Communication methods and devices

Abstract

The present disclosure provides a communication method and device for reducing training data transmission overhead and improving performance of an artificial intelligence model, the method including: acquiring first data and first information from a first node, the first information indicating a data augmentation method for the first data, and determining a first training data set for a model based on the first data and the first information.

Description

[Technical Field]

[0001] [Related Applications] This application claims priority to Chinese Patent Application No. 202111564623.8, filed with the State Intellectual Property Office of China on December 20, 2021, entitled "COMMUNICATION METHOD AND APPARATUS," which is incorporated herein by reference in its entirety.

[0002] [Technical field] The present disclosure relates to the field of communication technologies, and more particularly to communication methods and devices. [Background technology]

[0003] In wireless communication networks, e.g., mobile communication networks, the services supported by the network are becoming increasingly diverse, and therefore the requirements that must be met are becoming increasingly diverse. For example, networks are required to be able to support ultra-high speeds, ultra-low latency, and / or extremely large connections. This characteristic makes network planning, network configuration, and / or resource scheduling increasingly complex. Furthermore, as network capabilities become increasingly powerful, e.g., supporting increasingly higher spectrum and new technologies such as high-order multiple-input multiple-output (MIMO) technology, beamforming, and / or beam management, network energy conservation has become a hot research topic. These new requirements, scenarios, and characteristics pose unprecedented challenges to network planning, operation and maintenance, and efficient operation. To address this challenge, artificial intelligence (AI) technologies can be introduced into wireless communication networks to implement network intelligence. Based on this, how to effectively implement AI in networks is a worthy research issue. Summary of the Invention

[0004] The present disclosure provides a communication method and apparatus that reduces the overhead of transmitting training data and improves the performance of artificial intelligence models.

[0005] According to a first aspect, the present disclosure provides a communication method including: acquiring first data from a first node; and determining a first training dataset for a model based on the first data and a data augmentation method for the first data. Optionally, the data augmentation method for the first data may be indicated by the first node. For example, the communication method may alternatively be understood as: acquiring first data and first information from a first node, the first information indicating a data augmentation method for the first data; and determining a first training dataset for a model based on the first data and the first information alternatively understood as follows. Optionally, the data augmentation method for the first data may be agreed upon in a protocol.

[0006] The data augmentation method for the first data may be understood as one or more data augmentation methods adapted to the first data, and the data augmentation method may be described as a data augmentation method, a data processing method, a data pre-processing method, or another name.

[0007] In this design, a small amount of training data is transmitted, and the actual training data used for model training is generated by a data augmentation method, which can reduce the overhead of training data transmission and improve the efficiency and performance of model training.

[0008] In a possible design, the data augmentation method for the first data is determined based on the type of the first data and / or the model application scenario, which may be further described as a scenario corresponding to the first training data.

[0009] In a possible design, the first information includes one or more identifiers, the identifiers indicating a data extension method for the first data.

[0010] In a possible design, the first information includes parameters of a data augmentation method for the first data.

[0011] In one possible design, the communication method includes transmitting second information to the first node, the second information indicating at least one of the following: a type of the first data, a scenario corresponding to the first training data set, a data amount of the first data, or a data augmentation method supported by a second node, the second node being used to train the model; In this design, the second information indicates related requirements for obtaining the first data, and helps obtain appropriate training data, i.e., improves the effectiveness of the model training data.

[0012] In one possible design, the first data type may include channel data.

[0013] In a possible design, a scenario corresponding to the first training data set is used to determine the collection range of the first data. For example, the first data may be data of a scenario corresponding to the first data set. In this example, the scenario corresponding to the first training data set can also be described as a scenario corresponding to the first data.

[0014] In one possible design, the communication method further includes acquiring second data from the first node, the model being acquired by training based on the first training data set, and the second data being used to perform update training of the model. Optionally, a trigger condition of the method includes that the performance of the model does not satisfy a performance requirement, i.e., the second data is acquired from the first node when the performance of the model does not satisfy a performance requirement. The first node may determine that the performance of the model does not satisfy the performance requirement, or a node that trains the model, e.g., the second node, may determine that the performance of the model does not satisfy the performance requirement.

[0015] In one possible design, the step of obtaining second data from the first node comprises: sending third information to the first node, the third information indicating that performance of the model does not meet the performance requirements and / or the third information being used to request the second data; receiving the second data from the first node; It may further include:

[0016] In this design, the model training is first performed using the training data obtained by the data augmentation method, and then more training data is obtained for updating the model training based on the performance of the model, which can further guarantee the performance of the model while reducing the overhead of transmitting training data.

[0017] In a possible design, the communication method further comprises obtaining, from the first node, information indicative of the performance requirements.

[0018] In a possible design, the method further includes a step of transmitting fourth information to the first node, the fourth information being used to request the data extension method for the first data.

[0019] In a possible design, the fourth information includes an indication of a first data extension method; The first information includes acknowledgement information, and the acknowledgement information indicates that the data extension method of the first data includes the first data extension method; or The first information includes negative response information and indication information of a second data augmentation method, where the negative response information indicates that the data augmentation method for the first data does not include the first data augmentation method, and the data augmentation method for the first data includes the second data augmentation method. In this design, the first node determines whether the data augmentation method to be used is applicable to the first data, and then indicates the data augmentation method. This ensures that a data augmentation method adapted to the first data is actually used, which helps improve the performance of the model.

[0020] According to a second aspect, the present disclosure provides a communication method, comprising: determining first data and first information, the first information indicating a data extension method for the first data; transmitting the first data and the first information to a second node, wherein the first data and the first information are used to determine a first training data set for a model; The present invention provides a method comprising:

[0021] For details of the data extension method for the first data, please refer to the first aspect, and the details will not be described again here.

[0022] In one possible design, the method further includes acquiring second information, the second information indicating at least one of the following: a type of the first data, a scenario corresponding to the first training data set, a data amount of the first data, or a data augmentation method supported by the second node. The first data and the first information in the above design may be determined by the second information.

[0023] For a description of the first data type and the scenario corresponding to the first training data set, please refer to the first aspect, and the details will not be described again here.

[0024] In a possible design, the method further includes transmitting second data to the second node, the second data being used to perform model update training. Specifically, the second data can be transmitted to the second node if the model's performance does not meet the performance requirements. Optionally, the first node may determine that the model's performance does not meet the performance requirements, or a node training the model, e.g., the second node, may determine that the model's performance does not meet the performance requirements.

[0025] In a possible design, before the step of transmitting the second data to the second node, the method further comprises: receiving third information from the second node, the third information indicating that performance of the model does not meet the performance requirements and / or the third information being used to request the second data; Further includes:

[0026] In a possible design, information indicative of performance requirements is transmitted to the second node.

[0027] In a possible design, the method further includes receiving fourth information from the second node, the fourth information being used to request the data extension method for the first data.

[0028] For details of the fourth information and the first information, please refer to the first aspect, and the details will not be described again here.

[0029] According to a third aspect, the present disclosure provides a communications device. The communications device may be a second node, a device within the second node, or a device for use with the second node. In design, the communications device may include modules that correspond one-to-one to the methods / operations / steps / actions described in the first aspect. The modules may be implemented by hardware circuits, software, or a combination of software and hardware circuits. In a possible design, the communications device may include a processing module and a communications module.

[0030] In one example, the communication module is configured to obtain first data from the first node, and the processing module is configured to determine a first training data set for a model based on the first data and a data augmentation method for the first data.

[0031] Specifically, a data augmentation method for the first data may be indicated by the first node. In another example, the communication module is configured to obtain first data and first information from a first node, the first information indicating a data augmentation method for the first data, and the processing module is configured to determine a first training dataset for a model based on the first data and the first information.

[0032] Specifically, the data extension method of the first data may be agreed upon in a protocol.

[0033] In a possible design, the communication module is configured to send second information to the first node, the second information indicating at least one of the following: a type of the first data, a scenario corresponding to the first training data set, a data amount of the first data, or a data augmentation method supported by a second node, the second node being used to train the model; Further includes:

[0034] In a possible design, the communication module is configured to receive second data from the first node, the model is obtained by training based on the first training data set, and the second data is used to perform update training of the model. Optionally, if the performance of the model does not meet a performance requirement, the communication module is configured to receive the second data from the first node. The first node may determine that the performance of the model does not meet the performance requirement, or a node training the model, e.g., the second node, may determine that the performance of the model does not meet the performance requirement.

[0035] In a possible design, the communication module transmits third information to the first node, the third information indicating that performance of the model does not meet the performance requirements, and / or the third information is used to request the second data; receiving the second data from the first node; It is further configured as follows.

[0036] In a possible design, the communication module is configured to obtain information indicative of the performance requirements from the first node.

[0037] In a possible design, the communication module is configured to transmit fourth information to the first node, the fourth information being used to request the data extension method for the first data.

[0038] For details of the fourth information and the first information, please refer to the first aspect, and the details will not be described again here.

[0039] According to a fourth aspect, the present disclosure provides a communications device. The communications device may be a first node, a device in the first node, or a device used with the first node. In design, the communications device may include modules that correspond one-to-one to the methods / operations / steps / actions described in the second aspect. The modules may be implemented by hardware circuits, software, or a combination of software and hardware circuits. In a possible design, the communications device may include a processing module and a communications module. An example is as follows: the processing module is configured to determine first data and first information, the first information indicating a data augmentation method for the first data; The communication module is configured to transmit the first data and the first information to a second node, the first data and the first information being used to determine a first training data set for a model.

[0040] In a possible design, the processing module is further configured to obtain second information using the communication module, the second information indicating at least one of the following: a type of the first data, a scenario corresponding to the first training dataset, a data amount of the first data, or a data augmentation method supported by the second node, and the communication module is further configured to determine the first data and the first information based on the second information.

[0041] In a possible design, the communication module is further configured to transmit second data to the second node, where the second data is used to perform model update training. Specifically, the second data can be transmitted to the second node if the model's performance does not meet the performance requirements. Optionally, the first node may determine that the model's performance does not meet the performance requirements, or a node that trains the model, e.g., the second node, may determine that the model's performance does not meet the performance requirements.

[0042] In a possible design, before transmitting the second data to the second node, the communication module: and further configured to receive third information from the second node, the third information indicating that performance of the model does not meet the performance requirements and / or the third information being used to request the second data.

[0043] In a possible design, the communication module is further configured to transmit information indicative of the performance requirements to the second node.

[0044] In a possible design, the communication module is further configured to receive fourth information from the second node, the fourth information being used to request the data extension method for the first data.

[0045] For details of the fourth information and the first information, please refer to the first aspect, and the details will not be described again here.

[0046] According to a fifth aspect, the present disclosure provides a communications device. The communications device includes a processor configured to perform the method according to the first aspect. The communications device may further include a memory configured to store instructions and data. The memory is coupled to the processor. Execution of the instructions stored in the memory enables the processor to perform the method described in the first aspect. The communications device may further include a communications interface. The communications interface is used by the device to communicate with another device. For example, the communications interface may be a transceiver, a circuit, a bus, a module, a pin, or another type of communications interface. In a possible design, the communications device includes a memory configured to store program instructions and a processor configured to obtain first data from a first node via the communications interface, the processor further configured to determine a first training data set for a model based on the first data and a data augmentation method for the first data.

[0047] In another possible design, the communications device includes a memory configured to store program instructions; and a processor configured to obtain first data and first information from a first node via the communications interface, the first information indicating a data augmentation method for the first data, and the processor is further configured to determine a first training data set for a model based on the first data and the first information.

[0048] According to a sixth aspect, the present disclosure provides a communications device. The communications device includes a processor configured to perform the method according to the second aspect. The communications device may further include a memory configured to store instructions and data. The memory is coupled to the processor. Execution of the instructions stored in the memory enables the processor to perform the method described in the second aspect. The communications device may further include a communications interface. The communications interface is used by the device to communicate with another device. For example, the communications interface may be a transceiver, a circuit, a bus, a module, a pin, or another type of communications interface. In a possible design, the communications device includes a memory configured to store program instructions and a processor configured to determine first data and first information, the first information indicating a data augmentation method for the first data, the processor further configured to transmit the first data and the first information to a second node via the communications interface, the first data and the first information being used to determine a first training dataset for a model.

[0049] According to a seventh aspect, the present disclosure provides a communication system including a communication device according to the third or fourth aspect and a communication device according to the fourth or sixth aspect.

[0050] According to an eighth aspect, the present disclosure further provides a computer program which, when run on a computer, enables the computer to carry out the method provided in the first or second aspect.

[0051] According to a ninth aspect, the present disclosure further provides a computer program product comprising instructions which, when executed on a computer, enable the computer to perform the method provided in the first or second aspect.

[0052] According to a tenth aspect, the present disclosure further provides a computer-readable storage medium storing a computer program or instructions which, when executed by a computer, enable the computer to perform the method provided in the first or second aspect.

[0053] According to an eleventh aspect, the present disclosure further provides a chip, the chip being configured to read a computer program stored in the memory and to perform the method provided in the first or second aspect.

[0054] According to a twelfth aspect, the present disclosure further provides a chip system, the chip system including a processor configured to support a computer device in performing the method provided in the first or second aspect. In a possible design, the chip system further includes a memory configured to store programs and data required for the computer device. The chip system may include a chip, or may include a chip and other discrete components. [Brief explanation of the drawings]

[0055] [Figure 1A] 1 is a diagram of the structure of a communication system. [Figure 1B] 1 is a diagram of the structure of a communication system.

[0056] [Figure 2A] 1 is a diagram of a neuron structure.

[0057] [Figure 2B] FIG. 1 is a diagram showing the layer relationships of a neural network.

[0058] [Figure 3A] FIG. 1 is a diagram of an AI application framework according to the present disclosure.

[0059] [Figure 3B]1 is a diagram of several network architectures. [Figure 3C] 1 is a diagram of several network architectures. [Figure 3D] 1 is a diagram of several network architectures. [Figure 3E] 1 is a diagram of several network architectures.

[0060] [Figure 4] 1 is a schematic flow chart of a communication method according to the present disclosure.

[0061] [Figure 5A] FIG. 1 is a diagram of a channel flipping method. [Figure 5B] FIG. 1 is a diagram of a channel flipping method.

[0062] [Figure 6A] FIG. 1 is a diagram of a channel shifting method. [Figure 6B] FIG. 1 is a diagram of a channel shifting method.

[0063] [Figure 7] FIG. 10 is a diagram illustrating a noise addition processing method.

[0064] [Figure 8A] FIG. 1 is a diagram of an inter-channel permutation method. [Figure 8B] FIG. 1 is a diagram of an inter-channel permutation method.

[0065] [Figure 9] 1 is a schematic flow chart of a communication method according to the present disclosure. [Figure 10] 1 is a schematic flow chart of a communication method according to the present disclosure.

[0066] [Figure 11] FIG. 10 is a diagram of model training effectiveness according to the present disclosure.

[0067] [Figure 12]1 is a diagram of the structure of a communication device according to the present disclosure. [Figure 13] 1 is a diagram of the structure of a communication device according to the present disclosure. DETAILED DESCRIPTION OF THE INVENTION

[0068] To clarify the objectives, technical solutions, and advantages of the present disclosure, the following describes the present disclosure in detail with reference to the accompanying drawings.

[0069] The present disclosure relates to at least one (item) referring to one or more (items). A plurality of (items) means two (items) or more than two (items). The term "and / or" describes an association relationship describing associated objects and indicates that three relationships may exist. For example, A and / or B may represent the following three cases: only A is present, both A and B are present, and only B is present. The character " / " typically indicates an "or" relationship between associated objects. Furthermore, terms such as first and second may be used to describe objects in the present disclosure, but it should be understood that these objects are not limited by these terms. These terms are simply used to distinguish objects from one another.

[0070] The terms "including," "having," and other variations thereof referred to in the description of this disclosure are intended to cover a non-exclusive inclusion. For example, a process, method, system, product, or apparatus comprising a series of steps or units is not limited to the listed steps or units, and optionally further includes other steps or units not listed, or optionally further includes other specific steps or units of the process, method, product, or apparatus. It should be noted that in this disclosure, terms such as "example" or "for example" are used to denote example, illustration, or explanation. Any method or design approach described as "example" or "for example" in the examples of this disclosure should not be described as being preferred or having more advantages than another method or design approach. Rather, the words "example," "for example," etc. are intended to present the relevant concept in a particular way.

[0071] The techniques provided in this disclosure can be applied to various communication systems. For example, the communication system may be a third generation (3G) system. rd 3G (third generation) communication systems (e.g., universal mobile telecommunications system, UMTS), 4th generation (4G) th generation (4G) communication systems (e.g., long term evolution (LTE) systems), fifth generation (5 th The 5G communication system may be a 5G (first generation) communication system, a worldwide interoperability for microwave access (WiMAX) or a wireless local area network (WLAN) system, a system that integrates multiple systems, or a future communication system, such as a 6G communication system. A 5G communication system may also be called a new radio (NR) system.

[0072] A network element in a communication system can transmit signals to or receive signals from another network element. A signal may include information, signaling, data, etc. A network element may alternatively be referred to as an entity, a network entity, a device, a communication device, a communication module, a node, a communication node, etc. A network element is used as an example for purposes of illustration in this disclosure. For example, a communication system may include at least one terminal device and at least one access network device. The access network device may transmit downlink signals to the terminal device, and / or the terminal device may transmit uplink signals to the access network device. Furthermore, when a communication system includes multiple terminal devices, it may be understood that the multiple terminal devices may transmit signals to each other; in other words, both the signal-transmitting network element and the signal-receiving network element may be terminal devices.

[0073] 1A shows a communication system. For example, the communication system includes an access network device 110 and two terminal devices, namely, terminal device 120 and terminal device 130. At least one of terminal device 120 and terminal device 130 can transmit uplink data to access network device 110, and access network device 110 can receive the uplink data. The access network device can transmit downlink data to at least one of terminal device 120 and terminal device 130.

[0074] The terminal device and the access network device in FIG. 1A will be described in detail below.

[0075] (1) Access network equipment

[0076] The access network device may be a base station (BS). The access network device may also be called a network device, an access node (AN), or a radio access node (RAN). The access network device may be connected to a core network (LTE core network or 5G core network), and may provide wireless access services to terminal devices. The access network device includes, but is not limited to, at least one of the following: a next generation NodeB (gNB) in 5G, an open radio access network (O-RAN), or a module included in an access network device, an evolved NodeB (eNB), a radio network controller (RNC), a NodeB (NodeB, NB), a base station controller (BSC), a base transceiver station (BTS), a home base station (e.g., home evolved NodeB or home NodeB, HNB), a base band unit (BBU), a transmission reception point (TRP), a transmitting point (TP), and / or a mobile switching center. Alternatively, the access network device may be a radio unit (RU), a central unit (CU), a distributed unit (DU), a central unit control plane (CU-CP) node, or a central unit user plane (CU-UP) node.Alternatively, the access network device may be a relay station, an access point, an in-vehicle device, a wearable device, an access network device in a future evolved public land mobile network (PLMN), etc.

[0077] In the present disclosure, a communication device that realizes the functions of an access network device may be an access network device, a network device having some of the functions of an access network device, or a device that can support the realization of the functions of the access network device, such as a chip system, a hardware circuit, a software module, or a hardware circuit and a software module. The device may be mounted on the access network device or used in conjunction with the access network device. In the method of the present disclosure, an example will be described in which the communication device configured to realize the functions of the access network device is an access network device.

[0078] (2) Terminal device

[0079] A terminal device may also be referred to as a terminal, user equipment (UE), mobile station (MS), mobile terminal (MT), etc. A terminal device may be a device that provides voice and / or data connectivity to a user. A terminal device may communicate with one or more core networks through access network devices. A terminal device may include a handheld device with wireless connectivity, a separate processing device connected to a wireless modem, an in-vehicle device, etc. Alternatively, a terminal device may be a portable, pocket-sized, handheld, computer-integrated, or in-vehicle mobile device. Some examples of terminal devices are personal communication service (PCS) phones, cordless phones, session initiation protocol (SIP) phones, wireless local loop (WLL) stations, personal digital assistants (PDAs), wireless network cameras, mobile phones, tablet computers, notebook computers, palmtop computers, mobile internet devices (MIDs), wearable devices such as smart watches, virtual reality (VR) devices, augmented reality (AR) devices, wireless terminals in industrial control, terminals in vehicular internet systems, wireless terminals in self driving, wireless terminals in smart grids, wireless terminals in transportation safety, wireless terminals in smart cities such as smart fuels, terminals in high-speed railways, and wireless terminals in smart homes such as smart audio, smart coffee machines, smart printers, etc.

[0080] In the present disclosure, a communication device configured to realize the functions of a terminal device may be a terminal device, a terminal device having some functions of a terminal, or a device that can help realize the functions of a terminal device, such as a chip system. The device may be installed in the terminal device or used in conjunction with the terminal device. In the present disclosure, a chip system may include a chip, or may include a chip and other individual components. The technical solution provided in the present disclosure will be described using an example in which the communication device configured to realize the functions of a terminal device is a terminal device or UE.

[0081] (3) Protocol layer structure between access network equipment and terminal equipment

[0082] Communications between the access network device and the terminal device follow a predetermined protocol layer structure. The protocol layer structure may include a control plane protocol layer structure and a user plane protocol layer structure. For example, the control plane protocol layer structure may include protocol layer functions such as 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, and a physical layer. For example, the user plane protocol layer structure may include protocol layer functions such as a PDCP layer, an RLC layer, a MAC layer, and a physical layer. In a possible implementation, a service data adaptation protocol (SDAP) layer may be further included above the PDCP layer.

[0083] Optionally, the protocol layer structure between the access network device and the terminal device may further include an artificial intelligence (AI) layer used to transmit data related to AI functions.

[0084] Data transmission between an access network device and a terminal device is used as an example. Data transmission must pass through user plane protocol layers, such as the SDAP layer, PDCP layer, RLC layer, MAC layer, and physical layer. The SDAP layer, PDCP layer, RLC layer, MAC layer, and physical layer can be collectively referred to as the access stratum. Because data transmission direction includes transmission or reception, each layer is further divided into a transmitter and a receiver. Downlink data transmission is used as an example. After receiving data from a higher layer, the PDCP layer transmits the data to the RLC layer and MAC layer. The MAC layer generates a transport block, which then transmits it over the air via the physical layer. Data is encapsulated accordingly at each layer. For example, data received by a layer from a higher layer is considered to be encapsulated in a protocol data unit (PDU) as the service data unit (SDU) of that layer and then forwarded to the next layer.

[0085] For example, the terminal device may further include an application layer and a non-access stratum. The application layer may be used to provide services to an application installed in the terminal device. For example, downlink data received by the terminal device may be sequentially transmitted from the physical layer to the application layer and provided to the application by the application layer. As another example, the application layer may obtain data generated by an application, sequentially transmit the data to the physical layer, and transmit the data to another communication device. The non-access stratum may be configured to transfer user data. For example, the non-access stratum may transfer uplink data received from the application layer to the SDAP layer, or transfer downlink data received from the SDAP layer to the application layer.

[0086] (4) Access network device structure

[0087] An access network device may include a central unit (CU) and distributed units (DUs). Multiple DUs may be controlled by one CU in a centralized manner. For example, the interface between a CU and a DU may be called an F1 interface. The control plane (CP) interface is F1-C, and the user plane (UP) interface is F1-U. The CU and DU may be divided based on the protocol layer of a wireless network. For example, the functions of the PDCP layer and protocol layers above the PDCP layer may be located in the CU, and the functions of the protocol layers below the PDCP layer (e.g., the RLC layer and the MAC layer) may be located in the DU. As another example, the functions of the protocol layers above the PDCP layer may be located in the CU, and the functions of the PDCP layer and protocol layers below the PDCP layer may be located in the DU.

[0088] The division of processing functions between the CU and DU based on the protocol layers described above is merely an example, and other divisions are possible. For example, the CU or DU may be divided to accommodate more protocol layer functions. As another example, the CU or DU may be divided to accommodate some of the processing functions of the protocol layers. In one design, some of the RLC layer functions and functions of protocol layers above the RLC layer are configured in the CU, and the remaining RLC layer functions and functions of protocol layers below the RLC layer are configured in the DU. Other designs divide the CU or DU functions based on service type or other system requirements. For example, one design divides them based on latency. Functions whose processing time must meet latency requirements are assigned to the DU, and functions that do not need to meet latency requirements are assigned to the CU. Another design also has the CU perform one or more core network functions. For example, the CU may be located on the network side to facilitate centralized management. Another design involves remotely locating the RU of the DU. The RU has radio frequency functionality.

[0089] Optionally, the DU and RU may be distinguished by physical layer (PHY). For example, the DU may implement upper layer functions of the PHY layer, and the RU may implement lower layer functions of the PHY layer. When used for transmission, the PHY layer functions include cyclic redundancy check (CRC) code addition, channel coding, rate matching, scrambling, modulation, layer mapping, precoding, resource mapping, physical antenna mapping, and / or radio frequency transmission functions. When used for reception, the PHY layer functions include CRC, channel decoding, rate dematching, descrambling, demodulation, layer demapping, channel detection, resource demapping, physical antenna demapping, and / or radio frequency reception functions. The upper layer functions of the PHY layer may include some of the functions of the PHY layer, e.g., some of the functions are closer to the MAC layer. The lower layer functions of the PHY layer may include other parts of the functions of the PHY layer, e.g., some of the functions are closer to the radio frequency functions. For example, the upper layer functions of the PHY layer may include CRC code addition, channel coding, rate matching, scrambling, modulation, and layer mapping, and the lower layer functions of the PHY layer may include precoding, resource mapping, physical antenna mapping, and radio frequency transmission functions. Alternatively, the upper layer functions of the PHY layer may include CRC code addition, channel coding, rate matching, scrambling, modulation, layer mapping, and precoding. The lower layer functions of the PHY layer may include resource mapping, physical antenna mapping, and radio frequency transmission functions.

[0090] For example, the functions of the CU may be implemented by one entity or by different entities. For example, the functions of the CU may be further divided. That is, the control plane and the user plane may be separated and implemented by different entities, a control plane CU entity (i.e., a CU-CP entity) and a user plane CU entity (i.e., a CU-UP entity). The CU-CP entity and the CU-UP entity may be coupled to the DU to jointly complete the functions of the access network device.

[0091] In the above architecture, signaling generated by the CU may be transmitted to the terminal device through the DU, or signaling generated by the terminal device may be transmitted to the CU through the DU. For example, signaling in the RRC or PDCP layer is ultimately processed as signaling in the physical layer and transmitted to the terminal device, or converted from signaling received from the physical layer. Based on such an architecture, signaling in the RRC or PDCP layer may be considered to be transmitted using the DU, or transmitted using the DU and the RU.

[0092] Optionally, any of the DU, CU, CU-CP, CU-UP, and RU may be a software module, a hardware structure, or a combination of a software module and a hardware structure. This is not limited thereto. Different entities may exist in different forms, but this is not limited thereto. For example, the DU, CU, CU-CP, and CU-UP are software modules, and the RU is a hardware structure. These modules and the methods performed by these modules are also within the scope of protection of the present disclosure.

[0093] 1A is used merely as an example, and the present disclosure is not limited thereto. In actual applications, the communication system may further include more terminal devices and more access network devices, and may further include other network elements, such as core network devices and / or network elements configured to implement artificial intelligence functions.

[0094] The method provided in the present disclosure can be used for communication between an access network device and a terminal device, and can also be used for communication between other communication devices, for example, communication between a macro base station and a micro base station in a wireless backhaul link, or, as another example, communication between two terminal devices in a sidelink (SL), but this is not limited thereto.

[0095] The method provided in this disclosure relates to artificial intelligence (AI). AI can be implemented using various possible technologies, for example, machine learning technology. In this disclosure, the communication system may also include a network element that implements an artificial intelligence function. For example, an AI function (AI module or AI entity) can be configured in an existing network element in the communication system to implement AI-related operations. For example, in a 5G new radio (NR) system, the existing network elements may be access network devices (e.g., gNBs), terminal devices, core network devices, network management systems, etc. Based on the actual requirements of an operator's network operation, the network management system can classify network management tasks into three types: operation, administration, and maintenance. A network management system may also be referred to as an operation, administration, and maintenance (OAM) network element, or OAM for short. Operation mainly involves performing daily tasks such as analysis, forecasting, planning, and configuration for networks and services. Maintenance mainly involves daily operational activities such as testing or fault management performed on networks and network services. The network management system can detect the operating status of the network, optimize network connections and performance, improve network operation stability, and reduce network maintenance costs. Alternatively, an independent network element may be introduced into the communication system to perform AI-related operations. The independent network element may be referred to as an AI network element, AI node, etc. The name of the network element is not limited in this disclosure. The AI ​​network element may be directly connected to an access network device in the communication system or indirectly connected to the access network device via a third-party network element.The third-party network element may be a core network element, such as an authentication management function (AMF) network element or a user plane function (UPF) network element, an OAM, a cloud server, or other network elements. This is not limited thereto. See, for example, FIG. 1B. The communication system shown in FIG. 1A introduces an AI network element 140. The network element performing AI-related operations is a network element with built-in AI functions or the aforementioned AI network element. This is not limited in this embodiment disclosure. AI-related operations may also be referred to as AI functions. For a specific description of AI functions, see below.

[0096] In the following, for ease of understanding, AI terms in this disclosure will be explained with reference to A1 to A3, but these explanations do not limit the present disclosure.

[0097] A1: AI model

[0098] An AI model is a specific implementation of an AI function. An AI model represents a mapping relationship between the model's input and output. The AI ​​model may be a neural network, a linear regression model, a decision tree model, a support vector machine (SVM), a Bayesian network, a Q-learning model, or other machine learning model. In this disclosure, an AI function may include at least one of data collection (collection of training data and / or inference data), data preprocessing, model training (also referred to as model learning), model information release (model information construction), model validation, model inference, or inference result release. Inference may also be referred to as prediction. In this disclosure, an AI model may be referred to as a model for short.

[0099] Traditional communication systems require the design of communication modules with extensive expert knowledge, but deep learning communication systems based on machine learning techniques (such as neural networks) can automatically discover implicit pattern structures from large datasets, establish mapping relationships between data, and achieve better performance than traditional modeling methods.

[0100] A2: Neural networks

[0101] Neural networks are a specific implementation of AI and machine learning techniques. According to the universal approximation theorem, neural networks can theoretically approximate any continuous function, giving neural networks the ability to learn any mapping.

[0102] Neural networks are inspired by the neuron structure of the brain. For example, each neuron performs a weighted sum operation on its inputs and outputs the result using an activation function. Figure 2A shows a diagram of a neuron structure. Let the inputs of a neuron be:

number

number

number

number

number

number

[0103] Neural networks typically include multiple layers, each of which can contain one or more neurons. The depth and / or width of a neural network can be increased to improve the neural network's representational capabilities and provide more powerful information extraction and abstract modeling capabilities for complex systems. The depth of a neural network is the number of layers it contains, and the number of neurons in each layer can be referred to as its layer width. In one implementation, a neural network includes an input layer and an output layer. The input layer of a neural network performs neural processing on the received input information and forwards the processing results to the output layer, which obtains the neural network's output results. In another implementation, a neural network includes an input layer, a hidden layer, and an output layer. See FIG. 2B. The input layer of a neural network performs neural processing on the received input information and forwards the processing results to intermediate hidden layers. The hidden layers perform calculations on the received processing results and obtain calculation results. The hidden layers forward the calculation results to the output layer or adjacent hidden layers. Finally, the output layer obtains the neural network's output results. A neural network may include one hidden layer, or multiple hidden layers may be connected in series, but this is not limited thereto.

[0104] The neural network in the present disclosure is, for example, a deep neural network (DNN). Based on the network configuration method, the DNN may include a feedforward neural network (FNN), a convolutional neural network (CNN), and a recurrent neural network (RNN).

[0105] A3: Training data and inference data

[0106] Training data may include the input of an AI model, or may include the input and target output (label) of an AI model, and is used to train the AI ​​model. For example, training data includes multiple training samples, and each training sample is a single input to a neural network. Training data can be understood as a set of training samples and is sometimes called a training dataset. A training dataset is one of the important parts of machine learning. Model training essentially involves learning some features of the training data so that the output is as close as possible to the target output of the AI ​​model. For example, the difference between the output of the AI ​​model and the target output is as small as possible. The target output is sometimes called a label. The composition and selection of the training dataset determine to some extent the performance of the trained AI model.

[0107] Furthermore, in the training process of an AI model (such as a neural network), a loss function can be defined. The loss function describes the difference between the output value of the AI ​​model and a target output value. This disclosure does not limit the specific form of the loss function. The training process of an AI model is a process of adjusting the parameters of the AI ​​model so that the value of the loss function is less than a threshold or so that the value of the loss function meets a target requirement. For example, if the AI ​​model is a neural network, adjusting the parameters of the neural network includes adjusting at least one of the number of layers of the neural network, the width, the weights of the neurons, or the parameters of the activation functions of the neurons.

[0108] The inference data can be used as the input of the trained AI model and is used for AI model inference. In the model inference process, the inference data is input into the AI ​​model, and the corresponding output can be obtained as the inference result.

[0109] A4: AI model design

[0110] AI model design mainly includes a data collection phase (e.g., collecting training data and / or inference data), a model training phase, and a model inference phase. It may also include an inference result application phase. Figure 3A shows an AI application framework. In the data collection phase, a data source is used to provide training data and inference data. In the model training phase, the training data provided by the data source is analyzed or trained to obtain an AI model. The AI ​​model represents a mapping relationship between the model's input and output. Obtaining an AI model through learning by a model training host corresponds to obtaining a mapping relationship between the model's input and output through learning using the training data. In the model inference phase, the AI ​​model trained in the model training phase is used to perform inference based on the inference data provided by the data source, thereby obtaining an inference result. This phase can also be understood as follows: inference data is input to the AI ​​model, and output is obtained using the AI ​​model. The output is the inference result. The inference result may indicate configuration parameters used (acted upon) by an actor object and / or operations performed by the actor object. The inference result is released in the inference result application phase. For example, the inference results may be uniformly planned by an actor entity. For example, the actor entity may send the inference results to one or more actor objects (e.g., a core network device, an access network device, or a terminal device) for action. As another example, the actor entity may further feed back the performance of the model to a data source to facilitate subsequent update training of the model.

[0111] In a wireless communication system, the steps shown in FIG. 3A can be implemented by the same network element. That is, one network element may be responsible for data collection, model training, model inference, and release of inference results. Alternatively, different steps can be implemented by different network elements or nodes. For example, the node performing the data collection step may be different from the node performing the model training step. To make a model more accurate or have stronger generalization capabilities, a large amount of training data is required for model training. However, when data such as training data is transmitted between different nodes, air interface overhead is incurred. The larger the amount of training data transmitted, the higher the air interface overhead.

[0112] Based on this, the present disclosure provides a communication method for reducing the wireless interface overhead of data transmission and maximizing the efficiency and performance of model training. In the present disclosure, the data source of FIG. 3A can provide a relatively small amount of training data to the model training host. During the model training phase, the model training host can generate training data actually used for model training based on the small amount of training data provided by the data source and a data augmentation method for the small amount of training data, and perform model training. The small amount of training data refers to the training data used for model training, but does not mean that the amount of training data is small. In other words, it can be understood that the amount of training data is smaller than the amount of training data used for model training. Optionally, as shown in FIG. 3A, the data augmentation method for the small amount of training data can be provided by the data source, or the data augmentation method for the small amount of training data can be a predetermined method that can be used to perform augmentation processing on the small amount of training data.

[0113] The following describes a network architecture to which the communication method provided in the present disclosure can be applied, with reference to FIGS. 3B to 3E.

[0114] As shown in FIG. 3B , in a first possible embodiment, the access network device includes a near-real-time radio access network intelligent controller (RAN intelligent controller, RIC) module configured to perform model training and / or inference. For example, the near-real-time RIC may acquire network-side and / or terminal-side information from at least one of the CU, DU, and RU. The information may include a small amount of training data, instruction information for a data augmentation method for the small amount of training data, or inference data. For example, the near-real-time RIC may be configured to determine training data to be used for model training based on the small amount of training data and the data augmentation method for the small amount of training data, and perform model training. Furthermore, the near-real-time RIC may perform inference using the trained model. Furthermore, optionally, the near-real-time RIC may submit inference results to at least one of the CU, DU, and RU. Optionally, the CU and the DU may exchange inference results. Optionally, the DU and the RU may exchange inference results. For example, the near-real-time RIC submits the inference results to the DU, and the DU submits the inference results to the RU.

[0115] As shown in FIG. 3B , in a second possible implementation, the access network device may include a non-real-time RIC configured to perform model training and inference (optionally, the non-real-time RIC may be located in an OAM or core network device). For example, the non-real-time RIC may obtain network-side and / or terminal-side information from at least one of the CU, DU, and RU. The information may include a small amount of training data, instruction information for a data augmentation method for the small amount of training data, or inference data. For example, the non-real-time RIC is configured to determine training data to be used for model training based on the small amount of training data and the data augmentation method for the small amount of training data, and perform model training. Furthermore, the non-real-time RIC may perform inference using the trained model. Furthermore, optionally, the non-real-time RIC may submit inference results to at least one of the CU, DU, and RU. Optionally, the CU and the DU may exchange inference results. Optionally, the DU and the RU may exchange inference results. For example, the non-real-time RIC submits the inference results to the DU, and the DU submits the inference results to the RU.

[0116] As shown in FIG. 3B , in a third possible implementation, the access network device includes a near-real-time RIC, and the non-real-time RIC is located outside the access network device (optionally, the non-real-time RIC may be located in an OAM or core network device). As with the second possible implementation, the non-real-time RIC may be configured to perform model training and / or inference. And / or as with the first possible implementation, the near-real-time RIC may be configured to perform model training and / or inference. And / or the near-real-time RIC may obtain AI model information from the non-real-time RIC, obtain network-side and / or terminal-side information from at least one of the CU, DU, and RU, and obtain an inference result using the information and the AI ​​model information. Optionally, the near-real-time RIC may submit the inference result to at least one of the CU, DU, and RU. Optionally, the CU and DU may exchange the inference result. Optionally, the DU and RU may exchange the inference result. For example, the near-real-time RIC submits the inference result to the DU, and the DU submits the inference result to the RU. For example, the near-real-time RIC is configured to determine training data to be used for model training based on a small amount of training data and a data augmentation method for the small amount of training data, perform model training, and perform inference using the trained model. For example, the non-real-time RIC is configured to determine training data to be used for model training based on a small amount of training data and a data augmentation method for the small amount of training data, perform model training, and perform inference using the trained model. For example, the non-real-time RIC is configured to determine training data to be used for model training based on a small amount of training data and a data augmentation method for the small amount of training data, perform model training, and submit the trained model to the near-real-time RIC. The near-real-time RIC performs inference using the trained model.

[0117] 3C is an exemplary diagram of a network architecture to which the method provided in the present disclosure can be applied. Compared with FIG. 3B, in FIG. 3C, the CU is separated into a CU-CP and a CU-UP.

[0118] 3D is an example diagram of a network architecture to which the methods provided in the present disclosure can be applied. As shown in FIG. 3D, optionally, the access network device includes one or more AI entities, and the functions of the AI ​​entities are similar to the functions of the near-real-time RIC described above. Optionally, the OAM includes one or more AI entities, and the functions of the AI ​​entities are similar to the functions of the non-real-time RIC described above. Optionally, the core network device includes one or more AI entities, and the functions of the AI ​​entities are similar to the functions of the non-real-time RIC described above. If the OAM and the core network device each include an AI entity, the models obtained by training by the AI ​​entities of the OAM and the core network device are different and / or the models used for inference are different.

[0119] In this disclosure, the models differ with respect to at least one of the structural parameters of the model (eg, the weights and / or the number of layers of the model), the input parameters of the model, or the output parameters of the model.

[0120] FIG. 3E is an exemplary diagram of a network architecture to which the methods provided in the present disclosure can be applied. Compared with FIG. 3D, in FIG. 3E, the access network device is separated into a CU and a DU. Optionally, the CU includes an AI entity, whose function is similar to that of the near-real-time RIC described above. Optionally, the DU includes an AI entity, whose function is similar to that of the near-real-time RIC described above. If the CU and DU each include an AI entity, the models obtained by training by the AI ​​entities of the CU and DU are different and / or the models used for inference are different. Optionally, the CU in FIG. 3E may be further separated into a CU-CP and a CU-UP. Optionally, one or more AI models may be located in the CU-CP. And / or one or more AI models may be located in the CU-UP. Optionally, in FIG. 3D or FIG. 3E, the OAM of the access network device and the OAM of the core network device may be located separately.

[0121] In the present disclosure, one model can be used to derive one or more parameters through inference. The learning processes for different models can be located on different devices or nodes, or can be located on the same device or node. The inference processes for different models can be located on different devices or nodes, or can be located on the same device or node.

[0122] The following describes in detail the communication methods provided in the present disclosure with reference to Solutions 1 to 3. The steps or operations included in these methods are merely examples, and other operations or various modifications of operations may also be implemented in the present disclosure. Also, these steps may be implemented in a sequence different from the sequence presented in the present disclosure, and all operations in the present disclosure may not be implemented.

[0123] Solution 1:

[0124] 4 shows a communication method, in which the first node corresponds to the node that performs the data collection step, or the first node corresponds to a node that can transfer or manage data collected in the data collection step, and the second node corresponds to the node that performs the model training step. The method mainly includes the following steps:

[0125] S401: A second node acquires first data and first information from a first node, and the first information indicates a data extension method for the first data.

[0126] The first node may be a data collection node and have the function of collecting training data (which can also be described as collecting or measuring, for example). The second node may obtain related data from one or more data collection nodes. For example, the first node in S401 may be a node capable of providing first data and first information from multiple data collection nodes. Alternatively, the first node may be a data management node, which is connected to one or more data collection nodes, and the data collection node may have the function of collecting training data, and the data management node may manage the training data collected by the data collection nodes connected to the data management node. The second node may obtain related data from one or more data management nodes. For example, the first node in S401 may be a node capable of providing first data and first information from multiple data management nodes. Data collection nodes connected to different data management nodes may or may not overlap. This is not a limitation in the present disclosure. The second node is used to train a model, or is called a model training host. Examples are shown below, but are not limited thereto. The first node may be a terminal device, and the second node may be an access network device, a CU of an access network device in a separated architecture, or a DU of an access network device in a separated architecture. The first node may be an access network device, a CU of an access network device in a separated architecture, or a DU of an access network device in a separated architecture, and the second node is a terminal device. The first node may be a terminal device, an access network device, a CU of an access network device in a separated architecture, or a DU of an access network device in a separated architecture, and the second node is an AI network element or a core network element.The first node may be a core network device, and the second node may be an access network device, a CU of the access network device in a separated architecture, or a DU of the access network device in a separated architecture. Alternatively, the first node may be a third party node that can provide data (e.g., training data), and the second node may be a terminal device, an access network device, a CU or DU of the access network device in a separated architecture, an AI network element, or a core network element.

[0127] Optionally, the first node may transmit the first data and the first information to the second node using one signaling, or may transmit the first data and the first information separately using different signaling. When the first data and the first information are delivered using different signaling, an association relationship may be established between the first data and the first information. That is, the first information indicates a data extension method for the first data.

[0128] Specifically, the first data may be a small amount of training data stored in the first node or obtainable by the first node. The data augmentation method for the first data may be understood as a data augmentation method applicable when performing an augmentation process on the first data, or may be referred to as a data augmentation method adapted to the first data. Optionally, the first node may add an identifier of all or part of the data augmentation method adapted to the first data to the first information to indicate the data augmentation method for the first data to the second node. Both the first node and the second node may determine the corresponding data augmentation method based on the identifier of the data augmentation method. Alternatively, the first node may add other information to the first information to indicate the data augmentation method for the first data to the second node. This is not a limitation of the present disclosure.

[0129] Optionally, different data types are adapted to different data augmentation methods, or data in different application scenarios are adapted to different data augmentation methods. It can be understood that this method does not exclude the possibility that different data types are adapted to the same data augmentation method. For example, a data augmentation method adapted to a first data type is different from a data augmentation method adapted to a second data type, and a data augmentation method adapted to a second data type is the same as a data augmentation method adapted to a third data type. This method does not exclude data in different application scenarios being adapted to the same data augmentation method. For example, a data augmentation method adapted to data in a first application scenario is the same as a data augmentation method adapted to data in a second application scenario, and a data augmentation method adapted to data in a second application scenario is different from a data augmentation method adapted to data in a third application scenario. In this specification, an application scenario can be understood as an application scenario of a model or an application field of a model.

[0130] For example, for image data in the field of image processing, available data augmentation methods include at least one of geometric transformation (e.g., rotation, cropping, and / or scaling), color conversion, noise addition, etc. For example, the first data may include an image, and a data augmentation method for the image may include rotation, and a subsequent first training data set may include three new images obtained by performing 90-degree, 180-degree, and 270-degree rotations on the image. Optionally, the first training data set may further include the image in the first data.

[0131] For example, for language data in the field of natural language processing, available data augmentation methods include synonym replacement, back-translation, etc. For example, the first data includes Chinese segments, and the corresponding data augmentation method is back-translation. For example, the Chinese segments are translated into English, and then the English is translated back into Chinese to obtain new Chinese segments. In this example, the first training data set can include the new Chinese segments, and optionally, the first training data set can further include the Chinese segments in the first data.

[0132] For example, for channel data, available data augmentation methods include, but are not limited to, one or more of the following: channel flipping, channel intercept, channel scaling, channel shifting, noise addition, inter-channel substitution, virtual transmission, or data augmentation performed using a generative AI model. For ease of understanding, the data augmentation methods applicable to channel data will be described below with reference to the accompanying drawings.

[0133] Channel flipping may mean that the channel element with the lowest index and the channel element with the highest index are swapped in position, the channel element with the second lowest index and the channel element with the second highest index are swapped in position, and so on, in one or more dimensions, such as the time domain, the delay domain, the frequency domain, the Doppler domain, the antenna domain (or spatial domain), or the angle domain (or beam domain). For example, FIG. 5A shows a frequency-domain channel including 72 subcarriers. The frequency-domain channel, with subcarrier indexes in the order of 1 to 72, may be flipped in the frequency-domain dimension; in other words, a new frequency-domain channel, as shown in FIG. 5B, may be generated with subcarrier indexes in the order of 72 to 1. In this example, the first data set includes the frequency-domain channel shown in FIG. 5A, and the first training data set includes the new frequency-domain channel shown in FIG. 5B, and optionally further includes the frequency-domain channel shown in FIG. 5A.

[0134] Channel intercept may mean that multiple consecutive channel elements having consecutive indices are intercepted in one or more dimensions, such as the time domain, the delay domain, the frequency domain, the Doppler domain, the antenna domain (or spatial domain), or the angle domain (or beam domain). For example, if channel elements of 36 consecutive subcarriers starting from index 11 to index 46 are intercepted from the 72 subcarriers shown in FIG. 5A, a new frequency-domain channel including the channel elements of the 36 subcarriers may be formed. In this example, the first data includes the frequency-domain channel shown in FIG. 5A, and the first training data set includes the new frequency-domain channel including the channel elements of the 36 subcarriers.

[0135] Channel scaling may refer to extracting multiple channel elements at equal intervals based on the channel element index in one or more dimensions, such as the time domain, delay domain, frequency domain, Doppler domain, antenna domain (or spatial domain), or angle domain (or beam domain). For example, the above channel elements are specifically subcarriers. Also, among the 72 subcarriers shown in FIG. 5A , subcarriers with indices 10, 12, 14, ..., 70 are extracted every two indices from subcarriers 10 to 70 to form a new frequency-domain channel including 30 subcarriers. Alternatively, the same amount of channel elements are interpolated for every two channel elements with adjacent indices based on the channel element index in one or more dimensions, such as the time domain, delay domain, frequency domain, Doppler domain, antenna domain (or spatial domain), or angle domain (or beam domain). The interpolation method may be polynomial interpolation, linear interpolation, cubic interpolation, etc. For example, if new channel data is inserted into every two subcarriers with adjacent indices among the 72 subcarriers shown in Figure 5A, a new frequency domain channel including 143 subcarriers is formed. In this example, the first data includes the frequency domain channel shown in Figure 5A, and the first training data set includes the new frequency domain channel including 143 subcarriers.

[0136] Channel shifting may refer to a global shift of channel elements toward a lower or higher index, or a periodic shift, in one or more dimensions of the delay domain, Doppler domain, and / or angular domain (or beam domain). For example, FIG. 6A shows a delay-domain channel including 81 delay taps. Shifting the delay-domain channel by 10 delay taps toward a higher index results in a new delay-domain channel including 91 delay taps, as shown in FIG. 6B. In this example, the first data set includes the delay-domain channel shown in FIG. 6A, and the first training data set includes the new delay-domain channel shown in FIG. 6B, and optionally further includes the delay-domain channel shown in FIG. 6A.

[0137] Noise addition may mean that noise according to a specific distribution is superimposed in one or more dimensions, such as the time domain, the delay domain, the frequency domain, the Doppler domain, the antenna domain (or the spatial domain), and the angle domain (or the beam domain). For example, white Gaussian noise with a mean of 0 and a variance of 0.01 is superimposed on the delay-domain channel shown in Figure 6A to obtain a new delay-domain channel shown in Figure 7. In this example, the first data includes the delay-domain channel shown in Figure 6A, and the first training data set includes the new delay-domain channel shown in Figure 7, and optionally further includes the delay-domain channel shown in Figure 6A.

[0138] Inter-channel permutation may mean exchanging multiple channel elements having the same index in two channel data in one or more dimensions of the delay domain, Doppler domain, and angle domain (or beam domain). For example, for the delay-domain channel shown in FIG. 8A (a) and the delay-domain channel shown in FIG. 8A (b), the channel elements having the first 10 delay tap indices of the two channels are exchanged. After the exchange, FIG. 8A (a) becomes the new delay-domain channel shown in FIG. 8B (a), and FIG. 8A (b) becomes the new delay-domain channel shown in FIG. 8B (b). In this example, the first data includes the delay-domain channel shown in FIG. 8A, and the first training data set includes the new delay-domain channel shown in FIG. 8B, and optionally further includes the delay-domain channel shown in FIG. 8A.

[0139] The virtual transmission may refer to obtaining multiple virtual transmitted signals (data signals and / or reference signals) based on a preset method or a randomly generated method, and then obtaining virtual received signals corresponding to the multiple transmitted signals through calculation based on the multiple transmitted signals and channel data. Alternatively, the virtual transmitted signals (data signals and / or reference signals) may be obtained based on a preset method or a randomly generated method, and then precoding the multiple virtual transmitted signals using a precoding vector (also referred to as a precoding matrix or beam vector) to obtain multiple precoded transmitted signals. The virtual received signals corresponding to the multiple precoded transmitted signals are obtained through calculation based on the multiple precoded transmitted signals and channel data. Specifically, virtual received signals corresponding to all virtual transmitted signals may be generated based on one channel data, or multiple groups of {virtual transmitted signals, virtual received signals, channel data} may be obtained by generating virtual received signals corresponding to each virtual transmitted signal based on each channel data in the multiple channel data. For example, a virtual transmitted signal set S is preset, and S includes the virtual transmitted signals. According to a channel data set H (H includes b channel data), the virtual transmission model is expressed as y=h*s+n, where y is a received signal, h is any channel data in the channel data set H, s is any signal in the virtual transmitted signal set, and n is noise. In this case, a received signal set can be generated, and the received signal set includes a*b received signals. a and b are positive integers. In this example, the first data can include one or more of the above-mentioned channel data, and the first training data set includes multiple groups of {virtual transmitted signal, virtual received signal, channel data}. In another example, a virtual transmitted signal set S and a precoding matrix set W are preset, where S includes a virtual transmitted signal, and W includes q precoding matrices. a and q are positive integers. Optionally, a is equal to 1.According to a channel data set H (H includes b channel data), the virtual transmission model can be expressed as y=h*w*s+n, where y is a received signal, b is a positive integer, h is any channel data in the channel data set H, w is any precoding vector in W, s is any signal in the virtual transmitted signal set, and n is noise. In this case, a received signal set can be generated, and the received signal set includes a*b*q received signals. In this example, the first data can include one or more of the aforementioned channel data, and the first training data set includes multiple groups of {virtual transmitted signal, precoding matrix, virtual received signal, channel data}. In the present disclosure, the positive integer can be 1, 2, 3, 4, or a larger integer. This is not limited.

[0140] Data augmentation is performed by using a generative AI model, such as a generative adversarial network (GAN), to generate more training data that has the same distribution as the existing training data (i.e., the aforementioned first data). In this example, the first training data set includes the aforementioned more training data and, optionally, may further include the first data.

[0141] Optionally, the first information may further include parameters of a data augmentation method, and the second node may specifically perform a corresponding data augmentation process on the first data based on the parameters of the data augmentation method. The aforementioned data augmentation method corresponding to channel data is used as an example. The first information may further include one or more of the following: some or all of the information of the generative AI model; dimensions of data augmentation; at least two of the start index, length, and end index of channel elements in the channel intercept; channel scaling granularity, such as sampling interval, amount of channel elements for interpolation, or interpolation method; channel shift distance; noise distribution information; inter-channel substitution pattern; information about the virtual transmit signal and / or precoding vector used in the virtual transmission (e.g., the value and / or length of the precoding vector); channel format and / or format conversion method in the dataset; a collection range corresponding to the first data, such as a cell edge, a cell center, or a multipath delay range; etc.

[0142] S402: The second node determines a first training data set for the model based on the first data and the first information.

[0143] Specifically, the second node determines a first training data set for the model based on the first data and the data augmentation method of the first data, where the data augmentation method of the first data corresponds to the data augmentation method indicated by the first information of S401.

[0144] Optionally, the first information indicates a plurality of data augmentation methods adapted to the first data. The second node can perform a data augmentation process on the first data based on some or all of the data augmentation methods in the plurality of data augmentation methods to obtain a first training dataset for the model. For example, the second node can determine a data processing method actually used in the data augmentation method indicated by the first information based on the data augmentation methods supported by the second node, model training requirements, and / or model application scenarios. It can be understood that the actually used data processing method includes some or all of the data augmentation methods in the data augmentation method indicated by the first information, and then perform a data augmentation process on the first data based on the actually used data processing method. During the data augmentation process, a first training dataset that meets the model training requirements and / or model application scenarios can be obtained by considering the second node's ability to support the data augmentation methods. During the data augmentation process, the effectiveness of the data augmentation process can be ensured by considering the second node's ability to support the data augmentation methods.

[0145] Optionally, the first training data set of the model may include the first data and new data obtained by performing a data augmentation process on the first data, or the first training data set of the model may include only new data obtained by performing a data augmentation process on the first data, which is not a limitation of the present disclosure.

[0146] S403: The second node trains a model based on the first training dataset.

[0147] Specifically, when training the model, the second node may use supervised learning or unsupervised learning. This is not limited. The associated loss function is not limited and may be determined by factors such as the structural type of the model, the first training dataset of the model, and / or the model application scenario. Some examples of structural types of the model are CNN, RNN, and FNN.

[0148] Optionally, in FIG. 4, after S403, the second node may further acquire second data from the first node and perform model update training using the second data, or the second node may acquire further training data to perform model update training. The update training process may be applied when the performance of the model obtained by training based on the first training data set does not satisfy the performance requirements, but is not limited to this. The performance requirements may also be referred to as reference performance. The performance requirements may be predetermined performance indicators or may be indicated by the first node to the second node. For example, the first node may send information indicating the performance requirements to the second node. Specifically, the first node may determine the performance requirements based on the performance of another model trained in the past.

[0149] For example, the second node may determine whether the performance of the model obtained by training based on the first training data set meets the performance requirements. If the performance of the model does not meet the performance requirements, the second node may obtain second data from the first node. Alternatively, the second node may transmit model training results to the first node. The model training results indicate the performance of the model obtained by training based on the first training data set. The first node may determine whether the performance of the model meets the performance requirements, and if the performance of the model does not meet the performance requirements, may transmit second data to the second node.

[0150] For example, after S403, FIG. 4 further shows the following optional steps S404 to S406.

[0151] S404: The second node transmits the third information to the first node.

[0152] In any implementation, the third information may indicate a model training result. That is, the third information indicates the performance of the model obtained by training based on the first training dataset. Specifically, the third information may include a result value of a loss function after model training is completed. That is, the second node may use the third information to indicate the result of the loss function after model training is completed to reflect the performance of the model. Alternatively, the third information may specifically indicate the test performance of the model obtained by training on a predefined or preconfigured test set. For example, the third information may include a parameter value corresponding to the test performance. For example, the third information may include a performance level corresponding to the test performance, where the performance level is, for example, good / bad or acceptable / unacceptable. For example, the third information may include indicative information of the performance level corresponding to the test performance. For example, if the third information includes a first value, it indicates that the performance level corresponding to the test performance is good. Alternatively, if the third information includes a second value, it indicates that the performance level corresponding to the test performance is poor. Alternatively, the third information may specifically indicate whether the model obtained by training meets predefined or preconfigured performance requirements. For example, if the third information includes a third value, it indicates that the model obtained by training satisfies a predefined or preconfigured performance requirement. Alternatively, if the third information includes a fourth value, it indicates that the model obtained by training does not satisfy a predefined or preconfigured performance requirement. Alternatively, the third information may specifically indicate that the model obtained by training satisfies multiple predefined or preconfigured performance levels. For example, the model satisfies a first performance level among the multiple preconfigured performance levels, and the third information includes an identifier of the first performance level among the multiple preconfigured performance levels.

[0153] In any other implementation, the third information indicates that the model does not meet the performance requirements and / or is used to request the second data. Specifically, the third information includes a fourth value indicating that the model obtained by training does not meet predefined or preconfigured performance requirements, and / or the third information includes a first flag bit (or a first field), the value of which may be predefined, indicating that the second data is requested.

[0154] S405: The first node transmits the second data to the second node.

[0155] Specifically, the first node may determine, based on the third information, that the model does not meet the performance requirement. The model is obtained by training based on the first training dataset.

[0156] S406: The second node performs update training on the model based on the second data and the first training dataset.

[0157] In the present disclosure, the second node receives a small amount of training data and a small amount of data augmentation method from the first node and generates training data that is actually used for model training, thereby reducing the overhead of transmitting training data and improving the efficiency and performance of model training.

[0158] Solution 2:

[0159] 9 shows a communication method, in which the first node corresponds to the node that performs the data collection step, or the first node corresponds to a node that can transfer or manage data collected in the data collection step, and the second node corresponds to the node that performs the model training step. The method mainly includes the following steps:

[0160] S901: The second node transmits second information to the first node.

[0161] The definitions of the first node and the second node can be understood by referring to S401, and the details will not be described again here in this disclosure.

[0162] Specifically, the second information indicates at least one of the following: The first data type may be, for example, one or more of channel data, received signal data, received signal strength data, received signal quality data, received signal power data, location data, movement trajectory data, and interference signal data. The location data or movement trajectory data may be one or more of information permitted by a user, information obtained by desensitization, or information separated from user privacy information. The channel data type may be specifically classified into specific types of channel data, such as time domain channel data, frequency domain channel data, delay domain channel data, spatial domain channel data, and angle domain channel data. Alternatively, the first data type may be image data or language data as shown in S401.

[0163] The scenario corresponding to the first training dataset can also be understood as a model application scenario. The scenario corresponding to the first training dataset can be used to determine the collection range of the first data. Optionally, the first data may be data of the scenario corresponding to the first training dataset. However, the first data may not be data of the scenario corresponding to the first training dataset. However, the first data can be converted into data of the scenario corresponding to the first training dataset after performing a data augmentation process on the first data using a related data augmentation method (e.g., indicated by the first information). Examples of the scenario corresponding to the first training dataset include scenarios classified based on wireless signal reception strength, such as a cell edge or a cell center; scenarios classified based on wireless channel environments, such as a scenario with high scattering or a scenario with low scattering; scenarios classified based on measured multipath delay ranges, such as a scenario with long multipath delay or a scenario with short multipath delay; or actual locations such as a shopping mall or a high-speed train.

[0164] Regarding the amount of data of the first data, specifically, the second node may include instruction information in the second information, such as the size of the training dataset supported by the second node, the computing capability of the second node, or the memory capability of the second node, and instruct the first node to determine the amount of data of the first data based on the above instruction information.

[0165] The second node supports the following data expansion methods: Optionally, the second information may be understood as capability information of the second node and may instruct the first node to provide the first data and data augmentation method that is compatible with the capabilities of the second node.

[0166] S902: The first node transmits the first data and the first information to the second node.

[0167] Specifically, the first node may determine the first data to be transmitted based on the second information. For example, if the second information includes information indicating the amount of the first data, the first node may determine the amount (also called the size) of the first data to be transmitted based on the second information. And / or, if the second information includes information indicating a data extension method supported by the second node, the first node may determine the data extension method indicated by the first information based on the second information.

[0168] Also, the first data and the first information can be understood by referring to the description of S401, and the details will not be described again here in this disclosure.

[0169] S903: The second node determines a first training dataset for the model based on the first data and the first information.

[0170] For the implementation of this step, please refer to S402 above, and the details will not be described again here in this disclosure.

[0171] S904: The second node trains a model based on the first training dataset.

[0172] Optionally, as described in FIG. 4, in FIG. 9, after S904, the second node may further obtain second data from the first node and perform model update training using the second data, or it may be understood that the second node obtains further training data to perform model update training.

[0173] For example, after S904, FIG. 9 further shows the following optional steps S905-S907.

[0174] S905: The second node transmits the third information to the first node.

[0175] For the definition of the third information, please refer to the explanation of S404 for understanding, and the details will not be described again here in this disclosure.

[0176] S906: The first node transmits second data to the second node.

[0177] Specifically, the first node may determine, based on the third information, that the model does not meet the performance requirement. The model is obtained by training based on the first training dataset.

[0178] S907: The second node performs update training on the model based on the second data and the first training dataset.

[0179] In the present disclosure, the first node considers the capabilities of the second node and provides the second node with appropriate training data and appropriate data augmentation methods, allowing the second node to generate training data that is actually used for model training, thereby reducing the overhead of training data transmission and improving the efficiency and performance of model training.

[0180] Solution 3:

[0181] 10 shows a communication method, in which the first node corresponds to the node that performs the data collection step, or the first node corresponds to a node that can transfer or manage data collected in the data collection step, and the second node corresponds to the node that performs the model training step. The method mainly includes the following steps:

[0182] Optionally, in S1001, the second node transmits second information to the first node.

[0183] If this step is performed, please refer to S401 above for the implementation of this step, and the details will not be described again here in this disclosure.

[0184] Furthermore, the definitions of the first node and the second node can be understood with reference to S401, and the details will not be described again here in this disclosure.

[0185] S1002: The first node transmits the first data to the second node.

[0186] The determination of the first data may be understood by referring to the description of S902. Details will not be described again here in this disclosure. S1003: The second node sends fourth information to the first node, and the fourth information is used to request a data extension method for the first data.

[0187] Note that S1003 and S1002 may be performed in any order. Alternatively, S1003 may be performed before S1002. Alternatively, S1002 may be performed before S1003. For example, when S1002 is performed before S1003, the reason for performing S1003 is that the data volume of the first data does not meet the data volume requirement for model training, and the second node can request a data augmentation method for the first data from the first node by transmitting fourth information.

[0188] Optionally, the fourth information includes an indication of a first data augmentation method, which may be understood as a data augmentation method that the second node intends (plans or expects) to use for the first data.

[0189] S1004: The first node transmits first information to the second node, the first information indicating a data extension method for the first data.

[0190] When the fourth information includes indication information of the first data extension method, the first node can further determine whether the first data extension method is a data extension method adapted to the first data.

[0191] If the first data extension method is applicable to the first data, the first node may include acknowledgement information in the first information, where the acknowledgement information indicates that the data extension method of the first data includes the first data extension method. Alternatively, if the first data extension method is not applicable to the first data, the first node may include negative acknowledgement information and / or indication information of a second data extension method in the first information, where the data extension method of the first data does not include the first data extension method.

[0192] Specifically, the first information may include only negative acknowledgment information, and the negative acknowledgment information indicates that the data extension method of the first data does not include the first data extension method. In this case, the second node may re-determine the data extension method that the second node intends to use for the first data and, based on step S1003, initiate a request to the first node to transmit the fourth information until the first node indicates that the data extension method of the first data includes the data extension method that the second node intends to use. Alternatively, the first information may include only indication information of the second data extension method. In this case, the first node may implicitly indicate that the data extension method of the first data does not include the first data extension method and may understand that the data extension method of the first data includes the second data extension method. In this case, the second node may determine, based on the first information, that the data extension method of the first data includes the second data extension method. Alternatively, the first information may include the aforementioned negative acknowledgment information and indication information of the second data extension method. In this case, the first information including negative response information can be understood as the first node explicitly indicating that the data extension method of the first data does not include the first data extension method, and the first information including instruction information for the second data extension method can be understood as the first node indicating that the data extension method of the first data includes the second data extension method. In this case, the second node may determine, based on the first information, that the data extension method of the first data includes the second data extension method.

[0193] S1005: The second node determines a first training data set for the model based on the first data and the first information.

[0194] For the implementation of this step, please refer to S402 above, and the details will not be described again here in this disclosure.

[0195] S1006: The second node trains a model based on the first training dataset.

[0196] Optionally, as described in Fig. 4, in Fig. 10, after S1006, the second node may further obtain second data from the first node and perform model update training using the second data, or it may be understood that the second node obtains further training data to perform model update training. For example, after S1006, Fig. 10 further shows the following optional steps S1007 to S1009.

[0197] S1007: The second node transmits the third information to the first node.

[0198] For the definition of the third information, please refer to the explanation of S404 for understanding, and the details will not be described again here in this disclosure.

[0199] S1008: The first node transmits the second data to the second node.

[0200] Specifically, the first node may determine, based on the third information, that the model does not meet the performance requirement. The model is obtained by training based on the first training dataset.

[0201] S1009: The second node performs update training on the model based on the second data and the first training dataset.

[0202] In the present disclosure, the second node obtains a small amount of training data provided by the first node, and then obtains an adapted data augmentation method to generate training data that is actually used for model training, thereby reducing the overhead of training data transmission and improving the efficiency and performance of model training.

[0203] Furthermore, Figure 11 shows the model training effects of Solutions 1, 2, and 3. Solution 1 uses a large amount of training data for model training. Solution 2 uses a small amount of training data and a small amount of data augmentation. Solution 3 uses a small amount of training data for model training. For example, in a channel estimation scenario, the horizontal coordinate represents the signal-to-noise ratio (SNR), and the vertical coordinate represents model performance. The error between the estimated channel and the actual channel can be expressed, for example, as the normalized mean square error (NMSE). A smaller NMSE indicates that the estimated channel is closer to the actual channel and therefore better model performance. Solution 1 uses a model obtained by training on 50,000 different channel samples. Solution 2 uses 1,000 channel samples from the 5,000 channel samples used in Solution 1 to generate 50,000 training data using the noise-added data augmentation method, and then trains on the 50,000 training data. Solution 3 uses a model obtained by training on 1,000 different channel samples. From Figure 11, we can see that the performance of the model obtained by training with Solution 2 is close to that of the model obtained by training with Solution 1. In other words, by using an appropriate data augmentation method, it is possible to obtain similar performance to an AI model trained with a large amount of training data by training with only a small amount of training data, thereby significantly reducing the overhead of collecting, storing, and transmitting training data.

[0204] The above describes the methods provided in the present disclosure individually from the perspective of the interaction between a first node and a second node. To realize the functions in these methods, the first node and the second node include hardware structures and / or software modules, and the functions can be realized in the form of a hardware structure, a software module, or a combination of a hardware structure and a software module. Whether the functions among the above functions are implemented using a hardware structure, a software module, or a combination of a hardware structure and a software module depends on the specific application and design constraints of the technical solution.

[0205] Based on a similar concept, please refer to Fig. 12. The present application provides a communication device 1200. The communication device 1200 includes a processing module 1201 and a communication module 1202. The communication device 1200 may be a first node, or a communication device used in the first node or adapted for use in the first node, capable of implementing a communication method performed on the first node side. Alternatively, the communication device 1200 may be a second node, or a communication device used in the second node or adapted for use in the second node, capable of implementing a communication method performed on the second node side.

[0206] The communication module may also be referred to as a transceiver module, a transceiver, a transceiver machine, a transceiver device, etc. The processing module may also be referred to as a processor, a processing board, a processing unit, a processing device, etc. Optionally, the communication module is configured to perform transmitting and receiving operations at the first node side or the second node side in the above-mentioned method. A component in the communication module configured to implement a receiving function may be considered a receiving unit, and a component in the communication module configured to implement a transmitting function may be considered a transmitting unit. In other words, the communication module includes a receiving unit and a transmitting unit.

[0207] When the communication device 1200 is used in a first node, the processing module 1201 may be configured to perform the processing function of the first node in the embodiment shown in Figure 4, 9 or 10, and the communication module 1202 may be configured to perform the transmitting and receiving function of the first node in the embodiment shown in Figure 4, 9 or 10. The communication device can also be understood with reference to the third aspect in the overview and possible designs in the third aspect.

[0208] When the communication device 1200 is used in a second node, the processing module 1201 may be configured to perform the processing function of the second node in the embodiment shown in Figure 4, 9 or 10, and the communication module 1202 may be configured to perform the transmitting and receiving function of the second node in the embodiment shown in Figure 4, 9 or 10. The communication device can also be understood with reference to the fourth aspect in the overview and possible designs in the fourth aspect.

[0209] It should be noted that the communication module and / or the processing module may be implemented using virtual modules. For example, the processing module may be implemented using a software functional unit or a virtual device, and the communication module may be implemented using a software functional unit or a virtual device. Alternatively, the processing module or the communication module may be implemented using an entity device. For example, when implemented using a chip / chip circuit, the communication module may be an input / output circuit and / or a communication interface, and perform input operations (corresponding to the receiving operations described above) and output operations (corresponding to the transmitting operations described above). The processing module is an integrated processor, a microprocessor, or an integrated circuit.

[0210] The division into modules in the present disclosure is merely an example and is a division into logical functions, and other divisions may be used in actual implementation. Furthermore, the functional modules in the embodiments of the present disclosure may be integrated into one processor, or each module may exist physically alone, or two or more modules may be integrated into one module. The integrated modules may be implemented in the form of hardware or in the form of software functional modules.

[0211] Based on the same technical concept, the present disclosure further provides a communication device 1300. For example, the communication device 1300 may be a chip or a chip system. Optionally, in the present disclosure, a chip system may include a chip, or may include a chip and another individual component.

[0212] The communications device 1300 may be configured to implement the functionality of a network element in the communications system shown in FIG. 1A or 1B. The communications device 1300 may include at least one processor 1310. The processor 1310 is coupled to a memory. Optionally, the memory may be located within the device, the memory may be integrated with the processor, or the memory may be located external to the device. For example, the device 1300 may further include at least one memory 1320. The memory 1320 stores computer programs, computer programs or instructions, and / or data necessary to implement any of the aforementioned embodiments. The processor 1310 may execute the computer programs stored in the memory 1320 to achieve the method of any of the aforementioned embodiments.

[0213] The communication device 1300 may further include a communication interface 1330, through which the communication device 1300 may exchange information with other devices. For example, the communication interface 1330 may be a transceiver, a circuit, a bus, a module, a pin, or another type of communication interface. If the communication device 1300 is a chip-type device or circuit, the communication interface 1330 in the device 1300 may alternatively be an input / output circuit that inputs information (also referred to as received information) and outputs information (also referred to as transmitted information). The processor may be an integrated processor, a microprocessor, an integrated circuit, or a logic circuit. The processor may determine output information based on the input information.

[0214] A coupling in this disclosure may be an indirect coupling or communication connection between devices, units, or modules, and may be electrical, mechanical, or another form used for information interaction between the devices, units, and modules. The processor 1310 can operate in cooperation with the memory 1320 and the communication interface 1330. The specific connection medium between the processor 1310, the memory 1320, and the communication interface 1330 is not limited in this disclosure.

[0215] Optionally, refer to Figure 13. The processor 1310, memory 1320, and communication interface 1330 are connected to each other via a bus 1340. The bus 1340 may be a Peripheral Component Interconnect (PCI) bus, an Extended Industry Standard Architecture (EISA) bus, etc. The bus may be categorized as an address bus, a data bus, a control bus, etc. For ease of presentation, only one thick line is used to represent a bus in Figure 13, but this does not mean that there is only one line or only one type of bus.

[0216] In this disclosure, a processor may be a general-purpose processor, a digital signal processor, an application-specific integrated circuit, a field programmable gate array or other programmable logic device, a discrete gate or transistor logic device, or a discrete hardware component. The processor may implement or execute the methods, steps, and logic block diagrams disclosed in this disclosure. A general-purpose processor may be a microprocessor or any conventional processor, etc. The steps of the methods disclosed with reference to this disclosure may be performed directly by a hardware processor, or may be performed using a combination of hardware and software modules within the processor.

[0217] In the present disclosure, memory may be a non-volatile memory such as a hard disk drive (HDD) or a solid-state drive (SSD), or a volatile memory such as a random-access memory (RAM). Memory is, but is not limited to, any other medium capable of carrying or storing expected program code in the form of instructions or data structures and accessible by a computer. Memory in the present disclosure may alternatively be a circuit or any other device capable of implementing memory functions and configured to store program instructions and / or data.

[0218] In a possible implementation, the communication device 1300 may be used in a second node. Specifically, the communication device 1300 may be the second node, or a device supporting the second node and capable of implementing the functionality of the second node in any of the aforementioned embodiments. The memory 1320 stores computer programs (or instructions) and / or data for implementing the functionality of the second node in any of the aforementioned embodiments. The processor 1310 may execute computer programs stored in the memory 1320 to achieve the method performed by the second node in any of the aforementioned embodiments. When used in a second node, the communication interface of the communication device 1300 may be configured to interact with the first node, transmit information to the first node, or receive information from the first node.

[0219] In another possible implementation, the communication device 1300 may be used in a first node. Specifically, the communication device 1300 may be the first node, or a device supporting the first node and capable of implementing the functionality of the first node in any of the aforementioned embodiments. The memory 1320 stores computer programs (or instructions) and / or data for implementing the functionality of the first node in any of the aforementioned embodiments. The processor 1310 can execute the computer programs stored in the memory 1320 to achieve the method performed by the first node in any of the aforementioned embodiments. When used in a first node, the communication interface of the communication device 1300 may be configured to interact with a second node, transmit information to the second node, or receive information from the second node.

[0220] The communication device 1300 provided in this embodiment may be used in a second node to achieve a method performed by the second node, or may be used in a first node to achieve a method performed by the first node. Therefore, for the technical effects that can be achieved by this embodiment, please refer to the above-mentioned method examples. The details will not be described again here.

[0221] Based on the above-mentioned embodiments, the present disclosure provides a communication system including a second node and a first node, wherein the second node and the first node can implement the communication method provided in the embodiments shown in FIG.

[0222] All or part of the technical solutions provided in the present disclosure can be implemented by using software, hardware, firmware, or any combination thereof. When software is used to implement the technical solutions, all or part of the technical solutions can be implemented in the form of a computer program product. The computer program product includes one or more computer instructions. When loaded and executed on a computer, the computer program instructions generate, in whole or in part, the procedures or functions according to the present disclosure. The computer may be a general-purpose computer, a special-purpose computer, a computer network, a first node, a second node, or another programmable device. The computer instructions may be stored in a computer-readable storage medium or transmitted from a computer-readable storage medium to another computer-readable storage medium. For example, the computer instructions may be transmitted from a website, computer, server, or data center to another website, computer, server, or data center via wired (e.g., coaxial cable, optical fiber, or digital subscriber line (DSL)) or wireless (e.g., infrared, radio, or microwave) transmission. A computer-readable storage medium may be a data storage device, for example, any available medium that can be accessed by a computer, or a server or data center that integrates one or more available media. The available medium may be magnetic media (e.g., floppy disks, hard disks, or magnetic tapes), optical media (e.g., digital video discs (DVDs)), semiconductor media, etc.

[0223] In this disclosure, embodiments may be cross-referenced, provided that no logical contradiction exists. For example, methods and / or terms in method embodiments may be cross-referenced. For example, functions and / or terms in apparatus embodiments may be cross-referenced. For example, functions and / or terms in apparatus embodiments and method embodiments may be cross-referenced.

[0224] Obviously, those skilled in the art can make various modifications and variations to the present disclosure without departing from the scope of the present disclosure. The present disclosure intends to cover these modifications and variations of the present disclosure if they fall within the scope of the protection defined by the following claims of the present disclosure and their equivalent technologies.

Claims

1. A wireless communication method performed by a second node, the method comprising: acquiring first data and first information from a first node, the first information indicating a data extension method for the first data; determining a first training data set for a model based on the first data and the first information; sending second information to the first node, the second information indicating at least one of: a type of the first data, a scenario corresponding to the first training dataset, a data amount of the first data, or a data augmentation method supported by the second node, the second node being used to train the model; A method comprising:

2. acquiring second data from the first node if the performance of the model does not meet a performance requirement, the model being obtained by training based on the first training dataset, and the second data being used to perform update training of the model; The method of claim 1 further comprising:

3. The step of acquiring second data from the first node includes: sending third information to the first node, the third information indicating that performance of the model does not meet the performance requirements and / or the third information being used to request the second data; receiving the second data from the first node; The method of claim 2 , comprising:

4. obtaining information indicative of the performance requirements from the first node; The method of claim 2 further comprising:

5. transmitting fourth information to the first node, the fourth information being used to request the data extension method of the first data; The method of claim 1 further comprising:

6. the fourth information includes instruction information of a first data extension method, The first information includes acknowledgement information, and the acknowledgement information indicates that the data extension method of the first data includes the first data extension method; or the first information includes negative acknowledgement information and instruction information of a second data extension method, the negative acknowledgement information indicating that the data extension method of the first data does not include the first data extension method, and the data extension method of the first data includes the second data extension method; The method of claim 5.

7. A wireless communication method performed by a first node, comprising: determining first data and first information, the first information indicating a data extension method for the first data; transmitting the first data and the first information to a second node, wherein the first data and the first information are used to determine a first training data set for a model; acquiring second information, the second information indicating at least one of: a type of the first data, a scenario corresponding to the first training dataset, a data amount of the first data, or a data augmentation method supported by the second node; A method comprising:

8. transmitting second data to the second node, the second data being used to perform model update training; The method of claim 7 further comprising:

9. Prior to the step of transmitting second data to the second node, the method further comprises: receiving third information from the second node, the third information indicating that performance of the model does not meet a performance requirement and / or the third information being used to request the second data; The method of claim 8 further comprising:

10. transmitting information indicative of the performance requirements to the second node; 10. The method of claim 9 further comprising:

11. receiving fourth information from the second node, the fourth information being used to request the data extension method for the first data; The method of claim 7 further comprising:

12. the fourth information includes instruction information of a first data extension method, The first information includes acknowledgement information, and the acknowledgement information indicates that the data extension method of the first data includes the first data extension method; or the first information includes negative acknowledgement information and instruction information of a second data extension method, the negative acknowledgement information indicating that the data extension method of the first data does not include the first data extension method, and the data extension method of the first data includes the second data extension method; The method of claim 11.

13. A communication device configured to implement the method according to any one of claims 1 to 6.

14. A communications device configured to implement the method according to any one of claims 7 to 12.

15. A communications device comprising a processor, the processor coupled to a memory, the processor configured to perform the method of any one of claims 1 to 6.

16. A communications device comprising a processor, said processor coupled to a memory, said processor configured to perform the method of any one of claims 7 to 12.

17. A communication system comprising a communication device configured to perform a method according to any one of claims 1 to 6 and a communication device configured to perform a method according to any one of claims 7 to 12.

18. A computer-readable storage medium storing instructions that, when executed on a computer, enable the computer to perform the method of any one of claims 1 to 6 or any one of claims 7 to 12.

19. A computer program, which when run on a computer, enables the computer to carry out the method according to any one of claims 1 to 6 or any one of claims 7 to 12.

Images