Methods and apparatus for constructing topologies for artificial intelligence or machine learning

Abstract

Aspects of the present disclosure provide methods and apparatus for configuring a topology for artificial intelligence or machine learning (AI / ML) in a wireless communication network to solve problems such as data heterogeneity in traditional federated learning-based training processes. The topology may be centrally configured by a network device or may be self-organized by nodes in the topology using announcement and discovery messaging. Nodes in the topology may be configured as Type 1 nodes configured to collect and aggregate multiple AI / ML models to obtain a first-type AI / ML model, or as Type 2 nodes configured to obtain a second-type AI / ML model with its own training data without aggregation operations. The configured topology includes connections between at least one Type 1 node and zero or more Type 2 nodes and / or connections between at least two Type 1 nodes.

Description

[Technical Field]

[0001] This disclosure relates generally to wireless communication, and in specific embodiments to methods and apparatus for constructing topologies for artificial intelligence or machine learning (AI / ML). [Background technology]

[0002] Artificial intelligence technologies can be applied to communications, including AI-based communications in the physical layer and / or AI-based communications in the media access control (MAC) layer. For example, in the physical layer, AI-based communications can be aimed at optimizing component design and / or improving algorithmic performance. In the MAC layer, AI-based communications can be aimed at leveraging AI capabilities to learn, predict, and / or make decisions to solve complex optimization problems with the best possible strategies and / or optimal solutions. For example, it can optimize functionality in the MAC layer.

[0003] In some implementations, an AI architecture in a wireless network can include multiple nodes. These nodes can be organized in either a centralized or distributed configuration, and both can be located in the access network, core network, edge computing system, or third-party network. Centralized training and computing architectures are sometimes limited by significant communication overhead and strict user data privacy. Distributed training and computing architectures can include several frameworks, such as distributed machine learning and federative learning.

[0004] However, communication in wireless communication systems, including communications related to AI training across multiple nodes, typically occurs over less-than-ideal channels. For example, less-than-ideal conditions such as electromagnetic interference, signal degradation, phase delay, fading, and other non-ideal conditions can attenuate and / or distort communication signals, or interfere with or reduce the system's communication capabilities.

[0005] Traditional AI training processes typically rely on hybrid automatic repeat request (HARQ) feedback and retransmission processes to ensure that data communicated between devices involved in AI training is successfully received. However, the communication overhead and delay associated with such retransmissions can be problematic.

[0006] Furthermore, the processing capacity and / or availability of training data in an AI training process can vary significantly between different nodes / devices. This means that the capabilities of different nodes participating productively in the AI ​​training process can differ considerably. In practice, such differences mean that the training delay in AI training processes involving multiple nodes / devices, such as distributed learning or federative learning-based AI training processes, will be dominated by the node / device with the greatest delay due to communication and / or computational delays.

[0007] Federated learning, also known as collaborative learning, is a machine learning technique that trains algorithms across multiple distributed edge devices or servers. Each distributed edge device or server holds local data samples, but these cannot be exchanged with other devices or servers. Federated learning techniques are the opposite of traditional centralized machine learning methods in the following respect: In federated learning methods, local data samples are not shared, whereas in traditional centralized machine learning methods, all local datasets are uploaded to a single server.

[0008] In a federated learning-based (FL-based) AI training process, a network node / device / node initializes a global AI model, samples a group of user devices, and broadcasts the global AI model parameters to the user devices. Each user device then initializes its local AI model using the global AI model parameters and updates (trains) its local AI model using its own data. Each user device can then report the parameters of its updated local AI model to the network device, which then aggregates the updated parameters reported by the user devices and updates the global AI model. The procedure described above is one iteration of a conventional FL-based AI model training procedure. The network device and participating user devices typically perform multiple iterations until the AI ​​model converges sufficiently to meet one or more training goals / criteria and the AI ​​model is complete.

[0009] However, different user devices participating in an FL-based AI training process may observe different training datasets that do not necessarily represent all distributions of training data observed by other user devices participating in the FL-based AI training process. In other words, the training data may not be independently and identically distributed (non-iid) across devices participating in a conventional FL-based AI training process. Non-iid training data across devices has been shown to degrade convergence speed and model accuracy in conventional FL-based AI training processes. Therefore, data heterogeneity is a common problem in conventional FL-based AI training processes.

[0010] For these and other reasons, there is a need for new protocols and signaling mechanisms that will enable the implementation of new AI-enabled applications and processes while minimizing the signaling and communication overhead and latency associated with existing AI training procedures. [Overview of the project]

[0011] Traditional federated learning-based (FL-based) AI training processes have limitations. For example, as mentioned above, data heterogeneity is a common problem in traditional FL-based AI training processes. Also, the server node needs to collect a large amount of training data (e.g., gradients of client node updates) from multiple related client nodes. Furthermore, in traditional FL-based AI training processes, the server node and related client nodes must have the same AI / ML model structure. However, since different local nodes may support different AI / ML model structures, data heterogeneity can be a problem in traditional FL-based AI training processes.

[0012] Aspects of this disclosure provide solutions to overcome the aforementioned limitations, for example, in certain embodiments, methods and apparatus for constructing topologies for artificial intelligence or machine learning (AI / ML).

[0013] According to a first broad aspect of this disclosure, a method for configuring a topology for artificial intelligence or machine learning (AI / ML) in a wireless communication network is provided. The method according to the first broad aspect of this disclosure may include the step of receiving information from a node, including a report relating to the node's AI / ML capabilities. The method according to the first broad aspect of this disclosure may further include the step of configuring a node based on the received information, wherein configuring a node includes configuring a node type for the node, where the configured node type is one of a plurality of node types. For example, the plurality of node types may include Type 1, which represents a node configured to collect a plurality of AI / ML models and aggregate the collected AI / ML models to obtain a first type AI / ML model, or Type 2, which represents a node configured to obtain a second type AI / ML model having a set of training data without aggregation operations. The method according to the first broad aspect of this disclosure may further include the step of configuring one or more other nodes associated with the configured node based on the configured node type. The configured topology supports AI / ML model transfer over the wireless interface of the wireless communication network and includes connections between at least one Type 1 node and zero or more Type 2 nodes, and / or connections between at least two Type 1 nodes.

[0014] In some embodiments of the methods according to the first broad aspect of this disclosure, the configured node type indicates that the node is a type 1 node, and one or more other nodes are as follows: One or more Type 2 nodes, A second type 1 node that provides a first type AI / ML model of a second type 1 node to the node, or A third type 1 node that receives the first type AI / ML model from the node, It includes at least one of the following.

[0015] In some embodiments of the methods according to the first broad aspect of this disclosure, one or more other nodes include one or more Type 2 nodes, and the steps of configuring the nodes are: This includes the step of configuring nodes to collect each Type 2 AI / ML model from one or more Type 2 nodes.

[0016] In some embodiments of the methods according to the first broad aspect of this disclosure, the configured node type indicates that the node is a type 2 node, and one or more other nodes include one or more type 1 nodes connected to the node.

[0017] In some embodiments of the methods according to the first broad aspect of this disclosure, the information further includes requests made by nodes configured as Type 1 nodes.

[0018] In some embodiments of the methods according to the first broad aspect of this disclosure, a node is connected to one or more other nodes via a sidelink using device-to-device (D2D) communication, via a network device, or via an interface between network devices.

[0019] In some embodiments of the methods according to the first broad aspect of this disclosure, the node is a user device (UE), repeater, base station (BS), transmit / receive point (TRP), edge device, network system, or integrated access backhaul (IAB) node.

[0020] According to a second broad aspect of this disclosure, a method for configuring a topology for artificial intelligence or machine learning (AI / ML) in a wireless communication network is provided. The method according to the second broad aspect of this disclosure may include the step of establishing an aggregated connection with a second node based on aggregated acknowledgment messages. The topology configured according to the method according to the second broad aspect of this disclosure supports AI / ML model transfer over the wireless interface of the wireless communication network, and: A connection between at least one Type 1 node and zero or more Type 2 nodes, Connection between at least two Type 1 nodes, It includes at least one of the following: A Type 1 node is configured to collect multiple AI / ML models and aggregate the collected AI / ML models to obtain a Type 1 AI / ML model, while a Type 2 node is configured to obtain a Type 2 AI / ML model with a set of training data without performing any aggregation operations.

[0021] Optionally, before establishing an aggregation connection with a second node, the method further includes the step of the first node sending an aggregation message to the second node. After sending the aggregation message, the first node may receive an acknowledgment message from the second node.

[0022] Furthermore, the steps to be established are selective.

[0023] In a method according to a second broad aspect of the present disclosure, the message for aggregation is a discovery message containing aggregation information used by the second node to discover the first node.

[0024] In some embodiments of the methods according to the second broad aspect of this disclosure, the aggregated information used by the second node includes the aggregation capabilities of the first node.

[0025] In some embodiments of the methods according to the second broad aspect of this disclosure, the discovery message includes at least one of a model acquisition indicator or a reference AI / ML model.

[0026] In some embodiments of the methods according to the second broad aspect of this disclosure, the model collection indicator includes a distillation indicator, an expansion indicator, or one of a distillation and expansion indicator.

[0027] In some embodiments of the Method relating to a second broad aspect of the Disclosure, the message for aggregation is an aggregation request message, which includes information or an aggregation request relating to a second type AI / ML model of a first node.

[0028] In some embodiments of the Method relating to a second broad aspect of the present disclosure, the first node is a Type 2 node, and the aggregate request message includes information relating to a Type 2 AI / ML model of the first node.

[0029] In some embodiments of the Method relating to a second broad aspect of this disclosure, information relating to the second type AI / ML model of the first node is as follows: Information related to the neural network of the second type AI / ML model in the first node, Information regarding the size of the second type AI / ML model in the first node, or Information regarding the complexity of the second type AI / ML model in the first node. It includes at least one of the following.

[0030] In some embodiments of the Method relating to a second broad aspect of the Disclosure, the first node is a Type 1 node, and the aggregation request message includes an aggregation request.

[0031] In some embodiments of the Method relating to a second broad aspect of the Disclosure, the second node is one of several other nodes in a wireless communication network, the first node can be connected to only one of several other nodes, and when a processor executable instruction is executed, the processor... The first node further includes a processor-executable instruction that causes it to send each of the confirmations to at least one of several other nodes, informing them whether a connection to the first node has been successfully established.

[0032] In some embodiments of the methods according to the second broad aspect of this disclosure, the first node is a type 2 node and the second node is a type 1 node.

[0033] In some embodiments of the method according to a second broad aspect of the present disclosure, the first node is a first type 1 node, the second node is a second type 1 node, and the second type 1 node provides the first type AI / ML model of the second node to the first node or receives the first type AI / ML model of the first node from the first node.

[0034] In some embodiments of the Method relating to a second broad aspect of the present disclosure, the first node is a Type 1 node that is associated only with one or more other Type 1 nodes.

[0035] In some embodiments of the Method relating to a Second Broader Aspect of the Disclosure, the step of transmitting a message for aggregation includes the step of broadcasting, groupcasting, or unicasting a message for aggregation at predetermined intervals.

[0036] According to a third broad aspect of this disclosure, the Specified Method provides for configuring a topology for artificial intelligence or machine learning (AI / ML) in a wireless communication network. The method according to the third broad aspect of this disclosure may include the step of a first node receiving an aggregation message from a second node. After receiving the aggregation message, the first node may decide whether to connect with the second node and send a response to the second node indicating that decision. The method according to the third broad aspect of this disclosure may further include the step of selectively establishing an aggregation connection with the second node based on that decision. The topology configured according to the method according to the third broad aspect of this disclosure supports AI / ML model transfer over the wireless interface of the wireless communication network, and: A connection between at least one Type 1 node and zero or more Type 2 nodes, Connection between at least two Type 1 nodes, It includes at least one of the following: A Type 1 node is configured to collect multiple AI / ML models and aggregate the collected AI / ML models to obtain a Type 1 AI / ML model, while a Type 2 node is configured to obtain a Type 2 AI / ML model with a set of training data without performing any aggregation operations.

[0037] In a method according to a third broad aspect of the present disclosure, the message for aggregation is a discovery message containing aggregation information used by the first node to discover the second node.

[0038] In some embodiments of the method according to the third broad aspect of this disclosure, the aggregated information used by the first node includes the aggregation capabilities of the second node.

[0039] In some embodiments of the methods according to the third broad aspect of this disclosure, the discovery message includes at least one of a model acquisition indicator or a reference AI / ML model.

[0040] In some embodiments of the methods according to the third broad aspect of this disclosure, the model collection indicator includes a distillation indicator, an expansion indicator, or one of a distillation and expansion indicator.

[0041] In some embodiments of the Method relating to a third broad aspect of the Disclosure, the message for aggregation is an aggregation request message, which includes information or an aggregation request relating to a second type AI / ML model of a first node.

[0042] In some embodiments of the Method relating to a second broad aspect of the present disclosure, the second node is a Type 2 node, and the aggregate request message includes information relating to a Type 2 AI / ML model of the second node.

[0043] In some embodiments of the Method relating to a third broad aspect of this disclosure, information relating to the second type AI / ML model of the second node is as follows: Information related to the neural network of the second type AI / ML model in the second node. Information regarding the size of the second type AI / ML model in the second node, or Information regarding the complexity of the second type AI / ML model in the second node. It includes at least one of the following.

[0044] In some embodiments of the Method relating to a third broad aspect of the Disclosure, the first node decides whether or not to connect with the second node based on the aggregation capacity of the first node and information relating to the second type AI / ML model of the second node.

[0045] In some embodiments of the Method relating to a third broad aspect of the present disclosure, the second node is a Type 1 node, and the aggregation request message includes an aggregation request.

[0046] In some embodiments of the method according to a third broad aspect of the present disclosure, the second node may be connected to only one node, and the processor executable instruction further includes a processor executable instruction that, when executed, causes the processor to receive an acknowledgment from the second node indicating whether a connection between the second node and the first node has been successfully established.

[0047] In some embodiments of the methods according to the third broad aspect of this disclosure, the second node is a type 2 node and the first node is a type 1 node.

[0048] In some embodiments of the method according to a third broad aspect of the present disclosure, the second node is a first type 1 node, the first node is a second type 1 node, and the second type 1 node provides the first type AI / ML model of the first node to the second node or receives the first type AI / ML model of the second node from the second node.

[0049] In some embodiments of the Method relating to a third broad aspect of the present disclosure, the second node is a Type 1 node that is related only to one or more other Type 1 nodes.

[0050] Corresponding equipment and devices for carrying out this method are disclosed.

[0051] For example, according to another aspect of the present disclosure, a device is provided that includes a processor and a memory for storing processor-executable instructions, which, when executed, cause the processor to perform a method according to the first, second, or third broad aspects of the present disclosure described above.

[0052] In another aspect of this disclosure, an apparatus is provided comprising one or more units for carrying out any aspect of the methods disclosed herein. The term “unit” is used in a broad sense and can refer to any of the various names, including, for example, module, component, element, means, etc. A unit can be implemented using hardware, software, firmware, or any combination thereof.

[0053] In some aspects of this disclosure, heterogeneous AI / ML functionality can be enabled on various network devices and user devices, and heterogeneous AI / ML model transfer can be supported via the wireless interface of a wireless communication network.

[0054] In some aspects of this disclosure, the topology within a wireless communication network can be centrally configured by network devices (e.g., base stations (BS), transmit and receive points (TRP)) or network systems.

[0055] In some aspects of this disclosure, the topology within a wireless communication network can be autonomously configured by network equipment (e.g., base stations (BS)), transmit and receive points (TRPs) and / or user equipment using discovery procedures as described herein, thereby supporting flexible topology configuration.

[0056] In some aspects of this disclosure, topologies within wireless communication networks configured by the various methods described herein support parallel AI / ML model aggregation, reduce AI / ML model routing latency, and improve the robustness of AI / ML model routing. [Brief explanation of the drawing]

[0057] For the sake of illustrative purposes, refer to the following appended drawings showing exemplary embodiments of the present application.

[0058] [Figure 1] This is a simplified schematic diagram of a communication system as an example.

[0059] [Figure 2] Here is another example of a communication system.

[0060] [Figure 3] Examples of electronic equipment (ED), ground-based transmitting / receiving points (T-TRP), and non-ground transmitting / receiving points (NT-TRP) are shown.

[0061] [Figure 4] This shows an example unit or module within the device.

[0062] [Figure 5] This invention illustrates four EDs that communicate with network devices within a communication system.

[0063] [Figure 6A] An example of a neural network having multiple layers of neurons according to an embodiment of this disclosure is shown.

[0064] [Figure 6B] Examples of neurons that can be used as building blocks for neural networks according to embodiments of this disclosure are shown.

[0065] [Figure 7] This shows the star topology used in conventional associative learning (FL) procedures.

[0066] [Figure 8] This document demonstrates the differences between conventional FL-based AI / ML model training procedures and the AI / ML model learning scheme of this disclosure.

[0067] [Figure 9A] Two different types of nodes used for training AI / ML models according to embodiments of this disclosure are shown. [Figure 9B] Two different types of nodes used for training AI / ML models according to embodiments of this disclosure are shown.

[0068] [Figure 10] An exemplary self-organizing topology according to embodiments of the present disclosure is shown.

[0069] [Figure 11] This disclosure shows a configured topology that supports exemplary AI / ML model transfer according to embodiments of this disclosure.

[0070] [Figure 12] This disclosure provides an example of a procedure for establishing an aggregate connection between an aggregate node and a base node when configuring an AI / ML topology in a wireless communication network, according to an embodiment of this disclosure.

[0071] [Figure 13] This disclosure provides an example of a procedure for establishing an aggregate connection between two aggregate nodes when configuring an AI / ML topology in a wireless communication network, according to an embodiment of this disclosure.

[0072] [Figure 14] This invention provides another example of a procedure for establishing aggregate connections between aggregate nodes and base nodes when configuring an AI / ML topology in a wireless communication network.

[0073] [Figure 15] This embodiment of the disclosure provides another example of the procedure for establishing an aggregate connection between two aggregate nodes when configuring an AI / ML topology in a wireless communication network.

[0074] [Figure 16] An example of a configured topology that supports flexible communication between aggregation nodes according to an embodiment of this disclosure is shown.

[0075] [Figure 17] An example of a configured topology that supports flexible communication between aggregation nodes and base nodes according to an embodiment of this disclosure is shown.

[0076] Similar reference numerals may be used in different diagrams to indicate similar components. [Modes for carrying out the invention]

[0077] In this disclosure, “data collection” refers to the process of collecting data by network nodes, management entities, or user equipment (UE) for the purpose of training artificial intelligence (AI) / machine learning (ML) models, data analysis, and inference.

[0078] In this disclosure, "AI / ML model" refers to a data-driven algorithm that applies AI / ML technology to generate an output set based on an input set.

[0079] In this disclosure, "AI / ML model training" refers to the process of training an AI / ML model by learning input / output relationships in a data-driven manner and obtaining the trained AI / ML model for inference.

[0080] In this disclosure, "AI / ML inference" refers to the process of generating an output set based on an input set using a trained AI / ML model.

[0081] In this disclosure, "online training" refers to a situation in which the machine learning program is not running and is not taking in new information in real time.

[0082] In this disclosure, "offline training" refers to a situation in which a machine learning program is operating in real time on incoming data.

[0083] In this disclosure, "On-UE training" refers to online / offline training conducted in the Unreal Engine (UE).

[0084] In this disclosure, "on-network training" refers to online / offline training conducted over a network.

[0085] In this disclosure, "UE-side (AI / ML) model" refers to an AI / ML model in which inference is performed entirely on the UE.

[0086] In this disclosure, "network-side (AI / ML) model" refers to an AI / ML model in which inference is performed entirely on the network.

[0087] In this disclosure, “model transfer” refers to the delivery of an AI / ML model via a wireless interface, which is either the parameters of a model structure known at the receiving end, or a new model having those parameters. The delivery may include a complete model or a partial model.

[0088] In this disclosure, “model download” refers to the transfer of a model from the network to the UE.

[0089] In this disclosure, “model upload” refers to the transfer of a model from the UE to the network.

[0090] In this disclosure, “model deployment” means delivering a fully developed and tested model runtime image to the target UE / gNodeB (gNB) on which inference will be performed.

[0091] In this disclosure, “Federated learning / federated training” refers to a machine learning technique that trains AI / ML models across multiple distributed edge nodes (e.g., UEs, gNBs) that perform local model training using local data samples. This technique requires the exchange of multiple models but does not require the exchange of local data samples.

[0092] In this disclosure, "model monitoring" refers to the procedure for monitoring the inference performance of an AI / ML model.

[0093] In this disclosure, "model update" means retraining or fine-tuning an AI / ML model through online / offline training in order to improve model inference performance.

[0094] For illustrative purposes, specific examples of embodiments will be described in detail below with accompanying diagrams.

[0095] The embodiments described herein provide sufficient information to carry out the subject matter of the claims and illustrate methods for carrying out such subject matter. A person skilled in the art will understand the concepts of the subject matter of the claims and recognize the applications of these concepts not specifically addressed herein. It should be understood that these concepts and applications are within the scope of the disclosed and appended claims.

[0096] Furthermore, it will be understood that any module, component, or device that performs instructions as disclosed herein includes, or is otherwise accessible from, one or more non-temporary computer / processor-readable instruction storage media for storing information such as computer / processor-readable instructions, data structures, program modules, and / or other data. A non-exclusive list of examples of non-temporary computer / processor-readable storage media includes magnetic cassettes, magnetic tapes, magnetic disk storage devices or other magnetic storage devices, optical disks such as compact disk read-only memory (CD-ROM), digital video discs or digital versatile discs (i.e., DVDs), Blu-ray Discs®, or other optical storage devices, volatile and non-volatile, removable and non-removable media implemented in any way or technique, random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory, or other memory technologies. Such non-temporary computer / processor storage media may be part of a device, or may be accessible or connectable to it. Computer / processor-readable / executable instructions for implementing the applications or modules described herein may be stored or otherwise held in such non-temporary computer / processor-readable storage media.

[0097] Examples of communication systems and devices Referring to Figure 1, a simplified schematic diagram of a communication system is provided as an illustrative example, not limited to the above. The communication system 100 includes a radio access network 120. The radio access network 120 may be a next-generation (e.g., 6G or later) radio access network or a legacy (5G, 4G, 3G, 2G, etc.) radio access network. One or more communication electrical devices (EDs) 110a to 110j (generally referred to as 110) may be interconnected with each other or connected to one or more network nodes (170a, 170b, collectively referred to as 170) within the radio access network 120. The core network 130 may be part of the communication system and may depend on or be independent of the radio access technology used in the communication system 100. The communication system 100 also includes a public switched telephone network (PSTN) 140, the Internet 150, and other networks 160.

[0098] Figure 2 shows an exemplary communication system 100. Generally, the communication system 100 enables multiple wireless or wired elements to communicate data and other content. The purpose of the communication system 100 may be to deliver content such as voice, data, video, and / or text via broadcast, multicast, and unicast, etc. The communication system 100 may operate by sharing resources such as carrier spectral bandwidth among its components. The communication system 100 may include a terrestrial communication system and / or a non-terrestrial communication system. The communication system 100 may provide a wide range of communication services and applications, such as earth monitoring, remote sensing, passive sensing and positioning, navigation and tracking, autonomous delivery and mobility. The communication system 100 may provide high availability and robustness through the joint operation of the terrestrial and non-terrestrial communication systems. For example, integrating a non-terrestrial communication system (or its components) into a terrestrial communication system may result in what is considered a heterogeneous network involving multiple layers. Compared to conventional communication networks, heterogeneous networks can achieve better overall performance through efficient multilink collaboration, more flexible function sharing, and faster physical layer link switching between terrestrial and non-terrestrial networks.

[0099] Terrestrial and non-terrestrial communication systems may be considered subsystems of the communication system. In this example, the communication system 100 includes electronic devices (EDs) 110a-110d (generally referred to as ED110), radio access networks (RANs) 120a-120b, a non-terrestrial communication network 120c, a core network 130, a public switched telephone network (PSTN) 140, the Internet 150, and other networks 160. The RANs 120a-120b include each base station (BS) 170a-170b, collectively referred to as terrestrial transmit and receive points (T-TRPs) 170a-170b. The non-terrestrial communication network 120c includes access nodes 120c, collectively referred to as non-terrestrial transmit and receive points (NT-TRPs) 172.

[0100] Any ED110 may be alternatively or additionally configured to interface with, access, or communicate with other T-TRP170a-170b and NT-TRP172, the Internet 150, the core network 130, the PSTN 140, other networks 160, or any combination thereof. In some examples, ED110a may communicate with T-TRP170a for uplink and / or downlink transmissions via interface 190a. In some examples, ED110a, 110b, and 110d may communicate directly with each other via one or more sidelink radio interfaces 190b. In some examples, ED110D may communicate with NT-TRP172 for uplink and / or downlink transmissions via interface 190c.

[0101] The radio interfaces 190a and 190b may use any suitable radio access technology or similar communication technology. For example, the communication system 100 may implement one or more channel access methods on the radio interfaces 190a and 190b, such as code division multiple access (CDMA), time division multiple access (TDMA), frequency division multiple access (FDMA), orthogonal FDMA (OFDMA), or single-carrier FDMA (SC-FDMA). The radio interfaces 190a and 190b may utilize other higher-dimensional signal spaces, including combinations of orthogonal and / or non-orthogonal dimensions.

[0102] The wireless interface 190c can enable communication between the ED110d and one or more NT-TRP172s via a wireless link or simply a link. In some examples, the link is a dedicated connection for unicast transmission, a connection for broadcast transmission, or a connection between a group of EDs and one or more NT-TRPs for multicast transmission.

[0103] RAN120a and 120b communicate with the core network 130 to provide ED110a, 110b, and 110c with various services, including voice, data, and other services. RAN120a and 120b, and / or the core network 130, may or may not be directly serviced by the core network 130, and may or may not employ the same radio access technology as RAN120a, RAN120b, or both, and may or may not employ the same radio access technology. The core network 130 also functions as a gateway access between (i) RAN120a and 120b, or ED110a, 110b, and 110c, or both, and (ii) other networks (such as PSTN 140, the Internet 150, or other networks 160). Furthermore, some or all of ED110a, 110b, and 110c may include the ability to communicate with different radio networks via different radio links using different radio technologies and / or protocols. Instead of wireless communication (or in addition to it), ED110a, 110b, and 110c may communicate with a service provider or switch (not shown) and the Internet 150 via a wired communication channel. PSTN 140 may include a circuit-switched telephone network for providing plain old telephone service (POTS). The Internet 150 may include a network of computers and subnets (intranets) or both, and may incorporate protocols such as the Internet Protocol (IP), Transmission Control Protocol (TCP), and User Datagram Protocol (UDP). ED110a, 110b, and 110c may be multimode devices capable of operating according to multiple wireless access technologies and incorporating multiple transceivers necessary to support such technologies.

[0104] Figure 3 shows another example of the ED110 and base stations 170a, 170b, and / or 170c. The ED110 is used to connect people, things, machines, etc. The ED110 can be widely used in various scenarios, such as cellular communication, device-to-device (D2D), vehicle-to-vehicle (V2X), peer-to-peer (P2P), machine-to-machine (M2M), machine-to-machine communication (MTC), Internet of Things (IoT), virtual reality (VR), augmented reality (AR), industrial control, autonomous driving, telemedicine, smart grids, smart furniture, smart offices, smart wearables, smart transportation, smart cities, drones, robots, remote sensing, passive sensing, positioning, navigation and tracking, automated delivery and movement, etc.

[0105] Each ED110 represents any end-user device suitable for wireless operation, and may include, in particular, devices such as user equipment / devices (UE), wireless transceiver units (WTRU), mobile stations, fixed or mobile subscriber units, mobile phones, stations (STA), machine-type communication (MTC) equipment, personal digital assistants (PDAs), smartphones, laptops, computers, tablets, wireless sensors, consumer electronic devices, smartbooks, vehicles, automobiles, trucks, buses, trains, or IoT devices, industrial equipment, or devices (communication modules, modems, chips, etc.). Future generations of ED110 may be referred to using other terms. Base stations 170a and 170b are T-TRPs and will hereafter be referred to as T-TRP170. As also shown in Figure 3, the NT-TRP will hereafter be referred to as NT-TRP1702. Each ED110 connected to the T-TRP170 and / or NT-TRP172 may be started dynamically or semi-statically (i.e., established, activated, or enabled), stopped (released, deactivated, or disabled), and / or configured in response to one or more of the connection availability and the need for the connection.

[0106] ED110 includes a transmitter 201 and a receiver 203 coupled to one or more antennas 204. Only one antenna 204 is shown. Alternatively, one, some, or all of the antennas may be a panel. The transmitter 201 and receiver 203 may be integrated, for example, as a transceiver. The transceiver is configured to modulate data or other content for transmission by at least one antenna 204 or a NIC (Network Interface Controller). The transceiver is also configured to demodulate data or other content received by at least one antenna 204. Each transceiver includes any suitable structure for generating signals for wireless or wired transmission and / or processing signals received wirelessly or wired. Each antenna 204 includes any suitable structure for sending and receiving wireless or wired signals.

[0107] The ED110 includes at least one memory 208. The memory 208 stores instructions and data used, generated, or collected by the ED110. For example, the memory 208 may store software instructions or modules configured to implement some or all of the functions and / or embodiments described herein and performed by the processing unit 210. Each memory 208 includes any suitable volatile and / or non-volatile storage and retrieval device. Any suitable type of memory may be used, such as random access memory (RAM), read-only memory (ROM), hard disk, optical disk, subscriber identity module (SIM) card, memory stick, secure digital (SD) memory card, or on-processor cache.

[0108] The ED110 further includes one or more input / output devices (not shown) or interfaces (such as a wired interface to the Internet 150 in Figure 1). The input / output devices 206 enable interaction with the user or other devices in the network. Each input / output device includes any structure suitable for providing information to or receiving information from the user, including network interface communication, such as a speaker, microphone, keypad, keyboard, display, or touchscreen.

[0109] ED110 further includes a processor 210 that performs operations including operations related to preparing transmission for uplink transmission to NT-TRP172 and / or T-TRP170, operations related to processing downlink transmission received from NT-TRP172 and / or T-TRP170, and / or operations related to processing sidelink transmission to and from another ED110. Processing operations related to preparing transmission for uplink transmission may include operations such as coding, modulation, transmit beamforming, and generation of transmit symbols. Processing operations related to processing downlink transmission may include operations such as receive beamforming, demodulation, and decoding of received symbols. Depending on the embodiment, the downlink transmission may be received by a receiver 203 using receive beamforming, and the processor 210 may extract signaling from the downlink transmission (e.g., by detection and / or decoding of the signal). An example of signaling may be a reference signal transmitted by NT-TRP172 and / or T-TRP170. In some embodiments, the processor 276 performs transmit beamforming and / or receive beamforming based on beam direction indications received from the T-TRP 170, such as beam angle information (BAI). In some embodiments, the processor 210 may perform operations related to network access (such as initial access) and / or downlink synchronization, such as detecting the synchronization sequence and decoding and retrieving system information. In some embodiments, the processor 210 may perform channel estimation using reference signals received from, for example, the NT-TRP 172 and / or the T-TRP 170.

[0110] Although not shown, the processor 210 may form part of the transmitter 201 and / or receiver 203. Although not shown, the memory 208 may form part of the processor 210.

[0111] The processing components of the processor 210, and the transmitter 201 and receiver 203, may each be implemented by one or more identical or different processors configured to execute instructions stored in memory (e.g., in memory 208). Alternatively, some or all of the processing components of the processor 210, and the transmitter 201 and receiver 203 may be implemented using dedicated circuits such as a programmed field-programmable gate array (FPGA), a graphical processing unit (GPU), or an application-specific integrated circuit (ASIC).

[0112] In some implementations, the T-TRP170 may be known by other names such as base station, base transceiver station (BTS), radio base station, network node, network equipment, network-side equipment, transmit / receive node, node B, evolved node B (eNodeB or eNB), home eNodeB, next-generation node B (gNB), transmission point (TP), site controller, access point (AP), or radio router, relay station, remote radio head, ground node, ground network equipment, or ground base station, base band unit (BBU), remote radio unit (RRU), active antenna unit (AAU), remote radio head (RRH), central unit (CU), distribute unit (DU), positioning node, etc. The T-TRP170 may also be a macro BS, pico BS, relay node, donor node, etc., or a combination thereof. T-TRP170 may refer to the aforementioned device or the equipment within the aforementioned device (such as a communication module, modem, or chip).

[0113] In some embodiments, the components of T-TRP170 may be distributed. For example, some modules of T-TRP170 may be located away from the facility housing the T-TRP170 antenna and may be coupled to the facility housing the antenna via a communication link (not shown) also known as a fronthaul, such as a common public radio interface (CPRI). Thus, in some embodiments, the term T-TRP170 may refer to a network-side module that performs processing operations such as determining the location of the ED110, resource allocation (scheduling), message generation, and coding / decoding, and is not necessarily part of the facility housing the T-TRP170 antenna. The module may also be coupled to other T-TRPs. In some embodiments, T-TRP170 may actually be multiple T-TRPs working together to serve the ED110, for example, via coordinated multipoint transmission.

[0114] The T-TRP170 includes at least one receiver 254 coupled to one or more antennas 256 and at least one transmitter 252. Only one antenna 256 is shown. Alternatively, one, some, or all of the antennas may be a panel. The transmitter 252 and receiver 254 may be integrated as a transceiver. The T-TRP170 further includes a processor 260 that performs operations including operations related to preparing transmission for downlink transmission to ED110, processing uplink transmission received from ED110, preparing transmission for backhaul transmission to NT-TRP172, and processing transmission received from NT-TRP172 via backhaul. Processing operations related to preparing transmission for downlink or backhaul transmission may include operations such as coding, modulation, precoding (e.g., MIMO precoding), transmit beamforming, and generation of transmission symbols. Processing operations related to processing transmission received uplink or via backhaul may include operations such as receive beamforming, demodulation, and decoding of received symbols. The processor 260 can also perform operations related to network access (such as initial access) and / or downlink synchronization, such as generating the contents of a synchronization signal block (SSB) and generating system information. In some embodiments, the processor 260 also generates beam direction indications, such as BAI, which can be scheduled for transmission by the scheduler 253. The processor 260 performs other network-side processing operations described herein, such as determining the position of the ED 110 and determining the placement of the NT-TRP 172. In some embodiments, the processor 260 can generate signaling to constitute, for example, one or more parameters of the ED 110 and / or one or more parameters of the NT-TRP 172. Any signaling generated by the processor 260 is transmitted by the transmitter 252. Note that the term “signaling” as used herein may alternatively be referred to as control signaling.Dynamic signaling may be transmitted on a control channel, such as a physical downlink control channel (PDCCH), while static or semi-static upper-layer signaling may be included in packets transmitted on a data channel, such as a physical downlink shared channel (PDSCH).

[0115] The scheduler 253 may be coupled to the processor 260. The scheduler 253 may be contained within or operate separately from the T-TRP 170 and may schedule uplink, downlink, and / or backhaul transmissions, including issuing scheduling grants and / or configuring scheduling-free ("configured grant") resources. The T-TRP 170 further includes memory 258 for storing information and data. Memory 258 stores instructions and data used, generated, or collected by the T-TRP 170. For example, memory 258 may store software instructions or modules configured to implement some or all of the functions and / or embodiments described herein and performed by the processor 260.

[0116] Although not shown, the processor 260 may form part of the transmitter 252 and / or receiver 254. Also, although not shown, the processor 260 may implement a scheduler 253. Although not shown, memory 258 may form part of the processor 260.

[0117] The processing components of processor 260, scheduler 253, and transmitter 252 and receiver 254 may each be implemented by one or more identical or different processors configured to execute instructions stored in memory, for example, in memory 258. Alternatively, some or all of the processing components of processor 260, scheduler 210, and transmitter 252 and receiver 254 may be implemented using dedicated circuitry such as FPGA, GPU, or ASIC.

[0118] Although NT-TRP172 is illustrated as a drone for illustrative purposes only, NT-TRP172 can be implemented in any suitable non-terrestrial form. NT-TRP172 may also be known by other names in some implementations, such as a non-terrestrial node, non-terrestrial network equipment, or non-terrestrial base station. NT-TRP172 includes a transmitter 272 and a receiver 274 coupled to one or more antennas 280. Only one antenna 280 is shown; or one, some, or all of the antennas may be a panel. The transmitter 272 and receiver 274 may be integrated as a transceiver. NT-TRP172 further includes a processor 276 that performs operations including preparing transmissions for downlink transmission to ED110, processing uplink transmissions received from ED110, preparing transmissions for backhaul transmission to T-TRP170, and processing transmissions received from T-TRP170 via backhaul. Processing operations related to preparing a transmission for downlink or backhaul transmission may include operations such as coding, modulation, precoding (e.g., MIMO precoding), transmit beamforming, and generation of transmission symbols. Processing operations related to processing a transmission received on the uplink or via backhaul may include operations such as receive beamforming, demodulation, and decoding of received symbols. In some embodiments, the processor 276 performs transmit beamforming and / or receive beamforming based on beam direction information (e.g., BAI) received from the T-TRP 170. In some embodiments, the processor 276 can generate signaling to configure one or more parameters of the ED 110, for example. In some embodiments, the NT-TRP 172 implements physical layer processing but does not implement higher-layer functions such as functions in the medium access control (MAC) or radio link control (RLC) layer. This is just an example, and more generally, the NT-TRP 172 may implement higher-layer functions in addition to physical layer processing.

[0119] The NT-TRP172 further includes a memory 278 for storing information and data. Although not shown, a processor 276 may form part of the transmitter 272 and / or receiver 274. Although not shown, the memory 278 may form part of the processor 276.

[0120] The processor 276 and the processing components of the transmitter 272 and receiver 274 may each be implemented by one or more identical or different processors configured to execute instructions stored in memory, for example, in memory 278. Alternatively, some or all of the processing components of the processor 276 and the transmitter 272 and receiver 274 may be implemented using dedicated circuitry such as a pre-programmed FPGA, GPU, or ASIC. In some embodiments, the NT-TRP 172 may actually be multiple NT-TRPs working together to provide service to the ED 110, for example, via cooperative multipoint transmission.

[0121] Please note that "TRP" as used here may refer to either T-TRP or NT-TRP.

[0122] T-TRP170, NT-TRP172, and / or ED110 may include other components, but these are omitted for clarity.

[0123] One or more steps of the methods of the embodiments provided herein may be performed by a corresponding unit or module, as shown in Figure 4. Figure 4 shows a unit or module in a device such as ED110, T-TRP170, or NT-TRP172. For example, a signal may be transmitted by a transmitting unit or transmitting module. For example, a signal may be transmitted by a transmitting unit or transmitting module. A signal may be received by a receiving unit or receiving module. A signal may be processed by a processing unit or processing module. Other steps may be performed by an artificial intelligence (AI) or machine learning (ML) module. Each unit or module may be implemented using hardware, one or more components or devices that run software, or a combination thereof. For example, one or more units or modules may be integrated circuits such as programmed FPGAs, GPUs, or ASICs. If modules are implemented using software for execution by a processor, for example, they may be retrieved by the processor, as needed, in whole or in part, in single or multiple instances, individually or together for processing, and the modules themselves may contain instructions for further deployment and instantiation.

[0124] Further details regarding ED110, T-TRP170, and NT-TRP172 are known to those skilled in the art; therefore, these details are omitted here.

[0125] Control signaling is described herein in several embodiments. Control signaling may instead be referred to as signaling, or control information, or configuration information, or configuration. In some cases, control signaling may be dynamically indicated at the physical layer, for example, in a control channel. An example of dynamically indicated control signaling is information transmitted by physical layer control signaling, such as downlink control information (DCI). Control signaling may be semi-statically indicated, for example, in RRC signaling or MAC control elements (CEs). Dynamic indication may be at a lower layer, such as physical layer / layer 1 signaling (DCI, etc.), rather than at a higher layer (e.g., other than RRC signaling or MAC CEs). Semi-static indication may be at a semi-static signaling. When used herein, semi-static signaling may refer to non-dynamic signaling, such as upper-layer signaling, RRC signaling, and / or MAC CE. When used herein, dynamic signaling may refer to dynamic signaling, such as physical layer control signaling transmitted at the physical layer, such as DCI.

[0126] A wireless interface generally comprises a number of components and associated parameters that collectively specify how transmissions are sent and / or received over a wireless communication link between multiple communication devices. For example, a wireless interface may include one or more components that define one or more waveforms, frame structures, multiplex access schemes, protocols, coding schemes, and / or modulation schemes for transmitting information (e.g., data) over a wireless communication link. A wireless communication link may support a link between a wireless access network and user equipment (e.g., a "Uu" link), and / or a wireless communication link may support a link between devices, such as between two user devices (e.g., a "side link"), and / or a wireless communication link may support a link between a non-terrestrial (NT) communication network and user equipment (UE). The following are some examples of the above components. The waveform component can specify the shape and format of the signal being transmitted. Waveform options can include orthogonal multiplexing and non-orthogonal multiplexing waveforms. Non-exclusive examples of such waveform options include orthogonal frequency division multiplexing (OFDM), filtered OFDM (f-OFDM), time windowing OFDM, filter bank multicarrier (FBMC), universal filtered multicarrier (UFMC), generalized frequency division multiplexing (GFDM), wavelet packet modulation (WPM), fast Nyquist (FTN) waveforms, and low peak to average power ratio waveforms (low PAPR WF). • A frame structure component can specify the configuration of a frame or a group of frames. A frame structure component can indicate one or more of the following parameters: time, frequency, pilot signature, code, or other parameters of a frame or group of frames. Further details about frame structure are described below. The multiple access scheme component can specify multiple access technology options, including techniques that define how communication devices share a common physical channel. For example, Time Division Multiple Access (TDMA), Frequency Division Multiple Access (FDMA), Code Division Multiple Access (CDMA), Single Carrier Frequency Division Multiple Access (SC-FDMA), Low Density Signature Multicarrier Code Division Multiple Access (LDS-MC-CDMA), Non-Orthogonal Multiple Access (NOMA), Pattern Division Multiple Access (PDMA), Lattice Partition Multiple Access (LPMA), Resource Spread Multiple Access (RSMA), and Sparse Code Multiple Access. Furthermore, multiple access technology options, including SCMA, may include scheduled access and unscheduled access, also known as unauthorized access; non-orthogonal multiplexing access and orthogonal multiplexing access via dedicated channel resources (e.g., not shared among multiple communication devices); competition-based shared channel resources and non-competition-based shared channel resources; and cognitive radio-based access. The Hybrid Automatic Repeat Request (HARQ) protocol component can specify how transmission and / or retransmission occurs. Non-exclusive examples of transmission and / or retransmission mechanism options include specifying the scheduled data pipe size, the signaling mechanism for transmission and / or retransmission, and the retransmission mechanism. The coding and modulation components may specify how the transmitted information is encoded / decoded and modulated / demodulated for transmission / reception. Coding can represent methods of error detection and forward error correction. Non-exclusive examples of coding options include turbo trellis codes, turbo product codes, fountain codes, low-density parity check codes, and polarity codes. Modulation can simply represent a constellation (including, for example, modulation techniques and sequences), or more specifically, various types of advanced modulation methods such as hierarchical modulation and low PAPR modulation.

[0127] In some embodiments, the radio interface may be a "one-size-fits-all concept." For example, components within the radio interface cannot be changed or adapted once the radio interface is defined. In some embodiments, only a limited set of parameters or modes of the radio interface can be configured, such as cyclic prefix (CP) length or multiple input multiple output (MIMO) mode. In some embodiments, the radio interface design may provide a unified or flexible framework for supporting frequency bands below and above 6 GHz (e.g., mmWave) for both licensed and unlicensed access. As an example, the flexibility of a configurable radio interface provided by scalable number theory and symbolic duration can enable optimization of transmission parameters for different spectral bands and different services / devices. As another example, a unified radio interface may be self-contained within a frequency domain, and a frequency domain self-contained design may support more flexible radio access network (RAN) slicing through channel resource sharing between different services in both frequency and time.

[0128] Frame structure A frame structure is a feature of the wireless communication physical layer that defines the time-domain signal transmission structure, enabling, for example, timing reference and timing adjustment of basic time-domain transmission units. Wireless communication between communication devices may occur over time-frequency resources controlled by the frame structure. The frame structure is sometimes called a wireless frame structure.

[0129] Depending on the frame structure and / or the configuration of frames within the frame structure, frequency division duplex (FDD), / or time division duplex (TDD), and / or full duplex (FD) communication may be possible. FDD communication occurs when transmission in different directions (e.g., uplink and downlink) occurs in different frequency bands. TDD communication occurs when transmission in different directions (e.g., uplink and downlink) occurs over different time periods. FD communication occurs when transmission and reception occur using the same time-frequency resources. In other words, the device can transmit and receive simultaneously using the same frequency resources.

[0130] As an example of a frame structure, there is a long-term evolution (LTE) frame structure with the following specifications: Each frame has a duration of 10 milliseconds, each frame has 10 subframes, each subframe has a duration of 1 millisecond, each subframe has 2 slots, each slot has a duration of 0.5 milliseconds, each slot is for the transmission of 7 OFDM symbols (assuming a typical CP), each OFDM symbol has a symbol duration and a specific bandwidth (or partial bandwidth or bandwidth partition) related to the number of subcarriers and subcarrier spacing, the frame structure is based on OFDM waveform parameters such as subcarrier spacing and CP length (CPs have fixed-length or limited-length options), and the switching gap between the uplink and downlink in TDD must be an integer multiple of the OFDM symbol duration.

[0131] Another example of a frame structure is the frame structure of new radio (NR) with the following specifications: Multiple subcarrier intervals are supported, each subcarrier interval corresponding to a specific number theory; the frame structure is number theory dependent, but in all cases the frame length is set to 10ms and consists of 10 subframes, each 1ms long; slots are defined as 14 OFDM symbols, and slot lengths are number theory dependent. For example, the NR frame structure with a normal CP 15kHz subcarrier interval ("Number Theory 1") is different from the NR frame structure with a normal CP 30kHz subcarrier interval ("Number Theory 2"). The slot length for the 15kHz subcarrier interval is 1ms, and the slot length for the 30kHz subcarrier interval is 0.5ms. The NR frame structure is more flexible than the LTE frame structure.

[0132] Another example of a frame structure is a flexible frame structure for use in 6G networks and beyond. In a flexible frame structure, a symbol block may be defined as the minimum duration that can be scheduled within the flexible frame structure. A symbol block can be a transmission unit having optional redundant parts (e.g., CP parts) and informational parts (e.g., data parts). OFDM symbols are an example of a symbol block. Symbol blocks may be alternatively called symbols. Embodiments of a flexible frame structure include different parameters that may be configurable, such as frame length, subframe length, and symbol block length. A non-exhaustive list of possible configurable parameters in some embodiments of a flexible frame structure includes: (1) Frame: The frame length does not need to be limited to 10 milliseconds, and the frame length is configurable and can change over time. In some embodiments, each frame includes one or more downlink synchronous channels and / or one or more downlink broadcast channels, and each synchronous channel and / or broadcast channel may be transmitted in different directions by different beamforming. The frame length is of multiple possible values ​​and can be configured based on the application scenario. For example, an autonomous vehicle may require relatively fast initial access, in which case the frame length may be set to 5 milliseconds for the autonomous vehicle application. As another example, a smart meter in a home may not require fast initial access, in which case the frame length may be set to 20 ms for the smart meter application. (2) Subframe duration: Depending on the implementation, subframes may or may not be defined in a flexible frame structure. For example, a frame may be defined to include slots but not subframes. In frames where subframes are defined, the duration of the subframe may be configurable, for example, in the case of time domain alignment. For example, a subframe may be configured to have a length of 0.1 milliseconds, 0.2 milliseconds, 0.5 milliseconds, 1 millisecond, 2 milliseconds, or 5 milliseconds. In some embodiments, if a subframe is not required in a particular scenario, the length of the subframe may be defined as the same as the length of the frame, or it may not be defined at all. (3) Slot Configuration: Depending on the implementation, slots may or may not be defined in a flexible frame structure. In frames in which slots are defined, the slot definition (e.g., in terms of duration and / or number of symbol blocks) may be configurable. In one embodiment, the slot configuration is common to all UEs or groups of UEs. In this case, the slot configuration information can be transmitted to the UEs on a broadcast channel or a common control channel. In other embodiments, the slot configuration may be UE-specific, in which case the slot configuration information can be transmitted on a UE-specific control channel. In some embodiments, the slot configuration signaling can be transmitted together with the frame configuration signaling and / or the subframe configuration signaling. In other embodiments, the slot configuration can be transmitted independently of the frame configuration signaling and / or the subframe configuration signaling. Generally, the slot configuration may be system-common, base station-common, UE group-common, or UE-specific. (4) Subcarrier spacing (SCS): SCS is a parameter of scalable number theory and can be in the range of 15 kHz to 480 kHz. SCS may vary with the spectral frequency and / or maximum UE velocity to minimize the effects of Doppler shift and phase noise. In some examples, there may be separate transmit and receive frames, and the SCS of symbols in the receive frame structure may be constructed independently of the SCS of symbols in the transmit frame structure. The SCS in the receive frame may be different from the SCS in the transmit frame. In some examples, the SCS of each transmit frame may be half the SCS of each receive frame. If the SCS between the receive and transmit frames is different, for example, if a more flexible symbol period is implemented by using the inverse discrete Fourier transform (IDFT) instead of the fast Fourier transform (FFT), the difference does not necessarily have to scale in multiples of 2. Additional examples of frame structures can be used with different SCSs. (5) Flexible transmission duration of basic transmission units: A basic transmission unit may be a symbol block (alternatively called a symbol) that generally includes a redundant portion (called CP) and an information (e.g., data) portion, although in some embodiments the CP may be omitted from the symbol block. The CP length may be flexible and configurable. The CP length may be fixed within a frame or flexible within a frame, and the CP length may, in some cases, change dynamically per frame, or per frame group, or per subframe, or per slot, or per scheduling. The information (e.g., data) portion may be flexible and configurable. Another possible parameter relating to a symbol block that can be defined is the ratio of the CP duration to the information (e.g., data) duration. In some embodiments the symbol block length may be adjusted according to channel conditions (e.g., multipath delay, Doppler), and / or delay requirements, and / or available duration. As another example, the symbol block length may be adjusted to fit the available duration within a frame. (6) Flexible switching gap: A frame may include both a downlink portion for downlink transmission from the base station and an uplink portion for uplink transmission from the UE. A gap called a switching gap may exist between each uplink and downlink portion. The switching gap length (duration) may be configurable. The switching gap duration may be fixed within a frame or flexible within a frame, and the switching gap duration may, in some cases, change dynamically per frame, per frame group, per subframe, per slot, or per scheduling.

[0133] Cell / Carrier / Bandwidth Part (BWP) / Occupied Bandwidth Equipment such as base stations can provide coverage to the entire cell. Radio communication with equipment can occur over one or more carrier frequencies. Carrier frequencies are called carriers. Carriers are sometimes called component carriers (CCs). Carriers can be characterized by their bandwidth and reference frequency, for example, the carrier's center or lowest or highest frequency. Carriers may be on licensed or unlicensed spectrum. Radio communication with equipment can also occur over one or more bandwidth parts (BWPs). For example, a carrier may have one or more BWPs. More generally, radio communication with equipment can occur over spectrum. Spectrum may contain one or more carriers and / or one or more BWPs.

[0134] A cell may include one or more downlink resources and optionally one or more uplink resources. Alternatively, a cell may include one or more uplink resources and optionally one or more downlink resources. Alternatively, a cell may include both one or more downlink resources and one or more uplink resources. For example, a cell may include only one downlink carrier / BWP, or only one uplink carrier / BWP, or multiple downlink carriers / BWPs, or multiple uplink carriers / BWPs, or one downlink carrier / BWP and one uplink carrier / BWP, or one downlink carrier / BWP and multiple uplink carriers / BWPs, or multiple downlink carriers / BWPs and one uplink carrier / BWP, or multiple downlink carriers / BWPs and multiple uplink carriers / BWPs. In some embodiments, a cell may, instead or additionally, include one or more sidelink resources, including sidelink transmit and receive resources.

[0135] A BWP is a set of consecutive or non-consecutive frequency subcarriers on a carrier, or a set of consecutive or non-consecutive frequency subcarriers on multiple carriers, and these can have one or more carriers.

[0136] In some embodiments, a carrier may have one or more BWPs, for example, a carrier with a bandwidth of 20 MHz may contain one BWP, or a carrier with a bandwidth of 80 MHz may contain two adjacent consecutive BWPs, and so on. In other embodiments, a BWP may have one or more carriers, for example, a BWP may have a bandwidth of 40 MHz and contain two adjacent consecutive carriers, each having a bandwidth of 20 MHz. In some embodiments, a BWP may contain a non-contiguous spectral resource comprising a plurality of non-contiguous carriers, the first of which may be in the mmW band, the second in the low band (e.g., the 2 GHz band), the third in the THz band (if present), and the fourth in the visible light band (if present). Resources within one carrier belonging to a BWP may be consecutive or non-contiguous. In some embodiments, a BWP has non-contiguous spectral resources on one carrier.

[0137] Wireless communication may occur via occupied bandwidth. Occupied bandwidth may be defined as the width of a frequency band such that it is lower than the lower limit of the frequency and higher than the upper limit of the frequency, and the average power emitted is equal to a specified percentage β / 2 of the total average transmitted power, for example, β / 2 being 0.5%.

[0138] The carrier, BWP, or occupied bandwidth may be dynamically defined by network equipment (e.g., base stations) in physical layer control signaling such as Downlink Control Information (DCI), or semi-statically in radio resource control (RRC) signaling or medium access control (MAC) layers, or predefined based on the application scenario, or determined by the UE as a function of other parameters known to the UE, or fixed by a standard, for example.

[0139] Artificial Intelligence (AI) and / or Machine Learning (ML) The number of new devices in future wireless networks is expected to increase exponentially, and the functionality of these devices is expected to become increasingly diverse. Furthermore, many new applications / use cases are anticipated to emerge, each with more diverse quality-of-service requirements than those of 5G. These could lead to new key performance indicators (KPIs) for the highly challenging future wireless networks (e.g., 6G networks). AI technologies, such as ML (deep learning, etc.), are being introduced into telecommunications applications with the aim of improving system performance and efficiency.

[0140] Furthermore, antenna and bandwidth capabilities continue to advance, enabling more and / or better communication over wireless links. In addition, progress continues in the areas of computer architecture and computing power, including the introduction of general-purpose graphics processing units (GP-GPUs). Future generations of communication equipment may have higher computing and / or communication capabilities than previous generations, thereby enabling the adoption of AI for implementing wireless interface components. Future generations of networks may also have access to more accurate and / or new information (compared to previous networks) that could form the basis for input to AI models. Examples include the physical speed / velocity of the equipment, the equipment's link budget, the equipment's channel conditions, the capabilities and / or supported service types of one or more equipment, sensing information, and / or positioning information. To acquire sensing information, a TRP can transmit a signal to a target object (e.g., a suspicious UE), and based on the reflection of the signal, the TRP or another network device can calculate the angle (for beamforming of the equipment), the distance of the equipment from the TRP, and / or Doppler shift information. Positioning information, sometimes referred to as localization, can be obtained in various ways, including positioning reports from the UE (such as reports of the UE's GPS coordinates), the use of positioning reference signals (PRS), the use of the aforementioned sensing, and tracking and / or predicting the device's position.

[0141] AI technology (including ML technology) can be applied to communications including AI-based communications in the physical layer and / or AI-based communications in the MAC layer. In the physical layer, AI communications can be applied to optimize component design and / or improve algorithmic performance. For example, AI can be applied in relation to the implementation of channel coding, channel modeling, channel estimation, channel decoding, modulation, demodulation, MIMO, waveforms, multiple access, optimization and updating of physical layer element parameters, beamforming, tracking, sensing, and / or positioning. In the MAC layer, AI communications can be applied to leverage AI capabilities to learn, predict, and / or make decisions to solve complex optimization problems with the best possible strategies and / or optimal solutions. For example, it can optimize functions in the MAC layer. For example, AI can be applied to implement intelligent TRP management, intelligent beam management, intelligent channel resource allocation, intelligent power control, intelligent spectrum utilization, intelligent MCS, intelligent HARQ strategy, and / or intelligent transmit / receive mode adaptation.

[0142] In some embodiments, the AI ​​architecture may include multiple nodes. These nodes can be organized in either a centralized or distributed configuration, and both can be located in an access network, core network, edge computing system, or third-party network. Centralized training and computing architectures are sometimes limited by significant communication overhead and strict user data privacy. Distributed training and computing architectures may include several frameworks, such as distributed machine learning and federative learning. In some embodiments, the AI ​​architecture may include an intelligent controller that can run as a single agent or multiple agents based on joint or individual optimization. It is desirable that the new protocols and signaling mechanisms allow for the personalization of corresponding interface links with customized parameters to meet specific requirements, while minimizing signaling overhead through personalized AI techniques and maximizing the overall spectral efficiency of the system.

[0143] In some embodiments of this specification, novel protocols and signaling mechanisms are provided to operate within different operating modes for AI training, including between training modes and normal operating modes, and to switch between them, and to provide measurement and feedback to accommodate different measurements and information that may need to be fed back depending on the implementation.

[0144] AI training Referring again to Figures 1 and 2, embodiments of the present disclosure can be used to implement AI training involving two or more communication devices within a communication system 100. For example, Figure 5 shows four EDs communicating with a network device 452 within the communication system 100 according to one embodiment. The four EDs are each shown as different UEs, hereafter referred to as UEs 402, 404, 406, and 408. However, EDs do not necessarily have to be UEs.

[0145] The network device 452 is part of a network (e.g., a wireless access network 120). Depending on the implementation, the network device 452 may be located in an access network, a core network, an edge computing system, or a third-party network. The network device 452 may be (or part of) a T-TRP or a server. In one example, the network device 452 may be (or implemented in) a T-TRP 170 or an NT-TRP 172. In another example, the network device 452 may be a T-TRP control unit and / or an NT-TRP control unit capable of managing the T-TRP 170 or NT-TRP 172. In some embodiments, the components of the network device 452 may be distributed. UEs 402, 404, 406, and 408 may communicate directly with the network device 452, for example, if the network device 452 is part of a T-TRP that provides services to UEs 402, 404, 406, and 408. Alternatively, UE402, 404, 406, and 408 may communicate with network device 452 via one or more intermediate components, such as T-TRP and / or NT-TRP. For example, network device 452 may transmit and / or receive information (e.g., control signaling, data, training sequences, etc.) to and from one or more of UE402, 404, 406, and 408 via backhaul links and wireless channels interposed between network device 452 and UE402, 404, 406, and 408.

[0146] Each UE402, 404, 406, and 408 includes, as described above, its respective processor 210, memory 208, transmitter 201, receiver 203, and one or more antennas 204 (or alternatively, a panel). Only the processor 210, memory 208, transmitter 201, receiver 203, and antenna 204 of UE402 are illustrated for simplification, but the other UE404, 406, and 408 also include the same respective components.

[0147] For each UE402, 404, 406, and 408, the communication link between that UE and each TRP in the network is a radio interface. A radio interface generally includes a number of components and associated parameters that collectively specify how transmissions are sent and / or received over a radio medium.

[0148] The UE's processor 210 in Figure 5 implements one or more radio interface components on the UE side. The radio interface components configure and / or implement transmission and / or reception via the radio interface. Examples of radio interface components are described herein. Radio interface components may reside in the physical layer, for example, a channel encoder (or decoder) implementing the coding component of the radio interface for the UE, and / or a modulator (or demodulator) implementing the modulation component of the radio interface for the UE, and / or a waveform generator implementing the waveform component of the radio interface for the UE. Radio interface components may reside in or be part of a higher layer, such as the MAC layer, for example, a module implementing channel prediction / tracking, and / or a module implementing a retransmission protocol (e.g., implementing the HARQ protocol component of the UE's radio interface). The processor 210 also directly performs (or controls the UE to perform) the UE-side operations described herein.

[0149] The network device 452 includes a processor 454, memory 456, and input / output device 458. The processor 454 performs or instructs other network devices (e.g., T-TRP) to perform one or more radio interface components on the network side. The radio interface components may be performed differently on the network side for one UE compared to another UE. The processor 454 directly performs (or controls network components to perform) the network-side operations described herein.

[0150] The processor 454 may be implemented by one or more identical or different processors configured to execute instructions stored in memory (e.g., in memory 456). Alternatively, part or all of the processor 454 may be implemented using dedicated circuitry such as a pre-programmed FPGA, GPU, or ASIC. Memory 456 can be implemented by volatile and / or non-volatile storage. Any suitable type of memory can be used, such as RAM, ROM, hard disk, optical disk, or on-processor cache.

[0151] The input / output device 458 enables interaction with other devices by receiving (inputting) and transmitting (outputting) information. In some embodiments, the input / output device 458 may be implemented by a transmitter and / or receiver (or transceiver), and / or one or more interfaces (such as a wired interface to an internal network or the internet). In some implementations, the input / output device 458 may be implemented by a network interface, which may be implemented as a network interface card (NIC), and / or a computer port (e.g., a physical outlet to which a plug or cable is connected), and / or a network socket, depending on the implementation.

[0152] The network device 452 and UE402 are capable of performing one or more AI-enabled processes. In particular, in the embodiment shown in Figure 5, the network device 452 and UE402 include ML modules 410 and 460, respectively. ML module 410 is implemented by the processor 210 of UE402, and ML module 460 is implemented by the processor 454 of network device 452. Therefore, in Figure 5, ML module 410 is shown as being within the processor 210, and ML module 460 is shown as being with the processor 454. ML modules 410 and 460 execute one or more AI / ML algorithms to perform one or more AI-enabled processes, such as AI-enabled link adaptation to optimize the communication link between the network and UE402.

[0153] ML modules 410 and 460 can be implemented using AI models. The term AI model can refer to a computer algorithm configured to accept defined input data and output defined inference data, the algorithm's parameters (e.g., weights) can be updated and optimized through training (e.g., by using a training dataset or by using actually collected data). AI models can be implemented using one or more neural networks (e.g., deep neural networks (DNNs), recurrent neural networks (RNNs), convolutional neural networks (CNNs), and combinations thereof), and using various neural network architectures (e.g., autoencoders, generative adversarial networks, etc.). Various techniques can be used to train AI models in order to update and optimize their parameters. For example, backpropagation is a common technique for training DNNs, where a loss function is calculated between the inference data generated by the DNN and a target output from something (e.g., ground truth data). The gradient of the loss function is calculated with respect to the DNN parameters, and the calculated gradient is used to update the parameters with the aim of minimizing the loss function (e.g., by using a gradient descent algorithm).

[0154] In some embodiments, an AI model encompasses a neural network used in machine learning. A neural network consists of multiple computational units (also called neurons) arranged in one or more layers. The process of receiving input in an input layer and producing output in an output layer is sometimes called forward propagation. In forward propagation, each layer receives input (inputs can have the appropriate data format, such as vectors, matrices, or multidimensional arrays) and performs computations to produce an output (the output may have a different dimension than the input). The computations performed by a layer typically involve applying the input to a set of weights (e.g., multiplication). With the exception of the first layer of the neural network (i.e., the input layer), the input to each layer is the output of the previous layer. A neural network may include one or more layers between the first layer (i.e., the input layer) and the last layer (i.e., the output layer), called internal layers or hidden layers. For example, Figure 6A shows an example of neural network 600, which includes an input layer, an output layer, and two hidden layers. In this example, we can see that the output of each of the three neurons in the input layer of neural network 600 is contained in the input vectors to each of the three neurons in the first hidden layer. Similarly, the output of each of the three neurons in the first hidden layer is included in the input vector to each of the three neurons in the second hidden layer, and the output of each of the three neurons in the second hidden layer is included in the input vector to each of the two neurons in the output layer. As mentioned above, the basic computational unit in a neural network is the neuron, as shown in 650 in Figure 6A. Figure 6B shows an example of neuron 650 that can be used as a building block for neural network 600. As shown in Figure 6B, in this example, neuron 650 takes a vector as input and performs an inner product with the weight association vector. The final output z of the neuron is the result of the activation function f() of the dot product. Various neural networks can be designed with various architectures (e.g., various numbers of layers, with each layer performing various functions).

[0155] A neural network is trained to optimize its parameters (e.g., weights). This optimization is performed in an automated manner and is sometimes called machine learning. Training a neural network involves propagating input data samples forward to generate output values ​​(e.g., called predicted or estimated output values) and comparing the generated output values ​​to known or desired target values ​​(e.g., ground truth values). A loss function is defined to quantitatively represent the difference between the generated output values ​​and the target values. The goal of training a neural network is to minimize the loss function. Backpropagation is an algorithm for training a neural network. Backpropagation is used to adjust (also called updating) the values ​​of parameters (e.g., weights) within the neural network so that the calculated loss function becomes smaller. In backpropagation, the gradient of the loss function with respect to the parameters to be optimized is calculated, and the parameters are updated to decrease the loss function using a gradient algorithm (e.g., gradient descent). Backpropagation is performed iteratively over a large number of iterations until the loss function converges or is minimized. After the training conditions are met (for example, the loss function converges, or a predefined number of training iterations are performed), the neural network is considered trained. The trained neural network may be unfolded (or performed) to generate output data inferred from the input data. In some embodiments, training of the neural network may continue even after the neural network has been unfolded, so that the parameters of the neural network are repeatedly updated with the latest training data.

[0156] Referring again to Figure 5, in some embodiments, the UE402 and the network device 452 may exchange information for training purposes. The information exchanged between the UE402 and the network device 452 may be implementation-specific and may not have a human-understandable meaning (for example, it may be intermediate data generated during the execution of an ML algorithm). Furthermore, or instead, the information exchanged may not be predefined by a standard. For example, bits may be exchanged, but the bits may not be associated with a predefined meaning. In some embodiments, the network device 452 may provide or instruct the UE402 to one or more parameters used in the ML module 410 implemented in the UE402. As an example, the network device 452 may transmit or instruct updated neural network weights implemented in the neural network executed by the ML module 410 on the UE side in an attempt to optimize one or more aspects of modulation and / or coding used for communication between the UE402 and T-TRP or NT-TRP.

[0157] In some embodiments, UE402 may implement the AI ​​itself, for example, by performing learning, but in other embodiments, UE402 does not have to perform learning itself and can work in conjunction with the network-side AI implementation by assisting other devices (such as network devices or other AI-enabled UEs) in training the AI ​​model (such as a neural network or other ML algorithm) by receiving the configuration of the AI ​​model (such as a neural network or other ML algorithm) implemented by the ML module 410 from the network and / or by providing requested measurement results or observations. For example, in some embodiments, UE402 itself does not have to implement learning or training, but UE402 may receive trained configuration information of the ML model determined by the network device 452 and run the model.

[0158] The example in Figure 5 assumes network-side AI / ML capabilities, but the network itself may not perform training / learning; instead, the UE may perform the learning / training itself, possibly using a dedicated training signal transmitted from the network. In other embodiments, end-to-end (E2E) learning may be implemented by the UE and network device 452.

[0159] By using AI, for example, by implementing an AI model as described above, various processes such as link adaptation can be made AI-enabled. To realize the AI-enabled processes according to the embodiments of this disclosure, several examples of possible AI / ML training processes and wireless information exchange procedures between devices during the training phase are described below.

[0160] Referring again to Figure 5, in the case of federated learning (FL), the network device 452 can initialize the global AI / ML model implemented by the ML module 460, sample a group of UEs such as the four UEs 402, 404, 406, and 408 shown in Figure 5, and broadcast the parameters of the global AI / ML model to the UEs. Then, each of UEs 402, 404, 406, and 408 can initialize its own local AI / ML model using the parameters of the global AI / ML model and update (train) its own local AI / ML model using its own data. Then, each of UEs 402, 404, 406, and 408 can report the updated parameters of their local AI / ML model to the network device 452. The network device 452 can then aggregate the updated parameters reported from UEs 402, 404, 406, and 408 and update the global AI / ML model. The procedure described above is one iteration of the FL-based AI / ML model training procedure. The network devices and UE402, 404, 406, and 408 typically perform multiple iterations until the AI / ML model converges sufficiently to meet one or more training objectives / criteria and the AI / ML model is complete.

[0161] Figure 7 shows an example of a star topology used in conventional FL procedures. In the star topology shown in Figure 7, there is a master node / device 700 (e.g., server, TRP, BS) and several client nodes / devices 701-708. The master node 700 initializes the master AI / ML model, samples a group of user client nodes 701-708, and distributes the master model (MM) to the client nodes 701-708. Next, the client nodes 701-708 initialize local AI / ML models M1-M8 using the master AI / ML model and update (train) the local AI / ML models M1-M8 using their own data. Next, the client nodes 701-708 report the updated local AI / ML models to the master node 700. The master node 700 aggregates the updated AI / ML models M1-M8 reported from the client nodes 701-708 and updates the master AI / ML model.

[0162] Traditional FL-based AI / ML model training procedures have several limitations. In a star topology used in FL-based AI / ML model training procedures, the master node 700 needs to collect a large amount of training data (e.g., gradients of client node updates) from each client node 701-708. Traditional FL-based AI training procedures require that the AI / ML model structures (e.g., the master AI / ML model and local AI / ML models) be identical. However, data heterogeneity can be a problem in traditional FL-based AI training procedures because different client nodes may support different AI / ML model structures.

[0163] Aspects of this disclosure provide solutions to overcome the aforementioned limitations, for example, in certain embodiments, methods and apparatus for configuring a topology for artificial intelligence or machine learning (AI / ML) in a wireless communication network. The configured topology can support heterogeneous AI / ML model transmission over the wireless interface of the wireless communication network. The methods and apparatus described herein can be implemented and deployed taking into account the computing power of each network node and the potential scale-in / scale-out for AI / ML model learning and inference.

[0164] The AI / ML model training scheme presented in this disclosure is distinct from conventional FL-based AI / ML model training schemes. Figure 8 illustrates the difference between a conventional FL-based AI / ML model training procedure and the AI / ML model training scheme of this disclosure.

[0165] As shown in Figure 8, in the FL-based AI / ML model training scheme, the master model 810 does not move. Instead, a large training dataset 815 (e.g., gradients of client node updates) may be transferred from client nodes 811a to a master node / device 811 that updates the master AI / ML model 810 based on the collected data. In other words, in the FL-based AI / ML model training scheme, the training dataset 815 follows the AI / ML model 810. In contrast, in the AI / ML model learning scheme of this disclosure, the AI / ML model 820 follows the training dataset that remains at each network edge or node / device 821. In other words, the AI / ML model 820 (e.g., a deep neural model) is routed around the network edges or nodes / devices 821, and the deep learning algorithm is executed locally at each node / device 821.

[0166] In FL-based AI / ML model training, there is only one master node / device 811 that collects and aggregates AI / ML models. Only this master node / device 811 can be called an aggregation node. On the other hand, in the AI / ML model learning scheme of this disclosure, any node in the topology can collect and aggregate AI / ML models, so there may be multiple aggregation nodes (e.g., node 821). In practice, unlike FL-based AI / ML model training where nodes 811 and 811a are arranged in a star topology, multiple nodes / devices 821 are arranged in a self-organizing topology as shown in Figure 8.

[0167] A further difference between FL-based AI / ML model training and the AI / ML model learning scheme of this disclosure lies in the type of data / information exchanged between nodes. In FL-based AI / ML model training, a large training dataset 815 (e.g., gradients of client node updates) may be exchanged between client nodes 811a and master nodes / devices 811. In the AI / ML model learning scheme of this disclosure, AI / ML models 820 are exchanged between nodes 821.

[0168] In some embodiments, any node in the network may be a node that collects one or more AI / ML models and aggregates the collected AI / ML models to generate a new or updated AI / ML model (e.g., node 821 in Figure 8). Such a node may be a Type 1 node. A Type 1 node is a node configured to collect multiple AI / ML models and aggregate the collected AI / ML models in order to obtain a new or updated AI / ML model (e.g., a first type AI / ML model). A Type 1 node may also be referred to as an aggregated AI / ML node or aggregate node hereafter or elsewhere in this disclosure. Each aggregated node (or Type 1 node) may be communicated and / or operablely connected or associated with another aggregated node (or Type 1 node). Each aggregated node (or Type 1 node) may be communicated and / or operablely connected or associated with one or more other nodes. One or more other nodes may be Type 2 nodes. A Type 2 node is a node configured to train an AI / ML model (e.g., a Type 2 AI / ML model) on a set of training data (such as local data) without performing aggregation operations. A Type 2 node may also be referred to as a base AI / ML node or base node below or elsewhere in this disclosure. In some embodiments, some aggregation nodes (or Type 1 nodes) may not be associated with base nodes (or Type 2 nodes) (i.e., they may be associated only with other aggregation nodes (or Type 1 nodes)).

[0169] In some embodiments, AI / ML models may be transferred between nodes (e.g., BS, TRP, UE) according to a self-organizing topology. In this regard, training gradients are not transmitted or shared between nodes (such as aggregation nodes or base nodes) at the network level. This is because training gradients are large and controllable in the computation process is difficult.

[0170] In some embodiments, there may be at least two types of AI / ML models. A first type AI / ML model is an AI / ML model generated by an aggregation node based on one or more AI / ML models collected by the aggregation node, for example. A first type AI / ML model may be referred to as a common AI / ML model below or elsewhere in this disclosure. A second type AI / ML model is an AI / ML model trained by a base AI / ML node without aggregation operations. A second type AI / ML model may be referred to as a local AI / ML model below or elsewhere in this disclosure. Note that common AI / ML models and local AI / ML models may have different names. Those skilled in the art will readily understand that an AI / ML model should be considered a common AI / ML model or a local AI / ML model if it has essentially similar characteristics to a common AI / ML model or a local AI / ML model, or if it is generated / trained / updated in essentially similar ways to a common AI / ML model or a local AI / ML model as described herein, even if the AI / ML model is referred to in different ways.

[0171] Aspects of this disclosure provide methods for configuring or discovering topologies for artificial intelligence or machine learning (AI / ML). In some embodiments, there are two modes for topology configuration or topology discovery in a wireless communication network. In the first mode, a network device (e.g., a BS or TRP) configures a topology that supports heterogeneous AI / ML model transmission over the wireless interface of the wireless communication network. Such a configuration may be called a centralized configuration. In the second mode, a network device (such as a BS or TRP) and / or user device can autonomously configure topologies within the wireless communication network. In some embodiments, an aggregation node declares its aggregation capability to other nodes. To this end, an aggregation node may send a discovery message containing aggregation information to be used by other nodes for discovery of the aggregation node. In some embodiments, a node sends an aggregation request message (e.g., an aggregation request) to other nodes. The node sending the aggregation request message may be an aggregation node or a base node. The aggregation request message may contain an aggregation request or requested aggregation information that other nodes may be interested in discovering.

[0172] In some embodiments, the topology can be configured to process the aggregation of AI / ML models in parallel.

[0173] In a self-organizing topology, there are multiple nodes that perform communication and computing functions. These nodes can receive at least one computing model (AI / ML model) and transmit at least one computing model (AI / ML model). The received and transmitted models are not identical entities, but they can have the same or different data, information, and / or neural network structures. The received and transmitted AI / ML models can include one or more parameters related to multiple parameters, such as graph models, parameter models, model tables, model algorithms, and databases.

[0174] Nodes within a network can be classified into two node types with respect to the training of AI / ML models. Figures 9A and 9B illustrate two different types of nodes used for training AI / ML models according to embodiments of this disclosure. Figure 9A shows one type of node, which is a Type 2 node, or base AI / ML node, or base node, and Figure 9B shows another type of node, which is a Type 1 node, or aggregated AI / ML node, or aggregated node. Type 2 / base nodes may also be called local AI / ML nodes or local nodes.

[0175] A base AI / ML node can receive one or more common AI / ML models from other nodes (such as aggregation nodes) and train its own (customized) AI / ML model. The customized AI / ML model may be called a local AI / ML model. The base node trains its local AI / ML model using its own AI / ML model training algorithm, with the help of the (current) common AI / ML models received from one or more aggregation nodes (e.g., information about distillation and / or expansion). In some embodiments, the neural network (NN) structure of the local AI / ML model may be the same as at least some of the NN structures of the received common AI / ML model. In some embodiments, the NN structure of the local AI / ML model may be different from all of the NN structures of the received common AI / ML model. Once the training of the local AI / ML model is complete, the base node sends the local AI / ML model to one or more other nodes (such as aggregation nodes).

[0176] An aggregation node may receive one or more AI / ML models from other nodes in the network. The received AI / ML models include one or more local AI / ML models from the relevant base node and / or one or more common AI / ML models from other aggregation nodes. The aggregation AI / ML node shown in Figure 9B receives local AI / ML models, but in some embodiments, the aggregation node does not need to receive local AI / ML models from the base node (i.e., it only receives common AI / ML models from other aggregation nodes). In some embodiments, some or all of the received AI / ML models have the same NN structure. In some embodiments, all of the received AI / ML models have different NN structures. After collecting AI / ML models from other nodes, the aggregation node aggregates the collected AI / ML models (common AI / ML models, local AI / ML models) to generate a new or updated common AI / ML model. The aggregation node then sends the new / updated AI / ML model to one or more other nodes.

[0177] In some embodiments, the common AI / ML model is a predefined or preconfigured AI / ML model.

[0178] As described above, the common AI / ML model and local AI / ML models involved in the AI / ML model learning process have different NN structures. For example, if an aggregation node receives one common AI / ML model and three local AI / ML models, the received AI / ML models may all have different NN structures. The common AI / ML model may be a 6-layer deep neural network (DNN) model, the first local AI / ML model may be a 4-layer DNN model, the second local AI / ML model may be an 8-layer DNN model, and the third local AI / ML model may be a convolutional neural network (CNN). The NN structure can be considered as the model structure of the AI / ML model.

[0179] The self-organizing topology of the present disclosure includes connections between aggregating nodes and one or more basic nodes and / or connections between multiple aggregating nodes.

[0180] Any network device / apparatus can operate as an aggregating node within the network. The aggregating node may be communicatively and / or operably connected to one or more basic nodes. Such a connection may indicate that the aggregating node collects local AI / ML models from the associated basic nodes. However, note that in some embodiments, some aggregating nodes may not be communicatively and / or operably connected to any basic node. Such aggregating nodes may not collect local AI / ML models and may simply receive a common AI / ML model from other aggregating nodes.

[0181] Note that any aggregating node and / or any basic node can be, for example, a UE, a repeater, a base station (BS), a transmission and reception point (TRP), an edge device, an edge computing system, or a network system.

[0182] FIG. 10 shows an exemplary self-organizing topology 1000 according to an embodiment of the present disclosure. Topology 1000 includes connections between aggregating nodes and connections between aggregating nodes and basic nodes. Regarding the connections between aggregating nodes, each of aggregating nodes 901 to 908 is communicatively and operably connected to its adjacent aggregating nodes. For example, as shown in FIG. 10, aggregating node 901 is communicatively and operably connected to aggregating nodes 902 and 908. In this example, this connection indicates that aggregating node 901 j may receive the common AI / ML model M and transmit its own common AI / ML model M1 to aggregating node 902.

[0183] Regarding the connections between aggregation nodes, each of aggregation nodes 901 to 908 is connected to the base node in a communicative and operational manner in this example. Specifically, aggregation node 901 is connected to base nodes 901a and 901b in a communicative and operational manner. These connections allow aggregation node 901 to receive the common AI / ML model M from node 908. jThe data is sent to the base nodes 901a and 901b, indicating that local AI / ML models may be collected from the base nodes 901a and / or 901b. Similarly, the aggregation node 902 is connected to the base nodes 902a and 902b in a communicative and operational manner. These connections indicate that the aggregation node 902 may send the common AI / ML model M1 received from node 901 to the base nodes 902a and 902b, and collect local AI / ML models from the base nodes 902a and / or 902b. Furthermore, the aggregation node 903 is connected to the base nodes 903a and 903b in a communicative and operational manner. These connections indicate that the aggregation node 903 may send the common AI / ML model M2 received from node 902 to the base nodes 903a and 903b, and collect local AI / ML models from the base nodes 903a and / or 903b. Furthermore, aggregation node 904 is connected to base nodes 904a and 904b in a communicative and operational manner. These connections indicate that aggregation node 904 may transmit the common AI / ML model M3 received from node 903 to base nodes 904a and 904b, and collect local AI / ML models from base nodes 904a and / or 904b. Furthermore, aggregation node 905 is connected to base nodes 905a and 905b in a communicative and operational manner. These connections indicate that aggregation node 905 may transmit the common AI / ML model M4 received from node 904 to base nodes 905a and 905b, and collect local AI / ML models from base nodes 905a and / or 905b. Furthermore, aggregation node 906 is connected to base nodes 906a and 906b in a communicative and operational manner. These connections indicate that aggregation node 906 may transmit the common AI / ML model M5 received from node 905 to base nodes 906a and 906b, and collect local AI / ML models from base nodes 906a and / or 906b. Furthermore, aggregation node 907 is connected to base nodes 907a and 907b in a communicative and operational manner.These connections indicate that aggregation node 907 may transmit the common AI / ML model M6 received from node 906 to base nodes 907a and 907b, and collect local AI / ML models from base nodes 907a and / or 907b. Furthermore, aggregation node 908 is connected to base nodes 908a and 908b in a communicative and operational manner. These connections indicate that aggregation node 908 receives the common AI / ML model M6 from node 907. i This is sent to base nodes 908a and 908b, indicating that local AI / ML models may be collected from base nodes 908a and / or 908b.

[0184] Each base node in Figure 10 is communicatively and operationally connected to only one aggregation node, but in some embodiments, a base node may be communicatively and operationally connected to multiple aggregation nodes. Furthermore, each aggregation node in Figure 10 is communicatively and operationally connected to several base nodes, but in some embodiments, some aggregation nodes may not be communicatively and / or operationally connected to any base node. Such aggregation nodes may not collect local AI / ML models, but simply receive common AI / ML models from other aggregation nodes.

[0185] If an aggregation node i exists in the network, aggregation node i may receive information related to each AI / ML model from one or more aggregation nodes that are communicatively and operationally connected to aggregation node i. Each aggregation node that transmits information related to its own AI / ML model may be called the previous aggregation node connected to aggregation node i. Aggregation node i may transmit information related to its own AI / ML model to one or more other aggregation nodes that are communicatively and operationally connected to aggregation node i. Each aggregation node that receives information related to the AI / ML model of aggregation node i may be called the next aggregation node connected to aggregation node i. In the network shown in Figure 10, for aggregation node 902, the previous aggregation node is aggregation node 901, and the next aggregation node is aggregation node 903.

[0186] While topology 1000 in Figure 10 shows that each aggregation node has one previous aggregation node and one next aggregation node, it should be noted that, as shown in Figure 16, an aggregation node may have one or more previous aggregation nodes and one or more next aggregation nodes. Figure 16 shows that each aggregation node may have one or more previous aggregation nodes and one or more next aggregation nodes.

[0187] As described above, aspects of this disclosure provide methods for configuring AI / ML topologies. Several methods for configuring AI / ML topologies in wireless communication networks are shown below and elsewhere in this disclosure.

[0188] According to several aspects of this disclosure, a wireless communication network or network device (e.g., BS, TRP) constitutes an AI / ML topology in the wireless communication network. To constitute the topology, the network or network device receives information from nodes, including reports on the nodes' AI / ML capabilities. The network or network device can then configure nodes based on the received information. Node configuration may include configuring node types for the nodes. For example, the configured node type may be one of several node types, including at least a Type 1 node type (e.g., aggregation node) and a Type 2 node type (e.g., base node). As described above, a Type 1 node is configured to collect multiple AI / ML models and aggregate the collected AI / ML models in order to obtain a new or updated AI / ML model (e.g., a Type 1 AI / ML model). A Type 2 node is configured to train an AI / ML model (e.g., a Type 2 AI / ML model) on a set of training data (e.g., local data) without performing aggregation operations. After the network or network device has configured the nodes, the network or network device also configures one or more other nodes associated with the configured nodes. A network or network device can configure one or more other nodes in different ways based on the node type of the configured node.

[0189] In this way, a network or network device can configure the topology of AI / ML within a wireless communication network. The configured topology can support the transfer of AI / ML models via the wireless interface of the wireless communication network. The AI / ML model transfer may be a heterogeneous AI / ML model transfer, and the transfer of AI / ML models between different nodes in the topology can include AI / ML models having different neural network structures. The AI / ML model transfer can include the delivery of a complete AI / ML model or a partial AI / ML model. The configured topology can include at least one of the connections between at least one aggregation node and zero or more basic nodes, or the connections between at least two aggregation nodes.

[0190] FIG. 11 shows a configured topology that supports an exemplary AI / ML model transfer according to an embodiment of the present disclosure. Referring to FIG. 11, node 1101 i-1 , 1101 i , 1101 i+1 and each of 1101a to 1101f may be a user equipment (UE), a repeater, a base station (BS), a transmit receive point (TRP), an edge device, a network system, or an integrated access backhauled (IAB) node. Network 1100 or network device 1150 receives information including a report on the AI / ML capabilities from each of the nodes 1101 i-1 , 1101 i , 1101 i+1 , 1101a to 1101f in order to configure a topology that supports AI / ML model transfer. For example, network 1100 or network device 1150 receives information including a report on the AI / ML capabilities from the nodes, from node 1101 i . Node 1101 i-1 , 1101 i , 1101 i+1They may optionally submit an aggregation node request (i.e., a request to be configured as an aggregation node). Each node's aggregation node request may be included in information that includes a report on the node's AI / ML capabilities.

[0191] Network 1100 or network device 1150 has each node (i.e., node 1101 i-1 , 1101 i , 1101 i+1 , and each of 1101a to 1101f) constitute the node type. In particular, the network 1100 or network device 1150 constitutes the node type of each node. The node type of each node may be configured based on information received from that node, including a report on the node's AI / ML capabilities. In some embodiments, the node type of each node may be configured as type 1 (such as an aggregate node type) or type 2 (such as a base node type). For example, in the example shown in Figure 11, node 1101 i-1 , 1101 i , 1101 i+1 The node type of node 1 may be configured as type 1, and the node types of nodes 1101a to 1101f may be configured as type 2.

[0192] Node (i.e., Node 1101) i-1 , 1101 i , 1101 i+1 If (1101a to 1101f) is configured as described above, the network 1100 or network device 1150 may further configure one or more other nodes related to the configured node. For example, one or more other nodes may be configured differently based on the node type of the configured node.

[0193] For example, aggregation node 1101 i If it is configured by network 1100 or network device 1150, then base node 1101a, base node 1101b, aggregation node 1101 i-1 , or aggregation node 1101 i+1At least one of these may be comprised of network 1100 or network device 1150. Basic nodes 1101a and 1101b are node 1101 i The basic nodes 1101a and 1101b may be configured to connect to node 1101 in a communicative and operational manner. i It may be configured to send to: Aggregation node 1101 i-1 Node 1101 i It may be configured as the aggregation node before node 1101, and as a result node 1101 i-1 Common AI / ML model CM i-1 to node 1101 i Send to: Aggregation node 1101 i+1 Node 1101 i It may be configured as the next aggregation node, resulting in node 1101 i+1 Common AI / ML model CM i to node 1101 i Received from.

[0194] In another example, if the base nodes 1101a and 1101b are configured by network 1100 or network device 1150, then the aggregation node 1101 i The network may also be composed of network 1100 or network device 1150. For example, aggregation node 1101 i It may be configured to communicate and operate with the base nodes 1101a and 1101b. Figure 11 shows that the base nodes 1101a and 1101b are connected to one aggregation node (i.e., node 1101 i This indicates that the nodes are connected to the aggregation node in a communicative and operational manner, but in some embodiments, as shown in Figure 17, several base nodes (for example, node 1701 in Figure 17) may be connected to a plurality of aggregation nodes in a communicative and operational manner. In some embodiments, when base nodes 1101a and 1101b are configured, the aggregation node 1101 i It may be configured to collect local AI / ML models from the base nodes 1101a and 1101b, respectively.

[0195] Node 1101 included in network 1100 i-1 , 1101 i , 1101 i+1 It should be noted that 1101a, 1101b and other nodes may be connected in a communicable and operational manner within the topology via sidelinks using device-to-device (D2D) communication, via network devices (e.g., BS, TRP, etc.), or via interfaces between network devices. For example, in some embodiments, the aggregation node 1101 i The associated basic nodes 1101a and 1101b may be connected in a communicative and operational manner via side links using device-to-device (D2D) communication. In some embodiments, the aggregation node 1101 i and 1101 i+1 It can be connected in a communicative and operational manner via the network device 1150. In some embodiments, the aggregation node 1101 i and 1101 i+1 (These may be connected in a communicative and operational manner via network device 1150 and other network devices (not shown in Figure 11). In such a case, aggregation node 1101 i and 1101 i+1 It can be connected in a communicative and operational manner via an interface between the network device 1150 and other network devices.

[0196] According to several aspects of this disclosure, network nodes, such as user devices, can autonomously configure a topology for AI / ML in a wireless communication network. To configure a topology for AI / ML, a node (first node) sends an aggregation message to another node (second node). The second node decides whether to connect to the first node in a communicative and operational manner in order to send an AI / ML model to the first node or to receive an AI / ML model from the first node. The second node then sends a response to the first node indicating its decision. After the first node receives the response (e.g., an aggregation acknowledgment message), the first and second nodes selectively establish an aggregation connection between them based on the second node's decision. For example, if the second node decides to connect to the first node, the second node can send a positive aggregation acknowledgment message to the first node. After the first node receives the positive aggregation acknowledgment message, the first and second nodes can establish a connection between them. If the second node decides not to connect with the first node, it can send a negative aggregate acknowledgment message to the first node. In such a scenario, if the first node receives the negative aggregate acknowledgment message, the connection between the first and second nodes will not be established.

[0197] In this way, a network (e.g., user equipment) can autonomously configure the AI / ML topology within a wireless communication network. The configured topology can support AI / ML model transfer via the wireless interface of the wireless communication network. The AI / ML model transfer may be heterogeneous AI / ML model transfer. The AI / ML model transfer may include the delivery of a complete or partial AI / ML model. The configured topology may include at least one connection between at least one aggregation node and zero or more base nodes, or at least one connection between at least two aggregation nodes.

[0198] To establish an aggregation connection with other nodes, an aggregation node can send information to other nodes indicating its aggregation capabilities. In this case, the aggregation node may be called a declarative node, and the other nodes may be called monitoring nodes. A declarative node is a node that sends aggregation information that other nodes (such as monitoring nodes) can use to discover the declarative node. A monitoring node is a node that monitors the aggregation information sent by the declarative node. A monitoring node may be a base node or another aggregation node (i.e., not a declarative node). For example, if the monitoring node is another aggregation node, the monitoring node may be a previous aggregation node of the declarative node. In some embodiments, if the previous aggregation node is not discovered by the discovery procedure, the aggregation node (i.e., the declarative node) may send a request to a network device (e.g., BS, TRP) to obtain the latest common AI / ML model, rather than receiving the latest common AI / ML model from the previous aggregation node.

[0199] Figure 12 shows an example of the procedure for establishing an aggregate connection between an aggregate node and a base node when configuring an AI / ML topology in a wireless communication network according to an embodiment of the present disclosure. In Figure 12, aggregate node 1101 i The first node is the declaration node, while the base nodes 1101a and 1101b are monitoring nodes.

[0200] In step 1210, aggregation node 1101 i This declares the aggregation information. Specifically, aggregation node 1101 i The system sends discovery messages to the base nodes 1101a and 1101b. In some embodiments, the aggregation node 1101 iDiscover messages can be broadcast, groupcast, or unicast at predetermined intervals, for example, via broadcast / groupcast / unicast RRC, MAC-CE, or DCI, or via an interface between TRP / BS. Discover messages include aggregated information used by base nodes 1101a and 1101b for the discovery of the declaring node. For example, a discover message may include a model acquisition indicator, a reference AI / ML model, or both. The model acquisition indicator may include a distillation indicator, an expansion indicator, or both distillation and expansion indicators. Distillation and expansion operations related to AI / ML models are described in more detail below.

[0201] In step 1220, the base nodes 1101a and 1101b are aggregated by the aggregation node 1101 i If they are interested in the discovery message received from node 1101, the base nodes 1101a and 1101b can read and process the received discovery message. Then, the base nodes 1101a and 1101b will process the discovery message from node 1101 i You may decide whether or not to connect in a way that allows communication and operation.

[0202] In step 1230, the basic nodes 1101a and 1101b each send their own responses to the aggregation node 1101 i Each response can be sent to the aggregation node 1101 by the base node 1101a or 1101b. i This indicates whether a decision has been made to connect to the aggregate node 1101. In some embodiments, the response is either a positive aggregate acknowledgment message or a negative aggregate acknowledgment message. For example, base node 1101a connects to aggregate node 1101. i It was decided to connect to the aggregation node 1101, and the base node 1101b connected to the aggregation node 1101. i You may decide not to connect to it. In this case, the base node 1101a will connect to the aggregation node 1101. i The base node 1101b sends a positive aggregate acknowledgment message (e.g., ACK) to the node, and the aggregate node 1101 iSend a negative aggregation confirmation response message (e.g., NACK).

[0203] In step 1240, after the aggregation node 1101 i receives a response, the aggregation node 1101 i selectively establishes an aggregation connection with one or both of the basic nodes 1101a and 1101b based on the received response. In the illustrated example, both of the basic nodes 1101a and 1101b respond with a positive aggregation confirmation response message in step 1230, and thus, the aggregation node 1101 i assumes in step 1240 that it establishes an aggregation connection with each of the basic nodes 1101a and 1101b. On the other hand, the aggregation node 1101 i may not establish an aggregation connection with the associated basic node that responds with a negative aggregation confirmation response message in step 1230.

[0204] Using the above discovery procedure, after the aggregation connection between the aggregation node 1101 i and the basic nodes 1101a and 1101b is established, the aggregation node 1101 i can collect local AI / ML models from the basic nodes 1101a and 1101b and perform an aggregation operation to generate a new or updated common AI / ML model.

[0205] For example, referring again to FIG. 11, after the aggregation connection between the aggregation node 1101 i and the basic nodes 1101a and 1101b is established, the node 1101 i can receive the AI / ML model CM i-1 from the node 1101 i-1 . For example, the AI / ML model CM i-1 may be communicated by the node 1101 i-1 by broadcast, groupcast, or unicast signaling, e.g., broadcast / groupcast / unicast RRC, MAC-CE or DCI, or an interface between TRPs / BSS. After receiving the AI / ML model CM i-1 , the node 1101i This involves an AI / ML model CM being added to one or more of its associated nodes 1101a and 1101b. i-1 You may send this. For example, AI / ML model CM i-1 Node 1101 i Communication may be conducted via broadcast, groupcast, or unicast signaling, such as broadcast / groupcast / unicast RRC, MAC-CE or DCI, or an interface between TRP / BS.

[0206] Node1101 i AI / ML model CM sent by i-1 After receiving the data, the related nodes 1101a and 1101b can each train their own AI / ML models. Each of these AI / ML models may be one local AI / ML model of related nodes 1101a and 1101b. Each of related nodes 1101a and 1101b can use its own training dataset, its own AI / ML algorithm (e.g., the training algorithm for the AI / ML model), or node 1101 i AI / ML model commercial received from i-1 At least one of these can be used to train your (related) AI / ML model. In some embodiments, training is performed at node 1101 i AI / ML model commercial received from i-1 Based on this, it may be carried out by transfer learning, knowledge distillation and / or knowledge augmentation. AI / ML Model CM i-1 The related AI / ML models of related nodes 1101a and 1101b may have the same input and / or output types. However, the AI / ML model CM i-1 The related AI / ML models of related nodes 1101a and 1101b may have different neural network (NN) structures. In other words, the related AI / ML models of related nodes 1101a and 1101b may be AI / ML model CM i-1 It is not necessary to have an NN structure equivalent to the NN structure of the other NN.

[0207] In some embodiments, node 1101 i may send an indicator to collect AI / ML models from associated nodes 1101a and 1101b. For example, the indicator may be communicated by node 1101 i via broadcast, groupcast, or unicast signaling, such as broadcast / groupcast / unicast RRC, MAC-CE, or DCI, or an interface between TRPs / BSs. This indicator may be referred to as a model collection indicator. Node 1101 i may send a model collection indicator to associated nodes 1101a and 1101b to request reports related to the associated AI / ML models of each of the associated nodes 1101a and 1101b. In some embodiments, the reports related to the associated AI / ML models of associated nodes 1101a and 1101b are generated by associated nodes 1101a and 1101b based on the model collection indicator. In such embodiments, the model collection indicator may be a dynamic indicator or an event trigger indicator. For example, if the model collection indicator is a dynamic indicator, node 1101 i may receive reports related to the associated AI / ML models of each of the associated nodes 1101a and 1101b from the associated nodes 1101a and 1101b at node 1101 iIt can be received at each pre-set time. On the other hand, if the model acquisition indicator is an event trigger indicator, related nodes 1101a and 1101b can each send their own reports when a specific event is triggered. An event can be triggered when training of each related AI / ML model of related nodes 1101a and 1101b is completed, or when the performance of each related AI / ML model of the related node exceeds a specific performance metric. The specific performance metric can be pre-set based on at least one of accuracy or precision. If the performance of one of the related AI / ML models does not exceed a pre-set performance metric (for example, accuracy is lower than a given precision metric), the report related to that AI / ML model will not be generated and / or node 1101 i It may not be sent (for example, there is no local AL / ML model report submission).

[0208] In some embodiments, each of the related nodes 1101a and 1101b optionally transmits an acknowledgment indicator for the transmission of its related AI / ML model. For example, the acknowledgment indicator may be communicated via PUCCH or PUSCH, a sidelink channel, or an interface between TRP / BS. A positive acknowledgment indicator (e.g., ACK) indicates that a report related to the related AI / ML model is being transmitted. A negative acknowledgment indicator (e.g., NACK) indicates that a report related to the related AI / ML model is not being transmitted (e.g., the related AI / ML model is not reported). In some embodiments, the acknowledgment indicator may be transmitted before the transmission of the report related to the related AI / ML model. In some other embodiments, the acknowledgment indicator may be included in the report related to the related AI / ML model.

[0209] Subsequently, related nodes 1101a and 1101b will send reports related to the related AI / ML models of related nodes 1101a and 1101b to node 1101 iIt may be sent. In some embodiments, the reports related to the associated AI / ML models of associated nodes 1101a and 1101b may include at least one of information related to the associated AI / ML models of each of the associated nodes 1101a and 1101b, information related to the training data of the associated AI / ML models of each of the associated nodes 1101a and 1101b, or information related to the performance of the associated AI / ML models of each of the associated nodes 1101a and 1101b. The information related to each associated AI / ML model can include the NN structure (NN algorithm, width, depth, etc.) of the associated nodes 1101a and 1101b and / or one or more AI / ML parameters (weights, biases, activation functions, etc.). The information related to the training data of each associated AI / ML model can include information related to the amount (volume) of training data and / or the training data distribution. The information related to the performance of each associated AI / ML model can include accuracy, precision, recall, loss information (e.g., mean cross-entropy loss), the verification / test dataset composed by node 1100 i (or other aggregation nodes or BS). Note that in some embodiments, the information related to the training data of each associated AI / ML model and / or the information related to the performance of each associated AI / ML model can assist in the aggregation operation of the aggregation node. For example, the training data information and / or performance information can be used to determine the aggregation weights of the AI / ML model.

[0210] Node 1101 i After receiving the reports related to the associated AI / ML models of each of the associated nodes 1101a and 1101b, node 1101 i generates an AI / ML model CM i based on the received reports. The AI / ML model CM i may be an updated common AI / ML model. The AI / ML models CM i-1 and CM i may have the same NN structure. In some embodiments, node 1101 iThis involves using aggregation algorithms (e.g., model distillation, model expansion) to perform AI / ML model CM. i This generates. In some embodiments, node 1200 i It may generate its own AI / ML model. This AI / ML model may be a local AI / ML model. This AI / ML model may be generated without reports related to the respective related AI / ML models of related nodes 1101a and 1101b.

[0211] Next, node1101 i This is an AI / ML model commercial. i to node 1101 i+1 You may send it to: AI / ML Model CM i This may include one or more AI / ML model parameters (weights, biases, etc.), but the AI / ML model CM i It does not need to include information related to the NN structure. In other words, AI / ML model CM i The NN structure may be pre-configured, and / or the AI / ML model CM. i The NN structure is at node 1101 i+1 It may be already known, so AI / ML model CM i Information related to the NN structure can be found at node 1101. i From node 1101 i+1 It does not have to be sent to node 1200. In some embodiments, node 1200 i AI / ML model CM to be sent i This may be a complete AI / ML model or a partial AI / ML model.

[0212] As described above, the related AI / ML models of related nodes 1101a and 1101b are AI / ML model CM i-1 and AI / ML model CM i It is not limited to having an NN structure equivalent to the NN structure of the NN structure. In other words, heterogeneous AI / ML model aggregation can be enabled by configuring a topology for AI / ML in a wireless communication network according to embodiments of the present disclosure. For example, in some embodiments, node 1101 iby the previous node 1101 i-1 AI / ML model commercial received from i-1 And, node 1101 i Based on reports received from its associated nodes 1101a and 1101b, the AI / ML model CM was generated by the AI / ML model. i Both can include a 4-layer NN, but the local AI / ML models generated by related nodes 1101a and / or 1101b can have different NN structures. For example, the local AI / ML model generated by related node 1101a may include a 5-layer NN, while the local AI / ML model generated by related node 1101b may include a 3-layer NN.

[0213] Node1101 i After receiving reports related to the local AI / ML models of its associated nodes 1101a and 1101b, it then uses the AI / ML model CM based on the local AI / ML models generated by associated nodes 1101a and 1101b. i Aggregation operations (e.g., model distillation, model expansion) can be performed to generate the AI / ML model CM. Each of the local AI / ML models generated by related nodes 1101a and 1101b can be used to generate the AI / ML model CM. i Since the generation of may have different importance (importance), each of the local AI / ML models generated by related nodes 1101a and 1101b may, when aggregating the local AI / ML models, for example, W a and W b They may be weighted according to the respective weights, such as W. a and W b This may indicate the importance of the local AI / ML models generated by related nodes 1101a and 1101b. Generated AI / ML model CM i This could be an updated common AI / ML model. CM i The NN structure is an AI / ML model CM i-1 The NN structure may be the same as the above. In this case, the AI / ML model CM iAI / ML model commercial i-1 Both can, for example, include a 4-layer neural network.

[0214] However, as mentioned above, AI / ML model CM i The local AI / ML models LM1, LM2, and LM3 may have different neural network (NN) structures. The local AI / ML models generated by related nodes 1101a and 1101b are AI / ML model CM i To generate it, it is not necessary to have an NN structure equivalent to that of the common AI / ML model. In other words, it becomes possible to aggregate heterogeneous AI / ML models on a self-organizing topology.

[0215] In some embodiments, an aggregation node receives one or more heterogeneous AI / ML models from one or more related base nodes that are communicatively and operationally connected to the aggregation node. The aggregation node can distill and / or expand the received heterogeneous AI / ML models. The aggregation can then aggregate the distilled and / or expanded AI / ML models to generate a new or updated common AI / ML model. In some embodiments, the aggregation node can obtain the average of the distilled AI / ML models and / or the average of the expanded AI / ML models. After obtaining the common AI / ML model, the aggregation node can transmit the new or updated common AI / ML model to one or more other nodes, which may be the next aggregation node.

[0216] As described above, an aggregation node may distill and / or expand an AI / ML model if the NN structure of the aggregation node's associated base nodes differs from the NN structure of the aggregation node's common AI / ML model. Otherwise, if the NN structure of the aggregation node's associated base nodes is the same as the NN structure of the aggregation node's common AI / ML model, the distillation and / or expansion operation does not need to be performed. Distillation and / or expansion may also be part of the aggregation operation of the aggregation node. Distillation may involve generating a smaller AI / ML model from an AI / ML model received by a node, and expansion may involve generating a larger AI / ML model from an AI / ML model received by a node. Thus, distillation of an aggregation node may involve the aggregation node generating a common AI / ML smaller than an AI / ML model received from another node (e.g., a local AI / ML model received from an associated base node). Similarly, expansion of an aggregation node may involve the aggregation node generating a common AI / ML model larger than an AI / ML model received from another node (e.g., a local AI / ML model received from an associated base node).

[0217] For the purpose of illustrating distillation and / or expansion, if the first AI / ML model is larger than the second AI / ML model, the first AI / ML model may have more floating-point operations, a larger total number of parameters, more trainable parameters, a larger required buffer size, a larger width, a larger depth, or any combination thereof. Similarly, for the purpose of illustrating distillation and / or expansion, if the first AI / ML model is smaller than the second AI / ML model, the first AI / ML model may have fewer floating-point operations, a smaller total number of parameters, fewer trainable parameters, a smaller required buffer size, a smaller width, a smaller depth, or any combination thereof.

[0218] Figure 13 shows an example of the procedure for establishing an aggregation connection between two aggregation nodes when configuring an AI / ML topology in a wireless communication network according to an embodiment of the present disclosure. In Figure 13, aggregation node 1101 i This is the declaration node, and the aggregation node 1101 i-1 and 1101 k This is a monitoring node. Aggregation node 1101 i-1 is aggregation node 1101 i It could also be the aggregation node before it.

[0219] In Step 1310, aggregation node 1101 i This declares the aggregation information. Specifically, aggregation node 1101 i is aggregation node 1101 i-1 and 1101 k A discovery message is sent to the aggregation node 1101. In some embodiments, i The discovery message may be broadcast, groupcast, or unicast at predetermined intervals. The discovery message is sent to the aggregation node 1101 for the discovery of the declaring node. i-1 and 1101 k Includes aggregation information used by, for example, aggregation node 1101 i-1 and 1101 k The aggregated information used by aggregate node 1101 i It may also include the ability to aggregate.

[0220] In step 1320, aggregation node 1101 i-1 and 1101 k Aggregation node 1101 i If you are interested in discovery messages received from aggregate node 1101 i-1 and 1101 k It can read and process the received discovery message. Next, aggregation node 1101 i-1 and 1101 k Node 1101 i You may decide whether or not to connect in a way that allows communication and operation.

[0221] In Step 1330, aggregation node 1101 i-1 and 1101 k Each of them aggregates its own response to node 1101 i Each response can be sent to aggregation node 1101. i-1 Or 1101 k Aggregation node 1101 i This indicates whether a decision has been made to connect to it. In some embodiments, the response is either a positive aggregate acknowledgment message or a negative aggregate acknowledgment message. For example, aggregate node 1101 i-1 is aggregation node 1101 i It was decided to connect to aggregate 1101 k is aggregation node 1101 i You may decide not to connect to it. In this case, aggregate node 1101 i-1 is aggregation node 1101 i Send a positive aggregate acknowledgment message (e.g., ACK) to the node, and aggregate 1101 k is aggregation node 1101 i Send a negative aggregate acknowledgment message (e.g., NACK) to it.

[0222] In some embodiments, an aggregation node may be connected to communicate and / or operate with only one other aggregation node. For example, aggregation nodes in a network may have a one-to-one relationship with other aggregation nodes. In such embodiments, the declaring aggregation node sends a conflict confirmation to the monitoring aggregation node to inform it whether the connection to the declaring node has been successfully established. In Figure 13, aggregation node 1101 i Aggregation node 1101 i-1 and 1101 k If it can only connect to one of the following, aggregate node 1101 i In step 1335, optionally, the aggregation node 1101 i-1 and 1101 k Send a conflict check to aggregate node 1101 i Notifies whether the connection was successful.

[0223] In step 1340, aggregation node 1101 i Aggregation node 1101 i-1 Assuming that it can only connect to the aggregate node 1101, i Based on the received response, conflict check, or both, the aggregation node 1101 i-1 Establish an aggregate connection with the aggregate node 1101. i and aggregation node 1101 k No aggregate connection is established between them.

[0224] Aggregation node 1101 i and aggregation node 1101 i-1 After an aggregate connection is established between them, aggregate node 1101 i is aggregation node 1101 i-1 It can receive common AI / ML models from there.

[0225] To establish an aggregation connection with other nodes, a node can send an aggregation request message (e.g., an aggregation request) to the other node. In this case, the node that sends the aggregation request message can be called the discovering node, and the other node that receives the aggregation request message can be called the discovered node. If the discovering node is a base node, the discovering node can send an aggregation request message to the associated aggregation node. If the discovering node is an aggregation node, the discovering node can send an aggregation request message to other aggregation nodes to request the other aggregation node as the next hop (e.g., requesting the other aggregation to be the next aggregation node). In some embodiments, if the next aggregation node is not discovered by the discovery procedure, the aggregation node (discovering node) can send the aggregated common AI / ML model to a network device (e.g., BS, TRP).

[0226] Figure 14 shows another example of the procedure for establishing an aggregate connection between an aggregation node and a base node when configuring an AI / ML topology in a wireless communication network according to an embodiment of the present disclosure. In Figure 14, the base node 1101a is a discovery node, and the aggregation node 1101 i and 1101i+1 This is the node being discovered.

[0227] In step 1410, the base node 1101a is the aggregation node 1101 i and 1101 i+1 A request for aggregation is sent to the base node 1101a. In some embodiments, the base node 1101a may broadcast, groupcast, or unicast the request for aggregation at predetermined intervals. The request for aggregation may include requested aggregation information that the base node 1101a is interested in discovering. If the base node 1101a is a discovering node, the request for aggregation may include information related to the base node 1101a's local AI / ML model. The information related to the base node 1101a's local AI / ML model may include information related to the neural network of the base node 1101a's local AI / ML model, information related to the size of the base node 1101a's local AI / ML model, or information related to the complexity of the base node 1101a's local AI / ML model. In some embodiments, the information related to the size and / or complexity of the local AI / ML model may be the relative size and / or complexity of the local AI / ML model compared to a reference AI / ML model.

[0228] In step 1420, aggregation node 1101 i and 1101 i+1 If the aggregation node 1101 is interested in the aggregation request message received from the base node 1101a, then the aggregation node 1101 i and 1101 i+1 It can read and process the received aggregation request message. Next, aggregation node 1101 i and 1101 i+1 It may decide whether or not to connect to the base node 1101a in a way that allows communication and operation. For example, the discovered aggregation node 1101 i If the discovery base node 1101a has a distillation function and reports a larger AI / ML model than the reference AI / ML model, then the aggregation node 1101 iIt decides to establish a communicative and operational connection with the base node 1101a. i+1 If the discovery base node 1101a has a distillation function and reports an AI / ML model smaller than the reference AI / ML model, then the aggregation node 1101 i+1 It decides to establish a communicative and operational connection with the base node 1101a.

[0229] In step 1430, aggregation node 1101 i and 1101 i+1 Each can send its own response to the base node 1101a. Each response is sent to the aggregation node 1101a. i Or 1101 i+1 This indicates whether it has decided to connect to the base node 1101a. In some embodiments, the response is either a positive aggregate acknowledgment message or a negative aggregate acknowledgment message. For example, the discovered aggregate node 1101 i It has a distillation function, and the discovery aggregation node 1101 i+1 It has extension capabilities, and since discovery base node 1101a reports a larger AI / ML model than the reference AI / ML model, aggregation node 1101 i It decided to connect to the base node 1101a, and the aggregation node 1101 i+1 It may decide not to connect to the base node 1101a. In this case, the aggregation node 1101 i The base node 1101a sends a positive aggregate acknowledgment message (e.g., ACK), and the aggregate node 1101 i+1 It sends a negative aggregate acknowledgment message (e.g., NACK) to the base node 1101a.

[0230] In step 1435, the base node 1101a is optionally the aggregation node 1101 i and 1101 i+1 You can also send a conflict check to verify that each is successfully connected to base node 1101a.

[0231] In step 1440, the base node 1101a, based on the received response, conflict confirmation, or both, will aggregate node 1101 i Establish an aggregate connection between the base node 1101a and the aggregate node 1101. i+1 No aggregate connection is established between them.

[0232] The discovery procedure identified aggregation node 1101a and aggregation node 1101 i After an aggregate connection is established between them, aggregate node 1101 i This can collect local AI / ML models from the base node 1101a, perform aggregation operations, and generate new or updated common AI / ML models.

[0233] Figure 15 shows another example of the procedure for establishing an aggregate connection between two aggregate nodes when configuring an AI / ML topology in a wireless communication network according to an embodiment of the present disclosure. In Figure 15, aggregate node 1101 i This is a discovery node, and aggregation node 1101 i+1 and 1101 k This is the discovered node. Aggregation node 1101 i+1 is aggregation node 1101 i It could also be the next aggregation node.

[0234] In step 1510, aggregation node 1101 i is aggregation node 1101 k and 1101 i+1 A request for aggregation is sent to the aggregation node 1101. In some embodiments, the aggregation node 1101 i For example, aggregate request messages can be broadcast, groupcast, or unicast at predetermined intervals via broadcast / groupcast / unicast RRC, MAC-CE, DCI, or an interface between TRP / BS. The aggregate request message is sent to the aggregate node 1101 i It can include aggregate requests from other sources.

[0235] In step 1520, aggregation node 1101 kand 1101 i+1 Aggregation node 1101 i If you are interested in the aggregation request message received from aggregation node 1101 k and 1101 i+1 It can read and process the received aggregation request message. Next, aggregation node 1101 k and 1101 i+1 is aggregation node 1101 i You may decide whether or not to connect in a way that allows communication and operation.

[0236] In step 1530, aggregation node 1101 k and 1101 i+1 Each of them aggregates its own response to node 1101 i Each response can be sent to aggregation node 1101. k Or 1101 i+1 Aggregation node 1101 i This indicates whether a decision has been made to connect to it. In some embodiments, the response is either a positive aggregate acknowledgment message or a negative aggregate acknowledgment message. For example, aggregate node 1101 i+1 is aggregation node 1101 i It was decided to connect to aggregate 1101 k is aggregation node 1101 i You may decide not to connect to it. In this case, aggregate node 1101 i+1 is aggregation node 1101 i Send a positive aggregate acknowledgment message (e.g., ACK) to the node, and aggregate 1101 k is aggregation node 1101 i Send a negative aggregate acknowledgment message (e.g., NACK) to it.

[0237] In some embodiments, a discovery aggregation node may be connected communicatively and / or operationally to only one other aggregation node. For example, a discovery aggregation node in a network may have a one-to-one relationship with other aggregation nodes. In such embodiments, the discovery aggregation node sends a conflict confirmation to the discovered aggregation node to notify it whether a connection to the discovery node has been successfully established. In Figure 15, discovery aggregation node 1101 i The discovered aggregation node 1101 k and 1101 i+1 If it can only connect to one of the following, aggregate node 1101 i In step 1535, optionally, the discovered aggregation node 1101 k and 1101 i+1 Send a conflict check to aggregate node 1101 i Notifies whether the connection was successful.

[0238] In step 1540, aggregation node 1101 i Aggregation node 1101 i+1 Assuming that it can only connect to the aggregate node 1101, i Based on the received response, conflict check, or both, the aggregation node 1101 i+1 Establish an aggregate connection with the aggregate node 1101. i and aggregation node 1101 k No aggregate connection is established between them.

[0239] Aggregation node 1101 i and aggregation node 1101 i+1 After the aggregation connection between the two is established by the discovery procedure, the discovered aggregation node 1101 i+1 This is discovery aggregation node 1101 i It can receive common AI / ML models from there.

[0240] According to several aspects of this disclosure, a topology for AI / ML in a wireless communication network can be configured such that the aggregation of AI / ML models is processed in parallel across multiple aggregation nodes. In other words, a topology configured based on the methods shown in this disclosure may be flexible.

[0241] In some embodiments, the topology for AI / ML in a wireless communication network can be configured such that each aggregation node is connected to one or more aggregation nodes, as shown in Figure 16. Figure 16 shows an example of a configured topology 1600 that supports flexible communication between aggregation nodes according to embodiments of the present disclosure.

[0242] In the configured topology 1600, an aggregation node can send a common AI / ML model to one or more AI / ML aggregation nodes. For example, aggregation node 901 sends the common AI / ML model M1 to aggregation nodes 902, 903, 904, and 905. On the other hand, aggregation node 907 sends the common AI / ML model M i This is sent only to aggregation node 901. Note that the aggregation node in Figure 16 can transmit the common AI / ML model via broadcast, groupcast, or unicast signaling, for example, via broadcast / groupcast / unicast RRC, MAC-CE or DCI, or an interface between TRP / BS.

[0243] In the configured topology 1600, an aggregation node can receive common AI / ML models from multiple AI / ML aggregation nodes. For example, aggregation node 908 receives AI / ML models M2, M3, M4, and M5 from aggregation nodes 902, 903, 904, and 905.

[0244] Regarding the aggregation of AI / ML models, the aggregation operation is shown below using aggregation node 908. Aggregation node 908 receives AI / ML models M2, M3, M4, and M5 from aggregation nodes 902, 903, 904, and 905, respectively, as described above. Next, aggregation node 908 aggregates the received AI / ML models M2, M3, M4, and M5. In some embodiments, aggregation node 908 can calculate the average of AI / ML models M2, M3, M4, and M5 to obtain the aggregated AI / ML model. After obtaining the aggregated AI / ML model, aggregation node 908 sends the aggregated AI / ML model to its associated base nodes. In this example, the associated base nodes include three base nodes 908a, 908b, and 908c that are communicatively and operationally connected to aggregation node 908. The base nodes 908a, 908b, and 908c train their own local AI / ML models using the aggregated common AI / ML model sent from aggregation node 908. Once training is complete, aggregation node 908 collects the local AI / ML models from base nodes 908a, 908b, and 908c. Next, aggregation node 908 aggregates the collected local AI / ML models to create an updated common AI / ML model M. j This generates the model. The aggregation node 908 can then send the updated common AI / ML model to its next aggregation node 901.

[0245] In some embodiments, the topology for AI / ML in a wireless communication network can be configured such that each base node is connected to zero or more aggregation nodes, as shown in Figure 17. Figure 17 shows an example of a topology 1700 configured to support flexible communication between aggregation nodes and base nodes according to embodiments of the present disclosure.

[0246] In the configured topology 1700, each of the aggregation nodes 901-905, 907, and 908 is communicatively and operationally connected to one or more base nodes. For example, aggregation node 901 is communicatively and operationally connected to base nodes 901a and 1701. However, aggregation node 906 is not communicatively and operationally connected to any base nodes. Since aggregation node 906 has no associated base nodes, it can generate a common AI / ML model updated in a different way. For example, aggregation node 906 receives a common AI / ML model from another aggregation node (e.g., aggregation node 905) and generates its own local AI / ML model. Then, aggregation node 906 calculates the average of the received common AI / ML model and its own local AI / ML model to form an updated common AI / ML model. When aggregation node 906 obtains the updated common AI / ML model, it sends the updated AI / ML model to its next aggregation node (for example, aggregation node 907).

[0247] In the configured topology 1700, each base node is connected to one or more aggregation nodes in a communicative and operational manner. For example, base node 901a is connected to aggregation node 901 in a communicative and operational manner.

[0248] On the other hand, aggregation node 1701 is connected to two aggregation nodes 901 and 908 in a communicative and operational manner. The connection of the base node 1701 can be carried out using a network device (e.g., BS). For example, the network device broadcasts multiple local AI / ML models (each local AI / ML model has an AI / ML model identifier). The network device indicates a set of AI / ML model identifiers to be aggregated by each aggregation node. If the network device indicates AI / ML model identifiers, the same AI / ML model identifiers can be assigned to multiple aggregation nodes. Thus, the identifiers of local AI / ML models associated with base node 1701 can be assigned to both aggregation nodes 901 and 908.

[0249] In some aspects of this disclosure, heterogeneous AI / ML functionality can be enabled on various network devices and user devices, and heterogeneous AI / ML model transfer can be supported via the wireless interface of a wireless communication network.

[0250] In some aspects of this disclosure, the topology within a wireless communication network can be centrally configured by network devices (e.g., base stations (BS), transmit and receive points (TRP)) or network systems.

[0251] In some aspects of this disclosure, the topology within a wireless communication network can be autonomously configured by network equipment (e.g., base stations (BS)), transmit and receive points (TRPs) and / or user equipment using discovery procedures as described herein, thereby supporting flexible topology configuration.

[0252] In some aspects of this disclosure, topologies within wireless communication networks configured by the various methods described herein support parallel AI / ML model aggregation, reduce AI / ML model routing latency, and improve the robustness of AI / ML model routing.

[0253] Examples of apparatus for carrying out the various methods described herein (e.g., ED or UE and TRP or network equipment) are also disclosed.

[0254] For example, the first device may include a memory for storing processor-executable instructions and a processor for executing the processor-executable instructions. When the processor executes the processor-executable instructions, the processor may be made to perform one or more method steps of the device described herein, for example in relation to Figures 11 to 15. For example, the processor may cause the device to communicate via a radio interface in an operating mode by performing an operation consistent with that operating mode. For example, it may perform necessary measurements as configured for the operating mode, generate content from those measurements, prepare uplink transmissions, process downlink transmissions, perform coding, decoding, etc., configure and / or direct transmit / receive on the RF chain and antenna.

[0255] It should be noted that the expression "at least one of A or B" as used herein is interchangeable with the expression "A and / or B". This represents a list in which one can select either A or B, or both A and B. Similarly, the expression "at least one of A, B, or C" as used herein is interchangeable with "A and / or B and / or C" or "A, B, and / or C". This represents a list in which one can select either A or B or C, or A and B, or both A and C, or both B and C, or all of A, B, and C. The same principle applies to longer lists of the same form.

[0256] While the present invention has been described in relation to its particular features and embodiments, various modifications and combinations can be made without departing from the invention. Accordingly, the description and drawings should be considered merely illustrative of some embodiments of the invention as defined by the appended claims, and are intended to cover any modifications, variations, combinations, or equivalents within the scope of the invention. Thus, while the invention and its merits have been described in detail, various modifications, substitutions, and changes can be made herein without departing from the invention as defined by the appended claims. Furthermore, the scope of this application is not intended to be limited to specific embodiments of the processes, machines, manufactures, compositions of substances, means, methods, and steps described in the specification. As will be readily apparent to those skilled in the art from the disclosure of the invention, existing or subsequently developed processes, machines, manufactures, compositions of substances, means, methods, or steps that perform substantially the same functions or achieve substantially the same results as the corresponding embodiments described herein can be utilized in accordance with the invention. Accordingly, the appended claims are intended to include within their scope such processes, machines, manufactures, compositions of substances, means, methods, or steps.

[0257] Furthermore, any module, component, or device that executes instructions as illustrated herein includes, or is otherwise accessible, one or more non-temporary computer / processor-readable instruction storage media for storing information such as computer / processor-readable instructions, data structures, program modules, and / or other data. A non-exclusive list of examples of non-temporary computer / processor-readable storage media includes magnetic cassettes, magnetic tapes, magnetic disk storage devices or other magnetic storage devices, optical disks such as compact disk read-only memory (CD-ROM), digital video discs or digital versatile discs (DVDs), Blu-ray Discs®, or other optical storage devices, volatile and non-volatile, removable and non-removable media implemented in any way or technique, random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory, or other memory technologies. Such non-temporary computer / processor storage media may be part of a device, or may be accessible or connectable to it. Any application or module described herein may be implemented using computer / processor-readable / executable instructions stored or held by such non-temporary computer / processor-readable storage media.

[0258] Definition of an abbreviation LTE: Long Term Evolution NR: New Radio BWP: Bandwidth part BS: Base Station CA: Carrier Aggregation CC: Component Carrier CG: Cell Group CSI: Channel state information CSI-RS: Channel state information Reference Signal DC: Dual Connectivity DCI: Downlink control information DL: Downlink DL-SCH: Downlink shared channel EN-DC: E-UTRA NR dual connectivity with MCG using E-UTRA and SCG using NR gNB: Next generation (or 5G) base station HARQ-ACK Hybrid automatic repeat request acknowledgement MCG: Master cell group MCS: Modulation and coding scheme MAC-CE: Medium Access Control-Control Element PBCH: Physical broadcast channel PCell: Primary cell PDCCH: Physical downlink control channel PDSCH: Physical downlink shared channel PRACH: Physical Random Access Channel PRG: Physical resource block group PSCell: Primary SCG Cell PSS: Primary synchronization signal PUCCH: Physical uplink control channel PUSCH: Physical uplink shared channel RACH: Random access channel RAPID: Random access preamble identity RB: Resource block RE: Resource element RRM: Radio resource management RMSI: Remaining system information RS: Reference signal RSRP: Reference signal received power RRC: Radio Resource Control SCG: Secondary cell group SFN: System frame number SL: Sidelink SCell: Secondary Cell SPS: Semi-persistent scheduling SR: Scheduling request SRI: SRS resource indicator SRS: Sounding reference signal SSS: Secondary synchronization signal SSB: Synchronization Signal Block SUL: Supplement Uplink TA: Timing advance TAG: Timing advance group TUE : Target UE UCI: Uplink control information UE: User Equipment UL: Uplink UL-SCH: Uplink shared channel

Claims

1. A method for configuring a topology for artificial intelligence or machine learning (AI / ML) in a wireless communication network, wherein the method is: The steps include receiving information from a node, including a report related to the node's AI / ML capabilities, A step of configuring the node based on the received information, wherein the step of configuring the node includes a step of configuring the node type of the node, the configured node type is one of a plurality of node types, and the plurality of node types are Type 1 indicates a node configured to collect multiple AI / ML models and aggregate the collected AI / ML models to obtain a first type AI / ML model, or Type 2 shows a node configured to obtain a second type of AI / ML model using a set of training data without performing aggregation operations. Steps including, The steps of configuring one or more other nodes associated with the configured node based on the configured node type, Includes, The configured topology supports the transfer of AI / ML models via the wireless interface of the wireless communication network, as follows: A connection between at least one Type 1 node and zero or more Type 2 nodes, Connection between at least two Type 1 nodes, A method that includes at least one of the following.

2. The configured node type indicates that the node is a type 1 node, and the one or more other nodes are as follows: One or more Type 2 nodes, A second type 1 node that provides a first type AI / ML model of a second type 1 node to the node, or A third type 1 node that receives a first type AI / ML model from the aforementioned node, The method according to claim 1, comprising at least one of the following.

3. The one or more other nodes include the one or more Type 2 nodes, and the step of configuring the node is: The method according to claim 2, comprising the step of configuring the node to collect each second type AI / ML model from one or more type 2 nodes.

4. The method according to claim 1, wherein the configured node type indicates that the node is a type 2 node, and the one or more other nodes include one or more type 1 nodes connected to the node.

5. The method according to claim 1, further comprising the requirement that the information is configured as a node of type 1 by the node.

6. The method according to claim 1, wherein the node is a user device (UE), a repeater, a base station (BS), a transmit / receive point (TRP), an edge device, a network system, or an integrated access backhaul (IAB) node.

7. A method for configuring a topology for artificial intelligence or machine learning (AI / ML) in a wireless communication network, wherein the method is: The first node establishes an aggregated connection with the second node based on an aggregated acknowledgment message, The configured topology supports the transfer of AI / ML models via the wireless interface of the wireless communication network, as follows: A connection between at least one Type 1 node and zero or more Type 2 nodes, Connection between at least two Type 1 nodes, Includes at least one of the following: A method wherein a Type 1 node is configured to collect multiple AI / ML models and aggregate the collected AI / ML models to obtain a first type AI / ML model, and a Type 2 node is configured to obtain a second type AI / ML model using a set of training data without performing an aggregation operation.

8. The method according to claim 7, wherein the message for aggregation is a discovery message containing aggregation information used by the second node to discover the first node.

9. The method according to claim 8, wherein the discovery message includes at least one of a model acquisition indicator or a reference AI / ML model.

10. The method according to claim 7, wherein the message for aggregation is an aggregation request message, and the aggregation request message includes information or an aggregation request related to the second type AI / ML model of the first node.

11. The method according to claim 10, wherein the first node is a type 2 node, and the aggregation request message includes information relating to the second type AI / ML model of the first node.

12. The information relating to the second type AI / ML model of the first node is as follows: Information related to the neural network of the second type AI / ML model of the first node, Information regarding the size of the second type AI / ML model of the first node, or Information regarding the complexity of the second type AI / ML model in the first node, The method according to claim 11, comprising at least one of the following.

13. A network device for configuring a topology for artificial intelligence or machine learning (AI / ML) in a wireless communication network, wherein the network device includes a processor and a memory for storing processor-executable instructions, and when the processor-executable instructions are executed, the processor, The node receives information including reports related to the node's AI / ML capabilities. Based on the received information, the node is configured, and configuring the node includes configuring the node type of the node, the configured node type is one of a plurality of node types, and the plurality of node types are Type 1 indicates a node configured to collect multiple AI / ML models and aggregate the collected AI / ML models to obtain a first type AI / ML model, or Type 2 refers to a node configured to obtain a second type AI / ML model using a set of training data without performing aggregation operations. Includes, Based on the configured node type, configure one or more other nodes related to the configured node, The configured topology is a network device that supports AI / ML model transfer via the wireless interface of the wireless communication network and includes at least one of the following: connections between at least one type 1 node and zero or more type 2 nodes, or connections between at least two type 1 nodes.

14. The configured node type indicates that the node is a type 1 node, and the one or more other nodes are as follows: One or more Type 2 nodes, A second type 1 node that provides a first type AI / ML model of a second type 1 node to the node, or A third type 1 node that receives a first type AI / ML model from the aforementioned node, The network device according to claim 13, comprising at least one of the following.

15. The one or more other nodes include the one or more Type 2 nodes, and the nodes are composed of, The network device according to claim 14, comprising configuring the nodes to collect each of the second type AI / ML models from one or more type 2 nodes.

16. The network device according to claim 13, wherein the configured node type indicates that the node is a type 2 node, and the one or more other nodes include one or more type 1 nodes connected to the node.

17. The network device according to any one of claims 13 to 16, further comprising the requirement that the information is configured as a Type 1 node by the node.

18. The network device according to any one of claims 13 to 16, wherein the node is a user device (UE), a repeater, a base station (BS), a transmit / receive point (TRP), an edge device, a network system, or an integrated access backhaul (IAB) node.

19. An apparatus comprising one or more units for performing the method according to any one of claims 1 to 6.

20. A device for a first node that constitutes a topology for artificial intelligence or machine learning (AI / ML) in a wireless communication network, wherein the device is Processor and The memory that stores the processor's executable instructions, The processor executes the processor instruction, which includes the instruction, and when the processor executes the instruction, the processor Based on the aggregated acknowledgment message, establish an aggregated connection with the second node. The configured topology supports the transfer of AI / ML models via the wireless interface of the wireless communication network, as follows: A connection between at least one Type 1 node and zero or more Type 2 nodes, Connection between at least two Type 1 nodes, Includes at least one of the following: A device in which a Type 1 node is configured to collect multiple AI / ML models and aggregate the collected AI / ML models to obtain a Type 1 AI / ML model, and a Type 2 node is configured to obtain a Type 2 AI / ML model using a set of training data without performing an aggregation operation.

21. The apparatus according to claim 20, wherein the message for aggregation is a discovery message containing aggregation information used by the second node to discover the first node.

22. The apparatus according to claim 21, wherein the discovery message includes at least one of a model acquisition indicator or a reference AI / ML model.

23. The apparatus according to claim 20, wherein the message for aggregation is an aggregation request message, and the aggregation request message includes information or an aggregation request related to the second type AI / ML model of the first node.

24. The apparatus according to claim 23, wherein the first node is a type 2 node, and the aggregation request message includes information relating to the second type AI / ML model of the first node.

25. The information relating to the second type AI / ML model of the first node is as follows: Information related to the neural network of the second type AI / ML model of the first node, Information regarding the size of the second type AI / ML model of the first node, or Information regarding the complexity of the second type AI / ML model in the first node, The apparatus according to claim 24, comprising at least one of the following.

26. An apparatus comprising one or more units for performing the method according to any one of claims 7 to 12.

27. ​​A computer-readable medium storing instructions, wherein, when executed by a computer, the instructions cause the computer to execute the method according to any one of claims 1 to 12.

Images