Feature information transmission method, conversion information determination method, device and communication equipment

Abstract

The application discloses a feature information transmission method, a conversion information determination method, a device and communication equipment, and belongs to the technical field of communication. The feature information transmission method comprises the following steps: a first node processes first information into first feature information based on a first AI network model; and the first node sends second information to a second node, wherein the second information comprises the first feature information or second feature information, and the second feature information is converted feature information of the first feature information.

Description

Technical Field

[0001] This application belongs to the field of communication technology, specifically relating to a method for transmitting feature information, a method for determining converted information, an apparatus, and a communication device. Background Technology

[0002] In related technologies, methods for transmitting channel characteristic information using artificial intelligence (AI) network models have been studied.

[0003] The AI ​​network model can include an encoding part (i.e., an encoding AI network model) and a decoding part (i.e., a decoding AI network model). The encoding AI network model is used to encode channel information into channel feature information, and the decoding AI network model is used to recover channel information from the channel feature information output by the encoding AI network model. Thus, in order to match the encoding AI network model and the decoding AI network model, the encoding AI network model and the decoding AI network model need to be jointly trained on the same device. Then, the jointly trained encoding AI network model is transmitted to the terminal, and the jointly trained decoding AI network model is transmitted to the base station.

[0004] As can be seen from the above, in related technologies, there is a problem that the transmission overhead of terminals and / or base stations increases due to the transmission of AI network models. Summary of the Invention

[0005] This application provides a feature information transmission method, a transformation information determination method, an apparatus, and a communication device, which can reduce the transmission overhead of AI network models.

[0006] Firstly, a method for transmitting feature information is provided, the method comprising:

[0007] The first node, based on the first AI network model, processes the first information into first feature information;

[0008] The first node sends second information to the second node. The second information includes the first feature information or the second feature information, wherein the second feature information is the feature information converted from the first feature information.

[0009] Secondly, a feature information transmission device is provided, applied to a first node, the device comprising:

[0010] The first processing module is used to process the first information into first feature information based on the first AI network model;

[0011] The first sending module is used to send second information to the second node. The second information includes the first feature information or the second feature information, wherein the second feature information is feature information converted from the first feature information.

[0012] Thirdly, a method for transmitting feature information is provided, including:

[0013] The second node receives second information from the first node, wherein the second information includes first feature information or second feature information, wherein the first feature information is feature information obtained by processing the first information based on the first AI network model of the first node, and the second feature information is feature information transformed from the first feature information;

[0014] The second node performs recovery processing on the second feature information based on the fourth AI network model to obtain the first information. In the case where the second information includes the first feature information, the second feature information is obtained by the second node transforming the first feature information.

[0015] Fourthly, a feature information transmission device is provided, applied to a second node, the device comprising:

[0016] The first receiving module is configured to receive second information from the first node, wherein the second information includes first feature information or second feature information, wherein the first feature information is feature information obtained by processing the first information based on the first AI network model of the first node, and the second feature information is feature information converted from the first feature information;

[0017] The second processing module is used to recover the second feature information based on the fourth AI network model to obtain the first information. In the case where the second information includes the first feature information, the second feature information is obtained by the second node transforming the first feature information.

[0018] Fifthly, a method for determining transformation information is provided, including:

[0019] The third node obtains the third feature information obtained by the first AI network model after processing the training sample data, and obtains the fourth feature information obtained by the second AI network model after processing the training sample data, wherein the first AI network model is the AI ​​network model possessed by the first node, and the second AI network model is the AI ​​network model possessed by the second node.

[0020] The third node determines conversion information based on the third feature information and the fourth feature information, wherein the conversion information is used to convert the third feature information corresponding to the target training sample into the fourth feature information corresponding to the target training sample, and the target training sample is any sample in the training sample data;

[0021] The third node sends the conversion information to at least one of the first node and the second node.

[0022] Sixthly, a conversion information determining device is provided, applied to a third node, the device comprising:

[0023] The first acquisition module is used to acquire the third feature information obtained by the first AI network model after processing the training sample data, and to acquire the fourth feature information obtained by the second AI network model after processing the training sample data, wherein the first AI network model is the AI ​​network model possessed by the first node, and the second AI network model is the AI ​​network model possessed by the second node.

[0024] The first determining module is used to determine conversion information based on the third feature information and the fourth feature information, wherein the conversion information is used to convert the third feature information corresponding to the target training sample into the fourth feature information corresponding to the target training sample, and the target training sample is any sample in the training sample data;

[0025] The second sending module is used to send the conversion information to at least one of the first node and the second node.

[0026] In a seventh aspect, a communication device is provided, the communication device including a processor and a memory, the memory storing a program or instructions executable on the processor, the program or instructions, when executed by the processor, implementing the steps of the method as described in the first, third, or fifth aspect.

[0027] Eighthly, a communication device is provided, including a processor and a communication interface, wherein the processor is used to process first information into first feature information based on a first AI network model; the communication interface is used to send second information to a second node, the second information including the first feature information or second feature information, wherein the second feature information is feature information converted from the first feature information; or...

[0028] The communication interface is used to receive second information from the first node, wherein the second information includes first feature information or second feature information, wherein the first feature information is feature information obtained after processing the first information based on a first AI network model possessed by the first node, and the second feature information is feature information transformed from the first feature information; the processor is used to perform recovery processing on the second feature information based on a fourth AI network model to obtain the first information, wherein, if the second information includes the first feature information, the second feature information is obtained by the second node through transformation processing of the first feature information; or...

[0029] The communication interface is used to acquire third feature information obtained by the first AI network model after processing the training sample data, and to acquire fourth feature information obtained by the second AI network model after processing the training sample data, wherein the first AI network model is the AI ​​network model possessed by the first node, and the second AI network model is the AI ​​network model possessed by the second node; the processor is used to determine conversion information based on the third feature information and the fourth feature information, wherein the conversion information is used to convert the third feature information corresponding to the target training sample into the fourth feature information corresponding to the target training sample, wherein the target training sample is any sample in the training sample data; the communication interface is also used to send the conversion information to at least one of the first node and the second node.

[0030] A ninth aspect provides a communication system comprising: a first node, a second node, and a third node, wherein the first node is configured to perform the steps of the feature information transmission method as described in the first aspect, the second node is configured to perform the steps of the feature information transmission method as described in the third aspect, and the third node is configured to perform the steps of the conversion information determination method as described in the fifth aspect.

[0031] In a tenth aspect, a readable storage medium is provided, on which a program or instructions are stored, which, when executed by a processor, implement the steps of the method described in the first aspect, or the steps of the method described in the third aspect, or the steps of the method described in the fifth aspect.

[0032] Eleventhly, a chip is provided, the chip including a processor and a communication interface, the communication interface being coupled to the processor, the processor being configured to run programs or instructions to implement the method as described in the first aspect, or the method as described in the third aspect, or the method as described in the fifth aspect.

[0033] In a twelfth aspect, a computer program / program product is provided, which is stored in a storage medium and is executed by at least one processor to implement the steps of the feature information transmission method as described in the first aspect, or the computer program / program product is executed by at least one processor to implement the steps of the feature information transmission method as described in the third aspect, or the computer program / program product is executed by at least one processor to implement the steps of the conversion information determination method as described in the fifth aspect.

[0034] In this embodiment, a first node processes first information into first feature information based on a first AI network model. The first node sends second information to a second node, the second information including either the first feature information or the second feature information, wherein the second feature information is the feature information transformed from the first feature information. When compressing and restoring the first information between the first node and the second node using an AI network model, if the AI ​​network models used by the first node and the second node are mismatched, resulting in the first feature information compressed by the first node being unusable or unrecoverable by the second node, the first node can transform the first feature information to send second feature information that it can use or recover to the second node. Alternatively, the first node can send the first feature information to the second node, and the second node can transform the first feature information into second feature information that its AI network model can use or recover, and then use the AI ​​network model to recover the second feature information. In this way, the first node and the second node can independently use or train their respective AI network models, thereby reducing the overhead caused by transmitting AI network models. At the same time, the first node and the second node do not need to interact with their respective AI network models, thereby improving the information security of the first node and the second node. Attached Figure Description

[0035] Figure 1 This is a schematic diagram of the structure of a wireless communication system that can be applied to the embodiments of this application;

[0036] Figure 2 This is a flowchart of a feature information transmission method provided in an embodiment of this application;

[0037] Figure 3 This is a schematic diagram of the AI ​​network model of the first node and the second node in the embodiments of this application;

[0038] Figure 4 This is a flowchart of another feature information transmission method provided in the embodiments of this application;

[0039] Figure 5 This is a flowchart of a method for determining conversion information provided in an embodiment of this application;

[0040] Figure 6 This is a schematic diagram of the structure of a feature information transmission device provided in an embodiment of this application;

[0041] Figure 7 This is a schematic diagram of another feature information transmission device provided in the embodiments of this application;

[0042] Figure 8 This is a schematic diagram of the structure of a conversion information determination device provided in an embodiment of this application;

[0043] Figure 9 This is a schematic diagram of the structure of a communication device provided in an embodiment of this application. Detailed Implementation

[0044] The technical solutions of the embodiments of this application will be clearly described below with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of this application. All other embodiments obtained by those skilled in the art based on the embodiments of this application are within the scope of protection of this application.

[0045] The terms "first," "second," etc., used in the specification and claims of this application are used to distinguish similar objects and not to describe a specific order or sequence. It should be understood that such terms can be used interchangeably where appropriate so that embodiments of this application can be implemented in orders other than those illustrated or described herein, and the objects distinguished by "first" and "second" are generally of the same class, not limited in number; for example, a first object can be one or more. Furthermore, in the specification and claims, "and / or" indicates at least one of the connected objects, and the character " / " generally indicates that the preceding and following objects are in an "or" relationship.

[0046] It is worth noting that the technologies described in this application are not limited to Long Term Evolution (LTE) / LTE-Advanced (LTE-A) systems, but can also be used in other wireless communication systems, such as Code Division Multiple Access (CDMA), Time Division Multiple Access (TDMA), Frequency Division Multiple Access (FDMA), Orthogonal Frequency Division Multiple Access (OFDMA), Single-carrier Frequency Division Multiple Access (SC-FDMA), and other systems. The terms "system" and "network" in this application are often used interchangeably, and the described technologies can be used with the systems and radio technologies mentioned above, as well as with other systems and radio technologies. The following description describes New Radio (NR) systems for illustrative purposes, and NR terminology is used in most of the following description; however, these technologies can also be applied to applications beyond NR systems, such as 6th generation (6G) radio systems. th Generation 6G communication system.

[0047] Figure 1This diagram illustrates a block diagram of a wireless communication system applicable to embodiments of this application. The wireless communication system includes a terminal 11 and a network-side device 12. Terminal 11 can be a mobile phone, tablet computer, laptop computer, personal digital assistant (PDA), handheld computer, netbook, ultra-mobile personal computer (UMPC), mobile internet device (MID), augmented reality (AR) / virtual reality (VR) device, robot, wearable device, vehicle-mounted device (VUE), pedestrian terminal (PUE), smart home (home devices with wireless communication capabilities, such as refrigerators, televisions, washing machines, or furniture), game console, personal computer (PC), ATM, or self-service machine, etc. Wearable devices include: smartwatches, smart bracelets, smart headphones, smart glasses, smart jewelry (smart bracelets, smart chains, smart rings, smart necklaces, smart anklets, smart anklets, etc.), smart wristbands, smart clothing, etc. It should be noted that the specific type of terminal 11 is not limited in this embodiment. Network-side equipment 12 may include access network equipment or core network equipment. Access network equipment 12 may also be referred to as radio access network equipment, radio access network (RAN), radio access network function, or radio access network unit. Access network equipment 12 may include base stations, WLAN access points, or WiFi nodes, etc. Base stations may be referred to as Node B, evolved Node B (eNB), access point, base transceiver station (BTS), radio base station, radio transceiver, Basic Service Set (BSS), Extended Service Set (ESS), home B node, home evolved B node, Transmitting Receiving Point (TRP), or any other suitable term in the field, as long as the same technical effect is achieved. The base station is not limited to specific technical terms. It should be noted that in this application embodiment, only a base station in an NR system is used as an example for description, and the specific type of base station is not limited.

[0048] Artificial intelligence has been widely applied in various fields. AI network models can be implemented in various ways, such as neural networks, decision trees, support vector machines, and Bayesian classifiers. This application uses neural networks as an example, but does not limit the specific type of AI network model.

[0049] Generally, the AI ​​algorithms and network models selected vary depending on the type of problem to be solved. The main method for improving 5G network performance using AI network models is to enhance or replace existing algorithms or processing modules with neural network-based algorithms and models. In specific scenarios, neural network-based algorithms and models can achieve better performance than deterministic algorithms. Commonly used neural networks include deep neural networks, convolutional neural networks, and recurrent neural networks. Existing AI tools can be used to build, train, and validate neural networks.

[0050] In related technologies, AI network models include an encoding part and a decoding part. The encoding part (also called the encoding AI network model or encoder, i.e., the AI ​​network model used in the encoder) and the decoding part (also called the decoding AI network model or decoder, i.e., the AI ​​network model used in the decoder) are usually trained jointly. After joint training, the encoding part is sent to the first node and the decoding part is sent to the second node. The first node can be a terminal or network-side device in a wireless communication network. The first node encodes the information to be transmitted based on the encoding part to obtain the feature information of the information, and sends the feature information to the second node. The second node can then recover the received feature information based on the decoding part. The second node can be a terminal or network-side device in the wireless communication network that is different from the first node.

[0051] As can be seen from the above, the relevant technologies require joint training of the encoding AI network model and the decoding AI network model, and transmission of the encoding AI network model and / or the decoding AI network model to deploy the encoding AI network model on the first node and the decoding AI network model on the second node.

[0052] In this embodiment, the encoding AI network model used by the first node and the decoding AI network model used by the second node can be trained independently, thus eliminating the need to transmit the encoding and / or decoding AI network models between the first and second nodes. Furthermore, the encoding AI network model used by the first node and the decoding AI network model used by the second node may not be compatible. For example, the feature information output by the encoding AI network model used by the first node may not be decoded by the decoding AI network model used by the second node, or the dimension or size of the feature information output by the encoding AI network model used by the first node may not match the input information of the decoding AI network model used by the second node. In this embodiment, the feature information output by the encoding AI network model used by the first node can be transformed at either the first or second node so that the transformed feature information can be used or processed by the second node.

[0053] The following description, in conjunction with the accompanying drawings, details the feature information transmission method, feature information transmission device, and communication equipment provided in this application through some embodiments and application scenarios.

[0054] Please see Figure 2 The present application provides a feature information transmission method, the execution subject of which is a first node, such as... Figure 2 As shown, the feature information transmission method executed by the first node may include the following steps:

[0055] Step 201: The first node processes the first information into first feature information based on the first AI network model.

[0056] Step 202: The first node sends second information to the second node. The second information includes the first feature information or the second feature information, wherein the second feature information is the feature information converted from the first feature information.

[0057] In this embodiment, the first node can be an information sender and the second node can be an information receiver. The first information is the information that the first node needs to transmit to the second node, such as channel information. The channel information may specifically include channel matrix information or precoding matrix information. For ease of explanation, the channel information in this embodiment is usually given as channel matrix information, which does not constitute a specific limitation.

[0058] In implementation, the first node can be a terminal or a network-side device, wherein the network-side device can be an access network device or a core network device, and the second node can be a different terminal or network-side device from the first node. For example: the first node is a terminal and the second node is an access network device; or, the first node is an access network device and the second node is a core network device; or, the first node is a core network device and the second node is another core network device, etc. For ease of explanation, this application embodiment uses the first node as a terminal and the second node as an access network device as an example for illustration, which does not constitute a specific limitation.

[0059] The aforementioned first AI network model can be an encoded AI network model, which is used to compress and / or encode the first information that the first node needs to transmit to obtain first feature information, and then transmit the first feature information. In this way, transmitting the first feature information can reduce transmission overhead and improve the security of the transmitted information compared to transmitting the first information.

[0060] In implementation, the second node can use a decoding AI network model (hereinafter referred to as the fourth AI network model in the second embodiment) to decompress and / or decode the feature information to recover the first information. However, the encoding AI network model used by the first node and the decoding AI network model used by the second node are not jointly trained. Therefore, the encoding AI network model used by the first node and the decoding AI network model used by the second node may not match. In this case, it is necessary to convert the first feature information output by the encoding AI network model used by the first node to obtain the second feature information that can be used or recovered by the decoding AI network model used by the second node.

[0061] In one possible implementation, the first feature information can be converted into second feature information at the first node and transmitted to the second node. In this way, the second node can directly input the second feature information into its own decoding AI network model to recover the aforementioned first information.

[0062] In a specific implementation, the first AI network model and the transformation AI network model can be merged into a single AI network model. For example, the first AI network model and the transformation AI network model can be jointly trained, so that the merged AI network model can output second feature information when the first information is input. The transformation AI network model can convert the first feature information into the second feature information.

[0063] Of course, in implementation, the first node can also use an independent transformation AI network model or a non-AI transformation method to transform the first feature information output by the first AI network model into the second feature information.

[0064] In another possible implementation, the first node can directly send the first feature information to the second node, and the second node can convert and restore the first feature information. In this way, after the second node converts the first feature information into the second feature information, it can input the second feature information into its own decoding AI network model to restore the first information.

[0065] In a specific implementation, the decoding AI network model and the transformation AI network model of the second node can be merged into a single AI network model. For example, the fourth AI network model and the transformation AI network model can be jointly trained, so that the merged AI network model can output first information when the first feature information is input. The transformation AI network model can convert the first feature information into second feature information.

[0066] Of course, in implementation, the second node can also use an independent conversion AI network model or a non-AI conversion method to convert the received first feature information into second feature information that matches its own decoding AI network model.

[0067] Optionally, before the first node processes the first information into first feature information based on the first AI network model, the method further includes:

[0068] The first node is used to train the first AI network model.

[0069] In this implementation, the first node independently trains the first AI network model. For example, the first node performs joint training to obtain the first AI network model and the decoding AI network model that matches the first AI network model (in the following embodiments, the decoding AI network model that matches the first AI network model is referred to as the third AI network model). In use, the first node can retain only the first AI network model and delete the decoding AI network model that matches the first AI network model.

[0070] Similarly, the second node can also independently train the decoding AI network model. For example, the second node can perform joint training to obtain the encoding AI network model and the decoding AI network model that matches the encoding AI network model. In the following embodiments, the encoding AI network model of the second node is referred to as the second AI network model, and the decoding AI network model of the second node is referred to as the fourth AI network model. In use, the second node can retain only the fourth AI network model and delete the encoding AI network model that matches the fourth AI network model.

[0071] For example: Figure 3As shown, assuming the first node uses encoder 1 and the second node uses decoder 2, the decoder 2 matches encoder 2, that is, the feature information output by encoder 2 can be recovered by decoder 2, but the feature information output by encoder 1 cannot be recovered by decoder 2. At this time, the conversion module can be used to convert the feature information output by encoder 1 to obtain the same feature information as the feature information output by encoder 2. The conversion module can be configured with conversion information for converting the feature information output by encoder 1 into the feature information output by encoder 2.

[0072] Optionally, if the second node is a network-side device, the network-side device can obtain the uplink channel estimation result of the nearest Sounding Reference Signal (SRS) triggered by the Channel State Information (CSI) report as the uplink channel matrix, and train the decoding AI network model based on the uplink channel matrix. In this way, the network-side device can obtain the uplink channel matrices of at least two terminals in the same cell, so that the trained decoding AI network model is applicable to the uplink channels of at least two or all terminals in the cell. Based on the mutual difference between the uplink channel and the downlink channel, the decoding AI network model is also applicable to the downlink channels of at least two or all terminals in the cell.

[0073] In other words, network-side equipment can use the same decoding AI network model to decode channel feature information reported by different terminals within the same cell. However, each terminal can correspond to different conversion information. For example, suppose that a cell of the same base station includes terminal A and terminal B, and terminal A performs compression encoding on the channel matrix based on encoding AI network model a to obtain the first channel feature information a, and terminal B performs compression encoding on the channel matrix based on encoding AI network model b to obtain the first channel feature information b. Then, the first channel feature information a can be converted using conversion AI network model a to obtain the second channel feature information that the base station's decoding AI network model can process, and the first channel feature information b can be converted using conversion AI network model b to obtain the second channel feature information that the base station's decoding AI network model can process.

[0074] Of course, in practice, the first node may also use other methods to obtain the first AI network model, such as obtaining the first AI network model issued by the core network device. Similarly, the decoded AI network model of the second node can be trained independently by the second node, or the second node can also obtain the decoded AI network model from the core network device. However, the decoded AI network model obtained by the second node may not match the encoded AI network model obtained by the first node. For example, the core network device may issue an encoded AI network model to the first node according to the capabilities of the first node, and issue a decoded AI network model to the second node according to the capabilities of the second node. In this case, neither the first node nor the second node knows the AI ​​network model used by the other.

[0075] As an optional implementation, before the first node sends the first information to the second node, the method further includes:

[0076] The first node acquires conversion information, wherein the conversion information is used to convert the feature information output by the first AI network model into feature information matched by the AI ​​network model of the second node;

[0077] The first node converts the first feature information into the second feature information based on the conversion information.

[0078] The conversion information can be used to convert the first feature information into the second feature information. For example, the conversion information may include at least one of the following:

[0079] The first instruction information is used to indicate the unit array;

[0080] The second indication information is used to indicate the sparse matrix;

[0081] The third indication information is used to indicate relevant information of the transformation AI network model, which is used to transform the first feature information into the second feature information.

[0082] In one possible implementation, when the first feature information and the second feature information are feature matrices of the same bit size, the positions of "1" and / or "0" in the first feature information can be adjusted using an identity matrix or a sparse matrix to obtain the second feature information.

[0083] In another possible implementation, an AI network model can be used to convert the first feature information into the second feature information. This AI network model is called a transformation AI network model.

[0084] It is worth noting that the number of iterations required to train the transformation AI network model is less than the number of iterations required to train the encoding and / or decoding AI network model. In other words, the training sample data required to train the transformation AI network model is less, and the training complexity is lower. In addition, the transformation AI network model has a smaller model size, and the overhead required to transfer the transformation AI network model is less. Moreover, transferring the transformation AI network model does not involve the encoding and decoding implementation process of each node. Even if it is transferred between the first node, the second node, or even the third node, the confidentiality between the first node and the second node will not be reduced.

[0085] In this embodiment, the first node performs the conversion between the first feature information and the second feature information based on the acquired conversion information.

[0086] In practice, the aforementioned conversion information can be determined by the first node, or the first node can receive the conversion information from other nodes (the second or third node).

[0087] Scene 1

[0088] The first node receives the transformation information from the third node, wherein the third node can be a device other than the first node and the second node. For example, the third node is a device that distributes the AI ​​network model to the first node and the second node, or a device that the first node and the second node both trust. In this way, the third node can know the relevant information of the AI ​​network model used by the first node and the relevant information of the AI ​​network model used by the second node.

[0089] For example, the third node can know the structure, parameters, network model file, and feature information processed by the AI ​​network model used by the first and second nodes respectively, and determine how to transform the first and second feature information accordingly.

[0090] Taking the transformation information including a transformation AI network model as an example, in this embodiment, the transformation AI network model can be trained by a third node. The training sample data required to train the transformation AI network model can come from at least one of the following: a first node, a second node, a node dedicated to collecting training sample data, and a node that does not have the first AI network model or the second AI network model.

[0091] For example, the first node sends a first AI network model to the third node, and the second node sends a second AI network model to the third node. The third node can then input the acquired training sample data into both the first and second AI network models respectively, and analyze the processing results of the first and second AI network models on the same training sample data to obtain a transformed AI network model. Assuming the first AI network model's processing result on the training sample data is the third feature information, and the second AI network model's processing result is the fourth feature information, then when the transformed AI network model inputs the third feature information, it can output the fourth feature information corresponding to the same training sample data.

[0092] Of course, in practice, the first node might directly send the third feature information to the third node, and the second node might directly send the fourth feature information to the third node. Alternatively, one of the first and second nodes might send training sample data and the feature information obtained after processing the training sample data to the third node, while the other of the first and second nodes sends the encoded AI network model to the third node. This also enables the third node to obtain the third and fourth feature information, where the third feature information corresponds one-to-one with the fourth feature obtained from the same training sample data. In this way, when training the transformation AI network model, the third node can use the third feature information as the input to the transformation AI network model and use the corresponding fourth feature information as the output target of the transformation AI network model.

[0093] Scene 2

[0094] The first node determines the transformation information based on auxiliary information related to the first AI network model and the second AI network model.

[0095] Taking the transformation information including the transformation AI network model as an example, in this embodiment, the transformation AI network model can be trained by the first node.

[0096] In one possible implementation, the first node can also train the transformation AI network model based on the third and fourth feature information in the same way as the third node trains the transformation AI network model.

[0097] In another possible implementation, the first node can jointly train the first AI network model and the transformed AI network model based on the fourth feature information and training sample data. During the joint training process, the input information of the AI ​​network model is the training sample data, and the output target is the fourth feature information.

[0098] Optionally, the first node trains the transformation AI network model, including:

[0099] The first node acquires third feature information and receives fourth feature information from the second node, wherein the third feature information is feature information obtained by processing the training sample data based on the first AI network model, and the fourth feature information is feature information obtained by processing the training sample data based on the second AI network model of the second node.

[0100] The first node trains the transformation AI network model based on the third feature information and the fourth feature information, wherein the third feature information serves as the input of the transformation AI network model and the fourth feature information serves as the output target of the transformation AI network model.

[0101] The aforementioned third feature information can be the output information of the first AI network model when the first node inputs the training sample data into the first AI network model.

[0102] It is worth noting that the first node also needs to acquire fourth feature information when training the AI ​​network model. This fourth feature information can come from the second node. For example, the second node and the first node may have prior knowledge of the training sample data and its arrangement or identifier. The second node processes the training sample data into fourth feature information based on its own encoded AI network model and sends this fourth feature information to the first node. This could be done by sending the fourth feature information corresponding to each training sample data according to the aforementioned arrangement of the training sample data, or by indicating the corresponding training sample data or identifier when sending the fourth feature information to the first node. In this way, after the first node can process the aforementioned training sample data into third feature information based on the first AI network model, the correspondence between the third and fourth feature information can be determined based on the same training sample data, training sample data arranged in the same position, and training sample data with the same identifier.

[0103] Of course, the first node can also know the second node's encoded AI network model in advance, and process the training sample data into the fourth feature information based on the second node's encoded AI network model. Here, the method by which the first node obtains the fourth feature information is not specifically limited.

[0104] Scene 3

[0105] The first node receives the transformation information from the second node, wherein the second node can determine the transformation information based on auxiliary information related to the first AI network model and the second AI network model;

[0106] Taking the transformation information, including the transformation AI network model, as an example, in this scenario, the transformation AI network model can be trained by the second node.

[0107] It should be noted that the process of the second node training the transformation AI network model in this embodiment is similar to the process of the first node training the AI ​​network model in Scenario 2 above. The difference is that in this embodiment, when the second node trains the transformation AI network model, it needs to obtain at least one of the following from the first node: third feature information, training sample data, and the first AI network model. The specific process of the second node training the transformation AI network model can be referred to the process of the first node training the transformation AI network model in Scenario 2, and will not be repeated here.

[0108] Optionally, when the above-mentioned transformation AI network model is trained by a second node or a third node, the first node needs to provide auxiliary information to the node training the transformation AI network model. This auxiliary information is used to assist the node training the transformation AI network model in obtaining third feature information.

[0109] For example, before the first node receives the third indication information from the third node or the second node, the method further includes:

[0110] The first node sends third information to the third node or the second node, wherein the third information includes relevant information of the first AI network model, or the third information includes training sample data and third feature information, or the third information includes the third feature information;

[0111] Wherein, the third feature information is the feature information obtained after processing the training sample data based on the first AI network model, the transformation AI network model is trained based on at least one of the third feature information and the fourth feature information, and the fourth feature information is the feature information obtained after processing the training sample data based on the second AI network model possessed by the second node.

[0112] The relevant information of the first AI network model may include at least one of the model structure information, model parameter information, and model file of the first AI network model. The node that trains and transforms the AI ​​network model can realize model inference of the first AI network model based on the relevant information of the first AI network model. For example, it can process the training sample data based on the first AI network model to obtain the third feature information.

[0113] Furthermore, the training sample data used in training the transformation AI network model may be pre-stored by a third node or come from a fourth node, wherein the fourth node may include at least one of the following:

[0114] The first node;

[0115] The second node;

[0116] A node specifically designed for collecting training sample data;

[0117] Nodes that do not have the first AI network model and the second AI network model.

[0118] In the case where the training sample data comes from the first node, the training sample data can be the channel information detected by the first node.

[0119] Optionally, the feature information transmission method further includes:

[0120] The first node estimates the first channel information of the first channel, wherein the training sample data includes the first channel information.

[0121] In this embodiment, the conversion AI network model can be trained based on the channel information of the actual channel estimated by the first node. This can improve the matching degree between the conversion AI network model and the actual channel state of the first node.

[0122] It should be noted that when the conversion AI network model is trained by the third or second node, if the first node only sends the third feature information to the node training the conversion AI network model, the set of training sample data can be fixed, and both the first node and the node training the conversion AI network model are aware of this training sample data set. For example, a fixed training sample dataset can be set up through a protocol and periodic updates. The order of the training sample data in this training sample dataset is fixed, such as the first third feature information reported by the first node corresponding to the first training sample data in the training sample dataset, and so on. In this way, the node training the conversion AI network model can train the conversion AI network model based on the training sample data corresponding to the third feature information. For example, based on the correspondence between the third feature information and the training sample data, and the correspondence between the fourth feature information and the training sample data, the third feature information and the fourth feature information corresponding to the same training sample data can be correlated, so that the correlated third feature information and the fourth feature information can be used as the input information and label for training the conversion AI network model, respectively.

[0123] Of course, in implementation, the first node can also report the ID of the training sample data corresponding to the third feature information at the same time as reporting the third feature information. This ID can represent the ID of the training sample data used in the training sample data set. In this way, the node training the transformation AI network model can find the corresponding training sample data based on this ID, thereby completing the training of the transformation AI network model.

[0124] Similarly, when the training sample data comes from the second node, this training sample data can also be the channel information detected by the second node. For example, assuming the first node is the terminal and the second node is the base station, the training sample data can include the channel information of the downlink channel estimated by the terminal, or the training sample data can include the channel information of the uplink channel estimated by the base station. Based on the dissimilarity of the uplink and downlink channels, the channel information of the uplink channel is close to the channel information of the corresponding downlink channel, thereby making the conversion AI network model trained based on the channel information of the uplink channel match the downlink channel state of the terminal.

[0125] Of course, the nodes that train the transformation AI network model can also obtain training sample data through nodes dedicated to collecting training sample data or nodes that do not have the first AI network model and the second AI network model, which can reduce the complexity of collecting training sample data.

[0126] It is important to note that when the first node sends the third feature information to the node that converts the AI ​​network model, the node that converts the AI ​​network model can obtain the fourth feature information, which is either the training sample data corresponding to the third feature information or the training sample data corresponding to the same third feature information. For example, assuming the third node trains the converting AI network model, the third node can obtain the third feature information from the first node and the fourth feature information from the second node, and the third and fourth feature information are obtained based on the same training sample data; or, assuming the second node jointly trains the converting AI network model and the fourth AI network model, the second AI node can obtain the third feature information and the training sample data, and thereby jointly train an AI network model that can recover the training sample data from the third feature information.

[0127] In this embodiment, a first node processes first information into first feature information based on a first AI network model. The first node sends second information to a second node, the second information including either the first feature information or the second feature information, wherein the second feature information is the feature information transformed from the first feature information. When compressing and restoring the first information between the first node and the second node using an AI network model, if the AI ​​network models used by the first node and the second node are mismatched, resulting in the first feature information compressed by the first node being unusable or unrecoverable by the second node, the first node can transform the first feature information to send second feature information that it can use or recover to the second node. Alternatively, the first node can send the first feature information to the second node, and the second node can transform the first feature information into second feature information that its AI network model can use or recover, and then use the AI ​​network model to recover the second feature information. In this way, the first node and the second node can independently use or train their respective AI network models, thereby reducing the overhead caused by transmitting AI network models. At the same time, the first node and the second node do not need to interact with their respective AI network models, thereby improving the information security of the first node and the second node.

[0128] Please see Figure 4 Another feature information transmission method provided in this application embodiment is executed by a second node, such as... Figure 4 As shown, the feature information transmission method executed by the second node may include the following steps:

[0129] Step 401: The second node receives second information from the first node, wherein the second information includes first feature information or second feature information, wherein the first feature information is feature information obtained by processing the first information based on the first AI network model of the first node, and the second feature information is feature information transformed from the first feature information.

[0130] Step 402: The second node performs recovery processing on the second feature information based on the fourth AI network model to obtain the first information. Wherein, if the second information includes the first feature information, the second feature information is obtained by the second node transforming the first feature information.

[0131] like Figure 4 The method embodiments shown are similar to those described above. Figure 2 The method embodiments shown correspond to those described, except that, as Figure 2 The execution entity of the method embodiment shown is the first node, while as Figure 4The execution entity of the method embodiment shown is the second node, and the specific meanings of the first node, second node, second information, first feature information, second feature information, first AI network model, and fourth AI network model in this embodiment can be found in the following reference: Figure 2 The explanations and descriptions in the method embodiments shown will not be repeated here.

[0132] As an optional implementation, before the second node performs recovery processing on the second feature information based on the fourth AI network model, the feature information transmission method further includes:

[0133] The second node is used to train the fourth AI network model.

[0134] In implementation, the second node can jointly train the second AI network model and the fourth AI network model, and retain the fourth AI network model for decoding the received feature information; or, the second node can independently train the fourth AI network model; or, the second node can jointly train the transformation AI network model and the fourth AI network model.

[0135] The meanings of the second AI network model and the transformation AI network model can be found in the following reference: Figure 2 The explanations and descriptions in the method embodiments shown will not be repeated here.

[0136] Optionally, the first AI network model and the third AI network model are jointly trained, and the third AI network model is matched with the first AI network model and used to recover the feature information processed by the first AI network model.

[0137] And / or,

[0138] The second AI network model is obtained by joint training with the fourth AI network model. The second AI network model is the AI ​​network model that the second node has that matches the second feature information. The fourth AI network model matches the second AI network model and is used to restore the feature information processed by the second AI network model.

[0139] As an optional implementation, before the second node performs recovery processing on the second feature information based on the fourth AI network model, the feature information transmission method further includes:

[0140] The second node acquires conversion information, wherein the conversion information is used to convert the feature information output by the first AI network model into feature information matched by the AI ​​network model of the second node;

[0141] The second node converts the first feature information into the second feature information based on the conversion information.

[0142] In this implementation, the conversion between the first feature information and the second feature information is achieved by the second node.

[0143] As an optional implementation, the conversion information includes at least one of the following:

[0144] The first instruction information is used to indicate the unit array;

[0145] The second indication information is used to indicate the sparse matrix;

[0146] The third indication information is used to indicate relevant information of the transformation AI network model, which is used to transform the first feature information into the second feature information.

[0147] As an optional implementation, when the conversion information includes the third indication information, the second node acquires the conversion information, including:

[0148] The second node is used to train the transformation AI network model; or,

[0149] The second node receives the third indication information from the third node or the first node, wherein the third node is a device other than the first node and the second node.

[0150] As an optional implementation, the second node trains the transformation AI network model, including:

[0151] The second node acquires fourth feature information and receives third feature information from the first node, wherein the third feature information is feature information obtained after processing the training sample data based on the first AI network model, and the fourth feature information is feature information obtained after processing the training sample data based on the second AI network model possessed by the second node.

[0152] The second node trains the transformation AI network model based on the third feature information and the fourth feature information, wherein the third feature information serves as the input of the transformation AI network model and the fourth feature information serves as the output target of the transformation AI network model.

[0153] In this embodiment, the second node serves as the node for training the conversion AI network model. The training sample data required for training the conversion AI network model can come from at least one of the first, second, and third nodes. Furthermore, the specific process by which the second node trains the conversion AI network model based on third and fourth feature information is as follows: Figure 2In the method embodiment shown, the specific process of the first node training the transformation AI network model based on the third and fourth feature information is similar and will not be repeated here.

[0154] As an optional implementation, before the second node receives the third indication information from the third node or the first node, the feature information transmission method further includes:

[0155] The second node sends fourth information to the third node or the first node, wherein the fourth information includes relevant information about the second AI network model possessed by the second node, or the fourth information includes training sample data and fourth feature information, or the fourth information includes the fourth feature information;

[0156] The fourth feature information is the feature information obtained after processing the training sample data based on the second AI network model. The transformation AI network model is trained based on at least one of the third feature information and the fourth feature information. The third feature information is the feature information obtained after processing the training sample data based on the first AI network model.

[0157] In this implementation, the first node or the third node is used as the node for training the transformation AI network model. The specific process can be referred to as follows: Figure 2 The corresponding descriptions in the method embodiments shown will not be repeated here.

[0158] As an optional implementation, the feature information transmission method further includes:

[0159] The second node estimates the second channel information of the second channel, wherein the training sample data includes the second channel information.

[0160] As an optional implementation, when the second node is a network-side device, the second node receives second information from the first node, including:

[0161] The second node receives a target channel state information (CSI) report from at least one terminal, wherein the first feature information includes channel feature information of the target channel carried in the target CSI report, and the at least one terminal includes the first node;

[0162] The method further includes:

[0163] The second node acquires the third channel information of the SRS associated with the target channel;

[0164] The second node trains the fourth AI network model based on the third channel information.

[0165] Among them, the fourth AI network model trained based on the third channel information can recover the channel feature information reported by the terminal. Each terminal uses its own first AI network model to encode the detected channel information to obtain the channel feature information.

[0166] In this embodiment, the network-side device can obtain the uplink channel estimation result of the nearest Sounding Reference Signal (SRS) triggered by the Channel State Information (CSI) report as the uplink channel matrix, and train the decoding AI network model based on the uplink channel matrix. In this way, the network-side device can obtain the uplink channel matrices of at least two terminals in the same cell, so that the trained decoding AI network model is applicable to the uplink channels of at least two or all terminals in the cell. Based on the mutuality of uplink and downlink channels, the decoding AI network model is also applicable to the downlink channels of at least two or all terminals in the cell.

[0167] It should be noted that although the same decoding AI network model of the network-side device is applicable to the feature information reported by different terminals, the feature information reported by the terminal can still be converted and processed in advance using a conversion AI network model that corresponds one-to-one with each terminal, so as to obtain the feature information that the same decoding AI network model can process.

[0168] The feature information transmission method executed by the second node provided in this application embodiment is similar to... Figure 2 The feature information transmission method executed by the first node provided in the illustrated method embodiment, together with the first node and the second node, can jointly achieve the beneficial effect of enabling the first node and the second node to independently use or train their respective AI network models, thereby reducing the overhead caused by transmitting AI network models. At the same time, the first node and the second node do not need to interact with their respective AI network models, thereby improving the information security of the first node and the second node.

[0169] The feature information transmission method provided in this application can be executed by a feature information transmission device. This application uses a feature information transmission device executing the feature information transmission method as an example to illustrate the feature information transmission device provided in this application.

[0170] Please see Figure 5 The feature information transmission device provided in this application embodiment can be a device within a first node, such as... Figure 5 As shown, the feature information transmission device 500 may include the following modules:

[0171] The first processing module 501 is used to process the first information into first feature information based on the first AI network model;

[0172] The first sending module 502 is used to send second information to the second node. The second information includes the first feature information or the second feature information, wherein the second feature information is feature information converted from the first feature information.

[0173] Optionally, the feature information transmission device 500 further includes:

[0174] The first training module is used to train the first AI network model.

[0175] Optionally, the first AI network model and the third AI network model are jointly trained, and the third AI network model is matched with the first AI network model and used to recover the feature information processed by the first AI network model.

[0176] And / or,

[0177] The second AI network model is obtained by joint training with the fourth AI network model. The second AI network model is the AI ​​network model that the second node has that matches the second feature information. The fourth AI network model matches the second AI network model and is used to restore the feature information processed by the second AI network model.

[0178] Optionally, the feature information transmission device 500 further includes:

[0179] The second acquisition module is used to acquire conversion information, wherein the conversion information is used to convert the feature information output by the first AI network model into feature information matched by the AI ​​network model of the second node;

[0180] The first conversion module is used to convert the first feature information into the second feature information according to the conversion information.

[0181] Optionally, the conversion information includes at least one of the following:

[0182] The first instruction information is used to indicate the unit array;

[0183] The second indication information is used to indicate the sparse matrix;

[0184] The third indication information is used to indicate relevant information of the transformation AI network model, which is used to transform the first feature information into the second feature information.

[0185] Optionally, when the conversion information includes the third indication information, the second acquisition module is used to:

[0186] The transformation AI network model is obtained through training; or,

[0187] Receive the third indication information from a third node or the second node, wherein the third node is a device other than the first node and the second node.

[0188] Optionally, the second acquisition module includes:

[0189] The first acquisition unit is used to acquire third feature information and receive fourth feature information from the second node, wherein the third feature information is feature information obtained by processing the training sample data based on the first AI network model, and the fourth feature information is feature information obtained by processing the training sample data based on the second AI network model of the second node.

[0190] The first training unit is used to train the transformation AI network model based on the third feature information and the fourth feature information, wherein the third feature information is used as the input of the transformation AI network model and the fourth feature information is used as the output target of the transformation AI network model.

[0191] Optionally, the feature information transmission device 500 further includes:

[0192] The third sending module is used to send third information to the third node or the second node, wherein the third information includes relevant information of the first AI network model, or the third information includes training sample data and third feature information, or the third information includes the third feature information;

[0193] Wherein, the third feature information is the feature information obtained after processing the training sample data based on the first AI network model, the transformation AI network model is trained based on at least one of the third feature information and the fourth feature information, and the fourth feature information is the feature information obtained after processing the training sample data based on the second AI network model possessed by the second node.

[0194] Optionally, the feature information transmission device 500 further includes:

[0195] The first channel estimation module is used to estimate the first channel information of the first channel, wherein the training sample data includes the first channel information.

[0196] The feature information transmission device 500 provided in this application embodiment can achieve the following: Figure 2The various processes implemented by the first node in the method embodiment shown can achieve the same beneficial effects, and will not be described again here to avoid repetition.

[0197] Please see Figure 6 Another feature information transmission device provided in this application embodiment can be a device within the second node, such as... Figure 6 As shown, the feature information transmission device 600 may include the following modules:

[0198] The first receiving module 601 is used to receive second information from the first node, wherein the second information includes first feature information or second feature information, wherein the first feature information is feature information obtained by processing the first information based on the first AI network model of the first node, and the second feature information is feature information converted from the first feature information.

[0199] The second processing module 602 is used to perform recovery processing on the second feature information based on the fourth AI network model to obtain the first information, wherein, when the second information includes the first feature information, the second feature information is obtained by the second node transforming the first feature information.

[0200] Optionally, the feature information transmission device 600 further includes:

[0201] The second training module is used to train the fourth AI network model.

[0202] Optionally, the first AI network model and the third AI network model are jointly trained, and the third AI network model is matched with the first AI network model and used to recover the feature information processed by the first AI network model.

[0203] And / or,

[0204] The second AI network model is obtained by joint training with the fourth AI network model. The second AI network model is the AI ​​network model that the second node has that matches the second feature information. The fourth AI network model matches the second AI network model and is used to restore the feature information processed by the second AI network model.

[0205] Optionally, the feature information transmission device 600 further includes:

[0206] The third acquisition module is used to acquire conversion information, wherein the conversion information is used to convert the feature information output by the first AI network model into feature information matched by the AI ​​network model of the second node;

[0207] The second conversion module is used to convert the first feature information into the second feature information according to the conversion information.

[0208] Optionally, the conversion information includes at least one of the following:

[0209] The first instruction information is used to indicate the unit array;

[0210] The second indication information is used to indicate the sparse matrix;

[0211] The third indication information is used to indicate relevant information of the transformation AI network model, which is used to transform the first feature information into the second feature information.

[0212] Optionally, when the conversion information includes the third indication information, the third acquisition module is used to:

[0213] The transformation AI network model is obtained through training; or,

[0214] Receive the third indication information from a third node or the first node, wherein the third node is a device other than the first node and the second node.

[0215] Optionally, the third acquisition module includes:

[0216] The second acquisition unit is used to acquire fourth feature information and receive third feature information from the first node, wherein the third feature information is feature information obtained by processing the training sample data based on the first AI network model, and the fourth feature information is feature information obtained by processing the training sample data based on the second AI network model of the second node.

[0217] The second training unit is used to train the transformation AI network model based on the third feature information and the fourth feature information, wherein the third feature information is used as the input of the transformation AI network model and the fourth feature information is used as the output target of the transformation AI network model.

[0218] Optionally, the feature information transmission device 600 further includes:

[0219] The fourth sending module is used to send fourth information to the third node or the first node, wherein the fourth information includes relevant information about the second AI network model possessed by the second node, or the fourth information includes training sample data and fourth feature information, or the fourth information includes the fourth feature information;

[0220] The fourth feature information is the feature information obtained after processing the training sample data based on the second AI network model. The transformation AI network model is trained based on at least one of the third feature information and the fourth feature information. The third feature information is the feature information obtained after processing the training sample data based on the first AI network model.

[0221] Optionally, the feature information transmission device 600 further includes:

[0222] The second channel estimation module is used to estimate the second channel information of the second channel, wherein the training sample data includes the second channel information.

[0223] Optionally, when the second node is a network-side device, the first receiving module 601 is used to:

[0224] Receive a target channel state information (CSI) report from at least one terminal, wherein the first feature information includes channel feature information of the target channel carried in the target CSI report, and the at least one terminal includes the first node;

[0225] The feature information transmission device 600 also includes:

[0226] The fourth acquisition module is used to acquire the third channel information of the SRS related to the target channel;

[0227] The third training module is used to train the fourth AI network model based on the third channel information.

[0228] The feature information transmission device 600 provided in this application embodiment can achieve the following: Figure 4 The various processes implemented by the second node in the method embodiment shown can achieve the same beneficial effects, and will not be described again here to avoid repetition.

[0229] The feature information transmission device in this application embodiment can be an electronic device, such as an electronic device with an operating system, or a component in an electronic device, such as an integrated circuit or a chip. The electronic device can be a terminal or a network-side device, or other devices besides terminals and network-side devices. For example, the terminal can include, but is not limited to, the type of terminal 11 listed above, and the network-side device can include, but is not limited to, the type of network-side device 12 listed above. Other devices can be servers, network-attached storage (NAS), etc., and this application embodiment does not specifically limit them.

[0230] The feature information transmission device provided in this application embodiment can achieve... Figure 2 or Figure 4The various processes implemented in the method embodiment shown achieve the same technical effect, and will not be described again here to avoid repetition.

[0231] Please see Figure 7 The present application provides a method for determining conversion information, in which the execution entity is a third node, such as... Figure 7 As shown, the method for determining the transformation information executed by the third node may include the following steps:

[0232] Step 701: The third node obtains the third feature information obtained by the first AI network model after processing the training sample data, and obtains the fourth feature information obtained by the second AI network model after processing the training sample data, wherein the first AI network model is the AI ​​network model possessed by the first node, and the second AI network model is the AI ​​network model possessed by the second node.

[0233] Step 702: The third node determines conversion information based on the third feature information and the fourth feature information, wherein the conversion information is used to convert the third feature information corresponding to the target training sample into the fourth feature information corresponding to the target training sample, and the target training sample is any sample in the training sample data.

[0234] Step 703: The third node sends the conversion information to at least one of the first node and the second node.

[0235] like Figure 7 The meanings and functions of the first node, second node, third node, first AI network model, second AI network model, third feature information, fourth feature information, transformation information, and target training samples in the method embodiment shown can be referred to as follows: Figure 2 Explanation of the method embodiments shown.

[0236] In this embodiment, the implementation method for the third node to determine the transformation information is mainly described. For example, the third node updates the third feature information and the fourth feature information to train the transformation AI network model, and then sends it to the first node or the second node. The process of the third node determining the transformation information can be referred to as follows: Figure 2 In the method embodiment shown, the explanation of the third node determining the conversion information will not be repeated here.

[0237] As an optional implementation, the conversion information includes at least one of the following:

[0238] The first instruction information is used to indicate the unit array;

[0239] The second indication information is used to indicate the sparse matrix;

[0240] The third indication information is used to indicate relevant information of the transformed AI network model, wherein the transformed AI network model is used to transform the first feature information into the second feature information, wherein the first feature information matches the first AI network model, and the second feature information matches the second AI network model.

[0241] As an optional implementation, the third node acquires the third feature information obtained by the first AI network model after processing the training sample data, including:

[0242] The third node receives fifth information from the first node, the fifth information including relevant information about the first AI network model;

[0243] The third node inputs the training sample data into the first AI network model and obtains the third feature information output by the first AI network model;

[0244] And / or,

[0245] The third node acquires the fourth feature information obtained after the second AI network model processes the training sample data, including:

[0246] The third node receives sixth information from the second node, the sixth information including relevant information about the second AI network model;

[0247] The third node inputs the training sample data into the second AI network model and obtains the fourth feature information output by the second AI network model.

[0248] In this implementation, the third node can obtain the encoded AI network models of the first and second nodes respectively, and process the training sample data based on the encoded AI network models of the first and second nodes respectively, so as to obtain the third feature information and the fourth feature information required to determine the transformation information.

[0249] As an optional implementation, the training sample data is pre-stored in the third node, or the training sample data comes from a fourth node, the fourth node comprising at least one of the following:

[0250] The first node;

[0251] The second node;

[0252] A node specifically designed for collecting training sample data;

[0253] Nodes that do not have the first AI network model and the second AI network model.

[0254] As an optional implementation, the third node acquires the third feature information obtained by the first AI network model after processing the training sample data, including:

[0255] The third node receives training sample data from the first node and the third feature information, wherein the third feature information is the feature information output by the first AI network model after the training sample data is input into the first AI network model;

[0256] The third node acquires the fourth feature information obtained after the second AI network model processes the training sample data, including:

[0257] The third node inputs the training sample data into the second AI network model and obtains the fourth feature information output by the second AI network model.

[0258] As an optional implementation, the third node acquires the fourth feature information obtained after the second AI network model processes the training sample data, including:

[0259] The third node receives training sample data from the second node and the fourth feature information, wherein the fourth feature information is the feature information output by the second AI network model after the training sample data is input into the second AI network model;

[0260] The third node acquires the third feature information obtained by the first AI network model after processing the training sample data, including:

[0261] The third node inputs the training sample data into the first AI network model and obtains the third feature information output by the first AI network model.

[0262] For example, suppose the first node is a terminal, the second node is a base station, and the third node can be a device trusted by both the terminal and the base station. In this case, the scheme for training the AI ​​network model on the third node could be:

[0263] The terminal and base station send relevant information from their respective encoded AI network models to the third node. The third node uses its stored channel information to obtain the encoded third and fourth feature information based on the encoded AI network models of the terminal and base station, respectively. It then uses the third feature information as input and the fourth feature information (the so-called label) as input to train the transformation AI network model; or...

[0264] The terminal sends the actual estimated channel information and the third feature information processed by the terminal's coding AI network model to the third node. The base station sends the relevant information of its own coding AI network model to the third node. The third node processes the channel information reported by the terminal through the base station's coding AI network model to obtain the fourth feature information. The third feature information is used as input, and the fourth feature information is used as a label to train the transformation AI network model.

[0265] In this embodiment, the third node can obtain the third feature information obtained by the first AI network model of the first node after processing the training sample data, and obtain the fourth feature information obtained by the second AI network model of the second node after processing the training sample data. Based on the difference between the two, the conversion information is determined so that the output result of the first AI network model can be converted into the output result of the second AI network model.

[0266] The feature information transmission method provided in this application can be executed by a feature information transmission device. This application uses a feature information transmission device executing the feature information transmission method as an example to illustrate the feature information transmission device provided in this application.

[0267] Please see Figure 8 The conversion information determination device provided in this application embodiment can be a device within a third node, such as... Figure 8 As shown, the conversion information determining device 800 may include the following modules:

[0268] The first acquisition module 801 is used to acquire the third feature information obtained by the first AI network model after processing the training sample data, and to acquire the fourth feature information obtained by the second AI network model after processing the training sample data, wherein the first AI network model is the AI ​​network model possessed by the first node, and the second AI network model is the AI ​​network model possessed by the second node.

[0269] The first determining module 802 is used to determine conversion information based on the third feature information and the fourth feature information, wherein the conversion information is used to convert the third feature information corresponding to the target training sample into the fourth feature information corresponding to the target training sample, and the target training sample is any sample in the training sample data;

[0270] The second sending module 803 is used to send the conversion information to at least one of the first node and the second node.

[0271] Optionally, the conversion information includes at least one of the following:

[0272] The first instruction information is used to indicate the unit array;

[0273] The second indication information is used to indicate the sparse matrix;

[0274] The third indication information is used to indicate relevant information of the transformed AI network model, wherein the transformed AI network model is used to transform the first feature information into the second feature information, wherein the first feature information matches the first AI network model, and the second feature information matches the second AI network model.

[0275] Optionally, the first acquisition module 801 includes:

[0276] The first receiving unit is configured to receive fifth information from the first node, the fifth information including relevant information of the first AI network model;

[0277] The first processing unit is used to input training sample data into the first AI network model and obtain the third feature information output by the first AI network model.

[0278] And / or,

[0279] The first acquisition module 801 also includes:

[0280] The second receiving unit is used to receive sixth information from the second node, the sixth information including relevant information of the second AI network model;

[0281] The second processing unit is used to input training sample data into the second AI network model and obtain the fourth feature information output by the second AI network model.

[0282] Optionally, the training sample data is pre-stored in the third node, or the training sample data comes from a fourth node, the fourth node including at least one of the following:

[0283] The first node;

[0284] The second node;

[0285] A node specifically designed for collecting training sample data;

[0286] Nodes that do not have the first AI network model and the second AI network model.

[0287] Optionally, the first acquisition module 801 includes:

[0288] The third receiving unit is used to receive training sample data and the third feature information from the first node, wherein the third feature information is feature information output by the first AI network model after the training sample data is input into the first AI network model;

[0289] The third processing unit is used to input the training sample data into the second AI network model and obtain the fourth feature information output by the second AI network model.

[0290] Optionally, the first acquisition module 801 includes:

[0291] The fourth receiving unit is used to receive training sample data and the fourth feature information from the second node, wherein the fourth feature information is feature information output by the second AI network model after the training sample data is input into the second AI network model;

[0292] The fourth processing unit is used to input the training sample data into the first AI network model and obtain the third feature information output by the first AI network model.

[0293] The conversion information determining device 800 provided in this application embodiment can achieve the following: Figure 7 The various processes implemented by the third node in the method embodiment shown can achieve the same beneficial effects, and will not be described again here to avoid repetition.

[0294] Optional, such as Figure 9 As shown, this application embodiment also provides a communication device 900, including a processor 901 and a memory 902. The memory 902 stores a program or instructions that can run on the processor 901. For example, when the communication device 900 acts as a first node, the program or instructions executed by the processor 901 implement the following... Figure 2 The various steps of the illustrated method embodiment can achieve the same technical effect. When the communication device 900 acts as a second node, the program or instructions executed by the processor 901 implement the following... Figure 4 The various steps of the method embodiment shown can achieve the same technical effect. When the communication device 900 acts as a third node, the program or instructions executed by the processor 901 implement the following... Figure 7 The steps of the method embodiment shown are the same and can achieve the same technical effect. To avoid repetition, they will not be described again here.

[0295] This application also provides a communication device, including a processor and a communication interface.

[0296] In one implementation, when the communication device acts as a first node, the processor is used to process first information into first feature information based on a first AI network model; the communication interface is used to send second information to a second node, the second information including either the first feature information or the second feature information, wherein the second feature information is feature information converted from the first feature information; or...

[0297] In another embodiment, when the communication device acts as a second node, the communication interface is used to receive second information from the first node, wherein the second information includes first feature information or second feature information, wherein the first feature information is feature information obtained by processing the first information based on a first AI network model possessed by the first node, and the second feature information is feature information converted from the first feature information; the processor is used to perform recovery processing on the second feature information based on a fourth AI network model to obtain the first information, wherein, when the second information includes the first feature information, the second feature information is obtained by the second node by converting the first feature information.

[0298] In another embodiment, when the communication device acts as a third node, the communication interface is used to acquire third feature information obtained by the first AI network model after processing the training sample data, and to acquire fourth feature information obtained by the second AI network model after processing the training sample data, wherein the first AI network model is an AI network model possessed by the first node, and the second AI network model is an AI network model possessed by the second node; the processor is used to determine conversion information based on the third feature information and the fourth feature information, wherein the conversion information is used to convert the third feature information corresponding to the target training sample into the fourth feature information corresponding to the target training sample, wherein the target training sample is any sample in the training sample data; the communication interface is also used to send the conversion information to at least one of the first node and the second node.

[0299] This communication device embodiment is similar to... Figure 2 or Figure 4 or Figure 7 The method shown corresponds to the embodiment. Figure 2 or Figure 4 or Figure 7 The various implementation processes and methods of the illustrated method embodiments can be applied to the communication device embodiments and achieve the same technical effects.

[0300] This application embodiment also provides a readable storage medium storing a program or instructions that, when executed by a processor, implement... Figure 2 or Figure 4 or Figure 7 The various processes of the method embodiments shown can achieve the same technical effect, and will not be described again here to avoid repetition.

[0301] The processor is the processor in the terminal described in the above embodiments. The readable storage medium includes computer-readable storage media, such as computer read-only memory (ROM), random access memory (RAM), magnetic disk, or optical disk.

[0302] This application embodiment also provides a chip, the chip including a processor and a communication interface, the communication interface being coupled to the processor, the processor being used to run programs or instructions to implement, as described above. Figure 2 or Figure 4 or Figure 7 The various processes of the method embodiments shown can achieve the same technical effect, and will not be described again here to avoid repetition.

[0303] It should be understood that the chip mentioned in the embodiments of this application may also be referred to as a system-on-a-chip, system chip, chip system, or system-on-a-chip, etc.

[0304] This application embodiment also provides a computer program / program product, which is stored in a storage medium and executed by at least one processor to implement the following: Figure 2 or Figure 4 or Figure 7 The various processes of the method embodiments shown can achieve the same technical effect, and will not be described again here to avoid repetition.

[0305] This application also provides a communication system, including: a first node, a second node, and a third node, wherein the first node can be used to perform, for example... Figure 2 The steps of the feature information transmission method shown are described, and the second node can be used to perform, as follows: Figure 4 The steps of the feature information transmission method shown are described, and the third node can be used to perform, as follows: Figure 7 The steps of the method for determining conversion information are shown.

[0306] It should be noted that, in this document, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Without further limitations, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes that element. Furthermore, it should be noted that the scope of the methods and apparatuses in the embodiments of this application is not limited to performing functions in the order shown or discussed, but may also include performing functions substantially simultaneously or in the reverse order, depending on the functions involved. For example, the described methods may be performed in a different order than described, and various steps may be added, omitted, or combined. Additionally, features described with reference to certain examples may be combined in other examples.

[0307] Through the above description of the embodiments, those skilled in the art can clearly understand that the methods of the above embodiments can be implemented by means of software plus necessary general-purpose hardware platforms. Of course, they can also be implemented by hardware, but in many cases the former is a better implementation method. Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, can be embodied in the form of a computer software product. This computer software product is stored in a storage medium (such as ROM / RAM, magnetic disk, optical disk) and includes several instructions to cause a terminal (which may be a mobile phone, computer, server, air conditioner, or network device, etc.) to execute the methods described in the various embodiments of this application.

[0308] The embodiments of this application have been described above with reference to the accompanying drawings. However, this application is not limited to the specific embodiments described above. The specific embodiments described above are merely illustrative and not restrictive. Those skilled in the art can make many other forms under the guidance of this application without departing from the spirit and scope of the claims, and all of these forms are within the protection scope of this application.

Claims

1. A method for transmitting feature information, characterized in that, include: The first node, based on the first AI network model, processes the first information into first feature information; The first node sends second information to the second node. The second information includes the first feature information or the second feature information. The second feature information is the feature information after the first feature information is converted. The second feature information is obtained by the second node through recovery processing based on the fourth AI network model to obtain the first information. When the second information includes the first feature information, the first feature information is transformed by the second node to obtain the second feature information; or, when the second information includes the second feature information, the second feature information is transformed by the first node to obtain the first feature information.

2. The method according to claim 1, characterized in that, Before the first node processes the first information into first feature information based on the first AI network model, the method further includes: The first node is used to train the first AI network model.

3. The method according to claim 2, characterized in that, The first AI network model and the third AI network model are jointly trained. The third AI network model is matched with the first AI network model and is used to recover the feature information processed by the first AI network model. And / or, The second AI network model is obtained by joint training with the fourth AI network model. The second AI network model is the AI ​​network model that the second node has that matches the second feature information. The fourth AI network model matches the second AI network model and is used to restore the feature information processed by the second AI network model.

4. The method according to any one of claims 1 to 3, characterized in that, Before the first node sends the first information to the second node, the method further includes: The first node acquires conversion information, wherein the conversion information is used to convert the feature information output by the first AI network model into feature information matched by the AI ​​network model of the second node; The first node converts the first feature information into the second feature information based on the conversion information.

5. The method according to claim 4, characterized in that, The conversion information includes at least one of the following: The first instruction information is used to indicate the unit array; The second indication information is used to indicate the sparse matrix; The third indication information is used to indicate relevant information of the transformation AI network model, which is used to transform the first feature information into the second feature information.

6. The method according to claim 5, characterized in that, When the conversion information includes the third indication information, the first node acquires the conversion information, including: The first node is used to train the transformation AI network model; or, The first node receives the third indication information from the third node or the second node, wherein the third node is a device other than the first node and the second node.

7. The method according to claim 6, characterized in that, The first node trains the transformation AI network model, which includes: The first node acquires third feature information and receives fourth feature information from the second node, wherein the third feature information is feature information obtained by processing the training sample data based on the first AI network model, and the fourth feature information is feature information obtained by processing the training sample data based on the second AI network model of the second node. The first node trains the transformation AI network model based on the third feature information and the fourth feature information, wherein the third feature information serves as the input of the transformation AI network model and the fourth feature information serves as the output target of the transformation AI network model.

8. The method according to claim 6, characterized in that, Before the first node receives the third indication information from the third node or the second node, the method further includes: The first node sends third information to the third node or the second node, wherein the third information includes relevant information of the first AI network model, or the third information includes training sample data and third feature information, or the third information includes the third feature information; Wherein, the third feature information is the feature information obtained after processing the training sample data based on the first AI network model, the transformation AI network model is trained based on at least one of the third feature information and the fourth feature information, and the fourth feature information is the feature information obtained after processing the training sample data based on the second AI network model possessed by the second node.

9. The method according to claim 7 or 8, characterized in that, The method further includes: The first node estimates the first channel information of the first channel, wherein the training sample data includes the first channel information.

10. A method for transmitting feature information, characterized in that, include: The second node receives second information from the first node, wherein the second information includes first feature information or second feature information, wherein the first feature information is feature information obtained by processing the first information based on the first AI network model of the first node, and the second feature information is feature information transformed from the first feature information; The second node performs recovery processing on the second feature information based on the fourth AI network model to obtain the first information. Wherein, if the second information includes the first feature information, the second feature information is obtained by the second node transforming the first feature information; or, if the second information includes the second feature information, the second feature information is obtained by the first node transforming the first feature information.

11. The method according to claim 10, characterized in that, Before the second node performs recovery processing on the second feature information based on the fourth AI network model, the method further includes: The second node is used to train the fourth AI network model.

12. The method according to claim 11, characterized in that, The first AI network model and the third AI network model are jointly trained. The third AI network model is matched with the first AI network model and is used to recover the feature information processed by the first AI network model. And / or, The second AI network model is obtained by joint training with the fourth AI network model. The second AI network model is the AI ​​network model that the second node has that matches the second feature information. The fourth AI network model matches the second AI network model and is used to restore the feature information processed by the second AI network model.

13. The method according to any one of claims 10 to 12, characterized in that, Before the second node performs recovery processing on the second feature information based on the fourth AI network model, the method further includes: The second node acquires conversion information, wherein the conversion information is used to convert the feature information output by the first AI network model into feature information matched by the AI ​​network model of the second node; The second node converts the first feature information into the second feature information based on the conversion information.

14. The method according to claim 13, characterized in that, The conversion information includes at least one of the following: The first instruction information is used to indicate the unit array; The second indication information is used to indicate the sparse matrix; The third indication information is used to indicate relevant information of the transformation AI network model, which is used to transform the first feature information into the second feature information.

15. The method according to claim 14, characterized in that, When the conversion information includes the third indication information, the second node acquires the conversion information, including: The second node is used to train the transformation AI network model; or, The second node receives the third indication information from the third node or the first node, wherein the third node is a device other than the first node and the second node.

16. The method according to claim 15, characterized in that, The second node trains the transformation AI network model, including: The second node acquires fourth feature information and receives third feature information from the first node, wherein the third feature information is feature information obtained after processing the training sample data based on the first AI network model, and the fourth feature information is feature information obtained after processing the training sample data based on the second AI network model possessed by the second node. The second node trains the transformation AI network model based on the third feature information and the fourth feature information, wherein the third feature information serves as the input of the transformation AI network model and the fourth feature information serves as the output target of the transformation AI network model.

17. The method according to claim 15, characterized in that, Before the second node receives the third indication information from the third node or the first node, the method further includes: The second node sends fourth information to the third node or the first node, wherein the fourth information includes relevant information about the second AI network model possessed by the second node, or the fourth information includes training sample data and fourth feature information, or the fourth information includes the fourth feature information; The fourth feature information is the feature information obtained after processing the training sample data based on the second AI network model. The transformation AI network model is trained based on at least one of the third feature information and the fourth feature information. The third feature information is the feature information obtained after processing the training sample data based on the first AI network model.

18. The method according to claim 16 or 17, characterized in that, The method further includes: The second node estimates the second channel information of the second channel, wherein the training sample data includes the second channel information.

19. The method according to claim 16 or 17, characterized in that, When the second node is a network-side device, the second node receives second information from the first node, including: The second node receives a target channel state information (CSI) report from at least one terminal, wherein the first feature information includes channel feature information of the target channel carried in the target CSI report, and the at least one terminal includes the first node; The method further includes: The second node acquires the third channel information of the SRS associated with the target channel; The second node trains the fourth AI network model based on the third channel information.

20. A method for determining conversion information, characterized in that, include: The third node obtains the third feature information obtained by the first AI network model after processing the training sample data, and obtains the fourth feature information obtained by the second AI network model after processing the training sample data, wherein the first AI network model is the AI ​​network model possessed by the first node, and the second AI network model is the AI ​​network model possessed by the second node. The third node determines conversion information based on the third feature information and the fourth feature information, wherein the conversion information is used to convert the third feature information corresponding to the target training sample into the fourth feature information corresponding to the target training sample, and the target training sample is any sample in the training sample data; The third node sends the conversion information to at least one of the first node and the second node; Wherein, at least one of the first node and the second node performs transformation processing on the first feature information based on the transformation information to obtain the second feature information, the first feature information is obtained by the first node processing the first information based on the first AI network model, and the second feature information is used by the second node to perform recovery processing based on the fourth AI network model to obtain the first information.

21. The method according to claim 20, characterized in that, The conversion information includes at least one of the following: The first instruction information is used to indicate the unit array; The second indication information is used to indicate the sparse matrix; The third indication information is used to indicate relevant information of the transformed AI network model, wherein the transformed AI network model is used to transform the first feature information into the second feature information, wherein the first feature information matches the first AI network model, and the second feature information matches the second AI network model.

22. The method according to claim 20, characterized in that, The third node acquires the third feature information obtained by the first AI network model after processing the training sample data, including: The third node receives fifth information from the first node, the fifth information including relevant information about the first AI network model; The third node inputs the training sample data into the first AI network model and obtains the third feature information output by the first AI network model; And / or, The third node acquires the fourth feature information obtained after the second AI network model processes the training sample data, including: The third node receives sixth information from the second node, the sixth information including relevant information about the second AI network model; The third node inputs the training sample data into the second AI network model and obtains the fourth feature information output by the second AI network model.

23. The method according to claim 22, characterized in that, The training sample data is pre-stored in the third node, or the training sample data comes from a fourth node, the fourth node including at least one of the following: The first node; The second node; A node specifically designed for collecting training sample data; Nodes that do not have the first AI network model and the second AI network model.

24. The method according to claim 20, characterized in that, The third node acquires the third feature information obtained by the first AI network model after processing the training sample data, including: The third node receives training sample data from the first node and the third feature information, wherein the third feature information is the feature information output by the first AI network model after the training sample data is input into the first AI network model; The third node acquires the fourth feature information obtained after the second AI network model processes the training sample data, including: The third node inputs the training sample data into the second AI network model and obtains the fourth feature information output by the second AI network model.

25. The method according to claim 20, characterized in that, The third node acquires the fourth feature information obtained after the second AI network model processes the training sample data, including: The third node receives training sample data from the second node and the fourth feature information, wherein the fourth feature information is the feature information output by the second AI network model after the training sample data is input into the second AI network model; The third node acquires the third feature information obtained by the first AI network model after processing the training sample data, including: The third node inputs the training sample data into the first AI network model and obtains the third feature information output by the first AI network model.

26. A feature information transmission device, characterized in that, Applied to the first node, the device includes: The first processing module is used to process the first information into first feature information based on the first AI network model; The first sending module is used to send second information to the second node. The second information includes the first feature information or the second feature information, wherein the second feature information is the feature information after the first feature information is converted, and the second feature information is obtained by the second node through recovery processing based on the fourth AI network model to obtain the first information. When the second information includes the first feature information, the first feature information is transformed by the second node to obtain the second feature information; or, when the second information includes the second feature information, the second feature information is transformed by the first node to obtain the first feature information.

27. A feature information transmission device, characterized in that, Applied to the second node, the device includes: The first receiving module is configured to receive second information from the first node, wherein the second information includes first feature information or second feature information, wherein the first feature information is feature information obtained by processing the first information based on the first AI network model of the first node, and the second feature information is feature information converted from the first feature information; The second processing module is used to recover the second feature information based on the fourth AI network model to obtain the first information. In the case where the second information includes the first feature information, the second feature information is obtained by the second node converting the first feature information; or, in the case where the second information includes the second feature information, the second feature information is obtained by the first node converting the first feature information.

28. A device for determining conversion information, characterized in that, Applied to a third node, the device includes: The first acquisition module is used to acquire the third feature information obtained by the first AI network model after processing the training sample data, and to acquire the fourth feature information obtained by the second AI network model after processing the training sample data, wherein the first AI network model is the AI ​​network model possessed by the first node, and the second AI network model is the AI ​​network model possessed by the second node. The first determining module is used to determine conversion information based on the third feature information and the fourth feature information, wherein the conversion information is used to convert the third feature information corresponding to the target training sample into the fourth feature information corresponding to the target training sample, and the target training sample is any sample in the training sample data; The second sending module is used to send the conversion information to at least one of the first node and the second node; Wherein, at least one of the first node and the second node performs transformation processing on the first feature information based on the transformation information to obtain the second feature information, the first feature information is obtained by the first node processing the first information based on the first AI network model, and the second feature information is used by the second node to perform recovery processing based on the fourth AI network model to obtain the first information.

29. A communication device, characterized in that, It includes a processor and a memory, the memory storing a program or instructions that can run on the processor, the program or instructions being executed by the processor to implement the steps of the feature information transmission method as claimed in any one of claims 1 to 9, or the steps of the feature information transmission method as claimed in any one of claims 10 to 19, or the steps of the conversion information determination method as claimed in any one of claims 20 to 25.

30. A readable storage medium, characterized in that, The readable storage medium stores a program or instructions that, when executed by a processor, implement the steps of the feature information transmission method as described in any one of claims 1 to 9, or the steps of the feature information transmission method as described in any one of claims 10 to 19, or the steps of the conversion information determination method as described in any one of claims 20 to 25.

Images