Communication method and device, storage medium and program product

Abstract

The invention discloses a communication method and device, a storage medium and a program product. The method comprises: a first node sends first indication information and / or second indication information to a second node, the first indication information being used for indicating data set state information, and the second indication information being used for indicating model state information; the second node determines a communication strategy and a communication resource according to the first indication information and / or the second indication information; and the second node sends third indication information to the first node, the third indication information being used for indicating a communication policy and a communication resource, and the communication policy and the communication resource being determined according to the first indication information and / or the second indication information. According to the scheme of the invention, the data set state information and / or the model state information of the first node are / is acquired to determine the communication strategy and the communication resource, and the communication strategy and the communication resource of the first node are indicated, so that the communication reliability is improved.

Description

Communication method, device, storage medium and program product Technical Field

[0001] The present application relates to the field of communication technology, and in particular to a communication method, device, storage medium, and program product. Background Art

[0002] The information transmitted between distributed nodes and the central node is collectively referred to as AI traffic data. When distributed learning is performed within the L1 / L2 protocol layers, such as the physical (PHY) layer or the medium access control (MAC) layer of a wireless communication system, this AI traffic data is generated within the corresponding L1 / L2 protocol layers and is distinct from the service data streams delivered by the application layer.

[0003] For data flows at upper layers like the application and network layers, the new radio (NR) standard describes its transmission method as follows: Quality of service (QoS) flows are constructed and mapped step-by-step to the service data adaptation protocol (SDAP), packet data convergence protocol (PDCP), radio link control (RLC), MAC, and PHY layers in the protocol stack, ultimately transmitting over the RF air interface. However, this standard only provides a complete protocol stack processing process for upper-layer data and cannot be used for data flows generated at lower protocol layers like the PHY and MAC layers.

[0004] In view of this, how to transmit the artificial intelligence business data generated by the PHY / MAC layer is an urgent problem to be solved.

[0005] Summary of the Invention

[0006] The present application provides a communication method, device, storage medium, and program product to implement the transmission of artificial intelligence service data generated by the PHY / MAC layer.

[0007] In a first aspect, a communication method is provided, the method comprising: sending first indication information and / or second indication information, the first indication information being used to indicate data set status information, the second indication information being used to indicate model status information; and receiving third indication information, the third indication information being used to indicate a communication strategy and communication resources, the communication strategy and the communication resources being determined based on the first indication information and / or the second indication information. The method may be performed by a first device, which may be a communication device or a component in the communication device, such as a chip (system). The first device may be a first node in a network system.

[0008] In this aspect, the first node reports the data set status information and / or model status information, so that the second node determines the communication strategy and communication resources based on the data set status information and / or model status information reported by the first node, and instructs the first node on the communication strategy and communication resources, thereby realizing the transmission of artificial intelligence service data generated by the PHY / MAC layer.

[0009] In a possible implementation, the data set status information includes at least one of the following information: data type, sampling frequency, sampling location information, sample timestamp, sample quantity, and sample accuracy.

[0010] In this implementation, the data set status information refers to how the first node obtains the model input data. This data set status information affects the communication strategy and communication resource determination of the artificial intelligence data packet.

[0011] In another possible implementation, the model status information includes at least one of the following information: model type, model identification, sub-model order, model version number, model description, model update timestamp, model update location information, model accuracy, and model performance.

[0012] In this implementation, the model state information refers to a description of the model used by the first node. The model state information affects the communication strategy and communication resource determination of the artificial intelligence data packet.

[0013] In another possible implementation, the communication strategy is determined according to at least two of the first indication information, the second indication information, and channel state information.

[0014] In this implementation, the channel state between the first node and the second node also affects the communication strategy adopted by the first node. Therefore, the communication strategy is determined based on at least two of the first indication information, the second indication information and the channel state information, thereby improving the accuracy of the communication strategy.

[0015] In another possible implementation, the communication strategy includes at least one of the following strategies: a quantization strategy, a source coding strategy, a modulation coding strategy, a source-channel joint coding strategy, and a real or complex number processing strategy.

[0016] In another possible implementation, the method further includes: reporting the size of the artificial intelligence service data packet; wherein the communication resources are determined based on at least two of the size of the artificial intelligence service data packet, the data set status information, and the model status information.

[0017] In this implementation, communication resources are associated with at least two of the size of the artificial intelligence business data packet, the data set status information, and the model status information. Communication resources are determined based on at least two of the size of the artificial intelligence business data packet, the data set status information, and the model status information, thereby improving the accuracy of the determined communication resources.

[0018] In another possible implementation, the method further includes: generating an artificial intelligence service data packet according to the third indication information; and sending the artificial intelligence service data packet according to the third indication information.

[0019] In this implementation, the first node generates an artificial intelligence service data packet and can send the artificial intelligence service data packet to the second node or the third node.

[0020] In another possible implementation, the method further includes: receiving the artificial intelligence service data packet according to the third indication information.

[0021] In this implementation, after the second node determines the third indication information, it may also send the artificial intelligence service data packet to the first node based on the third indication information. Alternatively, the second node instructs the third node to send the artificial intelligence service data packet to the first node.

[0022] In yet another possible implementation, the artificial intelligence service data packet is communicated via a physical learning channel.

[0023] In this implementation, the physical learning channel is a new physical channel defined for artificial intelligence service data at the PHY / MAC layer.

[0024] In another possible implementation, communication through the physical learning channel includes at least one of the following processing operations: quantization, source coding; or communication through the physical learning channel includes at least one of the following processing operations: quantization, source-channel joint coding; or communication through the physical learning channel includes the following processing operations: real or complex number processing.

[0025] In this implementation, AI service data is generally in real or complex form and cannot be directly processed by bit-level channel coding. Therefore, the data processing flow of the physical learning channel requires the addition of operations such as quantization, source coding, joint source-channel coding, and real or complex number processing.

[0026] In another possible implementation, the artificial intelligence service data packet includes a header and a payload, and the payload includes at least one of the following data: parameters or gradients of a machine learning model, inference results, intermediate results of inference to a segmentation layer, gradients passed back to the segmentation layer, and logarithmic scores.

[0027] In another possible implementation, the packet header includes at least one of the following: content indication information, content description information, compression method, and packaging method.

[0028] In this implementation, by including at least one of the above information in the packet header, the receiving end can accurately depacketize the packet based on the information in the packet header.

[0029] In a second aspect, a communication method is provided, comprising: receiving first indication information and / or second indication information, wherein the first indication information is used to indicate data set status information, and the second indication information is used to indicate model status information; determining a communication strategy and communication resources based on the first indication information and / or the second indication information; and sending third indication information, wherein the third indication information is used to indicate the communication strategy and the communication resources. The method can be performed by a second device, which can be a communication device or a component in the communication device, such as a chip (system). The second device can be a second node in a network system.

[0030] In this aspect, the second node determines the communication strategy and communication resources by obtaining the data set status information and / or model status information reported by the first node, and instructs the first node on the communication strategy and communication resources, thereby realizing the transmission of artificial intelligence service data generated by the PHY / MAC layer.

[0031] In a possible implementation, the data set status information includes at least one of the following information: data type, sampling frequency, sampling location information, sample timestamp, sample quantity, and sample accuracy.

[0032] In another possible implementation, the model status information includes at least one of the following information: model type, model identification, sub-model order, model version number, model description, model update timestamp, model update location information, model accuracy, and model performance.

[0033] In another possible implementation, the communication strategy is determined according to at least two of the first indication information, the second indication information, and channel state information.

[0034] In another possible implementation, the communication strategy includes at least one of the following strategies: a quantization strategy, a source coding strategy, a modulation coding strategy, a source-channel joint coding strategy, and a real or complex number processing strategy.

[0035] In another possible implementation, determining the communication resources based on the first indication information and / or the second indication information includes: obtaining the size of the artificial intelligence service data packet; and determining the communication resources based on at least two of the size of the artificial intelligence service data packet, the first indication information, and the second indication information.

[0036] In another possible implementation, obtaining the size of the artificial intelligence business data packet includes: determining the size of the artificial intelligence business data packet based on the correspondence between at least one of the task type and the business type and the data packet size; or receiving the reported size of the artificial intelligence business data packet.

[0037] In another possible implementation, the method further includes: receiving an artificial intelligence service data packet sent according to the third indication information.

[0038] In another possible implementation, the method further includes: generating an artificial intelligence service data packet according to the third indication information; and sending the artificial intelligence service data packet according to the third indication information.

[0039] In another possible implementation, the method further includes: instructing a third node to send the artificial intelligence service data packet according to the third indication information.

[0040] In yet another possible implementation, the artificial intelligence service data packet is communicated via a physical learning channel.

[0041] In another possible implementation, communication through the physical learning channel includes at least one of the following processing operations: quantization, source coding; or communication through the physical learning channel includes at least one of the following processing operations: quantization, source-channel joint coding; or communication through the physical learning channel includes the following processing operations: real or complex number processing.

[0042] In another possible implementation, the artificial intelligence service data packet includes a header and a payload, and the payload includes at least one of the following data: parameters or gradients of a machine learning model, inference results, intermediate results of inference to a segmentation layer, gradients passed back to the segmentation layer, and logarithmic scores.

[0043] In another possible implementation, the packet header includes at least one of the following: content indication information, content description information, compression method, and packaging method.

[0044] In a third aspect, a communication method is provided, comprising: obtaining artificial intelligence service data generated by a physical (PHY) layer and / or a media access control (MAC) layer; and transmitting the artificial intelligence service data via a first radio bearer. The method may be performed by a first apparatus, which may be a communication device or a component within the communication device, such as a chip (system). The first apparatus may be a first node in a network system.

[0045] In this aspect, a new protocol stack processing flow is defined. For the artificial intelligence service data generated at the PHY / MAC layer, after it is input into the first wireless bearer, the artificial intelligence service data can be sent to the second node through the first wireless bearer, thereby realizing the communication of the artificial intelligence service data generated at the PHY / MAC layer.

[0046] In one possible implementation, sending the artificial intelligence service data through the first radio bearer includes: processing the artificial intelligence service data at at least one of the following protocol layers: packet data convergence protocol PDCP layer, radio link control RLC layer, MAC layer, PHY layer; and sending the processed artificial intelligence service data.

[0047] In another possible implementation, the first radio bearer is a learning radio bearer or a first data radio bearer.

[0048] In another possible implementation, the method is applied to a terminal, and the method further includes: receiving first signaling, where the first signaling is used to indicate configuration of the first radio bearer.

[0049] In another possible implementation, the method is applied to a network device, and the method further includes: sending a first signaling, where the first signaling is used to indicate configuration of the first radio bearer.

[0050] In this implementation, the first radio bearer is a newly defined learning radio bearer or a first data radio bearer, and the first radio bearer can be configured through the first signaling.

[0051] In a fourth aspect, a communication method is provided, comprising: receiving artificial intelligence service data via a first radio bearer; and outputting the artificial intelligence service data to a physical (PHY) layer and / or a media access control (MAC) layer. The method may be performed by a second apparatus, which may be a communication device or a component within the communication device, such as a chip (system). The second apparatus may be a second node in a network system.

[0052] In this aspect, a new protocol stack processing flow is defined. After receiving the artificial intelligence service data through the first wireless bearer, the artificial intelligence service data is output to the PHY / MAC layer, thereby realizing the communication of the artificial intelligence service data generated at the PHY / MAC layer.

[0053] In one possible implementation, outputting the artificial intelligence service data to the physical PHY layer and / or the media access control MAC layer includes: processing the artificial intelligence service data at at least one of the following protocol layers: PHY layer, MAC layer, radio link control RLC layer, packet data convergence protocol PDCP layer; and outputting the processed artificial intelligence service data to the PHY layer and / or MAC layer.

[0054] In another possible implementation, the first radio bearer is a learning radio bearer or a first data radio bearer.

[0055] In another possible implementation, the method is applied to a network device, and the method further includes: sending a first signaling, where the first signaling is used to indicate configuration of the first radio bearer.

[0056] In another possible implementation, the method is applied to a terminal, and the method further includes: receiving first signaling, where the first signaling is used to indicate configuration of the first radio bearer.

[0057] In this implementation, the first radio bearer is a newly defined learning radio bearer or a first data radio bearer, and the first radio bearer can be configured through the first signaling.

[0058] In a fifth aspect, a communication device is provided that can implement the communication method described in the first aspect. For example, the communication device can be a chip or a first node. The method can be implemented through software, hardware, or hardware executing corresponding software.

[0059] The communication device includes a transceiver unit and a processing unit, wherein the transceiver unit is used to send first indication information and / or second indication information, the first indication information is used to indicate data set status information, and the second indication information is used to indicate model status information; and the transceiver unit is also used to receive third indication information, the third indication information is used to indicate communication strategy and communication resources, and the communication strategy and the communication resources are determined based on the first indication information and / or the second indication information.

[0060] In a possible implementation, the data set status information includes at least one of the following information: data type, sampling frequency, sampling location information, sample timestamp, sample quantity, and sample accuracy.

[0061] In another possible implementation, the model status information includes at least one of the following information: model type, model identification, sub-model order, model version number, model description, model update timestamp, model update location information, model accuracy, and model performance.

[0062] In another possible implementation, the communication strategy is determined according to at least two of the first indication information, the second indication information, and channel state information.

[0063] In another possible implementation, the communication strategy includes at least one of the following strategies: a quantization strategy, a source coding strategy, a modulation coding strategy, a source-channel joint coding strategy, and a real or complex number processing strategy.

[0064] In another possible implementation, the transceiver unit is further used to report the size of the artificial intelligence service data packet; wherein the communication resources are determined based on at least two of the size of the artificial intelligence service data packet, the data set status information, and the model status information.

[0065] In another possible implementation, the processing unit is used to generate an artificial intelligence service data packet according to the third indication information; and the transceiver unit is further used to send the artificial intelligence service data packet according to the third indication information.

[0066] In another possible implementation, the transceiver unit is further configured to receive the artificial intelligence service data packet according to the third indication information.

[0067] In yet another possible implementation, the artificial intelligence service data packet is communicated via a physical learning channel.

[0068] In another possible implementation, communication through the physical learning channel includes at least one of the following processing operations: quantization, source coding; or communication through the physical learning channel includes at least one of the following processing operations: quantization, source-channel joint coding; or communication through the physical learning channel includes the following processing operations: real or complex number processing.

[0069] In another possible implementation, the artificial intelligence service data packet includes a header and a payload, and the payload includes at least one of the following data: parameters or gradients of a machine learning model, inference results, intermediate results of inference to a segmentation layer, gradients passed back to the segmentation layer, and logarithmic scores.

[0070] In another possible implementation, the packet header includes at least one of the following: content indication information, content description information, compression method, and packaging method.

[0071] In a sixth aspect, a communication device is provided that can implement the communication method described in the second aspect. For example, the communication device can be a chip or a second node. The method can be implemented through software, hardware, or hardware executing corresponding software.

[0072] The communication device includes a transceiver unit and a processing unit, wherein the transceiver unit is used to receive first indication information and / or second indication information, the first indication information is used to indicate data set status information, and the second indication information is used to indicate model status information; the processing unit is used to determine the communication strategy and communication resources based on the first indication information and / or the second indication information; and the transceiver unit is also used to send third indication information, and the third indication information is used to indicate the communication strategy and the communication resources.

[0073] In a possible implementation, the data set status information includes at least one of the following information: data type, sampling frequency, sampling location information, sample timestamp, sample quantity, and sample accuracy.

[0074] In another possible implementation, the model status information includes at least one of the following information: model type, model identification, sub-model order, model version number, model description, model update timestamp, model update location information, model accuracy, and model performance.

[0075] In another possible implementation, the communication strategy is determined according to at least two of the first indication information, the second indication information, and channel state information.

[0076] In another possible implementation, the communication strategy includes at least one of the following strategies: a quantization strategy, a source coding strategy, a modulation coding strategy, a source-channel joint coding strategy, and a real or complex number processing strategy.

[0077] In another possible implementation, the processing unit is further used to obtain the size of the artificial intelligence service data packet; and the processing unit is further used to determine the communication resources based on at least two of the size of the artificial intelligence service data packet, the first indication information, and the second indication information.

[0078] In another possible implementation, the processing unit is further used to determine the size of the artificial intelligence business data packet based on the correspondence between at least one of the task type and the business type and the data packet size; or the transceiver unit is further used to receive the reported size of the artificial intelligence business data packet.

[0079] In another possible implementation, the transceiver unit is further configured to receive an artificial intelligence service data packet sent according to the third indication information.

[0080] In another possible implementation, the processing unit is further used to generate an artificial intelligence service data packet according to the third indication information; and the transceiver unit is further used to send the artificial intelligence service data packet according to the third indication information.

[0081] In another possible implementation, the processing unit is further configured to instruct the third node to send the artificial intelligence service data packet according to the third indication information.

[0082] In yet another possible implementation, the artificial intelligence service data packet is communicated via a physical learning channel.

[0083] In another possible implementation, communication through the physical learning channel includes at least one of the following processing operations: quantization, source coding; or communication through the physical learning channel includes at least one of the following processing operations: quantization, source-channel joint coding; or communication through the physical learning channel includes the following processing operations: real or complex number processing.

[0084] In another possible implementation, the artificial intelligence service data packet includes a header and a payload, and the payload includes at least one of the following data: parameters or gradients of a machine learning model, inference results, intermediate results of inference to a segmentation layer, gradients passed back to the segmentation layer, and logarithmic scores.

[0085] In another possible implementation, the packet header includes at least one of the following: content indication information, content description information, compression method, and packaging method.

[0086] In a seventh aspect, a communication device is provided that can implement the communication method described in the third aspect. For example, the communication device can be a chip or a first node. The method can be implemented through software, hardware, or hardware executing corresponding software.

[0087] The communication device includes a transceiver unit and a processing unit, wherein the processing unit is used to obtain artificial intelligence service data generated by the physical PHY layer and / or the media access control MAC layer; and the transceiver unit is used to send the artificial intelligence service data through the first wireless bearer.

[0088] In one possible implementation, the processing unit is used to process the artificial intelligence service data at at least one of the following protocol layers: packet data convergence protocol (PDCP) layer, radio link control (RLC) layer, MAC layer, and PHY layer; and the transceiver unit is used to send the processed artificial intelligence service data.

[0089] In another possible implementation, the first radio bearer is a learning radio bearer or a first data radio bearer.

[0090] In another possible implementation, the communication device is a terminal, and the transceiver unit is further configured to receive a first signaling, where the first signaling is used to instruct configuration of the first radio bearer.

[0091] In another possible implementation, the communication device is a network device, and the transceiver unit is further configured to send a first signaling, where the first signaling is used to instruct configuration of the first radio bearer.

[0092] In an eighth aspect, a communication device is provided that can implement the communication method described in the fourth aspect. For example, the communication device can be a chip or a second node. The method can be implemented through software, hardware, or hardware executing corresponding software.

[0093] The communication device includes a transceiver unit and a processing unit, wherein the transceiver unit is used to receive artificial intelligence service data through a first radio bearer; and the processing unit is used to output the artificial intelligence service data to a physical PHY layer and / or a media access control MAC layer.

[0094] In one possible implementation, the processing unit is used to process the artificial intelligence service data at at least one of the following protocol layers: PHY layer, MAC layer, radio link control RLC layer, packet data convergence protocol PDCP layer; and the processing unit is also used to transmit the processed artificial intelligence service data to the PHY layer and / or MAC layer.

[0095] In another possible implementation, the first radio bearer is a learning radio bearer or a first data radio bearer.

[0096] In another possible implementation, the communication device is a network device, and the transceiver unit is further configured to send a first signaling, where the first signaling is used to instruct configuration of the first radio bearer.

[0097] In another possible implementation, the communication device is a terminal, and the transceiver unit is further configured to receive a first signaling, where the first signaling is used to instruct configuration of the first radio bearer.

[0098] Exemplarily, the processing unit may be a processor, and the transceiver unit may be a transceiver or a communication interface. It may also include a receiving unit and a sending unit, wherein the receiving unit may be a receiver or an input interface, and the sending unit may be a transmitter or a transmitter, or an output interface. Optionally, a storage unit may be included, and the storage unit may be a memory. It is understood that if the communication device is a chip provided in a device, the transceiver unit may be an input / output interface of the chip, such as an input / output circuit, a pin, etc.

[0099] In a ninth aspect, a communication device is provided, comprising a processor for executing the method described in the first aspect or any one of the implementations of the first aspect.

[0100] In a tenth aspect, a communication device is provided, comprising a processor for executing the method described in the second aspect or any one of the implementations of the second aspect.

[0101] In an eleventh aspect, a communication device is provided, comprising a processor for executing the method described in the third aspect or any one of the implementations of the third aspect.

[0102] In a twelfth aspect, a communication device is provided, comprising a processor for executing the method described in the fourth aspect or any one of the implementations of the fourth aspect.

[0103] In the thirteenth aspect, a communication device is provided, comprising a processor, a memory, and instructions stored in the memory and executable on the processor, wherein when the instructions are executed, the communication device executes the method described in the first aspect or any one implementation of the first aspect.

[0104] In the fourteenth aspect, a communication device is provided, comprising a processor, a memory, and instructions stored in the memory and executable on the processor, wherein when the instructions are executed, the communication device executes the method described in the second aspect or any one implementation of the second aspect.

[0105] In the fifteenth aspect, a communication device is provided, comprising a processor, a memory, and instructions stored in the memory and executable on the processor, wherein when the instructions are executed, the communication device executes the method described in the third aspect or any one implementation of the third aspect.

[0106] In the sixteenth aspect, a communication device is provided, comprising a processor, a memory, and instructions stored in the memory and executable on the processor, wherein when the instructions are executed, the communication device executes the method described in the fourth aspect or any one of the implementations of the fourth aspect.

[0107] In the seventeenth aspect, a communication system is provided, which includes the communication device of the fifth aspect and the communication device of the sixth aspect.

[0108] In the eighteenth aspect, a communication system is provided, which includes the communication device of the seventh aspect and the communication device of the eighth aspect.

[0109] In the nineteenth aspect, a computer-readable storage medium is provided, in which a computer program or instruction is stored. When a computer executes the computer program or instruction, the methods described in the above aspects are implemented.

[0110] In the twentieth aspect, a computer program product comprising instructions is provided, which, when executed on a communication device, causes the communication device to execute the methods described in the above aspects. BRIEF DESCRIPTION OF THE DRAWINGS

[0111] FIG1 is a schematic diagram of a communication system involved in this application;

[0112] FIG2 is a schematic diagram of a communication system according to a specific scenario provided by an embodiment of the present application;

[0113] FIG3 is a schematic diagram of another communication system involved in this application;

[0114] FIG4 is a schematic diagram of a communication system according to another specific scenario provided by an embodiment of the present application;

[0115] FIG5 is a schematic diagram of a federated learning architecture provided in an embodiment of the present application;

[0116] FIG6 is a schematic diagram of a segmentation learning architecture provided in an embodiment of the present application;

[0117] FIG7 is a schematic diagram of the architecture of a federal distillation system provided in an embodiment of the present application;

[0118] FIG8 is a schematic diagram of a transmission of application layer data provided by the prior art;

[0119] FIG9 is a schematic diagram of UCI / DCI communication provided in an embodiment of the present application;

[0120] FIG10 is a schematic diagram of the data processing flow of an existing physical channel;

[0121] FIG11 is a flow chart of a communication method provided in an embodiment of the present application;

[0122] FIG12 is a schematic diagram of communication of artificial intelligence service data provided by an embodiment of the present application;

[0123] FIG13a is a schematic diagram of a data processing flow of a physical learning channel provided in an embodiment of the present application;

[0124] FIG13b is a schematic diagram of a data processing flow of another physical learning channel provided in an embodiment of the present application;

[0125] FIG13c is a schematic diagram of a data processing flow of another physical learning channel provided in an embodiment of the present application;

[0126] FIG14 is a flow chart of another communication method provided in an embodiment of the present application;

[0127] FIG15 is a flow chart of another communication method provided in an embodiment of the present application;

[0128] FIG16 is a schematic diagram of a protocol stack processing flow for artificial intelligence business data provided by an embodiment of the present application;

[0129] FIG17 is a schematic structural diagram of a communication device provided in an embodiment of the present application;

[0130] FIG18 is a schematic diagram of the structure of a simplified terminal provided in an embodiment of the present application;

[0131] Figure 19 is a schematic structural diagram of a simplified network device provided in an embodiment of the present application. DETAILED DESCRIPTION

[0132] This application can be applied to the fifth generation (5 th generation, 5G) mobile communication system, the sixth generation (6 th generation, 6G) mobile communication systems, future evolved mobile communication systems, satellite communication systems, short-range communication systems or other communication systems, etc., are not limited in this application.

[0133] Figure 1 shows a schematic diagram of a communication system involved in this application, which includes one or more first nodes 101 (two first nodes are shown in the figure) and one or more second nodes 102 (one second node is shown in the figure). This communication system can be applied to scenarios such as cellular networks. For example, the first node 101 can be a distributed node, such as a terminal or a network device; the second node 102 can be a central node, such as a network device or a terminal.

[0134] As shown in Figure 2, a schematic diagram of a communication system for a specific scenario provided in an embodiment of the present application is provided, wherein the first node 101 is a terminal and the second node 102 is a network device. The first node 101 sends first indication information and / or second indication information to the second node 102, where the first indication information is used to indicate data set status information, and the second indication information is used to indicate model status information; the second node 102 determines the communication strategy and communication resources based on the first indication information and / or the second indication information; and the second node 102 sends third indication information to the first node 101, where the third indication information is used to indicate the communication strategy and communication resources, and the communication strategy and communication resources are determined based on the first indication information and / or the second indication information. The first node 101 can communicate artificial intelligence service data packets with the second node 102 based on the communication strategy and communication resources indicated by the third indication information.

[0135] FIG3 is a schematic diagram of another communication system involved in the present application, which includes a first node 201, a second node 202, and a third node 203. The communication system can be applied to scenarios such as vehicle-to-everything (V2X) and device-to-device (D2D) communication. For example, the first node 201 and the third node 203 can be distributed nodes, such as terminals; and the second node 202 can be a central node, such as a network device. The first node 201 sends first indication information and / or second indication information to the second node 202, where the first indication information is used to indicate dataset status information, and the second indication information is used to indicate model status information. The second node 202 determines a communication strategy and communication resources based on the first indication information and / or the second indication information. Furthermore, the second node 202 sends third indication information to the first node 201, where the third indication information is used to indicate a communication strategy and communication resources, where the communication strategy and communication resources are determined based on the first indication information and / or the second indication information. The first node 201 can communicate artificial intelligence service data packets with the third node 203 based on the communication strategy and communication resources indicated by the third indication information.

[0136] In the communication system shown in FIG2 , a terminal is a device with wireless transceiver capabilities that can be deployed on land (including indoors or outdoors), and can be handheld, wearable, or vehicle-mounted; can also be deployed on water, such as on a ship; and can also be deployed in the air, such as on an airplane, a balloon, or a satellite. The terminal can be a mobile phone, a tablet computer, a computer with wireless transceiver capabilities, a wearable device, a drone, a helicopter, an airplane, a ship, a robot, a robotic arm, a smart home device, a virtual reality (VR) terminal, an augmented reality (AR) terminal, a wireless terminal in industrial control, a wireless terminal in self-driving, a whole vehicle, a functional module in a vehicle, a wireless terminal in remote medical care, a wireless terminal in a smart grid, a wireless terminal in transportation safety, a wireless terminal in a smart city (e.g., a street light), a wireless terminal in a smart home, etc. The embodiments of the present application do not limit the application scenarios. A terminal may also be sometimes referred to as user equipment (UE), access terminal, UE unit, mobile station, mobile station, remote station, remote terminal, mobile device, terminal, wireless communication device, UE agent, or UE apparatus, etc. The embodiments of the present application do not limit the specific technology and specific device form adopted by the terminal.

[0137] A network device may be any device with wireless transceiver functions, including but not limited to: a base station (NodeB), an evolved base station (eNodeB), a base station in a 5G communication system, a base station or access network device in a future communication system, an access node in a WiFi system, a wireless relay node, a wireless backhaul node, etc. The network device may also be a wireless controller in a cloud radio access network (CRAN) scenario. The network device may also be a small station, a transmission reference point (TRP), etc. The embodiments of the present application do not limit the specific technology and specific device form adopted by the network device.

[0138] The communication system may also include core network equipment (not shown). The core network equipment primarily provides user connections, manages users, and carries services, acting as a bearer network and providing an interface to external networks. Establishing user connections includes functions such as mobility management (MM), call management (CM), switching / routing, and recorded notifications (combined with intelligent network services to establish connections to intelligent network peripheral devices). User management includes user descriptions, quality of service (QoS), user communication accounting, a virtual home environment (VHE) (interactions with the intelligent network platform provide a virtual home environment), and security (security measures provided by the authentication center include security management of mobile services and security processing for access to external networks). Bearer connections include connections to the external public switched telephone network (PSTN), external circuit data networks and packet data networks, the Internet, intranets, and mobile short message service (SMS) servers. The basic services that core network equipment can provide include mobile office, e-commerce, communications, entertainment services, travel and location-based services, telemetry services - simple messaging services (monitoring and control), etc.

[0139] Optionally, in an embodiment of the present application, a terminal or network device includes a hardware layer, an operating system layer running on the hardware layer, and an application layer running on the operating system layer. The hardware layer includes hardware such as a central processing unit (CPU), a memory management unit (MMU), and memory (also called main memory). The operating system can be any one or more computer operating systems that implement business processing through processes, such as a Linux operating system, a Unix operating system, an Android operating system, an iOS or Windows operating system, etc. The application layer includes applications such as browsers, address books, word processing software, and instant messaging software. In addition, the embodiment of the present application does not specifically limit the specific structure of the execution subject of the method provided in the embodiment of the present application. It can communicate according to the method provided in the embodiment of the present application by running a program that records the code of the method provided in the embodiment of the present application. For example, the execution subject of the method provided in the embodiment of the present application can be a terminal or network device, or a functional module in a terminal or network device that can call and execute a program.

[0140] In other words, the relevant functions of the terminal or network device in the embodiments of the present application can be implemented by a single device, or by multiple devices together, or by one or more functional modules within a single device, and the embodiments of the present application do not specifically limit this. It is understood that the above functions can be network elements in hardware devices, software functions running on dedicated hardware, or a combination of hardware and software, or virtualized functions instantiated on a platform (e.g., a cloud platform).

[0141] The communication between the network device and the terminal in the communication system shown in FIG2 can also be represented in another form. As shown in FIG4, which is a schematic diagram of a communication system according to another specific scenario provided by an embodiment of the present application, terminal 40 includes a processor 401, a memory 402, and a transceiver 403. Transceiver 403 includes a transmitter 4031, a receiver 4032, and an antenna 4033. Network device 41 includes a processor 411, a memory 412, and a transceiver 413. Transceiver 413 includes a transmitter 4131, a receiver 4132, and an antenna 4133. Receiver 4032 can be configured to receive transmission control information via antenna 4033, and transmitter 4031 can be configured to send transmission feedback information to network device 41 via antenna 4033. Transmitter 4131 can be configured to send transmission control information to terminal 40 via antenna 4133, and receiver 4132 can be configured to receive transmission feedback information sent by terminal 40 via antenna 4133.

[0142] The processor 401 / processor 411 may be a CPU, a microprocessor, an application-specific integrated circuit (ASIC), or one or more integrated circuits for controlling the execution of the program of the present application.

[0143] Memory 402 / memory 412 can be a device with a storage function. For example, it can be a read-only memory (ROM) or other types of static storage devices that can store static information and instructions, a random access memory (RAM) or other types of dynamic storage devices that can store information and instructions, or an electrically erasable programmable read-only memory (EEPROM), a compact disc read-only memory (CD-ROM) or other optical disc storage, optical disc storage (including compact disc, laser disc, optical disc, digital versatile disc, Blu-ray disc, etc.), a magnetic disk storage medium or other magnetic storage device, or any other medium that can be used to carry or store desired program code in the form of instructions or data structures and can be accessed by a computer, but is not limited thereto. The memory can exist independently and be connected to the processor via a communication line. The memory can also be integrated with the processor.

[0144] Memory 402 / memory 412 is used to store computer-executable instructions for executing the solution of the present application, and is controlled by processor 401 / processor 411. Processor 401 / processor 411 is used to execute the computer-executable instructions stored in memory 402 / memory 412, thereby implementing the information transmission method provided in the embodiments of the present application.

[0145] Alternatively, in an embodiment of the present application, the processor 401 / processor 411 may also perform processing-related functions in the information transmission method provided in the following embodiments of the present application.

[0146] The computer-executable instructions in the embodiments of the present application may also be referred to as application code, which is not specifically limited in the embodiments of the present application.

[0147] It should be noted that the terms "system" and "network" in the embodiments of the present application can be used interchangeably.

[0148] With the advent of the big data era, every device generates massive amounts of raw data in various forms every day. This data is isolated and exists in every corner of the world. Traditional centralized learning requires each edge device to transmit local data to a central server, which then uses the collected data for model training and learning. However, this architecture has been increasingly limited by the following factors:

[0149] (1) Edge devices are widely distributed in various regions and corners of the world. These devices will continuously generate and accumulate huge amounts of raw data at a rapid rate. If the center needs to collect raw data from all edge devices, it will inevitably bring huge communication losses and computing power requirements.

[0150] (2) As real-life scenarios become more complex, more and more learning tasks require edge devices to make timely and effective decisions and provide feedback. Traditional centralized learning involves uploading large amounts of data, which inevitably leads to a large degree of latency and makes it unable to meet the real-time requirements of actual task scenarios.

[0151] (3) Considering issues such as industry competition, user privacy and security, and complex administrative procedures, centralized data integration will face increasing resistance and constraints. Therefore, system deployment will increasingly tend to store data locally, while edge devices themselves will perform local model calculations.

[0152] Therefore, how to design a machine learning framework that enables artificial intelligence (AI) systems to share their data more efficiently and accurately while meeting data privacy, security, and regulatory requirements has become an important issue in the current development of AI.

[0153] The following solutions are currently proposed:

[0154] Federated Learning:

[0155] The concept of federated learning (FL) effectively solves the current difficulties faced by the development of artificial intelligence. It fully protects the privacy and security of user data by promoting the collaboration of various edge devices (also called "distributed nodes") and central servers (also called "central nodes") to efficiently complete the learning task of the model. As shown in Figure 5, a schematic diagram of the architecture of federated learning provided by an embodiment of the present application is shown. The FL architecture is the most widely used training architecture in the current FL field. The FedAvg algorithm is the basic algorithm of FL. Its algorithm flow is roughly as follows:

[0156] (1) The central node initializes the model to be trained And broadcast it to all distributed nodes.

[0157] (2) In the t∈[1, T]th round, the distributed node k∈[1, K] is based on the local data set For the received global model Perform E rounds of training to obtain local training results Report it to the central node.

[0158] (3) The central node aggregates and collects the local training results from all (or some) clients. Assume that the client set that uploads the local model in round t is The center will use the number of samples of the corresponding client as the weight to perform weighted averaging to obtain a new global model. The specific update rule is: The center then sends the latest version of the global model Broadcast to all client devices for a new round of training.

[0159] (4) Repeat steps (2) and (3) until the model finally converges or the number of training rounds reaches the upper limit.

[0160] In addition to reporting local models You can also use the local gradient of training After reporting, the central node averages the local gradients and updates the global model according to the direction of the average gradient.

[0161] As you can see, in the FL framework, datasets exist on distributed nodes. Distributed nodes collect local datasets, perform local training, and report the local training results (models or gradients) to the central node. The central node itself does not have a dataset; it is only responsible for fusing the training results of distributed nodes to obtain a global model and send it to the distributed nodes.

[0162] Segmentation Learning:

[0163] As shown in Figure 6, a schematic diagram of the architecture of split learning provided by an embodiment of the present application is shown. In split learning, the complete neural network model is divided into two parts (i.e., two sub-networks), one part is deployed on the distributed nodes, and the other part is deployed on the central node. The place where the complete neural network is split is called the "split layer". During forward reasoning, the distributed node inputs the local data into the local sub-network, reasons to the split layer, and sends the result Fk of the split layer to the central node through the communication link. The central node inputs the received Fk ​​into another sub-network deployed by itself, and continues forward reasoning to obtain the final reasoning result. In the gradient reverse transfer of model training, the gradient is reversely transferred to the split layer through the sub-network of the central node to obtain the reverse transfer result Gk, and then the central node sends Gk to the distributed node, and continues to perform gradient reverse transfer on the sub-network of the distributed node.

[0164] As you can see, the forward inference and gradient backpropagation process for segmentation learning involves only one distributed node and one central node. The trained subnetwork on a distributed node can be stored locally on the distributed node or on a dedicated model storage server. When a new distributed node joins the learning system, it can first download the trained subnetwork from the distributed node and then use its local data for further training.

[0165] Federal Distilling:

[0166] As shown in Figure 7, a schematic diagram of the architecture of federated distillation provided in an embodiment of the present application is provided. Federated distillation is also a feasible distributed learning method. Similar to federated learning, multiple distributed nodes interact with the central node to jointly complete the training of the machine learning model. Unlike federated learning, the information exchanged between the distributed nodes and the central node in federated distillation is logits, which are not model parameters or gradients in federated learning. Its workflow is similar to that of federated learning and will not be repeated here.

[0167] From the above introduction to federated learning, segmented learning, and federated distillation, we can see that different learning architectures require different information to be transmitted between participating nodes. For example, in federated learning, the parameters or gradients of the machine learning model are transmitted between the central node and the distributed nodes; in segmented learning, the distributed nodes transmit to the central node the intermediate results of inference to the segmentation layer, and the central node transmits to the distributed nodes the gradients reversely transferred to the segmentation layer; in federated distillation, the logits are transmitted between the central node and the distributed nodes. In this application, the above-mentioned information transmitted between nodes is collectively referred to as AI traffic. When distributed learning is used in the L1 / L2 protocol layer such as the physical layer or MAC layer of a wireless communication system (that is, the task of the model is a physical layer or MAC layer task, and the input and / or output of the model is from or sent to the physical layer or MAC layer), the above-mentioned AI traffic will be generated in the corresponding L1 / L2 protocol layer. They are different from the business data streams sent down by the application layer, and a new communication method needs to be designed.

[0168] However, existing communication solutions only provide protocol stack processing for upper-layer data such as the application layer / network layer, and cannot be used for data flows generated at lower protocol layers such as the physical layer / MAC layer, such as the aforementioned AI traffic. Figure 8 shows a schematic diagram of application layer data transmission provided by existing technologies. For data flows at upper layers such as the application layer / network layer, the new radio (NR) standard transmission method is described as follows: QoS flows are constructed and mapped step by step to the SDAP / PDCP / RLC / MAC / PHY layers in the protocol stack, and finally sent through the RF air interface. The functions of each protocol layer are as follows:

[0169] SDAP: maintains the mapping of QoS flows to data radio bearers (DRBs);

[0170] PDCP: data integrity and encryption;

[0171] RLC: caching, automatic repeat-request (ARQ);

[0172] MAC: radio resource management (RRM), hybrid automatic repeat-request (HARQ), multiplexing;

[0173] PHY: Builds transmission channels and performs channel coding, modulation, resource mapping, and other processing.

[0174] In addition, as shown in FIG9 , which is a schematic diagram of communication of uplink control information (UCI) and downlink control information (DCI), UCI / DCI is existing data to be transmitted generated at the physical layer / MAC layer.

[0175] Figure 10 illustrates the data processing flow for existing physical channels. The UCI / DCI is transmitted via a dedicated physical channel (control channel) after undergoing specialized processing as shown in Figure 10. In existing cellular network standards (Long Term Evolution (LTE) and NR), UCI / DCI has a fixed format consisting of multiple fields, with the meaning of the bits in each field defined by the standard. Taking the NR format0_0 DCI as an example, it includes: identifier (identifier), frequency domain location (F), time domain location (T), hopping, MCS, new data (newData), redundancy version (RV), HARQ, TPC, UL / SUL and other fields; while UCI may include channel state information (CSI) (channel quality indicator (CQI) / rank indication (RI) / precoding matrix indicator (PMI) / length indicator (LI) / CRI), acknowledgement (ACK) / non-acknowledgement (NACK) and other contents. UCI / DCI is shorter than ordinary data and has higher importance.

[0176] However, the dedicated physical channels and related processing used to transmit UCI / DCI, as shown in Figure 10, are only suitable for shorter bit sequences. This AI traffic may include larger data packets, such as machine learning model parameters, gradients, and even training data. The transmission of these packets may require unpacking and repackaging, which is not supported by the physical channels of existing protocols.

[0177] In view of this, the present application provides a communication solution, in which a first node sends first indication information and / or second indication information to a second node, wherein the first indication information is used to indicate data set status information, and the second indication information is used to indicate model status information; the second node determines a communication strategy and communication resources based on the first indication information and / or the second indication information; and the second node sends third indication information to the first node, wherein the third indication information is used to indicate a communication strategy and communication resources, and the communication strategy and communication resources are determined based on the first indication information and / or the second indication information. By adopting the solution of the present application, the communication strategy and communication resources are determined by obtaining the data set status information and / or model status information of the first node, and indicating the communication strategy and communication resources to the first node, thereby improving the reliability of communication.

[0178] As shown in FIG11, a flow chart of a communication method provided in an embodiment of the present application is applied to the communication system shown in FIG1. ​​Exemplarily, the method may include the following steps:

[0179] S1101. The first node sends first indication information and / or second indication information to the second node.

[0180] Correspondingly, the second node receives the first indication information and / or the second indication information.

[0181] The first indication information is used to indicate dataset state information (DSI), and the second indication information is used to indicate model state information (MSI).

[0182] Exemplarily, the first node may send first indication information to the second node; or the first node may send second indication information to the second node; or the first node may send first indication information and second indication information to the second node.

[0183] Exemplarily, the first node may report the first indication information and / or the second indication information through a data channel, a control channel or other dedicated channels.

[0184] Dataset status information refers to how the first node obtains the model's input data. This data set status information influences the communication strategy and resource determination of the AI ​​data packet. This data set status information includes at least one of the following: data type, sampling frequency, sampling location information, sample timestamp, sample number, and sample accuracy.

[0185] For example, regarding data types, the physical layer includes at least one of the following types of data: channel, signal strength, signal to interference plus noise ratio (SINR), and the MAC layer includes at least one of the following types of data: scheduling, resource allocation, buffer status, and link adaptation.

[0186] The sampling frequency can be a number of minislots / slots / subframes / frames / seconds.

[0187] The sampling location information may be represented by any of the following methods: cell identifier, base station distance, or global positioning system (GPS) location, which is not limited in this application.

[0188] Sample precision can be represented by at least one of the following: double-precision floating point number (double), 64-bit floating point number (float64), floating point number (float), 32-bit floating point number (float32), 16-bit floating point number (float16), 8-bit floating point number (float8), integer (int), 64-bit integer (int64), 32-bit integer (int32), 16-bit integer (int16), 8-bit integer (int8), binary number (bin).

[0189] Exemplarily, the first indication information is used to indicate dataset status information, and may indicate each piece of the at least one dataset status information separately or jointly. For each piece of the at least one dataset status information, for example, for a data type, multiple bits may be used to indicate multiple data types. For example, if there are eight data types, a three-bit binary number may be used to indicate the eight data types.

[0190] The model status information refers to a description of the model used by the first node. The model status information includes at least one of the following information: model type, model identifier, sub-model order, model version number, model description, model update timestamp, model update location information, model accuracy, and model performance.

[0191] The model type includes at least one of the following: fully connected neural network (FCNN), convolutional neural networks (CNN), recurrent neural networks (RNN), transformer, and hybrid model.

[0192] The model identifier is used to index the corresponding model and task.

[0193] The order of submodels is used to indicate whether the submodels and their order are used to build the complete model.

[0194] The model version number is used for version synchronization.

[0195] Model description includes parameter quantity, inference operation quantity, etc.

[0196] Model performance refers to the inference performance on the reference dataset.

[0197] Exemplarily, the second indication information is used to indicate model status information, and may indicate each piece of the at least one piece of model status information separately or jointly. For each piece of the at least one piece of model status information, for example, for a model type, multiple bits may be used to indicate multiple data types. For example, if there are six model types, a three-bit binary number may be used to indicate the six model types.

[0198] S1102. The second node determines a communication strategy and communication resources according to the first indication information and / or the second indication information.

[0199] In the communication system shown in Figure 1, the second node may communicate with one or more first nodes. After receiving the first indication information and / or the second indication information, the second node schedules the first node and determines the first node to which the artificial intelligence service data needs to be sent uplink.

[0200] In addition, the channel state between the first node and the second node has a certain impact on the determination of the first node that needs to send artificial intelligence service data uplink. Therefore, the first node that needs to send artificial intelligence service data uplink can also be determined based on at least two of the first indication information, the second indication information, and the channel state information.

[0201] The second node can perform channel estimation to determine the channel state information between the first node and the second node; alternatively, the second node can also obtain the channel state information between the first node and the second node through channel reciprocity, feedback of channel state information by the first node, etc.

[0202] The second node schedules the first node in the following ways, but is not limited to:

[0203] One approach is to schedule the first node with good channel conditions;

[0204] Another method is to schedule a first node whose data set meets the requirements, for example, a first node whose one or more parameters such as scheduling data type, sampling frequency, sampling location, sampling time, sample quantity, and sample accuracy meet the requirements of a second node.

[0205] Another method is to schedule a first node whose model meets the requirements, for example, scheduling a first node whose one or more parameters such as model type, model identifier, model description, model update time, model update location, model accuracy, and model performance meet the requirements;

[0206] Another implementation is to schedule the first node with good channel conditions and a data set that meets the requirements;

[0207] Another implementation is to schedule the first node with good channel conditions and a model that meets the requirements;

[0208] Another implementation is to schedule the first node with good channel conditions, a data set that meets the requirements, and a model that meets the requirements.

[0209] After the second node determines the first node that needs to send artificial intelligence service data uplink, it determines the communication strategy used by the first node for uplink transmission.

[0210] The communication strategy includes at least one of the following strategies: a quantization strategy, a source coding strategy, a modulation and coding scheme (MCS), a source-channel joint coding strategy, and a real or complex number processing strategy.

[0211] The data set status information and the model status information affect the communication strategy adopted by the first node. The second node can determine the above communication strategy based on the first indication information, the second indication information, or the first indication information and the second indication information.

[0212] Furthermore, the channel status between the first and second nodes has a certain impact on the communication strategy adopted by the first and second nodes. For example, if the channel conditions are good, a higher quantization accuracy can be selected; if the channel conditions are poor, a lower quantization accuracy can be selected. For another example, if the channel conditions are good, a higher MCS level can be selected; if the channel conditions are poor, a lower MCS level can be selected. Therefore, the communication strategy can also be determined based on at least two of the first indication information, the second indication information, and the channel status information.

[0213] The second node may also determine, based on the first indication information and / or the second indication information, the communication resources that the first node can use or need to use.

[0214] The communication resources that the first node can use or need to use are related to the size of the AI ​​service data packet that the first node needs to transmit uplink. For example, the second node can first obtain the size of the AI ​​service data packet and then determine the communication resources based on at least two of the size of the AI ​​service data packet, the first indication information, and the second indication information.

[0215] There are two ways to obtain the size of the AI ​​service data packet:

[0216] One approach is to predefine a fixed data packet size for each type of AI service data packet for each type of AI task, considering the limited number of AI-related tasks at the PHY / MAC layer, as shown in Table 1 below:

[0217] Table 1 Fixed-size artificial intelligence service data packets

[0218]

[0219] As shown in Table 1, the AI ​​task is channel state information (CSI) compression, and the type of the artificial intelligence service data packet is data, and the size of the corresponding artificial intelligence service data packet is S1; the AI ​​task is CSI compression, and the type of the artificial intelligence service data packet is model parameters, and the size of the corresponding artificial intelligence service data packet is S2; and so on.

[0220] The second node can obtain the task type of the AI ​​task and the type of the AI ​​service data packet. Therefore, the size of the AI ​​service data packet (i.e., the amount of data to be communicated) can be determined based on the correspondence between at least one of the task type and the service type and the data packet size. For example, if the second node obtains that the AI ​​task type is beam management and the AI ​​service data packet type is model parameter, the second node can determine the size of the AI ​​service data packet to be A2 based on the correspondence in Table 1 above.

[0221] After the second node determines the size of the artificial intelligence service data packet, it can determine the communication resources based on at least two of the size of the artificial intelligence service data packet, the first indication information, and the second indication information.

[0222] In one example, the second node calculates the number of modulation symbols based on the size of the artificial intelligence service data packet, the quantization strategy, the source coding strategy, and the MCS, and then determines the communication resources based on the number of modulation symbols.

[0223] In another example, the second node calculates the number of modulation symbols based on the size of the artificial intelligence service data packet, the quantization strategy, the source-channel joint coding strategy, and the modulation strategy, and then determines the communication resources based on the number of modulation symbols.

[0224] In another example, the second node calculates the number of real symbols to be communicated based on the size of the artificial intelligence service data packet and the real or complex number processing strategy, and then determines the communication resources based on the number of real symbols to be communicated.

[0225] Another way is that when the size of the artificial intelligence business data packet is not limited to a fixed size in a manner similar to Table 1 as described above, the first node can report the size of the artificial intelligence business data packet to the second node, and the second node receives the size of the artificial intelligence business data packet reported by the first node.

[0226] S1103. The second node sends third indication information to the first node.

[0227] Correspondingly, the first node receives the third indication information.

[0228] The third indication information is used to indicate the above-mentioned communication strategy and communication resources.

[0229] Exemplarily, the second node may indicate the scheduled first node through downlink control information (DCI) or learning control information (LCI), and indicate the communication strategy and communication resources used for its uplink transmission. The second node may use a reserved field in an existing DCI format to carry the above-mentioned communication strategy and communication resources; or use a newly defined DCI format to carry the above-mentioned communication strategy and communication resources; or use a newly defined LCI to carry the above-mentioned communication strategy and communication resources. The LCI may be a newly defined information format for distributed learning.

[0230] Exemplarily, the DCI / LCI may be scrambled using a dedicated radio network temporary identity (RNTI), such as a learning-radio network temporary identity (L-RNTI).

[0231] After receiving the DCI / LCI, the first node uses a dedicated RNTI to descramble the DCI / LCI. If the descrambling is successful, the DCI / LCI is sent to the first node.

[0232] Optionally, the method may further include the following steps (indicated by dotted lines in the figure):

[0233] S1104. The first node generates an artificial intelligence service data packet according to the third indication information.

[0234] After receiving the third indication information, the first node generates an artificial intelligence service data packet according to the third indication information. Exemplarily, the first node generates an artificial intelligence service data packet according to the communication strategy and communication resources indicated by the third indication information. For example, when the communication resources required for the artificial intelligence service data packet are greater than the communication resources indicated by the third indication information, the first node may unpack the artificial intelligence service data packet and split it into several sub-data packets so that each sub-data packet complies with the communication resources indicated by the third indication information. For another example, the first node quantizes the artificial intelligence service data packet according to the quantization strategy (a type of communication strategy) indicated by the third indication information to comply with the quantization strategy indicated by the third indication information. For another example, the first node modulates the artificial intelligence service data packet according to the modulation and coding strategy (a type of communication strategy) indicated by the third indication information to comply with the modulation and coding strategy indicated by the third indication information.

[0235] Among them, the artificial intelligence business data packet includes a header and a payload.

[0236] The above-mentioned load includes at least one of the following data: parameters or gradients of the machine learning model, inference results, intermediate results of inference to the segmentation layer, gradients passed back to the segmentation layer, and logarithmic scores.

[0237] The above-mentioned header includes at least one of the following:

[0238] Content indication information: used to indicate the at least one type of data included in the payload;

[0239] Content description information: used to indicate the timestamp, acquisition location, version information, etc. of the at least one type of data included in the payload;

[0240] Compression method: used to indicate the pruning location of the load (i.e., the location where the AI ​​service data packet is split), whether distillation is performed, etc.

[0241] Packet assembly method: used to indicate the serial number of a complete artificial intelligence business data packet and the serial number of the sub-packet after the complete artificial intelligence business data packet is split (when the data volume of the artificial intelligence business data packet is large, it can be sent through multiple sub-packets, and the serial number of each sub-packet is included in the packet header, which is used for the second node to assemble the packets after receiving multiple sub-packets).

[0242] S1105. The first node sends an artificial intelligence service data packet to the second node according to the third indication information.

[0243] Correspondingly, the second node receives the artificial intelligence service data packet sent according to the third indication information.

[0244] Exemplarily, the first node sends the artificial intelligence service data packet to the second node on the communication resources indicated by the third indication information according to the communication strategy indicated by the third indication information.

[0245] Furthermore, the above-mentioned artificial intelligence business data packets communicate through a physical learning channel.

[0246] As shown in FIG12 , which is a communication diagram of artificial intelligence service data provided by an embodiment of the present application, the physical learning channel is a new physical channel defined for artificial intelligence service data at the PHY / MAC layer, and may include:

[0247] Uplink: physical uplink learning channel (PULCH);

[0248] Downlink: physical downlink learning channel (PDLCH);

[0249] Physical broadcast learning channel (PBLCH).

[0250] Exemplarily, the above-mentioned PULCH / PDLCH / PBLCH may be carried by dedicated time-frequency resources.

[0251] The main difference between the processing of the physical learning channel described above and the existing technology (as shown in Figure 10) is that AI service data is generally in real or complex form and cannot be directly processed by bit-level channel coding. Therefore, operations such as quantization, source coding, source-channel joint coding, and real or complex number processing need to be added to the data processing flow of the physical learning channel. Possible implementation methods of the processing operation of the physical learning channel are:

[0252] Implementation method 1): Communication through a physical learning channel includes at least one of the following processing operations: quantization, source coding. For example, as shown in FIG13a, a data processing flow diagram of a physical learning channel provided in an embodiment of the present application is provided, and communication through a physical learning channel includes the following processing operations: quantization, source coding, channel coding, rate matching, scrambling, modulation, layer mapping, precoding, resource mapping, orthogonal frequency division multiplexing (OFDM) signal generation, etc. Among them, channel coding and subsequent processing operations can reuse the corresponding processing operations as shown in FIG10, or specially designed corresponding processing operations can be used (for example, using channel coding, modulation, and signal generation methods that are different from existing standards). It can be seen that in the processing flow shown in FIG13a, compared with the existing standard, the above-mentioned physical learning channel adds "quantization" and "source coding" processing operations. Among them, if the artificial intelligence service data is a real number, quantization refers to quantizing the real value contained in the artificial intelligence service data into several real numbers with a limited number of possibilities. If the AI ​​service data is complex, quantization refers to quantizing the real part of the complex number contained in the AI ​​service data into a limited number of real numbers, and quantizing the imaginary part of the complex number contained in the AI ​​service data into a limited number of real numbers. Quantization methods such as uniform quantization, non-uniform quantization, and vector quantization can be used. Source coding refers to encoding the quantized real value into a bit sequence. Source coding methods such as Huffman coding, arithmetic coding, and LZ coding can be used.

[0253] Implementation method 2): Communication through the physical learning channel includes at least one of the following processing operations: quantization, source channel joint coding. For example, as shown in Figure 13b, a data processing flow diagram of another physical learning channel provided in an embodiment of the present application is provided. Communication through the physical learning channel includes the following processing operations: quantization, source channel joint coding, rate matching, scrambling, modulation, layer mapping, precoding, resource mapping, OFDM signal generation, etc. Among them, rate matching and subsequent processing operations can reuse the corresponding processing operations in Figure 10, or use specially designed corresponding processing operations (for example, using channel coding, modulation, and signal generation methods that are different from existing standards). It can be seen that in the processing operations shown in Figure 13b, compared with the existing standards, the above-mentioned physical learning channel adds "quantization" and "source channel joint coding" processing operations and reduces "channel coding" processing operations. Among them, the concept of quantization operation can refer to the description of the above-mentioned implementation method 1). Source channel joint coding refers to encoding the real value obtained by quantization into a bit sequence. The source channel joint coding operation combines source coding and channel coding. Through this process, the quantized output is compressed and the reliability of the transmission is improved.

[0254] Implementation method 3): Communication through the physical learning channel includes the following processing operations: real number processing. For example, as shown in Figure 13c, which is a data processing flow diagram of another physical learning channel provided in an embodiment of the present application, communication through the physical learning channel includes the following processing operations: real number processing, layer mapping, precoding, resource mapping, OFDM signal generation, etc. Among them, the layer mapping and subsequent processing operations can reuse the corresponding processing operations as shown in Figure 10, or use specially designed corresponding processing operations (for example, using channel coding, modulation, and signal generation methods that are different from existing standards). It can be seen that in the processing operations shown in Figure 13c, compared with the existing standards, the above-mentioned physical learning channel adds the "real or complex number processing" operation and reduces the "channel coding", "rate matching", "scrambling", "modulation" and other processing operations. The real or complex number processing operation refers to mapping the real and imaginary parts of the real or complex values ​​contained in the artificial intelligence business data into a set of real number sequences, and directly performing subsequent processing such as layer mapping for transmission.

[0255] For example, a real or complex number processing operation may map the real and imaginary parts of an input real or complex value to a real number sequence. For example, if the input is a real value, the mapping method may be to directly equalize the input (i.e., the output is equal to the input) or to map the input real value to a finite number of real values ​​(e.g., numbers less than or equal to -1 are mapped to -1, numbers between -1 and 0 are mapped to -0.5, numbers between 0 and 1 are mapped to 0.5, and numbers greater than or equal to 1 are mapped to 1).

[0256] The real number sequence output by the real or complex number processing operation is mapped to time-frequency-space transmission resources through operations such as layer mapping, precoding, and resource mapping. After signal generation, the resulting signal is transmitted. Multiple signals transmitted simultaneously by a first node on the same time-frequency-space transmission resource are superimposed and received by a second node. The second node can then process the superimposed signals to obtain the required information. This is the process of over-the-air computing.

[0257] Exemplarily, the header and payload of the above-mentioned artificial intelligence service data packet can be sent separately. For example, the bit sequence contained in the header can be sent through the existing physical channel as shown in Figure 10 after channel coding, rate matching, scrambling, modulation, etc.; the payload can be sent through the above-mentioned physical learning channel.

[0258] After receiving the above artificial intelligence service data packet, the second node unpacks the artificial intelligence service data packet according to the packet header to obtain the required artificial intelligence service data packet.

[0259] According to a communication method provided in an embodiment of the present application, a communication strategy and communication resources are determined by obtaining data set status information and / or model status information of a first node, and the communication strategy and communication resources are indicated to the first node. The first node communicates with the second node according to the communication strategy and communication resources, thereby improving the reliability of communication.

[0260] In another embodiment, step S1104 can be replaced by the second node generating an artificial intelligence service data packet based on the third indication information. For details on how to generate the artificial intelligence service data packet, please refer to the description of step S1104 above. Step S1105 can also be replaced by the second node sending the artificial intelligence service data packet to the first node based on the third indication information. Accordingly, the first node receives the artificial intelligence service data packet sent based on the third indication information. That is, after the second node determines the third indication information, it itself sends the artificial intelligence service data packet to the first node based on the third indication information. The communication of the artificial intelligence service data packet can be referred to the description of step S1105 above and will not be repeated here.

[0261] FIG14 is a flow chart of another communication method provided in an embodiment of the present application, which is applied to the communication system shown in FIG3. Exemplarily, the method may include the following steps:

[0262] S1401. A first node sends first indication information and / or second indication information to a second node, wherein the first indication information is used to indicate dataset status information, and the second indication information is used to indicate model status information.

[0263] Correspondingly, the second node receives the first indication information and / or the second indication information.

[0264] The specific implementation of this step can refer to step S1101 shown in FIG11 , and will not be described in detail here.

[0265] S1402. The second node determines a communication strategy and communication resources according to the first indication information and / or the second indication information.

[0266] The specific implementation of this step can refer to step S1102 shown in FIG11 , and will not be described in detail here.

[0267] In which, when determining the first node that needs to send artificial intelligence service data uplink according to the first indication information, the second indication information and the channel state information, and determining the communication strategy according to the first indication information, the second indication information and the channel state information, the channel state information includes the channel state information between the first node and the third node; or the channel state information includes the channel state information between the first node, the second node and the third node.

[0268] S1403: The second node sends third indication information to the first node, where the third indication information is used to indicate a communication strategy and communication resources.

[0269] Correspondingly, the first node receives the third indication information.

[0270] The specific implementation of this step can refer to step S1103 shown in FIG11 , and will not be described in detail here.

[0271] Optionally, the method may further include the following steps (indicated by dotted lines in the figure):

[0272] S1404. The first node generates an artificial intelligence service data packet based on the third indication information.

[0273] The specific implementation of this step can refer to step S1104 shown in FIG11 , and will not be described in detail here.

[0274] S1405. The first node sends an artificial intelligence service data packet to the third node according to the third indication information.

[0275] Correspondingly, the third node receives the artificial intelligence service data packet.

[0276] Unlike the embodiment shown in Figure 11, this method is applied to the communication system shown in Figure 3. The first node communicates with the third node based on the communication strategy and communication resources indicated by the third indication information sent by the second node. For example, the first node sends an artificial intelligence service data packet to the third node based on the third indication information.

[0277] According to a communication method provided in an embodiment of the present application, a communication strategy and communication resources are determined by obtaining data set status information and / or model status information of a first node, and the communication strategy and communication resources are indicated to the first node. The first node communicates with a third node according to the communication strategy and communication resources, thereby improving the reliability of communication.

[0278] In another embodiment, step S1403 can be replaced by the second node sending third indication information to the third node. Accordingly, the third node receives the third indication information. Step S1404 can be replaced by the third node generating an artificial intelligence service data packet based on the third indication information. Step S1405 can be replaced by the third node sending an artificial intelligence service data packet to the first node based on the third indication information. Accordingly, the first node receives the artificial intelligence service data packet.

[0279] The above embodiments describe methods for communicating artificial intelligence service data at low-level protocol stacks such as the physical layer and / or MAC layer. The following embodiments describe methods for communicating artificial intelligence service data at high-level protocol stacks:

[0280] As shown in Figure 15, a flow chart of another communication method provided in an embodiment of the present application is applied to the communication system shown in Figure 1 or Figure 3. The communication system involves a first node and a second node. The first node can be a terminal and the second node can be a network device; alternatively, the first node can be a network device and the second node can be a terminal. Exemplarily, the method may include the following steps:

[0281] S1501. The first node obtains artificial intelligence service data generated by the physical layer and / or MAC layer.

[0282] The first node uses machine learning in an L1 / L2 protocol layer, such as a physical layer or a MAC layer, of a wireless communication system, and obtains artificial intelligence service data generated by the physical layer and / or the MAC layer. The meaning and content of the artificial intelligence service data can be referred to in the aforementioned embodiments.

[0283] S1502. The first node sends artificial intelligence service data to the second node through the first radio bearer.

[0284] Correspondingly, the second node receives the artificial intelligence service data through the first radio bearer.

[0285] For example, a protocol stack processing flow for artificial intelligence service data is shown in Figure 16, in which an artificial intelligence service data control (ATC) layer is added to the protocol stack. The ATC layer is used to input the artificial intelligence service data generated by the physical layer / MAC layer into the first radio bearer.

[0286] Exemplarily, another protocol stack processing flow for artificial intelligence service data may also be to add a function of inputting the artificial intelligence service data generated by the physical layer / MAC layer into the first wireless bearer in the existing module or protocol stack, so that the existing module or protocol stack with this newly added function can input the artificial intelligence service data generated by the physical layer / MAC layer into the first wireless bearer.

[0287] Exemplarily, the first radio bearer is a newly defined learning radio bearer (LRB) or a first data radio bearer. The first data radio bearer is a new data radio bearer (newDRB).

[0288] After the ATC layer or an existing module or protocol stack with this added function inputs the artificial intelligence service data into the first radio bearer, the first node processes the artificial intelligence service data at at least one of the following protocol layers: PDCP layer, RLC layer, MAC layer, and PHY layer, to obtain processed artificial intelligence service data. Specifically, the first node performs data integrity protection and encryption on the artificial intelligence service data at the PDCP layer; the first node caches, segments, and reassembles the artificial intelligence service data processed at the PDCP layer at the RLC layer; the first node performs mapping, multiplexing / demultiplexing, scheduling, HARQ, and logical channel priority setting between logical channels and transport channels on the MAC layer for the artificial intelligence service data processed at the RLC layer; and the first node performs channel coding, modulation, resource mapping, and other processing on the artificial intelligence service data processed at the MAC layer at the PHY layer. The first node then transmits the processed artificial intelligence service data to the second node via a data channel, control channel, or dedicated channel.

[0289] Correspondingly, the second node receives the processed artificial intelligence business data.

[0290] S1503. The second node outputs the artificial intelligence service data to the physical layer and / or MAC layer.

[0291] After the second node receives the processed artificial intelligence service data through the radio frequency module, the second node processes the artificial intelligence service data through at least one of the following protocol layers: the PHY layer, the MAC layer, the RLC layer, and the PDCP layer. Processing the artificial intelligence service data through at least one protocol layer may mean processing the artificial intelligence service data through one or more of the PHY layer, the MAC layer, the RLC layer, and the PDCP layer. Exemplarily, the second node processes the artificial intelligence service data sequentially through the PHY layer, the MAC layer, the RLC layer, and the PDCP layer.

[0292] Then, the second node outputs the processed artificial intelligence service data to the PHY layer and / or MAC layer.

[0293] In one example, the second node can input the processed artificial intelligence service data to the ATC layer through the first radio bearer, and the ATC layer outputs the artificial intelligence service data input from the first radio bearer to the PHY layer and / or MAC layer.

[0294] In another example, the second node may also input the processed artificial intelligence service data into an existing module or protocol layer through the first radio bearer. The existing module or protocol layer adds the function of outputting the artificial intelligence service data input from the first radio bearer to the PHY layer and / or MAC layer, thereby outputting the artificial intelligence service data input from the first radio bearer to the PHY layer and / or MAC layer.

[0295] The processing flow of the second node is the reverse process of the processing flow of Figure 16, and will not be repeated here.

[0296] According to a communication method provided in an embodiment of the present application, a new protocol stack processing flow is defined. For artificial intelligence service data generated at the PHY / MAC layer, after being input into the first wireless bearer through the ATC layer, the artificial intelligence service data can be sent to the second node through the first wireless bearer, thereby realizing the communication of artificial intelligence service data generated at the PHY / MAC layer.

[0297] In another embodiment, in order to establish the first radio bearer, the following steps may also be included:

[0298] The network device sends a first signaling to the terminal, where the first signaling is used to instruct configuration of a first radio bearer.

[0299] Correspondingly, the terminal receives the first signaling.

[0300] In combination with the above embodiment, when the first node is a terminal and the second node is a network device, the second node sends the above first signaling to the first node; when the first node is a network device and the second node is a terminal, the first node sends the above first signaling to the second node.

[0301] Exemplarily, the first signaling is radio resource control (RRC) signaling. Taking the first radio bearer as an LRB as an example, configuring the first radio bearer includes adding an LRB and releasing an LRB. Specifically, adding / releasing an LRB in the radio bearer configuration (RadioBearerConfig) information element of the RRC signaling, the specific content of the information element is:

[0302]

[0303]

[0304] For each LRB to be added (LRB-ToAddMod) in the LRB list to be added (LRB-ToAddModList), check its LRB identifier (LRB-Identity field) (exemplarily, a number can be used for identification),

[0305] a) If the LRB corresponding to the LRB-Identity field is not in the LRB list of the current terminal

[0306] i. Establish a PDCP entity (reestablishPDCP) and configure it according to the PDCP configuration (pdcp-Config) field in LRB-ToAddMod. The configuration method is the same as the existing PDCP entity configuration method;

[0307] b) If the LRB corresponding to the LRB-Identity field is already in the LRB list of the current terminal

[0308] i. If the reestablishPDCP field is true (true), the PDCP entity is re-established and configured according to the pdcp-Config field in LRB-ToAddMod. The configuration method is the same as the existing PDCP entity configuration method;

[0309] ii. If the recoverPDCP field is true, the LRB pair corresponding to the LRB-Identity field is restored.

[0310] The corresponding PDCP entity.

[0311] The process of releasing the LRB includes: releasing the LRB corresponding to the LRB-Identity field in the LRB to be released list (LRB-ToReleaseList) and its corresponding PDCP entity.

[0312] An example LRB establishment process is as follows. The meanings of the fields that are the same as those in existing communication standards are explained in the existing communication standards and are not repeated here.

[0313] The network device sends a radio resource control reconfiguration (RRCReconfiguration) signaling to the terminal:

[0314]

[0315] The RRCReconfiguration signaling includes a radio resource control reconfiguration (rrcReconfiguration) field, which includes radio resource control reconfiguration-information elements (RRCReconfiguration-IEs). The radio resource control reconfiguration field is used to configure the establishment of the LRB.

[0316] Furthermore, the RRCReconfiguration-IEs field is:

[0317]

[0318]

[0319] The radio resource control reconfiguration-information element further includes a radio bearer configuration (radioBearerConfig) field, which is used to configure the establishment of LRB.

[0320] Furthermore, the RadioBearerConfig field is:

[0321]

[0322] The radio bearer configuration field includes a list of LRBs to be added (lrb-ToAddModList) and a list of LRBs to be released (lrb-ToReleaseList). The contents of the LRB-ToAddModList and LRB-ToReleaseList fields are as described above, describing the LRBs to be established and released.

[0323] After receiving the RRCReconfiguration signaling, the terminal establishes or releases the LRB according to the contents of the LRB-ToAddModList and LRB-ToReleaseList fields.

[0324] Another possible example of an LRB establishment process for resuming a suspended RRC connection is as follows, wherein the meanings of the fields that are the same as those in the existing communication standard are explained in the existing communication standard and are not repeated here:

[0325] The network device sends a radio resource control resumption (RRCResume) signaling to the terminal:

[0326]

[0327] The radio resource control resume signaling includes a radio resource control resume (rrcResume) field, which in turn includes radio resource control resume-information elements (RRCResume-IEs). The radio resource control resume field is used to resume a suspended RRC connection.

[0328] Furthermore, the RRCResume-IEs field is

[0329]

[0330]

[0331] The radio resource control recovery-information element includes a radio bearer configuration (radioBearerConfig) field. The radio bearer configuration field further includes a radio bearer configuration (RadioBearerConfig) field.

[0332] Furthermore, the RadioBearerConfig field is:

[0333]

[0334] The RadioBearerConfig field includes a list of LRBs to be added (lrb-ToAddModList) and a list of LRBs to be released (lrb-ToReleaseList). The LRB-ToAddModList and LRB-ToReleaseList fields contain the contents described above, describing the LRBs to be established and released.

[0335] After receiving the RRCResume signaling, the terminal establishes or releases the LRB according to the contents of the LRB-ToAddModList and LRB-ToReleaseList fields, and the process is as described above.

[0336] It can be understood that in the above embodiments, the methods and / or steps implemented by the first node can also be implemented by components (such as chips or circuits) that can be used for the first node; the methods and / or steps implemented by the second node can also be implemented by components (such as chips or circuits) that can be used for the second node.

[0337] The above primarily describes the solutions provided by the embodiments of the present application from the perspective of interaction between various network elements. Accordingly, the embodiments of the present application also provide a communication device for implementing the various methods described above. The communication device can be the first node in the method embodiments described above, or a component that can be used by the first node; alternatively, the communication device can be the second node or the first node in the method embodiments described above, or a component that can be used by the second node or the first node. It will be understood that, to implement the aforementioned functions, the communication device includes hardware structures and / or software modules corresponding to the respective functions. Those skilled in the art will readily appreciate that, in conjunction with the various exemplary units and algorithm steps described in the embodiments disclosed herein, the present application can be implemented in hardware or a combination of hardware and computer software. Whether a function is implemented in hardware or in a hardware-driven manner by computer software depends on the specific application and design constraints of the technical solution. Professionals and technicians may use different methods to implement the described functions for each specific application, but such implementations should not be considered beyond the scope of this application.

[0338] In the embodiment of the present application, the functional modules of the communication device can be divided according to the above method embodiment. For example, each functional module can be divided according to each function, or two or more functions can be integrated into one processing module. The above integrated modules can be implemented in the form of hardware or in the form of software functional modules. It should be noted that the division of modules in the embodiment of the present application is schematic and is only a logical functional division. In actual implementation, there may be other division methods.

[0339] Based on the same concept of the above communication method, the present application also provides the following communication device:

[0340] As shown in FIG17 , it is a schematic diagram of the structure of a communication device provided in an embodiment of the present application. The communication device 1700 includes: a transceiver unit 1701 and a processing unit 1702.

[0341] In one example, when the communication device 1700 is used to implement the function of the first node in the above embodiment, the transceiver unit 1701 is used to implement the operation of the first node in steps S1101, S1103 and S1105 in the embodiment as shown in Figure 11; and the processing unit 1702 is used to implement step S1104 in the embodiment as shown in Figure 11.

[0342] When the communication device 1700 is used to implement the function of the second node in the above embodiment, the transceiver unit 1701 is used to implement the operation of the second node in steps S1101, S1103 and S1105 in the embodiment as shown in Figure 11; and the processing unit 1702 is used to implement step S1102 in the embodiment as shown in Figure 11.

[0343] In another example, when the communication device 1700 is used to implement the function of the first node in the above embodiment, the transceiver unit 1701 is used to implement the operation of the first node in steps S1401, S1403 and S1405 in the embodiment as shown in Figure 14; and the processing unit 1702 is used to implement step S1404 in the embodiment as shown in Figure 14.

[0344] When the communication device 1700 is used to implement the function of the second node in the above embodiment, the transceiver unit 1701 is used to implement the operation of the second node in steps S1401 and S1403 in the embodiment as shown in Figure 14; and the processing unit 1702 is used to implement step S1402 in the embodiment as shown in Figure 14.

[0345] In another example, when the communication device 1700 is used to implement the function of the first node in the above embodiment, the transceiver unit 1701 is used to implement the operation of the first node in step S1502 in the embodiment as shown in Figure 15; and the processing unit 1702 is used to implement step S1501 in the embodiment as shown in Figure 15.

[0346] When the communication device 1700 is used to implement the function of the second node in the above embodiment, the transceiver unit 1701 is used to implement the operation of the second node in step S1502 in the embodiment as shown in Figure 15; and the processing unit 1702 is used to implement step S1503 in the embodiment as shown in Figure 15.

[0347] The first node in the above embodiment can be a terminal. Figure 18 shows a simplified structural diagram of a terminal. For ease of understanding and illustration, in Figure 18, the terminal takes a mobile phone as an example. As shown in Figure 18, the terminal includes a processor, a memory, a radio frequency circuit, an antenna, and input and output devices. The processor is mainly used to process communication protocols and communication data, as well as to control the terminal, execute software programs, process software program data, etc. The memory is mainly used to store software programs and data. The radio frequency circuit is mainly used for converting baseband signals into radio frequency signals and processing radio frequency signals. The antenna is mainly used to transmit and receive radio frequency signals in the form of electromagnetic waves. Input and output devices, such as touch screens, display screens, keyboards, etc., are mainly used to receive data input by users and output data to users. It should be noted that some types of terminals may not have input and output devices.

[0348] When data needs to be sent, the processor performs baseband processing on the data to be sent and outputs the baseband signal to the RF circuit. The RF circuit performs RF processing on the baseband signal and then transmits the RF signal to the outside in the form of electromagnetic waves through the antenna. When data is sent to the terminal, the RF circuit receives the RF signal through the antenna, converts the RF signal into a baseband signal, and outputs the baseband signal to the processor. The processor converts the baseband signal into data and processes the data. For ease of explanation, only one memory and processor are shown in Figure 18. In an actual terminal product, there may be one or more processors and one or more memories. The memory may also be referred to as a storage medium or a storage device, etc. The memory may be set independently of the processor or integrated with the processor, and this embodiment of the application does not limit this.

[0349] In the embodiment of the present application, the antenna and radio frequency circuit with transceiver functions can be regarded as the receiving unit and sending unit of the terminal (also collectively referred to as the transceiver unit), and the processor with processing function can be regarded as the processing unit of the terminal. As shown in Figure 18, the terminal includes a transceiver unit 1801 and a processing unit 1802. The transceiver unit 1801 can also be called a receiver / transmitter (transmitter), a receiver / transmitter, a receiving / transmitting circuit, etc. The processing unit 1802 can also be called a processor, a processing board, a processing module, a processing device, etc. The transceiver unit 1801 is used to implement the functions of the transceiver unit 1701 in the embodiment shown in Figure 17; the processing unit 1802 is used to implement the functions of the processing unit 1702 in the embodiment shown in Figure 17.

[0350] In the above-mentioned embodiments, the second node can be a network device. Figure 19 shows a simplified schematic diagram of the structure of a network device. The network device includes a radio frequency signal transceiver and converter section and section 192, which in turn includes a transceiver unit 191. The radio frequency signal transceiver and converter section is primarily responsible for receiving and transmitting radio frequency signals and converting radio frequency signals to baseband signals; section 192 is primarily responsible for baseband processing and controlling the network device. Transceiver unit 191 can also be referred to as a receiver / transmitter, a receiver / transmitter, or a receiver / transmitter circuit. Section 192 is typically the control center of the network device and can be commonly referred to as a processing unit, which is responsible for controlling the network device to execute the steps performed by the second node in Figures 11, 14, and 15. For details, please refer to the description of the relevant sections above. Transceiver unit 191 can be used to implement the functions of transceiver unit 1701 in the embodiment shown in Figure 17, while section 192 is used to implement the functions of processing unit 1702 in the embodiment shown in Figure 17.

[0351] Section 192 may include one or more boards, each of which may include one or more processors and one or more memories. The processors are used to read and execute programs in the memories to implement baseband processing functions and control network devices. If multiple boards are present, the boards may be interconnected to increase processing power. As an optional embodiment, multiple boards may share one or more processors, multiple boards may share one or more memories, or multiple boards may simultaneously share one or more processors.

[0352] An embodiment of the present application further provides a computer-readable storage medium, in which a computer program or instruction is stored. When the computer program or instruction is executed, the method in the above embodiment is implemented.

[0353] An embodiment of the present application further provides a computer program product comprising instructions, which, when executed on a computer, enables the computer to execute the method in the above embodiment.

[0354] An embodiment of the present application also provides a communication system, including the above-mentioned communication device.

[0355] It should be noted that the above units or one or more of the units can be implemented by software, hardware or a combination of the two. When any of the above units or units is implemented in software, the software exists in the form of computer program instructions and is stored in a memory, and a processor can be used to execute the program instructions and implement the above method flow. The processor can be built into a system on chip (SoC) or ASIC, or it can be an independent semiconductor chip. In addition to the core used to execute software instructions to perform calculations or processing within the processor, it can further include necessary hardware accelerators, such as field programmable gate arrays (FPGAs), programmable logic devices (PLDs), or logic circuits that implement dedicated logic operations.

[0356] When the above units or units are implemented in hardware, the hardware can be any one or any combination of a CPU, a microprocessor, a digital signal processing (DSP) chip, a microcontroller unit (MCU), an artificial intelligence processor, an ASIC, a SoC, an FPGA, a PLD, a dedicated digital circuit, a hardware accelerator or a non-integrated discrete device, which can run the necessary software or not rely on the software to execute the above method flow.

[0357] Optionally, an embodiment of the present application further provides a chip system, comprising: at least one processor and an interface, wherein the at least one processor is coupled to a memory via the interface, and when the at least one processor executes a computer program or instruction in the memory, the chip system executes the method in any of the above method embodiments. Optionally, the chip system may be composed of a chip, or may include a chip and other discrete devices, which is not specifically limited in the embodiments of the present application.

[0358] It should be understood that in the description of this application, unless otherwise specified, " / " indicates that the objects associated with each other are in an "or" relationship. For example, A / B can mean A or B; where A and B can be singular or plural. Also, in the description of this application, unless otherwise specified, "multiple" means two or more than two. "At least one of the following" or similar expressions refers to any combination of these items, including any combination of single or plural items. For example, at least one of a, b, or c can mean: a, b, c, ab, ac, bc, or abc, where a, b, and c can be single or plural. In addition, to facilitate the clear description of the technical solutions of the embodiments of this application, in the embodiments of this application, words such as "first" and "second" are used to distinguish between identical or similar items with substantially the same functions and effects. Those skilled in the art will understand that words such as "first" and "second" do not limit the quantity or execution order, and words such as "first" and "second" do not necessarily mean different. At the same time, in the embodiments of this application, words such as "exemplary" or "for example" are used to indicate examples, illustrations, or descriptions. Any embodiment or design described as "exemplary" or "for example" in the embodiments of this application should not be interpreted as being more preferred or more advantageous than other embodiments or designs. Rather, the use of words such as "exemplary" or "for example" is intended to present the relevant concepts in a concrete manner to facilitate understanding.

[0359] In the above embodiments, all or part of the embodiments may be implemented by software, hardware, firmware, or any combination thereof. When implemented using a software program, all or part of the embodiments may be implemented in the form of a computer program product. The computer program product includes one or more computer instructions. When the computer program instructions are loaded and executed on a computer, all or part of the processes or functions described in the embodiments of the present application are generated. The computer may be a general-purpose computer, a special-purpose computer, a computer network, or other programmable device. The computer instructions may be stored in a computer-readable storage medium or transmitted from one computer-readable storage medium to another computer-readable storage medium. For example, the computer instructions may be transmitted from one website, computer, server, or data center to another website, computer, server, or data center via a wired (e.g., coaxial cable, optical fiber, digital subscriber line (DSL)) or wireless (e.g., infrared, wireless, microwave, etc.) method. The computer-readable storage medium may be any available medium that a computer can access or a data storage device such as a server or data center that includes one or more media integrated therein. The available medium may be a magnetic medium (eg, a floppy disk, a hard disk, a magnetic tape), an optical medium (eg, a DVD), or a semiconductor medium (eg, a solid state disk (SSD)).

[0360] Although the present application is described herein in conjunction with various embodiments, in the process of implementing the claimed application, those skilled in the art can understand and implement other changes to the disclosed embodiments by reviewing the drawings, the disclosure, and the appended claims. In the claims, the word "comprising" does not exclude other components or steps, and "a" or "an" does not exclude multiple situations. A single processor or other unit can implement several functions listed in the claims. Certain measures are recorded in different dependent claims, but this does not mean that these measures cannot be combined to produce good results.

Claims

1. A communication method, characterized in that: The method comprises: Sending first indication information and / or second indication information, where the first indication information is used to indicate dataset status information, and the second indication information is used to indicate model status information; Receive third indication information, where the third indication information is used to indicate a communication strategy and communication resources, where the communication strategy and the communication resources are determined according to the first indication information and / or the second indication information.

2. The method according to claim 1, characterized in that The data set status information includes at least one of the following information: data type, sampling frequency, sampling location information, sample timestamp, sample quantity, and sample accuracy.

3. The method according to claim 1, characterized in that The model status information includes at least one of the following information: model type, model identification, sub-model order, model version number, model description, model update timestamp, model update location information, model accuracy, and model performance.

4. The method according to any one of claims 1 to 3, characterized in that The communication strategy is determined according to at least two of the first indication information, the second indication information and channel state information.

5. The method according to claim 4, characterized in that The communication strategy includes at least one of the following strategies: quantization strategy, source coding strategy, modulation coding strategy, and source-channel joint coding strategy.

6. The method according to any one of claims 1 to 5, characterized in that The method further comprises: Report the size of the AI ​​service data packet; The communication resources are determined based on at least two of the size of the artificial intelligence service data packet, the data set status information, and the model status information.

7. The method according to any one of claims 1 to 6, characterized in that The method further comprises: Generate an artificial intelligence service data packet according to the third instruction information; According to the third indication information, the artificial intelligence service data packet is sent.

8. The method according to any one of claims 1 to 6, characterized in that The method further comprises: According to the third indication information, the artificial intelligence service data packet is received.

9. The method according to claim 7 or 8, characterized in that The artificial intelligence service data packets are communicated via a physical learning channel.

10. The method according to claim 9, characterized in that Communication via the physical learning channel includes at least one of the following processing operations: quantization, source coding; or Communication via the physical learning channel includes at least one of the following processing operations: quantization, source-channel joint coding; or Communication over the physical learning channel includes the following processing operations: Real number processing.

11. The method according to any one of claims 7 to 10, characterized in that The artificial intelligence service data packet includes a header and a payload, and the payload includes at least one of the following data: parameters or gradients of a machine learning model, inference results, intermediate results of inference to a segmentation layer, gradients passed back to a segmentation layer, and logarithmic scores.

12. The method according to claim 11, characterized in that The packet header includes at least one of the following: content indication information, content description information, compression method, and packaging method.

13. A communication method, characterized in that: The method comprises: receiving first indication information and / or second indication information, wherein the first indication information is used to indicate data set status information, and the second indication information is used to indicate model status information; Determining a communication strategy and communication resources according to the first indication information and / or the second indication information; Send third indication information, where the third indication information is used to indicate the communication strategy and the communication resources.

14. The method according to claim 13, characterized in that The data set status information includes at least one of the following information: data type, sampling frequency, sampling location information, sample timestamp, sample quantity, and sample accuracy.

15. The method according to claim 13, characterized in that The model status information includes at least one of the following information: model type, model identification, sub-model order, model version number, model description, model update timestamp, model update location information, model accuracy, and model performance.

16. The method according to any one of claims 13 to 15, characterized in that The communication strategy is determined according to at least two of the first indication information, the second indication information and channel state information.

17. The method according to claim 16, characterized in that The communication strategy includes at least one of the following strategies: quantization strategy, source coding strategy, modulation coding strategy, and source-channel joint coding strategy.

18. The method according to any one of claims 13 to 17, characterized in that The determining of communication resources according to the first indication information and / or the second indication information includes: Get the size of the AI ​​service data packet; The communication resource is determined according to at least two of the size of the artificial intelligence service data packet, the first indication information, and the second indication information.

19. The method according to claim 18, characterized in that Obtaining the size of the artificial intelligence service data packet includes: Determine the size of the artificial intelligence service data packet based on a correspondence between at least one of the task type and the service type and the data packet size; or Receive the reported size of the artificial intelligence service data packet.

20. The method according to any one of claims 13 to 19, characterized in that The method further comprises: Receive an artificial intelligence service data packet sent according to the third indication information.

21. The method according to any one of claims 13 to 19, wherein: The method further comprises: Generate an artificial intelligence service data packet according to the third instruction information; According to the third indication information, the artificial intelligence service data packet is sent.

22. The method according to claim 20 or 21, characterized in that The artificial intelligence service data packets are communicated via a physical learning channel.

23. The method according to claim 22, characterized in that Communication via the physical learning channel includes at least one of the following processing operations: quantization, source coding; or Communication via the physical learning channel includes at least one of the following processing operations: quantization, source-channel joint coding; or Communication over the physical learning channel includes the following processing operations: Real number processing.

24. The method according to any one of claims 20 to 23, characterized in that The artificial intelligence service data packet includes a header and a payload, and the payload includes at least one of the following data: parameters or gradients of a machine learning model, inference results, intermediate results of inference to a segmentation layer, gradients passed back to a segmentation layer, and logarithmic scores.

25. The method according to claim 24, characterized in that The packet header includes at least one of the following: content indication information, content description information, compression method, and packaging method.

26. A communication method, characterized in that: The method comprises: Acquire artificial intelligence service data generated by the physical PHY layer and / or the media access control MAC layer; The artificial intelligence service data is sent through the first radio bearer.

27. The method according to claim 26, characterized in that The sending the artificial intelligence service data through the first radio bearer includes: Processing the artificial intelligence service data on at least one of the following protocol layers: a packet data convergence protocol (PDCP) layer, a radio link control (RLC) layer, a MAC layer, and a PHY layer; Send processed AI business data.

28. The method according to claim 26 or 27, characterized in that The first radio bearer is a learning radio bearer or a first data radio bearer.

29. The method according to any one of claims 26 to 28, characterized in that The method further comprises: A first signaling is received, where the first signaling is used to instruct configuration of the first radio bearer.

30. A communication method, characterized in that: The method comprises: receiving artificial intelligence service data through the first radio bearer; The artificial intelligence service data is output to the physical PHY layer and / or the media access control MAC layer.

31. The method according to claim 30, wherein Outputting the artificial intelligence service data to the physical PHY layer and / or the media access control MAC layer includes: Processing the artificial intelligence service data on at least one of the following protocol layers: a PHY layer, a MAC layer, a radio link control RLC layer, and a packet data convergence protocol PDCP layer; The processed artificial intelligence service data is transmitted to the PHY layer and / or MAC layer.

32. The method according to claim 30 or 31, characterized in that The first radio bearer is a learning radio bearer or a first data radio bearer.

33. The method according to any one of claims 30 to 32, characterized in that The method further comprises: Sending first signaling, where the first signaling is used to indicate configuration of the first radio bearer.

34. A communication device, characterized in that: The device comprises: a transceiver unit; wherein: The transceiver unit is configured to send first indication information and / or second indication information, wherein the first indication information is used to indicate data set status information, and the second indication information is used to indicate model status information; The transceiver unit is further configured to receive third indication information, where the third indication information is configured to indicate a communication strategy and communication resources, where the communication strategy and the communication resources are determined based on the first indication information and / or the second indication information.

35. The device according to claim 34, characterized in that The data set status information includes at least one of the following information: data type, sampling frequency, sampling location information, sample timestamp, sample quantity, and sample accuracy.

36. The device according to claim 34, characterized in that The model status information includes at least one of the following information: model type, model identification, sub-model order, model version number, model description, model update timestamp, model update location information, model accuracy, and model performance.

37. The device according to any one of claims 34 to 36, characterized in that The communication strategy is determined according to at least two of the first indication information, the second indication information and channel state information.

38. The device according to claim 37, characterized in that The communication strategy includes at least one of the following strategies: quantization strategy, source coding strategy, modulation coding strategy, and source-channel joint coding strategy.

39. The device according to any one of claims 34 to 38, characterized in that The transceiver unit is further configured to report the size of the artificial intelligence service data packet; The communication resources are determined based on at least two of the size of the artificial intelligence service data packet, the data set status information, and the model status information.

40. The device according to any one of claims 34 to 39, characterized in that The device further comprises: a processing unit; The processing unit is configured to generate an artificial intelligence service data packet according to the third indication information; The transceiver unit is further configured to send the artificial intelligence service data packet according to the third indication information.

41. The device according to any one of claims 34 to 39, characterized in that The transceiver unit is further configured to receive the artificial intelligence service data packet according to the third indication information.

42. The device according to claim 40 or 41, characterized in that The artificial intelligence service data packets are communicated via a physical learning channel.

43. The device according to claim 42, characterized in that Communication via the physical learning channel includes at least one of the following processing operations: quantization, source coding; or Communication via the physical learning channel includes at least one of the following processing operations: quantization, source-channel joint coding; or Communication over the physical learning channel includes the following processing operations: Real number processing.

44. The device according to any one of claims 40 to 43, characterized in that The artificial intelligence service data packet includes a header and a payload, and the payload includes at least one of the following data: parameters or gradients of a machine learning model, inference results, intermediate results of inference to a segmentation layer, gradients passed back to a segmentation layer, and logarithmic scores.

45. The device according to claim 44, characterized in that The packet header includes at least one of the following: content indication information, content description information, compression method, and packaging method.

46. ​​A communication device, characterized in that The device comprises: a transceiver unit and a processing unit; wherein: The transceiver unit is configured to receive first indication information and / or second indication information, wherein the first indication information is used to indicate data set status information, and the second indication information is used to indicate model status information; The processing unit is configured to determine a communication strategy and communication resources according to the first indication information and / or the second indication information; The transceiver unit is further configured to send third indication information, where the third indication information is configured to indicate the communication strategy and the communication resources.

47. The device according to claim 46, characterized in that The data set status information includes at least one of the following information: data type, sampling frequency, sampling location information, sample timestamp, sample quantity, and sample accuracy.

48. The device according to claim 46, characterized in that The model status information includes at least one of the following information: model type, model identification, sub-model order, model version number, model description, model update timestamp, model update location information, model accuracy, and model performance.

49. The device according to any one of claims 46 to 48, characterized in that The communication strategy is determined according to at least two of the first indication information, the second indication information and channel state information.

50. The device according to claim 49, characterized in that The communication strategy includes at least one of the following strategies: quantization strategy, source coding strategy, modulation coding strategy, and source-channel joint coding strategy.

51. The device according to any one of claims 46 to 50, characterized in that The processing unit is further configured to obtain the size of the artificial intelligence service data packet; The processing unit is further used to determine the communication resources based on at least two of the size of the artificial intelligence service data packet, the first indication information, and the second indication information.

52. The device according to claim 51, characterized in that The processing unit is further configured to determine the size of the artificial intelligence service data packet based on a correspondence between at least one of the task type and the service type and the data packet size; or The transceiver unit is further configured to receive the reported size of the artificial intelligence service data packet.

53. The device according to any one of claims 46 to 52, characterized in that The transceiver unit is further configured to receive an artificial intelligence service data packet sent according to the third indication information.

54. The device according to any one of claims 46 to 52, characterized in that The processing unit is configured to generate an artificial intelligence service data packet according to the third indication information; The transceiver unit is further configured to send the artificial intelligence service data packet according to the third indication information.

55. The device according to claim 53 or 54, characterized in that The artificial intelligence service data packets are communicated via a physical learning channel.

56. The device according to claim 55, characterized in that Communication via the physical learning channel includes at least one of the following processing operations: quantization, source coding; or Communication via the physical learning channel includes at least one of the following processing operations: quantization, source-channel joint coding; or Communication over the physical learning channel includes the following processing operations: Real number processing.

57. The device according to any one of claims 53 to 56, characterized in that The artificial intelligence service data packet includes a header and a payload, and the payload includes at least one of the following data: parameters or gradients of a machine learning model, inference results, intermediate results of inference to a segmentation layer, gradients passed back to a segmentation layer, and logarithmic scores.

58. The device according to claim 57, characterized in that The packet header includes at least one of the following: content indication information, content description information, compression method, and packaging method.

59. A communication device, characterized in that The device comprises: a processing unit and a transceiver unit; wherein: The processing unit is used to obtain artificial intelligence service data generated by the physical PHY layer and / or the media access control MAC layer; The transceiver unit is configured to send the artificial intelligence service data through a first radio bearer.

60. The device according to claim 59, characterized in that The processing unit is further configured to process the artificial intelligence service data at at least one of the following protocol layers: a packet data convergence protocol (PDCP) layer, a radio link control (RLC) layer, a MAC layer, and a PHY layer; The transceiver unit is also used to send processed artificial intelligence business data.

61. The device according to claim 59 or 60, characterized in that The first radio bearer is a learning radio bearer or a first data radio bearer.

62. The device according to any one of claims 59 to 61, characterized in that The transceiver unit is further used to receive first signaling, where the first signaling is used to instruct configuration of the first radio bearer.

63. A communication device, characterized in that The device comprises: a transceiver unit and a processing unit; wherein: The transceiver unit is configured to receive artificial intelligence service data through the first radio bearer; The processing unit is used to output the artificial intelligence service data to the physical PHY layer and / or the media access control MAC layer.

64. The device according to claim 63, characterized in that The processing unit is further configured to process the artificial intelligence service data at at least one of the following protocol layers: a PHY layer, a MAC layer, a radio link control RLC layer, and a packet data convergence protocol PDCP layer; The processing unit is further configured to transmit the processed artificial intelligence service data to the PHY layer and / or MAC layer.

65. The device according to claim 63 or 64, characterized in that The first radio bearer is a learning radio bearer or a first data radio bearer.

66. The device according to any one of claims 63 to 65, characterized in that The transceiver unit is further used to send a first signaling, where the first signaling is used to instruct configuration of the first radio bearer.

67. A communication system, characterized in that The method comprises the communication device according to any one of claims 34 to 45 and the communication device according to any one of claims 46 to 58.

68. A communication system, characterized in that The method comprises the communication device according to any one of claims 59 to 62 and the communication device according to any one of claims 63 to 66.

69. A communication device, characterized in that The device comprises a processor configured to execute a program stored in a memory, and when the program is executed, the device executes the method according to any one of claims 1 to 33.

70. The device according to claim 69, characterized in that The memory is located external to the device.

71. A communication device, characterized in that The communication device comprises a processor, a memory, and instructions stored in the memory and executed on the processor, wherein when the instructions are executed, the communication device executes the method according to any one of claims 1 to 33.

72. A computer-readable storage medium comprising instructions, which, when executed on a computer, execute the method of any one of claims 1 to 33.

73. A computer program product, characterized in that When the computer is run, the method according to any one of claims 1 to 33 is executed.

Images