Communication method and communication device

Abstract

The invention provides a communication method and a communication device, and the communication method comprises the steps that first equipment obtains at least one first model, the first model can be a bilateral model or a unilateral model, the bilateral model is used for data coding and data decoding, and the unilateral model is used for data coding or data decoding; the first device may determine a second model based on the acquired at least one first model, the second model being a unilateral model. Through the method, the first device can determine the second model based on the obtained at least one first model, and only one AI model for data coding and / or one AI model for data decoding can be reserved in the network, so that the overhead of storing the AI models by the terminal device and / or the network device can be reduced.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of communication, and more particularly, to a communication method and a communication apparatus. BACKGROUND

[0002] In order to optimize air interface transmission, a working mode of a double-sided channel state information (CSI) compression feedback enhanced artificial intelligence (AI) model is proposed. In the working mode, the terminal device side collects CSI, compresses the collected CSI using a pre-trained AI compression model, and then transmits the compressed CSI to the network device side. The network device side decompresses the CSI using a pre-trained AI decompression model, and performs precoding for the downlink based on the decompressed CSI. Regarding the selection of the AI compression model and the AI decompression model, there are currently two schemes: Scheme one, each network device manufacturer can train a double-sided model suitable for the network devices produced by the manufacturer, the double-sided model including an AI compression model on the terminal device side and an AI decompression model on the network device side, and register the trained AI compression model on the terminal device side in the network for downloading by the terminal devices. Each terminal device downloads multiple AI compression models trained by multiple network device manufacturers, and selects the AI compression model trained by the manufacturer of the network device when connecting with the network device. Scheme two, each terminal device manufacturer can train a double-sided model suitable for the terminal devices produced by the manufacturer, the double-sided model including an AI compression model on the terminal device side and an AI decompression model on the network device side, and register the trained AI decompression model on the network device side in the network for downloading by the network devices. Each network device downloads multiple AI decompression models trained by multiple terminal device manufacturers, and selects the AI decompression model trained by the manufacturer of the terminal device when connecting with the terminal device. Since there are many manufacturers of network devices and terminal devices, the terminal devices in the above scheme one need to download and store many AI compression models, and the network devices in the above scheme two need to download and store many AI decompression models, resulting in large storage overhead. SUMMARY

[0003] The present application provides a communication method and a communication apparatus, which can reduce the overhead of AI model storage of terminal devices and network devices.

[0004] In a first aspect, a communication method is provided. The method can be applied to a first device side, i.e., the method can be executed by a first device, or can be executed by a component (such as a chip or a chip system or a circuit or a communication module) of the first device, which is not limited in the present application. Hereinafter, the first device will be mainly taken as an example for description.

[0005] The method can include: the first device obtaining at least one first model, the first model being a bilateral model or a first unilateral model, the bilateral model being used for data encoding and data decoding, and the first unilateral model being used for data encoding or data decoding; the first device determining a second model based on the at least one first model, the second model being used for data encoding or data decoding; and the first device sending first information, the first information being used for indicating the second model.

[0006] Through the above method, the first device can determine a second model based on the obtained at least one first model, and only one AI model used for data encoding and / or one AI model used for data decoding can be reserved in the network, thereby reducing the overhead of the terminal device and / or the network device in storing the AI model.

[0007] In combination with the first aspect, in some implementations of the first aspect, when the first model is the first unilateral model, the first device determining a second model based on the at least one first model includes: the first device determining a second model based on the at least one first model and at least one local model, the second model belonging to the at least one first model, and the local model being a second unilateral model.

[0008] If the first unilateral model is used for data encoding, the second unilateral model is used for data decoding; and if the first unilateral model is used for data decoding, the second unilateral model is used for data decoding.

[0009] Optionally, the first device can further determine a second model and a second model A based on the at least one first model and the at least one local model, the second model belonging to the at least one first model, and the second model A belonging to the at least one local model.

[0010] Through the above method, the first device can select a second model from the at least one first model for data encoding or data decoding, thereby reducing the overhead of the terminal device or the network device in storing the AI model used for data encoding or data decoding.

[0011] In combination with the first aspect, in some implementations of the first aspect, when the first model is the first unilateral model, the method further includes: the first device obtaining at least one third model, the third model being a second unilateral model; the first device determining a second model based on the at least one first model includes: the first device determining a second model and a first fourth model based on the at least one first model and the at least one third model, the second model belonging to the at least one first model, and the fourth model belonging to the at least one third model; and the sending of the first information includes: sending the first information and second information, the first information being used for indicating the second model, and the second information being used for indicating the fourth model.

[0012] If the first one-sided model is used for data encoding and the second one-sided model is used for data decoding, or if the first one-sided model is used for data decoding and the second one-sided model is used for data encoding.

[0013] Optionally, the first device can further determine a second model and a fourth model based on the at least one first model, the at least one third model, and the at least one local model, the at least one local model comprising the at least one first one-sided model and / or the at least one second one-sided model, the second model belonging to the at least one first one-sided model included in the at least one first model or the second model belonging to the at least one second one-sided model included in the at least one local model, and the fourth model belonging to the at least one third model or the fourth model belonging to the at least one second one-sided model included in the at least one local model.

[0014] Through the above method, the first device can select a second model and a fourth model from the at least one first one-sided model and the at least one second one-sided model respectively to complete data encoding and data decoding, thereby reducing the overhead of the terminal device and the network device in storing AI models for data encoding or data decoding.

[0015] In some implementations of the first aspect, the at least one third model is trained based on the at least one first model, or the at least one first model is trained based on the third model.

[0016] For example, if the at least one first model is a first one-sided model for data encoding, a model B in the at least one third model can be a second one-sided model for data decoding trained based on a model A in the at least one first model, or if the at least one first model is a first one-sided model for data decoding, a model B in the at least one third model can be a second one-sided model for data encoding trained based on a model A in the at least one first model.

[0017] For another example, if the at least one third model is a second one-sided model for data encoding, a model A in the at least one first model can be a first one-sided model for data decoding trained based on a model B in the at least one third model, or if the at least one third model is a second one-sided model for data decoding, a model A in the at least one first model can be a first one-sided model for data encoding trained based on a model B in the at least one third model.

[0018] Through the above method, the at least one first model and the at least one third model obtained by the first device can have higher performance in completing data encoding and data decoding, thereby improving the accuracy or model complexity of the second model and the fourth model determined by the first device.

[0019] With reference to the first aspect, in some implementations of the first aspect, when the first model is a first one-sided model, the second model is a second one-sided model.

[0020] If the first one-sided model is used for data encoding and the second one-sided model is used for data decoding, or if the first one-sided model is used for data decoding and the second one-sided model is used for data encoding.

[0021] Through the above method, the first device can determine a second one-sided model based on at least one first one-sided model, which can ensure that the terminal device and the network device in the network only retain one AI model for data encoding or one AI model for data decoding, thereby reducing the overhead of the terminal device and the network device for storing the AI model.

[0022] With reference to the first aspect, in some implementations of the first aspect, the method further includes: the first device sending third information, the third information being used for requesting to update the first model; and the first device obtaining the at least one first model including: the first device obtaining the at least one first model based on the third information, the first model being an updated model.

[0023] For example, the first device can periodically send the third information.

[0024] Through the above method, the first device can trigger real-time updating of the AI model in the network, thereby ensuring the accuracy of data encoding and data decoding.

[0025] With reference to the first aspect, in some implementations of the first aspect, the first information includes at least one of the following:

[0026] The model file, the training data set, the model structure, the loss function type, the training method, the parameter matrix list, the address of the device that can obtain the second model, and the like of the second model.

[0027] Through the above method, various methods for sending the model can be provided, and the flexibility of sending the model can be increased.

[0028] Secondly, a communication method is provided. The method can be applied to the second device side, that is, the method can be executed by the second device, or can be executed by a component (such as a chip or a chip system or a circuit or a communication module) of the second device, and the present application does not limit this. Hereinafter, the second device will be mainly taken as an example for description.

[0029] The method can include: the second device sending fourth information, the fourth information being used to indicate the first model, the first model being a bilateral model or a first unilateral model, the bilateral model being used for data encoding and data decoding, the first unilateral model being used for data encoding or data decoding, the first model being used for the first device to determine a second model, the second model being used for data encoding or data decoding.

[0030] Through the above method, the first device can determine a second model based on the obtained at least one first model, and only one AI model used for data encoding and / or one AI model used for data decoding can be reserved in the network, so that the overhead of the terminal device and / or the network device in storing the AI model can be reduced.

[0031] With reference to the second aspect, in some implementations of the second aspect, when the first model is the first unilateral model, the method further includes: the second device receiving first information, the first information being used to indicate the second model, the second model also being the first unilateral model.

[0032] Through the above method, the first device can select a second model from the at least one first model for data encoding or data decoding, so that the overhead of the terminal device or the network device in storing the AI model used for data encoding or data decoding can be reduced.

[0033] With reference to the second aspect, in some implementations of the second aspect, when the first model is the first unilateral model, the second model is a second unilateral model.

[0034] If the first unilateral model is used for data encoding, the second unilateral model is used for data decoding; if the first unilateral model is used for data decoding, the second unilateral model is used for data decoding.

[0035] Through the above method, the first device can determine a second unilateral model based on the at least one first unilateral model, and it can be ensured that the terminal device and the network device in the network only reserve one AI model used for data encoding or one AI model used for data decoding, so that the overhead of the terminal device and the network device in storing the AI model can be reduced.

[0036] With reference to the second aspect, in some implementations of the second aspect, the method further includes: the second device receiving third information, the third information being used to request updating the first model; and the second device sending the fourth information includes: the second device sending the fourth information based on the third information, the first model indicated by the fourth information being an updated model.

[0037] For example, the first device can periodically send the third information, and correspondingly, the second device can periodically receive the third information.

[0038] Through the method, the first device can trigger real-time updating of the AI model in the network, ensuring accuracy of data encoding and data decoding.

[0039] In some implementations of the second aspect, the fourth information includes at least one of the following:

[0040] The model file, the training data set, the model structure, the loss function type, the training method, the parameter matrix list, and the address of the device where the first model is available.

[0041] Through the method, various methods of sending the model can be provided, and flexibility of sending the model is increased.

[0042] In a third aspect, a communication apparatus is provided. The apparatus includes a transceiver configured to obtain at least one first model, the first model being a bilateral model or a first unilateral model, the bilateral model being used for data encoding and data decoding, and the first unilateral model being used for data encoding or data decoding; and a processor configured to determine a second model based on the at least one first model, the second model being used for data encoding or data decoding, and the transceiver is further configured to send first information, the first information being used to indicate the second model.

[0043] In some implementations of the third aspect, when the first model is the first unilateral model, the processor is configured to determine the second model based on the at least one first model, including determining the second model based on the at least one first model and at least one local model, the second model belonging to the at least one first model, and the local model being a second unilateral model.

[0044] If the first unilateral model is used for data encoding, the second unilateral model is used for data decoding; and if the first unilateral model is used for data decoding, the second unilateral model is used for data decoding.

[0045] Optionally, the processor is further configured to determine the second model and a second model A based on the at least one first model and the at least one local model, the second model belonging to the at least one first model, and the second model A belonging to the at least one local model.

[0046] In some implementations of the third aspect, when the first model is the first one-sided model, the transceiver is further configured to obtain at least one third model, the third model being a second one-sided model; and the processor is configured to determine the second model based on the at least one first model, including: the processor is configured to determine a second model and a first fourth model based on the at least one first model and the at least one third model, the second model belonging to the at least one first model, and the fourth model belonging to the at least one third model; and the transceiver is configured to transmit the first information, including: the transceiver is configured to transmit the first information and second information, the first information being used to indicate the second model, and the second information being used to indicate the fourth model.

[0047] In some implementations of the third aspect, when the first one-sided model is used for data encoding, the second one-sided model is used for data decoding; or when the first one-sided model is used for data decoding, the second one-sided model is used for data encoding.

[0048] Optionally, the processor is further configured to determine the second model and a fourth model based on the at least one first model, the at least one third model, and at least one local model, the at least one local model including the at least one first one-sided model and / or the at least one second one-sided model, the second model belonging to the at least one first one-sided model included in the at least one local model, and the fourth model belonging to the at least one third one-sided model included in the at least one local model.

[0049] In some implementations of the third aspect, the at least one third model is trained based on the at least one first model, or the at least one first model is trained based on the third model.

[0050] In some implementations of the third aspect, when the first model is the first one-sided model, the second model is a second one-sided model.

[0051] In some implementations of the third aspect, when the first one-sided model is used for data encoding, the second one-sided model is used for data decoding; or when the first one-sided model is used for data decoding, the second one-sided model is used for data encoding.

[0052] In some implementations of the third aspect, the transceiver is further configured to transmit third information used to request updating the first model; and the transceiver is configured to obtain the at least one first model, including: the transceiver is configured to obtain the at least one first model based on the third information, the first model being an updated model.

[0053] In some implementations of the third aspect, the first information includes at least one of:

[0054] a model file of the second model, a training data set, a model structure, a loss function type, a training method, a parameter matrix list, an address of a device where the second model is available, and the like.

[0055] The explanation and beneficial effects of the third aspect and any implementation manner of the third aspect can refer to the first aspect, and the third aspect will not be described again.

[0056] In a fourth aspect, a communication apparatus is provided. The apparatus includes a transceiver configured to transmit fourth information, the fourth information being used to indicate a first model, the first model being a bilateral model or a first unilateral model, the bilateral model being used for data encoding and data decoding, the first unilateral model being used for data encoding or data decoding, the first model being used for a first device to determine a second model, the second model being used for data encoding or data decoding.

[0057] With reference to the fourth aspect, in some implementation manners of the fourth aspect, when the first model is the first unilateral model, the transceiver is further configured to receive first information, the first information being used to indicate the second model, the second model also being the first unilateral model.

[0058] With reference to the fourth aspect, in some implementation manners of the fourth aspect, when the first model is the first unilateral model, the second model is a second unilateral model.

[0059] When the first unilateral model is used for data encoding, the second unilateral model is used for data decoding; when the first unilateral model is used for data decoding, the second unilateral model is used for data decoding.

[0060] With reference to the fourth aspect, in some implementation manners of the fourth aspect, the transceiver is further configured to receive third information, the third information being used to request to update the first model; and the transceiver configured to transmit the fourth information includes that the transceiver is configured to transmit the fourth information based on the third information, the fourth information indicating the first model being an updated model.

[0061] With reference to the fourth aspect, in some implementation manners of the fourth aspect, the fourth information includes at least one of the following:

[0062] a model file of the first model, a training data set, a model structure, a loss function type, a training method, a parameter matrix list, an address of a device where the first model is available.

[0063] The explanation and beneficial effects of the fourth aspect and any implementation manner of the fourth aspect can refer to the second aspect, and the fourth aspect will not be described again.

[0064] In a fifth aspect, a communication apparatus is provided. The apparatus is configured to perform the method in the first aspect or any possible implementation of the method in the first aspect, or the apparatus is configured to perform the method in the second aspect or any possible implementation of the method in the second aspect. Specifically, the apparatus can include a unit and / or module configured to perform the method in the first aspect or any possible implementation of the method in the first aspect, or the apparatus can include a unit and / or module configured to perform the method in the second aspect or any possible implementation of the method in the second aspect, such as a processing unit and / or a communication unit.

[0065] In an implementation, the apparatus is a communication device (e.g., the first device, or the second device). When the apparatus is a communication device, the communication unit can be a transceiver, or an input / output interface; the processing unit can be at least one processor. Optionally, the transceiver can be a transceiver circuit. Optionally, the input / output interface can be an input / output circuit.

[0066] In another implementation, the apparatus is a chip, chip system or circuit, or a communication module for a communication device (e.g., the first device, or the second device). When the apparatus is a chip, chip system or circuit for a communication device, the communication unit can be an input / output interface, an interface circuit, an output circuit, an input circuit, a pin or related circuitry, etc. on the chip, chip system or circuit; the processing unit can be at least one processor, a processing circuit or a logic circuit, etc.

[0067] In a sixth aspect, a communication apparatus is provided. The apparatus includes at least one processor configured to execute computer program or instructions to perform the method in the first aspect or any possible implementation of the method in the first aspect. Optionally, the apparatus further includes a memory configured to store the computer program or instructions. Optionally, the apparatus further includes a communication interface coupled to the processor, which can be configured to input the computer program or instructions to the processor, or output information in the processor.

[0068] In an implementation, the apparatus is a communication device (e.g., the first device, or the second device).

[0069] In another implementation, the apparatus is a chip, chip system or circuit, or a communication module for a communication device (e.g., the first device, or the second device).

[0070] In a seventh aspect, a processor is provided. The processor is configured to perform the method in the first aspect or the second aspect.

[0071] For the sending and obtaining / receiving operations involved by the processor, if no special description is made, or if it does not conflict with the actual role or internal logic in the related description, it can be understood as the processor output and receiving, input operations, and also can be understood as the sending and receiving operations performed by the radio frequency circuit and the antenna, and the present application does not limit this.

[0072] Optionally, the apparatus further comprises a memory for storing a program; correspondingly, the at least one processor is configured to execute the computer program or instructions in the memory.

[0073] Optionally, the apparatus further comprises a communication interface. The communication interface is coupled with the processor, and can be used for inputting information to the processor, or outputting information in the processor.

[0074] In an eighth aspect, a computer readable storage medium is provided, which stores program codes for execution by an apparatus, and the program codes comprise codes for executing the method in any possible implementation manner of the first aspect or the second aspect.

[0075] In a ninth aspect, a computer program product containing instructions which, when the computer program product is executed on a computer, cause the computer to execute the method in any possible implementation manner of the first aspect or the second aspect.

[0076] In a tenth aspect, a chip is provided, which comprises a processor and a communication interface. The processor reads instructions on a memory through the communication interface, and executes the method provided by the first aspect or the second aspect in any implementation manner.

[0077] Optionally, the chip is a Modem chip, also known as a baseband chip, or a system on chip (SoC) chip or a system in package (SIP) chip containing a modem core.

[0078] Optionally, as an implementation manner, the chip further comprises a memory, and the memory stores computer programs or instructions. The processor is configured to execute the computer programs or instructions on the memory, and when the computer programs or instructions are executed, the processor is configured to execute the method provided by the first aspect or the second aspect in any implementation manner.

[0079] In an eleventh aspect, a computer program product containing instructions is provided, which, when the computer program product is executed on a computer, causes the computer to execute the method provided by the first aspect or the second aspect in any implementation manner.

[0080] In a twelfth aspect, a communication system is provided, which comprises the first apparatus and / or the second apparatus as described above. BRIEF DESCRIPTION OF DRAWINGS

[0081] Figure 1 A communication architecture diagram applicable to the present application is shown.

[0082] Figure 2 A schematic flow chart of a communication method 200 provided by an embodiment of the present application is shown.

[0083] Figure 3 A schematic flow chart of a communication method 300 provided by an embodiment of the present application is shown.

[0084] Figure 4 A schematic flow chart of a communication method 400 provided by an embodiment of the present application is shown.

[0085] Figure 5 A schematic flow chart of a communication method 500 provided by an embodiment of the present application is shown.

[0086] Figure 6 A schematic block diagram of a communication apparatus 600 provided by an embodiment of the present application is shown.

[0087] Figure 7 A schematic block diagram of another communication apparatus 700 provided by an embodiment of the present application is shown.

[0088] Figure 8 A schematic block diagram of a chip system 800 provided by an embodiment of the present application is shown. DETAILED DESCRIPTION

[0089] The technical solutions in the embodiments of the present application will be described below with reference to the accompanying drawings.

[0090] In order to facilitate understanding of the embodiments of the present application, the following points are explained before the embodiments of the present application are introduced.

[0091] (1) In the present application, “for indicating” or “indicating” can include for directly indicating and for indirectly indicating, or in other words, “for indicating” or “indicating” can explicitly and / or implicitly indicate. For example, when describing that a certain information is for indicating information I, it can include that the information directly indicates I or indirectly indicates I, and it does not mean that I must be carried in the information.

[0092] (2) In the embodiments shown below, the first, second, third, fourth, and various numbers are only for distinguishing and do not limit the scope of the embodiments of the present application. For example, different messages are distinguished.

[0093] (3) In the embodiments of the present application, the words "example", "for example", "for instance", "e.g.", and "as an example" are used to indicate that the item being discussed is an example, instance, or illustration. Any embodiment or design solution described in the present application as "example" should not be interpreted as being more preferred or advantageous than other embodiments or design solutions. Rather, the use of the word "example" is intended to present concepts in a concrete manner.

[0094] (4) The terms "comprise", "comprising", "have", "having", and "include" and their variants mean "including but not limited to", unless otherwise expressly specified.

[0095] (5) In the embodiments of the present application, the description related to A sending a message, information or data to B, and B receiving the message, information or data from A, is intended to indicate which object the message, information or data is intended to send, and does not limit whether it is directly sent or indirectly sent via other nodes.

[0096] The technical solutions of the embodiments of the present application can be applied to various communication systems, including but not limited to: a 5th generation (5G) system or a new radio (NR) system, a long term evolution (LTE) system, a long term evolution-advanced (LTE-A) system, a wireless local area network (WLAN) system, a satellite communication system, an optical communication system, a microwave communication system, etc. It can also be applied to future communication systems, such as future communication networks, fusion systems of multiple systems, etc. In addition, it can also be applied to device to device (D2D) communication, vehicle-to-everything (V2X) communication, machine to machine (M2M) communication, machine type communication (MTC), and internet of things (IoT) communication systems or other communication systems. In addition, it can also be extended to similar wireless communication systems, such as wireless-fidelity (Wi-Fi), worldwide interoperability for microwave access (WIMAX), and 3rd generation partnership project (3GPP) related communication systems, etc., without limitation.

[0097] Next, a communication architecture applicable to embodiments of the present application is described.

[0098] Figure 1 is a schematic diagram of a communication architecture 100 applicable to embodiments of the present application. As shown in Figure 1 , the communication architecture 100 can include a network management system (NMS) or service management and orchestration function (SMO), an element management system (EMS), over the top (OTT), network devices, and terminal devices.

[0099] Specifically, the NMS is responsible for the operation, management, and maintenance functions of the network. It can also be referred to as a cross-domain management system. In the 3rd Generation Partnership Project (3GPP) network domain, the NMS directly manages the EMS.

[0100] The role of the SMO in the network architecture is similar to that of the NMS, and it is responsible for the operation, management, and maintenance of various network services and orchestration functions. In the open radio access network (ORAN) domain, the network elements directly managed by the SMO can be heterogeneous, such as EMS, network devices, network data analytics function (NWDAF), etc.

[0101] OTT refers to the manufacturer's server of network devices or the manufacturer's server of terminal devices. AI models or data information in the network can be opened to the manufacturer's server of terminal devices OTT through the NMS / SMO, and the manufacturer's server of terminal devices OTT can interact with the NMS / SMO. AI models or data information of terminal devices can interact with the manufacturer's server of terminal devices OTT.

[0102] The EMS is used to manage one or more network elements of a certain category. It can also be referred to as a domain management system or a single-domain management system. In the radio access network (RAN) domain, the EMS can comprehensively manage a single network device.

[0103] The network device is a device with wireless transceiving function, which is used for communication with terminal devices. The network device can be a node in the RAN, which can be referred to as a base station, and can also be referred to as a RAN node. It can be an evolved Node B (eNB or eNodeB) of long term evolution (LTE), a base station of 5G network such as gNodeB (gNB), or a base station in a public land mobile network (PLMN) evolved after 5G, a broadband network service gateway (BNG), a convergence switch, or a 3GPP access device, etc.

[0104] The above RAN can be configured as a RAN defined in 3GPP protocol, an O-RAN, or a cloud radio access network (C-RAN), etc. The network device can also include various forms of base stations, such as macro base stations, micro base stations (also known as small stations), relay stations, transmission reception points (TRPs), transmitting points (TPs), mobile switching centers, and devices that assume the function of a base station in device-to-device (D2D) communication, vehicle-to-everything (V2X) communication, machine-to-machine (M2M) communication, network devices in non-terrestrial networks (NTN), etc., without specific limitation.

[0105] The network device can also be a network element or module capable of implementing part of the functions of a base station, for example, the network device can be one or more of the following: a centralized unit (CU), a distributed unit (DU), or a radio unit (RU). The functions of the CU and the DU can be implemented by different network elements, or by the baseband unit (BBU) of the base station at the same time. Optionally, the CU can be further separated into a CU-control plane (CP) and a CU-user plane (UP). The functions of the RU can be implemented by the radio frequency device of the base station, for example, the radio frequency device of the base station can be a remote radio unit (RRU), a pico remote radio unit (pRRU), an active antenna unit (AAU), or other units, modules or devices with radio frequency processing functions, etc. The communication interface protocol between the BBU and the radio frequency device can be a common public radio interface (CPRI) interface protocol, an enhanced common public radio interface (eCPRI) interface protocol, or a fronthaul interface protocol between the DU and the RU in the O-RAN system, etc., without limitation.

[0106] The communication device for implementing the functions of the network device can be a network device, or a device capable of supporting the network device to implement the functions, such as a chip system. The device can be installed in the network device or used with the network device. The chip system in the embodiments of the present application can be composed of a chip, or can include a chip and other discrete devices.

[0107] The terminal device is a device with wireless transceiver function, which can be referred to as user equipment (UE), access terminal, subscriber unit, user station, mobile station, remote station, remote terminal, mobile device, user terminal, wireless communication device, user agent or user equipment. The terminal device can also be a satellite phone, a cellular phone, a smartphone, a wireless data card, a wireless modem, a machine type communication device, a cordless phone, a session initiation protocol (SIP) phone, a wireless local loop (WLL) station, a personal digital assistant (PDA), a customer-premises equipment (CPE), a smart point of sale (POS) machine, a handheld device with wireless communication function, a computing device or other processing device connected to a wireless modem, a vehicle-mounted device, a communication device carried on an aerial vehicle, a wearable device, a drone, a robot, a terminal in device-to-device (D2D) communication, a terminal in vehicle-to-everything (V2X) communication, a virtual reality (VR) terminal device, an augmented reality (AR) terminal device, a wireless terminal in industrial control, a wireless terminal in self driving, a wireless terminal in remote medical, a wireless terminal in smart grid, a wireless terminal in transportation safety, a wireless terminal in smart city, a wireless terminal in smart home, or a terminal device in a communication network evolved after 5G, etc. The present application does not make any limitation.

[0108] The communication device for implementing the function of the terminal device can be a terminal device, or a device capable of supporting the terminal device to implement the function, such as a chip system. The device can be installed in the terminal device or used with the terminal device. In the present application, the chip system can be composed of a chip, or can include a chip and other discrete devices.

[0109] It should be noted that the communication architecture 100 described above can include any number of EMSs, any number of OTT servers, any number of network devices, or any number of terminal devices, and the present application is not limited in this regard.

[0110] It should be noted that the communication architecture 100 described above is only an example, and the communication architecture applicable to the present application is not limited thereto. Any communication architecture capable of realizing the functions of the above-mentioned network elements is applicable to the present application. In other words, the communication architecture described in the present application is intended to more clearly illustrate the technical solutions of the present application and does not constitute a limitation on the technical solutions provided by the present application. Those skilled in the art can know that, as the system architecture evolves or new business scenarios emerge, the technical solutions provided by the present application are also applicable to similar technical problems.

[0111] In addition, the names of the network elements, the names of the messages / information, etc. involved in the present application are only examples. In future communication networks, these network elements, messages / information can also use other names, as long as these network elements, messages / information have the same or similar functions as the network elements, messages / information described in the present application and achieve the same or similar technical purposes, they should belong to the technical scope covered by the present application. For example, in future communication networks, part or all of the names of the above-mentioned network elements can be used as in 4G / 5G, or new names can be used.

[0112] In order to optimize air interface transmission, a working mode of a bilateral CSI compression feedback enhanced AI model is proposed. In the working mode, the terminal device side collects CSI, compresses the collected CSI using a pre-trained AI compression model, and then transmits the compressed CSI to the network device side. The network device side decompresses the CSI using a pre-trained AI decompression model, and performs precoding on the downlink based on the decompressed CSI. Regarding the selection of the AI compression model and the AI decompression model, there are currently two schemes: Scheme one, each network device manufacturer can train a bilateral model suitable for the network devices produced by the manufacturer, the bilateral model including an AI compression model on the terminal device side and an AI decompression model on the network device side, and register the trained AI compression model on the terminal device side in the network for terminal devices to download; each terminal device downloads multiple AI compression models trained by multiple network device manufacturers, and selects the AI compression model trained by the manufacturer of the network device when connecting with the network device. Scheme two, each terminal device manufacturer can train a bilateral model suitable for the terminal devices produced by the manufacturer, the bilateral model including an AI compression model on the terminal device side and an AI decompression model on the network device side, and register the trained AI decompression model on the network device side in the network for network devices to download; each network device downloads multiple AI decompression models trained by multiple terminal device manufacturers, and selects the AI decompression model trained by the manufacturer of the terminal device when connecting with the terminal device. Since there are many network device manufacturers and terminal device manufacturers, the terminal device in the above scheme one needs to download and store many AI compression models, and the network device in the above scheme two needs to download and store many AI decompression models, resulting in large storage overhead.

[0113] Based on the above technical problems, the present application provides a communication 200 which can reduce the overhead of the terminal device and the network device in storing AI models.

[0114] Figure 2 is a schematic flowchart of a communication method 200 provided by an embodiment of the present application. In this embodiment, the first device and the second device are taken as the interactive execution subject to illustrate the method, but the present application does not limit the interactive execution subject. For example, Figure 2 The first device in the above embodiment can also be a chip, a chip system, or a processor supporting the method that the first device can implement, and can also be a logic module or software capable of implementing all or part of the first device; the second device can also be a chip, a chip system, or a processor supporting the method that the second device can implement, and can also be a logic module or software capable of implementing all or part of the second device.

[0115] The communication method 200 can include the following steps:

[0116] S210, the second device sends fourth information to the first device, the fourth information being used for indicating the first model. Correspondingly, the first device receives the fourth information.

[0117] Specifically, the first model is a bilateral model or a first unilateral model, the bilateral model being used for data encoding and data decoding, and the first unilateral model being used for data encoding or data decoding.

[0118] The various models of the embodiments of the present application are AI models used for information transmission and respectively deployed on two sides of a communication device. The various models involved in the embodiments of the present application are described by taking data encoding and / or data decoding as examples. The various models of the embodiments of the present application can also be used for data compression and / or data decompression, etc., which is not limited in the present application.

[0119] Exemplarily, the first device can be a registration center for model registration, for example, the first device can be an NMS.

[0120] Exemplarily, the second device can be a vendor server of a terminal device or the second device can be an EMS for managing a single network device, which is not limited in the present application.

[0121] If the second device is a vendor server of a terminal device, the second device can also be replaced by a network function, a network element or a device related to the vendor of the terminal device, etc., which is described by taking the vendor server of the terminal device as an example in the embodiments of the present application. If the second device is an EMS for managing a single network device, the second device can also be replaced by a network function, a network element or a device related to the vendor of the network device, etc., which is described by taking the EMS for managing a single network device as an example in the embodiments of the present application, which is not limited in the present application.

[0122] Exemplarily, the fourth information is used for indicating an acquisition manner of the first model, and the fourth information can include at least one of the following:

[0123] A model file of the first model, a training data set, a model structure, a loss function type, a training method, a parameter matrix list, an address of a device where the first model can be acquired, etc.

[0124] S210 is described by taking one second device as an example. There can be at least one second device, and each of the at least one second device can indicate the first model to the first device.

[0125] S212, the first device acquires at least one first model.

[0126] S214, the first device determines a second model based on the at least one first model.

[0127] The first model and the second model can be divided into the following three possible cases:

[0128] Case 1: the first model is a bilateral model, and the second model is a first unilateral model or a second unilateral model.

[0129] Specifically, the first unilateral model can be used for data encoding or data decoding. If the first unilateral model is used for data encoding, the second unilateral model is used for data decoding; if the first unilateral model is used for data decoding, the second unilateral model is used for data encoding.

[0130] Case 2: the first model is a first unilateral model, and the second model is a first unilateral model.

[0131] In one implementation of Case 2, the first device determines the second model based on at least one first model and at least one local model, the at least one local model being a second unilateral model, and the second model belonging to the at least one first model.

[0132] Specifically, the first unilateral model can be used for data encoding or data decoding. If the first unilateral model is used for data encoding, the second unilateral model is used for data decoding; if the first unilateral model is used for data decoding, the second unilateral model is used for data encoding.

[0133] Case 3: the first model is a first unilateral model, and the second model is a second unilateral model.

[0134] Specifically, the first unilateral model can be used for data encoding or data decoding. If the first unilateral model is used for data encoding, the second unilateral model is used for data decoding; if the first unilateral model is used for data decoding, the second unilateral model is used for data encoding.

[0135] S216, the first device sends first information, the first information being used for indicating the second model.

[0136] Exemplarily, the first information is used for indicating an acquisition manner of the second model, and the first information can include at least one of the following:

[0137] A model file, a training data set, a model structure, a loss function type, a training method, a parameter matrix list, an address of a device where the second model can be acquired, and the like of the second model.

[0138] For Case 1 and Case 2 described above, the first device sends the first information to the second device; for Case 3 described above, the first device sends the first information to the third device.

[0139] In an example, if the second device is a vendor server of the terminal device and the third device is an EMS for managing a single network device, for the above-mentioned case 1 and case 2, the second device can deploy the second model indicated by the first device in the terminal device of the vendor to perform data encoding or data decoding by the terminal device; for the above-mentioned case 3, the third device can deploy the second model indicated by the first device in the network device to perform data encoding or data decoding by the network device.

[0140] In another example, if the second device is an EMS for managing a single network device and the third device is a vendor server of the terminal device, for the above-mentioned case 1 and case 2, the second device can deploy the second model indicated by the first device in the network device to perform data encoding or data decoding by the network device; for the above-mentioned case 3, the third device can deploy the second model indicated by the first device in the terminal device of the vendor to perform data encoding or data decoding by the terminal device.

[0141] Through the above-mentioned communication method 200, only one AI model can be reserved in the terminal device side and the network device side in the network, and the overhead of storing the AI model by the terminal device and the network device can be reduced.

[0142] The embodiments below respectively describe the technical solutions applicable to the above-mentioned case 1, case 2 and case 3 in detail. The embodiments below are exemplarily described with the first device as the NMS, the second device as the OTT#1 and the third device as the EMS#1, but it should be understood that the first device can also be the NMS, the second device can also be the EMS#1 and the third device can also be the OTT#1, which is not limited in the present application.

[0143] The communication method 300 shown below is an example of the technical solution applicable to the above-mentioned case 1, Figure 3 is an example of the flowchart of the communication method 300.

[0144] The communication method 300 can include the following steps:

[0145] S310, the OTT#1 performs model training based on the first data set to obtain a first model.

[0146] Specifically, the first model is a bilateral model, which can be used for data encoding and data decoding.

[0147] Specifically, the data included in the first data set corresponds to the function of the first model, that is, the data of the first data set is collected according to the function of the first model to be trained. Exemplarily, if the first model is a model for performing CSI compression or CSI decompression, the first data set can include original channel data collected from the terminal device and the network device, for example, original channel time-frequency resource matrix, etc.

[0148] The communication method 300 is described by taking an OTT as an example. There can be other OTTs. The other OTTs also perform receiving actions or sending actions similar to OTT #1, which are not described herein.

[0149] S312, OTT #1 sends fourth information to the NMS, the fourth information being used to indicate the first model. Accordingly, the NMS receives the fourth information.

[0150] Exemplarily, the fourth information can include at least one of the following:

[0151] The model file of the first model, the training data set, the model structure, the loss function type, the training method, the parameter matrix list, the address of the device where the first model can be obtained, etc.

[0152] Exemplarily, OTT #1 can indicate the first model to the NMS by the following four methods:

[0153] 1) Send the model file of the first model. For example, send the model file in the format of.bin,.pt,.json,.onnx, etc.

[0154] 2) Send the training data set, the model structure, the loss function type, and the training method of the first model. For example, the training data set of the first model can be the first data set described above, the loss function type can be the function expression of the loss function formula, the parameter value, etc., and the training method can be the optimizer type (Adam, SGD, etc.) of the training.

[0155] 3) Send the model structure of the first model and the parameter indication information. For example, the parameter indication information can be a parameter matrix list.

[0156] 4) Send the address of the device where the first model can be obtained.

[0157] S314, the NMS obtains at least one first model.

[0158] S316, the NMS determines a second model and a fourth model based on the at least one first model.

[0159] Specifically, the second model is a first one-sided model, and the fourth model is a second one-sided model. Exemplarily, if the first one-sided model is used for data encoding, the second one-sided model is used for data decoding; if the first one-sided model is used for data decoding, the second one-sided model is used for data encoding. Exemplarily, the first device performs simulation testing on each of the at least one first model, determines the accuracy rate of each of the at least one first model for the original channel recovery, and selects a first model with the best original channel recovery accuracy rate, and trains a second model and a fourth model according to the first model.

[0160] Optionally, the NMS can also determine a second model and a fourth model based on the at least one first model and the local bilateral model.

[0161] S318, the NMS sends first information to the OTT#1, the first information being used to indicate the second model. Accordingly, the OTT#1 receives the first information.

[0162] Exemplarily, the first information can include at least one of the following:

[0163] The model file of the second model, the training data set of the second model, the model structure of the second model, the loss function type of the second model, the training method of the second model, the parameter matrix list of the second model, the address of the device where the second model can be obtained, etc.

[0164] Exemplarily, the NMS can indicate the second model to the OTT#1 by the following four methods:

[0165] 1) sending the model file of the second model. For example, sending the model file in the format of.bin,.pt,.json,.onnx, etc.

[0166] 2) sending the training data set, the model structure, the loss function type, and the training method of the second model. For example, the loss function type can be the function expression of the loss function formula, the parameter value, etc., and the training method can be the optimizer type (Adam, SGD, etc.) of the training.

[0167] 3) sending the model structure of the second model and the parameter indication information. For example, the parameter indication information can be the parameter matrix list.

[0168] 4) sending the address of the device where the second model can be obtained.

[0169] S320, the OTT#1 deploys the second model to the terminal device of the manufacturer.

[0170] If the NMS indicates the second model to the OTT#1 by the method 1) or the method 4) shown in S318, the OTT#1 can directly obtain the second model; if the NMS indicates the second model to the OTT#1 by the method 2) or the method 3) shown in S318, the OTT#1 needs to perform model training based on the training data set, the model structure, the loss function type, the training method, and the parameter indication information of the second model to obtain the second model, and the OTT#1 can also perform model optimization, acceleration, and adaptation (for example, customizing application programming interfaces (APIs)) of the second model according to the hardware of the manufacturer, so that the second model can run in the chip and the operating system of the terminal device of the manufacturer.

[0171] Exemplarily, the OTT#1 can deploy the second model to the terminal device of the manufacturer through over the air (OTA) technology.

[0172] S322, the NMS sends second information to the EMS#1, the second information being used for indicating the fourth model. Correspondingly, the EMS#1 receives the second information.

[0173] Exemplarily, the second information can include at least one of the following:

[0174] The model file of the fourth model, the training data set, the model structure, the loss function type, the training method, the parameter matrix list, the address of the device where the fourth model can be acquired, etc.

[0175] Exemplarily, the NMS can indicate the fourth model to the EMS#1 through the following four methods:

[0176] 1) The model file of the fourth model is sent. For example, the model file in the format of.bin,.pt,.json,.onnx, etc. is sent.

[0177] 2) The training data set, the model structure, the loss function type, and the training method of the fourth model are sent. For example, the loss function type can be the function expression of the loss function formula, the parameter value, etc., and the training method can be the optimizer type (Adam, SGD, etc.) of the training.

[0178] 3) The model structure of the fourth model and the parameter indication information are sent. For example, the parameter indication information can be the parameter matrix list.

[0179] 4) The address of the device where the fourth model can be acquired is sent.

[0180] S324, the EMS#1 deploys the fourth model to the network device managed.

[0181] If the NMS indicates the fourth model to the EMS#1 through the method 1) or the method 4) shown in S322 above, the EMS#1 can directly acquire the fourth model; if the NMS indicates the fourth model to the EMS#1 through the method 2) or the method 3) shown in S322 above, the EMS#1 needs to perform model training based on the training data set, the model structure, the loss function type, the training method, and the parameter indication information of the fourth model to obtain the fourth model, and the EMS#1 can also perform model optimization, acceleration, and adaptation (for example, custom API) of the fourth model according to the hardware of the manufacturer of the network device, so that the model#2 can run in the chip and the operating system of the network device managed.

[0182] Through the communication method 300, the NMS can determine a set of AI models for data encoding and data decoding for the entire network, avoiding the terminal device and the network device from storing more AI models, and saving storage overhead.

[0183] The communication method 400 shown below is an example of a technical solution applicable to the above case 2, Figure 4 is an example of an exemplary flowchart of the communication method 400.

[0184] The communication method 400 can include the following steps:

[0185] S410, the OTT#1 performs model training based on the first data set to obtain a first model#1.

[0186] Specifically, the first model#1 is a first one-sided model, which can be used for data encoding or data decoding.

[0187] Specifically, the data included in the first data set corresponds to the function of the first model#1, that is, the data of the first data set is collected according to the function of the first model#1 to be trained. For example, if the first model#1 is a model for CSI compression or CSI decompression, the first data set can include original channel data collected from the terminal device and the network device, such as an original channel time-frequency resource matrix.

[0188] The communication method 400 is described by taking the manufacturer OTT of a terminal device as an example, and there can be other OTTs. Other OTTs also perform similar receiving actions or sending actions as the OTT#1, which are not described here.

[0189] Specifically, the OTT#1 can also assign an identification information#1 to the first model#1 obtained by training, and the identification information#1 includes the identification of the OTT#1 and the identification#1.

[0190] S412, the OTT#1 indicates the first model#1 to the EMS#1.

[0191] For example, the OTT#1 can indicate the first model#1 to the EMS#1 by the following four methods:

[0192] 1) Send the model file of the first model#1. For example, send the model file in the format of.bin,.pt,.json,.onnx, etc.

[0193] 2) Send the training data set, model structure, loss function type, and training method of the first model#1. For example, the training data set of the first model#1 can be the first data set described above, the loss function type can be the function expression, parameter value, etc. of the loss function formula, and the training method can be the optimizer type (Adam, SGD, etc.) of training.

[0194] 3) Send the model structure of the first model #1 and parameter indication information, for example, the parameter indication information can be a parameter matrix list.

[0195] 4) Send the address of the device that can obtain the first model #1.

[0196] The communication method 400 is described by taking an EMS for managing network devices as an example, and other EMSs can exist. Other EMSs also perform receiving actions or sending actions similar to EMS #1, which are not described here.

[0197] S414, the EMS #1 performs model training based on the at least one first model to obtain at least one third model.

[0198] Specifically, the third model is a second one-sided model. If the first one-sided model is used for data encoding, the second one-sided model is used for data decoding; if the first one-sided model is used for data decoding, the second one-sided model is used for data encoding.

[0199] Exemplarily, the EMS #1 performs model training based on the first model #1 from the OTT #1 to obtain a third model #1. The EMS #1 can assign identification information #2 to the third model #1, and the identification information #2 includes the identification of the OTT #1 and the identification #2. The identification of the OTT #1 can be considered as a unique virtual identification of the vendor OTT #1 of the terminal device, and the unique virtual identification of the OTT #1 can be assigned by the NMS.

[0200] Similarly, if there is another OTT (for example, OTT #2), the OTT #2 performs model training in the above S410 to obtain a first model #2, and assigns identification information #3 to the second model #2, and the identification information #3 includes the identification of the OTT #2 and the identification #3. The EMS #1 can perform model training based on the first model #2 from the OTT #2 to obtain a third model #2. The EMS #1 can assign identification information #4 to the third model #2, and the identification information #4 includes the identification of the OTT #2 and the identification #4.

[0201] Since the first model #1 and the third model #1 both include the identification of the OTT #1, it can be considered that the first model #1 and the third model #1 have a corresponding relationship, and it can also be understood that the first model #1 and the third model #1 can be used as a group of models for simulation testing in the following S428. Similarly, since the first model #2 and the third model #2 both include the identification of the OTT #2, it can be considered that the first model #2 and the third model #2 have a corresponding relationship, and it can also be understood that the first model #2 and the third model #2 can be used as a group of models for simulation testing in the following S428.

[0202] The S410-S414 shown above is a way for the OTT#1 to obtain at least one first model and for the EMS#1 to obtain at least one third model, the at least one third model in the way being trained based on the at least one first model.

[0203] The S416-S420 shown below is a way for the OTT#1 to obtain at least one first model and for the EMS#1 to obtain at least one third model, the at least one first model in the way being trained based on the at least one third model.

[0204] The S416, the EMS#1 performs model training based on the second data set to obtain a third model#1.

[0205] Specifically, the third model#1 is a second one-sided model, which can be used for data encoding or data decoding.

[0206] Specifically, the data included in the second data set corresponds to the function of the third model#1, that is, the data of the second data set is collected according to the function of the third model#1 to be trained. For example, if the third model#1 is a model for performing CSI compression or CSI decompression, the second data set can include original channel data collected from terminal devices and network devices, such as original channel time-frequency resource matrices, etc.

[0207] The communication method 400 is described by taking one EMS as an example, and there can be other EMSs. Other EMSs also perform similar receiving actions or sending actions as the EMS#1, which are not described here.

[0208] Specifically, the EMS#1 can also assign an identification information#1 to the trained third model#1, the identification information#1 including an identification of the EMS#1 and the identification#1.

[0209] The S418, the EMS#1 indicates the third model#1 to the OTT#1.

[0210] For example, the EMS#1 can indicate the third model#1 to the OTT#1 by the following four methods:

[0211] 1) Send a model file of the third model#1. For example, send a model file in.bin,.pt,.json,.onnx, etc.

[0212] 2) Send a training data set, model structure, loss function type, and training method of the third model#1. For example, the training data set of the third model#1 can be the second data set described above, the loss function type can be a function expression, parameter value, etc. of a loss function formula, and the training method can be an optimizer type (Adam, SGD, etc.) of training.

[0213] 3) send the model structure of the third model #1 and parameter indication information, for example, the parameter indication information can be a parameter matrix list.

[0214] 4) send the address of the device that the third model #1 can obtain.

[0215] The communication method 400 is described by taking one OTT as an example, and there can be other OTTs. The other OTTs also perform receiving actions or sending actions similar to OTT #1, which are not described here.

[0216] S420, OTT #1 performs model training based on at least one third model to obtain at least one first model.

[0217] Specifically, the first model is a first one-sided model. If the second one-sided model is used for data encoding, the first one-sided model is used for data decoding; if the second one-sided model is used for data decoding, the first one-sided model is used for data encoding.

[0218] Exemplarily, OTT #1 performs model training based on the third model #1 from EMS #1 to obtain a first model #1. OTT #1 can assign identification information #2 to the first model #1, and the identification information #2 includes the identification of EMS #1 and identification #2. The identification of EMS #1 can be considered as a unique virtual identification (Vendor virtual identification) of EMS #1 for managing a single network device, and the unique virtual identification (Vendor virtual identification) of EMS #1 can be assigned by NMS.

[0219] Similarly, if there is another EMS (for example, EMS #2), EMS #2 performs model training to obtain a third model #2 at S416, and assigns identification information #3 to the third model #2, and the identification information #3 includes the identification of EMS #2 and identification #3. OTT #1 can perform model training based on the third model #2 from EMS #2 to obtain a first model #2. OTT #1 can assign identification information #4 to the first model #2, and the identification information #4 includes the identification of EMS #2 and identification #4.

[0220] Since the third model #1 and the first model #1 both include the identification of EMS #1, it can be considered that the third model #1 and the first model #1 have a corresponding relationship, and it can also be understood that the third model #1 and the first model #1 can be used as a group of models for simulation test at S428. Similarly, since the third model #2 and the first model #2 both include the identification of EMS #2, it can be considered that the third model #2 and the first model #2 have a corresponding relationship, and it can also be understood that the third model #2 and the first model #2 can be used as a group of models for simulation test at S428. ​

[0221] S422, the OTT#1 indicates at least one first model to the NMS.

[0222] Exemplarily, in the above-mentioned manner one, the OTT#1 obtains the first model#1, and indicates the first model#1 to the NMS; in the above-mentioned manner two, the OTT#1 obtains the first model#1 and the first model#2, and indicates the first model#1 and the first model#2 to the NMS.

[0223] Exemplarily, the method that the OTT#1 indicates at least one first model to the NMS can refer to the four methods that the OTT#1 indicates the first model#1 to the EMS#1 in the above-mentioned S412, which will not be repeated here.

[0224] S424, the EMS#1 indicates at least one third model to the NMS.

[0225] Exemplarily, in the above-mentioned manner one, the EMS#1 obtains the third model#1 and the third model#2, and sends the third model#1 and the third model#2 to the NMS; in the above-mentioned manner two, the EMS#1 obtains the third model#1, and sends the third model#1 to the NMS.

[0226] Exemplarily, the method that the EMS#1 indicates at least one third model to the NMS can refer to the four methods that the EMS#1 indicates the third model#1 to the OTT#1 in the above-mentioned S418, which will not be repeated here.

[0227] The process that each OTT indicates a model to the NMS related by the embodiments of the present application can be considered as the process that each OTT registers a model to the NMS, and the process that each EMS indicates a model to the NMS can be considered as the process that each EMS registers a model to the NMS.

[0228] S426, the NMS obtains at least one first model and at least one third model.

[0229] S428, the NMS determines a second model and a fourth model based on at least one first model and at least one third model.

[0230] Specifically, at least one first model and at least one third model have a corresponding relationship, specifically: one of at least one first model has a corresponding relationship with multiple third models in at least one third model, or one of at least one third model has a corresponding relationship with multiple first models in at least one first model.

[0231] The at least one first model and the at least one third model can be grouped into multiple groups of models according to the correspondence. Taking the at least one first model and the at least one third model obtained in the above manner as an example, the first model #1 trained by the OTT #1 and the third model #1 trained by the EMS #1 based on the first model #1 are a first group of models, the second model #2 trained by the OTT #2 and the third model #2 trained by the EMS #1 based on the second model #2 are a second group of models, and so on.

[0232] The NMS performs simulation testing on each group of models in the multiple groups of models, determines at least one of multiple performance indicators of each group of models in the multiple groups of models, such as a compression rate, a decompression accuracy, a time required for a single calculation, a calculation efficiency, and the like, and selects a group of models with the best comprehensive performance. For example, the NMS determines that the first group of models has the best comprehensive performance indicator through simulation testing, the NMS determines the first model #1 in the first group of models as a second model, and determines the third model #1 in the first group of models as a fourth model. That is, the second model belongs to the at least one first model, and the fourth model belongs to the at least one third model. The evaluation indicator of the comprehensive performance can be one or more of the compression rate, the decompression accuracy, the time required for a single calculation, and the calculation efficiency. If the evaluation indicator of the comprehensive performance is multiple of the compression rate, the decompression accuracy, the time required for a single calculation, and the calculation efficiency, the comprehensive performance can be determined by weighted summation.

[0233] Optionally, the NMS can also determine a second model and a fourth model based on the at least one first model, the at least one third model, and the local model. At this time, the second model does not necessarily belong to the at least one first model, and the second model can be a local model. The fourth model also does not necessarily belong to the at least one third model, and the fourth model can also be a local model.

[0234] S430, the NMS sends first information to the OTT #1, and the first information is used to indicate the second model. Correspondingly, the OTT #1 receives the first information.

[0235] Exemplarily, the first information can include at least one of the following:

[0236] A model file of the second model, a training data set, a model structure, a loss function type, a training method, a parameter matrix list, an address of a device where the second model can be obtained, and the like.

[0237] Exemplarily, the NMS can indicate the second model to the OTT #1 by the following four methods:

[0238] 1) Send a model file of the second model. For example, send a model file in.bin,.pt,.json,.onnx, or the like.

[0239] 2) Send the training data set, model structure, loss function type, training method of the second model. For example, the loss function type can be the function expression of the loss function formula, parameter value, etc., and the training method can be the optimizer type (Adam, SGD, etc.) of the training.

[0240] 3) Send the model structure and parameter indication information of the second model. For example, the parameter indication information can be a parameter matrix list.

[0241] 4) Send the address of the device that can obtain the second model.

[0242] S432, OTT#1 deploys the second model to the terminal device of the manufacturer.

[0243] If the NMS indicates the second model to the OTT#1 through the method 1) or method 4) shown in S430 above, the OTT#1 can directly obtain the second model; if the NMS indicates the second model to the OTT#1 through the method 2) or method 3) shown in S430 above, the OTT#1 needs to train the model based on the training data set, model structure, loss function type, training method, parameter indication information, etc. of the second model to obtain the second model, and the OTT#1 can also perform model optimization, acceleration, adaptation (for example, custom API) of the second model according to the hardware of the manufacturer, so that the second model can run in the chip and operating system of the terminal device of the manufacturer.

[0244] Exemplarily, the OTT#1 can deploy the second model to the terminal device of the manufacturer through OTA technology.

[0245] S434, the NMS sends second information to the EMS#1, and the second information is used to indicate the fourth model. Correspondingly, the EMS#1 receives the second information.

[0246] Exemplarily, the second information can include at least one of the following:

[0247] The model file, training data set, model structure, loss function type, training method, parameter matrix list, address of the device that can obtain the fourth model, etc. of the fourth model.

[0248] Exemplarily, the NMS can indicate the fourth model to the EMS#1 through the following four methods:

[0249] 1) Send the model file of the fourth model. For example, send the model file in.bin,.pt,.json,.onnx, etc. format.

[0250] 2) send the training data set, model structure, loss function type, training method of the fourth model. For example, the loss function type can be a function expression of the loss function formula, parameter value, etc., and the training method can be the optimizer type (Adam, SGD, etc.) of the training.

[0251] 3) send the model structure and parameter indication information of the fourth model. For example, the parameter indication information can be a parameter matrix list.

[0252] 4) send the address of the fourth model obtainable device.

[0253] S436, EMS#1 deploys the fourth model to the managed network device.

[0254] Exemplarily, if the NMS indicates the fourth model to the EMS#1 through the method 1) or method 4) shown in S434 above, the EMS#1 can directly obtain the fourth model; if the NMS indicates the fourth model to the EMS#1 through the method 2) or method 3) shown in S434 above, the EMS#1 needs to train the model based on the training data set, model structure, loss function type, training method, parameter indication information, etc. of the fourth model to obtain the fourth model, and the EMS#1 can also perform model optimization, acceleration, adaptation (for example, custom API) of the fourth model according to the hardware of the network device manufacturer, so that the fourth model can run in the chip and operating system of the managed network device.

[0255] Through the above-mentioned communication method 400, the NMS can determine a set of AI models for data encoding and data decoding for the whole network, avoiding the terminal device and the network device to store more AI models, saving storage overhead.

[0256] The communication method 500 shown below is an example of a technical solution applicable to the above-mentioned case 3, Figure 5 is an example of an exemplary flowchart of the communication method 500.

[0257] The communication method 500 can include the following steps:

[0258] Optionally, S510, the NMS sends third information to the OTT#1. Correspondingly, the OTT#1 receives the third information.

[0259] Specifically, the third information is used to request to update the first model, and the first model is a first unilateral model, and the first unilateral model is used for data encoding or data decoding.

[0260] Exemplarily, the third information can be periodically sent, that is, the NMS can periodically request the OTT#1 to update the first model.

[0261] The communication method 500 is described by taking an OTT as an example. There can be other OTTs. The other OTTs also perform receiving actions or sending actions similar to OTT #1, which are not described here.

[0262] S512, OTT #1 performs model training based on the first data set to obtain a first model #1.

[0263] The execution sequence of S510 and S512 described above is not limited. For example, the NMS can first send the third information to the OTT #1, and then the OTT #1 performs model training based on the first data set to obtain the first model #1. Alternatively, the OTT #1 can first perform model training based on the first data set to obtain the first model #1, and then the NMS sends the third information to the OTT #1.

[0264] Specifically, the first model #1 is a first one-sided model, which can be used for data encoding or data decoding.

[0265] Specifically, the data included in the first data set corresponds to the function of the first model #1, that is, the data of the first data set is collected according to the function of the first model #1 to be trained. For example, if the first model #1 is a model for CSI compression or CSI decompression, the first data set can include original channel data collected from a terminal device or a network device, such as an original channel time-frequency resource matrix.

[0266] S514, the OTT #1 sends fourth information to the NMS, which is used to indicate the first model #1.

[0267] For example, the fourth information can include at least one of the following:

[0268] The model file of the first model #1, the training data set, the model structure, the loss function type, the training method, the parameter matrix list, the address of the device where the first model #1 can be obtained, and the like.

[0269] For example, the OTT #1 can indicate the first model #1 to the NMS by the following four methods:

[0270] 1) Send the model file of the first model #1. For example, send the model file in the format of.bin,.pt,.json,.onnx, etc.

[0271] 2) Send the training data set, the model structure, the loss function type, and the training method of the first model #1. For example, the training data set of the first model #1 can be the first data set described above, the loss function type can be the function expression of the loss function formula, the parameter value, etc., and the training method can be the optimizer type (Adam, SGD, etc.) of the training.

[0272] 3) Send the model structure of the first model #1 and parameter indication information, for example, the parameter indication information can be a parameter matrix list.

[0273] 4) Send the address of the first model #1 obtainable device.

[0274] S516, the NMS obtains at least one first model.

[0275] S518, the NMS determines a second model based on the at least one first model.

[0276] Specifically, the NMS performs model training based on the at least one first model to obtain a second model, and the second model is a second univariate model. If the first univariate model is used for data encoding, the second univariate model is used for data decoding; if the first univariate model is used for data decoding, the second univariate model is used for data encoding.

[0277] Illustratively, the NMS updates the local first model list based on the obtained at least one first model, and trains a second model jointly using all the first models. For example, the NMS generates a training data set #1 using all the first models and historical original channel time-frequency resource matrices, and performs model training based on the training data set #1 to obtain a second model.

[0278] S520, the NMS sends first information to the EMS #1, and the first information is used to indicate the second model.

[0279] Illustratively, the method that the NMS indicates the second model to the EMS #1 can refer to the four model indication methods of S514, which will not be described here.

[0280] The S518-S520 shown above is a way one in which the EMS #1 can obtain the second model, and the S522-S524 shown below is a way two in which the EMS #1 can obtain the second model.

[0281] S522, the NMS indicates at least one first model to the EMS #1. Accordingly, the EMS #1 obtains at least one first model.

[0282] Illustratively, the method that the NMS indicates at least one first model to the EMS #1 can refer to the three model indication methods of S514, which will not be described here.

[0283] S524, the EMS #1 determines a second model based on the at least one first model.

[0284] Specifically, the EMS #1 performs model training based on the at least one first model to obtain a second model, which is a second one-sided model. If the first one-sided model is used for data encoding, the second one-sided model is used for data decoding; if the first one-sided model is used for data decoding, the second one-sided model is used for data encoding.

[0285] For example, the EMS #1 updates the local first model list based on the obtained at least one first model, and jointly trains a second model using all the first models. For example, the EMS #1 generates a training data set #1 using all the first models and a historical original channel time-frequency resource matrix, and performs model training based on the training data set #1 to obtain a second model.

[0286] In S526, the EMS #1 deploys the second model to the network device.

[0287] For example, in the above-mentioned manner in which the EMS #1 can obtain the second model, if the NMS indicates the second model to the EMS #1 by using the method 1) or the method 4) of the four model sending methods, the EMS #1 can directly obtain the second model; if the NMS indicates the second model to the EMS #1 by using the method 2) or the method 3) of the four model sending methods, the EMS #1 needs to perform model training based on the training data set, the model structure, the loss function type, the training method, the parameter indication information, etc. of the second model to obtain the second model; the EMS #1 can also perform model optimization, acceleration, and adaptation (for example, customization of API) of the second model according to the manufacturer hardware of the network device, so that the second model can run in the chip and operating system of the network device of the manufacturer.

[0288] In S528, the OTT #1 deploys the first model #1 to the terminal device of the manufacturer.

[0289] For example, the OTT #1 can deploy the first model #1 trained by itself to the terminal device of the manufacturer by using the OTA technology.

[0290] Through the above-mentioned communication method 500, the NMS can train a network device side model for all network devices based on the terminal device side models provided by each terminal device manufacturer, so that all terminal devices and network devices in the network only need to reserve one AI model for data encoding or data decoding, thereby saving storage overhead.

[0291] It should be noted that, in the above-mentioned communication method 500, the NMS trains a network device side model for all network devices based on the terminal device side models provided by each terminal device manufacturer, and the NMS can also train a terminal device side model for all terminal devices based on the network device side models provided by each EMS for managing the network device, which is not limited in the present application.

[0292] It can be understood that some optional features in the embodiments of the present application can not depend on other features in some scenarios, or can be combined with other features in some scenarios, without limitation.

[0293] It can also be understood that the solutions in the embodiments of the present application can be reasonably combined, and the explanations or descriptions of various terms appearing in the embodiments can be mutually referenced or explained in various embodiments, without limitation.

[0294] It can also be understood that the methods and operations implemented by the devices (such as the first device and the second device) in the above method embodiments can also be implemented by components (such as chips or circuits) of the devices, without limitation.

[0295] The above, in combination with Figures 1 to 5 The method provided by the embodiments of the present application is described in detail. In the following, in combination with Figures 6 to 8 The device provided by the embodiments of the present application is described in detail. It should be understood that the description of the device embodiments corresponds to the description of the method embodiments, therefore, the contents not described in detail can be referred to the above method embodiments, and for brevity, will not be repeated here.

[0296] Referring to Figure 6 , Figure 6 is a schematic diagram of a communication device 600 provided by the embodiments of the present application. The device 600 includes a transceiver unit 610. The transceiver unit 610 can be used to implement the corresponding communication function. The transceiver unit 610 can also be referred to as a communication interface or a communication unit.

[0297] Optionally, the device 600 further includes a processing unit 620. The processing unit 620 can be used for processing. The function of the processing unit 620 can be implemented by one or more processors. Specifically, the processor can include a modem chip, or a system on chip SoC chip or a SIP chip containing a modem core.

[0298] Optionally, the device 600 can further include a storage unit, which can be used to store instructions and / or data. The processing unit 620 can read the instructions and / or data in the storage unit, so that the device implements the foregoing method embodiments.

[0299] Optionally, the transceiver unit 610 can include a receiving unit and a sending unit. The receiving unit can be used to perform receiving related operations (such as operations of receiving data or messages), and the sending unit can be used to perform sending related operations (such as operations of sending data or messages).

[0300] The first possible design is that the device 600 can be the first device in the foregoing embodiments, and the device 600 can implement the corresponding operations of the first device in the foregoing method embodiments. Figures 2 to 5The steps or procedures performed by the first device in the method embodiment. The transceiver 610 can be configured to perform the above Figures 2 to 5 The transceiver-related operations (e.g., operations of sending and / or receiving data or messages) of the first device or the NMS in the method embodiment. The processing unit 620 can be configured to perform the above Figures 2 to 5 The processing-related operations, or operations other than the transceiving (e.g., operations other than sending and / or receiving data or messages) of the first device or the NMS in the method embodiment.

[0301] The transceiver 610 can be configured to: obtain at least one first model; and the transceiver 610 can be further configured to send a second model.

[0302] The processing unit 620 can be configured to: determine a second model based on the at least one first model.

[0303] The second possible design, the apparatus 600 can be the second device in the foregoing embodiments, and the apparatus 600 can implement the steps or procedures corresponding to the above Figures 2 to 5 The steps or procedures performed by the second device or the OTT#1 or the EMS#1 in the method embodiment. The transceiver 610 can be configured to perform the above Figures 2 to 5 The transceiver-related operations (e.g., operations of sending and / or receiving data or messages) of the second device or the OTT#1 or the EMS#1 in the method embodiment. The processing unit 620 can be configured to perform the above Figures 2 to 5 The processing-related operations, or operations other than the transceiving (e.g., operations other than sending and / or receiving data or messages) of the second device or the OTT#1 or the EMS#1 in the method embodiment.

[0304] The transceiver 610 can be configured to: send a first model to the first device; and the transceiver 610 can be further configured to receive a second model from the first device.

[0305] It should be understood that the specific processes by which the units perform the above corresponding steps have been described in the above method embodiments, and thus will not be described here for brevity.

[0306] It should also be understood that the apparatus 600 herein is embodied in the form of a functional block diagram. The term "unit" herein can refer to an application specific integrated circuit (ASIC), an electronic circuit, a processor (for example, a shared processor, a dedicated processor, or a group processor, etc.) and a memory for executing one or more software or firmware programs, a combination of logical circuit and / or other suitable components supporting the described functions. In an optional example, those skilled in the art can understand that the apparatus 600 can be embodied as a communication device in the above-mentioned embodiments, and can be used to execute the processes and / or steps corresponding to the communication device in each of the above-mentioned method embodiments. To avoid repetition, details are not described here.

[0307] The apparatus 600 of each of the above-mentioned schemes has a function of implementing the corresponding steps performed by the communication device in the above-mentioned methods. The function can be implemented by hardware or by executing corresponding software by hardware. The hardware or software includes one or more modules corresponding to the above-mentioned functions; for example, the transceiver unit can be replaced by a transceiver (for example, the transmitting unit in the transceiver unit can be replaced by a transmitter, and the receiving unit in the transceiver unit can be replaced by a receiver), and other units, such as the processing unit, can be replaced by a processor, which respectively performs the transceiving operation and the related processing operation in each of the method embodiments.

[0308] In addition, the transceiver unit 610 described above can also be a transceiver circuit (for example, which can include a receiving circuit and a transmitting circuit), and the processing unit can be a processing circuit.

[0309] It should be noted that, Figure 6 The apparatus in the above-mentioned embodiments can be a communication device, or a chip or a chip system, for example, a modem chip or a system on chip (SoC) chip or a system in package (SIP) chip containing a modem core. The transceiver unit can be an input / output circuit, a communication interface; and the processing unit can be a processor or a microprocessor or an integrated circuit integrated on the chip. Here, no limitation is made.

[0310] Referring to Figure 7 , Figure 7 is a schematic diagram of another communication device 700 provided by the embodiments of the present application. The device 700 includes a processor 710, and the processor 710 is coupled with a memory 720, the memory 720 is used to store computer programs or instructions and / or data, and the processor 710 is used to execute the computer programs or instructions stored in the memory 720, or read the data stored in the memory 720, to execute the methods in the above-mentioned method embodiments.

[0311] Optionally, the processor 710 is one or more.

[0312] Optionally, the memory 720 is one or more.

[0313] Optionally, the memory 720 is integrated with the processor 710 or is separately arranged.

[0314] Optionally, as shown in Figure 7 the apparatus 700 further includes a transceiver 730 for signal receiving and / or sending. For example, the processor 710 is configured to control the transceiver 730 to perform signal receiving and / or sending. Exemplarily, the transceiver 730 can include a transmitter and / or a receiver, the transmitter is configured to perform sending operation, and the receiver is configured to perform receiving operation.

[0315] For example, the processor 710 can have the functions of the processing unit 620 shown in Figure 6 , the memory 720 can have the function of the storage unit, and the transceiver 730 can have the function of the transceiving unit 610 shown in Figure 6 .

[0316] As an option, the apparatus 700 is configured to implement the operations performed by the communication apparatus in the various method embodiments.

[0317] For example, the processor 710 is configured to execute the computer program or instructions stored in the memory 720 to implement the related operations of the terminal device or the network device in the various method embodiments.

[0318] It should be understood that the processor mentioned in the embodiments of the present application can be a central processing unit (CPU), and can also be other general-purpose processors, digital signal processors (DSP), application specific integrated circuits (ASIC), field programmable gate arrays (FPGA) or other programmable logic devices, discrete gates or transistor logic devices, discrete hardware components, etc. The general-purpose processor can be a microprocessor or the processor can also be any conventional processor.

[0319] It should also be understood that the memory referred to in the embodiments of the present application can be a volatile memory and / or a non-volatile memory. Among them, the non-volatile memory can be a read-only memory (ROM), a programmable read-only memory (PROM), an erasable programmable read-only memory (EPROM), an electrically EPROM (EEPROM) or a flash memory. The volatile memory can be a random access memory (RAM). For example, the RAM can be used as an external cache. As an example but not limitation, the RAM includes the following various forms: static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), double data rate SDRAM (DDR SDRAM), enhanced SDRAM (ESDRAM), synchronous link DRAM (SLDRAM) and direct memory bus random access memory (DRAM) (DRAM).

[0320] It should be noted that when the processor is a general processor, DSP, ASIC, FPGA or other programmable logic device, discrete gate or transistor logic device, discrete hardware component, the memory (storage module) can be integrated in the processor.

[0321] It should also be noted that the memory described herein is intended to include, but not limited to, these and any other suitable types of memory.

[0322] Referring to Figure 8 , Figure 8 is a schematic diagram of a chip system 800 provided by an embodiment of the present application. The chip system 800 (or also can be called a processing system) includes a logic circuit 810 and an input / output interface 820.

[0323] The logic circuit 810 can be a processing circuit in the chip system 800. The logic circuit 810 can be coupled to a storage unit, and invoke instructions in the storage unit, so that the chip system 800 can implement the methods and functions of the embodiments of the present application. The input / output interface 820 can be an input / output circuit in the chip system 800, and output information processed by the chip system 800, or input data or signaling information to be processed by the chip system 800.

[0324] Optionally, the logic circuit 810 can be implemented by one or more processors, including the one or more processors or processing portions in the one or more processors.

[0325] Optionally, the input / output interface 820 can include a transceiver, a transceiver, an input / output circuit or a communication interface.

[0326] As an option, the chip system 800 is configured to implement the operations performed by the communication device (e.g., the first device, and the second device) in the above various method embodiments.

[0327] For example, the logic circuit 810 is configured to implement the processing-related operations performed by the communication device (e.g., the first device, and the second device) in the above method embodiments; the input / output interface 820 is configured to implement the sending and / or receiving-related operations performed by the communication device (e.g., the first device, and the second device) in the above method embodiments, the input interface is configured to perform the receiving operation, and the output interface is configured to perform the sending operation.

[0328] The embodiments of the present application also provide a computer readable storage medium, which stores computer instructions for implementing the method performed by the communication device (e.g., the first device, and the second device) in the above various method embodiments.

[0329] For example, the computer program is executed by a computer, so that the computer can implement the method performed by the communication device (e.g., the first device, and the second device) in the above various method embodiments.

[0330] The embodiments of the present application also provide a computer program product, which includes instructions executed by a computer to implement the method performed by the communication device (e.g., the first device, and the second device) in the above various method embodiments.

[0331] The embodiments of the present application also provide a communication system, which includes the first device and / or the second device in the above various embodiments. For example, the system includes Figure 2 the first device and the second device in the above various embodiments.

[0332] The explanations and beneficial effects of the related contents in any of the above devices can refer to the corresponding method embodiments provided above, and will not be repeated here.

[0333] In several embodiments provided in the present application, it should be understood that the disclosed apparatus and method can be implemented in other manners. For example, the described apparatus embodiments are merely schematic. The division of the units is merely a logical function division. There can be another division manner for the actual implementation, for example, multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. In addition, the displayed or discussed mutual couplings or direct couplings or communication connections can be indirect couplings or communication connections through some interfaces, devices or units, and can be in electrical, mechanical or other forms.

[0334] In the embodiments described above, all or part of the embodiments can be implemented by software, hardware, firmware or any combination thereof. When implemented by software, all or part of the embodiments can 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 can be a general purpose computer, a special purpose computer, a computer network, or other programmable devices. For example, the computer can be a personal computer, a server, a network device, etc. The computer instructions can 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 can be transmitted from one website site, computer, server or data center to another website site, computer, server or data center through wired (such as coaxial cable, optical fiber, digital subscriber line (DSL)) or wireless (such as infrared, wireless, microwave, etc.) manner. The computer readable storage medium can be any available medium that can be accessed by a computer or a data storage device such as a server, data center, etc. integrated with one or more available media. The available media can be magnetic media (such as floppy disk, hard disk, magnetic tape), optical media (such as DVD), or semiconductor media (such as solid state disk (SSD), etc.). For example, the foregoing available media includes but is not limited to: U disk, mobile hard disk, read-only memory (ROM), random access memory (RAM), magnetic disk or optical disk, etc. Various media that can store program codes.

[0335] The above merely provides the specific implementation of the present application, but the protection scope of the present application is not limited to this. Any person skilled in the art can easily think of the changes or replacements within the technical range disclosed by the present application, which should be covered in the protection scope of the present application. Therefore, the protection scope of the present application should be subject to the protection scope of the claims.

Claims

1. A communication method, characterized in that, The method includes: Obtain at least one first model, wherein the first model is a bilateral model or a first unilateral model, the bilateral model is used for data encoding and data decoding, and the first unilateral model is used for data encoding or data decoding. A second model is determined based on the at least one first model, the second model being used for data encoding or data decoding; Send a first message, which is used to instruct the second model.

2. The method according to claim 1, characterized in that, When the first model is the first unilateral model, determining a second model based on the at least one first model includes: Based on the at least one first model and at least one local model, a second model is determined, wherein the second model belongs to the at least one first model, and the local model is a second unilateral model; If the first one-sided model is used for data encoding, the second one-sided model is used for data decoding; or, if the first one-sided model is used for data decoding, the second one-sided model is used for data encoding.

3. The method according to claim 1 or 2, characterized in that, When the first model is the first unilateral model, the method further includes: Obtain at least one third model, wherein the third model is a second unilateral model; Determining a second model based on the at least one first model includes: determining a second model and a fourth model based on the at least one first model and the at least one third model, wherein the second model belongs to the at least one first model and the fourth model belongs to the at least one third model; Sending the first information includes: sending the first information and the second information, wherein the first information is used to indicate the second model and the second information is used to indicate the fourth model; If the first one-sided model is used for data encoding, the second one-sided model is used for data decoding; or, if the first one-sided model is used for data decoding, the second one-sided model is used for data encoding.

4. The method according to claim 3, characterized in that, The at least one third model is trained based on the at least one first model, or the at least one first model is trained based on the at least one third model.

5. The method according to claim 1, characterized in that, When the first model is the first unilateral model, the second model is the second unilateral model. If the first one-sided model is used for data encoding, the second one-sided model is used for data decoding; or, if the first one-sided model is used for data decoding, the second one-sided model is used for data encoding.

6. The method according to any one of claims 1 to 5, characterized in that, The method further includes: Send a third message, the third message being used to request an update to the first model; The step of obtaining at least one first model includes: the first device obtaining the at least one first model based on the third information, wherein the first model is an updated model.

7. The method according to any one of claims 1 to 6, characterized in that, The first information includes at least one of the following: The second model includes the model file, training dataset, model structure, loss function type, training method, parameter matrix list, and the address of the device from which the second model can be obtained.

8. A communication method, characterized in that, The method includes: Send a fourth message, the fourth message being used to indicate a first model, the first model being either a bilateral model or a first unilateral model, the bilateral model being used for data encoding and data decoding, the first unilateral model being used for data encoding or data decoding, the first model being used by a first device to determine a second model, the second model being used for data encoding or data decoding.

9. The method according to claim 8, characterized in that, When the first model is the first unilateral model, the method further includes: Receive first information, which is used to indicate the second model, and the second model is also the first one-sided model.

10. The method according to claim 8, characterized in that, When the first model is the first unilateral model, the second model is the second unilateral model. If the first one-sided model is used for data encoding, the second one-sided model is used for data decoding; or, if the first one-sided model is used for data decoding, the second one-sided model is used for data encoding.

11. The method according to any one of claims 8 to 10, characterized in that, The method further includes: Receive third information, the third information being used to request an update to the first model; Sending the fourth information includes: the second device sending the fourth information based on the third information, wherein the first model indicated by the fourth information is an updated model.

12. The method according to any one of claims 8 to 11, characterized in that, The fourth piece of information includes at least one of the following: The first model includes its model file, training dataset, model structure, loss function type, training method, parameter matrix list, and the address of the device from which the first model can be obtained.

13. An apparatus, characterized in that, The apparatus includes a unit or module for performing the method of any one of claims 1 to 7, or the apparatus includes a unit or module for performing the method of any one of claims 8 to 12.

14. A system, characterized in that, include: A first device and a second device, wherein the first device is used to perform the method as described in any one of claims 1 to 7, and the second device is used to perform the method as described in any one of claims 8 to 12.

15. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores computer instructions that, when the computer program is run on a computer, cause the computer to perform the method as described in any one of claims 1 to 7, or cause the computer to perform the method as described in any one of claims 8 to 12.

16. A computer program product, characterized in that, The computer program product includes: computer program code, which, when run on a communication device, causes the device to perform the method as described in any one of claims 1 to 7, or causes the device to perform the method as described in any one of claims 8 to 12.

17. A chip, characterized in that, The chip includes a processor and a communication interface. The processor reads instructions stored in the memory through the communication interface and executes the method as described in any one of claims 1 to 7, or executes the method as described in any one of claims 8 to 12.

Images