

[0226] Information indicating the second part of the labels may include, for example, the index of each label in the second part of the labels.

[0227] In possible implementation 7, the index of the second part of the label is part or all of the indexes of multiple labels.

[0228] In possible implementation 8, the index of the second part of the label can be part or all of the index of the third part of the label.

[0229] In possible implementation 9, the index of the second part of the label can be understood as the index of the feature corresponding to the second part of the label, and the index of the feature corresponding to the second part of the label is part or all of the indexes of multiple features.

[0230] It should be understood that possible implementation 7 is similar to possible implementation 1 above, possible implementation 8 is similar to possible implementation 2 above, and possible implementation 9 is similar to possible implementation 3 above. Please refer to the description above, and it will not be repeated here.

[0231] Similar to how the information indicating the first part of the label can be a first bitmap, the information indicating the second part of the label can also be, for example, a second bitmap.

[0232] In a possible implementation 10, the number of bits in the second bitmap can be the number of tags included in a plurality of tags, and each bit in the second bitmap indicates one tag among the plurality of tags. The tags indicated by bits in the second bitmap that are 1 (or 0) can be the second part of the tags among the plurality of tags.

[0233] In possible implementation 11, the number of bits in the second bitmap can be the number of tags included in the third part of the tag, and each bit in the second bitmap indicates one tag in the third part of the tag. The tags indicated by the bits in the second bitmap that are 1 (or 0) can be the second part of the tag in the third part of the tag.

[0234] In possible implementation 12, the number of bits in the second bitmap can be the number of features included in a plurality of features, where each bit is used to indicate one feature among the plurality of features. The feature indicated by the bit in the second bitmap that takes a value of 1 (or 0) can be the feature corresponding to the second part label among the plurality of features. Then, the data providing device can determine the feature corresponding to the second part label based on the second bitmap, and thus determine the second part label.

[0235] It should be understood that possible implementation 10 is similar to possible implementation 4 above, possible implementation 11 is similar to possible implementation 5 above, and possible implementation 12 is similar to possible implementation 12 above. Please refer to the description above, and it will not be repeated here.

[0236] The information indicating whether to save the second part of the tag can be a Boolean type, for example. A true (or false) value indicates that the second part of the tag should be saved; conversely, a false (or true) value indicates that the second part of the tag should not be saved. The information indicating whether to save the second part of the tag can also be 1 bit, for example. A 1 (or 0) value indicates that the second part of the tag should be saved; a 0 (or 1) value indicates that the second part of the tag should not be saved. Indicating not to save the second part of the tag can also be replaced with indicating deletion of the second part of the tag, etc.

[0237] Case 3: The first instruction information includes instructions on deleting the first part of the labels from multiple labels in the training data and information on saving the second part of the labels from multiple labels.

[0238] The descriptions of the first and second part tags can be found in Case 1 and Case 2. The difference is that in Case 3, the first and second part tags do not contain the same tag; that is, if the first part tag includes a tag, then the second part tag does not include that tag.

[0239] Furthermore, the first instruction information may include, for example, information indicating the first part of the label and / or information indicating the deletion of the first part of the label; and information indicating the second part of the label and / or information indicating the saving of the second part of the label.

[0240] It should be understood that the instructions regarding the information for the first part of the label and the instructions for deleting the first part of the label can be found in the description of Case 1, and the instructions regarding the information for the second part of the label and the instructions for saving the second part of the label can be found in the description of Case 2, which will not be repeated here.

[0241] Based on the first instruction information, the data providing device can determine whether to delete the first part of the tags and / or whether to save the second part of the tags. It can be understood that if the first part of the tags consists of only a portion of multiple tags, deleting the first part of the tags can also be understood as saving the tags other than those in the first part. Therefore, if the first part of the tags consists of only a portion of multiple tags or the second part of the tags is not empty, the data providing device can also determine the duration or time period for saving the tags using the following two methods.

[0242] Method 1: The model training device instructs the data providing device to save the labels for a specified duration or time period.

[0243] In one possible implementation, method 500 further includes: the model training device sending second instruction information to the data providing device, the second instruction information including information indicating a first duration or information indicating a first time period, the first duration being used to indicate the duration for which the data providing device saves the tags, and the first time period being used to indicate the time period for which the data providing device saves the tags.

[0244] Among them, the tags in the saved tags can be tags that have been responded to, tags that have not been responded to, or all tags, and the saved tags can be tags other than the first part of the tags, or tags from the second part.

[0245] The first duration can be understood as the length of time the data providing device retains the tag. For example, the first duration could be 10 minutes. The first duration can also be understood as the minimum duration for which the data providing device retains the tag. The start time (or start position) of the first duration can be the moment the data providing device receives the second instruction information, so that the data providing device can determine whether to retain the tag within the first duration after receiving the second instruction information.

[0246] The first time period can be understood as a range of time during which the data providing device stores the tag, such as from one specified time to another. Information indicating the first time period may include at least two of the following: start time 1, storage duration 1, and end time 1, where start time 1 is the start time of the first time period, end time 1 is the end time of the first time period, and storage duration 1 is the length of the first time period. The data providing device can then determine the first time period based on at least two of these information.

[0247] Alternatively, the information indicating the first time period may include, for example, an end time 1. In this case, the first time period could be the time from when the second indication information is received from the data providing device to the end time 1. That is, the data providing device can determine to save the tag until the end time 1 based on the information indicating the first time period. For example, if the information indicating the first time period is 10:20, the data providing device can determine to save the tag until 10:20.

[0248] In this way, based on the second instruction information, the data providing device can determine that there are tags that need to be saved, and can determine the duration or time period for which the tags need to be saved.

[0249] Method 2: The data providing device instructs the model training device on the duration or time period for saving the labels.

[0250] In one possible implementation, method 500 further includes: the data providing device sending third instruction information to the model training device, the third instruction information including information indicating a second duration or information indicating a second time period, the second duration being used to indicate the duration for which the data providing device saves the tags, and the second time period being used to indicate the time period for which the data providing device saves the tags.

[0251] Similar to method 1, the tags in the saved tags can be tags that have received feedback, tags that have not received feedback, or all tags. The saved tags can be tags other than the first part of the tags, or tags from the second part.

[0252] The second duration can be understood as the length of time the data providing device retains the tags. Furthermore, the second duration can also be understood as an indication of the maximum duration for which the data providing device can retain tags; that is, the data providing device uses the second duration to indicate to the model training device that it can retain tags for a maximum of the second duration. The start time (or starting position) of the second duration can be the moment the model training device receives the third instruction information, allowing the model training device to determine that the data providing device may delete tags after the second duration following the receipt of the third instruction information.

[0253] The information indicating the second time period may include at least two of the following: start time 2, storage duration 2, and end time 2, where start time 2 is the start time of the second time period, end time 2 is the end time of the second time period, and storage duration 2 is the length of the second time period. The data providing device can then determine the second time period based on at least two of these items.

[0254] Alternatively, the information indicating the second time period could include, for example, an end time of 2. In this case, the second time period could be the time from when the third indication information is sent from the data providing device to the end time of 2. That is, the model training device can determine, based on the information indicating the second time period, that the data providing device might delete the label after the end time of 2. For example, if the information indicating the second time period indicates 10:20, the model training device can determine that the data providing device might delete the label after 10:20.

[0255] In this way, based on the third instruction information, the model training device can determine the duration or time period during which the data providing device can store the labels, so that the model training device can obtain the labels required for model training within that duration or time period, and thus enable the model training device to train the model based on the labels in a timely manner.

[0256] Furthermore, the third instruction information can be sent to the model training device before S503, enabling the model training device to determine the third part of the labels based on the second duration or the second time period. For example, if the second duration or the corresponding duration of the second time period is short, the number of labels included in the third part of the labels obtained by the model training device from the data provider can be larger. This allows the model training device to obtain relatively more labels before the data provider deletes them, thus enabling the model training device to train the model in a timely manner. Moreover, if the third part of the labels includes a larger number of labels, the probability of the model training device successfully training the model based on the third part of the labels is also higher, reducing the number of times the model training device needs to obtain labels from the data provider. This reduces the possibility of the model training device being unable to obtain labels from the data provider in a timely manner after the data provider has deleted them.

[0257] It is understood that the data providing device can send third instruction information to the model training device before S503, so that the model training device can determine the time and number of third part labels to be acquired. Furthermore, after S503, the data providing device can further instruct the model training device on the duration or time period for storing unreturned labels, as described in method 3 below.

[0258] It should be noted that the embodiments in Method 3 can be implemented alone; or Method 3 and Method 2 can be implemented in combination; or Method 3 can be implemented in combination with Method 2. This application does not make specific limitations in this regard.

[0259] Method 3: The data providing device instructs the model training device to save the duration or time period for which no feedback labels are provided.

[0260] In one possible implementation, method 500 further includes: the data providing device sending fourth indication information to the model training device, the fourth indication information including information indicating a third duration or information indicating a third time period, the third duration being used to indicate the duration for which the data providing device stores the fourth part of the labels, the third time period being used to indicate the time period for which the data providing device stores the fourth part of the labels, and the fourth part of the labels being labels other than the third part of the labels among a plurality of labels.

[0261] The fourth part of the label can also be called unfeedback data or unfeedback label, etc.

[0262] Similar to method 2, the data providing device instructs the model training device via a third duration that it may save the fourth part of the label for a third duration, or for a maximum of a third duration. The start time (or start position) of the third duration may be the moment when the model training device receives the fourth instruction information, so that the model training device can determine that the data providing device may delete the fourth part of the label after the third duration following the receipt of the fourth instruction information.

[0263] The information indicating the third time period may include at least two of the following: start time 3, storage duration 3, and end time 3, where start time 3 is the start time of the third time period, end time 3 is the end time of the third time period, and storage duration 3 is the length of the third time period. The data providing device can then determine the third time period based on at least two of these items. Alternatively, the information indicating the third time period may include end time 3, in which case the second time period could be the time from when the fourth indication information is sent from the data providing device to end time 3.

[0264] In this way, the model training device can determine the duration or time period for which the data provider retains the fourth part of the labels based on the fourth indication information. If the model training device fails to train the model based on the third part of the labels, it can retrieve the labels from the fourth part of the labels from the data provider before the data provider deletes them, thus helping the model training device to train the model in a timely manner.

[0265] It is understood that, in order to enable the model training device to obtain labels corresponding to some features from multiple features from the data providing device, thereby reducing the time-frequency domain resources consumed by label transmission; and / or, to enable the data providing device to collect labels corresponding to some features from multiple features, thereby reducing the power consumption of the data providing device, the data providing device will sequentially send features (e.g., multiple features in S501) to the model training device, and then send labels (e.g., the third part of the labels in S503) to the model training device. That is, the model training data obtains features and labels sequentially from the data providing device. The implementation method of S501, i.e., the method by which the model training device obtains multiple features, will be described in detail below.

[0266] In cases where multiple features are sent from the data providing device to the model training device, these multiple features can be data acquired by the data providing device, such as data measured or collected by the data providing device. Furthermore, the data providing device can also send multiple features to the model training device based on a request from the model training device.

[0267] In one possible implementation, prior to S501, method 500 further includes: the model training device sending request 2 to the data providing device, request 2 being used to request multiple features; correspondingly, the data providing device receiving request 2 from the model training device.

[0268] In this way, the data providing device can send multiple features to the model training device in a timely manner based on the needs of the model training device.

[0269] It should be understood that, prior to S501, model training devices could not determine exactly which data constituted multiple features. Therefore, Request 2, used to request multiple features, can be generic; Request 2 could be used to request features or measurement data, etc.

[0270] In addition, the model training device can also instruct the data provider to specify the conditions that the data to be fed back must meet.

[0271] In one possible implementation, method 500 may further include: the model training device sending fifth indication information to the data providing device, the fifth indication information indicating one or more of the following: the data type of multiple features, the data type of the labels corresponding to the multiple features, the number of multiple features, or the time period during which the data providing device collects (or measures, etc.) multiple features. Correspondingly, the data providing device receives the fifth indication information from the model training device.

[0272] Here, the labels corresponding to multiple features are the tags obtained by annotating multiple features. Data types can also be referred to as categories, etc. For example, in a channel measurement scenario, the model to be trained acquired by the model training device can be used to obtain initial channel state information, etc. Then, the data type of multiple features can be a channel vector, and the data type of the labels corresponding to multiple features can be the initial channel matrix (also called the original channel matrix), etc. Therefore, the data types of multiple features and the data types of the labels corresponding to multiple features can be represented, for example, as <channel vector, initial channel matrix>.

[0273] It should be understood that when the model training device sends the fifth instruction information to the data providing device, the model training device may not have yet received multiple features, and therefore cannot determine which specific data these multiple features are. Thus, the aforementioned multiple features can be used generically, and can be replaced with features, measurement data, or data, etc.

[0274] To indicate to the data providing device the type of data to be acquired, the fifth indication information can also be replaced with information indicating the type (or function) of the model to be trained. For example, the fifth indication information could indicate that the model to be trained is used to acquire initial channel state information, etc. In this way, the data providing device can also determine the type of data to be sent to the model training device based on the fifth indication information.

[0275] The number of features can be a positive integer, such as 5000. Based on the number of features, the data providing device can determine the data of the features to be measured. The time period for collecting multiple features can be indicated by any two of the following: the start time of collecting multiple features, the end time of collecting multiple features, or the duration of collecting multiple features. The data providing device can then determine the time period for collecting multiple features based on any two of these. Alternatively, the time period for collecting multiple features can be indicated by a duration of 1. Duration 1 can represent the time period for collecting multiple features as: within a duration of 1 after the data providing device receives the fifth indication information, or within a duration of 1 before the data providing device receives the fifth indication information, etc.

[0276] It should be understood that the fifth indication information may be carried in request 2 or sent via signaling different from request 2. Furthermore, the fifth indication information may include one or more of the following: indication information 1 for indicating the data type of multiple features, indication information 2 for indicating the data type of the tags corresponding to the multiple features, indication information 3 for indicating the quantity of multiple features, or indication information 4 for indicating the time period for collecting multiple features. Alternatively, indication information 1 and indication information 2 may also be a single piece of information. When the fifth indication information includes multiple indication information, these multiple indication information may also be sent via the same or different signaling. This application does not specifically limit this.

[0277] In addition, the model training device can instruct the data providing device to first provide features.

[0278] In one possible implementation, prior to S501, method 500 may further include: the model training device sending information to the data providing device for indicating feedback features and / or non-feedback labels. Correspondingly, the data providing device receives information from the model training device for indicating feedback features and / or non-feedback labels.

[0279] In this way, the data providing device can determine, based on this information, that it is necessary to send unlabeled data to the model training device first.

[0280] The information used to indicate feedback features and / or no-feedback labels may include: information indicating feedback features and / or information indicating no-feedback labels. This information can also be replaced with: information indicating feedback of unlabeled data or information indicating the feedback object, etc. The feedback object can be a feature, a label, or a feature and a label. When the information indicating the feedback object is information indicating a feature, it indicates feedback of the feature but no feedback of the label; when the information indicating the feedback object is information indicating a label, it indicates feedback of the label but no feedback of the feature; when the information indicating the feedback object is information indicating both a label and a feature, it indicates feedback of both the label and the feature. The information indicating the feedback object can be of Boolean type; for example, a true value indicates feedback of the feature but no feedback of the label, and a false value indicates feedback of both the feature and the label. To request multiple features, the information indicating the feedback object in this embodiment can be information indicating the feature.

[0281] It is understood that the second, third, and fourth indication information shown above, which indicate the duration or time period for saving the tag, as well as the first indication information, can be triggered based on the data providing device needing to save the tag.

[0282] For example, method 500 further includes: the model training device sending information to the data providing device indicating whether to save the labels. Correspondingly, the data providing device receives information from the model training device indicating whether to save the labels.

[0283] The information indicating whether to save the labels can be a Boolean type, where a true value indicates that the labels are saved, and a false value indicates that the labels are not saved. To facilitate the model training device in subsequently obtaining labels from the data provider, the information indicating whether to save the labels can be further specified, allowing the data provider to save the labels.

[0284] The information indicating whether to save labels can depend on the information indicating whether to provide feedback on features and / or not to provide feedback on labels. That is, the model training device only sends the information indicating whether to save labels to the data provider if the model training device sends information indicating feedback on features and / or not to provide feedback on labels. In other words, if the model training device requests features and labels, the data provider can send multiple features and their corresponding labels to the model training device simultaneously. In this case, the data provider can choose to delete or save the labels, and the model training device may not need to send the information indicating whether to save labels to the data provider. However, if the model training device requests features but not labels from the data provider, it may subsequently obtain some of the multiple labels from the data provider, and may obtain labels multiple times. Therefore, the model training device may send the information indicating whether to save labels to the data provider.

[0285] Alternatively, the information used to indicate whether to save labels may be independent of the information used to indicate whether to feed back features and / or not to feed back labels. For example, the protocol may stipulate that the information used to indicate whether to save labels may also indicate that the data providing device feeds back features and labels to the model training device.

[0286] It should be understood that information indicating feedback features and / or no feedback tags, and / or information indicating whether to save tags, can be carried in Request 2 for requesting multiple features, or can be sent via signaling different from Request 2. This application does not specifically limit this.

[0287] In addition, the model training device can obtain the third part of the labels from the data providing device in the following way, that is, S503 can be implemented in the following way.

[0288] It should be noted that in method 500, S503 can be triggered based on a request from the model training device or based on other events. In one possible implementation, before S503, method 500 further includes: the model training device sending request 3 to the data providing device, request 3 being used to request a third part of the label. Correspondingly, the data providing device receives request 3 from the model training device.

[0289] In this way, the data providing device can send the third part of the label to the model training device based on request 3.

[0290] It should be understood that the first instruction information may be carried in request 3 or sent via a different signaling than request 3, and this application does not specifically limit this.

[0291] Based on the above embodiments, the model training device can determine the third part of the labels to be acquired based on the maximum amount of label data that the data providing device can store. For example, before the model training device sends request 3 to the data providing device, method 500 may further include: the data providing device sending information to the model training device indicating a first data volume, where the first data volume is the maximum amount of label data that the data providing device can store. Correspondingly, the model training device receives the information indicating the first data volume from the data providing device and determines the third part of the labels based on the first data volume.

[0292] For example, when the first amount of data is large, the data providing device can store more labels, so the model training device can request fewer third-party labels; when the first amount of data is small, it means that the data providing device can only store a small amount of labels. In order for the model training device to obtain enough training data to train the model, the model training device can request more third-party labels from the data providing device, so that the data providing device can delete relatively more labels.

[0293] For example, if the initial data volume is small, and the model training device requests a large number of third-part labels from the data providing device, and the first-part labels indicated by the first instruction information can be, for example, already returned labels, then the data providing device can delete the third-part labels. This reduces the load on the data providing device and allows the model training device to acquire sufficient labels in a timely manner.

[0294] It should be noted that in the embodiments of this application, "larger" and "smaller" can be relative concepts. For example, in conjunction with the above example, assuming that data volume 1 is greater than data volume 2, then when the first data volume is data volume 1, the third part of the label requested by the model training device can be less than the third part of the label requested by the model training device when the first data volume is data volume 2. Alternatively, "larger" and "smaller" can also be determined based on a specific threshold or a preset threshold. For example, when the first data volume is greater than or equal to (or can be replaced by greater than) threshold 1, it can be determined that the first data volume is large; when the first data volume is less than or equal to (or can be replaced by less than) threshold 2, it can be determined that the first data volume is small, and threshold 1 is greater than or equal to threshold 2.

[0295] For example, the method 500 shown above can be... Figure 6 The process implementation is shown. That is, Figure 6 The illustrated process can be understood as one possible implementation of method 500, but Figure 6 The illustrated process does not constitute a limitation on the embodiments of this application.

[0296] Figure 6This is a flowchart illustrating a data transmission method 600 provided in an embodiment of this application. Method 600 can be applied to a communication system 1000, and method 600 includes the following steps:

[0297] S601, the model training device sends request 2 to the data providing device, request 2 requesting multiple features. Correspondingly, the data providing device receives request 2 from the model training device.

[0298] Among these features, several can refer to unlabeled data measured by the data providing device. That is, Request 2 can also be used to request measurement data or data measured by the data providing device.

[0299] Request 2 may also carry one or more of the following information: second instruction information, fifth instruction information indicating feedback features and / or no feedback on the tag, or information indicating whether to save the tag.

[0300] It should be understood that the above-mentioned information may not be included in request 2, and the above-mentioned information may be sent through different or the same signaling. Furthermore, for a description of the above-mentioned information, please refer to method 500, which will not be repeated here.

[0301] It is understandable that S601 can determine the need for model training based on the model training device.

[0302] For example, prior to S601, method 600 includes: a model training device acquiring a model; and the model training device determining that the model needs to be trained.

[0303] The model training device can obtain models from equipment vendor servers, network management systems, access network equipment units responsible for model distribution, or core network equipment network elements responsible for model distribution. These models can be, for example, models trained by equipment vendors, operators, or third-party manufacturers based on simulation data and / or live network data.

[0304] It should be noted that the model acquired by the model training device may not meet the device's requirements, thus requiring the device to train the acquired model. For example, the model training device may have already acquired some local data, which can be understood as data collected from other devices or data measured by the device itself. This local data can be used to train the model or test its performance. For instance, when the model is used for predicting channel feature information, the local data may include channel feature information.

[0305] Before running (or using) the model, the model training device may determine that the model does not match the local data and needs to be trained based on the local data; or, after running the model for a period of time, the model performance may degrade and fail to meet the requirements, causing the model training device to determine that the model needs to be trained. For example, in a channel measurement scenario, the channel feature information may change continuously, which may reduce the accuracy of the model in predicting the channel feature information and make it difficult to meet the requirements.

[0306] The determination by the model training equipment that the model needs to be trained can also be understood as: the model training equipment determines that training data needs to be collected from the data providing equipment in order to train the model.

[0307] S602. Based on request 2, the data providing device sends multiple features to the model training device. Correspondingly, the model training device receives multiple features from the data providing device.

[0308] It is understandable that before the data providing device receives request 2, the multiple features may be data stored in the data providing device, and the data providing device sends these multiple features to the model training device based on request 2. Alternatively, before the data providing device receives request 2, the multiple features may not be stored in the data providing device, in which case the data providing device may perform measurements based on request 2 and save the measured data, which includes multiple features; and the data providing device may send these multiple features to the model training device based on request 2.

[0309] It should be noted that, when request 2 carries the fifth indication information, or when the model training device sends the fifth indication information to the data providing device via other signaling before S602, the data providing device can acquire multiple features based on the data type indicated by the fifth indication information, the number of multiple features, and the time period for collecting multiple features. Furthermore, when request 2 carries information indicating feedback features and / or no feedback labels, or when the model training device sends information indicating feedback features and / or no feedback labels to the data providing device via other signaling before S602, the data providing device only provides feedback features to the model training device in S602.

[0310] It should be understood that Request 2 may carry third indication information, which indicates the duration or time period for which the data providing device retains the tag; alternatively, the third indication information may be transmitted via a message different from Request 2. When the data providing device sends the third indication information to the model training device, the second indication information sent by the model training device to the data providing device is optional.

[0311] S603, the model training device sends request 3 to the data providing device, request 3 being used to request the third part of the label. Correspondingly, the data providing device receives request 3 from the model training device.

[0312] The features corresponding to the third part of the label can also be understood as the high-value unlabeled data mentioned above. This third part of the label can be determined by the model training device based on multiple features. For example, the model training device can sequentially input each of the multiple features into the model to be trained and obtain the actual output data corresponding to each feature; based on the actual data corresponding to each feature, the model training device determines the high-value unlabeled data, which is the feature corresponding to the third part of the label. Then, the model training device can request the third part of the label from the data providing device.

[0313] For example, when the model to be trained is used to locate a terminal device, the model input can be the channel impulse responses (CIRs) between the terminal device and the network device, and the model output can be the location information of the terminal device, such as the coordinates of the terminal device. That is, the feature is the CIR, and the label is the location information. The model training device can, for example, determine high-value unlabeled data in the following ways.

[0314] In the first stage, N labeled data points are acquired, where the features can be CIR and the labels can be location information. Using a sampling method with replacement, K datasets are generated based on the N labeled data points, where K is greater than 1. A model is trained based on each of the K datasets, resulting in a total of K models.

[0315] Here, N is a positive integer. The N labeled data points can be, for example, the training data that the model training device has already obtained from the data provider. However, if the model training device fails to train the model well based on these N labeled data points, it can continue to identify new high-value unlabeled data points based on these N labeled data points, and then obtain the corresponding labels for these new high-value unlabeled data points from the data provider. In this way, the model training device can obtain more training data.

[0316] In the second stage, each unlabeled data point is input into the aforementioned K models to obtain the corresponding output value (i.e., location information), resulting in a total of K output values ​​for each unlabeled data point. Simultaneously, each unlabeled data point is input into the model to be trained, obtaining the corresponding output value (i.e., location information). For each unlabeled data point, the output value obtained from the model to be trained is compared with the K output values ​​obtained from the K models, yielding K differences. Based on these K differences, the step size (learning rate) of each of the K models is calculated.

[0317] In the third stage, for each unlabeled data point, the average of the corresponding K step sizes is taken to obtain the average value of each unlabeled data point. One or more unlabeled data points with the largest average value among all unlabeled data points are selected as high-value unlabeled data points.

[0318] It is understandable that before the data providing device receives request 3 from the model training device, the model training device may have already measured the labels corresponding to some or all of the multiple features, including a third part of the labels. In this case, when the third part of the labels is only a portion of the labels corresponding to some or all of the multiple features, the amount of label data sent by the data providing device to the model training device is relatively small, and the time-frequency domain resources consumed are relatively low.

[0319] Alternatively, if the model training device has not yet measured the third part of the label when the data providing device receives request 3 from the model training device, the data providing device can measure the third part of the label based on request 3. In this case, the data providing device can measure the third part of the label instead of the labels corresponding to all features, resulting in relatively low power consumption. It should be noted that in this case, the data providing device's ability to measure the third part of the label based on request 3 does not limit the data providing device to measuring only the third part of the label; the data providing device can also measure the remaining labels. This application does not impose specific limitations in this regard.

[0320] It should be understood that Request 3 may carry first instruction information, which instructs the deletion of the first part of the tag and / or the saving of the second part of the tag; or, the first instruction information may also be carried in a message different from Request 3, and this application does not specifically limit this.

[0321] S604. Based on request 3, the data providing device sends the third part of the label to the model training device. Correspondingly, the model training device receives the third part of the label from the data providing device.

[0322] It should be understood that the message transmitting the third part tag may also carry fourth indication information, which is used to indicate the duration or time period for which the data providing device retains the fourth part tag; or, the fourth indication information may also be carried in a message different from the message transmitting the third part tag, and this application does not specifically limit this.

[0323] S605. The data providing device deletes the first part of the tag and / or saves the second part of the tag based on the first instruction information, or the first instruction information and the second instruction information, or the first instruction information and the third instruction information.

[0324] In one possible implementation, the duration for which the tag is saved is the duration indicated by the second or third instruction information, or the time period for which the tag is saved is the time period indicated by the second or third instruction information.

[0325] S606, the model training device trains the model based on the third part of the label and the features corresponding to the third part of the label. The features corresponding to the third part of the label are some or all of the features among multiple features.

[0326] The third part of the label and its corresponding features can also be referred to as labeled data. Alternatively, the third part of the label and its corresponding features can be understood as part or all of the data in the training data.

[0327] If the model trained by the model training device based on the third part label and the features corresponding to the third part label meets the requirements, such as the accuracy of the trained model in predicting channel state information being greater than or equal to a preset threshold, this process ends.

[0328] If the model trained by the model training device based on the third part of the label and the features corresponding to the third part of the label does not meet the requirements, for example, if the accuracy of the trained model in predicting channel state information is less than a preset threshold, steps S603 to S606 can be repeated. During the repeated execution of steps S603 to S606, the label obtained by the model training device from the data providing device can be the same as, partially the same as, or completely different from the third part of the label; this application does not impose specific limitations in this regard.

[0329] It should be noted that in methods 500 or 600 above, regardless of whether the data providing device deletes labels, the data providing device can save multiple features so that it can subsequently measure the labels corresponding to some or all of the multiple features to provide training data to the model training device or other devices. Alternatively, after the data providing device sends multiple features to the model training device, the data providing device can delete the features corresponding to the measured labels to release some of the data providing device's content resources. Or, when the data providing device deletes labels, it can simultaneously delete the features corresponding to the deleted labels, etc. This application does not specifically limit this.

[0330] It should also be noted that the order of the methods listed above does not imply the order of execution. The execution order of each process should be determined by its function and internal logic.

[0331] The above text combined Figure 5 and Figure 6 The data transmission method of the embodiments of this application is described in detail below, in conjunction with Figures 7 to 8This application describes in detail the communication apparatus according to embodiments of the present application. The communication apparatus includes modules or units for performing each part of the above embodiments. The modules or units may be software, hardware, or a combination of software and hardware. The following is only a brief illustrative example of the communication apparatus; for details of the implementation, please refer to the description of the foregoing method embodiments, which will not be repeated below.

[0332] Figure 7 This is a schematic block diagram of a communication device 700 provided in an embodiment of this application. Figure 7 As shown, the communication device 700 includes a transceiver module 701.

[0333] In one possible implementation, the communication device 700 is used to implement the steps corresponding to the data providing device in the above-described method 500 or method 600.

[0334] The transceiver module 701 is used to receive first instruction information, which includes instructions on deleting a first part of the labels in the training data and / or saving a second part of the labels in the training data. The multiple labels include labels corresponding to some or all of the features in the training data. The training data is used for model training.

[0335] Optionally, the first instruction information includes at least one of the following: information indicating the first part of the label; information indicating the deletion of the first part of the label; information indicating the second part of the label; or, information indicating the saving of the second part of the label.

[0336] Optionally, the first part of the tags includes: all the tags among the multiple tags, or tags among the multiple tags that the device 700 has sent to the second communication device, or tags among the multiple tags that the device 700 has not sent to the second communication device; and / or, the second part of the tags includes: all the tags among the multiple tags, or tags among the multiple tags that the device 700 has sent to the second communication device, or tags among the multiple tags that the device 700 has not sent to the second communication device.

[0337] Optionally, the transceiver module 701 is further configured to: receive second indication information, the second indication information including information indicating a first duration or information indicating a first time period, the first duration being used to indicate the duration for which the device 700 stores the tag, and the first time period being used to indicate the time period for which the device 700 stores the tag.

[0338] Optionally, the transceiver module 701 is further configured to: send third indication information, the third indication information including information indicating a second duration or information indicating a second time period, the second duration being used to indicate the duration for which the device 700 stores the tag, and the second time period being used to indicate the time period for which the device 700 stores the tag.

[0339] Optionally, the transceiver module 701 is also configured to: send a third part tag, wherein the third part tag is some or all of the tags among a plurality of tags.

[0340] Optionally, the transceiver module 701 is further configured to: send fourth indication information, the fourth indication information including information indicating a third duration or information indicating a third time period, the third duration being used to indicate the duration for which the device 700 stores the fourth part of the tag, the third time period being used to indicate the time period for which the device 700 stores the fourth part of the tag, and the fourth part of the tag being a tag other than the third part of the tag among multiple tags.

[0341] Optionally, the transceiver module 701 is further configured to: send information indicating a first data volume, the first data volume being the maximum data volume of the tag stored in the device 700, and the third part of the tag being determined based on the first data volume.

[0342] Optionally, the transceiver module 701 is also used to: send multiple features.

[0343] Optionally, the transceiver module 701 is also configured to: receive information for indicating feedback features and / or non-feedback tags.

[0344] Optionally, the transceiver module 701 is also configured to: receive information indicating whether to save the tag.

[0345] Optionally, the device 700 further includes a processing module 702, which is used to: determine whether to delete the first part of the label and / or whether to save the second part of the label based on the first instruction information.

[0346] In another possible implementation, the communication device 700 includes a transceiver module 701 and a processing module 702. This communication device 700 is used to implement the steps corresponding to the data providing device in method 500 or method 600 described above.

[0347] Processing module 702 is used to: determine first instruction information, the first instruction information including information on deleting a first part of the labels in multiple labels in the training data and / or saving a second part of the labels in multiple labels, the multiple labels including labels corresponding to some or all of the features in multiple features in the training data, the training data being used for model training; transceiver module 701 is used to: send the first instruction information.

[0348] Optionally, the first instruction information includes at least one of the following: information indicating the first part of the label; information indicating the deletion of the first part of the label; information indicating the second part of the label; or, information indicating the saving of the second part of the label.

[0349] Optionally, the first part of the tags includes: all of the tags among the multiple tags, or tags among the multiple tags that the first communication device has sent to the device 700, or tags among the multiple tags that the first communication device has not sent to the device 700; and / or, the second part of the tags includes: all of the tags among the multiple tags, or tags among the multiple tags that the first communication device has sent to the device 700, or tags among the multiple tags that the first communication device has not sent to the device 700.

[0350] Optionally, the transceiver module 701 is further configured to: send second indication information, the second indication information including information indicating a first duration or information indicating a first time period, the first duration being used to indicate the duration for which the first communication device stores the tag, and the first time period being used to indicate the time period for which the first communication device stores the tag.

[0351] Optionally, the transceiver module 701 is further configured to: receive third indication information, the third indication information including information indicating a second duration or information indicating a second time period, the second duration being used to indicate the duration for which the first communication device stores the tag, and the second time period being used to indicate the time period for which the first communication device stores the tag.

[0352] Optionally, the transceiver module 701 is also configured to: receive a third part tag, wherein the third part tag is some or all of the tags among a plurality of tags.

[0353] Optionally, the transceiver module 701 is further configured to: receive fourth indication information, the fourth indication information including information indicating a third duration or information indicating a third time period, the third duration being used to indicate the duration for which the first communication device stores the fourth part of the tag, the third time period being used to indicate the time period for which the first communication device stores the fourth part of the tag, and the fourth part of the tag being a tag other than the third part of the tag among multiple tags.

[0354] Optionally, the transceiver module 701 is further configured to: receive information indicating a first data volume, wherein the first data volume is the maximum data volume of the tag stored by the first communication device, and the third part of the tag is determined based on the first data volume.

[0355] Optionally, the transceiver module 701 is also used to receive multiple features.

[0356] Optionally, the transceiver module 701 is also configured to: send information for indicating feedback features and / or no-feedback tags.

[0357] Optionally, the transceiver module 701 is also configured to: send information indicating whether to save the tag.

[0358] It should be understood that the communication device 700 here is embodied in the form of a functional module. The term "module" here can refer to application-specific integrated circuits (ASICs), electronic circuits, processors (e.g., shared processors, proprietary processors, or group processors, etc.) and memories for executing one or more software or firmware programs, integrated logic circuits, and / or other suitable components supporting the described functions. In an alternative example, those skilled in the art will understand that the communication device 700 can specifically be the model training device or data providing device in the above embodiments. The communication device 700 can be used to execute the various processes and / or steps corresponding to the model training device or data providing device in the above method embodiments; to avoid repetition, these will not be described again here.

[0359] The aforementioned communication device 700 has the function of implementing the corresponding steps performed by the model training device or data providing device in the above method; the above function can be implemented by hardware or by hardware executing corresponding software. The hardware or software includes one or more modules corresponding to the above function. In the embodiments of this application, Figure 7 The communication device 700 in the text can also be a chip, such as a SOC.

[0360] Figure 8 A schematic diagram of the structure of a communication device 800 provided in an embodiment of this application is shown. The communication device 800 includes a processor 801, a transceiver 802, and a memory 803. The processor 801, transceiver 802, and memory 803 communicate with each other via internal interconnection paths. The memory 803 stores instructions, such as computer program code, and the processor 801 executes the instructions stored in the memory 803 to control the transceiver 802 to send and / or receive signals.

[0361] It should be understood that the communication device 800 may specifically be the model training device or data providing device in the above embodiments, and may be used to execute the various steps and / or processes corresponding to the model training device or data providing device in the above method embodiments. Optionally, the memory 803 may include a read-only memory and a random access memory, and provide instructions and data to the processor. A portion of the memory may also include non-volatile random access memory. For example, the memory may also store device type information. The processor 801 may be used to execute instructions stored in the memory, and when the processor 801 executes instructions stored in the memory, the processor 801 is used to execute the various steps and / or processes of the above method embodiments. The transceiver 802 may include a transmitter 8021, a receiver 8022, and an antenna 8023. The transmitter 8021 may be used to implement the various steps and / or processes corresponding to the transceiver for performing the transmission action. For example, the transmitter 8021 may be used to transmit information to another device through the antenna 8023. The receiver 8022 may be used to implement the various steps and / or processes corresponding to the transceiver for performing the reception action. For example, receiver 8022 can be used to receive information from another device via antenna 8023.

[0362] It should be understood that, in the embodiments of this application, the processor may be a central processing unit (CPU), a microprocessor unit (MPU), a microcontroller unit (MCU), a graphics processing unit (GPU), or other general-purpose processors, digital signal processors (DSPs), application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. The general-purpose processor may be a microprocessor or any conventional processor.

[0363] In implementation, each step of the above method can be completed by integrated logic circuits in the processor's hardware or by instructions in software. The steps of the method disclosed in the embodiments of this application can be directly manifested as execution by a hardware processor, or as a combination of hardware and software modules within the processor. The software modules can reside in random access memory, flash memory, read-only memory, programmable read-only memory, electrically erasable programmable memory, registers, or other mature storage media in the art. This storage medium is located in memory, and the processor executes the instructions in the memory, combining them with its hardware to complete the steps of the above method. To avoid repetition, detailed descriptions are omitted here.

[0364] Figure 9 A schematic diagram of another communication device 900 provided in an embodiment of this application is shown. This communication device 900 can be a chip system, or it can be an apparatus configured with a chip system to implement the methods described in the above method embodiments. In this embodiment, the chip system can be composed of chips, or it can include chips and other discrete devices.

[0365] like Figure 9 As shown, the communication device 900 may include a processor 910, which can be used to execute computer programs or instructions in memory to perform the various steps and / or processes corresponding to the model training device or data providing device in the above method embodiments.

[0366] In one possible implementation, the communication device 900 further includes a communication interface 920. The communication interface 920 can be used to communicate with other devices via a transmission medium, thereby enabling the communication device 900 to communicate with other devices. The communication interface 920 may be, for example, a transceiver, an input / output interface, pins, a bus, a transceiver circuit, or a device capable of transmitting and receiving functions. The processor 910 can utilize the communication interface 920 to input and output data for executing the various steps and / or processes corresponding to the model training device or data providing device in the above method embodiments.

[0367] In one possible implementation, the communication device 900 further includes at least one memory 930 for storing program instructions and / or data. The memory 930 is coupled to the processor 910. The coupling in this embodiment is an indirect coupling or communication connection between devices, units, or modules, and can be electrical, mechanical, or other forms, for information exchange between devices, units, or modules. The processor 910 may operate in conjunction with the memory 930. The processor 910 may execute the program instructions stored in the memory 930.

[0368] Optionally, the memory 930 may be a memory disposed in the device 900. Exemplarily, the memory 930 may be integrated with the processor 910; or, the memory 930 may be disposed separately from the processor 910.

[0369] Optionally, the memory 930 may be a memory outside of the device 900. It may also be a memory outside of the communication device.

[0370] Figure 10 This is a schematic block diagram illustrating an O-RAN architecture according to an embodiment of this application. The O-RAN architecture can be understood as the architecture of O-RAN devices. Figure 10 The O-RAN architecture shown is merely an example; O-RAN equipment may also include... Figure 10 Other components besides those shown.

[0371] It should be understood that the access network equipment in the embodiments of this application, such as the data providing equipment and / or model training equipment for the access network equipment, can be O-RAN equipment, that is, it can adopt such... Figure 10 The architecture shown.

[0372] like Figure 10 As shown, the O-RAN equipment includes: Non-real-time radio access network intelligent controller (Non-RT RIC), Near-real-time radio access network intelligent controller (Near-RTRIC), O-RAN central unit control plane (O-CU-CP), O-RAN central unit user plane (O-CU-UP), O-RAN distributed unit (O-DU), and O-RAN evolved node B (O-eNB).

[0373] O-RAN devices can communicate with the core network (CN) via the backhaul link, or with terminal devices via the air interface.

[0374] This application also provides a computer-readable storage medium for storing a computer program for implementing the methods shown in the above-described method embodiments.

[0375] This application also provides a computer program product, which includes a computer program (also referred to as code or instructions) that, when run on a computer, enables the computer to perform the methods shown in the above-described method embodiments.

[0376] Those skilled in the art will recognize that the modules and algorithm steps of the various examples described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of this application.

[0377] Those skilled in the art will understand that, for the sake of convenience and brevity, the specific working processes of the systems, devices, and modules described above can be referred to the corresponding processes in the foregoing method embodiments, and will not be repeated here.

[0378] In the several embodiments provided in this application, it should be understood that the disclosed systems, apparatuses, and methods can be implemented in other ways. For example, the apparatus embodiments described above are merely illustrative; for instance, the division of modules is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple modules or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the coupling or direct coupling or communication connection shown or discussed may be through some interfaces; the indirect coupling or communication connection between apparatuses or modules may be electrical, mechanical, or other forms.

[0379] The modules described as separate components may or may not be physically separate. The components shown as modules may or may not be physical modules; that is, they may be located in one place or distributed across multiple network modules. Some or all of the modules can be selected to achieve the purpose of this embodiment according to actual needs.

[0380] In addition, the functional modules in the various embodiments of this application can be integrated into one processing module, or each module can exist physically separately, or two or more modules can be integrated into one module.

[0381] If the aforementioned functions are implemented as software functional modules and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this application, essentially, or the part that contributes to existing technology, or a portion of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of this application. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0382] The above description is merely a specific embodiment of this application, but the protection scope of the embodiments of this application is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the technical scope disclosed in the embodiments of this application should be included within the protection scope of the embodiments of this application. Therefore, the protection scope of the embodiments of this application should be determined by the protection scope of the claims.

Claims

1. A data transmission method, characterized in that, Applied to a first communication device, comprising: Receive first instruction information, the first instruction information including information on deleting a first part of the labels in a plurality of labels in the training data and / or saving a second part of the labels in the plurality of labels, the plurality of labels including labels corresponding to some or all of the features in a plurality of features in the training data, the training data being used for model training.

2. The method according to claim 1, characterized in that, The first indication information includes at least one of the following: Information indicating the first part of the label; Instructs the deletion of the information in the first part of the label; Information indicating the second part of the label; or, Instructions to save the information of the second part of the label.

3. The method according to claim 1 or 2, characterized in that, The first portion of the tags includes: all of the plurality of tags, or tags that the first communication device has sent to the second communication device, or tags that the first communication device has not sent to the second communication device; and / or, The second part of the tags includes: all the tags among the plurality of tags, or the tags among the plurality of tags that the first communication device has sent to the second communication device, or the tags among the plurality of tags that the first communication device has not sent to the second communication device.

4. The method according to any one of claims 1 to 3, characterized in that, The method further includes: Receive second indication information, the second indication information including information indicating a first duration or information indicating a first time period, the first duration being used to indicate the duration for which the first communication device stores the tag, and the first time period being used to indicate the time period for which the first communication device stores the tag.

5. The method according to any one of claims 1 to 4, characterized in that, The method further includes: Send a third indication message, which includes information indicating a second duration or information indicating a second time period. The second duration is used to indicate the duration for which the first communication device stores the tag, and the second time period is used to indicate the time period for which the first communication device stores the tag.

6. The method according to any one of claims 1 to 5, characterized in that, The method further includes: Send a third part of the tag, which is some or all of the multiple tags.

7. The method according to claim 6, characterized in that, The method further includes: Send a fourth indication message, which includes information indicating a third duration or information indicating a third time period. The third duration is used to indicate the duration for which the first communication device stores the fourth part of the tag, and the third time period is used to indicate the time period for which the first communication device stores the fourth part of the tag. The fourth part of the tag is the tag other than the third part of the tag among the plurality of tags.

8. The method according to claim 6 or 7, characterized in that, The method further includes: Information indicating a first data volume is sent, where the first data volume is the maximum data volume of the tag stored by the first communication device, and the third part of the tag is determined based on the first data volume.

9. The method according to any one of claims 1 to 8, characterized in that, Before receiving the first indication information, the method further includes: Send the aforementioned multiple features.

10. The method according to any one of claims 1 to 9, characterized in that, The method further includes: Receive information indicating feedback features and / or no-feedback labels.

11. The method according to any one of claims 1 to 10, characterized in that, The method further includes: Receive information indicating whether to save the tag.

12. The method according to any one of claims 1 to 11, characterized in that, The method further includes: Based on the first instruction information, determine whether to delete the first part of the label and / or whether to save the second part of the label.

13. A data transmission method, characterized in that, Applied to a second communication device, including: Determine first instruction information, which includes instructions on deleting a first portion of labels from multiple labels in training data and / or saving a second portion of labels from the multiple labels, wherein the multiple labels include labels corresponding to some or all features from multiple features in the training data, and the training data is used for model training; Send the first instruction message.

14. The method according to claim 13, characterized in that, The first indication information includes at least one of the following: Information indicating the first part of the label; Instructs the deletion of the information in the first part of the label; Information indicating the second part of the label; or, Instructions to save the information of the second part of the label.

15. The method according to claim 13 or 14, characterized in that, The first portion of the tags includes: all of the plurality of tags, or tags that the first communication device has sent to the second communication device, or tags that the first communication device has not sent to the second communication device; and / or, The second part of the tags includes: all the tags among the plurality of tags, or the tags among the plurality of tags that the first communication device has sent to the second communication device, or the tags among the plurality of tags that the first communication device has not sent to the second communication device.

16. The method according to any one of claims 13 to 14, characterized in that, The method further includes: Send a second instruction message, which includes information indicating a first duration or information indicating a first time period. The first duration is used to indicate the duration for which the first communication device stores the tag, and the first time period is used to indicate the time period for which the first communication device stores the tag.

17. The method according to any one of claims 13 to 16, characterized in that, The method further includes: Receive third indication information, the third indication information including information indicating a second duration or information indicating a second time period, the second duration being used to indicate the duration for which the first communication device stores the tag, and the second time period being used to indicate the time period for which the first communication device stores the tag.

18. The method according to any one of claims 13 to 17, characterized in that, The method further includes: Receive a third part of the tag, wherein the third part of the tag is some or all of the multiple tags.

19. The method according to claim 18, characterized in that, The method further includes: The system receives a fourth indication message, which includes information indicating a third duration or information indicating a third time period. The third duration is used to indicate the duration for which the first communication device stores the fourth part of the tag, and the third time period is used to indicate the time period for which the first communication device stores the fourth part of the tag. The fourth part of the tag is a tag other than the third part of the tag among the plurality of tags.

20. The method according to claim 18 or 19, characterized in that, The method further includes: The system receives information indicating a first data volume, which is the maximum data volume of the tag stored in the first communication device, and the third part of the tag is determined based on the first data volume.

21. The method according to any one of claims 13 to 20, characterized in that, Before receiving the first indication information, the method further includes: Receive the multiple features.

22. The method according to any one of claims 13 to 21, characterized in that, The method further includes: Send information to indicate feedback features and / or no-feedback labels.

23. The method according to any one of claims 13 to 22, characterized in that, The method further includes: Send information indicating whether to save the tag.

24. A communication device, characterized in that, include: It includes modules for performing the method as described in any one of claims 1 to 12, or the method as described in any one of claims 13 to 23.

25. A communication device, characterized in that, include: A processor coupled to a memory for storing a computer program, which, when invoked by the processor, causes the apparatus to perform the method of any one of claims 1 to 12, or the method of any one of claims 13 to 23.

26. A computer-readable storage medium, characterized in that, Used to store a computer program, the computer program including instructions for implementing the method as described in any one of claims 1 to 12, or the method as described in any one of claims 13 to 23.

27. A computer program product, the computer program product comprising instructions, characterized in that, When the instructions are executed on a computer, the computer causes the computer to implement the method as described in any one of claims 1 to 12, or the method as described in any one of claims 13 to 23.

Images