Communication method and related equipment

Abstract

The invention provides a communication method and related equipment, which are used for enabling the computing power of a communication node to be applied to artificial intelligence (AI) learning and can improve the implementation flexibility of different nodes. In the method, a first node receives first information, the first information indicates N pieces of configuration information, the N pieces of configuration information are used for configuring resources of N model parameters, and N is a positive integer; and the first node receives a first model parameter based on first configuration information in the N pieces of configuration information, wherein the first model parameter is used for determining a first neural network model corresponding to local information.

Description

A communication method and related equipment Technical Field

[0001] The present application relates to the field of communications, and in particular to a communication method and related equipment. Background Art

[0002] Wireless communication can be the transmission communication between two or more communication nodes without propagating through conductors or cables. The communication nodes generally include network devices and terminal devices.

[0003] Currently, in wireless communication systems, communication nodes generally possess both signal transceiver and computing capabilities. For example, network devices with computing capabilities primarily provide computing power to support signal transceiver capabilities (for example, calculating the time and frequency domain resources required to carry signals), enabling communication between the network device and other communication nodes.

[0004] However, in communication networks, communication nodes may have excess computing power beyond just supporting the aforementioned communication tasks. Therefore, how to utilize this computing power is a pressing technical issue.

[0005] Summary of the Invention

[0006] The present application provides a communication method and related equipment for enabling the computing power of communication nodes to be applied to artificial intelligence (AI) learning and improving the implementation flexibility of different nodes.

[0007] In a first aspect, the present application provides a communication method, which is executed by a first node, or the method is executed by some components in the first node (such as a processor, a chip or a chip system, etc.), or the method can also be implemented by a logic module or software that can realize all or part of the functions of the first node. In the first aspect and its possible implementation, the method is described as being executed by the first node. In this method, the first node receives first information, which indicates N configuration information, and the N configuration information is used to configure resources of N model parameters, where N is a positive integer; the first node receives a first model parameter based on the first configuration information in the N configuration information, and the first model parameter is used to determine a first neural network model corresponding to the local information.

[0008] Based on the above technical solution, the first information received by the first node indicates N configuration information, and the N configuration information is used to configure resources of N model parameters. Thereafter, the first node receives the first model parameter based on the resource configured by the first configuration information in the N configuration information, and the first node can determine the first neural network model corresponding to the local information based on the first model parameter. In other words, different first nodes can all obtain the neural network model corresponding to their respective local information based on the first information, and can subsequently participate in AI learning based on the locally obtained neural network model. Compared to the implementation method in which different nodes participate in AI learning based on randomized neural network models, in the above technical solution, different nodes can participate in AI learning based on the neural network model corresponding to their respective local information. Thus, when the communication node in the communication system acts as a node participating in AI learning, the computing power of the communication node can be applied to AI tasks in the learning system, and the implementation flexibility of different nodes can be improved.

[0009] In addition, the local information of different nodes is generally different. In the above technical solution, the first node can determine the neural network model corresponding to the local information based on the first information. To this end, different first nodes can participate in AI learning based on the neural network model adapted to the local information to improve the processing efficiency of the neural network model.

[0010] Optionally, AI learning may be replaced with machine learning, AI processing, AI neural network processing, etc.

[0011] It should be understood that the resources configured by the N configuration information include one or more of time domain resources, frequency domain resources, and space domain resources.

[0012] It should be understood that, among the N model parameters configured by the N configuration information, the model parameters are used to initialize the neural network model, or in other words, the model parameters include the initialization parameters of the neural network model. The initialization parameters may be parameters used to determine the use of the neural network model, or the initialization parameters may be parameters used to determine the start of neural network model training. In other words, for the first node, after the first node receives the first model parameters based on the first configuration information, the first node may perform initialization processing on the neural network model based on the first model parameters to obtain a first neural network model corresponding to the local information; or, the first node may use the first model parameters as the initialization parameters of the neural network model, and obtain the first neural network model after completing the local parameter initialization.

[0013] In a possible implementation manner of the first aspect, the local information includes at least one of a local resource overhead level, local AI capability information, and a local parameter scale.

[0014] Based on the above technical solution, local information can be implemented in a variety of ways to provide flexibility in solution implementation.

[0015] It should be understood that the local resource overhead level can be understood as the utilization rate of local computing resources, the utilization rate of local storage resources, the usage of local power consumption, etc. For example, the local computing resources may include resources of modules / devices such as a central processor unit (CPU), a microcontroller unit (MCU), or a graphics processing unit (GPU); the local storage resources may include resources of modules / devices such as a read-only memory (ROM) and a random access memory (RAM); the usage of local power consumption may include the health of a battery module, the current amount of power used by the battery module (or the percentage of the power), the current amount of power not used by the battery module (or the percentage of the power), etc.

[0016] It should be understood that local AI capability information may include local AI-related capability information, such as AI-related computing capability, storage capability, etc. Similarly, AI-related computing capability may include the computing capability of modules / devices such as CPU, MCU, and GPU, and AI-related storage capability may include the storage capability of modules / devices such as ROM and RAM.

[0017] It should be understood that the local parameter scale can be understood as the parameter scale of the local neural network model of the first node, including but not limited to the number of neural network layers of the neural network model, the number of neurons of the neural network model, the input and output of the neural network model, or the different computational complexity of the neural network model.

[0018] It should be noted that the local information of the first node can be implemented in a variety of ways. Accordingly, for the first node, after the first node receives the first model parameters, the first node can determine the first neural network model corresponding to the local information in a variety of ways.

[0019] In one implementation example, taking the example of the local information including the local resource overhead level, the local resource overhead level can be reflected by CPU utilization (or CPU occupancy); wherein the first node can determine the mapping relationship between different CPU utilizations and different neural network models through configuration or pre-configuration. Taking the example of the different neural network models including a first type of neural network model and a second type of neural network model, the first type of neural network model can be more complex than the second type of neural network model (for example, the scale of model parameters is larger, the number of neural network layers of the model is greater, etc.), and the mapping relationship can indicate that nodes with a CPU utilization of 0 to X% (X is greater than 0 and less than 100) correspond to the first type of neural network model, and indicate that nodes with a CPU utilization of X% to 100% correspond to the second type of neural network model. Thus, in the above technical solution, the first node can compare the local CPU utilization with X%. If the local CPU utilization is less than or equal to X%, the first neural network model obtained by the first node based on the first model parameters can be the first type of neural network model; if the local CPU utilization is greater than or equal to X%, the first neural network model obtained by the first node based on the first model parameters can be the second type of neural network model.

[0020] Optionally, in the above implementation example, the measurement of CPU utilization can be determined under the premise that the nodes execute the same or similar calculation parameters, that is, the CPU utilization is determined under relatively fair conditions. For example, for the training of two models, when the input parameters are the same, different nodes have different CPU utilizations required to complete a round of training. For example, when the CPU utilization of a certain node is low, it can be understood that the computing power of the CPU of the node is strong, and the initialization processing can be performed based on a larger scale of model parameters (or more complex model parameters); when the CPU utilization of a certain node is high, it can be understood that the computing power of the CPU of the node is weak, and the initialization processing can be performed based on a smaller scale of model parameters (or simpler model parameters). Accordingly, based on this difference, different model parameters can be configured for the different nodes, so that each node can obtain a differentiated (or personalized) initialized neural network model.

[0021] In another implementation example, taking the local information including local AI capability information as an example, the local AI capability information can be reflected by the storage capacity of the RAM; wherein the first node can determine the mapping relationship between different RAM storage capabilities and different neural network models by configuration or pre-configuration. Taking the different neural network models including the first type of neural network model, the second type of neural network model and the third type of neural network model as an example, the first type of neural network model can be more complex than the second type of neural network model (for example, the scale of model parameters is larger, the number of neural network layers of the model is greater, etc.), and the second type of neural network model can be more complex than the third type of neural network model; the mapping relationship can indicate that the node with a RAM storage capacity of 0 to Y gigabytes (GB) (Y is greater than 0) corresponds to the first type of neural network model, and indicates that the node with a RAM storage capacity greater than Y GB and less than or equal to Z GB (Z is greater than Y) corresponds to the second type of neural network model, and indicates that the node with a RAM storage capacity greater than Z GB corresponds to the third type of neural network model. Therefore, in the above technical solution, the first node can compare with Y and Z based on the storage capacity of the local RAM. When the storage capacity of the local RAM is less than or equal to Y GB, the first neural network model obtained by the first node based on the first model parameters can be a first-type neural network model; when the storage capacity of the local RAM is greater than Y GB and less than or equal to Z GB, the first neural network model obtained by the first node based on the first model parameters can be a second-type neural network model; when the storage capacity of the local RAM is greater than Z GB, the first neural network model obtained by the first node based on the first model parameters can be a third-type neural network model.

[0022] In a possible implementation manner of the first aspect, the model parameter includes at least one of a model weight, a model activation function, and a model bias.

[0023] Based on the above technical solution, the N model parameters of the resource transmission configured by the N configuration information include at least one of the above items to improve the flexibility of the solution implementation.

[0024] In a possible implementation manner of the first aspect, N is an integer greater than 1, and parameter scales of the N model parameters are different. The method further includes: the first node determining the first configuration information among the N configuration information based on the local information.

[0025] Based on the above technical solution, when the N configuration information indicated by the first information is greater than 1, the parameter scales of the N model parameters configured by the N configuration information are different, so that the first node can determine the first configuration information corresponding to the local information in the N configuration information based on the local information, so that the first node can obtain the model parameters corresponding to the local information based on the first configuration information.

[0026] It should be understood that the different parameter scales of the N model parameters can be understood as different requirements for the local information of the node. For example, if the local information includes the local resource overhead level, the model parameter with a larger parameter scale among the N model parameters requires the local resource overhead level to be below a threshold in order to provide more unused resources to process the model parameter with a larger parameter scale. For another example, if the local information includes local AI capability information, the model parameter with a larger parameter scale among the N model parameters requires the local AI capability information to be above a threshold in order to provide more AI resources to process the model parameter with a larger parameter scale.

[0027] Optionally, the parameter scales of the N model parameters are different and can be replaced by other descriptions, for example, the number of neural network layers corresponding to the N model parameters is different, the neurons corresponding to the N model parameters are different, the input and output corresponding to the N model parameters are different, the computational complexity of the N model parameters is different, or the computing power requirements of the N model parameters are different, or the storage capacity requirements of the N model parameters are different, etc.

[0028] In a possible implementation of the first aspect, the value of N is 1, wherein the first model parameter is used to configure a neural network model shared by at least two first nodes; the method also includes: the first node performs knowledge distillation processing on the neural network model shared by at least two first nodes based on the local information to obtain the first neural network model.

[0029] Based on the above technical solution, when the N configuration information indicated by the first information is one configuration information (i.e., the first configuration information), the first model parameters configured by the first configuration information are used to configure a neural network model shared by at least two first nodes, so that the at least two first nodes can obtain a common neural network model based on the first configuration information, and perform knowledge distillation processing on the common neural network model based on their respective local information to obtain a first neural network model corresponding to the local information.

[0030] It should be understood that the neural network model shared by at least two first nodes can be understood as a neural network model that can be recognized / processed by the at least two first nodes. Alternatively, the neural network model shared by at least two first nodes can be expressed as: a neural network model common to at least two first nodes, and a neural network model that can be recognized / processed by at least two first nodes.

[0031] In a possible implementation of the first aspect, the method further includes: the first node receives second information, the second information indicating M configuration information, the M configuration information being used to configure transmission resources of M model parameters, the parameter scales of the M model parameters being different, and M being an integer greater than 1; the first node receives a second model parameter based on the second configuration information in the M configuration information, wherein the parameter scale of the second model parameter is the same as the local parameter scale (or similar, for example, the difference between the parameter scale of the second model parameter and the local parameter scale is lower than a threshold).

[0032] Based on the above technical solution, the first node may also receive second information indicating M configuration information, and based on the second configuration information of the M configuration information, receive second model parameters with the same (or similar) parameter scale as the local parameters. In other words, different first nodes can each receive model parameters with the same (or similar) parameter scale as their respective local information via the second information, and perform AI processing (e.g., local training) based on the received model parameters.

[0033] It should be understood that the transmission resources configured by the M configuration information are used for different first nodes to receive one of the M model parameters, so that different first nodes can subsequently perform AI processing (such as local training) based on the received model parameters. Accordingly, the M model parameters can be understood as global model parameters of multiple first nodes, that is, the M model parameters can be replaced by M global model parameters.

[0034] Optionally, for any model parameter among the M model parameters, a gradient and / or a weight may be included. For example, the gradient and / or the weight are used for further processing of the initialized neural network model (e.g., the first neural network model described above), such as training the neural network model, updating the neural network model, etc.

[0035] Similarly, the parameter scales of the M model parameters are different and can be replaced by other descriptions, for example, the number of neural network layers corresponding to the M model parameters is different, the neurons corresponding to the M model parameters are different, the input and output corresponding to the M model parameters are different, the computational complexity of the M model parameters is different, or the computing power requirements of the M model parameters are different, or the storage capacity requirements of the M model parameters are different, etc.

[0036] In a possible implementation manner of the first aspect, any one of the M pieces of configuration information includes at least one of the following:

[0037] Indication information indicating parameter scales of model parameters configured by the one configuration information;

[0038] Indication information indicating the received power of the transmission resource configured by the one configuration information;

[0039] Indication information indicating a parameter type of a model parameter configured by the configuration information, where the parameter type includes a gradient and / or a weight;

[0040] Indication information indicating the frequency domain resources of the transmission resources configured by the one configuration information;

[0041] Indication information indicating the time domain resources of the transmission resources configured by the one configuration information.

[0042] Based on the above technical solution, any one of the M pieces of configuration information indicated by the second information may include at least one of the above items to enhance the flexibility of the solution implementation.

[0043] Optionally, any one of the N pieces of configuration information indicated by the first information includes at least one of the following:

[0044] Indication information indicating parameter scales of model parameters configured by the one configuration information;

[0045] Indication information indicating the received power of the transmission resource configured by the one configuration information;

[0046] Indication information indicating a parameter type of a model parameter configured by the configuration information, where the parameter type includes a gradient and / or a weight;

[0047] Indication information indicating the frequency domain resources of the transmission resources configured by the one configuration information;

[0048] Indication information indicating the time domain resources of the transmission resources configured by the one configuration information.

[0049] In a possible implementation of the first aspect, the method further includes: the first node receives third information, the third information indicating K configuration information, the K configuration information being used to configure transmission resources of K model parameters, the parameter scales of the K model parameters being different, and K being an integer greater than 1; the first node sends a third model parameter based on the third configuration information in the K configuration information, wherein the parameter scale of the third model parameter is the same as the local parameter scale (or similar, for example, the difference between the parameter scale of the third model parameter and the local parameter scale is lower than a threshold).

[0050] Based on the above technical solution, the first node may also receive third information indicating K configuration information, and based on the third configuration information of the K configuration information, receive third model parameters with the same (or similar) parameter scale as the local parameters. In other words, different first nodes can each send model parameters with the same (or similar) parameter scale as their respective local information via the third information, allowing the recipient of the model parameters to perform AI processing (e.g., parameter fusion processing) on ​​the model parameters sent by different first nodes.

[0051] It should be understood that the transmission resources configured by the K configuration information are used by different first nodes to transmit one of the K model parameters, so that the recipient of the model parameter can perform a fusion operation on the model parameters transmitted by the different first nodes. Accordingly, the K model parameters can be understood as the local model parameters of multiple first nodes, that is, the K model parameters can be replaced by K local model parameters.

[0052] Optionally, for any model parameter among the K model parameters, a gradient and / or a weight may be included. For example, the gradient and / or the weight are used for further fusion processing of the neural network model, such as deep fusion, weighted averaging, etc.

[0053] Similarly, the parameter scales of the K model parameters are different and can be replaced by other descriptions, for example, the number of neural network layers corresponding to the K model parameters is different, the neurons corresponding to the K model parameters are different, the input and output corresponding to the K model parameters are different, the computational complexity of the K model parameters is different, or the computing power requirements of the K model parameters are different, or the storage capacity requirements of the K model parameters are different, etc.

[0054] In a possible implementation manner of the first aspect, any one of the K pieces of configuration information includes at least one of the following:

[0055] Indication information indicating parameter scales of model parameters configured by the one configuration information;

[0056] Indication information indicating the transmit power of the transmission resource configured by the one configuration information;

[0057] Indication information indicating a parameter type of a model parameter configured by the configuration information, where the parameter type includes a gradient and / or a weight;

[0058] Indication information indicating the frequency domain resources of the transmission resources configured by the one configuration information;

[0059] Indication information indicating the time domain resource of the transmission resource configured by the one configuration information;

[0060] Indication information indicating the index of the transmission resource of the one configuration information in the transmission resources configured by the K configuration information.

[0061] It should be understood that when any one of the configuration information includes indication information indicating the index of the transmission resource of the one configuration information in the transmission resources configured by the K configuration information, it can be understood that the second node configures a transmission resource set (including at least one of time domain resources, frequency domain resources, and spatial domain resources) for the K configuration information, so that the first node can determine the corresponding transmission resource in the transmission resource set based on the index of the indication information.

[0062] Based on the above technical solution, any one of the K pieces of configuration information indicated by the third information may include at least one of the above items to enhance the flexibility of the solution implementation.

[0063] In a possible implementation of the first aspect, before the first node receives the first information, the method also includes: the first node receives fourth information, the fourth information is used to indicate the resource carrying the request information, and the request information is used to request the N configuration information; the first node sends the request information on the resource carrying the request information.

[0064] Based on the above technical solution, the first node can also receive fourth information for indicating the resources that carry the request information, and the request information is used to request the N configuration information, so that the first node can obtain the N configuration information through the first information after sending the request information based on the resources indicated by the fourth information.

[0065] In a possible implementation manner of the first aspect, the request information includes a first orthogonal sequence, where the first orthogonal sequence corresponds to the local information.

[0066] Optionally, the first orthogonal sequence is included in P orthogonal sequences, where the P orthogonal sequences correspond to different local information. In other words, different first nodes can each determine one of the P orthogonal sequences based on their respective local information and send the one of the orthogonal sequences to the second node, so that a recipient of the request information (e.g., the second node) can obtain the local information of each first node based on the received request information.

[0067] Optionally, when P is greater than 1, the different sequences can be orthogonal. That is, the P sequences can be expressed as P orthogonal sequences, so that the P orthogonal sequences can be identified by the receiver even when transmitted via the same resources. In other implementations, the P sequences can also be identified by the receiver in other ways, such as by staggering frequency or time domain resources. In other words, the aforementioned P orthogonal sequences can be replaced by P sequences, and the first orthogonal sequence can be replaced by the first sequence.

[0068] In addition, when the second node acts as the recipient of the request information, the second node can determine the local information of each first node based on one or more received orthogonal sequences, and can use the local information of each first node as the basis for determining at least one of N configuration information, M configuration information, and K configuration information, so that each configuration information sent by the second node can be adapted to the local information of each first node.

[0069] In addition, since different first nodes can indicate their own local parameter scales through the request information, overhead can be reduced.

[0070] Optionally, in addition to determining the resources carrying the first orthogonal sequence based on the fourth information, the first node may also determine the resources carrying the first orthogonal sequence in other ways (such as a preconfigured way), which is not limited here.

[0071] Optionally, the value of P may be greater than or equal to the value of N. For example, the first node may obtain the different parameter scales corresponding to the P orthogonal sequences respectively through configuration or pre-configuration, and the second node may determine the parameter scales of the respective first nodes by receiving request information from one or more first nodes. Accordingly, the second node may further generate the N configuration information based on the determined parameter scales of the respective first nodes. Generally, the P orthogonal sequences may have a redundant setting (i.e., P is greater than or equal to N) so that each first node can indicate its own local parameter scale through request information.

[0072] In a possible implementation of the first aspect, before the first node receives the first information, the method also includes: the first node receives fifth information, the fifth information is used to indicate the local information corresponding to the N model parameters respectively, and / or, the fifth information is used to indicate whether the first information includes the N configuration information.

[0073] Based on the above technical solution, the first node can also receive the fifth information so as to obtain relevant information of the N configuration information through the fifth information, and determine whether to receive the N configuration information based on the obtained information, which can avoid unnecessary detection by the first node to reduce overhead.

[0074] In a possible implementation of the first aspect, before the first node receives the third information, the method also includes: the first node receives sixth information, the sixth information is used to indicate the local information corresponding to the M model parameters respectively, and / or, the sixth information is used to indicate whether the third information includes the M configuration information.

[0075] Based on the above technical solution, the first node can also receive the sixth information so as to obtain relevant information of the M configuration information through the sixth information, and determine whether to receive the M configuration information based on the obtained information, which can avoid unnecessary detection by the first node to reduce overhead.

[0076] In a possible implementation of the first aspect, before the first node receives the second information, the method also includes: the first node receives seventh information, the seventh information is used to indicate the local information corresponding to the K model parameters respectively, and / or, the seventh information is used to indicate whether the second information includes the K configuration information.

[0077] Based on the above technical solution, the first node can also receive the seventh information so as to obtain relevant information of the K configuration information through the seventh information, and determine whether to receive the K configuration information based on the obtained information, which can avoid unnecessary detection by the first node to reduce overhead.

[0078] The second aspect of the present application provides a communication method, which is executed by a second node, or the method is executed by some components in the second node (such as a processor, a chip or a chip system, etc.), or the method can also be implemented by a logic module or software that can realize all or part of the functions of the second node. In the second aspect and its possible implementation, the method is described as being executed by the second node. In this method, the second node determines first information, and the first information indicates N configuration information, and the N configuration information is used to configure resources of N model parameters, where N is a positive integer; wherein one model parameter of the N model parameters is used to determine a neural network model corresponding to the local information of the first node; and the second node sends the first information.

[0079] Based on the above technical solution, the first information sent by the second node indicates N configuration information, and the N configuration information is used to configure resources of N model parameters. Thereafter, after receiving the first information, the first node can receive the first model parameters based on the resources configured by the first configuration information in the N configuration information, and the first node can determine the first neural network model corresponding to the local information based on the first model parameters. In other words, different first nodes can all obtain the neural network model corresponding to their respective local information based on the first information, and can subsequently participate in AI learning based on the locally obtained neural network model. Compared to the implementation method in which different nodes participate in AI learning based on randomized neural network models, in the above technical solution, different nodes can participate in AI learning based on the neural network model corresponding to their respective local information. Therefore, when the communication node in the communication system acts as a node participating in AI learning, the computing power of the communication node can be applied to AI tasks in the learning system, and the implementation flexibility of different nodes can be improved.

[0080] In addition, the local information of different nodes is generally different. In the above technical solution, the first node can determine the neural network model corresponding to the local information based on the first information. To this end, different first nodes can participate in AI learning based on the neural network model adapted to the local information to improve the processing efficiency of the neural network model.

[0081] Optionally, different first nodes may be distributed nodes, and the second node may be connected to one or more first nodes and serve as a central node (or control node, management node, etc.) of the one or more first nodes.

[0082] In a possible implementation manner of the second aspect, the local information includes at least one of a local resource overhead level, local AI capability information, and a local parameter scale.

[0083] Based on the above technical solution, local information can be implemented in a variety of ways to provide flexibility in solution implementation.

[0084] It should be understood that the local resource overhead level can be understood as the utilization rate of local computing resources, the utilization rate of local storage resources, the usage of local power consumption, etc. For example, the local computing resources may include resources of modules / devices such as a central processor unit (CPU), a microcontroller unit (MCU), or a graphics processing unit (GPU); the local storage resources may include resources of modules / devices such as a read-only memory (ROM) and a random access memory (RAM); the usage of local power consumption may include the health of a battery module, the current amount of power used by the battery module (or the percentage of the power), the current amount of power not used by the battery module (or the percentage of the power), etc.

[0085] It should be understood that local AI capability information may include local AI-related capability information, such as AI-related computing capability, storage capability, etc. Similarly, AI-related computing capability may include the computing capability of modules / devices such as CPU, MCU, and GPU, and AI-related storage capability may include the storage capability of modules / devices such as ROM and RAM.

[0086] In a possible implementation manner of the second aspect, the model parameter includes at least one of a model weight, a model activation function, and a model bias.

[0087] Based on the above technical solution, the N model parameters of the resource transmission configured by the N configuration information include at least one of the above items to improve the flexibility of the solution implementation.

[0088] In a possible implementation manner of the second aspect, N is an integer greater than 1, and parameter scales of the N model parameters are different.

[0089] Based on the above technical solution, when the N configuration information indicated by the first information is greater than 1, the parameter scales of the N model parameters configured by the N configuration information are different, so that the first node can determine the first configuration information corresponding to the local information in the N configuration information based on the local information, so that the first node can obtain the model parameters corresponding to the local information based on the first configuration information.

[0090] It should be understood that the different parameter scales of the N model parameters can be understood as different requirements for the local information of the node. For example, if the local information includes the local resource overhead level, the model parameter with a larger parameter scale among the N model parameters requires the local resource overhead level to be below a threshold in order to provide more unused resources to process the model parameter with a larger parameter scale. For another example, if the local information includes local AI capability information, the model parameter with a larger parameter scale among the N model parameters requires the local AI capability information to be above a threshold in order to provide more AI resources to process the model parameter with a larger parameter scale.

[0091] Optionally, the parameter scales of the N model parameters are different and can be replaced by other descriptions, for example, the number of neural network layers corresponding to the N model parameters is different, the neurons corresponding to the N model parameters are different, the input and output corresponding to the N model parameters are different, the computational complexity of the N model parameters is different, or the computing power requirements of the N model parameters are different, or the storage capacity requirements of the N model parameters are different, etc.

[0092] In a possible implementation of the second aspect, the value of N is 1.

[0093] Based on the above technical solution, when the N configuration information indicated by the first information is one configuration information (i.e., the first configuration information), the first model parameters configured by the first configuration information are used to configure a neural network model shared by at least two first nodes, so that the at least two first nodes can obtain a common neural network model based on the first configuration information, and perform knowledge distillation processing on the common neural network model based on their respective local information to obtain a first neural network model corresponding to the local information.

[0094] It should be understood that the neural network model shared by at least two first nodes can be understood as a neural network model that can be recognized / processed by the at least two first nodes. Alternatively, the neural network model shared by at least two first nodes can be expressed as: a neural network model common to at least two first nodes, and a neural network model that can be recognized / processed by at least two first nodes.

[0095] In a possible implementation of the second aspect, the method also includes: the second node sends second information, the second information indicates M configuration information, the M configuration information is used to configure transmission resources of M model parameters, the parameter scales of the M model parameters are different, and M is an integer greater than 1; the second node sends part or all of the M model parameters based on the M configuration information.

[0096] Based on the above technical solution, the first node may also receive second information indicating M configuration information, and based on the second configuration information of the M configuration information, receive second model parameters with the same (or similar) parameter scale as the local parameters. In other words, different first nodes can each receive model parameters with the same (or similar) parameter scale as their respective local information via the second information, and perform AI processing (e.g., local training) based on the received model parameters.

[0097] It should be understood that the transmission resources configured by the M configuration information are used for different first nodes to receive one of the M model parameters, so that different first nodes can subsequently perform AI processing (such as local training) based on the received model parameters. Accordingly, the M model parameters can be understood as global model parameters of multiple first nodes, that is, the M model parameters can be replaced by M global model parameters.

[0098] Optionally, for any model parameter among the M model parameters, a gradient and / or a weight may be included. For example, the gradient and / or the weight are used for further processing of the initialized neural network model (e.g., the first neural network model described above), such as training the neural network model, updating the neural network model, etc.

[0099] Similarly, the parameter scales of the M model parameters are different and can be replaced by other descriptions, for example, the number of neural network layers corresponding to the M model parameters is different, the neurons corresponding to the M model parameters are different, the input and output corresponding to the M model parameters are different, the computational complexity of the M model parameters is different, or the computing power requirements of the M model parameters are different, or the storage capacity requirements of the M model parameters are different, etc.

[0100] In a possible implementation manner of the second aspect, any one of the M pieces of configuration information includes at least one of the following:

[0101] Indication information indicating parameter scales of model parameters configured by the one configuration information;

[0102] Indication information indicating the received power of the transmission resource configured by the one configuration information;

[0103] Indication information indicating a parameter type of a model parameter configured by the configuration information, where the parameter type includes a gradient and / or a weight;

[0104] Indication information indicating the frequency domain resources of the transmission resources configured by the one configuration information;

[0105] Indication information indicating the time domain resources of the transmission resources configured by the one configuration information.

[0106] Based on the above technical solution, any one of the M pieces of configuration information indicated by the second information may include at least one of the above items to enhance the flexibility of the solution implementation.

[0107] In a possible implementation of the second aspect, the method also includes: the second node sends third information, the third information indicates K configuration information, the K configuration information is used to configure transmission resources of K model parameters, the parameter scales of the K model parameters are different, and K is an integer greater than 1; the second node receives part or all of the K model parameters based on the K configuration information.

[0108] Based on the above technical solution, the first node may also receive third information indicating K configuration information, and based on the third configuration information of the K configuration information, receive third model parameters with the same (or similar) parameter scale as the local parameters. In other words, different first nodes can each send model parameters with the same (or similar) parameter scale as their respective local information via the third information, allowing the recipient of the model parameters to perform AI processing (e.g., parameter fusion processing) on ​​the model parameters sent by different first nodes.

[0109] It should be understood that the transmission resources configured by the K configuration information are used by different first nodes to transmit one of the K model parameters, so that the recipient of the model parameter can perform a fusion operation on the model parameters transmitted by the different first nodes. Accordingly, the K model parameters can be understood as the local model parameters of multiple first nodes, that is, the K model parameters can be replaced by K local model parameters.

[0110] Optionally, for any model parameter among the K model parameters, a gradient and / or a weight may be included. For example, the gradient and / or the weight are used for further fusion processing of the neural network model, such as deep fusion, weighted averaging, etc.

[0111] Similarly, the parameter scales of the K model parameters are different and can be replaced by other descriptions, for example, the number of neural network layers corresponding to the K model parameters is different, the neurons corresponding to the K model parameters are different, the input and output corresponding to the K model parameters are different, the computational complexity of the K model parameters is different, or the computing power requirements of the K model parameters are different, or the storage capacity requirements of the K model parameters are different, etc.

[0112] In a possible implementation manner of the second aspect, any one of the K pieces of configuration information includes at least one of the following:

[0113] Indication information indicating parameter scales of model parameters configured by the one configuration information;

[0114] Indication information indicating the transmit power of the transmission resource configured by the one configuration information;

[0115] Indication information indicating a parameter type of a model parameter configured by the configuration information, where the parameter type includes a gradient and / or a weight;

[0116] Indication information indicating the frequency domain resources of the transmission resources configured by the one configuration information;

[0117] Indication information indicating the time domain resource of the transmission resource configured by the one configuration information;

[0118] Indication information indicating the index of the transmission resource of the one configuration information in the transmission resources configured by the K configuration information.

[0119] Based on the above technical solution, any one of the K pieces of configuration information indicated by the third information may include at least one of the above items to enhance the flexibility of the solution implementation.

[0120] In a possible implementation of the second aspect, before the second node sends the first information, the method also includes: the second node sends fourth information, the fourth information is used to indicate the resource carrying the request information, and the request information is used to request the N configuration information; the second node receives the request information on the resource carrying the request information.

[0121] Based on the above technical solution, the second node can also send fourth information for indicating the resources that carry the request information, and the request information is used to request the N configuration information, so that the first node can obtain the N configuration information through the first information after sending the request information based on the resources indicated by the fourth information.

[0122] In a possible implementation manner of the second aspect, the request information packet includes a first orthogonal sequence, and the first orthogonal sequence corresponds to the local information.

[0123] Optionally, the first orthogonal sequence is included in P orthogonal sequences, where the P orthogonal sequences correspond to different local information. In other words, different first nodes can each determine one of the P orthogonal sequences based on their respective local information and send the one of the orthogonal sequences to the second node, so that a recipient of the request information (e.g., the second node) can obtain the local information of each first node based on the received request information.

[0124] Optionally, when P is greater than 1, the different sequences can be orthogonal. That is, the P sequences can be expressed as P orthogonal sequences, so that the P orthogonal sequences can be identified by the receiver even when transmitted via the same resources. In other implementations, the P sequences can also be identified by the receiver in other ways, such as by staggering frequency or time domain resources. In other words, the aforementioned P orthogonal sequences can be replaced by P sequences, and the first orthogonal sequence can be replaced by the first sequence.

[0125] In addition, when the second node acts as the recipient of the request information, the second node can determine the local information of each first node based on one or more received orthogonal sequences, and can use the local information of each first node as the basis for determining at least one of N configuration information, M configuration information, and K configuration information, so that each configuration information sent by the second node can be adapted to the local information of each first node.

[0126] In addition, since different first nodes can indicate their own local parameter scales through the request information, overhead can be reduced.

[0127] Optionally, in addition to determining the resources carrying the first orthogonal sequence based on the fourth information, the first node may also determine the resources carrying the first orthogonal sequence in other ways (such as a preconfigured way), which is not limited here.

[0128] In a possible implementation of the second aspect, before the second node sends the first information, the method also includes: the second node sends fifth information, the fifth information is used to indicate the local information corresponding to the N model parameters respectively, and / or, the fifth information is used to indicate whether the first information includes the N configuration information.

[0129] Based on the above technical solution, the second node can also send the fifth information, so that the first node can obtain the relevant information of the N configuration information through the fifth information, and determine whether to receive the N configuration information based on the obtained information, which can avoid unnecessary detection by the first node to reduce overhead.

[0130] In a possible implementation of the second aspect, before the second node sends the third information, the method also includes: the second node sends sixth information, the sixth information is used to indicate the local information corresponding to the M model parameters respectively, and / or, the sixth information is used to indicate whether the third information includes the M configuration information.

[0131] Based on the above technical solution, the second node can also send the sixth information so as to obtain the relevant information of the M configuration information through the sixth information, and determine whether to receive the M configuration information based on the obtained information, thereby avoiding unnecessary detection by the first node to reduce overhead.

[0132] In a possible implementation of the second aspect, before the second node sends the second information, the method also includes: the second node sends seventh information, the seventh information is used to indicate the local information corresponding to the K model parameters respectively, and / or, the seventh information is used to indicate whether the second information includes the K configuration information.

[0133] Based on the above technical solution, the second node can also send the seventh information so as to obtain relevant information of the K configuration information through the seventh information, and determine whether to receive the K configuration information based on the obtained information, thereby avoiding unnecessary detection by the first node to reduce overhead.

[0134] A third aspect of the present application provides a communication method, which is performed by a first node, or the method is performed by some components in the first node (such as a processor, a chip or a chip system, etc.), or the method can also be implemented by a logic module or software that can realize all or part of the functions of the first node. In the third aspect and its possible implementation, the method is described as being performed by the first node. In this method, the first node receives second information, and the second information indicates M configuration information, and the M configuration information is used to configure the transmission resources of M model parameters, and the parameter scales of the M model parameters are different, and M is an integer greater than 1; the first node receives a second model parameter based on the second configuration information in the M configuration information, wherein the parameter scale of the second model parameter is the same as the local parameter scale (or similar, for example, the difference between the parameter scale of the second model parameter and the local parameter scale is lower than a threshold).

[0135] Based on the above technical solution, the second information received by the first node indicates M configuration information, and the M configuration information is used to configure the resources of M model parameters. Thereafter, the first node receives the second model parameter based on the resource configured by the second configuration information in the M configuration information, and the second model parameter is the same (or similar) as the local parameter scale. In other words, different first nodes can obtain model parameters with the same (or similar) parameter scale as the local one based on the first information, and can subsequently participate in AI learning based on model parameters with the same (or similar) parameter scale as the local one. Compared with the implementation method in which different nodes participate in AI learning based on randomized model parameters, in the above technical solution, different nodes can participate in AI learning based on model parameters with the same (or similar) parameter scale as the local one. Thus, when the communication node in the communication system acts as a node participating in AI learning, the computing power of the communication node can be applied to the AI ​​tasks in the learning system, and the implementation flexibility of different nodes can be improved.

[0136] In addition, the local parameter scales of different nodes may be different. In the above technical solution, the first node can participate in AI learning based on model parameters that are the same (or similar) to the local parameter scale. To this end, different first nodes can participate in AI learning based on model parameters that are adapted to local information to improve the processing efficiency of the neural network model.

[0137] In a possible implementation of the third aspect, the first node may determine the second configuration information from the M configuration information based on the local information, where the local information includes at least one of a local resource overhead level, local AI capability information, and a local parameter scale.

[0138] Based on the above technical solution, local information can be implemented in a variety of ways to provide flexibility in solution implementation.

[0139] It should be understood that the local resource overhead level can be understood as the utilization rate of local computing resources, the utilization rate of local storage resources, the usage of local power consumption, etc. For example, the local computing resources may include resources of modules / devices such as a central processor unit (CPU), a microcontroller unit (MCU), or a graphics processing unit (GPU); the local storage resources may include resources of modules / devices such as a read-only memory (ROM) and a random access memory (RAM); the usage of local power consumption may include the health of a battery module, the current amount of power used by the battery module (or the percentage of the power), the current amount of power not used by the battery module (or the percentage of the power), etc.

[0140] It should be understood that local AI capability information may include local AI-related capability information, such as AI-related computing capability, storage capability, etc. Similarly, AI-related computing capability may include the computing capability of modules / devices such as CPU, MCU, and GPU, and AI-related storage capability may include the storage capability of modules / devices such as ROM and RAM.

[0141] It should be understood that the transmission resources configured by the M configuration information are used for different first nodes to receive one of the M model parameters, so that different first nodes can subsequently perform AI processing (such as local training) based on the received model parameters. Accordingly, the M model parameters can be understood as global model parameters of multiple first nodes, that is, the M model parameters can be replaced by M global model parameters.

[0142] Optionally, for any model parameter among the M model parameters, a gradient and / or a weight may be included. For example, the gradient and / or the weight are used for further processing of the initialized neural network model (e.g., the first neural network model described above), such as training the neural network model, updating the neural network model, etc.

[0143] Similarly, the parameter scales of the M model parameters are different and can be replaced by other descriptions, for example, the number of neural network layers corresponding to the M model parameters is different, the neurons corresponding to the M model parameters are different, the input and output corresponding to the M model parameters are different, the computational complexity of the M model parameters is different, or the computing power requirements of the M model parameters are different, or the storage capacity requirements of the M model parameters are different, etc.

[0144] In a possible implementation manner of the third aspect, the model parameter includes at least one of a model weight, a model activation function, and a model bias.

[0145] Based on the above technical solution, the M model parameters of the resource transmission configured by the M configuration information include at least one of the above items to improve the flexibility of the solution implementation.

[0146] In a possible implementation manner of the third aspect, any one of the M pieces of configuration information includes at least one of the following:

[0147] Indication information indicating parameter scales of model parameters configured by the one configuration information;

[0148] Indication information indicating the received power of the transmission resource configured by the one configuration information;

[0149] Indication information indicating a parameter type of a model parameter configured by the configuration information, where the parameter type includes a gradient and / or a weight;

[0150] Indication information indicating the frequency domain resources of the transmission resources configured by the one configuration information;

[0151] Indication information indicating the time domain resources of the transmission resources configured by the one configuration information.

[0152] Based on the above technical solution, the first node can determine the second configuration information corresponding to the local information in M ​​configuration information based on the local information, so that the first node can obtain the model parameters corresponding to the local information based on the second configuration information.

[0153] In a possible implementation of the third aspect, the method also includes: the first node receives third information, the third information indicates K configuration information, the K configuration information is used to configure transmission resources of K model parameters, the parameter scales of the K model parameters are different, and K is an integer greater than 1; the first node sends a third model parameter based on the third configuration information in the K configuration information, wherein the parameter scale of the third model parameter is the same as the local parameter scale (or similar, for example, the difference between the parameter scale of the third model parameter and the local parameter scale is lower than a threshold).

[0154] Based on the above technical solution, the first node may also receive third information indicating K configuration information, and based on the third configuration information of the K configuration information, receive third model parameters with the same (or similar) parameter scale as the local parameters. In other words, different first nodes can each send model parameters with the same (or similar) parameter scale as their respective local information via the third information, allowing the recipient of the model parameters to perform AI processing (e.g., parameter fusion processing) on ​​the model parameters sent by different first nodes.

[0155] It should be understood that the transmission resources configured by the K configuration information are used by different first nodes to transmit one of the K model parameters, so that the recipient of the model parameter can perform a fusion operation on the model parameters transmitted by the different first nodes. Accordingly, the K model parameters can be understood as the local model parameters of multiple first nodes, that is, the K model parameters can be replaced by K local model parameters.

[0156] Optionally, for any model parameter among the K model parameters, a gradient and / or a weight may be included. For example, the gradient and / or the weight are used for further fusion processing of the neural network model, such as deep fusion, weighted averaging, etc.

[0157] Similarly, the parameter scales of the K model parameters are different and can be replaced by other descriptions, for example, the number of neural network layers corresponding to the K model parameters is different, the neurons corresponding to the K model parameters are different, the input and output corresponding to the K model parameters are different, the computational complexity of the K model parameters is different, or the computing power requirements of the K model parameters are different, or the storage capacity requirements of the K model parameters are different, etc.

[0158] In a possible implementation of the third aspect, any one of the K pieces of configuration information includes at least one of the following:

[0159] Indication information indicating parameter scales of model parameters configured by the one configuration information;

[0160] Indication information indicating the transmit power of the transmission resource configured by the one configuration information;

[0161] Indication information indicating a parameter type of a model parameter configured by the configuration information, where the parameter type includes a gradient and / or a weight;

[0162] Indication information indicating the frequency domain resources of the transmission resources configured by the one configuration information;

[0163] Indication information indicating the time domain resource of the transmission resource configured by the one configuration information;

[0164] Indication information indicating the index of the transmission resource of the one configuration information in the transmission resources configured by the K configuration information.

[0165] Based on the above technical solution, any one of the K pieces of configuration information indicated by the third information may include at least one of the above items to enhance the flexibility of the solution implementation.

[0166] In a possible implementation of the third aspect, before the first node receives the third information, the method also includes: the first node receives sixth information, the sixth information is used to indicate the local information corresponding to the M model parameters respectively, and / or, the sixth information is used to indicate whether the third information includes the M configuration information.

[0167] Based on the above technical solution, the first node can also receive the sixth information so as to obtain relevant information of the M configuration information through the sixth information, and determine whether to receive the M configuration information based on the obtained information, which can avoid unnecessary detection by the first node to reduce overhead.

[0168] In a possible implementation of the third aspect, before the first node receives the second information, the method also includes: the first node receives seventh information, the seventh information is used to indicate the local information corresponding to the K model parameters respectively, and / or, the seventh information is used to indicate whether the second information includes the K configuration information.

[0169] Based on the above technical solution, the first node can also receive the seventh information so as to obtain relevant information of the K configuration information through the seventh information, and determine whether to receive the K configuration information based on the obtained information, which can avoid unnecessary detection by the first node to reduce overhead.

[0170] A fourth aspect of the present application provides a communication method, which is performed by a second node, or the method is performed by some components in the second node (such as a processor, a chip or a chip system, etc.), or the method can also be implemented by a logic module or software that can realize all or part of the functions of the second node. In the fourth aspect and its possible implementation, the method is described as being performed by the second node. In this method, the second node determines second information, which indicates M configuration information, and the M configuration information is used to configure the transmission resources of M model parameters, and the parameter scales of the M model parameters are different, and M is an integer greater than 1; the second node sends the second information.

[0171] Based on the above technical solution, the second information sent by the second node indicates M configuration information, and the M configuration information is used to configure the resources of M model parameters. Thereafter, after the first node receives the second information, the first node receives the second model parameter based on the resources configured by the second configuration information in the M configuration information, and the second model parameter is the same (or similar) as the local parameter scale. In other words, different first nodes can all obtain model parameters with the same (or similar) parameter scale as the local parameter scale based on the first information, and can subsequently participate in AI learning based on model parameters with the same (or similar) parameter scale as the local parameter scale. Compared with the implementation method in which different nodes participate in AI learning based on randomized model parameters, in the above technical solution, different nodes can participate in AI learning based on model parameters with the same (or similar) parameter scale as the local parameter scale. Thus, when the communication node in the communication system acts as a node participating in AI learning, the computing power of the communication node can be applied to the AI ​​tasks in the learning system, and the implementation flexibility of different nodes can be improved.

[0172] In addition, the local parameter scales of different nodes may be different. In the above technical solution, the first node can participate in AI learning based on model parameters that are the same (or similar) to the local parameter scale. To this end, different first nodes can participate in AI learning based on model parameters that are adapted to local information to improve the processing efficiency of the neural network model.

[0173] It should be understood that the transmission resources configured by the M configuration information are used for different first nodes to receive one of the M model parameters, so that different first nodes can subsequently perform AI processing (such as local training) based on the received model parameters. Accordingly, the M model parameters can be understood as global model parameters of multiple first nodes, that is, the M model parameters can be replaced by M global model parameters.

[0174] Optionally, for any model parameter among the M model parameters, a gradient and / or a weight may be included. For example, the gradient and / or the weight are used for further processing of the initialized neural network model (e.g., the first neural network model described above), such as training the neural network model, updating the neural network model, etc.

[0175] Similarly, the parameter scales of the M model parameters are different and can be replaced by other descriptions, for example, the number of neural network layers corresponding to the M model parameters is different, the neurons corresponding to the M model parameters are different, the input and output corresponding to the M model parameters are different, the computational complexity of the M model parameters is different, or the computing power requirements of the M model parameters are different, or the storage capacity requirements of the M model parameters are different, etc.

[0176] In a possible implementation manner of the fourth aspect, the model parameter includes at least one of a model weight, a model activation function, and a model bias.

[0177] Based on the above technical solution, the M model parameters of the resource transmission configured by the M configuration information include at least one of the above items to improve the flexibility of the solution implementation.

[0178] In a possible implementation manner of the fourth aspect, any one of the M pieces of configuration information includes at least one of the following:

[0179] Indication information indicating parameter scales of model parameters configured by the one configuration information;

[0180] Indication information indicating the received power of the transmission resource configured by the one configuration information;

[0181] Indication information indicating a parameter type of a model parameter configured by the configuration information, where the parameter type includes a gradient and / or a weight;

[0182] Indication information indicating the frequency domain resources of the transmission resources configured by the one configuration information;

[0183] Indication information indicating the time domain resources of the transmission resources configured by the one configuration information.

[0184] Based on the above technical solution, the first node can determine the second configuration information corresponding to the local information in M ​​configuration information based on the local information, so that the first node can obtain the model parameters corresponding to the local information based on the second configuration information.

[0185] In a possible implementation of the fourth aspect, the method further includes: the second node sends third information, the third information indicates K configuration information, the K configuration information is used to configure transmission resources of K model parameters, the parameter scales of the K model parameters are different, and K is an integer greater than 1.

[0186] Based on the above technical solution, the second node may further transmit third information indicating K configuration information, so that the first node can receive third model parameters of the same (or similar) scale as the local parameters based on the third configuration information of the K configuration information. In other words, different first nodes can each transmit model parameters of the same (or similar) scale as the parameters of their respective local information via the third information, so that the recipient of the model parameters can perform AI processing (e.g., parameter fusion processing) on ​​the model parameters transmitted by different first nodes.

[0187] It should be understood that the transmission resources configured by the K configuration information are used by different first nodes to transmit one of the K model parameters, so that the recipient of the model parameter can perform a fusion operation on the model parameters transmitted by the different first nodes. Accordingly, the K model parameters can be understood as the local model parameters of multiple first nodes, that is, the K model parameters can be replaced by K local model parameters.

[0188] Optionally, for any model parameter among the K model parameters, a gradient and / or a weight may be included. For example, the gradient and / or the weight are used for further fusion processing of the neural network model, such as deep fusion, weighted averaging, etc.

[0189] Similarly, the parameter scales of the K model parameters are different and can be replaced by other descriptions, for example, the number of neural network layers corresponding to the K model parameters is different, the neurons corresponding to the K model parameters are different, the input and output corresponding to the K model parameters are different, the computational complexity of the K model parameters is different, or the computing power requirements of the K model parameters are different, or the storage capacity requirements of the K model parameters are different, etc.

[0190] In a possible implementation manner of the fourth aspect, any one of the K pieces of configuration information includes at least one of the following:

[0191] Indication information indicating parameter scales of model parameters configured by the one configuration information;

[0192] Indication information indicating the transmit power of the transmission resource configured by the one configuration information;

[0193] Indication information indicating a parameter type of a model parameter configured by the configuration information, where the parameter type includes a gradient and / or a weight;

[0194] Indication information indicating the frequency domain resources of the transmission resources configured by the one configuration information;

[0195] Indication information indicating the time domain resource of the transmission resource configured by the one configuration information;

[0196] Indication information indicating the index of the transmission resource of the one configuration information in the transmission resources configured by the K configuration information.

[0197] Based on the above technical solution, any one of the K pieces of configuration information indicated by the third information may include at least one of the above items to enhance the flexibility of the solution implementation.

[0198] In a possible implementation of the fourth aspect, before the second node sends the third information, the method also includes: the second node sends sixth information, the sixth information is used to indicate the local information corresponding to the M model parameters respectively, and / or, the sixth information is used to indicate whether the third information includes the M configuration information.

[0199] Based on the above technical solution, the second node can also send the sixth information so as to obtain the relevant information of the M configuration information through the sixth information, and determine whether to receive the M configuration information based on the obtained information, thereby avoiding unnecessary detection by the first node to reduce overhead.

[0200] In a possible implementation of the fourth aspect, before the second node sends the second information, the method also includes: the second node sends seventh information, the seventh information is used to indicate the local information corresponding to the K model parameters respectively, and / or, the seventh information is used to indicate whether the second information includes the K configuration information.

[0201] Based on the above technical solution, the second node can also send the seventh information so as to obtain relevant information of the K configuration information through the seventh information, and determine whether to receive the K configuration information based on the obtained information, thereby avoiding unnecessary detection by the first node to reduce overhead.

[0202] In a fifth aspect, the present application provides a communication method, which is executed by a first node, or the method is executed by some components in the first node (such as a processor, a chip or a chip system, etc.), or the method can also be implemented by a logic module or software that can realize all or part of the functions of the first node. In the fifth aspect and its possible implementation, the method is described as being executed by the first node. In this method, the first node receives a third message, which indicates K configuration information, and the K configuration information is used to configure the transmission resources of K model parameters. The parameter scales of the K model parameters are different, and K is an integer greater than 1; the first node sends a third model parameter based on the third configuration information in the K configuration information, wherein the parameter scale of the third model parameter is the same as the local parameter scale (or similar, for example, the difference between the parameter scale of the third model parameter and the local parameter scale is lower than a threshold).

[0203] Based on the above technical solution, the third information received by the first node indicates K configuration information, and the K configuration information is used to configure the resources of K model parameters. Thereafter, the first node sends the third model parameter based on the resource configured by the third configuration information in the K configuration information, and the third model parameter is the same (or similar) as the local parameter scale. In other words, different first nodes can obtain model parameters with the same (or similar) parameter scale as the local one based on the first information, and can subsequently participate in AI learning based on model parameters with the same (or similar) parameter scale as the local one. Compared with the implementation method in which different nodes participate in AI learning based on randomized model parameters, in the above technical solution, different nodes can participate in AI learning based on model parameters with the same (or similar) parameter scale as the local one. Thus, when the communication node in the communication system acts as a node participating in AI learning, the computing power of the communication node can be applied to the AI ​​tasks in the learning system, and the implementation flexibility of different nodes can be improved.

[0204] In addition, the local parameter scales of different nodes may be different. In the above technical solution, the first node can participate in AI learning based on model parameters that are the same (or similar) to the local parameter scale. To this end, different first nodes can participate in AI learning based on model parameters that are adapted to local information to improve the processing efficiency of the neural network model.

[0205] In a possible implementation manner of the fifth aspect, the first node may determine the second configuration information from the M configuration information based on the local information.

[0206] Based on the above technical solution, local information can be implemented in a variety of ways to provide flexibility in solution implementation.

[0207] It should be understood that the local resource overhead level can be understood as the utilization rate of local computing resources, the utilization rate of local storage resources, the usage of local power consumption, etc. For example, the local computing resources may include resources of modules / devices such as a central processor unit (CPU), a microcontroller unit (MCU), or a graphics processing unit (GPU); the local storage resources may include resources of modules / devices such as a read-only memory (ROM) and a random access memory (RAM); the usage of local power consumption may include the health of a battery module, the current amount of power used by the battery module (or the percentage of the power), the current amount of power not used by the battery module (or the percentage of the power), etc.

[0208] It should be understood that local AI capability information may include local AI-related capability information, such as AI-related computing capability, storage capability, etc. Similarly, AI-related computing capability may include the computing capability of modules / devices such as CPU, MCU, and GPU, and AI-related storage capability may include the storage capability of modules / devices such as ROM and RAM.

[0209] It should be understood that the transmission resources configured by the K configuration information are used by different first nodes to transmit one of the K model parameters, so that the recipient of the model parameter can perform a fusion operation on the model parameters transmitted by the different first nodes. Accordingly, the K model parameters can be understood as the local model parameters of multiple first nodes, that is, the K model parameters can be replaced by K local model parameters.

[0210] Optionally, for any model parameter among the K model parameters, a gradient and / or a weight may be included. For example, the gradient and / or the weight are used for further fusion processing of the neural network model, such as deep fusion, weighted averaging, etc.

[0211] Similarly, the parameter scales of the K model parameters are different and can be replaced by other descriptions, for example, the number of neural network layers corresponding to the K model parameters is different, the neurons corresponding to the K model parameters are different, the input and output corresponding to the K model parameters are different, the computational complexity of the K model parameters is different, or the computing power requirements of the K model parameters are different, or the storage capacity requirements of the K model parameters are different, etc.

[0212] In a possible implementation manner of the fifth aspect, the model parameter includes at least one of a model weight, a model activation function, and a model bias.

[0213] Based on the above technical solution, the K model parameters of the resource transmission configured by the K configuration information include at least one of the above items to improve the flexibility of the solution implementation.

[0214] In a possible implementation manner of the fifth aspect, any one of the K pieces of configuration information includes at least one of the following:

[0215] Indication information indicating parameter scales of model parameters configured by the one configuration information;

[0216] Indication information indicating the transmit power of the transmission resource configured by the one configuration information;

[0217] Indication information indicating a parameter type of a model parameter configured by the configuration information, where the parameter type includes a gradient and / or a weight;

[0218] Indication information indicating the frequency domain resources of the transmission resources configured by the one configuration information;

[0219] Indication information indicating the time domain resource of the transmission resource configured by the one configuration information;

[0220] Indication information indicating the index of the transmission resource of the one configuration information in the transmission resources configured by the K configuration information.

[0221] Based on the above technical solution, the first node can determine the third configuration information corresponding to the local information among the K configuration information based on the local information, so that the first node can send the model parameters corresponding to the local information based on the third configuration information.

[0222] In a possible implementation of the fifth aspect, before the first node receives the second information, the method also includes: the first node receives seventh information, the seventh information is used to indicate the local information corresponding to the K model parameters respectively, and / or, the seventh information is used to indicate whether the second information includes the K configuration information.

[0223] Based on the above technical solution, the first node can also receive the seventh information so as to obtain relevant information of the K configuration information through the seventh information, and determine whether to receive the K configuration information based on the obtained information, which can avoid unnecessary detection by the first node to reduce overhead.

[0224] In a sixth aspect of the present application, a communication method is provided, which is executed by a second node, or the method is executed by some components in the second node (such as a processor, a chip or a chip system, etc.), or the method can also be implemented by a logic module or software that can realize all or part of the functions of the second node. In the sixth aspect and its possible implementation, the method is described as being executed by the second node. In this method, the second node determines third information, and the third information indicates K configuration information, and the K configuration information is used to configure the transmission resources of K model parameters, and the parameter scales of the K model parameters are different, and K is an integer greater than 1; the second node sends the third information.

[0225] Based on the above technical solution, the third information sent by the second node indicates K configuration information, and the K configuration information is used to configure the resources of K model parameters. Thereafter, after the first node receives the third information, the first node sends the third model parameter based on the resource configured by the second configuration information in the K configuration information, and the third model parameter is the same (or similar) as the local parameter scale. In other words, different first nodes can all obtain model parameters with the same (or similar) parameter scale as the local one based on the first information, and can subsequently participate in AI learning based on model parameters with the same (or similar) parameter scale as the local one. Compared with the implementation method in which different nodes participate in AI learning based on randomized model parameters, in the above technical solution, different nodes can participate in AI learning based on model parameters with the same (or similar) parameter scale as the local one. Thus, when the communication node in the communication system acts as a node participating in AI learning, the computing power of the communication node can be applied to the AI ​​tasks in the learning system, and the implementation flexibility of different nodes can be improved.

[0226] In addition, the local parameter scales of different nodes may be different. In the above technical solution, the first node can participate in AI learning based on model parameters that are the same (or similar) to the local parameter scale. To this end, different first nodes can participate in AI learning based on model parameters that are adapted to local information to improve the processing efficiency of the neural network model.

[0227] It should be understood that the transmission resources configured by the K configuration information are used by different first nodes to transmit one of the K model parameters, so that the recipient of the model parameter can perform a fusion operation on the model parameters transmitted by the different first nodes. Accordingly, the K model parameters can be understood as the local model parameters of multiple first nodes, that is, the K model parameters can be replaced by K local model parameters.

[0228] Optionally, for any model parameter among the K model parameters, a gradient and / or a weight may be included. For example, the gradient and / or the weight are used for further fusion processing of the neural network model, such as deep fusion, weighted averaging, etc.

[0229] Similarly, the parameter scales of the K model parameters are different and can be replaced by other descriptions, for example, the number of neural network layers corresponding to the K model parameters is different, the neurons corresponding to the K model parameters are different, the input and output corresponding to the K model parameters are different, the computational complexity of the K model parameters is different, or the computing power requirements of the K model parameters are different, or the storage capacity requirements of the K model parameters are different, etc.

[0230] In a possible implementation manner of the sixth aspect, the model parameter includes at least one of a model weight, a model activation function, and a model bias.

[0231] Based on the above technical solution, the K model parameters of the resource transmission configured by the K configuration information include at least one of the above items to improve the flexibility of the solution implementation.

[0232] In a possible implementation manner of the sixth aspect, any one of the K pieces of configuration information includes at least one of the following:

[0233] Indication information indicating parameter scales of model parameters configured by the one configuration information;

[0234] Indication information indicating the transmit power of the transmission resource configured by the one configuration information;

[0235] Indication information indicating a parameter type of a model parameter configured by the configuration information, where the parameter type includes a gradient and / or a weight;

[0236] Indication information indicating the frequency domain resources of the transmission resources configured by the one configuration information;

[0237] Indication information indicating the time domain resource of the transmission resource configured by the one configuration information;

[0238] Indication information indicating the index of the transmission resource of the one configuration information in the transmission resources configured by the K configuration information.

[0239] Based on the above technical solution, the first node can determine the third configuration information corresponding to the local information among the K configuration information based on the local information, so that the first node can send the model parameters corresponding to the local information based on the third configuration information.

[0240] In a possible implementation of the sixth aspect, before the second node sends the second information, the method also includes: the second node sends seventh information, the seventh information is used to indicate the local information corresponding to the K model parameters respectively, and / or, the seventh information is used to indicate whether the second information includes the K configuration information.

[0241] Based on the above technical solution, the second node can also send the seventh information so as to obtain relevant information of the K configuration information through the seventh information, and determine whether to receive the K configuration information based on the obtained information, thereby avoiding unnecessary detection by the first node to reduce overhead.

[0242] In a seventh aspect of the present application, a communication device is provided, which is a first node, or the device is a partial component in the first node (such as a processor, chip, or chip system, etc.), or the device can also be a logic module or software that can implement all or part of the functions of the first node. In the seventh aspect and its possible implementation, the communication device is described as a first node, and the first node can be a terminal device or a network device.

[0243] The device includes a processing unit and a transceiver unit; the transceiver unit is used to receive first information, where the first information indicates N configuration information, and the N configuration information is used to configure resources of N model parameters, where N is a positive integer; the processing unit is used to determine the first configuration information among the N configuration information; the transceiver unit is also used to receive a first model parameter based on the first configuration information, and the first model parameter is used to determine a first neural network model corresponding to the local information.

[0244] In a possible implementation manner of the seventh aspect, the local information includes at least one of a local resource overhead level, local AI capability information, and a local parameter scale.

[0245] In a possible implementation manner of the seventh aspect, the model parameter includes at least one of a model weight, a model activation function, and a model bias.

[0246] In a possible implementation of the seventh aspect, N is an integer greater than 1, the parameter scales of the N model parameters are different, and the processing unit is further used to determine the first configuration information among the N configuration information based on the local information.

[0247] In a possible implementation of the seventh aspect, the value of N is 1, wherein the first model parameter is used to configure a neural network model shared by at least two first nodes; the processing unit is also used to perform knowledge distillation processing on the neural network model shared by the at least two first nodes based on the local information to obtain the first neural network model.

[0248] In a possible implementation of the seventh aspect, the transceiver unit is further used to receive second information, where the second information indicates M configuration information, and the M configuration information is used to configure transmission resources of M model parameters, where the parameter scales of the M model parameters are different, and M is an integer greater than 1; the processing unit is further used to determine the second configuration information among the M configuration information; the transceiver unit is further used to receive a second model parameter based on the second configuration information, where the parameter scale of the second model parameter is the same as (or similar to) the local parameter scale.

[0249] In a possible implementation manner of the seventh aspect, any one of the M pieces of configuration information includes at least one of the following:

[0250] Indication information indicating parameter scales of model parameters configured by the one configuration information;

[0251] Indication information indicating the received power of the transmission resource configured by the one configuration information;

[0252] Indication information indicating a parameter type of a model parameter configured by the configuration information, where the parameter type includes a gradient and / or a weight;

[0253] Indication information indicating the frequency domain resources of the transmission resources configured by the one configuration information;

[0254] Indication information indicating the time domain resources of the transmission resources configured by the one configuration information.

[0255] In a possible implementation of the seventh aspect, the transceiver unit is further used to receive third information, where the third information indicates K configuration information, where the K configuration information is used to configure transmission resources for K model parameters, where the parameter scales of the K model parameters are different, and K is an integer greater than 1; the processing unit is further used to determine the third configuration information among the K configuration information; the transceiver unit is further used to send a third model parameter based on the third configuration information, where the parameter scale of the third model parameter is the same as (or similar to) the local parameter scale.

[0256] In a possible implementation manner of the seventh aspect, any one of the K pieces of configuration information includes at least one of the following:

[0257] Indication information indicating parameter scales of model parameters configured by the one configuration information;

[0258] Indication information indicating the transmit power of the transmission resource configured by the one configuration information;

[0259] Indication information indicating a parameter type of a model parameter configured by the configuration information, where the parameter type includes a gradient and / or a weight;

[0260] Indication information indicating the frequency domain resources of the transmission resources configured by the one configuration information;

[0261] Indication information indicating the time domain resource of the transmission resource configured by the one configuration information;

[0262] Indication information indicating the index of the transmission resource of the one configuration information in the transmission resources configured by the K configuration information.

[0263] In a possible implementation of the seventh aspect, the transceiver unit is also used to receive fourth information, which is used to indicate the resources carrying the request information, and the request information is used to request the N configuration information; the transceiver unit is also used to send the request information on the resources carrying the request information.

[0264] In a possible implementation manner of the seventh aspect, the request information includes a first orthogonal sequence, and the first orthogonal sequence corresponds to the local information.

[0265] In a possible implementation of the seventh aspect, the transceiver unit is also used to receive fifth information, which is used to indicate the local information corresponding to the N model parameters respectively, and / or the fifth information is used to indicate whether the first information includes the N configuration information.

[0266] In the seventh aspect of the present application, the constituent modules of the communication device can also be used to execute the steps performed in each possible implementation method of the first aspect and achieve corresponding technical effects. For details, please refer to the first aspect and will not be repeated here.

[0267] In an eighth aspect of the present application, a communication device is provided, which is a second node, or the device is a partial component (such as a processor, chip, or chip system) in the second node, or the device can also be a logic module or software that can implement all or part of the functions of the second node. In the eighth aspect and its possible implementations, the communication device is described as a second node, and the second node can be a terminal device or a network device.

[0268] The device includes a processing unit and a transceiver unit; the processing unit is used to determine first information, where the first information indicates N configuration information, and the N configuration information is used to configure resources of N model parameters, where N is a positive integer; wherein one model parameter among the N model parameters is used to determine a neural network model corresponding to the local information of the first node; and the transceiver unit is used to send the first information.

[0269] In a possible implementation manner of the eighth aspect, the local information includes at least one of a local resource overhead level, local AI capability information, and a local parameter scale.

[0270] In a possible implementation manner of the eighth aspect, the model parameter includes at least one of a model weight, a model activation function, and a model bias.

[0271] In a possible implementation manner of the eighth aspect, N is an integer greater than 1, and the parameter scales of the N model parameters are different.

[0272] In a possible implementation of the eighth aspect, the value of N is 1.

[0273] In a possible implementation of the eighth aspect, the transceiver unit is also used to send second information, where the second information indicates M configuration information, and the M configuration information is used to configure transmission resources of M model parameters, where the parameter scales of the M model parameters are different, and M is an integer greater than 1; the transceiver unit is also used to send part or all of the M model parameters based on the M configuration information.

[0274] In a possible implementation manner of the eighth aspect, any one of the M pieces of configuration information includes at least one of the following:

[0275] Indication information indicating parameter scales of model parameters configured by the one configuration information;

[0276] Indication information indicating the received power of the transmission resource configured by the one configuration information;

[0277] Indication information indicating a parameter type of a model parameter configured by the configuration information, where the parameter type includes a gradient and / or a weight;

[0278] Indication information indicating the frequency domain resources of the transmission resources configured by the one configuration information;

[0279] Indication information indicating the time domain resources of the transmission resources configured by the one configuration information.

[0280] In a possible implementation of the eighth aspect, the transceiver unit is also used to send third information, where the third information indicates K configuration information, and the K configuration information is used to configure transmission resources of K model parameters, where the parameter scales of the K model parameters are different, and K is an integer greater than 1; the transceiver unit is also used to receive part or all of the K model parameters based on the K configuration information.

[0281] In a possible implementation manner of the eighth aspect, any one of the K pieces of configuration information includes at least one of the following:

[0282] Indication information indicating parameter scales of model parameters configured by the one configuration information;

[0283] Indication information indicating the transmit power of the transmission resource configured by the one configuration information;

[0284] Indication information indicating a parameter type of a model parameter configured by the configuration information, where the parameter type includes a gradient and / or a weight;

[0285] Indication information indicating the frequency domain resources of the transmission resources configured by the one configuration information;

[0286] Indication information indicating the time domain resource of the transmission resource configured by the one configuration information;

[0287] Indication information indicating the index of the transmission resource of the one configuration information in the transmission resources configured by the K configuration information.

[0288] In a possible implementation of the eighth aspect, the transceiver unit is also used to send fourth information, which is used to indicate the resources carrying the request information, and the request information is used to request the N configuration information; the transceiver unit is also used to receive the request information on the resources carrying the request information.

[0289] In a possible implementation manner of the eighth aspect, the request information includes a first orthogonal sequence, and the first orthogonal sequence corresponds to the local information.

[0290] In a possible implementation of the eighth aspect, the transceiver unit is also used to send fifth information, which is used to indicate the local information corresponding to the N model parameters respectively, and / or, the fifth information is used to indicate whether the first information includes the N configuration information.

[0291] In the eighth aspect of the present application, the constituent modules of the communication device can also be used to execute the steps performed in each possible implementation method of the second aspect and achieve corresponding technical effects. For details, please refer to the second aspect and will not be repeated here.

[0292] In a ninth aspect of the present application, a communication device is provided, which is a first node, or the device is a partial component in the first node (such as a processor, chip, or chip system, etc.), or the device can also be a logic module or software that can implement all or part of the functions of the first node. In the ninth aspect and its possible implementations, the communication device is described as a first node, and the first node can be a terminal device or a network device.

[0293] The device includes a processing unit and a transceiver unit; the transceiver unit is used to receive second information, the second information indicating M configuration information, the M configuration information being used to configure transmission resources of M model parameters, the parameter scales of the M model parameters being different, and M being an integer greater than 1; the processing unit is used to determine the second configuration information among the M configuration information; the transceiver unit is also used to receive a second model parameter based on the second configuration information, wherein the parameter scale of the second model parameter is the same as (or similar to) the local parameter scale.

[0294] In a possible implementation manner of the ninth aspect, the local information includes at least one of a local resource overhead level, local AI capability information, and a local parameter scale.

[0295] In a possible implementation of the ninth aspect, the model parameters include at least one of a model weight, a model activation function, and a model bias.

[0296] In a possible implementation manner of the ninth aspect, any one of the M pieces of configuration information includes at least one of the following:

[0297] Indication information indicating parameter scales of model parameters configured by the one configuration information;

[0298] Indication information indicating the received power of the transmission resource configured by the one configuration information;

[0299] Indication information indicating a parameter type of a model parameter configured by the configuration information, where the parameter type includes a gradient and / or a weight;

[0300] Indication information indicating the frequency domain resources of the transmission resources configured by the one configuration information;

[0301] Indication information indicating the time domain resources of the transmission resources configured by the one configuration information.

[0302] In a possible implementation of the ninth aspect, the transceiver unit is further used to receive third information, where the third information indicates K configuration information, and the K configuration information is used to configure transmission resources of K model parameters, where the parameter scales of the K model parameters are different, and K is an integer greater than 1; the transceiver unit is also used to send a third model parameter based on the third configuration information in the K configuration information, where the parameter scale of the third model parameter is the same as (or similar to) the local parameter scale.

[0303] In a possible implementation manner of the ninth aspect, any one of the K pieces of configuration information includes at least one of the following:

[0304] Indication information indicating parameter scales of model parameters configured by the one configuration information;

[0305] Indication information indicating the transmit power of the transmission resource configured by the one configuration information;

[0306] Indication information indicating a parameter type of a model parameter configured by the configuration information, where the parameter type includes a gradient and / or a weight;

[0307] Indication information indicating the frequency domain resources of the transmission resources configured by the one configuration information;

[0308] Indication information indicating the time domain resource of the transmission resource configured by the one configuration information;

[0309] Indication information indicating the index of the transmission resource of the one configuration information in the transmission resources configured by the K configuration information.

[0310] In the ninth aspect of the present application, the constituent modules of the communication device can also be used to execute the steps performed in each possible implementation method of the third aspect and achieve corresponding technical effects. For details, please refer to the third aspect and will not be repeated here.

[0311] In a tenth aspect, the present application provides a communication device, which is a second node, or a component of the second node (such as a processor, chip, or chip system), or a logic module or software capable of implementing all or part of the functions of the second node. In the tenth aspect and its possible implementations, the communication device is described as a second node, and the second node can be a terminal device or a network device.

[0312] The device includes a processing unit and a transceiver unit; the processing unit is used to determine second information, the second information indicates M configuration information, the M configuration information is used to configure transmission resources of M model parameters, the parameter scales of the M model parameters are different, and M is an integer greater than 1; the transceiver unit is used to send the second information.

[0313] In a possible implementation manner of the tenth aspect, the local information includes at least one of a local resource overhead level, local AI capability information, and a local parameter scale.

[0314] In a possible implementation manner of the tenth aspect, the model parameters include at least one of a model weight, a model activation function, and a model bias.

[0315] In a possible implementation manner of the tenth aspect, any one of the M pieces of configuration information includes at least one of the following:

[0316] Indication information indicating parameter scales of model parameters configured by the one configuration information;

[0317] Indication information indicating the received power of the transmission resource configured by the one configuration information;

[0318] Indication information indicating a parameter type of a model parameter configured by the configuration information, where the parameter type includes a gradient and / or a weight;

[0319] Indication information indicating the frequency domain resources of the transmission resources configured by the one configuration information;

[0320] Indication information indicating the time domain resources of the transmission resources configured by the one configuration information.

[0321] In a possible implementation of the tenth aspect, the processing unit is further configured to determine third information, where the third information indicates K pieces of configuration information, the K pieces of configuration information being used to configure transmission resources of K model parameters, where the K model parameters have different parameter scales, and K is an integer greater than 1;

[0322] In a possible implementation manner of the tenth aspect, any one of the K pieces of configuration information includes at least one of the following:

[0323] Indication information indicating parameter scales of model parameters configured by the one configuration information;

[0324] Indication information indicating the transmit power of the transmission resource configured by the one configuration information;

[0325] Indication information indicating a parameter type of a model parameter configured by the configuration information, where the parameter type includes a gradient and / or a weight;

[0326] Indication information indicating the frequency domain resources of the transmission resources configured by the one configuration information;

[0327] Indication information indicating the time domain resource of the transmission resource configured by the one configuration information;

[0328] Indication information indicating the index of the transmission resource of the one configuration information in the transmission resources configured by the K configuration information.

[0329] In the tenth aspect of the present application, the constituent modules of the communication device can also be used to execute the steps performed in each possible implementation method of the fourth aspect and achieve corresponding technical effects. For details, please refer to the fourth aspect and will not be repeated here.

[0330] In an eleventh aspect of the present application, a communication device is provided, which is a first node, or the device is a partial component of the first node (such as a processor, chip, or chip system), or the device can also be a logic module or software that can implement all or part of the functions of the first node. In the eleventh aspect and its possible implementations, the communication device is described as a first node, and the first node can be a terminal device or a network device.

[0331] The device includes a processing unit and a transceiver unit; the transceiver unit is used to receive third information, where the third information indicates K configuration information, and the K configuration information is used to configure transmission resources of K model parameters, where the parameter scales of the K model parameters are different, and K is an integer greater than 1; the processing unit is used to determine the third configuration information among the K configuration information; the transceiver unit is also used to send a third model parameter based on the third configuration information, where the parameter scale of the third model parameter is the same as (or similar to) the local parameter scale.

[0332] In a possible implementation of the eleventh aspect, the local information includes at least one of a local resource overhead level, local AI capability information, and a local parameter scale.

[0333] In a possible implementation of the eleventh aspect, the model parameters include at least one of a model weight, a model activation function, and a model bias.

[0334] In a possible implementation manner of the eleventh aspect, any one of the K pieces of configuration information includes at least one of the following:

[0335] Indication information indicating parameter scales of model parameters configured by the one configuration information;

[0336] Indication information indicating the transmit power of the transmission resource configured by the one configuration information;

[0337] Indication information indicating a parameter type of a model parameter configured by the configuration information, where the parameter type includes a gradient and / or a weight;

[0338] Indication information indicating the frequency domain resources of the transmission resources configured by the one configuration information;

[0339] Indication information indicating the time domain resource of the transmission resource configured by the one configuration information;

[0340] Indication information indicating the index of the transmission resource of the one configuration information in the transmission resources configured by the K configuration information.

[0341] In the eleventh aspect of the present application, the constituent modules of the communication device can also be used to execute the steps performed in each possible implementation method of the fifth aspect and achieve corresponding technical effects. For details, please refer to the fifth aspect and will not be repeated here.

[0342] In a twelfth aspect of the present application, a communication device is provided, which is a second node, or the device is a partial component of the second node (such as a processor, chip, or chip system), or the device can also be a logic module or software capable of implementing all or part of the functions of the second node. In the twelfth aspect and its possible implementations, the communication device is described as a second node, and the second node can be a terminal device or a network device.

[0343] The device includes a processing unit and a transceiver unit; the processing unit is used to determine third information, where the third information indicates K configuration information, and the K configuration information is used to configure transmission resources of K model parameters, where the parameter scales of the K model parameters are different, and K is an integer greater than 1; the transceiver unit is used to send the third information.

[0344] In a possible implementation of the twelfth aspect, the local information includes at least one of a local resource overhead level, local AI capability information, and a local parameter scale.

[0345] In a possible implementation of the twelfth aspect, the model parameters include at least one of a model weight, a model activation function, and a model bias.

[0346] In a possible implementation manner of the twelfth aspect, any one of the K pieces of configuration information includes at least one of the following:

[0347] Indication information indicating parameter scales of model parameters configured by the one configuration information;

[0348] Indication information indicating the transmit power of the transmission resource configured by the one configuration information;

[0349] Indication information indicating a parameter type of a model parameter configured by the configuration information, where the parameter type includes a gradient and / or a weight;

[0350] Indication information indicating the frequency domain resources of the transmission resources configured by the one configuration information;

[0351] Indication information indicating the time domain resource of the transmission resource configured by the one configuration information;

[0352] Indication information indicating the index of the transmission resource of the one configuration information in the transmission resources configured by the K configuration information.

[0353] In the twelfth aspect of the present application, the constituent modules of the communication device can also be used to execute the steps performed in each possible implementation method of the sixth aspect and achieve corresponding technical effects. For details, please refer to the sixth aspect and will not be repeated here.

[0354] A thirteenth aspect of an embodiment of the present application provides a communication device, comprising at least one processor, wherein the at least one processor is coupled to a memory; the memory is used to store programs or instructions; the at least one processor is used to execute the program or instructions so that the device implements the method described in the first aspect or any possible implementation method of the first aspect.

[0355] A fourteenth aspect of an embodiment of the present application provides a communication device, comprising at least one processor, wherein the at least one processor is coupled to a memory; the memory is used to store programs or instructions; the at least one processor is used to execute the program or instructions so that the device implements the method described in the aforementioned second aspect or any possible implementation method of the second aspect.

[0356] A fifteenth aspect of an embodiment of the present application provides a communication device, comprising at least one processor, wherein the at least one processor is coupled to a memory; the memory is used to store programs or instructions; the at least one processor is used to execute the program or instructions so that the device implements the method described in the third aspect or any possible implementation method of the third aspect.

[0357] A sixteenth aspect of an embodiment of the present application provides a communication device, comprising at least one processor, wherein the at least one processor is coupled to a memory; the memory is used to store programs or instructions; the at least one processor is used to execute the program or instructions so that the device implements the method described in the aforementioned fourth aspect or any possible implementation method of the fourth aspect.

[0358] A seventeenth aspect of an embodiment of the present application provides a communication device, comprising at least one processor, wherein the at least one processor is coupled to a memory; the memory is used to store programs or instructions; the at least one processor is used to execute the program or instructions so that the device implements the method described in the aforementioned fifth aspect or any possible implementation method of the fifth aspect.

[0359] In an eighteenth aspect, an embodiment of the present application provides a communication device, comprising at least one processor, wherein the at least one processor is coupled to a memory; the memory is used to store programs or instructions; the at least one processor is used to execute the program or instructions so that the device implements the method described in the aforementioned sixth aspect or any possible implementation method of the sixth aspect.

[0360] In a nineteenth aspect, an embodiment of the present application provides a communication device, comprising at least one logic circuit and an input / output interface; the logic circuit is used to execute the method described in the first aspect and any possible implementation thereof.

[0361] The twentieth aspect of an embodiment of the present application provides a communication device, comprising at least one logic circuit and an input / output interface; the logic circuit is used to execute the method described in the second aspect and any possible implementation thereof.

[0362] In aspect 21 of an embodiment of the present application, a communication device is provided, comprising at least one logic circuit and an input / output interface; the logic circuit is used to execute the method described in aspect 3 and any possible implementation thereof.

[0363] In aspect 22 of an embodiment of the present application, a communication device is provided, comprising at least one logic circuit and an input / output interface; the logic circuit is used to execute the method described in aspect 4 above and any possible implementation thereof.

[0364] In a twenty-third aspect, an embodiment of the present application provides a communication device, comprising at least one logic circuit and an input / output interface; the logic circuit is used to execute the method described in the fifth aspect and any possible implementation thereof.

[0365] In the twenty-fourth aspect of an embodiment of the present application, a communication device is provided, comprising at least one logic circuit and an input / output interface; the logic circuit is used to execute the method described in the sixth aspect and any possible implementation thereof.

[0366] In aspect 25 of an embodiment of the present application, a computer-readable storage medium is provided, which is used to store one or more computer-executable instructions. When the computer-executable instructions are executed by a processor, the processor executes the method described in any possible implementation of any one of aspects 1 to 6 above.

[0367] A twenty-sixth aspect of an embodiment of the present application provides a computer program product (or computer program). When the computer program product is executed by the processor, the processor executes the method described in any possible implementation of any one of the first to sixth aspects above.

[0368] A twenty-seventh aspect of an embodiment of the present application provides a chip system, which includes at least one processor for supporting a communication device to implement the functions involved in any possible implementation method of any aspect of the first to sixth aspects mentioned above.

[0369] In one possible design, the chip system may further include a memory for storing program instructions and data necessary for the first communication device. The chip system may be composed of a chip or may include a chip and other discrete components. Optionally, the chip system may further include an interface circuit that provides program instructions and / or data to the at least one processor.

[0370] The twenty-eighth aspect of an embodiment of the present application provides a communication system, which includes the communication device of the seventh aspect and the communication device of the eighth aspect, and / or, the communication system includes the communication device of the ninth aspect and the communication device of the tenth aspect, and / or, the communication system includes the communication device of the eleventh aspect and the communication device of the twelfth aspect, and / or, the communication system includes the communication device of the thirteenth aspect and the communication device of the fourteenth aspect, and / or, the communication system includes the communication device of the fifteenth aspect and the communication device of the sixteenth aspect, and / or, the communication system includes the communication device of the seventeenth aspect and the communication device of the eighteenth aspect, and / or, the communication system includes the communication device of the nineteenth aspect and the communication device of the twentieth aspect, and / or, the communication system includes the communication device of the twenty-first aspect and the communication device of the twenty-second aspect, and / or, the communication system includes the communication device of the twenty-third aspect and the communication device of the twenty-fourth aspect.

[0371] Among them, the technical effects brought about by any design method in the seventh aspect to the twenty-eighth aspect can refer to the technical effects brought about by the different design methods in the above-mentioned first aspect to the sixth aspect, and will not be repeated here. BRIEF DESCRIPTION OF THE DRAWINGS

[0372] FIG1a is a schematic diagram of a communication system provided by the present application;

[0373] FIG1b is another schematic diagram of the communication system provided by the present application;

[0374] FIG1c is another schematic diagram of the communication system provided by the present application;

[0375] FIG1d is a schematic diagram of the AI ​​processing process involved in this application;

[0376] FIG1e is another schematic diagram of the AI ​​processing process involved in this application;

[0377] FIG2a is another schematic diagram of the AI ​​processing process involved in this application;

[0378] FIG2 b is another schematic diagram of the AI ​​processing process involved in this application;

[0379] FIG2c is another schematic diagram of the AI ​​processing process involved in this application;

[0380] FIG2 d is another schematic diagram of the AI ​​processing process involved in this application;

[0381] FIG3 is an interactive schematic diagram of the communication method provided by this application;

[0382] FIG4a is another schematic diagram of the AI ​​processing process provided by this application;

[0383] FIG4 b is another schematic diagram of the AI ​​processing process provided by this application;

[0384] FIG4c is another schematic diagram of the AI ​​processing process provided by this application;

[0385] FIG5 is another interactive schematic diagram of the communication method provided by this application;

[0386] FIG6 is another interactive schematic diagram of the communication method provided by this application;

[0387] FIG7 is another schematic diagram of the AI ​​processing process provided by this application;

[0388] FIG8 is a schematic diagram of a communication device provided by the present application;

[0389] FIG9 is another schematic diagram of a communication device provided by the present application;

[0390] FIG10 is another schematic diagram of the communication device provided by the present application;

[0391] FIG11 is another schematic diagram of a communication device provided by the present application;

[0392] FIG12 is another schematic diagram of the communication device provided in this application. DETAILED DESCRIPTION

[0393] First, some of the terms used in the embodiments of the present application are explained to facilitate understanding by those skilled in the art.

[0394] (1) Terminal device: It can be a wireless terminal device that can receive network device scheduling and instruction information. The wireless terminal device can be a device that provides voice and / or data connectivity to the user, or a handheld device with wireless connection function, or other processing device connected to a wireless modem.

[0395] Terminal devices can communicate with one or more core networks or the Internet via a radio access network (RAN). Terminal devices can be mobile terminal devices, such as mobile phones (also known as "cellular" phones, mobile phones), computers, and data cards. For example, they can be portable, pocket-sized, handheld, computer-built-in, or vehicle-mounted mobile devices that exchange voice and / or data with the radio access network. Examples include personal communication service (PCS) phones, cordless phones, Session Initiation Protocol (SIP) phones, wireless local loop (WLL) stations, personal digital assistants (PDAs), tablet computers, and computers with wireless transceiver capabilities. Wireless terminal equipment can also be called system, subscriber unit, subscriber station, mobile station, mobile station (MS), remote station, access point (AP), remote terminal equipment (remote terminal), access terminal equipment (access terminal), user terminal equipment (user terminal), user agent, subscriber station (SS), customer premises equipment (CPE), terminal, user equipment (UE), mobile terminal (MT), etc.

[0396] As an example and not a limitation, in the embodiments of the present application, the terminal device may also be a wearable device. Wearable devices may also be referred to as wearable smart devices or smart wearable devices, etc., which are a general term for wearable devices that are intelligently designed and developed using wearable technology for daily wear, such as glasses, gloves, watches, clothing, and shoes. A wearable device is a portable device that is worn directly on the body or integrated into the user's clothes or accessories. Wearable devices are not only hardware devices, but also achieve powerful functions through software support, data interaction, and cloud interaction. Broadly speaking, wearable smart devices include those that are fully functional, large in size, and can achieve complete or partial functions without relying on smartphones, such as smart watches or smart glasses, etc., as well as those that only focus on a certain type of application function and need to be used in conjunction with other devices such as smartphones, such as various smart bracelets, smart helmets, and smart jewelry for vital sign monitoring.

[0397] The terminal may also be a drone, a robot, a terminal in device-to-device (D2D) communication, a terminal in vehicle-to-everything (V2X), a virtual reality (VR) terminal device, an augmented reality (AR) terminal device, a wireless terminal in industrial control, a wireless terminal in self-driving, a wireless terminal in remote medical, a wireless terminal in smart grid, a wireless terminal in transportation safety, a wireless terminal in smart city, a wireless terminal in smart home, etc.

[0398] In addition, the terminal device may also be a terminal device in a communication system that has evolved after the fifth generation (5G) communication system (e.g., a sixth generation (6G) communication system) or a terminal device in a future public land mobile network (PLMN). For example, the 6G network can further expand the form and function of 5G communication terminals. 6G terminals include but are not limited to vehicles, cellular network terminals (with integrated satellite terminal functions), drones, and Internet of Things (IoT) devices.

[0399] In an embodiment of the present application, the terminal device may also obtain AI services provided by the network device. Optionally, the terminal device may also have AI processing capabilities.

[0400] (2) Network equipment: It can be a device in a wireless network. For example, the network equipment can be a RAN node (or device) that connects a terminal device to a wireless network, which can also be called a base station. Currently, some examples of RAN equipment include: base station, evolved NodeB (eNodeB), gNB (gNodeB) in a 5G communication system, transmission reception point (TRP), evolved Node B (eNB), radio network controller (RNC), Node B (NB), home base station (e.g., home evolved Node B, or home Node B, HNB), base band unit (BBU), or wireless fidelity (Wi-Fi) access point AP, etc. In addition, in a network structure, the network equipment can include a centralized unit (CU) node, a distributed unit (DU) node, or a RAN device including a CU node and a DU node.

[0401] Alternatively, a RAN node can be a macro base station, micro base station, indoor base station, relay node, donor node, or a wireless controller in a cloud radio access network (CRAN) scenario. A RAN node can also be a server, wearable device, vehicle, or vehicle-mounted device. For example, the access network device in vehicle-to-everything (V2X) technology can be a roadside unit (RSU).

[0402] In another possible scenario, multiple RAN nodes collaborate to assist the terminal in achieving wireless access, and different RAN nodes respectively implement part of the functions of the base station. For example, the RAN node can be a centralized unit (CU), a distributed unit (DU), a CU-control plane (CP), a CU-user plane (UP), or a radio unit (RU). The CU and DU can be set separately, or they can be included in the same network element, such as a baseband unit (BBU). The RU can be included in a radio frequency device or radio frequency unit, such as a remote radio unit (RRU), an active antenna unit (AAU), or a remote radio head (RRH).

[0403] In different systems, CU (or CU-CP and CU-UP), DU or RU may also have different names, but those skilled in the art can understand their meanings. For example, in an open access network (open RAN, O-RAN or ORAN) system, CU may also be called O-CU (open CU), DU may also be called O-DU, CU-CP may also be called O-CU-CP, CU-UP may also be called O-CU-UP, and RU may also be called O-RU. For the convenience of description, this application takes CU, CU-CP, CU-UP, DU and RU as examples for description. Any unit of CU (or CU-CP, CU-UP), DU and RU in this application can be implemented by a software module, a hardware module, or a combination of a software module and a hardware module.

[0404] The communication between the access network device and the terminal device follows a certain protocol layer structure. The protocol layer may include a control plane protocol layer and a user plane protocol layer. The control plane protocol layer may include at least one of the following: a radio resource control (RRC) layer, a packet data convergence protocol (PDCP) layer, a radio link control (RLC) layer, a media access control (MAC) layer, or a physical (PHY) layer. The user plane protocol layer may include at least one of the following: a service data adaptation protocol (SDAP) layer, a PDCP layer, an RLC layer, a MAC layer, or a physical layer.

[0405] For the correspondence between network elements in the ORAN system and their achievable protocol layer functions, please refer to Table 1 below.

[0406] Table 1

[0407] The network device may be any other device that provides wireless communication functionality to the terminal device. The embodiments of this application do not limit the specific technology and device form used by the network device. For ease of description, the embodiments of this application do not limit this.

[0408] The network equipment may also include core network equipment, which may include, for example, a mobility management entity (MME), a home subscriber server (HSS), a serving gateway (S-GW), a policy and charging rules function (PCRF), and a public data network gateway (PDN gateway, P-GW) in a fourth generation (4G) network; and network elements such as an access and mobility management function (AMF), a user plane function (UPF), or a session management function (SMF) in a 5G network. In addition, the core network equipment may also include other core network equipment in a 5G network and a next generation network of a 5G network.

[0409] In an embodiment of the present application, the above-mentioned network device may also have a network node with AI capabilities, which can provide AI services for terminals or other network devices. For example, it can be an AI node on the network side (access network or core network), a computing power node, a RAN node with AI capabilities, a core network element with AI capabilities, etc.

[0410] In the embodiments of the present application, the apparatus for implementing the function of the network device may be the network device, or may be a device capable of supporting the network device in implementing the function, such as a chip system, which may be installed in the network device. In the technical solutions provided in the embodiments of the present application, the technical solutions provided in the embodiments of the present application are described by taking the network device as an example.

[0411] (3) Configuration and pre-configuration: In this application, configuration and pre-configuration are used simultaneously. Configuration refers to the network device / server sending some parameter configuration information or parameter values ​​to the terminal through messages or signaling, so that the terminal can determine the communication parameters or resources during transmission based on these values ​​or information. Pre-configuration is similar to configuration, and can be parameter information or parameter values ​​pre-negotiated between the network device / server and the terminal device, or parameter information or parameter values ​​used by the base station / network device or terminal device as specified in the standard protocol, or parameter information or parameter values ​​pre-stored in the base station / server or terminal device. This application does not limit this.

[0412] Furthermore, these values ​​and parameters can be changed or updated.

[0413] (4) The terms "system" and "network" in the embodiments of the present application can be used interchangeably. "Multiple" refers to two or more. "And / or" describes the association relationship of associated objects, indicating that three relationships can exist. For example, A and / or B can mean: A exists alone, A and B exist at the same time, and B exists alone, where A and B can be singular or plural. The character " / " generally indicates that the previous and next associated objects are in an "or" relationship. "At least one of the following" or similar expressions refers to any combination of these items, including any combination of single or plural items. For example, "at least one of A, B and C" includes A, B, C, AB, AC, BC or ABC. In addition, unless otherwise specified, the ordinal numbers such as "first" and "second" mentioned in the embodiments of the present application are used to distinguish multiple objects, and are not used to limit the order, timing, priority or importance of multiple objects.

[0414] (5) “Sending” and “receiving” in the embodiments of the present application indicate the direction of signal transmission. For example, “sending information to XX” can be understood as the destination of the information being XX, which can include direct sending through the air interface, as well as indirect sending through the air interface by other units or modules. “Receiving information from YY” can be understood as the source of the information being YY, which can include direct receiving from YY through the air interface, as well as indirect receiving from YY through the air interface from other units or modules. “Sending” can also be understood as the “output” of the chip interface, and “receiving” can also be understood as the “input” of the chip interface.

[0415] In other words, sending and receiving can be performed between devices, for example, between a network device and a terminal device, or can be performed within a device, for example, sending or receiving between components, modules, chips, software modules or hardware modules within the device through a bus, wiring or interface.

[0416] It is understandable that information may be processed between the source and destination of information transmission, such as coding, modulation, etc., but the destination can understand the valid information from the source. Similar expressions in this application can be understood similarly and will not be repeated.

[0417] (6) In the embodiments of the present application, "indication" may include direct indication and indirect indication, and may also include explicit indication and implicit indication. The information indicated by a certain information (such as the indication information described below) is called information to be indicated. In the specific implementation process, there are many ways to indicate the information to be indicated, such as but not limited to, directly indicating the information to be indicated, such as the information to be indicated itself or the index of the information to be indicated. The information to be indicated may also be indirectly indicated by indicating other information, wherein the other information is associated with the information to be indicated; or only a part of the information to be indicated may be indicated, while the other part of the information to be indicated is known or agreed in advance. For example, the indication of specific information may be achieved by means of the arrangement order of each information agreed in advance (such as predefined by the protocol), thereby reducing the indication overhead to a certain extent. The present application does not limit the specific method of indication. It is understandable that for the sender of the indication information, the indication information can be used to indicate the information to be indicated, and for the receiver of the indication information, the indication information can be used to determine the information to be indicated.

[0418] In this application, unless otherwise specified, the same or similar parts between the various embodiments can refer to each other. In the various embodiments of this application, and the various methods / designs / implementations in each embodiment, if there is no special explanation and logical conflict, the terms and / or descriptions between different embodiments and the various methods / designs / implementations in each embodiment are consistent and can be referenced to each other. The technical features in different embodiments and the various methods / designs / implementations in each embodiment can be combined to form new embodiments, methods, or implementations according to their inherent logical relationships. The following description of the implementation methods of this application does not constitute a limitation on the scope of protection of this application.

[0419] The present application can be applied to a long term evolution (LTE) system, a new radio (NR) system, or a communication system evolved after 5G (such as 6G, etc.). The communication system includes at least one network device and / or at least one terminal device.

[0420] Please refer to Figure 1a, which is a schematic diagram of a communication system in this application. Figure 1a exemplarily illustrates a network device and six terminal devices, namely terminal device 1, terminal device 2, terminal device 3, terminal device 4, terminal device 5, and terminal device 6. In the example shown in Figure 1a, terminal device 1 is a smart teacup, terminal device 2 is a smart air conditioner, terminal device 3 is a smart gas pump, terminal device 4 is a vehicle, terminal device 5 is a mobile phone, and terminal device 6 is a printer.

[0421] As shown in Figure 1a, the AI ​​configuration information sending entity can be a network device. The AI ​​configuration information receiving entity can be terminal devices 1-6. In this case, the network device and terminal devices 1-6 form a communication system. In this communication system, terminal devices 1-6 can send data to the network device, and the network device needs to receive data sent by terminal devices 1-6. At the same time, the network device can send configuration information to terminal devices 1-6.

[0422] For example, in Figure 1a, terminal devices 4 and 6 can also form a communication system. Terminal device 5 serves as a network device, i.e., the AI ​​configuration information sending entity; terminal devices 4 and 6 serve as terminal devices, i.e., the AI ​​configuration information receiving entities. For example, in a connected vehicle system, terminal device 5 sends AI configuration information to terminal devices 4 and 6, respectively, and receives data from them. Correspondingly, terminal devices 4 and 6 receive AI configuration information from terminal device 5 and send data to terminal device 5.

[0423] Taking the communication system shown in Figure 1a as an example, in addition to performing communication-related services, different devices (including between network devices and network devices, between network devices and terminal devices, and / or between terminal devices) may also perform AI-related services. For example, as shown in Figure 1b, taking the network device as a base station as an example, the base station can perform communication-related services and AI-related services with one or more terminal devices, and different terminal devices can also perform communication-related services and AI-related services. For another example, as shown in Figure 1c, taking the terminal devices including a TV and a mobile phone as an example, communication-related services and AI-related services can also be performed between the TV and the mobile phone.

[0424] The technical solution provided in this application can be applied to a wireless communication system (e.g., the system shown in FIG. 1a , FIG. 1b , or FIG. 1c ). For example, an AI network element can be introduced into the communication system provided in this application to implement some or all AI-related operations. The AI ​​network element can also be referred to as an AI node, AI device, AI entity, AI module, AI model, or AI unit, etc. The AI ​​network element can be a network element built into the communication system. For example, the AI ​​network element can be an AI module built into: an access network device, a core network device, a cloud server, or a network management (OAM) to implement AI-related functions. The OAM can be a network management device for a core network device and / or a network management device for an access network device. Alternatively, the AI ​​network element can also be an independently set network element in the communication system. Optionally, the terminal or the chip built into the terminal can also include an AI entity to implement AI-related functions.

[0425] The following is a brief introduction to artificial intelligence (AI) that may be involved in this application.

[0426] Artificial intelligence (AI) can imbue machines with human intelligence. For example, it can enable machines to simulate certain intelligent human behaviors using computer hardware and software. Machine learning methods can be used to achieve AI. In machine learning, a machine uses training data to learn (or train) a model. This model represents the mapping from input to output. The learned model can be used for inference (or prediction), meaning that the model can be used to predict the output corresponding to a given input. This output can also be called an inference result (or prediction result).

[0427] Machine learning can include supervised learning, unsupervised learning, and reinforcement learning. Among them, unsupervised learning can also be called unsupervised learning.

[0428] Supervised learning uses machine learning algorithms to learn the mapping relationship between sample values ​​and sample labels based on collected sample values ​​and sample labels, and then expresses this learned mapping relationship using an AI model. The process of training a machine learning model is the process of learning this mapping relationship. During training, sample values ​​are input into the model to obtain the model's predicted values. The model parameters are optimized by calculating the error between the model's predicted values ​​and the sample labels (ideal values). Once the mapping relationship is learned, the learned mapping can be used to predict new sample labels. The mapping relationship learned by supervised learning can include linear mappings or nonlinear mappings. Based on the type of label, the learning task can be divided into classification tasks and regression tasks.

[0429] Unsupervised learning uses algorithms to discover inherent patterns in collected sample values. One type of unsupervised learning algorithm uses the samples themselves as supervisory signals, meaning the model learns the mapping from one sample to another. This is called self-supervised learning. During training, the model parameters are optimized by calculating the error between the model's predictions and the samples themselves. Self-supervised learning can be used in signal compression and decompression recovery applications. Common algorithms include autoencoders and generative adversarial networks.

[0430] Reinforcement learning, unlike supervised learning, is a type of algorithm that learns problem-solving strategies through interaction with the environment. Unlike supervised and unsupervised learning, reinforcement learning problems lack explicit label data for "correct" actions. Instead, the algorithm must interact with the environment to obtain reward signals from the environment, and then adjust its decision-making actions to maximize the reward signal value. For example, in downlink power control, the reinforcement learning model adjusts the downlink transmit power of each user based on the overall system throughput fed back by the wireless network, hoping to achieve higher system throughput. The goal of reinforcement learning is also to learn the mapping between environmental states and optimal (e.g., optimal) decision-making actions. However, because the labels for "correct actions" cannot be obtained in advance, network optimization cannot be achieved by calculating the error between actions and "correct actions." Reinforcement learning training is achieved through iterative interaction with the environment.

[0431] A neural network (NN) is a specific model in machine learning technology. According to the universal approximation theorem, NNs can theoretically approximate any continuous function, enabling them to learn arbitrary mappings. Traditional communication systems require extensive expert knowledge to design communication modules. However, deep learning communication systems based on neural networks can automatically discover implicit patterns in massive data sets and establish mapping relationships between data, achieving performance superior to traditional modeling methods.

[0432] The idea of ​​a neural network originates from the neuronal structure of the brain. For example, each neuron performs a weighted sum operation on its input values ​​and outputs the result through an activation function. Figure 1d shows a schematic diagram of a neuron structure. Assume that the input of a neuron is x = [x0, x1, ..., x n ], and the weights corresponding to each input are w=[w,w1,…,w n ], where n is a positive integer, w i and x i It can be a decimal, an integer (such as 0, a positive integer or a negative integer, etc.), or a complex number. i As x i The weight of x i Weighted. The bias of the weighted sum of the input values ​​according to the weight is, for example, b. The activation function can take many forms. Assuming that the activation function of a neuron is: y = f(z) = max(0,z), then the output of the neuron is: For another example, if the activation function of a neuron is: y = f(z) = z, then the output of the neuron is: b can be a decimal, an integer (eg, 0, a positive integer, or a negative integer), or a complex number, etc. The activation functions of different neurons in a neural network can be the same or different.

[0433] Furthermore, neural networks generally include multiple layers, each of which may include one or more neurons. Increasing the depth and / or width of a neural network can improve its expressive power, providing more powerful information extraction and abstract modeling capabilities for complex systems. The depth of a neural network can refer to the number of layers it comprises, and the number of neurons in each layer can be referred to as the width of that layer. In one implementation, a neural network includes an input layer and an output layer. The input layer processes the input information received by the neural network through neurons, passing the processing results to the output layer, which then obtains the output of the neural network. In another implementation, a neural network includes an input layer, a hidden layer, and an output layer. The input layer processes the input information received by the neural network through neurons, passing the processing results to an intermediate hidden layer. The hidden layer performs calculations on the received processing results to obtain a calculation result, which is then passed to the output layer or the next adjacent hidden layer, which ultimately obtains the output of the neural network. A neural network can include one hidden layer or multiple hidden layers connected in sequence, without limitation.

[0434] A neural network is, for example, a deep neural network (DNN). Depending on how the network is constructed, a DNN can include a feedforward neural network (FNN), a convolutional neural network (CNN), and a recurrent neural network (RNN).

[0435] A characteristic of FNN networks is that neurons in adjacent layers are fully connected. This characteristic typically requires a large amount of storage space and results in high computational complexity. Figure 1e shows a schematic diagram of an FNN network.

[0436] CNN is a neural network specifically designed to process data with a grid-like structure. For example, time series data (discrete sampling along the time axis) and image data (discrete sampling along two dimensions) can both be considered grid-like data. CNNs do not utilize all input information at once for computation. Instead, they use a fixed-size window to intercept a portion of the information for convolution operations, significantly reducing the computational complexity of model parameters. Furthermore, depending on the type of information intercepted by the window (e.g., people and objects in an image represent different types of information), each window can use a different convolution kernel, enabling CNNs to better extract features from the input data.

[0437] RNNs are a type of DNN that utilizes feedback time series information. Their input consists of a new input value at the current moment and their own output value at the previous moment. RNNs are suitable for capturing temporally correlated sequence features and are particularly well-suited for applications such as speech recognition and channel coding.

[0438] During the machine learning model training process, a loss function can be defined. This function describes the gap or discrepancy between the model's output and the ideal target value. Loss functions can be expressed in various forms, and there are no restrictions on their specific form. The model training process can be viewed as adjusting some or all of the model's parameters to keep the loss function below a threshold or meet the target.

[0439] A model may also be referred to as an AI model, rule, or other name. An AI model can be considered a specific method for implementing an AI function. An AI model represents a mapping relationship or function between the input and output of a model. AI functions may include one or more of the following: data collection, model training (or model learning), model information release, model inference (or model reasoning, inference, or prediction, etc.), model monitoring or model verification, or inference result release, etc. AI functions may also be referred to as AI (related) operations, or AI-related functions.

[0440] The following is an exemplary description of the implementation process of the neural network with reference to the accompanying drawings.

[0441] 1. Fully connected neural network.

[0442] Also known as a multilayer perceptron (MLP), an MLP consists of an input layer (left), an output layer (right), and multiple hidden layers (center). Each layer of the MLP contains several nodes, called neurons. Neurons in adjacent layers are connected to each other.

[0443] Optionally, considering neurons in two adjacent layers, the output h of a neuron in the next layer is the weighted sum of all neurons x connected to it in the previous layer and passes through an activation function, which can be expressed as: h=f(wx+b).

[0444] Among them, w is the weight matrix, b is the bias vector, and f is the activation function.

[0445] Alternatively, the output of the neural network can be recursively expressed as: y = f n (w n f n-1 (…)+b n ).

[0446] Where n is the index of the neural network layer, 1<=n<=N, where N is the total number of neural network layers.

[0447] In other words, a neural network can be understood as a mapping from an input data set to an output data set. Neural networks are typically initialized randomly, and the process of obtaining this mapping from random w and b using existing data is called neural network training.

[0448] Optionally, the specific training method is to use a loss function to evaluate the output of the neural network. As shown in Figure 2b, the error can be backpropagated, and the neural network parameters (including w and b) can be iteratively optimized through gradient descent until the loss function reaches a minimum value, i.e., the "better point (e.g., optimal point)" in Figure 2b. It is understandable that the neural network parameters corresponding to the "better point (e.g., optimal point)" in Figure 2b can be used as the neural network parameters in the trained AI model information.

[0449] Alternatively, the gradient descent process can be expressed as:

[0450] Among them, θ is the parameter to be optimized (including w and b), L is the loss function, and η is the learning rate, which controls the step size of gradient descent. represents the derivative operation, represents the derivative of θ with respect to L.

[0451] Alternatively, the backpropagation process can utilize the chain rule for partial derivatives. As shown in Figure 2c, the gradient of the previous layer parameters can be recursively calculated from the gradient of the next layer parameters, which can be expressed as:

[0452] Among them, w ij is the weight of node j connecting to node i, s i is the weighted sum of the inputs to node i.

[0453] 2. Federated Learning (FL)

[0454] The concept of federated learning effectively addresses the current challenges facing AI development. While fully protecting user data privacy and security, it enables collaboration between edge devices and central servers to efficiently complete model learning tasks. As shown in Figure 2d, the FL architecture is the most widely used training architecture in the FL field. The FedAvg algorithm is the foundation of FL. Its algorithmic process is roughly as follows:

[0455] (1) The center initializes the model to be trained And broadcast it to all client devices.

[0456] (2) In the round t∈[1,T], client k∈[1,K] based on the local dataset For the received global model Perform E epochs of training to obtain local training results Report it to the central node.

[0457] (3) The central node aggregates and collects the local training results from all (or some) clients. Assume that the client set that uploads the local model in round t is The center will use the number of samples of the corresponding client as the weight to perform weighted averaging to obtain a new global model. The specific update rule is: The center then sends the latest version of the global model Broadcast to all client devices for a new round of training.

[0458] (4) Repeat steps (2) and (3) until the model finally converges or the number of training rounds reaches the upper limit.

[0459] In addition to reporting local models You can also use the local gradient of training After reporting, the central node averages the local gradients and updates the global model according to the direction of the average gradient.

[0460] As you can see, in the FL framework, datasets exist on distributed nodes. Distributed nodes collect local datasets, perform local training, and report the local training results (models or gradients) to the central node. The central node itself does not have a dataset; it is only responsible for fusing the training results of distributed nodes to obtain a global model and send it to the distributed nodes.

[0461] 3. Knowledge distillation.

[0462] A typical approach to heterogeneous model fusion is to convert models with different structures into a common structure through knowledge distillation, followed by fusion. Knowledge distillation involves transferring knowledge from one network to another, which can be either homogeneous or heterogeneous. Knowledge distillation can be used to transform a large network into a smaller one (for example, the parameter scale of the larger network is larger than that of the smaller network) while retaining performance close to that of the larger network. It can also transfer knowledge learned from multiple networks into a single network. Heterogeneous fusion involves a central node first collecting heterogeneous models from distributed nodes, distilling them into a homogeneous model, and then fusing them. The fused homogeneous model is then fed back to each distributed node, which then restores its own network structure to initiate the next round of fusion.

[0463] The technical solutions provided in this application can be applied to wireless communication systems (e.g., the systems shown in Figures 1a and 1b). In wireless communication systems, communication nodes generally have both signal transceiver capabilities and computing capabilities. For example, network devices with computing capabilities primarily provide computing power to support signal transceiver capabilities (e.g., calculating the time and frequency domain resources that carry the signal), thereby enabling communication between the network device and other communication nodes.

[0464] However, in communication networks, communication nodes may have excess computing power beyond just supporting the aforementioned communication tasks. Therefore, how to utilize this computing power is a pressing technical issue.

[0465] As an example, to meet the vision of intelligent and inclusive future, intelligence is likely to evolve further at the wireless network architecture level. In future communication systems, AI will likely be further integrated with wireless networks, enabling inherent network intelligence and terminal intelligence. For example, the integration of AI and wireless networks can be applied to the following new demands and scenarios: For example, more flexible and intelligent terminal connectivity, including but not limited to diverse terminal types, the Super Internet of Things (Supper IoT) (e.g., Supper IoT can include IoT, connected vehicles, industrial, and medical), massive connectivity, more flexible terminal connectivity, and terminals with inherent AI capabilities. Another example is inherent network intelligence: in addition to traditional communication connectivity services, the network will also provide computing and AI services to better support inclusive, real-time, and highly secure AI services. These new demands and scenarios will bring about changes in wireless network architecture and communication models.

[0466] Generally, the participating nodes of AI learning may include nodes in a plurality of distributed communication networks, such as terminal devices, network devices, etc. For example, in the network architecture with a central node shown in FIG2d, the central node may be a network device or a terminal device, and any distributed node may also be a network device or a terminal device. During the AI ​​learning process, different nodes may participate in AI learning based on a randomized neural network model, that is, after each node obtains the randomized model parameters, each node may initialize the neural network model locally based on the randomized model parameters to obtain an initialized neural network model, and execute the AI ​​learning process based on the initialized neural network model until the AI ​​neural network model converges to a preset condition (for example, the loss function of the neural network model in FIG2b indicates that the model converges to a better point (for example, the optimal point), or the number of iterations of the neural network model reaches a threshold, etc.).

[0467] However, the local information of different nodes may be different, for example, one or more of the computing resources of different nodes (such as resources of modules / devices such as CPU and GPU), the storage resources of different nodes (such as resources of modules / devices such as ROM and RAM), and the utilization rate of local computing resources of different nodes may be different. For this reason, in the process of AI learning based on randomized model parameters of the above-mentioned nodes, the initialized neural network model obtained by each node is randomized, which may lead to a slower convergence of the AI ​​neural network model, and thus affect the processing efficiency of AI learning.

[0468] In order to solve the above problems, the present application provides a communication method and related equipment for enabling the computing power of communication nodes to be applied to artificial intelligence (AI) learning and improving the implementation flexibility of different nodes.

[0469] Please refer to FIG3 , which is a schematic diagram of an implementation of the communication method provided in this application. The method includes the following steps.

[0470] It should be noted that, in FIG3 , the first node and the second node are used as examples of the execution subjects of the interactive diagram to illustrate the method, but the present application does not limit the execution subjects of the interactive diagram. For example, the execution subject of S301 in FIG3 and the corresponding implementation is the first node, and the execution subject may also be a chip, chip system, or processor that supports the first node to implement the method, or a logic module or software that can implement all or part of the functions of the first node. The second node in S301-S302 in FIG3 and the corresponding implementation may also be replaced by a chip, chip system, or processor that supports the second node to implement the method, or may be replaced by a logic module or software that can implement all or part of the functions of the second node.

[0471] S301. The second node sends first information, and the first node receives the first information accordingly. The first information is used to indicate N configuration information, and the N configuration information is used to configure resources of N model parameters, where N is a positive integer.

[0472] It should be noted that different first nodes may be distributed nodes, and the second node may be connected to one or more first nodes. The second node may serve as a central node (or control node, management node, etc.) for the one or more first nodes; or the second node may serve as a server, and one or more first nodes may serve as clients.

[0473] It should be understood that the first information is used to indicate N configuration information, which can be understood as: the first information includes the index of the N configuration information, so that the recipient of the first information can obtain the N configuration information based on the index; or, the first information includes the N configuration information, so that the recipient of the first information can obtain the N configuration information from the first information. In addition, the N configuration information is used to configure the resources of N model parameters, which can be understood as the N configuration information is used to configure the resources of N model parameters respectively, or the N configuration information has a one-to-one correspondence with the resources of N model parameters. In addition, the second node can send the N configuration information through one or more information / messages, that is, the N configuration information indicated by the first information can be completed through a single sending process or a multiple sending process. Accordingly, the first node can also receive the N configuration information through a single receiving process or a multiple receiving process.

[0474] Similarly, the second information mentioned later is used to indicate M configuration information, which can be understood as: the second information includes the index of the M configuration information, so that the recipient of the second information can obtain the M configuration information based on the index; or, the second information includes the M configuration information, so that the recipient of the second information can obtain the M configuration information from the M configuration information. In addition, the M configuration information is used to configure the resources of M model parameters, which can be understood as the M configuration information is used to configure the resources of M model parameters respectively, or the M configuration information corresponds one-to-one to the resources of M model parameters. In addition, the second node can send N configuration information through one or more information / messages, that is, the M configuration information indicated by the second information can be completed through a one-time sending process or a multiple-time sending process. Correspondingly, the first node can also receive the M configuration information through a one-time receiving process or a multiple-time receiving process.

[0475] Similarly, the third information mentioned later is used to indicate K configuration information, which can be understood as: the third information includes the index of the K configuration information, so that the recipient of the third information can obtain the K configuration information based on the index; or, the third information includes the K configuration information, so that the recipient of the third information can obtain the K configuration information from the K configuration information. In addition, the K configuration information is used to configure the resources of K model parameters, which can be understood as the K configuration information is used to configure the resources of K model parameters respectively, or the K configuration information corresponds one-to-one to the resources of K model parameters. In addition, the second node can send N configuration information through one or more information / messages, that is, the K configuration information indicated by the third information can be completed through a one-time sending process or a multiple-time sending process. Correspondingly, the first node can also receive the K configuration information through a one-time receiving process or a multiple-time receiving process.

[0476] In one possible implementation, in step S301, for any model parameter configured by the N pieces of configuration information indicated by the first information sent by the second node, the model parameter includes at least one of a model weight, a model activation function, and a model bias. Thus, the N model parameters transmitted by the resources configured by the N pieces of configuration information include at least one of the aforementioned items, thereby improving the flexibility of solution implementation.

[0477] S302. The first node receives a first model parameter based on the first configuration information among the N configuration information, where the first model parameter is used to determine a first neural network model corresponding to the local information.

[0478] Optionally, AI learning may be replaced with machine learning, AI processing, AI neural network processing, etc.

[0479] It should be understood that the resources configured by the N pieces of configuration information include one or more of time domain resources, frequency domain resources, and spatial domain resources. In step S302, the first node receiving the first model parameter based on the first piece of configuration information among the N pieces of configuration information can be understood as the first node receiving the first model parameter based on the resources configured by the first piece of configuration information among the N pieces of configuration information.

[0480] It should be understood that in step S301, among the N model parameters configured by the N configuration information indicated by the first information, the model parameters are used to initialize the neural network model, or in other words, the model parameters include the initialization parameters of the neural network model. The initialization parameters may be parameters used to determine the use of the neural network model, or the initialization parameters may be parameters used to determine the start of neural network model training. In other words, for the first node, after the first node receives the first model parameters based on the first configuration information in step S301, in step S302, the first node may perform initialization processing on the neural network model based on the first model parameters to obtain a first neural network model corresponding to the local information; or, in step S302, the first node may use the first model parameters as the initialization parameters of the neural network model, and obtain the first neural network model after completing the local parameter initialization.

[0481] In one possible implementation, the first model parameter received by the first node in step S302 is used to determine a first neural network model corresponding to local information, where the local information includes at least one of a local resource overhead level, local AI capability information, and a local parameter scale. Thus, the local information can be implemented in a variety of ways to provide flexibility in solution implementation.

[0482] It should be understood that the local resource overhead level can be understood as the utilization rate of local computing resources, the utilization rate of local storage resources, the usage of local power consumption, etc. For example, the local computing resources may include resources of modules / devices such as a central processor unit (CPU), a microcontroller unit (MCU), or a graphics processing unit (GPU); the local storage resources may include resources of modules / devices such as a read-only memory (ROM) and a random access memory (RAM); the usage of local power consumption may include the health of a battery module, the current amount of power used by the battery module (or the percentage of the power), the current amount of power not used by the battery module (or the percentage of the power), etc.

[0483] It should be understood that local AI capability information may include local AI-related capability information, such as AI-related computing capability, storage capability, etc. Similarly, AI-related computing capability may include the computing capability of modules / devices such as CPU, MCU, and GPU, and AI-related storage capability may include the storage capability of modules / devices such as ROM and RAM.

[0484] It should be understood that the local parameter scale can be understood as the parameter scale of the local neural network model of the first node, including but not limited to the number of neural network layers of the neural network model, the number of neurons of the neural network model, the input and output of the neural network model, or the different computational complexity of the neural network model.

[0485] It should be noted that the local information of the first node can be implemented in a variety of ways. Accordingly, for the first node, after the first node receives the first model parameters, the first node can determine the first neural network model corresponding to the local information in a variety of ways.

[0486] In one implementation example, taking the example of the local information including the local resource overhead level, the local resource overhead level can be reflected by CPU utilization; wherein the first node can determine the mapping relationship between different CPU utilizations and different neural network models through configuration or pre-configuration. Taking the example of the different neural network models including a first type of neural network model and a second type of neural network model, the first type of neural network model can be more complex than the second type of neural network model (for example, the scale of model parameters is larger, the number of neural network layers of the model is greater, etc.), and the mapping relationship can indicate that nodes with a CPU utilization of 0 to X% (X is greater than 0 and less than 100) correspond to the first type of neural network model, and indicate that nodes with a CPU utilization of X% to 100% correspond to the second type of neural network model. Thus, in the above technical solution, the first node can compare the local CPU utilization with X%. If the local CPU utilization is less than or equal to X%, the first neural network model obtained by the first node based on the first model parameters can be the first type of neural network model; if the local CPU utilization is greater than or equal to X%, the first neural network model obtained by the first node based on the first model parameters can be the second type of neural network model.

[0487] Optionally, in the above implementation example, the measurement of CPU utilization can be determined under the premise that the nodes execute the same or similar calculation parameters, that is, the CPU utilization is determined under relatively fair conditions. For example, for the training of two models, when the input parameters are the same, different nodes have different CPU utilizations required to complete a round of training. For example, when the CPU utilization of a certain node is low, it can be understood that the computing power of the CPU of the node is strong, and the initialization processing can be performed based on a larger scale of model parameters (or more complex model parameters); when the CPU utilization of a certain node is high, it can be understood that the computing power of the CPU of the node is weak, and the initialization processing can be performed based on a smaller scale of model parameters (or simpler model parameters). Accordingly, based on this difference, different model parameters can be configured for the different nodes, so that each node can obtain a differentiated (or personalized) initialized neural network model.

[0488] In another implementation example, taking the local information including local AI capability information as an example, the local AI capability information can be reflected by the storage capacity of the RAM; wherein the first node can determine the mapping relationship between different RAM storage capabilities and different neural network models by configuration or pre-configuration. Taking the different neural network models including the first type of neural network model, the second type of neural network model and the third type of neural network model as an example, the first type of neural network model can be more complex than the second type of neural network model (for example, the scale of model parameters is larger, the number of neural network layers of the model is greater, etc.), and the second type of neural network model can be more complex than the third type of neural network model; the mapping relationship can indicate that the node with a RAM storage capacity of 0 to Y gigabytes (GB) (Y is greater than 0) corresponds to the first type of neural network model, and indicates that the node with a RAM storage capacity greater than Y GB and less than or equal to Z GB (Z is greater than Y) corresponds to the second type of neural network model, and indicates that the node with a RAM storage capacity greater than Z GB corresponds to the third type of neural network model. Therefore, in the above technical solution, the first node can compare with Y and Z based on the storage capacity of the local RAM. When the storage capacity of the local RAM is less than or equal to Y GB, the first neural network model obtained by the first node based on the first model parameters can be a first-type neural network model; when the storage capacity of the local RAM is greater than Y GB and less than or equal to Z GB, the first neural network model obtained by the first node based on the first model parameters can be a second-type neural network model; when the storage capacity of the local RAM is greater than Z GB, the first neural network model obtained by the first node based on the first model parameters can be a third-type neural network model.

[0489] In one possible implementation, in step S301, the parameter scales of the N model parameters configured by the N pieces of configuration information indicated by the first information sent by the second node are different, and N is greater than 1. Before step S302, the method further includes: the first node determining the first configuration information from the N pieces of configuration information based on the local information. Specifically, when the N pieces of configuration information indicated by the first information are greater than 1, the parameter scales of the N model parameters configured by the N pieces of configuration information are different, so that the first node can determine the first configuration information corresponding to the local information from the N pieces of configuration information based on the local information, and the first node can obtain the model parameters corresponding to the local information based on the first configuration information.

[0490] It should be understood that the different parameter scales of the N model parameters can be understood as different requirements for the local information of the node. For example, if the local information includes the local resource overhead level, the model parameter with a larger parameter scale among the N model parameters requires the local resource overhead level to be below a threshold in order to provide more unused resources to process the model parameter with a larger parameter scale. For another example, if the local information includes local AI capability information, the model parameter with a larger parameter scale among the N model parameters requires the local AI capability information to be above a threshold in order to provide more AI resources to process the model parameter with a larger parameter scale.

[0491] Optionally, the parameter scales of the N model parameters are different and can be replaced by other descriptions, for example, the number of neural network layers corresponding to the N model parameters is different, the neurons corresponding to the N model parameters are different, the input and output corresponding to the N model parameters are different, the computational complexity of the N model parameters is different, or the computing power requirements of the N model parameters are different, or the storage capacity requirements of the N model parameters are different, etc.

[0492] As an implementation example, taking the second node as a network device and the first node as a terminal device as an example, the N model parameters may include the three model parameters in Figure 4a. These three model parameters can be used for initialization processing of three types of models in Figure 4a, namely, the Cat 0 model, the Cat 1 model, and the Cat 2 model in Figure 4a. As can be seen from the example shown in Figure 4a, different models have different numbers of neural network layers. Therefore, the terminal device can receive one of the N model parameters based on the locally deployed model. For example, the terminal device that deploys the type 0 model (i.e., the Cat0 terminal device in the figure) can receive the model parameters of the Cat 0 Model, the terminal device that deploys the type 1 model (i.e., the Cat1 terminal device in the figure) can receive the model parameters of the Cat 1 Model, and the terminal device that deploys the type 2 model (i.e., the Cat2 terminal device in the figure) can receive the model parameters of the Cat 2 Model, so that each terminal device can receive model parameters that match the local information, and then perform initialization processing based on the received model parameters. The initialized neural network model can be used for subsequent AI learning (e.g., model training).

[0493] In one possible implementation, in step S301, the N configuration information indicated by the first information sent by the second node is one configuration information (i.e., the first configuration information, and the value of N is 1). The first model parameters are used to configure a neural network model shared by at least two first nodes; after step S302, the method further includes: the first node performs knowledge distillation processing on the neural network model shared by the at least two first nodes based on the local information to obtain the first neural network model. Specifically, in the case where the N configuration information indicated by the first information is one configuration information (i.e., the first configuration information), the first model parameters configured by the first configuration information are used to configure the neural network model shared by at least two first nodes, so that the at least two first nodes can obtain a common neural network model based on the first configuration information, and perform knowledge distillation processing on the common neural network model based on their respective local information to obtain a first neural network model corresponding to the local information.

[0494] It should be understood that the neural network model shared by at least two first nodes can be understood as a neural network model that can be recognized / processed by the at least two first nodes. Alternatively, the neural network model shared by at least two first nodes can be expressed as: a neural network model common to at least two first nodes, and a neural network model that can be recognized / processed by at least two first nodes.

[0495] In one possible implementation, before the first node receives the first information in step S301, the method further includes: the first node receiving fourth information, the fourth information being used to indicate a resource carrying request information, the request information being used to request the N pieces of configuration information; and the first node sending the request information on the resource carrying the request information. Specifically, the first node may also receive fourth information indicating a resource carrying the request information, the request information being used to request the N pieces of configuration information, so that the first node can obtain the N pieces of configuration information through the first information after sending the request information based on the resource indicated by the fourth information.

[0496] Optionally, the request information includes a first orthogonal sequence, and the first orthogonal sequence corresponds to the local information.

[0497] Optionally, the first orthogonal sequence is included in P orthogonal sequences, where the P orthogonal sequences correspond to different local information. In other words, different first nodes can each determine one of the P orthogonal sequences based on their respective local information and send the one of the orthogonal sequences to the second node, so that a recipient of the request information (e.g., the second node) can obtain the local information of each first node based on the received request information.

[0498] Optionally, when P is greater than 1, the different sequences can be orthogonal. That is, the P sequences can be expressed as P orthogonal sequences, so that the P orthogonal sequences can be identified by the receiver even when transmitted via the same resources. In other implementations, the P sequences can also be identified by the receiver in other ways, such as by staggering frequency or time domain resources. In other words, the aforementioned P orthogonal sequences can be replaced by P sequences, and the first orthogonal sequence can be replaced by the first sequence.

[0499] In addition, when the second node acts as the recipient of the request information, the second node can determine the local information of each first node based on one or more received orthogonal sequences, and can use the local information of each first node as the basis for determining at least one of N configuration information, M configuration information, and K configuration information, so that each configuration information sent by the second node can be adapted to the local information of each first node.

[0500] In addition, since different first nodes can indicate their own local parameter scales through the request information, overhead can be reduced.

[0501] Optionally, in addition to determining the resources carrying the first orthogonal sequence based on the fourth information, the first node may also determine the resources carrying the first orthogonal sequence in other ways (such as a preconfigured way), which is not limited here.

[0502] Optionally, the value of P may be greater than or equal to the value of N. For example, the first node may obtain the different parameter scales corresponding to the P orthogonal sequences respectively through configuration or pre-configuration, and the second node may determine the parameter scales of the respective first nodes by receiving request information from one or more first nodes. Accordingly, the second node may further generate the N configuration information based on the determined parameter scales of the respective first nodes. Generally, the P orthogonal sequences may have a redundant setting (i.e., P is greater than or equal to N) so that each first node can indicate its own local parameter scale through request information.

[0503] In one possible implementation, before the first node receives the first information, the method further includes: the first node receiving fifth information, the fifth information being used to indicate local information corresponding to the N model parameters, and / or the fifth information being used to indicate whether the first information includes the N configuration information. Specifically, the first node may further receive the fifth information to obtain relevant information about the N configuration information through the fifth information, and determine whether to receive the N configuration information based on the obtained information, thereby avoiding unnecessary detection by the first node and reducing overhead.

[0504] As an implementation example, the request information, the fifth information, and the first information can be carried via different time domain resources. As shown in Figure 4b, multiple first nodes can send request information on the time domain resource where the request information is located, so that the second node can obtain the local parameter scale of each first node. Thereafter, the second node can send relevant information about the N configuration information indicated by the first information in the fifth information, so that the first node can determine whether to receive the N configuration information based on the fifth information, thereby avoiding unnecessary detection by the first node and reducing overhead.

[0505] In one possible implementation, in the method shown in Figure 3, the method further includes: the first node receives second information, the second information indicating M configuration information, the M configuration information being used to configure transmission resources for M model parameters, the parameter scales of the M model parameters being different, and M being an integer greater than 1; the first node receives a second model parameter based on the second configuration information in the M configuration information, wherein the parameter scale of the second model parameter is the same as the local parameter scale (or similar, for example, the difference between the parameter scale of the second model parameter and the local parameter scale is lower than a threshold). Specifically, the first node may also receive second information indicating M configuration information, and receive a second model parameter that is the same (or similar) as the local parameter scale based on the second configuration information of the M configuration information. In other words, different first nodes can receive model parameters that are the same (or similar) as the parameter scale of their respective local information through the second information, and perform AI processing (such as local training) based on the received model parameters.

[0506] It should be understood that the transmission resources configured by the M configuration information are used for different first nodes to receive one of the M model parameters, so that different first nodes can subsequently perform AI processing (such as local training) based on the received model parameters. Accordingly, the M model parameters can be understood as global model parameters of multiple first nodes, that is, the M model parameters can be replaced by M global model parameters.

[0507] Optionally, for any model parameter among the M model parameters, a gradient and / or a weight may be included. For example, the gradient and / or the weight are used for further processing of the initialized neural network model (e.g., the first neural network model described above), such as training the neural network model, updating the neural network model, etc.

[0508] Similarly, the parameter scales of the M model parameters are different and can be replaced by other descriptions, for example, the number of neural network layers corresponding to the M model parameters is different, the neurons corresponding to the M model parameters are different, the input and output corresponding to the M model parameters are different, the computational complexity of the M model parameters is different, or the computing power requirements of the M model parameters are different, or the storage capacity requirements of the M model parameters are different, etc.

[0509] In a possible implementation, any one of the M pieces of configuration information includes at least one of the following:

[0510] Indication information indicating parameter scales of model parameters configured by the one configuration information;

[0511] Indication information indicating the received power of the transmission resource configured by the one configuration information;

[0512] Indication information indicating a parameter type of a model parameter configured by the configuration information, where the parameter type includes a gradient and / or a weight;

[0513] Indication information indicating the frequency domain resources of the transmission resources configured by the one configuration information;

[0514] Indication information indicating the time domain resources of the transmission resources configured by the one configuration information.

[0515] Specifically, any one of the M pieces of configuration information indicated by the second information may include at least one of the above items to enhance the flexibility of solution implementation.

[0516] Optionally, any one of the N pieces of configuration information indicated by the first information received by the first node in step S301 may also include at least one of the following items to improve the flexibility of the solution implementation. The at least one item includes:

[0517] Indication information indicating parameter scales of model parameters configured by the one configuration information;

[0518] Indication information indicating the received power of the transmission resource configured by the one configuration information;

[0519] Indication information indicating a parameter type of a model parameter configured by the configuration information, where the parameter type includes a gradient and / or a weight;

[0520] Indication information indicating the frequency domain resources of the transmission resources configured by the one configuration information;

[0521] Indication information indicating the time domain resources of the transmission resources configured by the one configuration information.

[0522] In one possible implementation, before the first node receives the second information, the method further includes: the first node receiving sixth information, the sixth information being used to indicate local information corresponding to the M model parameters, and / or the sixth information being used to indicate whether the third information includes the M configuration information. Specifically, the first node may further receive the sixth information to obtain relevant information about the M configuration information through the sixth information, and determine whether to receive the M configuration information based on the obtained information, thereby avoiding unnecessary detection by the first node and reducing overhead.

[0523] In one possible implementation, in the method shown in Figure 3, the method further includes: the first node receives a third message, the third message indicates K configuration information, the K configuration information is used to configure the transmission resources of K model parameters, the parameter scales of the K model parameters are different, and K is an integer greater than 1; the first node sends a third model parameter based on the third configuration information in the K configuration information, wherein the parameter scale of the third model parameter is the same as the local parameter scale (or similar, for example, the difference between the parameter scale of the third model parameter and the local parameter scale is lower than a threshold). Specifically, the first node may also receive a third message indicating K configuration information, and receive a third model parameter that is the same (or similar) as the local parameter scale based on the third configuration information of the K configuration information. In other words, different first nodes can send model parameters that are the same (or similar) as the parameter scale of their respective local information through the third message, so that the recipient of the model parameter can perform AI processing (for example, parameter fusion processing) on ​​the model parameters sent by different first nodes.

[0524] It should be understood that the transmission resources configured by the K configuration information are used by different first nodes to transmit one of the K model parameters, so that the recipient of the model parameter can perform a fusion operation on the model parameters transmitted by the different first nodes. Accordingly, the K model parameters can be understood as the local model parameters of multiple first nodes, that is, the K model parameters can be replaced by K local model parameters.

[0525] Optionally, for any model parameter among the K model parameters, a gradient and / or a weight may be included. For example, the gradient and / or the weight are used for further fusion processing of the neural network model, such as deep fusion, weighted averaging, etc.

[0526] Similarly, the parameter scales of the K model parameters are different and can be replaced by other descriptions, for example, the number of neural network layers corresponding to the K model parameters is different, the neurons corresponding to the K model parameters are different, the input and output corresponding to the K model parameters are different, the computational complexity of the K model parameters is different, or the computing power requirements of the K model parameters are different, or the storage capacity requirements of the K model parameters are different, etc.

[0527] In one possible implementation, any one of the K pieces of configuration information includes at least one of the following:

[0528] Indication information indicating parameter scales of model parameters configured by the one configuration information;

[0529] Indication information indicating the transmit power of the transmission resource configured by the one configuration information;

[0530] Indication information indicating a parameter type of a model parameter configured by the configuration information, where the parameter type includes a gradient and / or a weight;

[0531] Indication information indicating the frequency domain resources of the transmission resources configured by the one configuration information;

[0532] Indication information indicating the time domain resource of the transmission resource configured by the one configuration information;

[0533] Indication information indicating the index of the transmission resource of the one configuration information in the transmission resources configured by the K configuration information.

[0534] It should be understood that when any one of the configuration information includes indication information indicating the index of the transmission resource of the one configuration information in the transmission resources configured by the K configuration information, it can be understood that the second node configures a transmission resource set (including at least one of time domain resources, frequency domain resources, and spatial domain resources) for the K configuration information, so that the first node can determine the corresponding transmission resource in the transmission resource set based on the index of the indication information.

[0535] Specifically, any one of the K pieces of configuration information indicated by the third information may include at least one of the above items to enhance the flexibility of solution implementation.

[0536] In one possible implementation, before the first node receives the third information, the method further includes: the first node receiving seventh information, the seventh information being used to indicate local information corresponding to the K model parameters, and / or the seventh information being used to indicate whether the second information includes the K configuration information. Specifically, the first node may further receive the seventh information to obtain relevant information about the K configuration information through the seventh information, and determine whether to receive the K configuration information based on the obtained information, thereby avoiding unnecessary detection by the first node and reducing overhead.

[0537] Optionally, the values ​​of N / M / K may be different from each other, or at least two of them may be the same, or all three may be the same, which is not limited here.

[0538] As an implementation example, take the case where the resources configured by M configuration information are used to transmit M global model parameters, and the resources configured by K configuration information are used to transmit N local model parameters. As shown in Figure 4c, when the values ​​of M and K are both 4, in the time domain, the first node of Cat0, the first node of Cat1, the first node of Cat2, and the first node of Cat3 can respectively receive one of the M global model parameters on the corresponding time domain resources, so that each first node can perform AI processing (such as AI training processing) locally based on the global model parameter. Thereafter, in the time domain, the first node of Cat0, the first node of Cat1, the first node of Cat2, and the first node of Cat3 can respectively send one of the K local model parameters on the corresponding time domain resources, so that the second node can perform AI processing (such as AI fusion processing) based on the received local model parameters.

[0539] Based on the technical solution shown in Figure 3, the first information received by the first node in step S301 indicates N configuration information, and the N configuration information is used to configure resources for N model parameters. Thereafter, the first node receives the first model parameter based on the resource configured by the first configuration information in the N configuration information in step S302, and the first node can determine the first neural network model corresponding to the local information based on the first model parameter. In other words, different first nodes can all obtain the neural network model corresponding to their respective local information based on the first information, and can subsequently participate in AI learning based on the locally obtained neural network model. Compared to the implementation method in which different nodes participate in AI learning based on randomized neural network models, in the above technical solution, different nodes can participate in AI learning based on the neural network model corresponding to their respective local information. Therefore, when the communication node in the communication system acts as a node participating in AI learning, the computing power of the communication node can be applied to AI tasks in the learning system, and the implementation flexibility of different nodes can be improved.

[0540] In addition, the local information of different nodes is generally different. In the above technical solution, the first node can determine the neural network model corresponding to the local information based on the first information. To this end, different first nodes can participate in AI learning based on the neural network model adapted to the local information to improve the processing efficiency of the neural network model.

[0541] Please refer to FIG5 , which is another schematic diagram of an implementation of the communication method provided in this application. The method includes the following steps.

[0542] It should be noted that, in FIG5 , the first node and the second node are used as examples of the execution subjects of the interactive diagram to illustrate the method, but the present application does not limit the execution subjects of the interactive diagram. For example, the execution subject of S501 in FIG5 and the corresponding implementation is the first node, and the execution subject may also be a chip, chip system, or processor that supports the first node to implement the method, or a logic module or software that can implement all or part of the functions of the first node. The second node in S501-S502 in FIG5 and the corresponding implementation may also be replaced by a chip, chip system, or processor that supports the second node to implement the method, or may be replaced by a logic module or software that can implement all or part of the functions of the second node.

[0543] S501. The second node sends second information, and the first node receives the second information accordingly. The second information is used to indicate M configuration information, and the M configuration information is used to configure transmission resources of M model parameters, where the parameter scales of the M model parameters are different, and M is an integer greater than 1.

[0544] S502. The first node receives a second model parameter based on second configuration information among the M configuration information, wherein a parameter scale of the second model parameter is the same as a local parameter scale (or similar, for example, a difference between the parameter scale of the second model parameter and the local parameter scale is less than a threshold).

[0545] Based on the technical solution shown in Figure 5, the second information received by the first node in step S501 indicates M configuration information, and the M configuration information is used to configure the resources of M model parameters. Thereafter, in step S502, the first node receives the second model parameter based on the resource configured by the second configuration information in the M configuration information, and the second model parameter is the same (or similar) as the local parameter scale. In other words, different first nodes can all obtain model parameters with the same (or similar) parameter scale as the local parameter scale based on the first information, and can subsequently participate in AI learning based on model parameters with the same (or similar) parameter scale as the local parameter scale. Compared with the implementation method in which different nodes participate in AI learning based on randomized model parameters, in the above technical solution, different nodes can participate in AI learning based on model parameters with the same (or similar) parameter scale as the local parameter scale. Thus, when the communication node in the communication system acts as a node participating in AI learning, the computing power of the communication node can be applied to the AI ​​task in the learning system, and the implementation flexibility of different nodes can be improved.

[0546] In addition, the local parameter scales of different nodes may be different. In the above technical solution, the first node can participate in AI learning based on model parameters that are the same (or similar) to the local parameter scale. To this end, different first nodes can participate in AI learning based on model parameters that are adapted to local information to improve the processing efficiency of the neural network model.

[0547] It should be noted that the technical solution shown in Figure 5 can refer to the description of the aforementioned Figure 3 and related embodiments, and achieve corresponding technical effects, such as local information, model parameters, M configuration information, third information, K configuration information, sixth information, seventh information, etc.

[0548] Please refer to FIG6 , which is another schematic diagram of an implementation of the communication method provided in this application. The method includes the following steps.

[0549] It should be noted that, in FIG6 , the first node and the second node are used as examples of the execution subjects of the interactive diagram to illustrate the method, but the present application does not limit the execution subjects of the interactive diagram. For example, the execution subject of S601 in FIG6 and the corresponding implementation is the first node, and the execution subject may also be a chip, chip system, or processor that supports the first node to implement the method, or a logic module or software that can implement all or part of the functions of the first node. The second node in S601-S602 in FIG6 and the corresponding implementation may also be replaced by a chip, chip system, or processor that supports the second node to implement the method, or may be replaced by a logic module or software that can implement all or part of the functions of the second node.

[0550] S601. The second node sends third information, and the first node receives the third information accordingly. The third information indicates K configuration information, the K configuration information being used to configure transmission resources of K model parameters, the K model parameters having different parameter scales, and K being an integer greater than 1.

[0551] S602. The first node sends a third model parameter based on the third configuration information among the K configuration information, wherein a parameter scale of the third model parameter is the same as a local parameter scale (or similar, for example, a difference between the parameter scale of the third model parameter and the local parameter scale is lower than a threshold).

[0552] Based on the technical solution shown in Figure 6, the third information received by the first node in step S601 indicates K configuration information, and the K configuration information is used to configure the resources of K model parameters. Thereafter, in step S602, the first node sends the third model parameter based on the resource configured by the third configuration information in the K configuration information, and the third model parameter is the same (or similar) as the local parameter scale. In other words, different first nodes can all obtain model parameters with the same (or similar) parameter scale as the local parameter scale based on the first information, and can subsequently participate in AI learning based on model parameters with the same (or similar) parameter scale as the local parameter scale. Compared with the implementation method in which different nodes participate in AI learning based on randomized model parameters, in the above technical solution, different nodes can participate in AI learning based on model parameters with the same (or similar) parameter scale as the local parameter scale. Thus, when the communication node in the communication system acts as a node participating in AI learning, the computing power of the communication node can be applied to the AI ​​task in the learning system, and the implementation flexibility of different nodes can be improved.

[0553] In addition, the local parameter scales of different nodes may be different. In the above technical solution, the first node can participate in AI learning based on model parameters that are the same (or similar) to the local parameter scale. To this end, different first nodes can participate in AI learning based on model parameters that are adapted to local information to improve the processing efficiency of the neural network model.

[0554] It should be noted that the technical solution shown in Figure 6 can refer to the description of Figure 3 and related embodiments above and achieve corresponding technical effects, such as local information, model parameters, third information, K configuration information, seventh information, etc.

[0555] As an implementation example of the technical solutions shown in Figures 3 to 6 above, as shown in Figure 7, relevant steps can be executed between the second node and the first node through the flow direction in the time domain shown in Figure 7.

[0556] In an implementation example, the second node can send the first information involved in the technical solution shown in Figure 3 to the first node through a resource grant (Grant) method of scheduling, so that the first node obtains model parameters based on the N configuration information indicated by the first information, and performs an initialization process based on the model parameters to obtain a first neural network model. Subsequently, an AI processing process can be performed based on the first neural network model (for example, AI training processing of the first node, AI fusion processing of the second node, etc.).

[0557] Optionally, before the second node schedules the first information, the second node may also send fourth information to the first node, so that the first node sends request information based on the resources configured by the fourth information. The request information can be used to trigger the second node to schedule the resources indicated by N configuration information.

[0558] In another implementation example, the second node can send the second information involved in the technical solution shown in Figure 3 / Figure 5 / Figure 6 to the first node through a resource grant (Grant) of scheduling, so that the first node receives the global model parameters based on the M configuration information indicated by the second information, and performs the AI ​​processing process based on the global model parameters (for example, AI training processing of the first node, AI fusion processing of the second node, etc.).

[0559] In another implementation example, the second node can send the third information involved in the technical solution shown in Figure 3 / Figure 5 / Figure 6 to the first node through a resource grant (Grant) of scheduling, so that the first node sends local model parameters based on the K configuration information indicated by the third information, and performs AI processing based on the global model parameters (for example, AI training processing of the first node, AI fusion processing of the second node, etc.).

[0560] Optionally, during the AI ​​processing process, the first node can feedback (Feedback) the local AI processing results to the second node, and accordingly, the second node can also adjust (Adjustment) the relevant information of the AI ​​processing (such as model gradient / model weight, etc.) to the first node.

[0561] Optionally, after the AI ​​processing, the second node may further send an indicator indicating that the AI ​​processing has terminated. In other words, the first node may determine that the AI ​​processing has terminated after detecting the indicator through event monitoring.

[0562] Referring to Figure 8, an embodiment of the present application provides a communication device 800, which can implement the functions of the first node or the second node (the first node or the second node is a terminal device or a network device) in the above-mentioned method embodiment, and thus can also achieve the beneficial effects possessed by the above-mentioned method embodiment. In the embodiment of the present application, the communication device 800 can be the first node (or the second node), or it can be an integrated circuit or component inside the first node (or the second node), such as a chip. The following embodiments are described by taking the communication device 800 as the first node (or the second node) as an example.

[0563] It should be noted that the transceiver unit 802 may include a sending unit and a receiving unit, which are respectively used to perform sending and receiving.

[0564] In one possible implementation, when the device 800 is used to execute the method executed by the first node in any of the aforementioned embodiments, the device 800 includes a processing unit 801 and a transceiver unit 802; the transceiver unit 802 is used to receive first information, where the first information indicates N configuration information, and the N configuration information is used to configure resources of N model parameters, where N is a positive integer; the processing unit 801 is used to determine the first configuration information among the N configuration information; the transceiver unit 802 is also used to receive a first model parameter based on the first configuration information, and the first model parameter is used to determine a first neural network model corresponding to the local information.

[0565] In a possible implementation, the local information includes at least one of a local resource overhead level, local AI capability information, and a local parameter scale.

[0566] In a possible implementation, the model parameter includes at least one of a model weight, a model activation function, and a model bias.

[0567] In a possible implementation, N is an integer greater than 1, the parameter scales of the N model parameters are different, and the processing unit 801 is further configured to determine the first configuration information among the N configuration information based on the local information.

[0568] In one possible implementation, the value of N is 1, wherein the first model parameter is used to configure a neural network model shared by at least two first nodes; the processing unit 801 is also used to perform knowledge distillation processing on the neural network model shared by at least two first nodes based on the local information to obtain the first neural network model.

[0569] In one possible implementation, the transceiver unit 802 is also used to receive second information, where the second information indicates M configuration information, and the M configuration information is used to configure transmission resources of M model parameters, where the parameter scales of the M model parameters are different, and M is an integer greater than 1; the processing unit 801 is also used to determine the second configuration information among the M configuration information; the transceiver unit 802 is also used to receive a second model parameter based on the second configuration information, where the parameter scale of the second model parameter is the same as (or similar to) the local parameter scale.

[0570] In a possible implementation, any one of the M pieces of configuration information includes at least one of the following:

[0571] Indication information indicating parameter scales of model parameters configured by the one configuration information;

[0572] Indication information indicating the received power of the transmission resource configured by the one configuration information;

[0573] Indication information indicating a parameter type of a model parameter configured by the configuration information, where the parameter type includes a gradient and / or a weight;

[0574] Indication information indicating the frequency domain resources of the transmission resources configured by the one configuration information;

[0575] Indication information indicating the time domain resources of the transmission resources configured by the one configuration information.

[0576] In one possible implementation, the transceiver unit 802 is also used to receive third information, where the third information indicates K configuration information, where the K configuration information is used to configure transmission resources for K model parameters, where the parameter scales of the K model parameters are different, and K is an integer greater than 1; the processing unit 801 is also used to determine the third configuration information among the K configuration information; the transceiver unit 802 is also used to send a third model parameter based on the third configuration information, where the parameter scale of the third model parameter is the same as (or similar to) the local parameter scale.

[0577] In one possible implementation, any one of the K pieces of configuration information includes at least one of the following:

[0578] Indication information indicating parameter scales of model parameters configured by the one configuration information;

[0579] Indication information indicating the transmit power of the transmission resource configured by the one configuration information;

[0580] Indication information indicating a parameter type of a model parameter configured by the configuration information, where the parameter type includes a gradient and / or a weight;

[0581] Indication information indicating the frequency domain resources of the transmission resources configured by the one configuration information;

[0582] Indication information indicating the time domain resource of the transmission resource configured by the one configuration information;

[0583] Indication information indicating the index of the transmission resource of the one configuration information in the transmission resources configured by the K configuration information.

[0584] In one possible implementation, the transceiver unit 802 is further used to receive fourth information, which is used to indicate the resources carrying the request information, and the request information is used to request the N configuration information; the transceiver unit 802 is also used to send the request information on the resources carrying the request information.

[0585] In a possible implementation manner, the request information includes a first orthogonal sequence, and the first orthogonal sequence corresponds to the local information.

[0586] In one possible implementation, the transceiver unit 802 is further used to receive fifth information, where the fifth information is used to indicate local information corresponding to the N model parameters respectively, and / or the fifth information is used to indicate whether the first information includes the N configuration information.

[0587] In one possible implementation, when the device 800 is used to execute the method executed by the second node in any of the aforementioned embodiments, the device 800 includes a processing unit 801 and a transceiver unit 802; the processing unit 801 is used to determine first information, where the first information indicates N configuration information, and the N configuration information is used to configure resources of N model parameters, where N is a positive integer; wherein one model parameter among the N model parameters is used to determine a neural network model corresponding to the local information of the first node; and the transceiver unit 802 is used to send the first information.

[0588] In a possible implementation, the local information includes at least one of a local resource overhead level, local AI capability information, and a local parameter scale.

[0589] In a possible implementation, the model parameter includes at least one of a model weight, a model activation function, and a model bias.

[0590] In a possible implementation, N is an integer greater than 1, and the parameter scales of the N model parameters are different.

[0591] In a possible implementation, the value of N is 1.

[0592] In one possible implementation, the transceiver unit 802 is also used to send second information, where the second information indicates M configuration information, and the M configuration information is used to configure transmission resources of M model parameters, where the parameter scales of the M model parameters are different, and M is an integer greater than 1; the transceiver unit 802 is also used to send part or all of the M model parameters based on the M configuration information.

[0593] In a possible implementation, any one of the M pieces of configuration information includes at least one of the following:

[0594] Indication information indicating parameter scales of model parameters configured by the one configuration information;

[0595] Indication information indicating the received power of the transmission resource configured by the one configuration information;

[0596] Indication information indicating a parameter type of a model parameter configured by the configuration information, where the parameter type includes a gradient and / or a weight;

[0597] Indication information indicating the frequency domain resources of the transmission resources configured by the one configuration information;

[0598] Indication information indicating the time domain resources of the transmission resources configured by the one configuration information.

[0599] In one possible implementation, the transceiver unit 802 is also used to send third information, where the third information indicates K configuration information, where the K configuration information is used to configure transmission resources for K model parameters, where the parameter scales of the K model parameters are different, and K is an integer greater than 1; the transceiver unit 802 is also used to receive part or all of the K model parameters based on the K configuration information.

[0600] In one possible implementation, any one of the K pieces of configuration information includes at least one of the following:

[0601] Indication information indicating parameter scales of model parameters configured by the one configuration information;

[0602] Indication information indicating the transmit power of the transmission resource configured by the one configuration information;

[0603] Indication information indicating a parameter type of a model parameter configured by the configuration information, where the parameter type includes a gradient and / or a weight;

[0604] Indication information indicating the frequency domain resources of the transmission resources configured by the one configuration information;

[0605] Indication information indicating the time domain resource of the transmission resource configured by the one configuration information;

[0606] Indication information indicating the index of the transmission resource of the one configuration information in the transmission resources configured by the K configuration information.

[0607] In one possible implementation, the transceiver unit 802 is also used to send fourth information, which is used to indicate the resources carrying the request information, and the request information is used to request the N configuration information; the transceiver unit 802 is also used to receive the request information on the resources carrying the request information.

[0608] In a possible implementation manner, the request information includes a first orthogonal sequence, and the first orthogonal sequence corresponds to the local information.

[0609] In one possible implementation, the transceiver unit 802 is further used to send fifth information, where the fifth information is used to indicate the local information corresponding to the N model parameters respectively, and / or the fifth information is used to indicate whether the first information includes the N configuration information.

[0610] In one possible implementation, when the device 800 is used to execute the method executed by the first node in any of the aforementioned embodiments, the device 800 includes a processing unit 801 and a transceiver unit 802; the transceiver unit 802 is used to receive second information, where the second information indicates M configuration information, and the M configuration information is used to configure transmission resources of M model parameters, where the parameter scales of the M model parameters are different, and M is an integer greater than 1; the processing unit 801 is used to determine the second configuration information among the M configuration information; the transceiver unit 802 is also used to receive a second model parameter based on the second configuration information, where the parameter scale of the second model parameter is the same as (or similar to) the local parameter scale.

[0611] In a possible implementation, the local information includes at least one of a local resource overhead level, local AI capability information, and a local parameter scale.

[0612] In a possible implementation, the model parameter includes at least one of a model weight, a model activation function, and a model bias.

[0613] In a possible implementation, any one of the M pieces of configuration information includes at least one of the following:

[0614] Indication information indicating parameter scales of model parameters configured by the one configuration information;

[0615] Indication information indicating the received power of the transmission resource configured by the one configuration information;

[0616] Indication information indicating a parameter type of a model parameter configured by the configuration information, where the parameter type includes a gradient and / or a weight;

[0617] Indication information indicating the frequency domain resources of the transmission resources configured by the one configuration information;

[0618] Indication information indicating the time domain resources of the transmission resources configured by the one configuration information.

[0619] In one possible implementation, the transceiver unit 802 is also used to receive third information, where the third information indicates K configuration information, and the K configuration information is used to configure transmission resources of K model parameters, where the parameter scales of the K model parameters are different, and K is an integer greater than 1; the transceiver unit 802 is also used to send a third model parameter based on the third configuration information in the K configuration information, where the parameter scale of the third model parameter is the same as (or similar to) the local parameter scale.

[0620] In one possible implementation, any one of the K pieces of configuration information includes at least one of the following:

[0621] Indication information indicating parameter scales of model parameters configured by the one configuration information;

[0622] Indication information indicating the transmit power of the transmission resource configured by the one configuration information;

[0623] Indication information indicating a parameter type of a model parameter configured by the configuration information, where the parameter type includes a gradient and / or a weight;

[0624] Indication information indicating the frequency domain resources of the transmission resources configured by the one configuration information;

[0625] Indication information indicating the time domain resource of the transmission resource configured by the one configuration information;

[0626] Indication information indicating the index of the transmission resource of the one configuration information in the transmission resources configured by the K configuration information.

[0627] In one possible implementation, when the device 800 is used to execute the method executed by the second node in any of the aforementioned embodiments, the device 800 includes a processing unit 801 and a transceiver unit 802; the processing unit 801 is used to determine the second information, where the second information indicates M configuration information, and the M configuration information is used to configure the transmission resources of M model parameters, where the parameter scales of the M model parameters are different, and M is an integer greater than 1; the transceiver unit 802 is used to send the second information.

[0628] In a possible implementation, the local information includes at least one of a local resource overhead level, local AI capability information, and a local parameter scale.

[0629] In a possible implementation, the model parameter includes at least one of a model weight, a model activation function, and a model bias.

[0630] In a possible implementation, any one of the M pieces of configuration information includes at least one of the following:

[0631] Indication information indicating parameter scales of model parameters configured by the one configuration information;

[0632] Indication information indicating the received power of the transmission resource configured by the one configuration information;

[0633] Indication information indicating a parameter type of a model parameter configured by the configuration information, where the parameter type includes a gradient and / or a weight;

[0634] Indication information indicating the frequency domain resources of the transmission resources configured by the one configuration information;

[0635] Indication information indicating the time domain resources of the transmission resources configured by the one configuration information.

[0636] In one possible implementation, the processing unit 801 is further configured to determine third information, where the third information indicates K pieces of configuration information, where the K pieces of configuration information are used to configure transmission resources of K model parameters, where the K model parameters have different parameter scales, and K is an integer greater than 1.

[0637] In one possible implementation, any one of the K pieces of configuration information includes at least one of the following:

[0638] Indication information indicating parameter scales of model parameters configured by the one configuration information;

[0639] Indication information indicating the transmit power of the transmission resource configured by the one configuration information;

[0640] Indication information indicating a parameter type of a model parameter configured by the configuration information, where the parameter type includes a gradient and / or a weight;

[0641] Indication information indicating the frequency domain resources of the transmission resources configured by the one configuration information;

[0642] Indication information indicating the time domain resource of the transmission resource configured by the one configuration information;

[0643] Indication information indicating the index of the transmission resource of the one configuration information in the transmission resources configured by the K configuration information.

[0644] In one possible implementation, when the device 800 is used to execute the method executed by the first node in any of the aforementioned embodiments, the device 800 includes a processing unit 801 and a transceiver unit 802; the transceiver unit 802 is used to receive third information, where the third information indicates K configuration information, and the K configuration information is used to configure transmission resources of K model parameters, where the parameter scales of the K model parameters are different, and K is an integer greater than 1; the processing unit 801 is used to determine the third configuration information among the K configuration information; the transceiver unit 802 is also used to send a third model parameter based on the third configuration information, where the parameter scale of the third model parameter is the same as (or similar to) the local parameter scale.

[0645] In a possible implementation, the local information includes at least one of a local resource overhead level, local AI capability information, and a local parameter scale.

[0646] In a possible implementation, the model parameter includes at least one of a model weight, a model activation function, and a model bias.

[0647] In one possible implementation, any one of the K pieces of configuration information includes at least one of the following:

[0648] Indication information indicating parameter scales of model parameters configured by the one configuration information;

[0649] Indication information indicating the transmit power of the transmission resource configured by the one configuration information;

[0650] Indication information indicating a parameter type of a model parameter configured by the configuration information, where the parameter type includes a gradient and / or a weight;

[0651] Indication information indicating the frequency domain resources of the transmission resources configured by the one configuration information;

[0652] Indication information indicating the time domain resource of the transmission resource configured by the one configuration information;

[0653] Indication information indicating the index of the transmission resource of the one configuration information in the transmission resources configured by the K configuration information.

[0654] In one possible implementation, when the device 800 is used to execute the method executed by the second node in any of the aforementioned embodiments, the device 800 includes a processing unit 801 and a transceiver unit 802; the processing unit 801 is used to determine third information, where the third information indicates K configuration information, and the K configuration information is used to configure transmission resources of K model parameters, where the parameter scales of the K model parameters are different, and K is an integer greater than 1; the transceiver unit 802 is used to send the third information.

[0655] In a possible implementation, the local information includes at least one of a local resource overhead level, local AI capability information, and a local parameter scale.

[0656] In a possible implementation, the model parameter includes at least one of a model weight, a model activation function, and a model bias.

[0657] In one possible implementation, any one of the K pieces of configuration information includes at least one of the following:

[0658] Indication information indicating parameter scales of model parameters configured by the one configuration information;

[0659] Indication information indicating the transmit power of the transmission resource configured by the one configuration information;

[0660] Indication information indicating a parameter type of a model parameter configured by the configuration information, where the parameter type includes a gradient and / or a weight;

[0661] Indication information indicating the frequency domain resources of the transmission resources configured by the one configuration information;

[0662] Indication information indicating the time domain resource of the transmission resource configured by the one configuration information;

[0663] Indication information indicating the index of the transmission resource of the one configuration information in the transmission resources configured by the K configuration information.

[0664] It should be noted that, for details on the information execution process of the units of the above-mentioned communication device 800, please refer to the description in the method embodiment shown above in this application, and no further details will be given here.

[0665] Please refer to Fig. 9, which is another schematic structural diagram of a communication device 900 provided in this application. The communication device 900 at least includes an input and output interface 902. The communication device 900 may be a chip or an integrated circuit.

[0666] Optionally, the communication device further includes a logic circuit 901 .

[0667] The transceiver unit 802 shown in FIG8 may be a communication interface, which may be the input / output interface 902 in FIG9 , which may include an input interface and an output interface. Alternatively, the communication interface may be a transceiver circuit, which may include an input interface circuit and an output interface circuit.

[0668] Optionally, the input-output interface 902 is used to receive first information, where the first information indicates N configuration information, and the N configuration information is used to configure resources of N model parameters, where N is a positive integer; the logic circuit 901 is used to determine the first configuration information among the N configuration information; the input-output interface 902 is also used to receive a first model parameter based on the first configuration information, and the first model parameter is used to determine a first neural network model corresponding to the local information.

[0669] Optionally, the logic circuit 901 is used to determine first information, which indicates N configuration information, and the N configuration information is used to configure resources of N model parameters, where N is a positive integer; wherein one model parameter among the N model parameters is used to determine a neural network model corresponding to the local information of the first node; and the input-output interface 902 is used to send the first information.

[0670] Optionally, the input and output interface 902 is used to receive second information, which indicates M configuration information, and the M configuration information is used to configure the transmission resources of M model parameters, the parameter scales of the M model parameters are different, and M is an integer greater than 1; the logic circuit 901 is used to determine the second configuration information among the M configuration information; the input and output interface 902 is also used to receive a second model parameter based on the second configuration information, wherein the parameter scale of the second model parameter is the same as (or similar to) the local parameter scale.

[0671] Optionally, the logic circuit 901 is used to determine the second information, which indicates M configuration information, and the M configuration information is used to configure the transmission resources of M model parameters, the parameter scales of the M model parameters are different, and M is an integer greater than 1; the input and output interface 902 is used to send the second information.

[0672] Optionally, the input and output interface 902 is used to receive third information, which indicates K configuration information, and the K configuration information is used to configure the transmission resources of K model parameters, the parameter scales of the K model parameters are different, and K is an integer greater than 1; the logic circuit 901 is used to determine the third configuration information among the K configuration information; the input and output interface 902 is also used to send a third model parameter based on the third configuration information, wherein the parameter scale of the third model parameter is the same as (or similar to) the local parameter scale.

[0673] Optionally, the logic circuit 901 is used to determine third information, which indicates K configuration information, and the K configuration information is used to configure the transmission resources of K model parameters, the parameter scales of the K model parameters are different, and K is an integer greater than 1; the input and output interface 902 is used to send the third information.

[0674] The logic circuit 901 and the input / output interface 902 may also execute other steps executed by the first node or the second node in any embodiment and achieve corresponding beneficial effects, which will not be described in detail here.

[0675] In a possible implementation, the processing unit 801 shown in FIG. 8 may be the logic circuit 901 in FIG. 9 .

[0676] Optionally, the logic circuit 901 may be a processing device, and the functions of the processing device may be partially or entirely implemented by software. The functions of the processing device may be partially or entirely implemented by software.

[0677] Optionally, the processing device may include a memory and a processor, wherein the memory is used to store a computer program, and the processor reads and executes the computer program stored in the memory to perform corresponding processing and / or steps in any one of the method embodiments.

[0678] Alternatively, the processing device may include only a processor. A memory for storing the computer program is located outside the processing device, and the processor is connected to the memory via circuits / wires to read and execute the computer program stored in the memory. The memory and processor may be integrated or physically separate.

[0679] Optionally, the processing device may be one or more chips, or one or more integrated circuits. For example, the processing device may be one or more field-programmable gate arrays (FPGAs), application-specific integrated circuits (ASICs), system-on-chips (SoCs), central processor units (CPUs), network processors (NPs), digital signal processors (DSPs), microcontroller units (MCUs), programmable logic devices (PLDs), or other integrated chips, or any combination of the above chips or processors.

[0680] Please refer to Figure 10, which shows the communication device 1000 involved in the above-mentioned embodiments provided in an embodiment of the present application. The communication device 1000 can specifically be a communication device serving as a terminal device in the above-mentioned embodiments. The example shown in Figure 10 is that the terminal device is implemented through the terminal device (or a component in the terminal device).

[0681] Herein, a possible logical structure diagram of the communication device 1000 is shown. The communication device 1000 may include but is not limited to at least one processor 1001 and a communication port 1002 .

[0682] The transceiver unit 802 shown in FIG8 may be a communication interface, which may be the communication port 1002 in FIG10 , which may include an input interface and an output interface. Alternatively, the communication port 1002 may be a transceiver circuit, which may include an input interface circuit and an output interface circuit.

[0683] Further optionally, the device may also include at least one of a memory 1003 and a bus 1004. In an embodiment of the present application, the at least one processor 1001 is used to control and process the actions of the communication device 1000.

[0684] In addition, the processor 1001 can be a central processing unit, a general-purpose processor, a digital signal processor, an application-specific integrated circuit, a field programmable gate array or other programmable logic device, a transistor logic device, a hardware component, or any combination thereof. It can implement or execute the various exemplary logic blocks, modules, and circuits described in conjunction with the disclosure of this application. The processor can also be a combination that implements computing functions, such as a combination of one or more microprocessors, a combination of a digital signal processor and a microprocessor, and so on. Those skilled in the art will clearly understand that for the convenience and brevity of description, the specific working processes of the systems, devices, and units described above can refer to the corresponding processes in the aforementioned method embodiments and will not be repeated here.

[0685] It should be noted that the communication device 1000 shown in Figure 10 can be specifically used to implement the steps implemented by the terminal device in the aforementioned method embodiment and achieve the corresponding technical effects of the terminal device. The specific implementation methods of the communication device shown in Figure 10 can refer to the description in the aforementioned method embodiment and will not be repeated here.

[0686] Please refer to Figure 11, which is a structural diagram of the communication device 1100 involved in the above-mentioned embodiments provided in an embodiment of the present application. The communication device 1100 can specifically be a communication device as a network device in the above-mentioned embodiments. The example shown in Figure 11 is that the network device is implemented through the network device (or a component in the network device), wherein the structure of the communication device can refer to the structure shown in Figure 11.

[0687] The communication device 1100 includes at least one processor 1111 and at least one network interface 1114. Further optionally, the communication device also includes at least one memory 1112, at least one transceiver 1113 and one or more antennas 1115. The processor 1111, the memory 1112, the transceiver 1113 and the network interface 1114 are connected, for example, via a bus. In an embodiment of the present application, the connection may include various interfaces, transmission lines or buses, etc., which are not limited in this embodiment. The antenna 1115 is connected to the transceiver 1113. The network interface 1114 is used to enable the communication device to communicate with other communication devices through a communication link. For example, the network interface 1114 may include a network interface between the communication device and the core network device, such as an S1 interface, and the network interface may include a network interface between the communication device and other communication devices (such as other network devices or core network devices), such as an X2 or Xn interface.

[0688] The transceiver unit 802 shown in FIG8 may be a communication interface, which may be the network interface 1114 in FIG11 , which may include an input interface and an output interface. Alternatively, the network interface 1114 may be a transceiver circuit, which may include an input interface circuit and an output interface circuit.

[0689] Processor 1111 is primarily used to process communication protocols and communication data, control the entire communication device, execute software programs, and process software program data, for example, to support the communication device in performing the actions described in the embodiments. The communication device may include a baseband processor and a central processing unit. The baseband processor is primarily used to process communication protocols and communication data, while the central processing unit is primarily used to control the entire terminal device, execute software programs, and process software program data. Processor 1111 in Figure 11 may integrate the functions of both a baseband processor and a central processing unit. Those skilled in the art will appreciate that the baseband processor and the central processing unit may also be independent processors interconnected via a bus or other technology. Those skilled in the art will appreciate that a terminal device may include multiple baseband processors to accommodate different network standards, multiple central processing units to enhance its processing capabilities, and various components of the terminal device may be connected via various buses. The baseband processor may also be referred to as a baseband processing circuit or a baseband processing chip. The central processing unit may also be referred to as a central processing circuit or a central processing chip. The functionality for processing communication protocols and communication data may be built into the processor or stored in memory as a software program, which is executed by the processor to implement the baseband processing functionality.

[0690] The memory is primarily used to store software programs and data. Memory 1112 can exist independently and be connected to processor 1111. Alternatively, memory 1112 can be integrated with processor 1111, for example, within a single chip. Memory 1112 can store program code for executing the technical solutions of the embodiments of the present application, and execution is controlled by processor 1111. The various computer program codes executed can also be considered drivers for processor 1111.

[0691] Figure 11 shows only one memory and one processor. In an actual terminal device, there may be multiple processors and multiple memories. The memory may also be referred to as a storage medium or a storage device. The memory may be a storage element on the same chip as the processor, i.e., an on-chip storage element, or an independent storage element, which is not limited in the present embodiment.

[0692] The transceiver 1113 can be used to support the reception or transmission of radio frequency signals between the communication device and the terminal. The transceiver 1113 can be connected to the antenna 1115. The transceiver 1113 includes a transmitter Tx and a receiver Rx. Specifically, one or more antennas 1115 can receive radio frequency signals. The receiver Rx of the transceiver 1113 is used to receive the radio frequency signal from the antenna, convert the radio frequency signal into a digital baseband signal or a digital intermediate frequency signal, and provide the digital baseband signal or digital intermediate frequency signal to the processor 1111 so that the processor 1111 can further process the digital baseband signal or digital intermediate frequency signal, such as demodulation and decoding. In addition, the transmitter Tx in the transceiver 1113 is also used to receive a modulated digital baseband signal or digital intermediate frequency signal from the processor 1111, convert the modulated digital baseband signal or digital intermediate frequency signal into a radio frequency signal, and transmit the radio frequency signal through one or more antennas 1115. Specifically, the receiver Rx can selectively perform one or more stages of down-mixing and analog-to-digital conversion on the RF signal to obtain a digital baseband signal or a digital intermediate frequency signal. The order of the down-mixing and analog-to-digital conversion processes is adjustable. The transmitter Tx can selectively perform one or more stages of up-mixing and digital-to-analog conversion on the modulated digital baseband signal or digital intermediate frequency signal to obtain a RF signal. The order of the up-mixing and digital-to-analog conversion processes is adjustable. The digital baseband signal and the digital intermediate frequency signal may be collectively referred to as digital signals.

[0693] The transceiver 1113 may also be referred to as a transceiver unit, a transceiver, a transceiver device, etc. Optionally, a device in the transceiver unit that implements a receiving function may be referred to as a receiving unit, and a device in the transceiver unit that implements a transmitting function may be referred to as a transmitting unit. That is, the transceiver unit includes a receiving unit and a transmitting unit. The receiving unit may also be referred to as a receiver, an input port, a receiving circuit, etc., and the transmitting unit may be referred to as a transmitter, a transmitter, or a transmitting circuit, etc.

[0694] It should be noted that the communication device 1100 shown in Figure 11 can be specifically used to implement the steps implemented by the network device in the aforementioned method embodiment, and to achieve the corresponding technical effects of the network device. The specific implementation methods of the communication device 1100 shown in Figure 11 can refer to the description in the aforementioned method embodiment, and will not be repeated here one by one.

[0695] Please refer to FIG12 , which is a schematic structural diagram of the communication device involved in the above-mentioned embodiment provided in an embodiment of the present application.

[0696] It can be understood that the communication device 120 includes, for example, modules, units, elements, circuits, or interfaces, which are appropriately configured together to implement the technical solutions provided in this application. The communication device 120 can be the RAN node, terminal, core network device, or other network device described above, or a component (e.g., a chip) in these devices, used to implement the method described in the following method embodiment. The communication device 120 includes one or more processors 121. The processor 121 can be a general-purpose processor or a dedicated processor. For example, it can be a baseband processor or a central processing unit. The baseband processor can be used to process communication protocols and communication data, and the central processing unit can be used to control the communication device (e.g., RAN node, terminal, or chip), execute software programs, and process data of software programs.

[0697] Optionally, in one design, the processor 121 may include a program 123 (sometimes also referred to as code or instructions), which may be executed on the processor 121 to cause the communication device 120 to perform the methods described in the following embodiments. In yet another possible design, the communication device 120 includes circuitry (not shown in FIG12 ).

[0698] Optionally, the communication device 120 may include one or more memories 122 on which a program 124 (sometimes also referred to as code or instructions) is stored. The program 124 can be run on the processor 121, so that the communication device 120 executes the method described in the above method embodiment.

[0699] Optionally, the processor 121 and / or the memory 122 may include an AI module 127, 128, which is used to implement AI-related functions. The AI ​​module may be implemented through software, hardware, or a combination of software and hardware. For example, the AI ​​module may include a wireless intelligent control (RIC) module. For example, the AI ​​module may be a near real-time RIC or a non-real-time RIC.

[0700] Optionally, data may be stored in the processor 121 and / or the memory 122. The processor and the memory may be provided separately or integrated together.

[0701] Optionally, the communication device 120 may further include a transceiver 125 and / or an antenna 126. The processor 121 may also be sometimes referred to as a processing unit, and controls the communication device (e.g., a RAN node or terminal). The transceiver 125 may also be sometimes referred to as a transceiver unit, a transceiver, a transceiver circuit, or a transceiver, and is configured to implement the transceiver functions of the communication device through the antenna 126.

[0702] The transceiver unit 802 shown in FIG8 may be a communication interface, which may be the transceiver 125 in FIG12 . The transceiver 125 may include an input interface and an output interface. Alternatively, the transceiver 125 may be a transceiver circuit, which may include an input interface circuit and an output interface circuit.

[0703] An embodiment of the present application also provides a computer-readable storage medium, which is used to store one or more computer-executable instructions. When the computer-executable instructions are executed by a processor, the processor executes the method described in the possible implementation of the first node or the second node in the above embodiment.

[0704] An embodiment of the present application also provides a computer program product (or computer program). When the computer program product is executed by the processor, the processor executes the method of the possible implementation of the above-mentioned first node or second node.

[0705] An embodiment of the present application also provides a chip system, which includes at least one processor for supporting a communication device to implement the functions involved in the possible implementation methods of the above-mentioned communication device. Optionally, the chip system also includes an interface circuit, which provides program instructions and / or data to the at least one processor. In one possible design, the chip system may also include a memory for storing program instructions and data necessary for the communication device. The chip system can be composed of a chip, or it can include a chip and other discrete devices, wherein the communication device can specifically be the first node or the second node in the aforementioned method embodiment.

[0706] An embodiment of the present application also provides a communication system, wherein the network system architecture includes a first node and a second node in any of the above embodiments, wherein the first node may be a terminal device or a network device, and the second node may also be a terminal device or a network device.

[0707] 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 merely schematic. For example, the division of the units is merely a logical function division. In actual implementation, there may be other division methods, such as multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be an indirect coupling or communication connection through some interfaces, devices or units, which can be electrical, mechanical or other forms.

[0708] The units described as separate components may or may not be physically separate, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed across multiple network units. Some or all of these units may be selected to achieve the purpose of this embodiment according to actual needs.

[0709] In addition, the functional units in the various embodiments of the present application can be integrated into a processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The above-mentioned integrated unit can be implemented in the form of hardware or in the form of a software functional unit. If the integrated unit is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present application is essentially or the contributing part or all or part of the technical solution can be embodied in the form of a software product, which is stored in a storage medium and includes several instructions for causing a computer device (which can be a personal computer, a server, or a network device, etc.) to execute all or part of the steps of the method described in the various embodiments of the present application. The aforementioned storage medium includes various media that can store program code, such as a USB flash drive, a mobile hard disk, a read-only memory (ROM), a random access memory (RAM), a magnetic disk or an optical disk.

Claims

1. A communication method, characterized in that: include: Receive first information, where the first information indicates N configuration information, where the N configuration information is used to configure resources of N model parameters, where N is a positive integer; A first model parameter is received based on a first configuration information among the N configuration information, where the first model parameter is used to determine a first neural network model corresponding to the local information.

2. The method according to claim 1, characterized in that The local information includes at least one of a local resource overhead level, local AI capability information, and a local parameter scale.

3. The method according to claim 1 or 2, characterized in that: The model parameters include at least one of a model weight, a model activation function, and a model bias.

4. The method according to any one of claims 1 to 3, characterized in that: N is an integer greater than 1, and the parameter scales of the N model parameters are different. The method further includes: The first configuration information is determined among the N configuration information based on the local information.

5. The method according to any one of claims 1 to 3, characterized in that: The value of N is 1, wherein the first model parameter is used to configure a neural network model shared by at least two first nodes; the method further includes: Based on the local information, the neural network model shared by the at least two first nodes is subjected to knowledge distillation processing to obtain the first neural network model.

6. The method according to any one of claims 1 to 5, characterized in that: The method further comprises: receiving second information, where the second information indicates M configuration information, where the M configuration information is used to configure transmission resources of M model parameters, where parameter scales of the M model parameters are different, and M is an integer greater than 1; A second model parameter is received based on second configuration information among the M configuration information, wherein a parameter scale of the second model parameter is the same as a local parameter scale.

7. The method according to claim 6, characterized in that Any one of the M pieces of configuration information includes at least one of the following: Indication information indicating parameter scale of a model parameter configured by the one configuration information; Indication information indicating the received power of the transmission resource configured by the one configuration information; Indication information indicating a parameter type of a model parameter configured by the one configuration information, wherein the parameter type includes a gradient and / or a weight; Indication information indicating frequency domain resources of transmission resources configured by the one configuration information; Indication information indicating the time domain resources of the transmission resources configured by the one configuration information.

8. The method according to any one of claims 1 to 7, characterized in that: The method further comprises: receiving third information, where the third information indicates K pieces of configuration information, where the K pieces of configuration information are used to configure transmission resources of K model parameters, where the parameter scales of the K model parameters are different, and K is an integer greater than 1; A third model parameter is sent based on third configuration information among the K configuration information, wherein a parameter scale of the third model parameter is the same as a local parameter scale.

9. The method according to claim 8, characterized in that Any one of the K pieces of configuration information includes at least one of the following: Indication information indicating parameter scale of a model parameter configured by the one configuration information; Indication information indicating the transmit power of the transmission resource configured by the one configuration information; Indication information indicating a parameter type of a model parameter configured by the one configuration information, wherein the parameter type includes a gradient and / or a weight; Indication information indicating frequency domain resources of transmission resources configured by the one configuration information; Indication information indicating the time domain resource of the transmission resource configured by the one configuration information; Indication information indicating the index of the transmission resource of the one configuration information in the transmission resources configured by the K configuration information.

10. The method according to any one of claims 1 to 9, characterized in that: Before receiving the first information, the method further includes: receiving fourth information, where the fourth information is used to indicate a resource that carries request information, where the request information is used to request the N pieces of configuration information; The request information is sent on the resource that carries the request information.

11. The method according to claim 10, characterized in that The request information includes a first orthogonal sequence, and the first orthogonal sequence corresponds to the local information.

12. The method according to any one of claims 1 to 11, characterized in that: Before receiving the first information, the method further includes: Receive fifth information, where the fifth information is used to indicate local information corresponding to the N model parameters respectively, and / or, the fifth information is used to indicate whether the first information includes the N configuration information.

13. A communication method, characterized in that: include: Determine first information, where the first information indicates N configuration information, where the N configuration information is used to configure resources of N model parameters, where N is a positive integer; wherein one model parameter among the N model parameters is used to determine a neural network model corresponding to the local information of the first node; The first information is sent.

14. The method according to claim 13, characterized in that The local information includes at least one of a local resource overhead level, local AI capability information, and a local parameter scale.

15. The method according to claim 13 or 14, characterized in that The model parameters include at least one of a model weight, a model activation function, and a model bias.

16. The method according to any one of claims 13 to 15, characterized in that N is an integer greater than 1, and the parameter scales of the N model parameters are different.

17. The method according to any one of claims 13 to 15, characterized in that The value of N is 1.

18. The method according to any one of claims 13 to 17, characterized in that The method further comprises: Sending second information, where the second information indicates M configuration information, where the M configuration information is used to configure transmission resources of M model parameters, where parameter scales of the M model parameters are different, and M is an integer greater than 1; Part or all of the M model parameters are sent based on the M configuration information.

19. The method according to claim 18, characterized in that Any one of the M pieces of configuration information includes at least one of the following: Indication information indicating parameter scale of a model parameter configured by the one configuration information; Indication information indicating the received power of the transmission resource configured by the one configuration information; Indication information indicating a parameter type of a model parameter configured by the one configuration information, wherein the parameter type includes a gradient and / or a weight; Indication information indicating frequency domain resources of transmission resources configured by the one configuration information; Indication information indicating the time domain resources of the transmission resources configured by the one configuration information.

20. The method according to any one of claims 13 to 19, characterized in that The method further comprises: Sending third information, where the third information indicates K pieces of configuration information, where the K pieces of configuration information are used to configure transmission resources of K model parameters, where the parameter scales of the K model parameters are different, and K is an integer greater than 1; Part or all of the K model parameters are received based on the K configuration information.

21. The method according to claim 20, characterized in that Any one of the K pieces of configuration information includes at least one of the following: Indication information indicating parameter scale of a model parameter configured by the one configuration information; Indication information indicating the transmit power of the transmission resource configured by the one configuration information; Indication information indicating a parameter type of a model parameter configured by the one configuration information, wherein the parameter type includes a gradient and / or a weight; Indication information indicating frequency domain resources of transmission resources configured by the one configuration information; Indication information indicating the time domain resource of the transmission resource configured by the one configuration information; Indication information indicating the index of the transmission resource of the one configuration information in the transmission resources configured by the K configuration information.

22. The method according to any one of claims 13 to 21, characterized in that Before sending the first information, the method further includes: Sending fourth information, where the fourth information is used to indicate a resource that carries request information, where the request information is used to request the N pieces of configuration information; The request information is received on the resource that carries the request information.

23. The method according to claim 22, characterized in that The request information includes a first orthogonal sequence, and the first orthogonal sequence corresponds to a local parameter scale.

24. The method according to any one of claims 13 to 23, characterized in that Before receiving the first information, the method further includes: Send fifth information, where the fifth information is used to indicate local information corresponding to the N model parameters respectively, and / or, the fifth information is used to indicate whether the first information includes the N configuration information.

25. A communication device, characterized in that: including a processing unit and a transceiver unit; The transceiver unit is used to receive first information, where the first information indicates N configuration information, and the N configuration information is used to configure resources of N model parameters, where N is a positive integer; The processing unit is used to determine the first configuration information among the N configuration information; The transceiver unit is also used to receive first model parameters based on the first configuration information, and the first model parameters are used to determine a first neural network model corresponding to the local information.

26. The device according to claim 25, characterized in that The local information includes at least one of a local resource overhead level, local AI capability information, and a local parameter scale.

27. The device according to claim 25 or 26, characterized in that The model parameters include at least one of a model weight, a model activation function, and a model bias.

28. The device according to any one of claims 25 to 27, characterized in that N is an integer greater than 1, the parameter scales of the N model parameters are different, and the processing unit is further used to determine the first configuration information among the N configuration information based on the local information.

29. The device according to any one of claims 25 to 27, characterized in that The value of N is 1, wherein the first model parameter is used to configure a neural network model shared by at least two first nodes; the processing unit is also used to perform knowledge distillation processing on the neural network model shared by the at least two first nodes based on the local information to obtain the first neural network model.

30. The device according to any one of claims 25 to 29, characterized in that The transceiver unit is further used to receive second information, where the second information indicates M configuration information, where the M configuration information is used to configure transmission resources of M model parameters, where the parameter scales of the M model parameters are different, and M is an integer greater than 1; The processing unit is further configured to determine second configuration information among the M configuration information; The transceiver unit is further configured to receive a second model parameter based on the second configuration information, wherein a parameter scale of the second model parameter is the same as a local parameter scale.

31. The device according to claim 30, characterized in that Any one of the M pieces of configuration information includes at least one of the following: Indication information indicating parameter scale of a model parameter configured by the one configuration information; Indication information indicating the received power of the transmission resource configured by the one configuration information; Indication information indicating a parameter type of a model parameter configured by the one configuration information, wherein the parameter type includes a gradient and / or a weight; Indication information indicating frequency domain resources of transmission resources configured by the one configuration information; Indication information indicating the time domain resources of the transmission resources configured by the one configuration information.

32. The device according to any one of claims 25 to 31, characterized in that The transceiver unit is further used to receive third information, where the third information indicates K configuration information, where the K configuration information is used to configure transmission resources of K model parameters, where the parameter scales of the K model parameters are different, and K is an integer greater than 1; The processing unit is further configured to determine third configuration information among the K configuration information; The transceiver unit is further configured to send a third model parameter based on the third configuration information, wherein a parameter scale of the third model parameter is the same as a local parameter scale.

33. The device according to claim 32, characterized in that Any one of the K pieces of configuration information includes at least one of the following: Indication information indicating parameter scale of a model parameter configured by the one configuration information; Indication information indicating the transmit power of the transmission resource configured by the one configuration information; Indication information indicating a parameter type of a model parameter configured by the one configuration information, wherein the parameter type includes a gradient and / or a weight; Indication information indicating frequency domain resources of transmission resources configured by the one configuration information; Indication information indicating the time domain resource of the transmission resource configured by the one configuration information; Indication information indicating the index of the transmission resource of the one configuration information in the transmission resources configured by the K configuration information.

34. The device according to any one of claims 25 to 33, characterized in that The transceiver unit is further used to receive fourth information, where the fourth information is used to indicate a resource that carries request information, where the request information is used to request the N pieces of configuration information; The transceiver unit is further configured to send the request information on the resource that carries the request information.

35. The device according to claim 34, characterized in that The request information includes a first orthogonal sequence, and the first orthogonal sequence corresponds to a local parameter scale.

36. The device according to any one of claims 25 to 35, characterized in that The transceiver unit is further used to receive fifth information, where the fifth information is used to indicate local information corresponding to the N model parameters respectively. And / or, the fifth information is used to indicate whether the first information includes the N configuration information.

37. A communication device, characterized in that: including a processing unit and a transceiver unit; The processing unit is used to determine first information, the first information indicates N configuration information, the N configuration information is used to configure resources of N model parameters, N is a positive integer; wherein one model parameter among the N model parameters is used to determine a neural network model corresponding to the local information of the first node; The transceiver unit is used to send the first information.

38. The device according to claim 37, characterized in that The local information includes at least one of a local resource overhead level, local AI capability information, and a local parameter scale.

39. The device according to claim 37 or 38, characterized in that The model parameters include at least one of a model weight, a model activation function, and a model bias.

40. The device according to any one of claims 37 to 39, characterized in that N is an integer greater than 1, and the parameter scales of the N model parameters are different.

41. The device according to any one of claims 37 to 39, characterized in that The value of N is 1.

42. The device according to any one of claims 37 to 41, characterized in that The transceiver unit is further used to send second information, where the second information indicates M configuration information, where the M configuration information is used to configure transmission resources of M model parameters, where the parameter scales of the M model parameters are different, and M is an integer greater than 1; The transceiver unit is further configured to send part or all of the M model parameters based on the M configuration information.

43. The device according to claim 42, characterized in that Any one of the M pieces of configuration information includes at least one of the following: Indication information indicating parameter scale of a model parameter configured by the one configuration information; Indication information indicating the received power of the transmission resource configured by the one configuration information; Indication information indicating a parameter type of a model parameter configured by the one configuration information, wherein the parameter type includes a gradient and / or a weight; Indication information indicating frequency domain resources of transmission resources configured by the one configuration information; Indication information indicating the time domain resources of the transmission resources configured by the one configuration information.

44. The device according to any one of claims 37 to 43, characterized in that The transceiver unit is further used to send third information, where the third information indicates K configuration information, where the K configuration information is used to configure transmission resources of K model parameters, where the parameter scales of the K model parameters are different, and K is an integer greater than 1; The transceiver unit is further configured to receive part or all of the K model parameters based on the K configuration information.

45. The device according to claim 44, characterized in that Any one of the K pieces of configuration information includes at least one of the following: Indication information indicating parameter scale of a model parameter configured by the one configuration information; Indication information indicating the transmit power of the transmission resource configured by the one configuration information; Indication information indicating a parameter type of a model parameter configured by the one configuration information, wherein the parameter type includes a gradient and / or a weight; Indication information indicating frequency domain resources of transmission resources configured by the one configuration information; Indication information indicating the time domain resource of the transmission resource configured by the one configuration information; Indication information indicating the index of the transmission resource of the one configuration information in the transmission resources configured by the K configuration information.

46. ​​The device according to any one of claims 37 to 45, characterized in that The transceiver unit is further used to send fourth information, where the fourth information is used to indicate a resource that carries request information, where the request information is used to request the N pieces of configuration information; The transceiver unit is further configured to receive the request information on the resource that carries the request information.

47. The device according to claim 46, characterized in that The request information includes a first orthogonal sequence, and the first orthogonal sequence corresponds to a local parameter scale.

48. The device according to any one of claims 37 to 47, characterized in that The transceiver unit is further used to send fifth information, where the fifth information is used to indicate local information corresponding to the N model parameters respectively, and / or the fifth information is used to indicate whether the first information includes the N configuration information.

49. A communication device, characterized in that: The device comprises at least one processor configured to execute computer programs or instructions in the memory to implement the method according to any one of claims 1 to 12.

50. A communication device, characterized in that: The device comprises at least one processor configured to execute computer programs or instructions in the memory to implement the method according to any one of claims 13 to 24.

51. A communication system, characterized in that: The system comprises a communication device as claimed in any one of claims 25 to 36, and a communication device for executing any one of claims 37 to 48; or The system comprises the communication device of claim 49 and the communication device of claim 50.

52. A computer-readable storage medium, characterized in that: The medium stores instructions, and when the instructions are executed by a computer, the method according to any one of claims 1 to 24 is implemented.

53. A computer program product, characterized in that The method comprises instructions which, when executed on a computer, cause the computer to perform the method according to any one of claims 1 to 24.

Images