WO2024055918A1

WO2024055918A1 - Data transmission method and apparatus, and device and storage medium

Info

Publication number: WO2024055918A1
Application number: PCT/CN2023/117912
Authority: WO
Inventors: 王四海; 秦城; 杨锐
Original assignee: 华为技术有限公司
Priority date: 2022-09-13
Filing date: 2023-09-11
Publication date: 2024-03-21
Also published as: CN117714309A

Abstract

A data transmission method and apparatus, and a device and a storage medium. The method comprises: a first communication apparatus receiving first indication information, which is sent by a second communication apparatus. The first indication information is used to indicate information related to first data for model processing, such that the first communication apparatus or a first processing apparatus is instructed to perform model processing on a first AI model, so as to execute a model processing process in a communication system.

Description

Data transmission methods, devices, equipment and storage media

This application requires the priority of the Chinese patent application submitted to the State Intellectual Property Office of China on September 13, 2022, with the application number 202211109372.9 and the application title "A data transmission method, terminal equipment, network equipment, device", and, The priority of the Chinese patent application submitted to the State Intellectual Property Office of China on November 4, 2022, with the application number 202211379250.1 and the application name "Data transmission method, device, equipment and storage medium", the entire content of which is incorporated by reference in in this application.

Technical field

The present application relates to the field of communication technology, and in particular, to a data transmission method, device, equipment and storage medium.

Background technique

At present, artificial intelligence (AI) or machine learning (ML) technology is introduced into wireless communication systems, such as model processing (such as model training, model verification, and model testing) of AI/ML models in wireless communication systems. etc.), so that AI/ML models can be applied to communication systems to improve network performance or efficiency. However, how to transmit data between communication devices to support the execution of model processing is an urgent problem that needs to be solved.

Contents of the invention

Embodiments of the present application provide a data transmission method, device, equipment and storage medium for data transmission between communication devices to support the execution of model processing.

In a first aspect, embodiments of the present application provide a data transmission method, including: a first communication device receiving first instruction information sent by a second communication device; the first communication device performing a first AI model based on the first data Model processing, the first data is the data received by the first communication device from the second communication device; or the first communication device sends the first instruction information to the first processing device.

In one design, the first indication information is used to indicate data characteristics of the first data. The second communication device sends the first instruction information to the first communication device, and uses the first instruction information to indicate the data characteristics of the first data for model processing on the first AI model, so as to facilitate receiving and using the first data to perform model processing on the first AI model. Model processing.

In a possible implementation, the data characteristics include at least one of the following: a sample category, which includes an input category and/or an expected output category; the data of the input category is used to input the first AI model, and the expected output category The data of the output category is used to indicate the expected output of the first AI model when the data of the input category is input, and the data of the input category corresponds to the data of the expected output category; the configuration information when the first data is generated; the The data dimension of the first data; the data type of the first data; the data quantity of the first data; the number of data transmitted each time the first data is transmitted in batches.

Through the data transmission method provided by this embodiment, the sample category of the first data, the configuration information when the first data is generated, the data dimensions of the first data, the data type of the first data, and the data quantity of the first data are first. When the data is transmitted in batches, at least one of the data quantities transmitted each time is indicated, so as to facilitate the first communication device to receive and use the first data.

In another design, the first instruction information is used to instruct the data set to which the first data belongs to perform model processing of the first AI model. The first data is received by the first communication device from the second communication device. Data, the model processing includes at least one of model training, model verification, and model testing. The first communication device determines the use of the first data in the model processing process, such as for model training, model verification or model testing, based on the data set to which the first data indicated by the first instruction information belongs, and then performs on the first AI model Corresponding model processing.

In a possible implementation, the first indication information includes at least one of the following: a type of data set to which the first data belongs. The type of data set includes at least one of a training data set, a verification data set, and a test data set. One type; the identification of the data set to which the first data belongs; and the proportion of data quantity in each type of data set.

Through the data transmission method provided in this embodiment, the first communication device can determine the use of the first data in the model processing process according to the data set to which the first data indicated by the first indication information belongs, for example, for model training, model validation or model Test, and then perform corresponding model processing on the first AI model.

In a possible implementation, the first indication information includes an identification of a data set to which the first data belongs; the data set to which the first data belongs is a data set saved in the first target device, and the first indication information It also includes first incremental data information, which is used to indicate adding the first data in the data set; the first target device is the first communication device or the first processing device.

Through the data transmission method provided in this embodiment, the first communication device adds the received first data to the corresponding data set based on the first incremental data information, so as to facilitate subsequent use of the first data to execute the corresponding model on the first AI model. deal with.

In another design, for example, the first indication information is used to indicate the data quantity of the second data. The second data includes the first data and the data received by the third communication device from the second communication device. The third communication device The second data is used to perform model processing on the first AI model, which facilitates the first processing device to determine the total amount of all data to be received for performing model processing on the first AI model, thereby improving communication stability.

In yet another design, for example, the first indication information is used to indicate the first AI model, so that the first target device determines the object to be processed by the model.

Optionally, the first indication information includes at least one of the following: an identification of the first AI model; model parameters of the first AI model; information indicating training or optimization training to obtain the first AI model.

Optionally, the first instruction information includes information indicating that the first AI model is obtained through optimized training; the first instruction information also includes an identification of the base model of the first AI model, and the identification of the base model includes the structure of the base model. Identification and/or parameter identification of the basic model; the first communication device training the first AI model based on the first data includes: the first communication device training the basic model based on the first data to obtain the first AI model.

In yet another design, for example, the first indication information is used to indicate the identity of the wireless communication network to which the first AI model belongs, so that the model-processed first AI model can run under the wireless communication network.

In yet another design, for example, the first indication information is used to indicate the round and/or batch of the first data in the training process of the target AI model, so as to facilitate the first communication device or the first processing device to determine The round and/or batch of model processing determines the batch and/or round to which the first data belongs, thereby improving the stability of data transmission.

Combined with any design of the above examples, the first instruction information is also used to instruct model processing of the first AI model. The model processing includes at least one of model training, model verification, and model testing. The first communication device or The first processing device may determine what model processing to perform on the first AI model.

Combined with any design of the above examples, the first indication information is also used to indicate that the model processing of the first AI model is started when at least one of the following meets the corresponding threshold: the number of received data in the first data accounts for the The ratio of the data quantity of the first data; the ratio of the data quantity of the received data in the second data to the data quantity of the second data; the data quantity of the received data in the first data; the data quantity of the received data in the second data The data quantity of the data; the duration of receiving the first data; the duration of receiving the second data. This prevents the first target device from waiting blindly and being unable to start training in time to obtain the first AI model that meets the performance requirements.

Combined with any design of the above examples, the first instruction information is used to instruct model verification of the first AI model, and the first instruction information is also used to indicate at least one verification indicator and/or the at least one verification indicator corresponds to The threshold value can flexibly control the first communication device or the first processing device to perform model verification on the first AI model.

Combined with any design of the above examples, the first instruction information is used to instruct model testing of the first AI model, and the first instruction information is also used to indicate at least one test indicator and/or at least one test indicator corresponding to The threshold value can flexibly control the first communication device or the first processing device to perform model testing on the first AI model.

Optionally, the first processing device is an upper layer OTT server on the network.

Optionally, the second communication device is a network device.

Optionally, the first communication device is a terminal device.

Combined with any design of the above examples, the target AI model includes the first AI model and the second AI model. The second AI model is model processed by the second target device. The second target device includes the second communication device or The method further includes: the first communication device sending second instruction information to the second communication device, the second instruction information being used to indicate at least one of the following: The data characteristics of the third data; the data quantity of the fourth data; the data set to which the third data belongs; the second AI model; the model processing of the second AI model; the wireless communication network to which the second AI model belongs The identification of the network; the round of the third data in the training process of the target AI model; the batch of the third data in the training process of the target AI model; wherein the third data is the first communication device In the data sent to the second communication device, the fourth data includes the third data. The second target device may perform model processing on the second AI model based on the indication of the second indication information to implement joint model processing of the bilateral models.

In a possible implementation, the training process of the target AI model includes multiple rounds of training processes, and each round of training process includes multiple batches of training processes. In each batch of the target AI model, During the training process, training process data is transmitted between the first communication device and the second communication device, and the training process data includes the first data and the third data.

Combined with any design of the above examples, the method further includes: the first communication device receiving the configuration information sent by the second communication device; or the first communication device sending the configuration information to the second communication device; wherein, the The configuration information is used to indicate at least one of the following: the duration of the transmission of the training process data; the number of batch samples, used to determine the number of training process data transmitted during a batch of training of the target AI model; the target AI The total number of rounds of model training; the data characteristics of the training process data; the data set of the training process data; the target AI model; the model processing instructions of the target AI model; and the identification of the wireless communication network to which the target AI model belongs. Aligning the configurations between the first communication device and the second communication device to facilitate subsequent data transmission and execution of the model processing process.

Combined with any design of the above examples, the method further includes: the first communication device sending first feedback information to the second communication device, the first feedback information being used to indicate at least one of the following: The receiving status includes successful reception or unsuccessful reception; the number of successfully received data in the target data; the ratio of the number of successfully received data in the target data to the number of data in the target data; wherein the target data includes the first data and/or the first indication information. Feedback whether the target data is successfully received improves the reliability of data transmission, thereby making the model processing process more reliable.

Combined with any design of the above examples, the method further includes: the first communication device sending first feedback information to the second communication device, the first feedback information being used to indicate at least one of the following: The reception status of data, which includes successful reception or unsuccessful reception; the number of successfully received data in the second data; the ratio of the number of successfully received data in the second data to the number of data in the target data.

In a possible implementation, the first feedback information includes a bitmap, and the bits in the bitmap are used to indicate the reception status of the target data and/or the second data. The target data includes the first data and /or the first instruction information.

Through the data transmission method provided by this embodiment, accurate indication of the data reception status is achieved.

In a possible implementation, the first communication device sends the first instruction information to the first processing device, and before the first communication device sends the first feedback information to the second communication device, the method further includes: The first communication device receives the first feedback information sent by the first processing device.

Combined with any design of the above examples, the method further includes: the first communication device sending third indication information to the second communication device, the third indication information being used to determine receipt of the first communication device sent by the second communication device. Instruction information and/or first data; wherein the third instruction information includes at least one of the following: identification information of the first communication device; the ability of the first communication device to perform model processing, where the model processing includes model training, model At least one of verification and model testing; whether the first communication device is connected to the first processing device; whether the first communication device has the ability to connect to the first processing device; the communication quality of the first communication device Measurement results; location information of the first communication device; battery information of the first communication device; moving speed of the first communication device; and the first communication device is in a charging state. The second communication device evaluates the capability of the terminal equipment in the serving cell for relay transmission based on the third indication information reported by the first communication device, and determines at least one first communication device that can be used for relay transmission, thereby improving the communication system reliability of the model processing process.

Combined with any design of the above examples, the method further includes: the first communication device sending fourth indication information to the first processing device; wherein the fourth indication information is used to indicate that the first communication device is not in the first processing device. The communication range of the second communication device; or, the fourth indication information is used to indicate that the fourth communication device is not within the communication range of the second communication device. When the communication device used for relay transmission leaves the communication range of the second communication device, the communication device for relay transmission is adjusted and the transmission status is aligned to reduce the impact of the communication device for relay transmission leaving the communication range on the model processing process and improve the model processing process. Communication stability, thereby improving the stability of the model processing process of the AI model.

In a possible implementation, the fourth indication information is used to indicate that the fourth communication device is not within the communication range of the second communication device, and the method further includes: the first communication device receiving a message sent by the second communication device. the fourth instruction information.

In a possible implementation, the fourth indication information is used to indicate that the fourth communication device is not located next to the second communication device. communication range, the method further includes: the first communication device sending the first indication information corresponding to the fourth communication device to the first processing device.

In a second aspect, embodiments of the present application provide a data transmission method, including: the second communication device determines first indication information, the first indication information is used to indicate the data characteristics of the first data, and/or, is used to indicate The data set to which the first data belongs performs model processing of the first AI model. The first data is the data received by the first communication device from the second communication device. The model processing includes model training, model verification, and model testing. At least one; the second communication device sends the first indication information to the first communication device.

In another design, for example, the first indication information is used to indicate the data quantity of the second data. The second data includes the first data and the data received by the third communication device from the second communication device. The third communication device The second data is used to perform model processing on the first AI model.

In yet another design, for example, the first indication information is used to indicate the first AI model.

In yet another design, for example, the first indication information is used to indicate the identity of the wireless communication network to which the first AI model belongs.

In yet another design, for example, the first indication information is used to indicate the round and/or batch of the first data in the training process of the target AI model.

Combined with any design of the above examples, the first instruction information is also used to instruct model processing of the first AI model. The model processing includes at least one of model training, model verification, and model testing.

Combined with any design of the above examples, the first indication information is also used to indicate that the model processing of the first AI model is started when at least one of the following meets the corresponding threshold: the number of received data in the first data accounts for the The ratio of the data quantity of the first data; the ratio of the data quantity of the received data in the second data to the data quantity of the second data; the data quantity of the received data in the first data; the data quantity of the received data in the second data The data quantity of the data; the duration of receiving the first data; the duration of receiving the second data.

Combined with any design of the above examples, the first instruction information is used to instruct model verification of the first AI model, and the first instruction information is also used to indicate at least one verification indicator and/or the at least one verification indicator corresponds to threshold.

Combined with any design of the above examples, the first instruction information is used to instruct model testing of the first AI model, and the first instruction information is also used to indicate at least one test indicator and/or at least one test indicator corresponding to threshold.

Optionally, the second communication device is a network device.

Optionally, the first communication device is a terminal device.

Combined with any design of the above examples, the target AI model includes the first AI model and the second AI model. The second AI model is model processed by the second target device. The second target device includes the second communication device or A second processing device is communicatively connected to the second communication device. The method further includes: the second communication device receiving second indication information sent by the first communication device, the second indication information being used to indicate at least one of the following: The data characteristics of the third data; the data quantity of the fourth data; the data set to which the third data belongs; the second AI model; the model processing of the second AI model; the identification of the wireless communication network to which the second AI model belongs; the The round of the third data in the training process of the target AI model; the batch of the third data in the training process of the target AI model; wherein the third data is communicated by the first communication device to the second The fourth data includes the third data sent by the device.

Combined with any design of the above examples, the method further includes: the second communication device sends configuration information to the first communication device; or, the second communication device receives the configuration information sent by the first communication device; wherein, the The configuration information is used to indicate at least one of the following: the duration of the transmission of the training process data; the number of batch samples, used to determine the number of training process data transmitted during a batch of training of the target AI model; the target AI The total number of rounds of model training; the data characteristics of the training process data; the data set of the training process data; the target AI model; the model processing instructions of the target AI model; and the identification of the wireless communication network to which the target AI model belongs.

Combined with any design of the above examples, the method further includes: the second communication device receiving first feedback information sent by the first communication device, the first feedback information being used to indicate at least one of the following: data in the target data The receiving status includes successful reception or unsuccessful reception; the number of successfully received data in the target data; the proportion of the number of successfully received data in the target data to the number of data in the target data; wherein, the target data includes the first a piece of data and/or the first indication information.

Combined with any design of the above examples, the method further includes: the second communication device receiving first feedback information sent by the first communication device, the first feedback information being used to indicate at least one of the following: The reception status of data, which includes successful reception or unsuccessful reception; the number of successfully received data in the second data; the ratio of the number of successfully received data in the second data to the number of data in the target data.

Combined with any design of the above examples, the method further includes: the second communication device receiving third instruction information sent by the first communication device, the third instruction information being used to determine to send the first instruction to the first communication device information and/or first data; wherein the third indication information includes at least one of the following: identification information of the first communication device; the ability of the first communication device to perform model processing, where the model processing includes model training and model verification. , at least one of model testing; whether the first communication device is connected to the first processing device; whether the first communication device has the ability to connect to the first processing device; communication quality measurement of the first communication device Result: the position information of the first communication device; the battery information of the first communication device; the moving speed of the first communication device; the first communication device is in a charging state.

In a possible implementation, the fourth indication information is used to indicate that the fourth communication device is not within the communication range of the second communication device. The method further includes: the first communication device sending the first processing device the First instruction information corresponding to the fourth communication device.

The beneficial effects of the data transmission method provided by the above-mentioned second aspect and each possible implementation manner of the above-mentioned second aspect can be referred to the beneficial effects brought by the above-mentioned first aspect and each possible implementation manner of the first aspect, in No further details will be given here.

In a third aspect, embodiments of the present application provide a communication device. The communication device is a first communication device, including: a transceiver module, configured to receive first indication information sent by a second communication device, where the first indication information is used to indicate Data characteristics of the first data, and/or, used to indicate that the data set to which the first data belongs is subjected to model processing of the first AI model. The first data is data received by the transceiver module from the second communication device. The model The processing includes at least one of model training, model verification, and model testing; the processing module is used to perform model processing of the first AI model based on the first data; or the transceiver module is also used to Send the first instruction information to the first processing device.

In another design, illustratively, the first indication information is used to indicate the data quantity of second data, where the second data includes the first data and data received by the third communication device from the second communication device.

Optionally, the first instruction information includes information indicating that the first AI model is obtained through optimized training; the first instruction information also includes an identification of the base model of the first AI model, and the identification of the base model includes the structure of the base model. Identification and/or parameter identification of the basic model; the processing module is specifically used to: train the basic model according to the first data to obtain the first AI model.

Optionally, the second communication device is a network device.

Optionally, the first communication device is a terminal device.

Combined with any design of the above examples, the target AI model includes the first AI model and the second AI model. The second AI model is model processed by the second target device. The second target device includes the second communication device or The second processing device is communicatively connected to the second communication device. The transceiver module is also configured to send second indication information to the second communication device. The second indication information is used to indicate at least one of the following: data characteristics of the third data. ;The data quantity of the fourth data; the data set to which the third data belongs; the second AI model; the model processing of the second AI model; the identification of the wireless communication network to which the second AI model belongs; the location of the third data in the The round in the training process of the target AI model; the batch number of the third data in the training process of the target AI model times; wherein, the third data is data sent by the transceiver module to the second communication device, and the fourth data includes the third data. The second target device may perform model processing on the second AI model based on the indication of the second indication information to implement joint model processing of the bilateral models.

In a possible implementation, the training process of the target AI model includes multiple rounds of training processes, and each round of training process includes multiple batches of training processes. In each batch of the target AI model, During the training process, training process data is transmitted between the transceiver module and the second communication device, and the training process data includes the first data and/or the third data.

Combined with any design of the above examples, the transceiver module is also used to receive the configuration information sent by the second communication device; or, the transceiver module is also used to send the configuration information to the second communication device; wherein the configuration information is used to indicate At least one of the following: the duration of the transmission of the training process data; the number of batch samples, used to determine the number of training process data transmitted during a batch of training of the target AI model; the round of training of the target AI model The total number; the data characteristics of the training process data; the data set of the training process data; the target AI model; the model processing instructions of the target AI model; and the identification of the wireless communication network to which the target AI model belongs. Aligning the configurations between the first communication device and the second communication device.

Combined with any design of the above examples, the transceiver module is also used to send first feedback information to the second communication device, where the first feedback information is used to indicate at least one of the following: the reception status of the data in the target data, the The reception status includes successful reception or unsuccessful reception; the number of successfully received data in the target data; the ratio of the number of successfully received data in the target data to the number of data in the target data; wherein the target data includes the first data and/or the first instruction information. Feedback whether the target data is successfully received.

Combined with any design of the above examples, the transceiver module is also used to send first feedback information to the second communication device, where the first feedback information is used to indicate at least one of the following: the reception status of the data in the second data. , the reception status includes successful reception or unsuccessful reception; the number of successfully received data in the second data; the proportion of the number of successfully received data in the second data to the number of data in the target data.

In a possible implementation, the transceiver module is specifically configured to receive the first feedback information sent by the first processing device.

Combined with any design of the above examples, the transceiver module is also used to send third indication information to the second communication device, and the third indication information is used to determine receipt of the first indication information and/or sent by the second communication device. Or the first data; wherein the third indication information includes at least one of the following: identification information of the first communication device; the ability of the first communication device to perform model processing, where the model processing includes model training, model verification, and model testing. At least one of; whether the first communication device is connected to the first processing device; whether the first communication device has the ability to connect to the first processing device; the communication quality measurement result of the first communication device; the The location information of the first communication device; the battery information of the first communication device; the moving speed of the first communication device; and the first communication device is in a charging state.

Combined with any design of the above examples, the transceiver module is also used to send fourth indication information to the first processing device; wherein the fourth indication information is used to indicate that the first communication device is not in the second communication device. communication range; or, the fourth indication information is used to indicate that the fourth communication device is not within the communication range of the second communication device.

In a possible implementation, the fourth indication information is used to indicate that the fourth communication device is not within the communication range of the second communication device, and the transceiver module is also used to receive the fourth indication sent by the second communication device. information.

In a possible implementation, the fourth indication information is used to indicate that the fourth communication device is not within the communication range of the second communication device, and the transceiver module is also used to send the fourth communication device to the first processing device. Corresponding first instruction information.

The beneficial effects of the communication device provided by the above-mentioned third aspect and each possible implementation manner of the above-mentioned third aspect can be referred to the beneficial effects brought by the above-mentioned first aspect and each possible implementation manner of the first aspect, here No longer.

In a fourth aspect, embodiments of the present application provide a communication device. The communication device is a second communication device. The communication device includes: a processing module for determining first indication information, the first indication information being used to indicate first data. The data characteristics, and/or, are used to indicate that the data set to which the first data belongs is subjected to model processing of the first AI model. The first data is the data received by the first communication device from the second communication device. The model processing It includes at least one of model training, model verification, and model testing; and a transceiver module for sending the first indication information to the first communication device.

In a possible implementation, the data feature includes at least one of the following: sample category, the sample category includes input Category and/or expected output category; the data of the input category is used to input the first AI model, the data of the expected output category is used to indicate the expected output of the first AI model when the data of the input category is input, the input The data of the category corresponds to the data of the expected output category; the configuration information when the first data is generated; the data dimension of the first data; the data type of the first data; the data quantity of the first data; the first The amount of data transferred each time when data is transferred in batches.

Optionally, the second communication device is a network device.

Optionally, the first communication device is a terminal device.

Combined with any design of the above examples, the target AI model includes the first AI model and the second AI model. The second AI model is model processed by the second target device. The second target device includes the second communication device or The second processing device is communicatively connected to the second communication device. The transceiver module is also used to receive second indication information sent by the first communication device. The second indication information is used to indicate at least one of the following: data of the third data. Characteristics; the data quantity of the fourth data; the data set to which the third data belongs; the second AI model; the model processing of the second AI model; the identification of the wireless communication network to which the second AI model belongs; the third data in The round in the training process of the target AI model; the batch of the third data in the training process of the target AI model; wherein the third data is the data sent by the first communication device to the second communication device , the fourth data includes the third data.

Combined with any design of the above examples, the transceiver module is also used to send configuration information to the first communication device; or, the transceiver module is also used to receive configuration information sent by the first communication device; wherein the configuration information is Indicates at least one of the following: the duration of the transmission of the training process data; the number of batch samples, used to determine the number of training process data transmitted during a batch of training of the target AI model; the number of training process data for the target AI model training The total number of rounds; the data characteristics of the training process data; the data set of the training process data; the target AI model; the model processing instructions of the target AI model; and the identification of the wireless communication network to which the target AI model belongs.

Combined with any design of the above examples, the transceiver module is also used to receive first feedback information sent by the first communication device, where the first feedback information is used to indicate at least one of the following: the reception status of the data in the target data, The receiving status includes successful reception or unsuccessful reception; the number of successfully received data in the target data; the ratio of the number of successfully received data in the target data to the number of data in the target data; wherein the target data includes the first data and/or or the first instruction information.

Combined with any design of the above examples, the transceiver module is also used to receive first feedback information sent by the first communication device, where the first feedback information is used to indicate at least one of the following: the reception status of the data in the second data. , the reception status includes successful reception or unsuccessful reception; the number of successfully received data in the second data; the proportion of the number of successfully received data in the second data to the number of data in the target data.

Combined with any design of the above examples, the transceiver module is also used to receive third indication information sent by the first communication device, and the third indication information is used to determine to send the first indication information and/or to the first communication device. first data; wherein the third indication information includes at least one of the following: identification information of the first communication device; the ability of the first communication device to perform model processing, where the model processing includes model training, model verification, and model testing. at least one of; whether the first communication device is connected to the first processing device; whether the first communication device has the ability to connect to the first processing device; the communication quality measurement result of the first communication device; the third Location information of a communication device; battery information of the first communication device; moving speed of the first communication device; and the first communication device is in a charging state.

The beneficial effects of the communication device provided by the above-mentioned fourth aspect and each possible implementation manner of the above-mentioned fourth aspect can be referred to the beneficial effects brought by the above-mentioned first aspect and each possible implementation manner of the first aspect, here No longer.

In a fifth aspect, embodiments of the present application provide a communication device, including: a processor and a memory. The memory is used to store a computer program. The processor is used to call and run the computer program stored in the memory to perform the following steps: Methods in the second aspect, the third aspect, the fourth aspect or each possible implementation manner.

In a sixth aspect, embodiments of the present application provide a chip, including: a processor, configured to call and run computer instructions from a memory, so that a device installed with the chip executes the first aspect, the second aspect, or each possible implementation. method within the method.

In a seventh aspect, embodiments of the present application provide a computer-readable storage medium for storing computer program instructions. The computer program causes the computer to execute the method in the first aspect, the second aspect, or each possible implementation manner.

In an eighth aspect, embodiments of the present application provide a computer program product, including computer program instructions, which cause a computer to execute the method in the first aspect, the second aspect, or each possible implementation manner.

Description of drawings

Figure 1 is an architectural schematic diagram of a mobile communication system applied in an embodiment of the present application;

Figure 2a is a schematic diagram of the training process of a bilateral model provided by the embodiment of the present application;

Figure 2b is a schematic diagram of the training process of another bilateral model provided by the embodiment of the present application;

FIG3 is a schematic diagram of a system architecture for data transmission provided in an embodiment of the present application;

Figure 4 is a schematic interactive flow chart of the data transmission method provided by the embodiment of the present application;

Figure 5 is a schematic interactive flow chart of the data transmission method provided by the embodiment of the present application;

Figure 6 is a schematic interactive flow chart of the data transmission method provided by the embodiment of the present application;

Figure 7 is a schematic block diagram of a communication device provided by an embodiment of the present application;

Figure 8 is another schematic block diagram of a communication device provided by an embodiment of the present application.

Detailed ways

The technical solutions in this application will be described below with reference to the accompanying drawings.

Figure 1 is a schematic architectural diagram of a mobile communication system applied in an embodiment of the present application. As shown in Figure 1, the mobile communication system includes a core network device 110, a network device 120 and at least one terminal device (terminal device 130 and terminal device 140 in Figure 1). Terminal equipment is connected to network equipment through wireless means, and network equipment is connected to core network equipment through wireless or wired means. Core network equipment and network equipment can be independent and different physical devices, or the functions of the core network equipment and the logical functions of the network equipment can be integrated on the same physical device, or part of the core network can be integrated into one physical device. Device functionality and functionality of some network devices. Terminal equipment can be fixed or movable. Figure 1 is only a schematic diagram. The communication system may also include other network equipment, such as wireless relay equipment and wireless backhaul equipment, which are not shown in Figure 1 . The embodiments of the present application do not limit the number of core network equipment, network equipment, and terminal equipment included in the mobile communication system.

In this embodiment of the present application, the network device may be any device with wireless transceiver functions. Network equipment includes but is not limited to: evolved node B (evolved Node B, eNB), home base station (for example, home evolved NodeB, or home Node B, HNB), baseband unit (baseband unit, BBU), wireless fidelity (wireless) Fidelity, WiFi) system access point (access point, AP), wireless relay node, wireless backhaul node, transmission point (transmission point, TP) or sending and receiving point (transmission and reception point, TRP), you can also It is a mobile switching center and a device that assumes base station functions in device-to-device (D2D), vehicle-to-everything (V2X), and machine-to-machine (M2M) communications. etc., it can also be a gNB in a 5G system, one or a group (including multiple antenna panels) of antenna panels of a base station in a 5G system, or it can also be a network node that constitutes a gNB or a transmission point, such as a BBU, or, Distributed unit (distributed unit, DU), etc., the embodiments of this application do not specifically limit this.

In some deployments, gNB may include centralized units (CUs) and DUs. CU and DU implement some functions of gNB respectively, and CU and DU can communicate through the F1 interface. The gNB may also include an active antenna unit (AAU). AAU can realize some physical layer processing functions, radio frequency processing and active antenna related functions.

It can be understood that the network device may be a device including one or more of a CU node, a DU node, and an AAU node. In addition, the CU can be divided into network equipment in the access network (radio access network, RAN), or the CU can be divided into network equipment in the core network (core network, CN), which is not limited in this application.

In the embodiment of this application, the terminal equipment may also be called user equipment (UE), 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.

The terminal device may be a device that provides voice/data connectivity to the user, such as a handheld device, a vehicle-mounted device, etc. with wireless connectivity capabilities. Currently, some examples of terminals include: mobile phones, tablets, computers with wireless transceiver functions (such as laptops, handheld computers, etc.), customer-premises equipment (CPE), smart phones Point of sale (POS) machines, mobile internet devices (MID), virtual reality (VR) devices, augmented reality (AR) devices, and industrial control (industrial control) Wireless terminals, wireless terminals in self-driving, wireless terminals in remote medical, wireless terminals in smart grid, wireless terminals in transportation safety, smart cities Wireless terminals and smart homes in (smart city) Wireless terminals, cellular phones, cordless phones, session initiation protocol (SIP) phones, wireless local loop (WLL) stations, personal digital assistants (PDAs) in (smart home) , handheld devices with wireless communication functions, computing devices or other processing devices connected to wireless modems, vehicle-mounted devices, wearable devices, terminal devices in 5G networks or terminal devices in systems evolved after 5G, etc.

Network equipment and terminal equipment can communicate through licensed spectrum (licensed spectrum), unlicensed spectrum (unlicensed spectrum), or both licensed spectrum and unlicensed spectrum. Network equipment and terminal equipment can communicate through spectrum below 6G, or through spectrum above 6G, or they can communicate using spectrum below 6G and spectrum above 6G at the same time. The embodiments of the present application do not limit the spectrum resources used between the network device and the terminal device.

It should be understood that this application does not limit the specific forms of network equipment and terminal equipment.

The communication method provided by this application can be applied to various communication systems, such as: Long Term Evolution (LTE) system, 5G mobile communication system, 6G and other mobile communication systems evolved after 5G. Among them, the 5G mobile communication system or the 6G mobile communication system may include non-standalone networking (non-standalone, NSA) and/or independent networking (standalone, SA).

The communication method provided by this application can also be applied to machine type communication (MTC), machine-to-machine communication long-term evolution technology (Long Term Evolution-machine, LTE-M), device to device (device to device, D2D) network , machine to machine (M2M) network, Internet of things (IoT) network or other networks.

In order to introduce AI/ML technology into the wireless communication system, the AI/ML model can be deployed in a terminal device (such as the terminal device 130 and/or 140 in Figure 1) and/or a network device (such as the network device 120 in Figure 1) (The following is unified into an AI model for ease of expression). Network performance or efficiency can be improved by running the AI model. Among them, AI models can include unilateral models and bilateral models:

1. Unilateral model: Deployed on terminal devices or network devices, the device deployed by the AI model performs inference by running the AI model. For example, the AI model is deployed on network devices as a positioning model. The terminal device reports location-related wireless channel measurement information to the network device. The network device inputs the location-related wireless channel measurement information into the AI model to locate the terminal device.

2. Two-sided (AI/ML) model: It is a pair of AI sub-models that perform joint reasoning. Each AI sub-model is deployed on network equipment and terminal equipment respectively to achieve joint reasoning. Among them, joint reasoning refers to AI reasoning that is jointly performed across terminal devices and network devices. For example, part of the reasoning is performed by the terminal device and the rest of the reasoning is performed by the network device. For example, the terminal device can perform data compression through the AI sub-model deployed by itself, and send the compressed data to the network device. The network device can decompress the data through the AI sub-model deployed by itself to restore the data sent by the terminal device. Data, in this case, the AI sub-model deployed on the terminal device can be implemented as an AI encoder, and the AI sub-model deployed on the network device can be implemented as an AI decoder. The AI encoder and AI decoder constitute a bilateral model.

Taking channel state information (CSI) encoding feedback as an example, the terminal uses an AI encoder to encode the CSI and sends the encoded CSI to the network device. The network device uses the corresponding AI decoder to decode, thus obtaining Restored CSI.

It should be noted that the unilateral model and the bilateral model are only named for convenience of distinction, and their specific naming does not limit the scope of protection of the present application.

It can be understood that before the terminal device and/or network device runs the AI model, the AI model can be obtained through model training, and the AI model can also be subject to model verification and/or model testing to ensure the accuracy of the AI model; After the terminal device and/or network device runs the AI model or during the process of running the AI model, model optimization can be performed through model training. Of course, the accuracy of the optimized AI model can also be determined through model verification and/or model testing.

Taking the implementation of the CSI coding feedback process through the bilateral model as an example, two exemplary descriptions of the model training process in the communication system are given below with reference to Figure 2a and Figure 2b.

Example 1: First, one side of the network device and the terminal device iteratively trains its own deployed AI sub-model, and then outputs the training data to the other side to train its own deployed AI sub-model based on the received training data.

Referring to Figure 2a, for example, the network device first performs training on the CSI reconstruction part (i.e., AI decoding). An initial AI encoder and an initial AI decoder are deployed in the network device. The network device inputs the training sample CSI into the initial AI encoder. , and the AI encoder encodes the CSI, and outputs the encoded CSI (i.e. z') to the initial AI decoder. The initial AI decoder decodes z' to obtain the restored CSI', which is determined based on the loss function. The difference between CSI and CSI' (that is, the loss value). When the training completion conditions are met (such as the loss value converging to the preset value or the number of iterations reaching the preset number), the network device ends the training process and obtains the AI encoder f ₃ and AI decoder f ₂ ; after the network device training is completed, the training sample CSI is input into the AI encoder f ₃ to obtain the encoded CSI (i.e. z), and the network device sends (CSI, z) as training data to the terminal device ;Further, the terminal device performs training on the CSI generation part (i.e., AI encoding) based on the training data. For example, the terminal device inputs the training sample CSI in the training data into the initial encoder, and the encoder outputs z", and determines it based on the loss function The difference between z” and z (that is, the loss value), when the training completion conditions are met (such as the loss value converging to the preset value or the number of iterations reaching the preset number), the terminal device ends the training process and obtains the AI encoder f ₁ . Among them, the encoder f ₃ is only used for the model processing process. In the actual AI model running stage, the network device directly inputs the encoded CSI sent by the terminal device into the AI decoder f ₂ to obtain the restored CSI.

This application does not limit the network device to always perform training of the AI sub-model first. For example, the terminal device first performs training on the CSI generation part (i.e., AI coding) to obtain the AI encoder f ₁ and the AI decoder f ₄ (not shown in the figure) that does not participate in the model operation, and then inputs the training sample CSI into the AI coding The decoder f ₁ obtains the encoded CSI (i.e., z), and sends (z, CSI) as training data to the network device, so that the network device is trained to obtain the AI decoder f ₂ based on the training data.

Example 2: Network devices and terminal devices jointly perform iterative training on their respective deployed AI sub-models. Network equipment and terminal equipment exchange training process data in forward propagation (FP) and back propagation (BP), and cyclically train AI sub-models on both sides.

Referring to Figure 2b, for example, an initial AI encoder is deployed in the terminal device, and an initial decoder is deployed in the network device. The terminal device inputs the training sample CSI into the initial AI encoder to obtain the encoded CSI (i.e., z); During the FP process, the terminal device sends z to the network device. The network device inputs the received z to the initial decoder to obtain the recovered CSI', and determines the difference between CSI and CSI' based on the loss function (that is, the loss value); in the BP process, the network device sends the derivative value (partial derivative value) of the loss value to z to the terminal device, so that the terminal device can adjust the parameters of the encoder based on the training process data fed back by the BP process; with the above process Execute multiple iterations. When the training completion conditions are met (such as the loss value converging to the preset value or the number of iterations reaching the preset number), the terminal device and network device end the training process and obtain the AI encoder f ₁ deployed in the terminal device. and AI decoder f ₂ deployed in network equipment.

When training a unilateral model in a communication system, if the AI model is deployed in a terminal device, the network device can send training data to the terminal device. The training data can include data of the input category and/or data of the expected output category, The data of the expected output category is used to indicate the expected output after the input category data is calculated by the AI model. Similarly, if the AI model is deployed in a network device, the terminal device can send the above training data to the network device.

The model verification and model testing process is similar to the above training process and will not be described in detail here. The difference is that the data interacted between the network device and the terminal device during the model verification process is used to implement model verification and can be called verification data or verification process data. , The data interacted between the network device and the terminal device during the model testing process is used to implement the model testing and can be called test data or test process data. In the following, for the convenience of description, model training, model verification, model testing, etc. are collectively referred to as model processing.

It should be noted that the embodiments of this application do not limit the data transmitted between the network device and the terminal device during model processing.

This application provides a data transmission method that transmits instruction information between a terminal device and a network device to instruct the above model processing process, thereby supporting the implementation of the model processing process.

Figure 3 is a schematic diagram of a data transmission system architecture provided by an embodiment of the present application. As shown in FIG. 3 , the system 200 at least includes a network device 210 and N terminal devices (such as 220-1 to 220-N), where N is a positive integer. Exemplarily, the network device 210 may communicate with at least one terminal device among N terminal devices (such as 220-1 to 220-N) to perform model processing on the above-mentioned network device and/or terminal device side.

In some embodiments, considering that model processing of AI models is a computationally intensive task, the computing power of the terminal device is limited, and performing calculation-intensive tasks will have a greater impact on the user experience, so the first device 230 can perform model processing on the terminal device side. In this case, the network device 210 communicates with the first device 230 through N terminal devices (such as 220-1 to 220-N) to collaboratively complete model processing. Further, when performing model processing on the AI model, a large amount of data (such as the above training process data) needs to be exchanged between the network device 210 and the first device 230. In order to reduce the risk of the terminal device acting as a relay device receiving data from the network device 210 and sending it to When the first device 230 sends data, the impact on user experience can be achieved by using at least two terminal devices (for example, N is greater than 1) as relays to realize data interaction between the network device 210 and the first device 230 .

It should be understood that each of the N terminal devices communicates with the network device independently. For example, the communication between the terminal device 220-1 and the network device 210, and the communication between the terminal device 220-2 and the network device 210 are independent. , both of which have no timing constraints. Similarly, the N terminal devices independently communicate with the first device.

The first device 230 may be a device associated with N terminal devices (such as 220-1 to 220-N). For example, it may be a device set by the manufacturer of the N terminal devices (such as 220-1 to 220-N). It is set by the chip manufacturer used by the terminal equipment (such as 220-1 to 220-N), or jointly set by the manufacturers of N terminal equipment (such as 220-1 to 220-N) or the chip manufacturers used. The first device can be composed of multiple devices to jointly complete the model processing task. Correspondingly, the N terminal devices (such as 220-1 to 220-N) may be from the same manufacturer or the same model, or the chips deployed by the N terminal devices (such as 220-1 to 220-N) may be from the same manufacturer or The same model, or N terminal devices (such as 220-1 to 220-N) deploy the same AI model, etc.

For similar considerations, the network device 210 can also perform model processing on the AI model deployed on the network device 210 through the second device 240 connected to it. In this case, N terminal devices (such as 220-1 to 220- N) can communicate with the second device 240 through the network device 210. This application does not limit the number of network devices connected to the second device 240.

Both the first device 230 and the second device 240 can be any electronic device with processing capabilities, such as a terminal device, a server (such as an over-the-top (OTT) server), etc.

It should be understood that information exchange between the network device and the terminal device can be based on the protocol of any of the above communication systems, for example, information exchange can be conducted through Radio Resource Control messages (Radio Resource Control, RRC). Between the terminal device and the first device, or between the network device and the second device, information can be exchanged based on an OTT communication protocol, such as: hyper text transfer protocol (HTTP), transmission control protocol (transmission control protocol) , TCP), User Datagram Protocol (UDP), Internet Protocol (IP).

In order to facilitate understanding of the embodiments of this application, the following points are explained.

First, in the embodiments shown below, the first, second, third and various numerical numbers are only for convenience of description and are not used to limit the scope of the embodiments of the present application. For example, distinguish between different communication devices, models, data, etc.

Second, the "protocol" involved in the embodiments of this application may refer to a standard protocol in the communication field, which may include, for example, LTE protocol, NR protocol, and related protocols applied in future communication systems. This application does not limit this.

Third, “at least one” refers to one or more, and “plurality” refers to two or more. "And/or" describes the association of associated objects, indicating that there can be three relationships, for example, A and/or B, which can mean: A exists alone, A and B exist simultaneously, and B exists alone, where A, B can be singular or plural. The character "/" generally indicates that the related objects are in an "or" relationship.

The communication method provided by the embodiment of the present application will be described below with reference to the accompanying drawings.

It should be understood that the following is only for convenience of understanding and explanation, and the method provided by the embodiment of the present application is mainly explained by taking the interaction between the first communication device and the second communication device as an example. The first communication device may be a terminal device or a component in a terminal device, such as a chip, a chip system, or other functional module capable of calling and executing a program. The second communication device may be a network device or a component in a network device, such as Chips, chip systems, or other functional modules that can call programs and execute them. For example, the first communication device may be 130 or 140 in Figure 1, and the network device may be 120 in Figure 1, or the terminal device may be any one of 220-1 to 220-N in Figure 3, and the network device is 210 in Figure 3; alternatively, the first communication device can be a network device or a component in a network device, and the second communication device can be a terminal device or a component in a terminal device. For example, the first communication device can be a network device in Figure 1 The network device 120, the second communication device can be the terminal device 130 or 140 in Figure 3, or the first communication device is the network device 210 in Figure 3, and the second communication device is the terminal device 220-1 to 220 in Figure 3 Any of 220-N.

In some embodiments, the interaction between the first communication device and the first processing device will be taken as an example to illustrate the method proposed in the embodiments of the present application. The method provided is explained. When the first communication device is implemented as a terminal device or a component in a terminal device, the first processing device can be implemented as the above-mentioned first device or a component in the first device; when the first communication device is implemented as a network device or a component in a network device , the first processing device may be implemented as a second device or a component in the second device.

In some embodiments, the method provided by the embodiments of the present application will be described with reference to the interaction between the second communication device and the second processing device as an example. When the second communication device is implemented as a terminal device or a component in a terminal device, the second processing device can be implemented as the above-mentioned first device or a component in the first device; when the second communication device is implemented as a network device or a component in a network device , the second processing device may be implemented as a second device or a component in the second device.

Figure 4 is a schematic interactive flow chart of the data transmission method 300 provided by the embodiment of the present application. As shown in Figure 4, the method 300 may include at least some of the following steps.

S310. The second communication device sends the first instruction information to the first communication device. Correspondingly, the first communication device receives the first indication information sent by the second communication device.

S321. The first communication device performs model processing on the first AI model based on the first data.

S322-1. The first communication device sends the first instruction information to the first processing device.

S322-2: The first processing device performs model processing on the first AI model based on the first data.

S331. The first processing device sends the first feedback information to the first communication device. Correspondingly, the first communication device receives the first feedback information sent by the first processing device.

S332. The first communication device sends the first feedback information to the second communication device. Correspondingly, the second communication device receives the first feedback information sent by the first communication device.

The method 300 provided by the embodiment of the present application includes at least S310 and S321, or at least includes S310, S322-1 and S322-2. Among them, when S321 is executed, S322-1 and S322-2 are not executed; when S322-1 and S322-2 are executed, S321 is not executed, so S321 is represented by a dotted line in Figure 4 . S331 is an optional process, so it is represented by a dotted line in Figure 4 .

The first AI model may be the unilateral model in the foregoing example, or the AI sub-model deployed on one side of the device in the bilateral model in the foregoing example.

The first data may be data sent by the second communication device to the first communication device. This application does not limit the order in which the second communication device sends the first data and the first instruction information to the first communication device. The first data may be training data or training process data in the model training process of the aforementioned example. Taking the CSI feedback shown in Figure 2a as an example, the first data may be training data (CSI,z), or it may be training data ( The CSI in CSI,z), or it can be the z in the training data (CSI,z). Taking the CSI feedback shown in Figure 2b as an example, the first data can be the training process data. Of course, the first data can also be verification process data in the model verification process, or the first data can also be test process data in the model testing process.

For convenience of description below, without distinguishing whether the execution subject of the model processing is the first communication device or the first processing device, the first communication device and/or the first processing device may be referred to as the first target device. It can be understood that when the first communication device performs model processing of the first AI model based on the first data, the first target device is the first communication device; or when the first communication device sends the first AI model to the first processing device In the case of the first data and the first instruction information, the first target device is the first processing device.

In this embodiment of the present application, the information indicated by the first indication information is used to assist in implementing the model processing process of the first AI model. The following provides an example rather than a limiting description of the content indicated by the first indication information.

Example 1: The first indication information is used to indicate the data characteristics of the first data:

Data characteristics may include at least one of the following:

1) Sample category, which sample category includes at least one of an input category and an expected output category; the data of the input category is used to input the first AI model, and the data of the expected output category is used to indicate that the first AI model is inputting the input The data of the category is the expected output, and the data of the input category corresponds to the data of the expected output category. Taking the CSI encoding feedback in Figure 2a as an example, the first data may include CSI, which is the data of the input category, or the first data may include z, which is the data of the expected output category, or the first data may include CSI and z, that is (CSI,z). It can be understood that when the first data only includes data of the input category or data of the expected output category in the training data, the remaining training data may be preset in the first target device, or may be obtained from the first target device. obtained from other devices.

2) The configuration information when the first data is generated may be related to the configuration of the second communication device and related to the generation of the first data. Taking CSI feedback as an example, the configuration information when the first data is generated may include antenna configuration, subband configuration, feedback bit number configuration, etc. of the second communication device. The configuration information when the first data is generated may reflect information such as the data dimension and data quantity of the generated first data.

3) The data dimension of the first data is the number of independent parameters (mathematics) or system degrees of freedom (physics) required to describe the state of the object, reflecting the organization of the data in the first data. Taking CSI feedback as an example, if the first data includes B1 CSI data, then each CSI data is a data sample, and B1 is the data number of the first data mentioned below; if each CSI data sample is C1×C2× The three-dimensional matrix of C3, then C1, C2, and C3 here are the data dimensions of the first data. Still taking CSI feedback as an example, if the first data includes B1 (CSI,z) data, then CSI and z each have a corresponding set of data dimensions, and the first data can have multiple sets of data dimensions. The data dimension of the first data may also be indicated in an indirect manner. Taking CSI feedback as an example, the first indication information may be information indicating that the CSI is a channel matrix H or that the CSI is a feature vector to achieve indirection of the data dimension of the first data. instruct.

4) The data type of the first data. For example: 8-bit integer (INT8), 16-bit floating-point (FP16), single-precision floating point, etc.

5) The data quantity of the first data. The number of data can be the number of all data, or the number of data with data samples as the basic unit. Taking CSI feedback as an example, if the first data includes B1 CSI data, it can be understood that each CSI data is a data sample. If each CSI data sample is a three-dimensional matrix of C1×C2×C3, then the data of the first data The quantity can be B1×C1×C2×C3 (the number of all data), or B1 (the number of data with data samples as the basic unit). When the first communication device executes the above-mentioned S321, the first data is all data used for model processing sent by the second communication device to the first communication device; when the first communication device executes the above-mentioned S322-1, then the first processing When the device executes S322-2 above, the first data is all data sent by the second communication device to the first processing device through the first communication device. It can be understood that if the second communication device communicates with the first processing device through at least two first communication devices, the first data is sent by the second communication device to the first processing device through at least two first communication devices for modelling. For part of all the processed data (such as the second data), the second communication device also sends the remaining part of the second data to the first processing device through other first communication devices among at least two first communication devices.

6) The number of data transmitted each time when the first data is transmitted in batches. It can be understood that when the amount of data is large, the second communication device may transmit the first data in batches. Optionally, the data stream transmitted each time may be the same or different, which is not limited in this application.

In the above example one, the second communication device sends the first indication information to the first communication device to indicate the data characteristics of the first data to facilitate receiving and using the first data to perform model processing on the first AI model.

Example 2: The first indication information is used to indicate the data quantity of the second data. As mentioned before, the second data is all data sent by the second communication device to the first processing device for model processing through at least two first communication devices. Therefore, Example 2 is suitable for performing the above S322-1 and S322-2, that is, the first communication device serves as a relay to send the first instruction information sent by the first communication device to the first processing device, and the first processing device performs the processing according to the The first indication information and the second data sent by at least two first communication devices are subjected to model processing.

Generally, when it is necessary to perform model processing on the first AI model through the first processing device, the first instruction information indicates the data quantity of the second data. However, this application is not limited to this. When the first communication device performs model processing on the first AI model, the first indication information may also indicate the data quantity of the second data.

It should be noted that when the first communication device performs model processing on the first AI model, the second data and the first data may be the same data. In this case, the data quantity of the second data indicated by the first indication information is equal to The data quantity of the first data is the same; or, the second communication device can use other communication devices (the communication device can be any terminal equipment) as a relay device to send the second data except the first data to the first communication device. data, so that the first communication device performs model processing on the first AI model based on the second data. In this case, the first indication information may indicate the data quantity of the second data.

Example 3: The first instruction information is used to instruct the data set to which the first data belongs to perform model processing of the first AI model.

In Example 3, the first indication information may include at least one of the following:

1) The type of the data set to which the first data belongs. The type of the data set includes at least one of a training set, a validation set, and a test set. So that the first target device receives the first data and determines what kind of model processing the first data is used for.

2) The identification of the data set to which the first data belongs. The identification of the data set to which the first data indicated by the first indication information may be the identification of the data set saved in the first target device, and the first target device may add the first data to the data set; or it may be the first target An identification of a data set that does not yet exist on the device and that the first target device can create and add the first data to.

3) The proportion of data quantity in each type of data set. After receiving the first data, the first target device may allocate the first data to different types of data sets based on the data quantity ratio of each type of data set, or allocate each data in the first data to different types of data sets. , so that each type of data set meets the data quantity ratio. One possible situation is that the network device does not indicate the type of data set to which the first data belongs. Then, after receiving all the data (such as the second data), the first target device allocates each data in the second data to different types. of data sets.

4) First incremental data information. The first incremental data information is used to indicate adding the first data to the data set to which the first data belongs. The data set to which the first data belongs is the data set saved in the first target device. For example, it can be through the first instruction information. The identity of the data set is determined.

In the above example three, the first communication device can determine the use of the first data in the model processing process, such as for model training, model verification or model testing, based on the data set to which the first data indicated by the first indication information belongs. Then corresponding model processing is performed on the first AI model.

Example 4: The first instruction information is used to instruct the first AI model. In order to facilitate the first target device to determine the object of model processing.

In Example 4, the first indication information may include at least one of the following:

1) Identification of the first AI model. Based on the identification of the first AI model, a corresponding relationship between the first data and the first AI model is established, so that the first target device determines that the object for model processing based on the first data is the first AI model.

2) Model parameters of the first AI model. Taking data compression transmission as an example, the model parameters of the first AI model may include, for example, a coding compression ratio and a quantization parameter. The quantization parameter may include information indicating whether to quantize and/or the degree of quantization.

3) Instruct training or optimization training to obtain the information of the first AI model. Among them, training to obtain the first AI model refers to performing model training on the initial model to obtain the first AI model; optimizing training to obtain the first AI model refers to optimizing or adjusting (tuning or finetuning) based on the trained first AI model. ).

If the first information indicates that the first AI model is obtained through optimized training, the first indication information may also indicate a basic model identifier, and the basic model identifier may also be a basic model structure identifier and/or a basic model parameter identifier. In this case, the first target device can train the basic model based on the first data to obtain the first AI model.

Example 5: The first indication information is used to indicate the identity of the wireless communication network to which the first AI model belongs, so that the model-processed first AI model can run under the wireless communication network. The identification of the wireless communication network may include, for example, the identification of the wireless communication network equipment manufacturer and/or the identification of the wireless communication network operator.

Example 6: The first instruction information is used to instruct model processing of the first AI model, so that the first target device determines whether to start model processing of the first AI model, and/or what type of processing is performed on the first AI model. model processing, etc.

If the first instruction information indicates model verification of the first AI model, the first instruction information is also used to indicate at least one verification indicator and/or a threshold corresponding to at least one verification indicator. For example: verification indicators can include minimum mean square error (MSE) and its deformations, cosine similarity (CS) and its deformations, cross entropy, Euclidean distance, recall rate, precision rate, F1 score ( F1Score) to understand the performance of the AI model at the current stage

If the first instruction information indicates performing model testing of the first AI model, the first instruction information is also used to indicate at least one test indicator and/or a threshold corresponding to at least one test indicator. For example: test indicators can include minimum mean square error (MSE) and its deformations, cosine similarity (CS) and its deformations, cross entropy, Euclidean distance, recall rate, precision rate, F1 score ( F1Score) to ensure that the first AI model trained by the first target device meets the performance requirements.

Example 7: The first instruction information is used to instruct the model processing of the first AI model to be started when at least one of the following meets the corresponding threshold:

The ratio of the number of received data in the first data to the number of data in the first data;

The ratio of the number of received data in the second data to the number of data in the second data;

The number of received data in the first data;

The number of received data in the second data;

The duration of receiving the first data, or in other words, the waiting time for receiving the first data;

The duration of receiving the second data, or in other words, the waiting time for receiving the second data.

It can be understood that if the first target device fails to receive only a small amount of data, it is still possible to train and obtain the first AI model that meets the performance requirements, or to complete the task of model verification or model testing. Through this instruction, it is possible to prevent the first target device from waiting blindly and being unable to start training in time to obtain the first AI model that meets the performance requirements.

Some or all of the various examples of the above first indication information can be combined with each other to obtain more possible first indication information. In other words, the first indication information can include part or all of the information in the above examples of the first indication information. Any kind of first indication information in any of the above examples can be sent as independent indication information, or can be combined with each other into one or more indication information and sent, which is not limited in this application.

It should also be understood that the first indication information in the above various examples may also be agreed in whole or in part by an agreement or agreed in advance.

Therefore, in the embodiment of the present application, the first communication device can implement model processing of the first AI model based on the instruction of the first instruction information, providing a reliable method for data transmission in the communication system to support the execution of the model processing. solution.

In some embodiments, the method 300 provided by the embodiment of this application also includes the above-mentioned S332. The first communication device sends the first feedback information to the second communication device to indicate the reception of the first data and/or the first indication information. For the convenience of description, target data is used to represent the first data and/or the first instruction information in the following. Instructions. As a first example, the first feedback information may be used to indicate at least one of the following:

1) The reception status of the data in the target data. The reception status includes successful reception or unsuccessful reception. Refers to indicating whether each data in the target data is successfully received, for example, indicating whether the first data is successfully received, indicating whether the identification of the first AI model is successfully received, indicating whether the identification of the data set to which the first data belongs is Was successfully received, whether the sample category was successfully received, etc.

2) The number of successfully received data in the target data;

3) The ratio of the number of successfully received data in the target data to the number of data in the target data;

It should be noted that when the above S321 is executed, whether the target data is successfully received refers to whether the target data is successfully received by the first communication device; when the above S322-1 and S322-2 are executed, whether the target data is successfully received. Successful reception refers to whether the target data is successfully received by the first processing device.

As a second example, when the above-mentioned S322-1 and S322-2 are executed in the embodiment of the present application, the first feedback information may be used to indicate at least one of the following:

1) The reception status of the data in the second data;

2) The number of successfully received data in the second data;

3) The ratio of the number of successfully received data in the second data to the number of data in the target data.

The first feedback information in the above-mentioned first example and the above-mentioned second example may be combined with each other. Various contents indicated in the first feedback information may be combined and sent as one first feedback information or multiple first feedback information, or may be sent as one first feedback information respectively, which is not limited in this application.

Optionally, the first feedback information includes a bitmap, and the bits in the bitmap are used to indicate the reception status of the target data and/or the second data. For example, each bit in the bitmap may respectively correspond to the first data, the identifier of the first AI model, the identifier of the data set to which the first data belongs, and the sample category. When the bit corresponding to the first data is "0", it indicates that the first One data reception fails. When the bit is "1", it indicates that the first data reception is successful. It should be understood that this application does not limit the correspondence between bit values and reception success or failure. For example, a bit of "0" may indicate reception success, and a bit of "1" may indicate reception failure.

In the case of executing the above S322-1 and S322-2, whether the target data is successfully received refers to whether the target data is successfully received by the first processing device. In some embodiments, the first processing device may send first feedback information to the first communication device (eg, S331 above). The first feedback information has been described in the above examples and will not be described again here. In other embodiments, the first communication device can learn whether the target data is successfully received by the first processing device through a communication protocol, such as the message confirmation mechanism of the transmission control protocol (transmission control protocol, TCP), that is, based on the acknowledgment character (acknowledge character) character, ACK), it can be known whether the target data is successfully received by the first processing device, without the need for the first processing device to send the first feedback information to the first communication device as in the above example. That is, there is no need to execute S331 in Figure 4 .

It should be understood that the first feedback information in the above various examples may be agreed in whole or in part by an agreement or agreed in advance.

In this embodiment, the first communication device sends first feedback information to the second communication device to feedback whether the target data is successfully received. Collection improves the reliability of data transmission, thereby making the model processing process more reliable.

Figure 5 is a schematic interactive flow chart of the data transmission method 400 provided by the embodiment of the present application. As shown in Figure 5, the method 400 may include at least some of the following steps.

S410: Configuration alignment is performed between the first communication device and the second communication device.

S420. The first communication device sends the second instruction information to the second communication device. Correspondingly, the second communication device receives the second indication information sent by the first communication device.

S431. The second communication device performs model processing of the second AI model based on the third data.

S432-1. The second communication device sends the second instruction information to the second processing device. Correspondingly, the second processing device receives the second indication information sent by the second communication device.

S432-2: The second processing device performs model processing on the second AI model based on the third data.

S440. The second processing device sends the second feedback information to the second communication device. Correspondingly, the second communication device receives the second feedback information sent by the second processing device.

S450. The second communication device sends the second feedback information to the first communication device. Correspondingly, the first communication device receives the second feedback information sent by the second communication device.

S460. The second communication device sends the first instruction information to the first communication device. Correspondingly, the first communication device receives the first indication information sent by the second communication device.

S471. The first communication device performs model processing on the first AI model based on the first data.

S472-1. The first communication device sends the first instruction information to the first processing device.

S472-2: The first processing device performs model processing on the first AI model based on the first data.

S481. The first processing device sends the first feedback information to the first communication device. Correspondingly, the first communication device receives the first feedback information sent by the first processing device.

S482. The first communication device sends the first feedback information to the second communication device. Correspondingly, the second communication device receives the first feedback information sent by the first communication device.

It should be noted that the method 400 provided in this embodiment is suitable for model processing of a bilateral model, and the bilateral model may implement the model processing process based on the method shown in Figure 2b. In this embodiment, the first AI model and the second AI model are included in the target AI model. For example, the first AI model and the second AI model may be sub-models of the target AI model respectively.

In the above S410, the configuration alignment between the first communication device and the second communication device can be achieved by the first communication device sending configuration information to the second communication device, or by the second communication device sending configuration information to the first communication device. Configuration information is implemented, or it can be agreed in the agreement or agreed in advance. For example, the first communication device may query the capability determination configuration information supported by the second communication device; or the second communication device may query the capability determination configuration information supported by the first communication device.

For example, the configuration information may include at least one of the following:

1) The waiting time for data transmission during training is t. That is, when the waiting time of the first communication device and/or the second communication device waiting for data transmission from the other party exceeds t, the currently ongoing model training process of the target AI model is released to avoid waiting indefinitely and affecting device performance.

2) Batch sample size (batch size), used to determine the number of training process data transmitted during the model training process of a batch of the target AI model, which can be used to align the model of the target AI model. The amount of data for each interaction during training

3) The total number of rounds (epochs) of the target AI model training. When the training rounds of the target AI model exceed the total number of rounds, the training process is stopped.

4) Data characteristics of training process data. The data characteristics indicating the first data in the previous example are similar and will not be described again here.

5) Data set of training process data. It is similar to the data set indicating the first data in the previous example, and will not be described again here.

6) Target AI model. It is similar to indicating the first AI model in the previous example and will not be described again here.

7) Model processing instructions for the target AI model. It is similar to indicating the first AI model in the previous example and will not be described again here.

8) The identification of the wireless communication network to which the target AI model belongs. It is similar to the identifier indicating the wireless communication network to which the first AI model belongs in the previous example, and will not be described again here.

It should be noted that the items included in the above configuration information can be combined into one or more configuration information for sending, or or are sent as independent configuration information.

In this embodiment, S460 to S482 are respectively similar to S310 to S332 in FIG. 4 . However, when the target AI model is trained based on the joint training method in Figure 2b, the first instruction information can also be used to indicate the epoch and/or batch of the first data in the training process of the target AI model. (batch). For example, the first indication information may include round information and/or batch information, the round information may be an epoch number, and the batch information may be a batch number. The training process of the target AI model includes multiple rounds of training processes. One round of model training based on multiple training data (such as all or part of the training data in the training data set) can be called a round. Each round of training The process includes multiple batches of training processes. During the training process of each batch of the target AI model, training process data is transmitted between the first communication device and the second communication device. The training process data includes the first data and the second communication device. Three data.

In addition, the first indication information can also be used to indicate the epoch and/or batch of the first data in the model verification process of the target AI model; the first indication information can also be used to indicate the epoch and/or batch of the first data in the model testing process of the target AI model. The above embodiment is illustrated using the application scenario of the bilateral model as an example, but it is still applicable to the application scenario of the aforementioned unilateral model. For example, the first indication information can be used to indicate the epoch and/or batch of the first data in the model processing process of the first AI model. In other words, the first indication information can indicate the epoch and/or batch of the first data when used for model processing.

It should be noted that the third data is data sent by the first communication device to the second communication device. This application does not limit the order in which the first communication device sends the third data and the second instruction information to the second communication device. The third data may be the training process data in the model training process of the aforementioned example. Taking the CSI feedback shown in Figure 2b as an example, the third data may be the training process data z. Of course, the third data can also be verification process data in the model verification process, or the first data can also be test process data in the model testing process.

The fourth data includes the above-mentioned third data, and the fourth data is all data sent by the first communication device to the second processing device for model processing through at least two second communication devices.

Regarding the above S420, it should be noted that the second indication information can be used to indicate at least one of the following:

1) Data characteristics of the third data. The data characteristics indicating the first data in the previous example are similar and will not be described again here.

2) The data quantity of the fourth data. The data quantity indicating the second data is similar to that in the previous example, and will not be described again here.

3) The data set to which the third data belongs. It is similar to indicating the data set to which the first data belongs in the previous example, and will not be described again here.

4) Second AI model. It is similar to indicating the first AI model in the previous example and will not be described again here.

5) Model processing of the second AI model. It is similar to the model processing indicating the first AI model in the previous example, and will not be described again here.

6) The identification of the wireless communication network to which the second AI model belongs. It is similar to the identifier indicating the wireless communication network to which the first AI model belongs in the previous example, and will not be described again here.

7) The round of the third data in the training process of the target AI model. It is similar to the round in the first instruction information and will not be described again here.

8) Batch information of the third data during the training process of the target AI model. It is similar to the batch information in the first instruction information and will not be described again here.

In this embodiment, the second communication device can implement model processing of the second AI model based on the instruction of the second instruction information, thereby providing a reliable method for performing data transmission in the communication system to support the execution of model processing of the bilateral model. s solution.

It should be noted that when S431 is executed, S432-1 and S432-2 are not executed; when S432-1 and S432-2 are executed, S431 is not executed, so S431 is represented by a dotted line in Figure 4 . S440 is an optional process, so it is represented by a dotted line in Figure 4 .

Among them, S431 to S450 have the same or similar implementation methods as S321 to S332 respectively. For the sake of simplicity, they will not be described again.

Based on the above-mentioned embodiment shown in Figure 4 or Figure 5, the embodiment of the present application also provides a solution for determining the terminal device for relay transmission, and a management solution for the terminal device to leave the service. This embodiment is combined with FIG. 6 , taking the second communication device as a network device as an example, and explains this embodiment on the basis of the embodiment shown in FIG. 4 .

It should also be understood that this application is not limited to the process of iterative training, first performing model processing on the second AI model based on the second instruction information, and then performing model processing on the first AI model based on the first instruction information. For example, model processing may be performed on the first AI model based on the first instruction information, and then model processing may be performed on the second AI model based on the second instruction information.

Figure 6 is a schematic interactive flow chart of the data transmission method 500 provided by the embodiment of the present application. As shown in Figure 6, the method 500 may include:

S510: The first communication device sends third instruction information to the second communication device, and accordingly, the second communication device receives the third instruction information sent by the first communication device. The third indication information is used to determine to receive the first indication information and/or the first data sent by the second communication device. In other words, the second communication device can determine whether the first communication device is used for relay transmission based on the third indication information. Implement communication between the second communication device and the first processing device. Wherein, the third instruction information includes at least one of the following:

1) Identification information of the first communication device. For example: terminal manufacturer identification, chip identification used, terminal model identification, terminal stakeholder identification, AI model identification used, etc. When the identities of the first communication devices are the same, the same AI model can be used and serve for AI model joint training.

2) The ability of the first communication device to perform model processing.

3) Whether the first communication device is connected to the first processing device. If the first communication device is connected to the first processing device, the first communication device has a basis for implementing relay transmission.

4) Whether the first communication device has the ability to connect to the first processing device. The first communication device has the ability to connect with the first processing device, indicating that the first communication device has the ability to relay transmission to achieve communication between the second communication device and the first processing device.

5) Communication quality measurement results of the first communication device. For example: signal to interference plus noise ratio (SINR), reference signal receiving power (RSRP), received signal strength indication (RSSI), reference signal receiving quality ( reference signal receiving quality, RSRQ), received signal code power (received signal code power, RSCP). The second communication device may select the first communication device with better signal quality based on the communication quality measurement result of the first communication device to ensure the data transmission effect.

6) The location information of the first communication device can help select the first communication device with better communication stability. For example, if the first communication device is on the highway, it may quickly leave the current serving cell through cell reselection, so it is not suitable for use. in relay transmission.

7) Battery information of the first communication device. The second communication device may select the first communication device with better connection stability based on the battery information of the first communication device. The battery information includes at least one of the remaining battery power, battery capacity, and battery model. For example, a first communication device whose remaining power is lower than a threshold cannot be used for relay transmission.

8) The moving speed of the first communication device. The second communication device may select a first communication device with higher communication stability with the second communication device based on the movement speed of the first communication device. For example, a first communication device moving at high speed may quickly leave the current serving cell through cell reselection. , not suitable for relay transmission.

9) The first communication device is in a charging state. For example, the first communication device that is being charged is less affected in the user's power experience and is more suitable for relay transmission.

In this embodiment, the second communication device evaluates the capability of the terminal equipment in the serving cell for relay transmission based on the third indication information reported by the first communication device, and determines at least one first communication device that can be used for relay transmission. , which improves the reliability of the model processing process in communication systems.

In some embodiments, method 500 may include:

S530: The first communication device sends fourth instruction information to the first processing device, and accordingly, the first processing device receives the fourth instruction information sent by the first communication device.

In the first implementation manner, the fourth indication information may be used to indicate that the first communication device itself is not within the communication range of the second communication device, or that the first communication device has left the serving cell provided by the second communication device. The first communication device may proactively notify the first processing device after sensing that it is not within the communication range of the second communication device.

In the second implementation manner, the fourth indication information is used to indicate that other terminal equipment (such as the fourth communication device) is not within the communication range of the second communication device, or that the fourth communication device has left the service provided by the second communication device. Service community.

In the above second implementation manner, method 500 may also include:

S520. The second communication device sends the fourth instruction information to the first communication device. Correspondingly, the first communication device receives the second communication device. The fourth instruction information sent by the messaging device. After sensing that the fourth communication device is not within the communication range of the second communication device itself, the second communication device notifies the first processing device through relay transmission by the first communication device.

Further, after receiving the fourth indication information, the first processing device can feed back to the second communication device the reception status of data relayed and transmitted by a communication device that is out of the communication range (such as the first communication device or the fourth communication device).

Optionally, in the case where the fourth indication information indicates that the first communication device itself is not within the communication range of the second communication device, the first processing device may use any other communication device for relay transmission (such as the fifth communication device). ) communicates with the second communication device; when the fourth indication information indicates that other terminal equipment (such as the fourth communication device) is not within the communication range of the second communication device, the first processing device can communicate with the second communication device through the first communication device. The communication device communicates. Alternatively, the first processing device may wait until the second communications device re-instructs the communications device to relay the transmission.

It should be noted that the above-mentioned S510 and S520 have no timing constraints with other steps in Figure 6, but are executed in response to the trigger event of the communication device being separated.

In this embodiment, when the communication device for relay transmission leaves the communication range of the second communication device, the communication device for relay transmission is adjusted and the transmission status is aligned to reduce the impact on the model processing process when the communication device for relay transmission leaves the communication range. The impact improves the stability of communication, thereby improving the stability of the model processing process of the AI model.

Figure 7 is a schematic block diagram of a communication device provided by an embodiment of the present application. As shown in Figure 7, the device 600 may include: a transceiver module 610 and a processing module 620.

Optionally, the communication device 600 may correspond to the first communication device in the above method embodiment.

Among them, the transceiver module 610 can be used to receive first indication information sent by a second communication device, and the first indication information is used to indicate data characteristics of the first data, and/or to indicate that the data set to which the first data belongs performs model processing of the first AI model, and the first data is data received by the transceiver module from the second communication device, and the model processing includes at least one of model training, model verification, and model testing; the processing module 620 can be used to perform model processing of the first AI model based on the first data; or, the transceiver module 610 is also used to send the first indication information to the first processing device.

It should be understood that the specific processes and effects of each module performing the above corresponding steps have been described in detail in the above method embodiments, and will not be described again for the sake of brevity.

Optionally, the communication device 600 may correspond to the second communication device in the above method embodiment.

The processing module 620 may be used to determine the first indication information, which is used to indicate the data characteristics of the first data, and/or is used to indicate the data set to which the first data belongs to perform the first AI model. Processing, the first data is data received by the first communication device from the second communication device, and the model processing includes at least one of model training, model verification, and model testing; the transceiver module 610 can be used to send the first communication data to the first communication device. The device sends the first instruction information.

It should be understood that the specific processes and effects of each unit performing the above corresponding steps have been described in detail in the above method embodiments, and will not be described again for the sake of brevity.

The transceiver module 610 in the communication device 600 may be implemented by a transceiver, for example, may correspond to the transceiver 710 in the communication device 700 shown in FIG. 8 , and the processing module 620 in the communication device 600 may be implemented by at least one processor. , for example, may correspond to the processor 720 in the communication device 700 shown in FIG. 8 .

When the communication device 600 is a chip or chip system configured in a communication device (such as a terminal device or a network device), the transceiver module 610 in the communication device 600 can be implemented through an input/output interface, a circuit, etc., the communication device 600 The processing module 620 in can be implemented by a processor, microprocessor or integrated circuit integrated on the chip or chip system.

Figure 8 is another schematic block diagram of a communication device provided by an embodiment of the present application. As shown in FIG. 8 , the communication device 700 may include: a transceiver 710 , a processor 720 and a memory 730 . Among them, the transceiver 710, the processor 720 and the memory 730 communicate with each other through internal connection paths. The memory 730 is used to store instructions, and the processor 720 is used to execute the instructions stored in the memory 730 to control the transceiver 710 to send signals and /or receive a signal.

It should be understood that the communication device 700 may correspond to the first communication device or the second communication device in the above method embodiment, and may be used to perform various steps performed by the first communication device or the second communication device in the above method embodiment. /or process. Optionally, the memory 730 may include read-only memory and random access memory and provide instructions and data to the processor. A portion of the memory may also include non-volatile random access memory. The memory 730 can be a separate device or integrated into the processor 720 . The processor 720 may be used to perform instructions stored in the memory 730, and when the processor 720 performs instructions stored in the memory, the processor 720 is used to perform the above-mentioned operations corresponding to the first communication device or the second communication device. Various steps and/or processes of method embodiments.

Optionally, the communication device 700 is the first communication device in the previous embodiment.

Optionally, the communication device 700 is the second communication device in the previous embodiment.

Among them, the transceiver 710 may include a transmitter and a receiver. The transceiver 710 may further include an antenna, and the number of antennas may be one or more. The processor 720, the memory 730 and the transceiver 710 may be devices integrated on different chips. For example, the processor 720 and the memory 730 can be integrated in the baseband chip, and the transceiver 710 can be integrated in the radio frequency chip. The processor 720, the memory 730 and the transceiver 710 may also be devices integrated on the same chip. This application does not limit this.

Optionally, the communication device 700 is a component configured in the first communication device, such as a chip, a chip system, etc.

Optionally, the communication device 700 is a component configured in the second communication device, such as a chip, a chip system, etc.

The transceiver 720 may also be a communication interface, such as an input/output interface, a circuit, etc. The transceiver 720, the processor 710 and the memory 730 can be integrated in the same chip, such as a baseband chip.

This application also provides a processing device, including at least one processor, the at least one processor being used to execute a computer program stored in the memory, so that the processing device performs the first communication device or the second communication device in the above method embodiment. method of proceeding.

An embodiment of the present application also provides a processing device, including a processor and an input and output interface. The input and output interface is coupled to the processor. The input and output interface is used to input and/or output information. The information includes at least one of instructions and data. The processor is used to execute a computer program, so that the processing device performs the method performed by the first communication device or the second communication device in the above method embodiment.

An embodiment of the present application also provides a processing device, including a processor and a memory. The memory is used to store a computer program, and the processor is used to call and run the computer program from the memory, so that the processing device performs the method performed by the first communication device or the second communication device in the above method embodiment.

Embodiments of the present application also provide a communication system, including a first communication device and a second communication device. The first communication device is used to perform the method on the first communication device side in any of the above method embodiments. The second communication device The device is configured to perform the method on the second communication device side in any of the above method embodiments.

It should be understood that the above-mentioned processing device may be one or more chips. For example, the processing device may be a field programmable gate array (FPGA), an application specific integrated circuit (ASIC), or a system on chip (SoC), or It can be a central processing unit (CPU), a network processor (NP), a digital signal processing circuit (DSP), or a microcontroller unit , MCU), it can also be a programmable logic device (PLD) or other integrated chip.

During the implementation process, each step of the above method can be completed by instructions in the form of hardware integrated logic circuits or software in the processor. The steps of the method disclosed in conjunction with the embodiments of the present application can be directly embodied in a hardware processor to be completed, or can be completed using a combination of hardware and software modules in the processor. The software module can be located in random access memory, flash memory, read-only memory, programmable read-only memory or electrically erasable programmable memory, registers and other mature storage media in this field. The storage medium is located in the memory, and the processor reads the information in the memory and completes the steps of the above method in combination with its hardware. To avoid repetition, it will not be described in detail here.

It should be noted that the processor in the embodiment of the present application may be an integrated circuit chip with signal processing capabilities. During the implementation process, each step of the above method embodiment can be completed through an integrated logic circuit of hardware in the processor or instructions in the form of software. The above-mentioned processor may be a general-purpose processor, a digital signal processor (DSP), an application-specific integrated circuit (ASIC), a field programmable gate array (FPGA) or other programmable logic devices, discrete gate or transistor logic devices, or discrete hardware components. . Each method, step, and logical block diagram disclosed in the embodiments of this application can be implemented or performed. A general-purpose processor may be a microprocessor or the processor may be any conventional processor, etc. The steps of the method disclosed in conjunction with the embodiments of the present application can be directly implemented as a hardware decoding processor, or can be completed using a combination of hardware and software modules in the decoding processor. The software module can be located in random access memory, flash memory, read-only memory, programmable read-only memory or electrically erasable programmable memory, registers and other mature storage media in this field. The storage medium is located in the memory, and the processor reads the information in the memory and completes the steps of the above method in combination with its hardware.

It can be understood that the memory in the embodiment of the present application may be a volatile memory or a non-volatile memory, or may include both volatile and non-volatile memories. Among them, non-volatile memory can be read-only memory (ROM), programmable ROM (PROM), erasable programmable read-only memory (erasable PROM, EPROM), electrically removable memory. Erase electrically programmable read-only memory (EPROM, EEPROM) or flash memory. Volatile memory can be random access memory (RAM), which is used as an external cache. By way of illustration, but not limitation, many forms of RAM are available, such as static random access memory (SRAM), dynamic random access memory (DRAM), synchronous dynamic random access memory (synchronous DRAM, SDRAM), double data rate synchronous dynamic random access memory (double data rate SDRAM, DDR SDRAM), enhanced synchronous dynamic random access memory (enhanced SDRAM, ESDRAM), synchronous link dynamic random access memory (synchlink DRAM, SLDRAM) ) and direct memory bus random access memory (direct rambus RAM, DR RAM). It should be noted that the memory of the systems and methods described herein is intended to include, but is not limited to, these and any other suitable types of memory.

According to the method provided by the embodiment of the present application, the present application also provides a computer program product. The computer program product includes: computer program code. When the computer program code is run on a computer, it causes the computer to perform the first step in the above method embodiment. A method performed by a communication device or a second communication device.

According to the method provided by the embodiments of the present application, the present application also provides a computer-readable storage medium. The computer-readable storage medium stores program code. When the program code is run on a computer, it causes the computer to perform the above method embodiments. A method performed by the first communication device or the second communication device.

According to the method provided by the embodiments of the present application, the present application also provides a communication system, which may include the aforementioned first communication device and second communication device.

The terms "component", "module", "system", etc. used in this specification are used to refer to computer-related entities, hardware, firmware, a combination of hardware and software, software, or software in progress. For example, a component may be, but is not limited to, a process, processor, object, executable file, executable thread, program and/or computer running on a processor. Through the illustrations, both applications running on the computing device and the computing device may be components. One or more components can reside in a process and/or thread of execution, and a component can be localized on one computer and/or distributed between 2 or more computers. Additionally, these components may be implemented from various computer-readable media having various data structures stored thereon. A component may, for example, be based on a signal having one or more data packets (eg, data from two components interacting with another component, a local system, a distributed system, and/or a network, such as the Internet, which interacts with other systems via signals) Communicate through local and/or remote processes.

Those of ordinary skill in the art will appreciate that the units and algorithm steps of each example described in conjunction with the embodiments disclosed herein can be implemented with 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. Skilled artisans may implement the described functionality using different methods for each specific application, but such implementations should not be considered beyond the scope of this application.

Those skilled in the art can clearly understand that for the convenience and simplicity of description, the specific working processes of the systems, devices and units described above can be referred to the corresponding processes in the foregoing method embodiments, and will not be described again here.

In the several embodiments provided in this application, it should be understood that the disclosed systems, devices and methods can be implemented in other ways. For example, the device embodiments described above are only illustrative. For example, the division of units is only a logical function division. In actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated. to another system, or some features can be ignored, or not implemented. On the other hand, the coupling or direct coupling or communication connection between each other shown or discussed may be through some interfaces, and the indirect coupling or communication connection of the devices or units may be in electrical, mechanical or other forms.

A unit described as a separate component may or may not be physically separate. A component shown as a unit may or may not be a physical unit, that is, it may be located in one place, or it may be distributed to multiple network units. Some or all of the units can be selected according to actual needs to achieve the purpose of the solution of this embodiment.

In addition, each functional unit in various embodiments of the present application can be integrated into one processing module, each unit can exist physically alone, or two or more units can be integrated into one unit.

If the above functions are implemented in the form of software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the part that the technical solution of the present application essentially contributes or the technology Part of the technical solution can be embodied in the form of a software product. The computer software product is stored in a storage medium and includes a number of instructions to cause a computer device (which can be a personal computer, a server, or a second communication device, etc.) to perform All or part of the steps of the methods in various embodiments of the present application. The aforementioned storage media include: U disk, mobile hard disk, ROM, RAM, magnetic disk or optical disk and other media that can store program codes.

The above are only specific embodiments of the present application, but the protection scope of the present application is not limited thereto. Any person familiar with the technical field can easily think of changes or substitutions within the technical scope disclosed in the present application. should be covered by the protection scope of this application. Therefore, the protection scope of this application should be subject to the protection scope of the claims.

Claims

A data transmission method, characterized by including:

The first communication device receives the first instruction information sent by the second communication device;

The first communication device performs model processing of the first AI model based on first data, the first data being data received by the first communication device from the second communication device; or,

The first communication device sends the first instruction information to the first processing device;

Wherein, the first indication information is used to indicate at least one of the following:

Data characteristics of the first data, the first data being data received by the first communication device from the second communication device;

The data set to which the first data belongs is subjected to model processing of the first artificial intelligence AI model, and the model processing includes at least one of model training, model verification, and model testing;

The first data is used for rounds and/or batches when performing model processing;

A data quantity of second data, the second data including the first data and data received by the third communication device from the second communication device.
The method according to claim 1, characterized in that the data characteristics include at least one of the following:

Sample category, the sample category includes input category and/or expected output category; the data of the input category is used to input the first AI model, and the data of the expected output category is used to indicate that the first AI model is in The expected output when data of the input category is input, the data of the input category corresponding to the data of the expected output category;

Configuration information when the first data is generated;

The data dimension of the first data;

The data type of the first data;

The data quantity of the first data;

The number of data transmitted each time the first data is transmitted in batches.
The method according to claim 1 or 2, characterized in that the first indication information includes at least one of the following:

The type of data set to which the first data belongs, the type of data set includes at least one of a training data set, a verification data set, and a test data set;

The identification of the data set to which the first data belongs;

The proportion of data quantity in each type of data set;

Wherein, the first indication information includes an identification of a data set to which the first data belongs, and the data set to which the first data belongs is a data set saved in the first target device. The first indication information also includes a third data set. An incremental data information, the first incremental data information is used to indicate adding the first data to the data set; the first target device is the first communication device or the first processing device.
The method according to any one of claims 1 to 3, characterized in that the first instruction information is also used to indicate that the model processing of the first AI model is started when at least one of the following meets a corresponding threshold:

The ratio of the number of received data in the first data to the number of data in the first data;

The ratio of the number of received data in the second data to the number of data in the second data;

The number of received data in the first data;

The number of received data in the second data;

The duration of receiving the first data;

The duration of receiving the second data.
The method according to any one of claims 1 to 4, characterized in that the first communication device is a terminal device, and/or the second communication device is a network device.
The method according to any one of claims 1 to 5, characterized in that the first processing device is an upper layer OTT server on the network.
A data transmission method, characterized by including:

the second communication device determines the first indication information;

The second communication device sends the first instruction information to the first communication device;

Wherein, the first indication information is used to indicate at least one of the following:

Data characteristics of first data, the first data being data received by the first communication device from the second communication device;

The data set to which the first data belongs is subjected to model processing of the first artificial intelligence AI model, and the model processing includes at least one of model training, model verification, and model testing;

The round and/or batch in which the first data is used for model processing;

A data quantity of second data, the second data including the first data and data received by the third communication device from the second communication device;

Wherein, the first data is data sent by the second communication device to the first communication device.
The method according to claim 7, characterized in that the data characteristics include at least one of the following:

Sample category, the sample category includes input category and/or expected output category; the data of the input category is used to input the first AI model, and the data of the expected output category is used to indicate that the first AI model is in The expected output when data of the input category is input, the data of the input category corresponding to the data of the expected output category;

Configuration information when the first data is generated;

The data dimension of the first data;

The data type of the first data;

The data quantity of the first data;

The number of data transmitted each time the first data is transmitted in batches.
The method according to claim 7 or 8, characterized in that the first indication information includes at least one of the following:

The type of data set to which the first data belongs, the type of data set includes at least one of a training data set, a verification data set, and a test data set;

The identification of the data set to which the first data belongs;

The proportion of data quantity in each type of data set;

Wherein, the first indication information includes an identification of a data set to which the first data belongs, and the data set to which the first data belongs is a data set saved in the first target device. The first indication information also includes a third data set. An incremental data information, the first incremental data information is used to indicate adding the first data in the data set; the first target device is the first communication device or is related to the first communication device Connected to the first processing device.
The method according to any one of claims 7 to 9, characterized in that the first instruction information is also used to indicate that the model processing of the first AI model is started when at least one of the following meets a corresponding threshold:

The ratio of the number of received data in the first data to the number of data in the first data;

The ratio of the data quantity of received data in the second data to the data quantity of the second data;

The number of received data in the first data;

The number of received data in the second data;

The duration of receiving the first data;

The duration for receiving the second data.
The method according to any one of claims 7 to 9, characterized in that the first communication device is a terminal device, and/or the second communication device is a network device.
A communication device, characterized by including:

A transceiver module configured to receive the first instruction information sent by the second communication device;

A processing module configured to perform model processing of the first AI model based on the first data, which is the data received by the transceiver module from the second communication device; or,

The transceiver module is also configured to send the first indication information to the first processing device;

Wherein, the first indication information is used to indicate at least one of the following:

Data characteristics of the first data, which are data received by the transceiver module from the second communication device;

The data set to which the first data belongs is subjected to model processing of the first artificial intelligence AI model, and the model processing includes at least one of model training, model verification, and model testing;

The first data is used for rounds and/or batches when performing model processing;

A data quantity of second data, the second data including the first data and a third communication device receiving data from the second communication device received data.
The device according to claim 12, wherein the data characteristics include at least one of the following:

Sample category, the sample category includes input category and/or expected output category; the data of the input category is used to input the first AI model, and the data of the expected output category is used to indicate that the first AI model is in The expected output when data of the input category is input, the data of the input category corresponding to the data of the expected output category;

Configuration information when the first data is generated;

The data dimension of the first data;

The data type of the first data;

The data quantity of the first data;

The number of data transmitted each time the first data is transmitted in batches.
The device according to claim 12 or 13, characterized in that the first indication information includes at least one of the following:

The type of data set to which the first data belongs, the type of data set includes at least one of a training data set, a verification data set, and a test data set;

The identification of the data set to which the first data belongs;

The proportion of data quantity in each type of data set;

Wherein, the first indication information includes an identification of a data set to which the first data belongs, and the data set to which the first data belongs is a data set saved in the first target device. The first indication information also includes a third data set. An incremental data information, the first incremental data information is used to indicate adding the first data to the data set; the first target device is the communication device or the first processing device.
The device according to any one of claims 12 to 14, wherein the first instruction information is also used to indicate that the model processing of the first AI model is started when at least one of the following meets a corresponding threshold:

The ratio of the number of received data in the first data to the number of data in the first data;

The ratio of the number of received data in the second data to the number of data in the second data;

The number of received data in the first data;

The number of received data in the second data;

The duration of receiving the first data;

The duration for receiving the second data.
The device according to any one of claims 12 to 15, characterized in that the first processing device is an OTT server.
A communication device, characterized by including:

A processing module, used to determine the first indication information;

A transceiver module, configured to send the first indication information to the first communication device;

Wherein, the first indication information is used to indicate at least one of the following:

Data characteristics of the first data, the first data being the data received by the first communication device from the transceiver module;

The data set to which the first data belongs is subjected to model processing of the first AI model, and the model processing includes at least one of model training, model verification, and model testing;

The round and/or batch in which the first data is used for model processing;

The data quantity of second data, the second data including the first data and the data received by the third communication device from the transceiver module;

Wherein, the first data is data sent by the transceiver module to the first communication device.
The device according to claim 17, characterized in that the data characteristics include at least one of the following:

Sample category, the sample category includes input category and/or expected output category; the data of the input category is used to input the first AI model, and the data of the expected output category is used to indicate that the first AI model is in The expected output when data of the input category is input, the data of the input category corresponding to the data of the expected output category;

Configuration information when the first data is generated;

The data dimension of the first data;

The data type of the first data;

The data quantity of the first data;

The number of data transmitted each time the first data is transmitted in batches.
The device according to claim 17 or 18, characterized in that the first indication information includes at least one of the following:

The type of data set to which the first data belongs, the type of data set includes at least one of a training data set, a verification data set, and a test data set;

The identification of the data set to which the first data belongs;

The proportion of data quantity in each type of data set;

Wherein, the first indication information includes an identification of a data set to which the first data belongs, and the data set to which the first data belongs is a data set saved in the first target device. The first indication information also includes a third data set. An incremental data information, the first incremental data information is used to indicate adding the first data in the data set; the first target device is the first communication device or is related to the first communication device Connected to the first processing device.
The device according to any one of claims 17 to 19, wherein the first instruction information is also used to indicate that the model processing of the first AI model is started when at least one of the following meets a corresponding threshold:

The ratio of the number of received data in the first data to the number of data in the first data;

The ratio of the data quantity of received data in the second data to the data quantity of the second data;

The number of received data in the first data;

The number of received data in the second data;

The duration of receiving the first data;

The duration for receiving the second data.
A communication device, characterized in that it includes: a processor and a memory, the memory is used to store a computer program, the processor is used to call and run the computer program stored in the memory, and execute any of claims 1 to 11. The method described in one item.
A chip, characterized in that it includes: a processor, configured to call and run computer instructions from a memory, so that a device equipped with the chip executes the method according to any one of claims 1 to 11.
A computer-readable storage medium, characterized in that it is used to store computer program instructions, and the computer program causes the computer to execute the method according to any one of claims 1 to 11.
A computer program product, characterized by comprising computer program instructions, which cause a computer to execute the method according to any one of claims 1 to 11.