Domain management node and device node

Abstract

The invention discloses a domain management node and a device node. The domain management node comprises a transceiver; the processor is coupled to the transceiver, and the processor is configured to obtain relevant information of the model set, and the relevant information of the model set at least comprises function description information of the air interface and application condition information of the air interface; generating a first message based on the related information of the model set; a first message is transmitted via the transceiver, the first message including support capability related indication information of the domain management node or related information of the model set.

Description

Technical Field

[0001] This application relates to the field of communication technologies, and in particular, to a domain management node and a device node. Background Art

[0002] As the complexity of communication systems continues to increase, methods based on rules and optimization algorithms have become difficult to meet the needs of modern communication. Models based on Artificial Intelligence (AI) or Machine Learning (ML) can learn and extract features from large amounts of data, thereby achieving intelligent management of communication systems, and then significantly improving communication efficiency, reducing power consumption, and optimizing resource allocation.

[0003] Therefore, in order to meet the wireless communication needs in emerging scenarios and strengthen the management of AI / ML models, it has become an indispensable part of communication systems. Summary of the Invention

[0004] This application mainly provides a domain management node and a device node. The technical solution of this application is implemented as follows:

[0005] In a first aspect, an embodiment of this application provides a domain management node, which includes:

[0006] A transceiver; and

[0007] A processor, which is coupled to the transceiver, wherein the processor is configured to:

[0008] Obtain relevant information of a model set, where the relevant information of the model set at least includes function description information of the radio interface and application condition information of the radio interface;

[0009] Generate a first message based on the relevant information of the model set;

[0010] Transmit the first message via the transceiver, where the first message includes indication information related to the support capabilities of the domain management node or the relevant information of the model set.

[0011] In a second aspect, an embodiment of this application provides a device node, which includes:

[0012] A transceiver; and

[0013] A processor, which is coupled to the transceiver, wherein the processor is configured to:

[0014] Receive a first message via a transceiver, where the first message includes indication information related to the support capabilities of a domain management node or related information of a model set; the first message is generated by the domain management node based on the related information of the model set, and the related information of the model set is obtained by the domain management node, and the related information of the model set at least includes function description information of the radio interface and application condition information of the radio interface.

[0015] In a third aspect, an embodiment of the present application provides a communication method, which is applied to a domain management node. The communication method includes:

[0016] Obtain related information of a model set, where the related information of the model set at least includes function description information of the radio interface and application condition information of the radio interface;

[0017] Generate a first message based on the related information of the model set;

[0018] Transmit the first message, where the first message includes indication information related to the support capabilities of the domain management node or related information of the model set.

[0019] In a fourth aspect, an embodiment of the present application provides a communication method, which is applied to a device node. The communication method includes:

[0020] Receive a first message, where the first message includes indication information related to the support capabilities of a domain management node or related information of a model set; the first message is generated by the domain management node based on the related information of the model set, and the related information of the model set is obtained by the domain management node, and the related information of the model set at least includes function description information of the radio interface and application condition information of the radio interface. Description of the Drawings

[0021] Figure 1 Schematic flow chart of a communication method provided by an embodiment of the present application Figure 1 ;

[0022] Figure 2 Schematic flow chart of a communication method provided by an embodiment of the present application Figure 2 ;

[0023] Figure 3 Schematic composition structure of a communication system provided by an embodiment of the present application Figure 1 ;

[0024] Figure 4 Schematic composition structure of a communication system provided by an embodiment of the present application Figure 2 ;

[0025] Figure 5 Schematic flow chart of a communication method provided by an embodiment of the present application Figure 3 ;

[0026] Figure 6Flow schematic of a communication method provided by an embodiment of the present application Figure 4 ;

[0027] Figure 7 Flow schematic of a communication method provided by an embodiment of the present application Figure 5 ;

[0028] Figure 8 Flow schematic of a communication method provided by an embodiment of the present application Figure 6 ;

[0029] Figure 9 Flow schematic of a communication method provided by an embodiment of the present application Figure 7 ;

[0030] Figure 10 Flow schematic of a communication method provided by an embodiment of the present application Figure 8 ;

[0031] Figure 11 Flow schematic of a communication method provided by an embodiment of the present application Figure 9 ;

[0032] Figure 12 Flow schematic of a communication method provided by an embodiment of the present application Figure 10 ;

[0033] Figure 13 Flow schematic of a communication method provided by an embodiment of the present application Figure 10 One;

[0034] Figure 14 Flow schematic of a communication method provided by an embodiment of the present application Figure 10 Two;

[0035] Figure 15 Schematic structural diagram of a device provided by an embodiment of the present application. Detailed implementation manner

[0036] In order to more comprehensively understand the features and technical content of the embodiments of the present application, the implementation of the embodiments of the present application will be described in detail below with reference to the accompanying drawings. The accompanying drawings are for reference and illustration purposes only and are not used to limit the embodiments of the present application.

[0037] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by those skilled in the technical field to which this application belongs. The terms used herein are only for the purpose of describing the embodiments of this application and are not intended to limit this application.

[0038] In the following description, reference is made to "some embodiments", which describe a subset of all possible embodiments. However, it is understood that "some embodiments" may be the same subset or different subsets of all possible embodiments and may be combined with each other without conflict.

[0039] It should also be noted that the terms "first", "second", and "third" involved in the embodiments of the present application are only used to distinguish similar objects and do not represent a specific order for the objects. It is understood that "first", "second", and "third" can be interchanged with a specific order or sequence when allowed, so that the embodiments of the present application described herein can be implemented in an order other than that illustrated or described herein.

[0040] In addition, the mention of "embodiments" in this document means that the specific features, structures, or characteristics described in connection with the embodiments may be included in at least one embodiment of the present application. The occurrence of this phrase at various positions in the specification does not necessarily refer to the same embodiment, nor is it an independent or alternative embodiment that is mutually exclusive with other embodiments. Those skilled in the art will explicitly and implicitly understand that the embodiments described herein can be combined with other embodiments.

[0041] The following provides an introduction to the related technologies of the present application.

[0042] In the 3rd Generation Partnership Project (3GPP) standardization organization, AI / ML models are also an important research topic, and corresponding research and standardization work have been carried out from the Core Network (CN) to the Radio Access Network (RAN). For the physical layer (RAN1), research has been conducted on using AI / ML models to enhance certain use cases of the air interface, including reducing pilot overhead in Beam Management (BM) through AI / ML models, improving positioning accuracy using AI / ML models, and compressed transmission and prediction of Channel State Information (CSI) based on AI / ML models, among other specific cases. Based on the performance evaluation results, the use cases involving beam management and improved positioning accuracy have started the corresponding standardization work in NR Release 19.

[0043] Currently, considering the standardization process and actual product implementation, in the scenario of enhancing the air interface, 3GPP has agreed to defer the discussion on model transmission to future releases (i.e., Release 20 / 21), and no issues related to AI / ML model transmission will be considered in the current solution discussion. Moreover, since 3GPP aims to ensure wide-area coverage and enable ubiquitous access and communication for user equipment, including a series of interface interactions such as the core network (CN) and radio access network (RAN), and involves network architectures and protocols such as the core network, the construction, management, and configuration of AI / ML are relatively complex.

[0044] Based on this, the embodiments of the present application provide a domain management node and a device node. Considering the future wireless communication system, especially in the indoor wireless coverage environment, a domain management node with functions such as model management, update, and storage is introduced. After the domain management node obtains the relevant information of the model set, it generates a first message and sends the first message to the device node. The first message may include indication information related to the support capabilities of the domain management node or the relevant information of the model set. In this way, the domain management node constructs a more concise model set and realizes efficient management of the model set. Moreover, the domain management node synchronizes the support capabilities of the domain management node for AI / ML models to the device node by transmitting and obtaining the relevant information of the model set, facilitating the device node to request the relevant information of the model set according to its own situation; or the domain management node directly synchronizes the relevant information of the model set, facilitating the device node to request AI / ML models according to its own situation, thereby realizing the transmission of AI / ML models between the domain management node and the device node.

[0045] The following further elaborates on the present application through the accompanying drawings and specific embodiments.

[0046] In a first aspect, in an embodiment of the present application, Figure 1 is a schematic flowchart of a communication method provided by an embodiment of the present application Figure 1 . This communication method can be applied to a domain management node. As Figure 1 shown, this method may include:

[0047] S101, obtain the relevant information of the model set; the relevant information of the model set at least includes the function description information of the air interface and the application condition information of the air interface.

[0048] S102, generate a first message based on the relevant information of the model set.

[0049] S103, transmit the first message; the first message includes indication information related to the support capabilities of the domain management node or the relevant information of the model set.

[0050] Second aspect, embodiments of the present application provide a communication method, which can be applied to a device node. The method may include:

[0051] Receiving a first message, the first message includes indication information related to the support capabilities of the domain management node or related information of the model set; the first message is generated by the domain management node based on the related information of the model set, and the related information of the model set is obtained by the domain management node. The related information of the model set at least includes function description information of the radio interface and application condition information of the radio interface.

[0052] Third aspect, embodiments of the present application provide a communication method, which is applied to a communication system including a domain management node and a device node. As Figure 2 shown, the method may include:

[0053] S301, the domain management node obtains related information of the model set, and the related information of the model set at least includes function description information of the radio interface and application condition information of the radio interface.

[0054] S302, the domain management node generates a first message based on the related information of the model set.

[0055] S303, the domain management node transmits the first message to the device node; correspondingly, the device node receives the first message. The first message includes indication information related to the support capabilities of the domain management node or related information of the model set.

[0056] As Figure 2 shown, in embodiments of the present application, the above step S301 may include:

[0057] S3011, the domain management node sends a model set query request to the storage module.

[0058] In embodiments of the present application, the domain management node supporting the application of wireless AI / ML models queries the storage module for related information of the AI / ML model set stored therein. This process may consider enhancing the basic service layer of the communication system and expanding the support for wireless AI / ML model set discovery and access services.

[0059] S3012, the storage module sends the related information of the model set to the domain management node.

[0060] In embodiments of the present application, when the storage module receives the information query request, it sends the related information of the stored wireless AI / ML model to the requesting domain management node.

[0061] It should be noted that the purpose of the above steps S3011 to S3012 is to enable the domain management node to obtain the description information of the model set. Therefore, it can be completed at the initial stage of building the communication system, or after the storage module is updated. Without executing step S3011, the storage module actively notifies the domain management node to update in the form of "registration" through step S3012.

[0062] Next, based on the foregoing embodiments, Figure 1 and Figure 2 the communication method shown will be described.

[0063] In the embodiments of the present application, the communication system includes at least one domain management node and at least one device node. As Figure 3 shown, taking the communication system including one domain management node and one device node as an example for illustration. Among them, the domain management node 401 may include a model management function module 4011, which is a node in the communication system that has functions such as management and storage for the AI / ML model set. The storage module 402 may also be referred to as the storage module of the AI / ML model set. As a logical functional unit, the storage module 402 can be implemented in any node in the communication system (such as the domain management node 401), or on a third-party server. It should be understood that in the subsequent process of interacting with the storage module 402, it is assumed that the domain management node accesses this functional module through the relevant protocol of the basic service layer of this communication system, and does not involve the air interface signaling enhancement solution provided by this application. In addition, the device node 403 may be a user equipment (UE). The device node 403 includes a model usage module 4031. The device node 403 is used to send messages such as queries, requests, and updates to the domain management node 401. After receiving the message from the device node 403, the domain management node 401 sends corresponding messages such as queries, updates, and storage to the storage module 402 to obtain messages such as status reports and downloads fed back by the storage module 402, and the domain management node 401 feeds back messages such as downloads, updates, and monitoring to the device node 403. Among them, when the storage module 402 runs in other nodes, the domain management node 401 can interact with this node based on the foregoing steps S3011 to S3012 to obtain relevant information of the model set; or, when the storage module 402 runs on the domain management node 401, the domain management node 401 can also interact with the storage module 402 based on the foregoing steps S3011 to S3012, or directly read the relevant information of the model set from the storage module 402.

[0064] It should be understood that the AI / ML model set can also be understood as the model set, and the two can be equivalent or replaced.

[0065] In some embodiments, the functional description information of the air interface is used to characterize the air interface module to which the model is applied; the application condition information of the air interface is used to characterize the system configuration and application scenario of the model.

[0066] In the embodiments of the present application, the model set includes at least one model, and the related information of the model set includes the related information of each model. The related information of the model set should at least include the functional description information of the air interface, and this information includes the description information of each model. The description information is used to describe the radio air interface module to which the corresponding model can be applied, such as channel estimation, channel state information compression, channel state information prediction, and beam prediction, etc. As Figure 4As shown, taking the example where the transmitting end applies the model set 501 of the transmitting end for air interface enhancement and the receiving end applies the model set 502 of the receiving end for air interface enhancement. Among them, the dotted line represents the radio air interface module that can apply its corresponding model set to enhance the air interface capability. For example, the Coding.&Mod 5011 at the transmitting end can use the model set 501 of the transmitting end to enhance the coding ability; for example, the Digital BF 5013 at the transmitting end can use the model set 501 of the transmitting end to generate beamforming parameters and use different models for different scenarios; for example, the RF chain 1 5015 at the transmitting end can use the model set 501 of the transmitting end, and the RF chain 2 5021 at the receiving end can apply the model set 502 of the receiving end to identify and eliminate the noise in the signal, thereby improving the clarity of the signal; for example, the Time / freq.Sync 5023 at the receiving end can apply the model set 502 of the receiving end to make the time synchronization and frequency synchronization more accurate; for example, the Channel / Interface Estimation (or Chan. / interf.Esti) 5024 can apply the model set 502 of the receiving end to estimate the characteristics of the wireless channel to ensure the accuracy of the data; for example, the Demod.&decoding 5026 can apply the model set 502 of the receiving end to optimize the modulation process of the original analog signal and the process of decoding and restoring it to the original data. Other radio air interface modules, such as the Layer mapping 5012, Inverse Fast Fourier Transform (iFFT) 5014 and adding Cyclic Prefix (CP), Fast Fourier Transform (FFT) 5022 and removing the added cyclic prefix, Multiple-Input Multiple-Output detection (MIMO detection) 5025, etc., can also apply the model set of their respective devices in some embodiments, which will not be listed one by one here.

[0067] In the embodiments of the present application, the relevant information of the model set may further include the application condition information of the radio interface, including the application condition information of each model. The application condition information can be used to describe the conditions under which the corresponding model can be applied. For example, it includes the system configuration parameters of the domain management node and the device node, as well as the corresponding application scenarios of the two nodes. Among them, the system configuration parameters are managed by the domain management node and represent the parameter configuration of the physical layer transmission, such as the number of antenna ports and the reference symbol configuration, etc.; the application scenarios include the signal-to-noise ratio (SNR) range of the received signal and the channel statistical characteristics, etc., which are not specifically limited herein.

[0068] It should be understood that the relevant information of the model set is the relevant information of all the models stored in the storage module. In the embodiments of the present application, after the domain management node obtains the relevant information of the model set, based on its own ability to provide performance improvement using the model, or rather, based on the radio interface module of the model applicable in the domain management node, it makes a decision to select the models supported by the domain management node and the relevant information of the supported models from the relevant information of the model set. Further, when the domain management node determines that it can support the application of at least one model, it generates a first message.

[0069] Among them, in some embodiments, the first message may include the indication information related to the support ability of the domain management node. In this case, the indication information related to the support ability can be an indication information implemented according to a specific standard format and occupying several bits, used to indicate to the device node that the domain management node supports the application of the model set, or it may further include indicating to the device node that the domain management node stores the model set. It should be noted that the domain management node can send it to all device nodes in the domain as part of a broadcast message.

[0070] Or, in some embodiments, the first message may include the relevant information of the supported model set selected by the domain management node from the foregoing relevant information of the model set, indicating to the device node the models it can support and the relevant information of these supported models.

[0071] An embodiment of the present application provides a communication method. After obtaining relevant information of a model set, a domain management node generates a first message and sends the first message to a device node. The first message may include indication information related to the support capabilities of the domain management node or relevant information of the model set. In this way, the domain management node constructs a more concise model set and realizes efficient management of the model set. Moreover, the domain management node synchronizes the support capabilities of the domain management node for AI / ML models to the device node by transmitting to obtain relevant information of the model set, facilitating the device node to request relevant information of the model set according to its own situation. Or the domain management node directly synchronizes relevant information of the model set, facilitating the device node to request AI / ML models according to its own situation, thereby realizing the transmission of AI / ML models between the domain management node and the device node.

[0072] In another embodiment of the present application, as Figure 5 shown, the communication system may further include:

[0073] S304. When the first message includes indication information related to the support capabilities of the domain management node, the device node sends a first request message; correspondingly, the domain management node receives the first request message sent by the device node.

[0074] In an embodiment of the present application, when the device node also supports the model set, based on the aforementioned indication information related to the support capabilities of the domain management node, the device node determines that the domain management node with which it communicates supports the application of the model set. Further, the device node sends a first request message (which may also be referred to as a model set query request) to the domain management node to request relevant information of the model set in the domain from the domain management node.

[0075] S305. In response to the first request message, the domain management node obtains the update situation of relevant information of the model set.

[0076] In an embodiment of the present application, after receiving the first request message from the device node, the domain management node may query the storage module to check whether there is an update to the model set. The storage module, according to the actual situation, feeds back to the domain management node whether the model set is updated, and in the case of an update, the relevant information of the latest model or the relevant information of the updated model.

[0077] As Figure 5 shown, in an embodiment of the present application, the above step S305 may include:

[0078] S3051. The domain management node sends a model set update query message to the storage module.

[0079] S3052. The storage module sends a model set update confirmation message to the domain management node.

[0080] S306. The domain management node sends a first response message based on the update status of the relevant information of the model set. Correspondingly, the device node receives the first response message. The first response message includes the relevant information of the model set.

[0081] In the embodiments of the present application, the domain management node may carry the relevant information of the model set, or the updated relevant information of the model set, or the relevant information of the updated model in the first response message and send it to the device node, depending on the specific interaction situation, which is not specifically limited herein.

[0082] It should be noted that when the domain management node determines that the relevant information of the model set has not been updated based on the foregoing step S305, it may send the relevant information of the obtained or selected model set to the device node. In the case of an update, the relevant information of the model set stored in the domain management node will be updated based on the update, and the updated relevant information of the model set will be sent to the device node.

[0083] It should also be noted that when the first message includes the relevant information of the model set, the device node may also send a first request message to the domain management node to inquire whether the relevant information of the model set in the first message has been updated. When the domain management node determines that the relevant information of the model set has been updated, it may send the relevant information of the updated model to the device node through the first response message. Or, in the case of no update, the domain management node sends a first response message in a special format or does not send a first response message.

[0084] In some embodiments, the relevant information of the model set further includes one or more of the following:

[0085] The identifier of the model;

[0086] The attribute information of the model. The attribute information of the model includes the parameter format of the model, the input information of the model, and the output information of the model;

[0087] The structure information of the model;

[0088] The parameter information of the model.

[0089] In the embodiments of the present application, at least one model is stored in the storage module, and each model is constructed based on the relevant information of the model set. For the convenience of unified management of the models in the model set, the relevant information of the model set may further include the identifier (Identifier) of each model. It should be understood that the storage module uniformly manages the identifiers of all models, and in a model set, each model has a unique identifier. Among them, the relevant information of the model set should at least include the identifier of the model.

[0090] In some embodiments, the relevant information of the model set further includes the attribute information of each model, which is used to describe the basic attributes of the corresponding model, such as the structure of the model, the number of parameters, the input information, the output information, and the sizes of the input information and the output information, etc., which are not enumerated one by one here.

[0091] In some embodiments, the relevant information of the model set further includes the complete data of each model. For example, the structural information of the model, such as the network or algorithm used, may further include the parameter information of the model, such as the parameter content of the input information or the parameter content of the output information, etc.

[0092] It should be noted that, based on the steps in the foregoing embodiments, the relevant information of the model set that the domain management node can send to the device node may include at least one or more of the model identifier, the model attribute information, the radio interface function description information, and the radio interface application condition information. That is to say, the relevant information of the model set obtained by the domain management node from the storage module and the relevant information of the model set sent by the domain management node to the device node may not include the complete data of the model, but first send a partial content similar to an index or a directory for the device node to obtain the complete data based on the method in the following embodiments when needed.

[0093] The embodiment of the present application provides a communication method. When the first message sent by the domain management node to the device node includes the indication information related to the support capabilities of the domain management node, the domain management node responds to the first request message sent by the device node, generates a first response message carrying the relevant information of the model set based on the update situation of the relevant information of the model set, and sends it to the device node. In this way, the device node sends the first request message to the domain management node only when it also supports the model set, reducing the processing overhead of other device nodes and the pressure of the domain management node to broadcast information.

[0094] In another embodiment of the present application, based on the method in the foregoing embodiments, after the device node obtains the relevant information of the model set, when the device node meets the application conditions of one or several of the models, or when the target model that the device node needs to use is a unilateral model (a model only applied to the device node) to improve the radio interface performance, the device node actively requests to use the target model from the domain management node and obtains the complete data of the target model from the domain management node. As Figure 6 shown, the communication method may further include:

[0095] S601, the device node transmits the device capability information of the device node to the domain management node; correspondingly, the domain management node receives the device capability information of the device node; the device capability information includes the storage situation of the models in the device node, the preferred model information, and the computing resource information.

[0096] In the embodiments of the present application, since some models require bilateral cooperation between the domain management node and the device node, the domain management node needs to understand the device capability information of the device node, such as whether the device node can support the model set, whether the relevant information of the model has been obtained in the device node, whether the complete data of the model is stored in the device node, the structure of the model preferred or supported by the device node, the operations of the model supported by the device node, and the computing resource information of the device node, including the computing power and storage space of the device node.

[0097] In some embodiments, as Figure 6 shown, before step S601, it may further include:

[0098] S605, the domain management node sends a model support capability query request message to the device node.

[0099] In the embodiments of the present application, in the scenario where the device node requests the complete data of the model from the domain management node, as an optional step, when the device node accesses the network, the domain management node sends a model support capability request message to the device node as part of the device node capability query. Alternatively, the device node can actively report the device capability information of the device node to the domain management node based on the foregoing step S601.

[0100] S602, the device node transmits a download request message via the transceiver; correspondingly, the domain management node receives the download request message.

[0101] In the embodiments of the present application, after the device node obtains the relevant information of the foregoing model set, by querying the relevant information of the model set, it determines whether to download and use the model, or to download and use the model for a certain wireless air interface module. If necessary, the device node sends a download request message to the domain management node. The download request message may include the identifier of the target model requested by the device node, or other description information of the model to be requested, such as the structure and application conditions of the model, and the domain management node determines the target model and feeds it back to the device node.

[0102] In some embodiments, as Figure 6 shown, the domain management node obtains the structure information and parameter information of the target model from the storage module, which may include:

[0103] S6031, the domain management node sends a model download request message to the storage module.

[0104] S6032, the storage module sends the complete data of the target model, such as the structure information and parameter information of the model, to the domain management node.

[0105] In an embodiment of the present application, after receiving a download request message from a device node, the domain management node determines a target model. If the domain management node does not store the structure information and parameter information of the target model, the domain management node sends a model download request message to the storage module to request the complete data of the target model from the storage module. Accordingly, the storage module sends the complete data of the target model to the domain management node.

[0106] S603. The domain management node generates a second message based on the device capability information of the device node, the download request message, and the foregoing structure information and parameter information for obtaining the target model.

[0107] It should be noted that after obtaining the device capability information of the device node, when the first request message in the device node does not contain the identifier of the target model, the domain management node can comprehensively determine a suitable target model according to the device capability information of the device node and other description information of the required model provided by the device node in the foregoing download request message, and generate a second message. The second message carries the structure information and parameter information of the target model pushed by the domain management node for the device node.

[0108] In some embodiments, the domain management node may also send other relevant information of the target model that the device node has not obtained to the device node in the second message.

[0109] Furthermore, the domain management node adds the complete data of the target model obtained from the storage module or stored by itself to the second message to generate the second message.

[0110] S604. The domain management node transmits the second message to the device node; accordingly, the device node receives the second message.

[0111] In an embodiment of the present application, after generating the second message, the domain management node sends the second message to the device node. In this way, the device node downloads the complete data of the target model from the domain management node, and the target model can be used to enhance at least one radio air interface module in the device node.

[0112] Or, in some embodiments, after the device node obtains the relevant information of the model set based on the foregoing embodiments, when the device node or the domain management node needs to use the target model as a bilateral model (used jointly when communicating between the two ends of the device node and the domain management node) to improve the air interface performance, as Figure 7 shown, this communication method may further include:

[0113] S605. The domain management node sends a model support capability query request message to the device node.

[0114] In an embodiment of the present application, when the domain management node needs to jointly use a target model that is a bilateral model with the device node, the domain management node needs to actively send a model support capability query request message to the device node, requesting the device node to send the capability information of the device node to determine whether the device node can support the target model that the domain management node needs to use.

[0115] S601. The device node transmits the device capability information of the device node to the domain management node; correspondingly, the domain management node receives the device capability information of the device node; the device capability information includes the storage status of the model in the device node, the preferred model information, and the computing resource information.

[0116] This step can refer to step S601 in the foregoing embodiment and will not be elaborated here.

[0117] S606. The domain management node transmits a model push request message to the device node; correspondingly, the device node receives the model push request message.

[0118] In an embodiment of the present application, the domain management node can determine whether the target model that the domain management node needs to use can be supported by the device node according to the device capability information fed back by the device node. When the device node supports it, the domain management node carries the relevant information of the target model, such as the identifier of the target model, in the model push request message and sends it to the device node; or, when the device node does not support it, the domain management node re-determines the target model based on the device capability information fed back by the device node and the wireless air interface module to be applied, and carries the relevant information of the target model in the model push request message and sends it to the device node.

[0119] It should be understood that the model push request message is used to inquire the device node whether the device node can receive and apply the corresponding model.

[0120] S607. The device node sends a push confirmation message to the domain management node; correspondingly, the domain management node receives the push confirmation message sent by the device node.

[0121] In an embodiment of the present application, the device node can query the relevant information of the model set stored in the device node based on the relevant information of the model in the model push request message, and in combination with its own device capability information, when supporting the target model in the model push request message sent by the domain management node, it can send a push confirmation message to the domain management node, which contains confirmation information.

[0122] In some embodiments, when the device node does not support the target model, it may not feedback a push confirmation message to the domain management node, or include unsupported information in the push confirmation message to notify the domain management node.

[0123] In some embodiments, such as Figure 7 shown, the domain management node obtains the structure information and parameter information of the target model from the storage module, which may include the foregoing steps S6031 to step S6032.

[0124] S608. When the domain management node receives the push confirmation message, based on the device capability information of the device node, the push confirmation message, and the foregoing obtained structure information and parameter information of the target model, a second message is generated.

[0125] In the embodiments of the present application, when the domain management node determines that the device node supports the target model based on the push confirmation message, the device capability information of the device node may be further compared with the relevant information of the target model to confirm whether the device node supports the target model. When the device node supports the target model, a second message is generated and sent to the device node and stored in the domain management node. Among them, the second message carries the complete data of the target model obtained from the storage module, including the structure information and parameter information of the target model, etc.

[0126] It should be noted that the domain management node may obtain the complete data of the target model from the storage module after step S607, or may obtain the complete data of the target model in S608 after determining the target model based on the device capability information of the device node. The order of the steps is not specifically limited herein.

[0127] S604. The domain management node transmits the second message to the device node; correspondingly, the device node receives the second message.

[0128] In this way, based on the embodiments of the present application, the domain management node pushes the complete data of the target model obtained from the storage module to the device node, enabling the device node to perform a pairing operation with the domain management node and improving the air interface performance.

[0129] The embodiments of the present application provide a communication method. When the device node needs to apply a unilateral target model, the device node requests the complete data of the target model from the domain management node according to the device capability information; or when the domain management node needs to apply a bilateral target model, the domain management node sends the complete data of the target model in the storage module to the device node according to the device capability information fed back by the device node or the confirmation information in the push confirmation message. In this way, the device node can obtain the complete data of the applicable target model according to the actual needs, avoiding occupying the channel resources by transmitting the complete data of all models in the model set during the broadcast process, and also reducing the occupation of the storage space of the domain management node and the device node.

[0130] In yet another embodiment of the present application, based on the foregoing embodiment, the device node obtains relevant information of the model set, and after obtaining the complete data of the target model by requesting from the domain management node or being actively pushed by the domain management node, it may not be used temporarily. When the device node or the domain management node meets the application conditions, system configuration, etc. in the relevant information of the model set, the device node actively or the domain management node configures the device node to activate the target model.

[0131] In some embodiments, when the device node meets the application condition information of the air interface or the function description information of the air interface in the relevant information of the model set, and the device node needs to activate the model to be activated of at least one radio air interface module applied to the device node, the device node may actively confirm with the domain management node whether the model to be activated can be activated. As Figure 8 shown, in this case, the method may include:

[0132] S701, the device node transmits a first activation request message to the domain management node; correspondingly, the domain management node receives the first activation request message sent by the device node, and the first activation request message includes the identifier of the model to be activated and the first activation time limit information.

[0133] In the embodiment of the present application, when the device node or the radio air interface module in the device node meets the application condition information of the air interface provided in the relevant information of the model set, the device node may consider activating the model applied to the radio air interface module to improve the air interface performance.

[0134] In this way, the device node may send a first activation request message to the domain management node, and the message carries the identifier of the model to be activated or other information of the model to be activated. The model to be activated may be the model applied to the radio air interface module that meets the application conditions, and the model to be activated may be obtained from the domain management node based on the foregoing embodiment. The first activation request message may further carry the first activation time limit information, which is used to characterize the activation time of the model to be activated expected by the device node, and may be expressed to the domain management node by means of the activation start time point, activation duration, etc.

[0135] S702, the domain management node transmits a first activation confirmation message to the device node; correspondingly, the device node receives the first activation confirmation message; the first activation confirmation message includes the confirmation information of the domain management node and the second activation time limit information, and the second activation time limit information is determined according to the first activation time limit information and the computing resource information of the domain management node.

[0136] In the embodiments of the present application, after receiving the first activation request message, the domain management node can determine whether the model to be activated requested by the device node is a unilateral model that the device node needs to apply, or a bilateral model that needs to be paired with the domain management node for use. Exemplarily, when the model to be activated is a bilateral model, the domain management node can select to determine the second activation time limit information by combining its own computing resource situation and the first activation time limit information given by the device node. The activation time in the second activation time limit information should be at least within the activation time in the first activation time limit information, or cover the activation time in the first activation time limit. The second activation time limit information is used to represent the activation time recommended by the domain management node to the device node, that is, the activation time that the domain management node can support, and can be expressed by information such as the activation start time point and the activation duration.

[0137] In addition, the domain management node sends the first activation request message and records information such as the identifier of the model to be activated, the identifier of the device node, and the activation time and activation status of the model to be activated sent to the device node.

[0138] S703. The device node performs activation processing on the model to be activated based on the second activation time limit information and the identifier of the model to be activated.

[0139] In the embodiments of the present application, the device node obtains the activation time in the second activation time limit information given by the domain management node and activates the model to be activated within the activation time.

[0140] In some embodiments, after the model to be activated is activated and applied to the corresponding radio air interface module, if the device node or the domain management node no longer meets the relevant information of the foregoing model, such as the application condition information of the air interface, the device node can confirm deactivation to the domain management node. As Figure 9 shown, in this case, the communication method may include:

[0141] S704. When the deactivation condition is met, the domain management node sends a deactivation request message to the device node; correspondingly, the device node receives the deactivation request message; the deactivation request message includes the identifier of the model to be deactivated and the deactivation time limit information.

[0142] In some embodiments, before step S704, it may further include:

[0143] S705. The domain management node sends a model deactivation indication message to the device node.

[0144] It should be noted that the model to be deactivated is a model in the device node that has been activated based on the steps in the foregoing embodiments and is working properly within the activation time. If certain events occur on the device node within the activation time, such as performance degradation, insufficient computing power, or scenario change, etc., which cause the device node to no longer meet the application condition information of the radio air interface, the model (model to be deactivated) applied to the foregoing radio air interface module will no longer be used. The device node executes step S705 and needs to send a deactivation indication message to the domain management node. This message should at least include the identifier of the model to be deactivated, and may also include deactivation time limit information and other relevant information of the model to be deactivated, etc., and notify the domain management node of the change situation of the device node.

[0145] In the embodiments of the present application, some predefined events occur on the domain management node, such as scenario change and model update. In the case where the domain management node and the device node use a bilateral model in a paired manner, then S705 is an optional step. The domain management node sends a deactivation request message to the device node by executing step S704. This message carries the identifier of the model to be deactivated and deactivation time limit information.

[0146] Among them, the deactivation time limit information can be expressed in the form of a deactivation time point.

[0147] S706, the device node performs deactivation processing on the model to be deactivated based on the deactivation time limit information.

[0148] In the embodiments of the present application, after receiving the deactivation request message sent by the domain management node, the device node obtains the deactivation time limit information in the deactivation request message and performs deactivation processing on the model to be deactivated at the given deactivation time point.

[0149] In some embodiments, the device node can be configured by the domain management node to be in a state of spontaneously activating the model. After completing the activation processing of the model to be activated within the configured time, it notifies the domain management node of the activation state of the model to be activated.

[0150] In some embodiments, after step S706, it may further include:

[0151] S707, the device node sends a deactivation confirmation message to the domain management node.

[0152] It should be noted that after successfully deactivating the model to be deactivated, the device node can send a confirmation message to the domain management node to enable the domain management node to update the activation state of the model to be deactivated recorded above.

[0153] In some embodiments, the device node may also be configured by the domain management node to deactivate a model that has been activated spontaneously, and after the deactivation process of the model to be deactivated is completed, send a notification message to the domain management node, which carries the current activation status of the model to be deactivated.

[0154] In some embodiments, the domain management node may also configure the device node to activate the model to be activated when the device node meets the air interface application condition information or the air interface function description information in the relevant information of the model set. As Figure 10 shown, in this case, the method may include:

[0155] S708, the domain management node sends a second activation request message to the device node; correspondingly, the device node receives the second activation request message; the second activation request message includes the identifier of the model to be activated and the third activation time limit information.

[0156] In the embodiments of the present application, when the domain management node meets the air interface application conditions of a certain radio air interface module, and / or when the domain management node determines that the device node meets the air interface application conditions of a certain radio air interface module, the domain management node may send a second activation request message to the device node, requesting to activate the model to be activated that can be applied to the radio air interface module that meets the application conditions, so as to improve the air interface performance.

[0157] Among them, the second activation request message indicates that it includes the identifier of the model to be activated and the second activation time limit information, and may also include other relevant information of the model to be activated. The second activation time limit information is used to characterize the activation time that the domain management node can support and hopes the device node to synchronize, for example, including information such as the start time point and activation duration of the model to be activated.

[0158] S709, the device node sends a second activation confirmation message to the domain management node; correspondingly, the domain management node receives the second activation confirmation message.

[0159] In the embodiments of the present application, after the device node receives the second activation request message, the device node comprehensively determines whether it can support activating the model to be activated at the activation time indicated by the third activation time limit information according to the computing resources, the activation status of the model to be activated, and whether the radio air interface module to which the model to be activated is applied in the device node meets the air interface application conditions. If the device node determines that it can support, the device node sends a second activation confirmation message to the domain management node, and the message carries confirmation information.

[0160] S710, the device node activates the model to be activated based on the third activation time limit information and the identifier of the model to be activated.

[0161] Further, the device node activates the model to be activated at the activation time indicated by the third activation time limit information, and after successful activation, synchronizes the activation status of the model to be activated to the domain management node.

[0162] It should be noted that, in the embodiments of the present application, the model to be activated actively requested by the domain management node to be activated by the device node is mostly a bilateral model in most cases. Therefore, the device node needs to clarify to the domain management node whether the model to be activated has been activated in order to be paired with the model on the corresponding side of the model to be activated in the domain management node for use, enhancing the air interface.

[0163] S704, when the deactivation condition is met, the domain management node sends a deactivation request message to the device node; correspondingly, the device node receives the deactivation request message; the deactivation request message includes the identifier of the model to be deactivated and the deactivation time limit information.

[0164] In some embodiments, before step S704, it may further include:

[0165] S705, the domain management node sends a model deactivation indication message to the device node.

[0166] S706, the device node performs a deactivation process on the model to be deactivated based on the deactivation time limit information.

[0167] In some embodiments, after step S706, it may further include:

[0168] S707, the device node sends a deactivation confirmation message to the domain management node.

[0169] The descriptions of the above steps S704 to S707 can refer to the foregoing embodiments and will not be elaborated here.

[0170] The embodiments of the present application provide a communication method. When the domain management node or the device node meets the application conditions of the air interface, the domain management node or the device node actively sends an activation request, and the device node performs an activation process on the model to be activated. When the domain management node or the device node no longer meets the application conditions of the air interface within the activation time, the device node confirms to the domain management node and performs a deactivation process on the model to be deactivated. In this way, when the model is a bilateral model, it can be ensured that the activation and deactivation of the model by the domain management node and the device node are synchronized, improving the stability and reliability of the communication system.

[0171] In another embodiment of the present application, based on the foregoing embodiments, after the device node completes the activation process on the model to be activated, when the activated model works normally, it is necessary to detect the performance of the model to be detected in the activated model, that is, to determine whether the performance is improved after using the model to be detected.

[0172] In some embodiments, such as Figure 11 shown, when the device node actively triggers the model performance detection, the method may include:

[0173] S801, the device node sends a performance detection request message to the domain management node; correspondingly, the domain management node receives the performance detection request message, and the performance detection request message includes the identifier of the model to be detected and the detection time limit information.

[0174] In the embodiments of the present application, based on the time point of the pre-configured detection period, or when an event triggers the device node to perform performance detection on an already activated and working model, the device node sends a performance detection request message to the domain management node. The message should at least include the identifier of the model to be detected, the detection time limit information, and the auxiliary data request information. Among them, the detection time limit information may include the detection start time point, the detection time length, or the detection period, etc.; the auxiliary data request information is used to represent the auxiliary data that the device node determines according to the relevant information of the model to be detected and requires the domain management node to provide. The auxiliary data, such as the label data for performance comparison that is ideally used to improve the estimation accuracy, may include, for example, the accurate location information of the user, a better channel estimation result, or the channel state information with high quantization accuracy.

[0175] S802, the domain management node sends performance detection auxiliary data to the device node; correspondingly, the device node receives the performance detection auxiliary data; the performance detection auxiliary data includes at least one label data information.

[0176] It should be noted that after receiving the performance detection request information, the domain management node confirms to the device node that the device node can perform performance detection on the model to be detected, and provides the corresponding detection auxiliary data for the device node according to the auxiliary data request information in the performance detection request information, such as the label data for performance comparison that is ideally used to improve the estimation accuracy module.

[0177] S803, the device node performs detection on the model to be detected to obtain a performance detection result.

[0178] In the embodiments of the present application, after receiving the confirmation and auxiliary data from the domain management node, the device node performs performance detection on the model to be detected at the detection time indicated by the aforementioned detection time limit information, and evaluates the performance of the model to be detected after being applied to the corresponding radio air interface module. The performance detection method may be, for example, to compare the output of the model to be detected with the ideal value or the output of the radio air interface module without applying the model to be detected.

[0179] S804, the device node sends the performance detection result to the domain management node; correspondingly, the domain management node receives the performance detection result.

[0180] When the device node actively triggers a performance detection and requests a performance detection from the domain management node, as an optional step, S804, the device node may send a performance detection result to the domain management node based on S804.

[0181] Alternatively, in some embodiments, the device node does not execute S804 and saves the performance detection result in a local memory.

[0182] In some embodiments, as Figure 12 shown, when the domain management node actively triggers a model performance detection, the method may include:

[0183] S805, the domain management node sends a performance detection request message to the device node; correspondingly, the device node receives the performance detection request message, and the performance detection request message includes an identifier of the model to be detected and detection time limit information.

[0184] In the embodiments of the present application, at a periodic detection time point or event trigger, the domain management node may trigger sending a performance detection request message to the device node, where at least the identifier of the model to be detected, detection time limit information, and auxiliary data are carried in the message.

[0185] S802, the domain management node sends performance detection auxiliary data to the device node; correspondingly, the device node receives the performance detection auxiliary data; the performance detection auxiliary data includes at least one tag data information.

[0186] S803, the device node performs a detection on the model to be detected to obtain a performance detection result.

[0187] S804, the device node sends the performance detection result to the domain management node; correspondingly, the domain management node receives the performance detection result.

[0188] It should be noted that the descriptions of steps S802 to step 804 in the embodiments of the present application may refer to the foregoing embodiments and will not be elaborated herein.

[0189] It should also be noted that when the domain management node actively triggers sending a performance detection request message to the device node, step S804 is necessary. After the device node completes the performance detection, it executes S804 to feedback the performance detection result to the domain management node.

[0190] The embodiments of the present application provide a communication method, where the device node is actively triggered by the device node or the domain management node to perform a performance detection on the model to be detected, and the device node optionally feedbacks the performance detection result to the domain management node. In this way, the activated model can be evaluated to determine whether the activated model is effective, improving the reliability of the communication system.

[0191] In another embodiment of the present application, the device node may also have the function of local wireless data collection. In this case, some device nodes with model training capabilities may update the model downloaded from the domain management node based on the foregoing embodiments in order to better adapt to the application scenario. Therefore, after updating the model, the device node can upload the updated model to the domain management node and the storage module. As Figure 13 shown, in this case, the method may include:

[0192] S901, the device node sends a model update request message to the domain management node; correspondingly, the domain management node receives the model update request message, and the model update request message includes the identifier of the model to be updated.

[0193] In the embodiment of the present application, after local training and model update of the device node, the device node can send a model update request message to the domain management node through the identifier of the model. At least the identifier of the model to be updated should be carried in this information, which is used to characterize the model that has been updated in the device node and needs to be synchronized to the domain management node and the storage module.

[0194] Among them, the model to be updated can be the model obtained by the device node from the storage module through the domain management node. The device node can update the complete data of the model, such as the structure, parameters or other relevant information of the model, which is not specifically limited here.

[0195] S902, when it is determined that the model to be updated needs to be updated, the domain management node sends a model update confirmation message to the device node; correspondingly, the device node receives the model update confirmation message.

[0196] In the embodiment of the present application, after receiving the model update request message, the domain management node can further confirm whether the model to be updated needs to be updated. For example, whether the model to be updated can also be applied to the domain management node or other device nodes, then it is necessary to perform a synchronous update.

[0197] If the domain management node confirms that the model to be updated needs to be updated, it sends a model update confirmation message to the device node. This message is used to instruct the device node to upload the relevant information of the model to be updated, or at least includes the part of the information that has been updated in the relevant information of the model to be updated.

[0198] S903, the device node sends the relevant information of the model to be updated to the domain management node; correspondingly, the domain management node receives the relevant information of the model to be updated.

[0199] In the embodiment of the present application, after receiving the model update confirmation message, the device node uploads the relevant information of the model to be updated to the domain management node, or at least includes the part of the information that has been updated in the relevant information of the model to be updated.

[0200] S904, the domain management node updates the model to be updated in the storage module based on the relevant information of the model to be updated.

[0201] In some embodiments, S904 may include:

[0202] S9041, the domain management node sends the relevant information of the model to be updated to the storage module; correspondingly, the storage module receives the relevant information of the model to be updated.

[0203] In the embodiments of the present application, the domain management node forwards the received relevant information of the model to be updated to the storage module. When the domain management node also applies the model to be updated currently, the model to be updated stored in the domain management node can also be updated synchronously.

[0204] S9042, the storage module updates the relevant information of the model set based on the relevant information of the model to be updated.

[0205] S9043, the storage module sends a first update confirmation message to the domain management node.

[0206] In the embodiments of the present application, after the storage module finishes updating the relevant information of the model set, it can send a first update confirmation message to the domain management node. This message carries at least confirmation information and can also carry the identifier of the model to be updated to notify the domain management node that the update has been completed.

[0207] S9044, the domain management node sends a second update confirmation message to the device node.

[0208] Further, based on the received first update confirmation message, the domain management node sends a second update request message to the device node to notify the device node that both the domain management node and the storage module have completed the update. And when the model to be updated is a bilateral model, it communicates with the device node based on the synchronized model to be updated.

[0209] In another embodiment of the present application, the device node updates the model to be updated in the storage module based on the foregoing embodiments to obtain an updated model. In this case, the storage module can also notify other device nodes using the model to be updated through the domain management node that the model to be updated has been updated and synchronize the relevant information of the model to be updated. As Figure 14 shown, the method may further include:

[0210] S1001, when there is an updated model in the storage module, the domain management node sends a model update indication message to the device node; correspondingly, the device node receives the model update indication message; the model update indication message includes the identifier of the updated model.

[0211] In some embodiments, before step S1001, the following steps may further be included:

[0212] S1002. The storage module sends a model set update indication message to the domain management node.

[0213] In the embodiments of the present application, after the model to be updated in the storage module is updated, the storage module synchronizes it as an updated model to each device node using the updated model, that is, the identifier of the updated model is the same as that of the model to be updated. The synchronization method may be to send a model set update indication message to the domain management node to indicate to the domain management node that there is an updated model in the model set, or to send the identifier of the updated model to the domain management node.

[0214] Further, when the domain management node receives the model set update indication message, based on the records in the domain management node, it sends a model update indication message to the device node that uses the updated model or has requested to download the updated model, or broadcasts a model update indication message to all device nodes in the domain. The message may carry the identifier of the updated model or an indication message that the model set has been updated.

[0215] S1003. The device node deactivates the updated model.

[0216] In the embodiments of the present application, if the device node is using the updated model, after receiving the model update indication message, the device node may deactivate the updated model based on the foregoing embodiments. If the device node has only downloaded and stored the model, this step may not be executed.

[0217] S1004. The domain management node sends a model update push request message to the device node; correspondingly, the device node receives the model update push request message.

[0218] In some embodiments, after successfully completing the deactivation, the device node may send a notification to the domain management node. The domain management node may further send a model update push request message to the device node, and the message should at least include the identifier of the updated model.

[0219] S1005. The device node sends a model update push confirmation message to the domain management node; correspondingly, the domain management node receives the model update push confirmation message.

[0220] In the embodiments of the present application, after obtaining the identifier of the updated model, the device node confirms whether it needs to receive the relevant information of the updated model according to its own situation. If it needs to receive, it sends a model update push confirmation message to the domain management node, and the message should at least include confirmation information and may also include the identifier of the updated model.

[0221] S1006, the domain management node obtains the updated model from the storage module based on the model update push confirmation message.

[0222] In some embodiments, step S1006 may include:

[0223] S1007, the domain management node sends a model set update request message to the storage module.

[0224] S1008, the storage module sends the relevant information of the updated model to the domain management node.

[0225] In the embodiments of the present application, after receiving the model update push message sent by the device node, the domain management node may send a model set update request message to the storage module to request the relevant information of the updated model from the storage module.

[0226] Furthermore, the storage module sends the relevant information of the updated model, or the partial information with updates in the relevant information of the updated model, to the domain management node.

[0227] S1009, the domain management node sends a model update message to the device node; correspondingly, the device node receives the model update message; the model update message includes the relevant information of the updated model.

[0228] Among them, the model update message carries the relevant information of the updated model obtained by the foregoing domain management node from the storage module, or the partial information with updates in the relevant information of the updated model.

[0229] S1010, the domain management node sends a model activation request message to the device node; correspondingly, the device node receives the model activation request message.

[0230] In some embodiments, after S1008, it may further include:

[0231] S1011, the device node sends a model activation confirmation message to the domain management node.

[0232] S1012, the device node performs an activation process on the updated model.

[0233] It should be noted that if the device node was using the updated model before the foregoing step S1001 and deactivated it based on the foregoing step S1003, and the domain management node needs to continue to pair with the device model to use the updated model, the domain management node may send a model activation request message to the device node to request the device node to activate the updated model. The model activation request message should at least include information such as the identifier of the updated model and the activation time of the updated model.

[0234] Furthermore, based on its actual situation, the device node sends a model activation confirmation message to the domain management node and activates the updated model at the agreed activation time.

[0235] It should also be noted that the interaction signaling between the domain management node and the device node in the embodiments of the present application and the above embodiments can be implemented in the radio link management, media access layer, and physical layer control signaling of the communication system.

[0236] The embodiment of the present application provides a communication method. The device node can update or train the downloaded model and send the updated model to the storage module through the domain management node. After the update, the storage module synchronizes the relevant information of the updated model to other device nodes using the model. In this way, it can ensure that all device nodes and domain management nodes in the communication system using the same model are synchronized, so as to achieve the effect of enhancing the air interface when paired.

[0237] In a fourth aspect, the embodiment of the present application provides a communication system, including a storage module, at least one domain management node, and at least one device node, wherein each domain management node communicates with at least some of the at least one device node.

[0238] Compared with the 3GPP system, the communication system provided in the embodiment of the present application (which can also be called the SparkLink system) is mainly applied to wireless short-distance communication in indoor scenarios. In terms of realizing air interface enhancement based on the AI / ML model, the communication system provided in the embodiment of the present application has the following characteristics and advantages:

[0239] (1) The wireless propagation environment is relatively stable. Since the application scenario is an indoor environment, the object distribution in this environment is relatively stable, and there are no high-speed moving objects. The signal propagation environment is relatively fixed. Therefore, the wireless channel is relatively stable within a certain period of time. The device can train the AI / ML model to learn and match the wireless environment, so that the model has better universality and achieves better performance.

[0240] (2) The protocol framework is relatively simple. The protocol framework of the communication system where the solution provided in the present application is located includes three parts: the basic application layer, the basic service layer, and the access layer.

[0241] Among them, the access layer technology consists of Synchronous Low-latency Broadband (SLB) technology, Synchronous Low Energy (SLE) technology, and Synchronous Link Positioning (SLP) technology that has been completed but not shown in the figure. The protocol includes physical layer transmission, media access layer, and link control layer to support the effective, fast, and reliable transmission of information.

[0242] Among them, the basic service layer is composed of a series of basic functional units, including a control plane composed of a device discovery functional unit, a general management functional unit (including connection management, measurement management, etc.), a service management functional unit, a Quality of Service (QoS) management functional unit, and a security management functional unit; the data transmission and adaptation functional unit is responsible for service plane functions such as data encapsulation. The basic application layer is composed of multiple general application service frameworks and a series of configuration documents for specific applications.

[0243] To implement the related capabilities of AI / ML, especially in enhancing the AI / ML capabilities of the air interface, it is necessary to strengthen the access layer, including modules such as encoding, modulation, and precoding at the transmitter, and basic modules such as synchronization, channel estimation, demodulation, and decoding at the receiver, and support is provided by adding certain functions to the basic service layer.

[0244] (3) The hardware differences of node devices are small. Due to having a unified authentication standard, the hardware device suppliers of the key air interface modules are not as diverse as those of 3GPP, resulting in relatively small differences in hardware implementation. Therefore, it is more convenient in terms of the basic hardware for physical layer signal processing of applying AI / ML models.

[0245] (4) Support for ultra-low latency and reliable air interface transmission design. The air interface design of the communication system provided by the embodiments of the present application ensures ultra-low latency and highly reliable physical layer transmission, which facilitates the transmission of parameters, configurations, data, and management related to AI / ML models. Therefore, it has advantages in implementing the management of AI / ML models for air interface applications.

[0246] In the embodiments of this application, the design of relevant functions and signaling interactions is aimed at enhancing the support for the management and operation of AI / ML models in the link control layer and the media access layer of the access layer / air interface of the communication system, so as to implement an enhanced solution for physical layer signal transmission and processing based on AI / ML models. The AI / ML model management function for improving the wireless air interface performance, which is oriented to in-domain device nodes in the main communication system, includes constructing an "AI / ML model set (which can also be called a model set)" and adding air interface signaling for "AI / ML model management". In terms of constructing an AI / ML model set that supports air interface enhancement, the system performs unified identification management on the models used, and describes information such as the basic information and application conditions of the models; in terms of AI / ML model management, a signaling process interaction for model operations between nodes based on model identification is designed.

[0247] In the embodiments of this application, in order to support the AI / ML capabilities introduced by the air interface, new functional units need to be introduced in the communication system: "Wireless AI / ML Model Management Function (i.e., Model Management Function Module 4011)", as Figure 3 shown. Among them, the Wireless AI / ML Model Management Function Module is responsible for in-domain wireless AI / ML model management and related operations, including querying, updating, and storing AI / ML models in the wireless AI / ML model set (which can also be called a storage module), receiving status information from the AI / ML model set, and downloading models, etc. At the same time, this module is also responsible for interacting with device nodes that support wireless AI / ML functions regarding model-related information, including downloading, updating, and monitoring wireless AI / ML models to device nodes, and receiving requests for querying, downloading, and updating models from device nodes. The interaction signaling of the communication system will be described below in conjunction with Embodiments 1 to 7.

[0248] Embodiment 1 Construction of Wireless AI / ML Model Set

[0249] In order to uniformly manage the AI / ML models (which can also be called models) applicable to the air interface, a wireless AI / ML model set needs to be introduced. The models in it are managed by being assigned a unified ID number, and each model is described by the following relevant information:

[0250] (1) Model identification (ID). The communication system performs unified numbering management on the AI / ML models among them. In an AI / ML model set, each AI / ML model has a unique encoding.

[0251] (2) Model attribute information. It is used to describe the basic attributes of the AI / ML model, including the model's structure, parameter format, input and output sizes, etc.

[0252] (3) Radio interface function description information of the model. It is used to describe the radio interface modules applicable to the AI / ML model, such as channel estimation, channel state information compression, channel state information prediction, and beam prediction, etc. For example, the radio interface module in the foregoing Figure 4 wireless radio interface module.

[0253] (4) Application condition information of the model's radio interface. It is used to describe the conditions under which the AI / ML model can be used, including system configuration and application scenarios. System configuration represents the parameter configuration of physical layer transmission, such as the number of antenna ports and reference symbol configuration, etc.; application scenarios include the received signal SNR range and channel statistical characteristics, etc.

[0254] (5) Complete data of the model. The complete data of the AI / ML model itself, including the model structure and all parameter values.

[0255] It should be noted that as a logical functional unit, the storage module can be implemented at any node in the system or a third-party server. In the subsequent interaction process related to this set, it is assumed that the domain management node accesses this functional module through the relevant protocol of the basic service layer of the communication system, and it does not involve the radio interface signaling enhancement solution of this application.

[0256] In the embodiments of this application, the radio interface signaling related to the AI / ML function includes the following aspects: indication of radio interface support for AI / ML enhancement, request and transmission of wireless AI / ML models, activation and deactivation of wireless AI / ML models, performance detection of wireless AI / ML models, and update of wireless AI / ML models, etc.

[0257] Embodiment 2 Wireless AI / ML Support Capability Indication

[0258] In the embodiments of this application, the domain management node needs to obtain the information of the wireless AI / ML model set of the communication system and send messages related to the wireless AI / ML support capability and model set to the device nodes within the domain. The specific signaling interaction between nodes is as Figure 5 shown, and specifically includes the following steps and interaction signaling:

[0259] Step 0 (S3011): The domain management node sends a model set query request to the wireless AI / ML model set. The domain management node that supports the wireless AI / ML function queries the wireless AI / ML model-related information stored in the wireless AI / ML model set in the communication system. This process can consider enhancing the basic service layer of the communication system and expanding the support for wireless AI / ML model set discovery and access services.

[0260] Step 1 (S3012): The wireless AI / ML model set sends model set-related information to the domain management node. When the wireless AI / ML model set receives an information query request, it sends the description information of the wireless AI / ML models in the set (part of the information in the related information), including the model identifier, model attributes, model functions, and model application conditions, etc., to the requesting domain management node.

[0261] The purpose of the above two steps is to enable the domain management node to obtain the description information of the wireless AI / ML model set in the system. Therefore, it can be completed at the initial stage of system setup or the domain management node can be actively notified to update when the wireless AI / ML model set is updated.

[0262] Step 2 (S303): The domain management node sends an indication message (the first message) supporting wireless AI / ML capabilities within the domain. The domain management node indicates to the nodes within the domain its support for wireless AI / ML capabilities or the existence of the description information of the available wireless AI / ML model set. This message can be sent to all nodes as part of the system broadcast message.

[0263] Step 3 (S304): The device node sends a wireless AI / ML model set information query request (the first request message) to the domain management node. When there is a device node in the domain that supports wireless AI / ML capabilities and obtains the indication information in Step 2, and learns that the domain supports wireless AI / ML capabilities and there is an available wireless AI / ML model set, this node can request the description information of the wireless AI / ML model set in the domain from the domain management node.

[0264] Step 4 (S3051): The domain management node queries the wireless AI / ML model set for update information (the model set update query message). After receiving the request from the device node, the domain management node can query the wireless AI / ML model set to check if there is an update.

[0265] Step 5 (S3052): The wireless AI / ML model set confirms the update information to the domain management node (the model set update confirmation message). The wireless AI / ML model set, based on the actual situation, feeds back to the domain management node whether the model set has been updated. If there is an update, it sends the latest set information or the updated information to the domain management node.

[0266] Step 6 (S306): The domain management node sends the wireless AI / ML model set description information (the first response message) to the requesting device node. The domain management node sends the (updated) wireless AI / ML model set description information to the requesting device node.

[0267] After these steps, the device nodes in the domain that support wireless AI / ML capabilities can obtain the indication information that the domain supports wireless AI / ML capabilities and the description information of the models in the wireless AI / ML model set.

[0268] Example 3 Wireless AI / ML Model Transmission Support

[0269] In an embodiment of the present application, when a device node needs to apply a wireless AI / ML model, the device node needs to request the domain management node to use the wireless AI / ML model to improve the performance of a certain wireless module. The specific signaling interactions between nodes include two scenarios: the device node actively requests the wireless AI / ML model (corresponding to Figure 6 ) and the domain management node pushes the wireless AI / ML model to the device node (corresponding to Figure 7 ).

[0270] In some embodiments, for the case where the device node requests a wireless AI / ML model, that is, when the device node needs to use a certain unilateral AI / ML model (i.e., a model only applied to the device node) to improve the air interface performance, this interaction process is required, as shown in Figure 6 , and includes the following steps and interaction signaling:

[0271] Step 0 (S605): The domain management node sends a wireless AI / ML support capability query request (model support capability query request message) to the device node. As an optional step, the domain management node can send a device wireless AI / ML capability support query to the device node. If the AI / ML capability is regarded as a part of the device wireless transmission capability, it can be directly reported to the domain management node in Step 1.

[0272] Step 1 (S601): The device node reports its wireless AI / ML support capability (device capability information of the device node) to the domain management node. When the device node receives a wireless AI / ML capability query request, or as a part of the device capability, the device node reports its wireless AI / ML support capability, including at least the following information:

[0273] - Whether there is a wireless AI / ML model and the description information of the model;

[0274] - Preferred AI / ML model structure;

[0275] - Computing power and storage space supporting wireless AI / ML operations.

[0276] Step 2 (S602): The device node sends a wireless AI / ML model download request (download request message) to the domain management node. When the device node needs to download and apply the AI / ML model for a certain module according to the description information of the wireless AI / ML model set, it sends a download request for the relevant model to the domain management node through the model ID.

[0277] Step 3 (S6031): The domain management node sends a corresponding wireless AI / ML model download request (model download request message) to the wireless AI / ML model set. When receiving a wireless AI / ML model download request from a device node, the domain management node requests the complete data of the wireless AI / ML model corresponding to the model ID from the wireless AI / ML model set.

[0278] Step 4 (S6032): The wireless AI / ML model set sends the requested wireless AI / ML model data to the domain management node. When receiving a wireless AI / ML model download request from the domain management node, the wireless AI / ML model set sends the complete data of the requested model to the domain management node, for example, including the structure information and parameter information of the model.

[0279] Step 5 (S604): The domain management node sends the requested wireless AI / ML model data (carried in the second message) to the device node. The domain management node sends the requested wireless AI / ML model data to the requesting device node.

[0280] Through these steps, the device node downloads a wireless AI / ML model from the wireless AI / ML model set that can be applied to enhance a certain air interface module.

[0281] In some embodiments, for the case where the domain management node pushes an AI / ML model to a device node, that is, when the device node needs to use a certain AI / ML model to communicate with the AI / ML model of the domain management node and use it jointly (i.e., a bilateral model) to improve the air interface performance, this interaction process is required, including the following steps and interaction signaling:

[0282] Step 0 (S605): The domain management node sends a wireless AI / ML support capability query request to the device node. The domain management node can send a query for the device's wireless AI / ML capability support to the device node. If the wireless AI / ML capability is regarded as part of the device capability, it can be directly reported to the domain management node in Step 1.

[0283] Step 1 (S601): The device node reports its wireless AI / ML support capability to the domain management node. When the device node receives a wireless AI / ML capability query request, or as part of the device capability, the device node reports its wireless AI / ML support capability, including at least the following information:

[0284] - Whether there is a wireless AI / ML model and the description information of the model;

[0285] - The preferred AI / ML model structure;

[0286] - The computing power and storage space supporting AI / ML operations.

[0287] Step 2 (S606): The domain management node sends a wireless AI / ML model push request to the device node. Based on the wireless AI / ML capabilities returned by the device node, the domain management node sends an AI / ML model push request including the model ID to this node, asking whether the device node can receive and apply the corresponding AI / ML model.

[0288] Step 3 (S607): The device node confirms the wireless AI / ML model push request (push confirmation message) to the domain management node. After the device node queries the model set description information through the requested model ID to confirm whether it can be supported, it sends a confirmation push request to the domain management node.

[0289] Step 4 (S6031): The domain management node sends a wireless AI / ML model download request to the wireless AI / ML model set. After receiving the device node's consent to receive the AI / ML model push, the domain management node requests the corresponding AI / ML model data from the AI / ML model set.

[0290] Step 5 (S6032): The wireless AI / ML model set sends the requested wireless AI / ML model data to the domain management node. The wireless AI / ML model set sends the requested wireless AI / ML model data to the domain management node.

[0291] Step 6 (S604): The domain management node sends the wireless AI / ML model data to the device node. The domain management node sends the requested wireless AI / ML model data to the requesting device node.

[0292] Through these steps, the domain management node pushes the AI / ML model downloaded from the AI / ML model set to the device node, enabling it to perform a pairing operation with the model in the management node to improve the air interface performance.

[0293] Example 4 Wireless AI / ML Model Activation

[0294] In the embodiments of this application, when the device node activates the AI / ML model applied to a certain module, it needs to confirm with the domain management node. The specific signaling interactions between the nodes include two scenarios: the device node actively activates the AI / ML model of this module and notifies the domain management node (corresponding to Figure 8 ) and the domain management node actively activates the AI / ML model of a certain module of the device node (corresponding to Figure 10 ).

[0295] In some embodiments, as Figure 8 shown, for the case where the device node activates the AI / ML model of a certain module, the following steps and signaling interactions are included:

[0296] Step 0 (S701): The device node sends an AI / ML model activation request (the first activation request message) for a certain module to the domain management node. When certain conditions are met (i.e., the model application conditions provided in the description information of the model set), the device node can consider activating the AI / ML model used by the module to improve performance. The node sends an activation request for the model to the domain management node, mainly including the model identifier and the first activation time limit information, such as information including the expected activation time (i.e., the activation start time point and how long to activate, etc.).

[0297] Step 1 (S702): The domain management node confirms the activation of the wireless AI / ML model to the device node (the first activation confirmation message). The domain management node confirms to the device node that the model can be activated, or gives the second activation time limit information, such as information including the recommended activation start time point and activation duration, etc., and records the model activation status of the node.

[0298] Step 2 (S703): The device node activates the corresponding wireless AI / ML model. The device node activates the corresponding wireless AI / ML model at the given time point according to the confirmed activation information.

[0299] In some embodiments, in certain scenarios, the device node can be configured by the domain management node to spontaneously activate the AI / ML model and notify the domain management node of the activation status of the model after activating a certain model.

[0300] In some embodiments, for the case where the domain management node activates the wireless AI / ML model of the device node, as Figure 10 shown, the following steps and interaction signaling can be included:

[0301] Step 0 (S708): The domain management node sends a wireless AI / ML model activation request (the second activation request message) to the device node. When certain conditions are met (i.e., the model application conditions provided in the description information of the model set), the domain management node considers sending an activation request for the AI / ML model of a certain module to the device node to improve performance, so it is necessary to send an activation request for the model to the device node, mainly including the model identifier and the third activation time limit information, such as information including the expected activation time (i.e., the activation start time point and how long to activate, etc.).

[0302] Step 1 (S709): The device node confirms the activation of the wireless AI / ML model to the domain management node (the second activation confirmation message). The device node confirms the model activation request to the domain management node according to the actual situation.

[0303] Step 2 (S710): The device node activates the corresponding wireless AI / ML model. The device node activates the corresponding wireless AI / ML model at the given time point according to the confirmed activation information.

[0304] In the scenario of activating the AI / ML model of the device node by the domain management node, it is more necessary to activate the bilateral model. Therefore, the device node needs to clarify with the device management node whether to activate a certain AI / ML model in order to successfully pair and use the model in the device management node.

[0305] Example 5 Performance Detection of Wireless AI / ML Model

[0306] In the embodiment of the present application, after the device node activates the AI / ML model of a certain module, it is necessary to detect the performance of the model to determine whether the performance is improved after using the model. The signaling interaction between nodes in this process includes two scenarios: the device node actively triggers the performance detection of the wireless AI / ML model (corresponding to Figure 11 ) and the domain management node triggers the performance detection of the wireless AI / ML model of the device node (corresponding to Figure 12 ).

[0307] In some embodiments, as Figure 11 shown, for the case where the device node actively triggers the performance detection of the wireless AI / ML model, it includes the following steps and signaling interactions:

[0308] Step 0 (S801): The device node sends a performance detection request (performance detection request message) for the AI / ML model of a certain module to the domain management node. When it is necessary to detect the model performance (periodic or event-triggered), the device node sends a performance detection request for the corresponding model to the domain management node through the model ID, including detection time limit information, such as the start time of detection and the time length or period, and expects the management node to provide auxiliary information according to the model description information, such as the label data for performance comparison of the ideal module for improving the estimation accuracy.

[0309] Step 1 (S802): The domain management node confirms the performance detection of the AI / ML model to the device node and sends the corresponding auxiliary data. The domain management node confirms that the device node can start the performance detection of the model and provides the corresponding auxiliary data according to the request, such as the label data for performance comparison of the ideal module for improving the estimation accuracy.

[0310] Step 2 (S803): The device node starts to perform the performance detection of the AI / ML model of the module and obtains the performance detection result. After receiving the confirmation and auxiliary data from the domain management node, the device node starts to evaluate the performance of the AI / ML model of the module at the given time, such as comparing the output of the activated model with the ideal value or the gap with the non-AI / ML model.

[0311] Step 3 (S804): The device node sends the performance detection result to the domain management node. As an optional step, the device node can send the performance detection result to the domain management node or store the result locally.

[0312] In some embodiments, for the case where the domain management node triggers the performance detection of the wireless AI / ML model of the device node, the following steps and interaction signaling are included:

[0313] Step 0 (S805): The domain management node sends an AI / ML model performance detection request (performance detection request message) for a certain module to the device node. When it is necessary to detect the model performance (periodically or event-triggered), the domain management node can trigger the performance detection request for the model corresponding to the model ID, including the detection start time and the time length or period, and provide auxiliary information according to the model description information, such as the ideal label data for performance comparison.

[0314] Step 1 (S802): The domain management node sends AI / ML model performance detection auxiliary data to the device node. The domain management node provides the corresponding auxiliary data to the device node, such as the ideal label data for comparison.

[0315] Step 2 (S803): The device node starts to perform the AI / ML model performance detection. After receiving the auxiliary data from the domain management node, the device node starts to evaluate the performance of the AI / ML model at the given time, such as comparing the output of the activated model with the ideal value or the gap from the non-AI / ML model.

[0316] Step 3 (S804): The device node sends the AI / ML model performance detection result to the domain management node. Since the detection process is triggered by the domain management node, after the detection is completed, the device node needs to send the performance detection result to the domain management node.

[0317] Embodiment 6 Wireless AI / ML Model Deactivation

[0318] When the device node deactivates the AI / ML model of a certain module, it needs to confirm with the domain management node. The specific signaling interaction between the nodes is as Figure 9 shown. The following steps and interaction signaling are included:

[0319] Step 0 (S705): The device node sends a wireless AI / ML model deactivation indication (model deactivation indication message) to the domain management node. When certain events occur, such as performance degradation, insufficient computing power, and scenario changes, etc., the device node needs to deactivate the AI / ML model applied to the module and needs to send a model deactivation indication to the domain management node based on the model ID.

[0320] Step 1 (S704): The domain management node sends a deactivation request message for the AI / ML model to the device node. In some predefined events, such as scenario changes and model updates, the domain management node needs to send a request to deactivate a certain AI / ML model to the device node that has activated the model, mainly including the model identifier and deactivation time limit information, including the information about the deactivation time point.

[0321] Step 2 (S706): The device node deactivates the corresponding wireless AI / ML model. After receiving the deactivation request instruction from the domain management node, the device node deactivates the AI / ML model at the given activation time point.

[0322] Step 3 (S707): The device node sends a deactivation confirmation message for the AI / ML model to the domain management node. After successfully deactivating the AI / ML model, the device node sends a confirmation message to the domain management node.

[0323] In some embodiments, in certain scenarios, the device node can be configured by the domain management node to spontaneously deactivate the AI / ML model and notify the domain management node of the deactivation status of the model after deactivating a certain model.

[0324] Embodiment Seven Wireless AI / ML Model Update

[0325] Benefiting from the local wireless data collection function, some device nodes with model training capabilities may update the used wireless AI / ML model in order to better adapt to the application scenario. Therefore, after updating the model, the device can consider uploading and updating it to the wireless AI / ML model set. The specific signaling interaction between nodes in this process is as Figure 13 shown. It includes the following steps and signaling interactions:

[0326] Step 0 (S901): The device node sends an AI / ML model update request message for a certain module to the domain management node. After the device node performs local training and model update, the device node can send an indication of the local wireless AI / ML model update to the domain management node through the model ID and request to update the AI / ML model set.

[0327] Step 1 (S902): The domain management node confirms the AI / ML model update request for this module to the device node (model update confirmation message). When the domain management node believes that the model updated by the device node needs to update the wireless AI / ML model set, it sends a confirmation message to the device node and instructs it to upload the updated model data or the updated model data.

[0328] Step 2 (S903): The device node sends wireless AI / ML model update data to the domain management node. The device node sends the updated wireless AI / ML model or partial update information to the domain management node.

[0329] Step 3 (S9041): The domain management node sends wireless AI / ML model update data to the AI / ML model set. The domain management node sends the updated wireless AI / ML model data or partial update information to the wireless AI / ML model set.

[0330] Step 4 (S9043): The wireless AI / ML model set sends a wireless AI / ML model update completion confirmation (first update confirmation message) to the domain management node. The wireless AI / ML model set sends confirmation information after the wireless AI / ML model is updated to the domain management node.

[0331] Step 5 (S9044): The domain management node sends a wireless AI / ML model update completion confirmation (second update confirmation message) to the device node. The domain management node sends confirmation information after the wireless AI / ML model is updated to the device node.

[0332] In some embodiments, when an AI / ML model in the AI / ML model set is updated, it is necessary to notify the device node running the model through the domain management node and update the corresponding model. The specific signaling interaction between nodes is as Figure 14 shown. It includes the following steps and signaling interactions:

[0333] Step 0 (S1002): The wireless AI / ML model set sends an AI / ML model update indication message to the domain management node. The wireless AI / ML model set sends an AI / ML model update indication to the domain management node, indicating that a certain model(s) in the set has been updated.

[0334] Step 1 (S1001): The domain management node sends a model update indication message to the device node running the AI / ML model. The domain management node sends a model update indication to the device node using the model according to the model information and the registered activated AI / ML models.

[0335] Step 2 (S1003): The device node deactivates the corresponding wireless AI / ML model. After receiving the information that the corresponding AI / ML model needs to be updated, the relevant device node deactivates the model.

[0336] Step 3 (S1004): The domain management node sends an AI / ML model update push request message to the device node. The domain management node sends an AI / ML model update push request to the device node.

[0337] Step 4 (S1005): The device node confirms the AI / ML model update push request (update push confirmation message) to the domain management node. The device node confirms the model update push request to the domain management node.

[0338] Step 5 (S1007): The domain management node sends the AI / ML model update data download request (model set update request message) to the AI / ML model set. The domain management node sends the updated AI / ML model or partial update information to the AI / ML model set.

[0339] Step 6 (S1008): The wireless AI / ML model set sends the requested wireless AI / ML model update data to the domain management node. The AI / ML model set sends the updated AI / ML model or partial update information to the domain management node.

[0340] Step 7 (S1009): The domain management node sends the AI / ML model update data (model update message) to the device node. The domain management node sends the updated AI / ML model or partial update information to the device node.

[0341] Step 8 (S1010): The domain management node sends the updated AI / ML model activation request message to the device node. The domain management node sends the activation request for the updated AI / ML model to the device node.

[0342] Step 9 (S1011): The device node confirms the activation of the updated AI / ML model to the management node. The device node, according to the actual situation, confirms the activation request for the updated model to the domain management node.

[0343] Step 10 (S1012): The device node activates the updated wireless AI / ML model. The device node activates the updated wireless AI / ML model.

[0344] It should be noted that all the above interaction signaling between the domain management node and the device node can be implemented in the radio link management, media access layer, and physical layer control signaling of the communication system.

[0345] The embodiments of the present application provide a communication system. Under similar hardware conditions, the communication system configuration model is more convenient than 3GPP. Moreover, the protocol architecture of the communication system provided by the embodiments of the present application is essentially different from that of 3GPP. The protocol design of the communication system is to ensure localized short-distance wireless connections and achieve effective and reliable data transmission between nodes within a certain range (i.e., the coverage range of the domain management node); 3GPP is to ensure wide-area coverage and achieve ubiquitous access and communication of user equipment, including a series of interface interactions such as the core network and radio access network. For enhancing the air interface using AI / ML models, the signaling interaction of the communication system should be implemented on the air interface and does not involve network architectures and protocols such as the core network. Additionally, currently, 3GPP is discussing solutions for AI / ML-based beam management, improving positioning accuracy, and compressing and transmitting channel state information. The goal is to design specific signaling enhancement solutions, such as enhanced CSI feedback report design for supporting beam management and enhanced positioning measurement data feedback. The overall framework still reuses the existing 3GPP frameworks, such as the CSI feedback framework and positioning-related protocols. The present application will consider the AI / ML capability support framework in the communication system and design the signaling interaction for the model itself. In summary, different from several related signaling enhancements for specific use cases in the 3GPP system, the present application considers implementing support for AI / ML capabilities in the protocol architecture of the communication system, especially the management related to AI / ML models and the design of operation interaction processes between nodes.

[0346] In a fifth aspect, to implement the above communication method, as Figure 15 shown, the embodiments of the present application provide a device 1500 (network unit or terminal device), which may include at least one processor 1501 and at least one transceiver 1502 coupled to at least one processor 1501. The transceiver 1502 may include at least one separate receiving circuit system and transmitting circuit system, or at least one integrated receiving circuit system and transmitting circuit system. The at least one processor 1501 may be a central processing unit (CPU), a microprocessor unit (MPU), a digital signal processor (DSP), or a field-programmable gate array (FPGA), etc.

[0347] According to some embodiments of the present application, when the device 1500 is a domain management node, the processor 1501 is configured to:

[0348] Obtain relevant information of the model set, where the relevant information of the model set includes at least function description information of the air interface and application condition information of the air interface;

[0349] Generate a first message based on the relevant information of the model set;

[0350] Transmit the first message via a transceiver, where the first message includes indication information related to the support capabilities of the domain management node or the relevant information of the model set.

[0351] In some embodiments, the processor is configured to:

[0352] When the first message includes indication information related to the support capabilities of the domain management node, receive a first request message via the transceiver, and in response to the first request message, obtain the update status of the relevant information of the model set;

[0353] Based on the update status of the relevant information of the model set, transmit a first response message via the transceiver, where the first response message includes the relevant information of the model set.

[0354] In some embodiments, the functional description information of the air interface is used to characterize the air interface module of the model application;

[0355] The application condition information of the air interface is used to characterize the system configuration and application scenario of the model.

[0356] In some embodiments, the relevant information of the model set further includes one or more of the following:

[0357] The identifier of the model;

[0358] The attribute information of the model; the attribute information of the model includes the parameter format of the model, the input information of the model, and the output information of the model;

[0359] The structure information of the model;

[0360] The parameter information of the model.

[0361] In some embodiments, the processor is configured to:

[0362] Receive the device capability information of the device node via the transceiver, where the device capability information includes the storage status of the model in the device node, the preferred model information, and the computing resource information.

[0363] In some embodiments, the processor is configured to:

[0364] Receive a download request message via the transceiver, obtain the structure information and parameter information of the target model based on the device capability information of the device node and the download request message, generate a second message, and transmit the second message via the transceiver;

[0365] Alternatively, a request message is pushed via a transceiver's transmission model. In the case where a push confirmation message is received via the transceiver, based on the device capability information of the device node and the push confirmation message, the structural information and parameter information of the target model are obtained, a second message is generated, and the second message is transmitted via the transceiver.

[0366] In some embodiments, the processor is configured to:

[0367] Receive a first activation request message via the transceiver, the first activation request message including an identifier of the model to be activated and first activation time limit information;

[0368] Transmit a first activation confirmation message via the transceiver, the first activation confirmation message including confirmation information of the domain management node and second activation time limit information, the second activation time limit information being determined according to the first activation time limit information and the computing resource information of the domain management node; the first activation confirmation message is used to instruct the device node to perform an activation process on the model to be activated based on the second activation time limit information and the identifier of the model to be activated;

[0369] In the case where a deactivation condition is met, transmit a deactivation request message via the transceiver, the deactivation request message including an identifier of the model to be deactivated and deactivation time limit information; the deactivation request message is used to instruct the device node to perform a deactivation process on the model to be deactivated based on the deactivation time limit information.

[0370] In some embodiments, the processor is configured to:

[0371] Transmit a second activation request message via the transceiver, the second activation request message including an identifier of the model to be activated and third activation time limit information;

[0372] Receive a second activation confirmation message via the transceiver; the second activation confirmation message is used to represent that the device node performs an activation process on the model to be activated based on the third activation time limit information and the identifier of the model to be activated;

[0373] In the case where a deactivation condition is met, transmit a deactivation request message via the transceiver, the deactivation request message including an identifier of the model to be deactivated and deactivation time limit information; the deactivation request message is used to instruct the device node to perform a deactivation process on the model to be deactivated based on the deactivation time limit information.

[0374] In some embodiments, the processor is configured to:

[0375] Receive or transmit a performance detection request message via the transceiver, the performance detection request message including an identifier of the model to be detected and detection time limit information;

[0376] Transmit performance detection auxiliary data via the transceiver, the performance detection auxiliary data including at least one label data information;

[0377] Receive a performance detection result via a transceiver, where the performance detection result is obtained by a device node detecting a model to be detected.

[0378] In some embodiments, the processor is configured to:

[0379] Receive a model update request message via a transceiver, where the model update request message includes an identifier of a model to be updated;

[0380] When it is determined to update the model to be updated, transmit a model update confirmation message via the transceiver;

[0381] Receive information related to the model to be updated via the transceiver, and update the model to be updated in the storage module based on the information related to the model to be updated.

[0382] In some embodiments, the processor is configured to:

[0383] When there is an updated model in the storage module, transmit a model update indication message via the transceiver, where the model update indication message includes an identifier of the updated model, and the model update indication message is used to instruct the device node to deactivate the updated model;

[0384] Transmit a model update push request message via the transceiver;

[0385] Receive a model update push confirmation message via the transceiver, and obtain the updated model from the storage module based on the model update push confirmation message;

[0386] Transmit a model update message and a model activation request message via the transceiver; the model update message includes information related to the updated model, and the model activation request message is used to instruct the device node to activate the updated model.

[0387] According to some embodiments of the present application, when device 1500 is a device node, the processor is configured to:

[0388] Receive a first message via the transceiver, where the first message includes indication information related to the support capabilities of the domain management node or information related to the model set; the first message is generated by the domain management node based on the information related to the model set, and the information related to the model set is obtained by the domain management node, and the information related to the model set at least includes functional description information of the radio interface and application condition information of the radio interface.

[0389] In some embodiments, the processor is configured to:

[0390] When the first message includes indication information related to the support capabilities of the domain management node, transmit a first request message via the transceiver, so that the domain management node, in response to the first request message, obtains the update status of the information related to the model set and generates a first response message;

[0391] Receive a first response message via a transceiver, the first response message including information related to a model set.

[0392] In some embodiments, the functional description information of the air interface is used to characterize the air interface module for model application;

[0393] The application condition information of the air interface is used to characterize the system configuration and application scenario of the model.

[0394] In some embodiments, the information related to the model set further includes one or more of the following:

[0395] The identifier of the model;

[0396] The attribute information of the model; the attribute information of the model includes the parameter format of the model, the input information of the model, and the output information of the model;

[0397] The structure information of the model;

[0398] The parameter information of the model.

[0399] In some embodiments, the processor is configured to:

[0400] Transmit, via a transceiver, the device capability information of the device node, the device capability information including the storage situation of the model in the device node, the preferred model information, and the computing resource information.

[0401] In some embodiments, the processor is configured to:

[0402] Transmit a download request message via a transceiver; receive a second message via the transceiver, the second message being generated by the domain management node based on the device capability information of the device node and the download request message to obtain the structure information and parameter information of the target model;

[0403] Alternatively, receive a model push request message via a transceiver, and receive a second message via the transceiver, the second message being generated by the domain management node based on the device capability information of the device node and the push confirmation message to obtain the structure information and parameter information of the target model in the case where a push confirmation message is received via the transceiver.

[0404] In some embodiments, the processor is configured to:

[0405] Transmit a first activation request message via a transceiver, the first activation request message including the identifier of the model to be activated and the first activation time limit information;

[0406] Receive a first activation confirmation message via a transceiver, the first activation confirmation message including the confirmation information of the domain management node and the second activation time limit information, the second activation time limit information being determined according to the first activation time limit information and the computing resource information of the domain management node;

[0407] Activate the model to be activated based on the second activation time limit information and the identifier of the model to be activated;

[0408] When the deactivation condition is met, receive a deactivation request message via the transceiver, where the deactivation request message includes the identifier of the model to be deactivated and deactivation time limit information;

[0409] Deactivate the model to be deactivated based on the deactivation time limit information.

[0410] In some embodiments, the processor is configured to:

[0411] Receive a second activation request message via the transceiver, where the second activation request message includes the identifier of the model to be activated and third activation time limit information;

[0412] Transmit a second activation confirmation message via the transceiver; the second activation confirmation message is used to indicate that the device node activates the model to be activated based on the third activation time limit information and the identifier of the model to be activated;

[0413] When the deactivation condition is met, receive a deactivation request message via the transceiver, where the deactivation request message includes the identifier of the model to be deactivated and deactivation time limit information;

[0414] Deactivate the model to be deactivated based on the deactivation time limit information.

[0415] In some embodiments, the processor is configured to:

[0416] Transmit or receive a performance detection request message via the transceiver, where the performance detection request message includes the identifier of the model to be detected and detection time limit information;

[0417] Receive performance detection auxiliary data via the transceiver, where the performance detection auxiliary data includes at least one tag data information;

[0418] Transmit a performance detection result via the transceiver, where the performance detection result is obtained by the device node detecting the model to be detected.

[0419] In some embodiments, the processor is configured to:

[0420] Transmit a model update request message via the transceiver, where the model update request message includes the identifier of the model to be updated;

[0421] When the domain management node determines to update the model to be updated, receive a model update confirmation message via the transceiver;

[0422] Transmit the relevant information of the model to be updated via the transceiver, so that the domain management node updates the model to be updated in the storage module based on the relevant information of the model to be updated.

[0423] In some embodiments, the processor is configured to:

[0424] When there is an updated model in the storage module, receive a model update indication message via the transceiver. The model update indication message includes the identifier of the updated model, and the model update indication message is used to instruct the device node to deactivate the updated model;

[0425] Receive a model update push request message via the transceiver;

[0426] Transmit a model update push confirmation message via the transceiver, and obtain the updated model from the storage module based on the model update push confirmation message;

[0427] Receive a model update message and a model activation request message via the transceiver; the model update message includes the relevant information of the updated model, and the model activation request message is used to instruct the device node to activate the updated model.

[0428] The description of the above device embodiments is similar to the description of the above method embodiments, and has similar beneficial effects to the method embodiments. For the technical details not disclosed in the device embodiments of the present application, please refer to the description of the method embodiments of the present application for understanding.

[0429] It should be noted that in the embodiments of the present application, if the above communication method is implemented in the form of software function modules and sold or used as an independent product, it can also be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the embodiments of the present application essentially or the part that contributes to the related technology can be embodied in the form of a software product. The computer software product is stored in a storage medium and includes several instructions for causing a computer device (which can be a personal computer, a server, or a network device, etc.) to execute all or part of the methods described in the embodiments of the present application. The foregoing storage medium includes: various media such as USB flash drives, mobile hard disks, read-only memories (ROMs), magnetic disks, or optical discs that can store program codes. In this way, the embodiments of the present application are not limited to any specific combination of hardware and software.

[0430] In a sixth aspect, to implement the above communication method, an embodiment of the present application provides an electronic device, including a memory and a processor. The memory stores a computer program that can run on the processor, and when the processor executes the program, it implements the steps in the communication method provided in the above embodiments.

[0431] In a seventh aspect, an embodiment of the present application provides a storage medium, that is, a computer-readable storage medium, on which a computer program is stored. When the computer program is executed by a processor, the steps in the communication method provided in the above embodiment are implemented.

[0432] An embodiment of the present application also provides a computer program product containing instructions. When it runs on a computer, it causes the computer to execute the steps in the method provided in the above method embodiment.

[0433] It should be understood that the descriptions of the above storage medium and device embodiments are similar to the descriptions of the above method embodiments and have similar beneficial effects to the method embodiments. For the technical details not disclosed in the storage medium, storage medium and device embodiments of the present application, please refer to the descriptions of the method embodiments of the present application for understanding.

[0434] It should be understood that the "one embodiment" or "an embodiment" or "some embodiments" mentioned throughout the specification means that the specific features, structures or characteristics related to the embodiment are included in at least one embodiment of the present application. Therefore, the appearances of "in one embodiment" or "in an embodiment" or "in some embodiments" throughout the specification do not necessarily refer to the same embodiment. In addition, these specific features, structures or characteristics can be combined in one or more embodiments in any suitable manner. It should be understood that in various embodiments of the present application, the magnitudes of the serial numbers of the above processes do not mean the order of execution. The order of execution of each process should be determined by its function and internal logic, and should not constitute any limitation to the implementation process of the embodiments of the present application. The serial numbers of the embodiments of the present application above are only for description and do not represent the advantages or disadvantages of the embodiments. The above descriptions of the various embodiments tend to emphasize the differences between the various embodiments. Their similarities or similarities can be referred to each other. For the sake of brevity, they will not be repeated herein.

[0435] It should also be noted that in the present application, the term "comprising", "including" or any other variant thereof is intended to cover a non-exclusive inclusion, so that a process, method, article or device including a series of elements not only includes those elements, but also includes other elements not expressly listed, or also includes elements inherent to such process, method, article or device. Without further limitation, the element defined by the statement "including one..." does not exclude the existence of additional identical elements in the process, method, article or device including the element.

[0436] The serial numbers of the above embodiments of the present application are only for description and do not represent the advantages or disadvantages of the embodiments.

[0437] The methods disclosed in several method embodiments provided by the present application can be arbitrarily combined without conflict to obtain new method embodiments.

[0438] The features disclosed in several product embodiments provided by this application can be arbitrarily combined without conflict to obtain new product embodiments.

[0439] The features disclosed in several method or device embodiments provided by this application can be arbitrarily combined without conflict to obtain new method embodiments or device embodiments.

[0440] The above is only a preferred embodiment of this application and is not intended to limit the protection scope of this application. Any modifications, equivalent replacements, improvements, etc. made within the spirit and principle of this application shall be included in the protection scope of this application.

Claims

1. A domain management node, the domain management node comprising: a transceiver; and a processor coupled to the transceiver, wherein the processor is configured to: obtain relevant information of a model set, the relevant information of the model set at least including function description information of an air interface and application condition information of the air interface; generate a first message based on the relevant information of the model set; transmit the first message via the transceiver, the first message including indication information related to the support capabilities of the domain management node or the relevant information of the model set.

2. The domain management node according to claim 1, wherein the processor is configured to: in the case where the first message includes indication information related to the support capabilities of the domain management node, receive a first request message via the transceiver, and in response to the first request message, obtain an update situation of the relevant information of the model set; transmit a first response message via the transceiver based on the update situation of the relevant information of the model set, the first response message including the relevant information of the model set.

3. The domain management node according to claim 1, wherein the relevant information of the model set further includes one or more of the following: the identifier of the model; the attribute information of the model; the attribute information of the model includes the parameter format of the model, the input information of the model, and the output information of the model; the structure information of the model; the parameter information of the model.

4. The domain management node according to claim 1, wherein the processor is configured to: receive device capability information of a device node via the transceiver, the device capability information including the storage situation of the model in the device node, the preferred model information, and the computing resource information.

5. The domain management node according to claim 4, wherein the processor is configured to: receive a download request message via the transceiver, obtain the structure information and parameter information of a target model based on the device capability information of the device node and the download request message, generate a second message, and transmit the second message via the transceiver; alternatively, transmit a model push request message via the transceiver, and in the case of receiving a push confirmation message via the transceiver, obtain the structure information and parameter information of the target model based on the device capability information of the device node and the push confirmation message, generate a second message, and transmit the second message via the transceiver.

6. The domain management node according to claim 1, wherein the processor is configured to: receive a first activation request message via the transceiver, the first activation request message including the identifier of a model to be activated and first activation time limit information; transmit a first activation confirmation message via the transceiver, the first activation confirmation message including confirmation information of the domain management node and second activation time limit information, the second activation time limit information being determined according to the first activation time limit information and the computing resource information of the domain management node; the first activation confirmation message is used to instruct the device node to perform activation processing on the model to be activated based on the second activation time limit information and the identifier of the model to be activated; When the deactivation condition is met, a deactivation request message is transmitted via the transceiver, and the deactivation request message includes the identifier of the model to be deactivated and deactivation time limit information; The deactivation request message is used to instruct the device node to perform a deactivation process on the model to be deactivated based on the deactivation time limit information.

7. The domain management node according to claim 1, wherein the processor is configured to: Transmit a second activation request message via the transceiver, and the second activation request message includes the identifier of the model to be activated and third activation time limit information; Receive a second activation confirmation message via the transceiver; the second activation confirmation message is used to represent that the device node performs an activation process on the model to be activated based on the third activation time limit information and the identifier of the model to be activated; When the deactivation condition is met, a deactivation request message is transmitted via the transceiver, and the deactivation request message includes the identifier of the model to be deactivated and deactivation time limit information; The deactivation request message is used to instruct the device node to perform a deactivation process on the model to be deactivated based on the deactivation time limit information.

8. The domain management node according to any one of claims 1 to 7, wherein the processor is configured to: Receive a model update request message via the transceiver, and the model update request message includes the identifier of the model to be updated; When it is determined to update the model to be updated, transmit a model update confirmation message via the transceiver; Receive the relevant information of the model to be updated via the transceiver, and update the model to be updated in the storage module based on the relevant information of the model to be updated.

9. The domain management node according to any one of claims 1 to 7, wherein the processor is configured to: When there is an updated model in the storage module, transmit a model update indication message via the transceiver, and the model update indication message includes the identifier of the updated model, and the model update indication message is used to instruct the device node to perform a deactivation process on the updated model; Transmit a model update push request message via the transceiver; Receive a model update push confirmation message via the transceiver, and obtain the updated model from the storage module based on the model update push confirmation message; Transmit a model update message and a model activation request message via the transceiver; the model update message includes the relevant information of the updated model, and the model activation request message is used to instruct the device node to perform an activation process on the updated model.

10. A device node, the device node includes: A transceiver; And A processor coupled to the transceiver, wherein the processor is configured to: Receive a first message via the transceiver, and the first message includes indication information related to the support capabilities of the domain management node or relevant information of the model set; the first message is generated by the domain management node based on the relevant information of the model set, and the relevant information of the model set is obtained by the domain management node, and the relevant information of the model set at least includes function description information of the radio interface and application condition information of the radio interface.

Images