Method 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] The present disclosure relates generally to wireless communications and, in particular embodiments, to methods and apparatus for configuring topologies for artificial intelligence or machine learning (AI / ML). [Background technology]

[0002] Artificial intelligence techniques can be applied to communications, including AI-based communications at the physical layer and / or AI-based communications at the medium access control (MAC) layer. For example, at the physical layer, AI-based communications can aim to optimize component design and / or improve algorithm performance. At the MAC layer, AI-based communications can aim to utilize AI capabilities to learn, predict, and / or make decisions to solve complex optimization problems with better strategies and / or optimal solutions possible. For example, functions at the MAC layer can be optimized.

[0003] In some implementations, an AI architecture in a wireless communication network can include multiple nodes. Here, the multiple nodes can be organized in one of two modes: centralized and distributed, both of which can be located in an access network, a core network, an edge computing system, or a third-party network. A centralized training and computing architecture can be limited by large communication overhead and strict user data privacy in some cases. A distributed training and computing architecture can include several frameworks, such as distributed machine learning and federated learning.

[0004] However, communications in wireless communication systems, including communications related to AI training across multiple nodes, typically occur over non-ideal channels. Non-ideal conditions, such as electromagnetic interference, signal degradation, phase delay, fading, and other non-ideal conditions, can attenuate and / or distort communication signals and disrupt or degrade the communication capabilities of the system.

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

[0006] Furthermore, the processing power and / or availability of training data for an AI training process can vary significantly among different nodes / devices. This means that the ability of different nodes to productively participate in the AI ​​training process can vary significantly. In practice, this variation means that the training delay of an AI training process involving multiple nodes / devices, such as a distributed learning or federated learning-based AI training process, will be dominated by the node / device with the largest 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 maintains local data samples, which cannot be exchanged with other devices or servers. Federated learning is the opposite of traditional centralized machine learning techniques in that local data samples are not shared in federated learning, whereas in traditional centralized machine learning, all local datasets are uploaded to a single server.

[0008] In a federated learning-based (FL-based) AI training process, a network node / device / nodes initialize a global AI model, sample a group of user devices, and broadcast 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 updated local AI model parameters to the network device, which then aggregates the updated parameters reported by the user devices and updates the global AI model. The foregoing procedure 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 has converged 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 data sets that may not be representative of all distributions of training data observed by other user devices participating in the FL-based AI training process. That is, the training data may not be independently and identically distributed (non-iid) among devices participating in a conventional FL-based AI training process. Non-iid training data among devices has been shown to reduce 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, new protocols and signaling mechanisms are desirable that enable the implementation of new AI-enabled applications and processes while minimizing the signaling and communication overhead and delays associated with existing AI training procedures. Summary of the Invention

[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, a server node needs to collect a large amount of training data sets (e.g., gradients of client node updates) from multiple associated client nodes. Furthermore, traditional FL-based AI training processes require the server node and associated client nodes to have the same AI / ML model structure. However, data heterogeneity can be a problem in traditional FL-based AI training processes because different local nodes may support different AI / ML model structures.

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

[0013] According to a first broad aspect of the present disclosure, provided herein is a method for configuring a topology for artificial intelligence or machine learning (AI / ML) in a wireless communication network. The method according to the first broad aspect of the present disclosure may include receiving information from a node, the information including a report related to the node's AI / ML capabilities. The method according to the first broad aspect of the present disclosure may further include configuring the node based on the received information, where configuring the node includes configuring a node type of the node, the configured node type being one of a plurality of node types. For example, the plurality of node types may include Type 1, which 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, which indicates a node configured to obtain a second-type AI / ML model with a set of training data without an aggregation operation. The method according to the first broad aspect of the present disclosure may further include 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 an air interface of the wireless communication network and includes a connection between at least one Type 1 node and zero or more Type 2 nodes and / or a connection between at least two Type 1 nodes.

[0014] In some embodiments of a method 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: one or more Type 2 nodes, a second Type 1 node that provides the first Type 1 AI / ML model of the second Type 1 node to the node; or a third Type 1 node that receives the first Type 1 AI / ML model of the node; It includes at least one of the following:

[0015] In some embodiments of the method according to the first broad aspect of the disclosure, the one or more other nodes include one or more Type 2 nodes, and configuring the node includes: Configuring the node to collect each second type AI / ML model from one or more type 2 nodes.

[0016] In some embodiments of a method according to a first broad aspect of the present disclosure, 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.

[0017] In some embodiments of a method according to the first broad aspect of this disclosure, the information further includes a request by the node configured as a Type 1 node.

[0018] In some embodiments of a method according to a first broad aspect of the 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 the network devices.

[0019] In some embodiments of the method according to the first broad aspect of the disclosure, the node is a user equipment (UE), a relay, a base station (BS), a transmission / reception point (TRP), an edge device, a network system, or an integrated access backhaul (IAB) node.

[0020] According to a second broad aspect of the present disclosure, there is provided herein a method for configuring a topology for artificial intelligence or machine learning (AI / ML) in a wireless communication network. The method according to the second broad aspect of the present disclosure may include establishing an aggregation connection with a second node based on an aggregation acknowledgement message. The topology configured by the method according to the second broad aspect of the present disclosure supports AI / ML model transfer over an air interface of the wireless communication network, and includes: a connection between at least one type 1 node and zero or more type 2 nodes, or A connection between at least two type 1 nodes, and A Type 1 node is a node 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 a node configured to obtain a second-type AI / ML model with a set of training data without performing an aggregation operation.

[0021] Optionally, before establishing the aggregation connection with the second node, the method further comprises sending a message for aggregation by the first node to the second node, After sending the message for aggregation, the first node may receive an acknowledgement message from the second node.

[0022] Furthermore, the establishing step is optional.

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

[0024] In some embodiments of the method according to the second broad aspect of the disclosure, the aggregation information used by the second node includes the aggregation capability of the first node.

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

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

[0027] In some embodiments of the method according to the second broad aspect of the disclosure, the message for aggregation is an aggregation request message, and the aggregation request message includes information or an aggregation request related to a second type AI / ML model of the first node.

[0028] In some embodiments of the method according to the second broad aspect of this disclosure, the first node is a Type 2 node, and the aggregation request message includes information related to a Type 2 AI / ML model of the first node.

[0029] In some embodiments of the method according to the second broad aspect of this disclosure, the information related to the second type AI / ML model of the first node includes: Information related to the neural network of the second type AI / ML model of the first node; Information about the size of the second type AI / ML model of the first node; or Information about the complexity of the second type AI / ML model of the first node, It includes at least one of the following:

[0030] In some embodiments of the method according to the second broad aspect of this 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 according to the second broad aspect of this disclosure, the second node is one of a plurality of other nodes in a wireless communication network, and the first node can connect to only one of the plurality of other nodes, the processor-executable instructions, when executed, cause the processor to: The method further includes processor-executable instructions for causing the first node to send a respective confirmation to at least one of the plurality of other nodes informing the at least one node whether a connection to the first node was successfully established.

[0032] In some embodiments of a method 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 a method according to a second broad aspect of the 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 second node's first Type 1 AI / ML model to the first node or receives the first Type 1 AI / ML model of the first node from the first node.

[0034] In some embodiments of the method according to the second broad aspect of this disclosure, the first node is a Type 1 node that is only related to one or more other Type 1 nodes.

[0035] In some embodiments of the method according to the second broad aspect of this disclosure, transmitting the message for aggregation includes broadcasting, groupcasting, or unicasting the message for aggregation at predetermined intervals.

[0036] According to a third broad aspect of the present disclosure, provided herein is a method 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 the present disclosure may include receiving, by a first node, a message for aggregation from a second node. After receiving the message for aggregation, the first node may determine whether to connect with the second node and transmit a response to the second node indicating the determination. The method according to the third broad aspect of the present disclosure may further include selectively establishing an aggregation connection with the second node based on the determination. The topology configured by the method according to the third broad aspect of the present disclosure supports AI / ML model transfer over an air interface of the wireless communication network and includes the following: a connection between at least one type 1 node and zero or more type 2 nodes, or A connection between at least two type 1 nodes, and A Type 1 node is a node 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 a node configured to obtain a second-type AI / ML model with a set of training data without performing an aggregation operation.

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

[0038] In some embodiments of a method according to the third broad aspect of the present disclosure, the aggregation information used by the first node includes an aggregation capability of the second node.

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

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

[0041] In some embodiments of the method according to the third broad aspect of the disclosure, the message for aggregation is an aggregation request message, and the aggregation request message includes information or an aggregation request related to a second type AI / ML model of the first node.

[0042] In some embodiments of the method according to the second broad aspect of the disclosure, the second node is a Type 2 node, and the aggregation request message includes information related to a second type AI / ML model of the second node.

[0043] In some embodiments of the method according to the third broad aspect of this disclosure, the information related to the second type AI / ML model of the second node includes: Information related to the neural network of the second type AI / ML model of the second node; Information about the size of the second type AI / ML model of the second node; or Information about the complexity of the second type AI / ML model of the second node; It includes at least one of the following:

[0044] In some embodiments of the method according to a third broad aspect of the disclosure, the first node determines whether to connect with the second node based on the aggregation capabilities of the first node and information related to the second type AI / ML model of the second node.

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

[0046] In some embodiments of a method according to a third broad aspect of the disclosure, the second node can connect to only one node, and the processor-executable instructions further include processor-executable instructions that, when executed, cause the processor to receive a confirmation from the second node informing whether the connection between the second node and the first node was successfully established.

[0047] In some embodiments of a method 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 a method according to a third broad aspect of the 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 according to the third broad aspect of this disclosure, the second node is a Type 1 node that is only related to one or more other Type 1 nodes.

[0050] Corresponding instruments and apparatus for carrying out the methods are disclosed.

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

[0052] According to another aspect of the present disclosure, there is provided an apparatus including one or more units for performing any of the method aspects disclosed in this disclosure. The term "unit" is used broadly and may be referred to by any of a variety of names including, for example, a module, a component, an element, a means, etc. A unit may be implemented using hardware, software, firmware, or any combination thereof.

[0053] Certain aspects of the present disclosure enable heterogeneous AI / ML capabilities in various network devices and user devices, and can support heterogeneous AI / ML model transfer over the air interface of a wireless communication network.

[0054] According to some aspects of the present disclosure, the topology within a wireless communication network can be centrally configured by a network device (e.g., a base station (BS), a transmit and receive point (TRP)), or a network system.

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

[0056] In accordance with some aspects of the present disclosure, a topology within a wireless communication network configured by various methods described in this disclosure supports parallel AI / ML model aggregation, reduces AI / ML model routing latency, and improves AI / ML model routing robustness. [Brief explanation of the drawings]

[0057] By way of example only, reference is made to the accompanying drawings, in which exemplary embodiments of the present application are shown.

[0058] [Figure 1] 1 is a simplified schematic diagram of a communication system according to an example;

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

[0060] [Figure 3] 1 shows an example of an electronic device (ED), a terrestrial transmitting / receiving point (T-TRP), and a non-terrestrial transmitting / receiving point (NT-TRP).

[0061] [Figure 4] 1 illustrates exemplary units or modules within the device.

[0062] [Figure 5] 1 illustrates four EDs communicating with a network device in a communication system according to an embodiment of the present disclosure.

[0063] [Figure 6A] 1 illustrates an example of a neural network having multiple layers of neurons, according to an embodiment of the present disclosure.

[0064] [Figure 6B] 1 illustrates an example of a neuron that can be used as a building block of a neural network, according to an embodiment of the present disclosure.

[0065] [Figure 7] 1 shows the star topology used in conventional federated learning (FL) procedures.

[0066] [Figure 8] 1 illustrates the difference between the conventional FL-based AI / ML model training procedure and the AI / ML model learning scheme of the present disclosure.

[0067] [Figure 9A] 1 illustrates two different types of nodes used to train an AI / ML model, according to an embodiment of the present disclosure. [Figure 9B] 1 illustrates two different types of nodes used to train an AI / ML model, according to an embodiment of the present disclosure.

[0068] [Figure 10] 1 illustrates an exemplary self-organizing topology according to an embodiment of the present disclosure.

[0069] [Figure 11] 1 illustrates a configured topology supporting an exemplary AI / ML model transfer, according to an embodiment of the present disclosure.

[0070] [Figure 12] 1 illustrates an example of a procedure for establishing an aggregation connection between an aggregation node and a basic node when configuring an AI / ML topology in a wireless communication network according to an embodiment of the present disclosure.

[0071] [Figure 13] 1 illustrates an example of a 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.

[0072] [Figure 14] 10 illustrates another example of a procedure for establishing an aggregation connection between an aggregation node and a basic node when configuring an AI / ML topology in a wireless communication network according to an embodiment of the present disclosure.

[0073] [Figure 15] 10 illustrates another example of a 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.

[0074] [Figure 16] 1 illustrates an example of a configured topology that supports flexible communication between aggregation nodes, according to an embodiment of the present disclosure.

[0075] [Figure 17] 1 illustrates an example of a configured topology that supports flexible communication between aggregation nodes and basic nodes, according to an embodiment of the present disclosure.

[0076] Similar reference numbers may be used in different figures to indicate similar components. DETAILED DESCRIPTION OF THE INVENTION

[0077] In this disclosure, "data collection" refers to the process of collecting data by a network node, management entity, or user equipment (UE) for the purposes of artificial intelligence (AI) / machine learning (ML) model training, data analysis, and inference.

[0078] In this disclosure, an "AI / ML model" refers to a data-driven algorithm that applies AI / ML techniques to generate a set of outputs based on a set of inputs.

[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 to obtain a trained AI / ML model for inference.

[0080] In this disclosure, "AI / ML inference" refers to the process of using a trained AI / ML model to generate a set of outputs based on a set of inputs.

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

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

[0083] In this disclosure, "on-UE training" refers to online / offline training at the UE.

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

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

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

[0087] In this disclosure, "model transfer" refers to the delivery of an AI / ML model over an air interface, either the parameters of the model structure known at the receiving end, or a new model with the parameters. The delivery can 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 a UE to a network.

[0090] In this disclosure, "model deployment" refers to the distribution of a fully developed and tested model runtime image to the target UE / gNodeB (gNB) where inference will be performed.

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

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

[0093] In this disclosure, "model updating" refers to retraining or fine-tuning an AI / ML model via online / offline training to improve model inference performance.

[0094] For purposes of explanation, specific examples of embodiments are described in detail below with reference to the accompanying figures.

[0095] The embodiments described herein present sufficient information to implement the claimed subject matter and illustrate how to implement such subject matter. Upon reading the following description in light of the accompanying figures, one skilled in the art will understand the concepts of the claimed subject matter and will recognize applications of these concepts not specifically addressed herein. These concepts and applications should be understood to be within the scope of the disclosure and the accompanying claims.

[0096] Additionally, it will be understood that any module, component, or device disclosed herein that executes instructions can include or otherwise access one or more non-transitory 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-exhaustive list of examples of non-transitory computer / processor-readable storage media includes magnetic cassettes, magnetic tape, magnetic disk storage or other magnetic storage devices, compact disk read-only memory (CD-ROM), digital video disk or digital versatile disk (i.e., DVD), optical disk such as Blu-ray Disc™ or other optical storage devices, volatile and non-volatile, removable and non-removable media implemented in any manner or technology, random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory, or other memory technology. Such non-transitory computer / processor storage media can be part of the device or can be accessible or connectable to it. Computer / processor readable / executable instructions for implementing the applications or modules described herein may be stored or otherwise maintained by such non-transitory computer / processor readable storage media.

[0097] Examples of communication systems and devices Referring to FIG. 1 , a simplified schematic diagram of a communication system is provided as a non-limiting illustrative example. The communication system 100 includes a radio access network 120. The radio access network 120 may be a next-generation (e.g., sixth-generation (6G) or later) radio access network or a legacy (5G, 4G, 3G, 2G, etc.) radio access network. One or more communication electric devices (EDs) 110a-110j (commonly referred to as 110) may be interconnected to each other or connected to one or more network nodes (170a, 170b, collectively referred to as 170) within the radio access network 120. A core network 130 may be part of the communication system and may be dependent or 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] FIG. 2 illustrates 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 provide content, such as voice, data, video, and / or text, via broadcast, multicast, unicast, and the like. The communication system 100 may operate by sharing resources, such as carrier spectrum bandwidth, among its components. The communication system 100 may include terrestrial and / or non-terrestrial communication systems. 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, and the like. The communication system 100 may provide high availability and robustness through the cooperation of the terrestrial and non-terrestrial communication systems. For example, integrating a non-terrestrial communication system (or components thereof) with a terrestrial communication system may result in what may be considered a heterogeneous network including multiple layers. Compared to traditional communication networks, heterogeneous networks can achieve better overall performance through efficient multi-link cooperation, more flexible function sharing, and faster physical layer link switching between terrestrial and non-terrestrial networks.

[0099] The terrestrial and non-terrestrial communication systems may be considered subsystems of a communication system. In this example, communication system 100 includes electronic devices (EDs) 110a-110d (commonly referred to as EDs 110), 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. RANs 120a-120b include respective base stations (BSs) 170a-170b, collectively referred to as terrestrial transmit and receive points (T-TRPs) 170a-170b. Non-terrestrial communication network 120c includes access nodes 120c, collectively referred to as non-terrestrial transmit and receive points (NT-TRPs) 172.

[0100] Any ED 110 may alternatively or additionally be configured to interface with, access, or communicate with other T-TRPs 170a-170b and NT-TRPs 172, the Internet 150, the core network 130, the PSTN 140, other networks 160, or any combination of the above. In some examples, ED 110a may communicate uplink and / or downlink transmissions with T-TRP 170a via interface 190a. In some examples, EDs 110a, 110b, and 110d may communicate directly with each other via one or more sidelink radio interfaces 190b. In some examples, ED 110D may communicate uplink and / or downlink transmissions with NT-TRP 172 via interface 190c.

[0101] Air interfaces 190a and 190b may use similar communication technologies, such as any suitable radio access technology. For example, communication system 100 may implement one or more channel access methods, 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), over air interfaces 190a and 190b. Air interfaces 190a and 190b may also utilize other high-dimensional signal spaces, including combinations of orthogonal and / or non-orthogonal dimensions.

[0102] The wireless interface 190c may enable communication between the ED 110d and one or more NT-TRPs 172 via a wireless link or simply a link. In some examples, the link is a dedicated connection for unicast transmissions, a connection for broadcast transmissions, or a connection between a group of EDs and one or more NT-TRPs for multicast transmissions.

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

[0104] 3 shows another example of the ED 110 and the base stations 170a, 170b, and / or 170c. The ED 110 is used to connect people, things, machines, etc. The ED 110 can be widely used in various scenarios, such as cellular communication, device-to-device (D2D), vehicle-to-everything (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 grid, smart furniture, smart office, smart wearable, smart transportation, smart city, drones, robots, remote sensing, passive sensing, positioning, navigation and tracking, autonomous delivery and movement, etc.

[0105] Each ED 110 represents any end-user device suitable for wireless operation and may include, among other devices, as previously described, a user equipment / device (UE), a wireless transmit / receive unit (WTRU), a mobile station, a fixed or mobile subscriber unit, a cell phone, a station (STA), a machine type communication (MTC) device, a personal digital assistant (PDA), a smartphone, a laptop, a computer, a tablet, a wireless sensor, a consumer electronic device, a smartbook, a vehicle, an automobile, a truck, a bus, a train, or an IoT device, industrial device, or equipment (such as a communication module, modem, chip, etc.). Future generations of EDs 110 may be referred to using other terminology. Base stations 170a and 170b are T-TRPs and will hereinafter be referred to as T-TRPs 170. As also shown in FIG. 3, an NT-TRP will hereinafter be referred to as NT-TRP 1702. Each ED110 connected to a T-TRP170 and / or NT-TRP172 may be dynamically or semi-statically activated (i.e., established, activated, or enabled), deactivated (released, deactivated, or disabled), and / or configured in response to one or more of connection availability and connection need.

[0106] The ED 110 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 panels. 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 network interface controller (NIC). 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 transmitting and receiving wireless or wired signals.

[0107] The ED 110 includes at least one memory 208. The memory 208 stores instructions and data used, generated, or collected by the ED 110. 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 may include 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, on-processor cache, etc.

[0108] ED 110 further includes one or more input / output devices (not shown) or interfaces (such as a wired interface to Internet 150 in FIG. 1). The input / output devices 206 enable interaction with a user or other devices in a network. Each input / output device includes any structure suitable for providing information to or receiving information from a user, such as a speaker, microphone, keypad, keyboard, display, touch screen, etc., including network interface communications.

[0109] The ED 110 further includes a processor 210 that performs operations including operations related to preparing a transmission for uplink transmission to the NT-TRP 172 and / or the T-TRP 170, operations related to processing for a downlink transmission received from the NT-TRP 172 and / or the T-TRP 170, and / or operations related to processing a sidelink transmission to or from another ED 110. Processing operations related to preparing a transmission for uplink transmission can include operations such as encoding, modulation, transmit beamforming, and generating symbols for transmission. Processing operations related to processing a downlink transmission can include operations such as receive beamforming, demodulation, and decoding received symbols. Depending on the embodiment, the downlink transmission is received by the receiver 203, possibly using receive beamforming, and the processor 210 can extract signaling from the downlink transmission (e.g., by detecting and / or decoding the signaling). An example of signaling may be a reference signal transmitted by the NT-TRP 172 and / or the T-TRP 170. In some embodiments, processor 276 performs transmit beamforming and / or receive beamforming based on beam direction instructions, e.g., beam angle information (BAI), received from T-TRP 170. In some embodiments, processor 210 may perform operations related to network access (e.g., initial access) and / or downlink synchronization, e.g., detecting synchronization sequences, decoding and obtaining system information, etc. In some embodiments, processor 210 may perform channel estimation, e.g., using reference signals received from NT-TRP 172 and / or T-TRP 170.

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

[0111] The processor 210 and the processing components of the transmitter 201 and receiver 203 may each be implemented by one or more of the same or different processors configured to execute instructions stored in a 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 special purpose circuitry, 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-TRP 170 may be known by other names such as a base station, base transceiver station (BTS), radio base station, network node, network equipment, network side equipment, transmitting / receiving 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 wireless router, relay station, remote radio head, terrestrial node, terrestrial network equipment, or terrestrial 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-TRP 170 may also be a macro BS, pico BS, relay node, donor node, etc., or a combination thereof. The T-TRP 170 may refer to such a device or a device within such a device (such as a communication module, modem, chip, etc.).

[0113] In some embodiments, the components of the T-TRP 170 may be distributed. For example, some of the modules of the T-TRP 170 may be located remotely from the facility housing the T-TRP 170's antenna and may be coupled to the facility housing the antenna via a communications link (not shown), also known as fronthaul, such as a common public radio interface (CPRI). Thus, in some embodiments, the term T-TRP 170 may refer to a network-side module that performs processing operations such as determining the location of the ED 110, resource allocation (scheduling), message generation, and encoding / decoding, and is not necessarily part of the facility housing the T-TRP 170's antenna. Modules may also be coupled to other T-TRPs. In some embodiments, the T-TRP 170 may actually be multiple T-TRPs operating together to provide service to the ED 110, for example, via coordinated multipoint transmission.

[0114] The T-TRP 170 includes at least one receiver 254 and at least one transmitter 252 coupled to one or more antennas 256. Only one antenna 256 is shown. Alternatively, one, some, or all of the antennas may be panels. The transmitter 252 and receiver 254 may be integrated as a transceiver. The T-TRP 170 further includes a processor 260 that performs operations including operations related to preparing a transmission for downlink transmission to the ED 110, processing an uplink transmission received from the ED 110, preparing a transmission for backhaul transmission to the NT-TRP 172, and processing a transmission received via the backhaul from the NT-TRP 172. Processing operations related to preparing a transmission for downlink or backhaul transmission may include operations such as encoding, modulation, precoding (e.g., MIMO precoding), transmit beamforming, and generating symbols for transmission. Processing operations related to processing a transmission received on the uplink or via the backhaul may include operations such as receive beamforming, demodulation, and decoding of received symbols. The processor 260 may also perform operations related to network access (e.g., initial access) and / or downlink synchronization, such as generating synchronization signal block (SSB) content, generating system information, etc. In some embodiments, the processor 260 also generates beam direction indications, such as a BAI, that may be scheduled for transmission by the scheduler 253. The processor 260 performs other network-side processing operations described herein, such as determining the location of the ED 110 and determining the placement location of the NT-TRP 172. In some embodiments, the processor 260 may generate signaling, for example, to configure 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 "signaling" as used herein may alternatively be referred to as control signaling.Dynamic signaling may be transmitted on a control channel, e.g., the physical downlink control channel (PDCCH), and static or semi-static higher layer signaling may be included in packets transmitted on a data channel, e.g., the physical downlink shared channel (PDSCH).

[0115] The scheduler 253 may be coupled to the processor 260. The scheduler 253 may be included within the T-TRP 170 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 grants") resources. The T-TRP 170 further includes a memory 258 for storing information and data. The memory 258 stores instructions and data used, generated, or collected by the T-TRP 170. For example, the 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 the receiver 254. Also, although not shown, the processor 260 may implement the scheduler 253. Although not shown, the 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 of the same or different processors configured to execute instructions stored in a memory, such as 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 an FPGA, GPU, or ASIC.

[0118] Although the NT-TRP 172 is illustrated as a drone for example only, the NT-TRP 172 may be implemented in any suitable non-terrestrial form. The NT-TRP 172 may also be known by other names, such as a non-terrestrial node, a non-terrestrial network device, or a non-terrestrial base station, in some implementations. The NT-TRP 172 includes a transmitter 272 and a receiver 274 coupled to one or more antennas 280. Only one antenna 280 is shown. Alternatively, one, some, or all of the antennas may be panels. The transmitter 272 and the receiver 274 may be integrated as a transceiver. The NT-TRP 172 further includes a processor 276 that performs operations, including operations related to preparing transmissions for downlink transmission to the ED 110, processing uplink transmissions received from the ED 110, preparing transmissions for backhaul transmission to the T-TRP 170, and processing transmissions received via the backhaul from the T-TRP 170. Processing operations related to preparing a transmission for downlink or backhaul transmission can include operations such as encoding, modulation, precoding (e.g., MIMO precoding), transmit beamforming, and generating symbols for transmission. Processing operations related to processing a transmission received on the uplink or over the backhaul can 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, for example, to configure one or more parameters of the ED 110. In some embodiments, the NT-TRP 172 implements physical layer processing but does not implement higher layer functionality, such as functionality at the Medium Access Control (MAC) or Radio Link Control (RLC) layers. This is by way of example only; more generally, the NT-TRP 172 can implement higher layer functionality in addition to physical layer processing.

[0119] The NT-TRP 172 further includes a memory 278 for storing information and data. Although not shown, the processor 276 may form part of the transmitter 272 and / or the 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 of the same or different processors configured to execute instructions stored in a memory, such as 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 programmed FPGA, GPU, or ASIC. In some embodiments, the NT-TRP 172 may actually be multiple NT-TRPs operating together to provide service to the ED 110, for example, via coordinated multipoint transmission.

[0121] Note that "TRP" as used herein may refer to T-TRP or NT-TRP.

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

[0123] One or more steps of the method of the embodiments provided herein may be performed by a corresponding unit or module according to FIG. 4. FIG. 4 illustrates units or modules within a device such as the ED 110, the T-TRP 170, or the NT-TRP 172. For example, a signal may be transmitted by a transmitting unit or a transmitting module. For example, a signal may be transmitted by a transmitting unit or a transmitting module. A signal may be received by a receiving unit or a receiving module. A signal may be processed by a processing unit or a 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 execute software, or a combination thereof. For example, one or more of the units or modules may be an integrated circuit such as a programmed FPGA, a GPU, or an ASIC. It will be understood that when modules are implemented using software, for example, for execution by a processor, they may be retrieved individually or together for processing by the processor, in whole or in part, in single or multiple instances, as needed, and the modules themselves may include instructions for further deployment and instantiation.

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

[0125] Control signaling is described herein in some embodiments. Control signaling may alternatively 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, on a control channel. An example of dynamically indicated control signaling is information transmitted in physical layer control signaling, such as downlink control information (DCI). Control signaling may be semi-statically indicated, for example, in RRC signaling or a MAC control element (CE). A dynamic indication may be an indication in a lower layer, such as physical layer / Layer 1 signaling (such as DCI), rather than in a higher layer (e.g., other than RRC signaling or MAC CE). A semi-static indication may be an indication in semi-static signaling. Semi-static signaling, as used herein, may refer to non-dynamic signaling, e.g., higher layer signaling, RRC signaling, and / or MAC CE. Dynamic signaling, as used herein, may refer to dynamic signaling, e.g., physical layer control signaling transmitted at the physical layer, such as DCI.

[0126] A radio interface generally includes numerous 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 radio interface may include one or more components that define one or more waveforms, frame structures, multiple access schemes, protocols, coding schemes, and / or modulation schemes for conveying information (e.g., data) over a wireless communication link. A wireless communication link may support a link between a radio access network and a user equipment (e.g., a "Uu" link), and / or a wireless communication link may support a device-to-device link (e.g., a "sidelink"), such as between two user equipments, 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 such components: The waveform component can specify the shape and format of the transmitted signal. Waveform options can include orthogonal and non-orthogonal multiple access waveforms. Non-limiting examples of such waveform options include Orthogonal Frequency Division Multiplexing (OFDM), Filtered OFDM (f-OFDM), Time Windowed OFDM, Filter Bank Multicarrier (FBMC), Universal Filtered Multicarrier (UFMC), Generalized Frequency Division Multiplexing (GFDM), Wavelet Packet Modulation (WPM), Faster Than Nyquist (FTN) waveforms, and low Peak to Average Power Ratio Waveforms (low PAPR WFs). The frame structure component may specify the composition of a frame or group of frames. The frame structure component may indicate one or more of the following: time, frequency, pilot signature, code, or other parameters of the frame or group of frames. Further details of the frame structure are described below. The multiple access method component can specify multiple access technology options, including techniques that define how communication devices share a common physical channel, such as 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 (SCA). Further multiple access technology options may include scheduled access and unscheduled access, also known as permissionless access, non-orthogonal multiple access and orthogonal multiple access, for example, via dedicated channel resources (e.g., not shared among multiple communication devices), contention-based shared channel resources and non-contention-based shared channel resources, and cognitive radio-based access. A hybrid automatic repeat request (HARQ) protocol component may specify how transmission and / or retransmission occurs. Non-limiting examples of transmission and / or retransmission mechanism options include those that specify a scheduled data pipe size, a signaling mechanism for transmission and / or retransmission, and a retransmission mechanism. The coding and modulation component 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-limiting examples of coding options include turbo trellis codes, turbo product codes, fountain codes, low-density parity check codes, and polar codes. Modulation can simply represent a constellation (e.g., including modulation technique and order), 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 limited parameters or modes of the radio interface, such as cyclic prefix (CP) length or multiple input multiple output (MIMO) mode, may be configurable. 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 one example, the flexibility of the configurable radio interface provided by scalable number theory and symbol duration may enable optimization of transmission parameters for different spectrum bands and different services / devices. As another example, the unified radio interface may be self-contained in the 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 The frame structure is a feature of the wireless communication physical layer that defines the time-domain signal transmission structure, e.g., allows timing reference and timing alignment 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 referred to as the radio frame structure.

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

[0130] An example of a frame structure is that of long-term evolution (LTE), which has the following specifications: each frame has a duration of 10 ms; each frame has 10 subframes, each with a duration of 1 ms; each subframe has two slots, each with a duration of 0.5 ms; each slot is for the transmission of seven OFDM symbols (assuming a normal CP); each OFDM symbol has a symbol duration and a specific bandwidth (or fractional bandwidth or bandwidth partition) related to the number of subcarriers and the subcarrier spacing; the frame structure is based on OFDM waveform parameters such as subcarrier spacing and CP length (CP has options for fixed or limited length); and the switching gap between uplink and downlink in TDD must be an integer multiple of the OFDM symbol duration.

[0131] Another example of a frame structure is the new radio (NR) frame structure, which has the following specifications: Multiple subcarrier spacings are supported, each corresponding to a different number theory; the frame structure depends on the number theory, but in all cases the frame length is set to 10 ms, consisting of 10 subframes of 1 ms each; a slot is defined as 14 OFDM symbols, and the slot length depends on the number theory. For example, the NR frame structure with normal CP 15 kHz subcarrier spacing ("number theory 1") differs from the NR frame structure with normal CP 30 kHz subcarrier spacing ("number theory 2"); the slot length for 15 kHz subcarrier spacing is 1 ms, and the slot length for 30 kHz subcarrier spacing is 0.5 ms. The NR frame structure is more flexible than the LTE frame structure.

[0132] Another example of a frame structure is a flexible frame structure, e.g., for use in 6G networks and beyond. In a flexible frame structure, a symbol block may be defined as the smallest period that can be scheduled in the flexible frame structure. A symbol block may be a transmission unit having an optional redundant portion (e.g., a CP portion) and an information (e.g., data) portion. An OFDM symbol is an example of a symbol block. A symbol block may alternatively be referred to as a symbol. Embodiments of a flexible frame structure include different parameters that may be configurable, such as, e.g., frame length, subframe length, symbol block length, etc. A non-exhaustive list of possible configurable parameters in some embodiments of a flexible frame structure includes the following: (1) Frame: The frame length need not be limited to 10 milliseconds; it is configurable and can change over time. In some embodiments, each frame includes one or more downlink synchronization channels and / or one or more downlink broadcast channels, and each synchronization channel and / or broadcast channel can be transmitted in a different direction by different beamforming. The frame length is a number of 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 can be set to 5 milliseconds for the autonomous vehicle application. As another example, a residential smart meter may not require fast initial access, in which case the frame length can be set as 20 ms for the smart meter application. (2) Subframe Duration: Subframes may or may not be defined in a flexible frame structure, depending on the implementation. For example, a frame may be defined to include slots but not subframes. In frames where subframes are defined, the duration of the subframes may be configurable, e.g., in the case of time-domain alignment. For example, subframes may be configured to have lengths of 0.1 ms, 0.2 ms, 0.5 ms, 1 ms, 2 ms, or 5 ms, etc. In some embodiments, if subframes are not required in a particular scenario, the length of the subframe may be defined to be the same as the length of the frame, or may not be defined. (3) Slot Configuration: Slots may or may not be defined in a flexible frame structure, depending on the implementation. 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 a group of UEs. In this case, slot configuration information may 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 slot configuration information may be transmitted on a UE-specific control channel. In some embodiments, slot configuration signaling may be transmitted together with frame configuration signaling and / or subframe configuration signaling. In other embodiments, the slot configuration may be transmitted independently from frame configuration signaling and / or subframe configuration signaling. In general, the slot configuration may be system-wide, base station-wide, UE group-wide, or UE-specific. (4) Subcarrier spacing (SCS): SCS is a parameter of scalable number theory, allowing for a possible range of SCS from 15 kHz to 480 kHz. SCS may vary with the frequency of the spectrum and / or the 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 configured independently from the SCS of symbols in the transmit frame structure. The SCS in the receive frames may be different from the SCS in the transmit frames. In some examples, the SCS of each transmit frame may be half the SCS of each receive frame. When the SCS between the receive and transmit frames differs, for example, if a more flexible symbol period is implemented using an inverse discrete Fourier transform (IDFT) instead of a fast Fourier transform (FFT), the difference does not necessarily need to be scaled by a factor of two. Additional example frame structures may be used with different SCSs. (5) Flexible Transmission Duration of Basic Transmission Unit: A basic transmission unit may be a symbol block (alternatively referred to as a symbol) that generally includes a redundant portion (referred to as a 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 may be flexible within a frame, and the CP length may dynamically change, in some cases, from frame to frame group, from subframe to subslot, or from scheduling to scheduling. The information (e.g., data) portion may be flexible and configurable. Another possible parameter related to a symbol block that may 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 time period. As another example, the symbol block length may be adjusted to fit the available time period within a frame. (6) Flexible Switch 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. Between each uplink and downlink portion, there may be a gap called a switching gap. The switching gap length (duration) may be configurable. The switching gap duration may be fixed within a frame or may be flexible within a frame, and the switching gap duration may change dynamically from frame to frame, frame group to frame, subframe to subframe, slot to slot, or scheduling to scheduling.

[0133] Cell / Carrier / Bandwidth Part (BWP) / Occupied Bandwidth A device, such as a base station, can provide coverage throughout a cell. Wireless communication with devices can occur over one or more carrier frequencies. A carrier frequency is referred to as a carrier. A carrier is sometimes referred to as a component carrier (CC). A carrier can be characterized by its bandwidth and a reference frequency, e.g., the carrier's center or lowest or highest frequency. A carrier may be on a licensed or unlicensed spectrum. Wireless communication with devices can also or alternatively occur over one or more bandwidth portions (BWPs). For example, a carrier may have one or more BWPs. More generally, wireless communication with devices can occur over a spectrum. A spectrum may include 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. By way of 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 alternatively or additionally include one or more sidelink resources, including sidelink transmission and reception resources.

[0135] A BWP is a set of contiguous or non-contiguous frequency subcarriers on a carrier, or a set of contiguous or non-contiguous frequency subcarriers on multiple carriers, which may have one or more carriers.

[0136] In some embodiments, a carrier can have one or more BWPs; for example, a carrier can have a 20 MHz bandwidth and include one BWP, or a carrier can have an 80 MHz bandwidth and include two adjacent, contiguous BWPs, etc. In other embodiments, a BWP can have one or more carriers; for example, a BWP can have a 40 MHz bandwidth and include two adjacent, contiguous carriers, each having a 20 MHz bandwidth. In some embodiments, a BWP can include non-contiguous spectrum resources including multiple non-contiguous carriers, where a first carrier of the non-contiguous carriers may be in the mmW band, a second carrier may be in a low band (e.g., the 2 GHz band), a third carrier (if present) may be in the THz band, and a fourth carrier (if present) may be in the visible light band. Resources within a carrier belonging to a BWP can be contiguous or non-contiguous. In some embodiments, a BWP has non-contiguous spectrum resources on a single carrier.

[0137] Wireless communication may occur over an occupied bandwidth, which may be defined as the width of a frequency band below a lower frequency limit and above an upper frequency limit, each emitting an average power equal to a specified percentage β / 2 of the total average transmitted power, e.g., β / 2=0.5%.

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

[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 device functions are expected to become increasingly diverse. Furthermore, many new applications / use cases are expected to emerge with more diverse quality of service requirements than those of 5G. These may result in new key performance indicators (KPIs) for future wireless networks (e.g., 6G networks) that will be extremely challenging. AI techniques, such as ML techniques (e.g., deep learning), are being introduced into telecommunications applications with the goal of improving system performance and efficiency.

[0140] Additionally, antenna and bandwidth capabilities continue to advance, enabling more and / or better communications over wireless links. Furthermore, advances continue in the areas of computer architecture and computing power, such as the introduction of general-purpose graphics processing units (GP-GPUs). Future generations of communications devices may have greater computing and / or communications capabilities than previous generations, thereby enabling the adoption of AI to implement air interface components. Future generations of networks may also have access to more accurate and / or new information (compared to previous networks) that may form the basis of input to AI models. For example, the physical speed / velocity at which the device is traveling, the device's link budget, the device's channel conditions, the capabilities and / or supported service types of one or more devices, sensing information, and / or positioning information. To obtain sensing information, the TRP can transmit a signal to a target object (e.g., a suspect UE), and based on the signal reflections, the TRP or another network device calculates the angle (for device beamforming), the device's distance from the TRP, and / or Doppler shift information. Positioning information, sometimes referred to as localization, can be obtained in a variety of ways, including positioning reports from the UE (e.g., reporting the UE's GPS coordinates), using positioning reference signals (PRS), using the sensing described above, and tracking and / or predicting the device's location.

[0141] AI techniques (including ML techniques) can be applied to communications, including AI-based communications at the physical layer and / or AI-based communications at the MAC layer. At the physical layer, AI communications can aim to optimize component design and / or improve algorithm performance. For example, AI can be applied in connection with implementing 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. At the MAC layer, AI communications can aim to utilize AI capabilities to learn, predict, and / or make decisions to solve complex optimization problems with possible better strategies and / or optimal solutions. For example, functions at the MAC layer can be optimized. 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 strategies, and / or intelligent transmit / receive mode adaptation.

[0142] In some embodiments, the AI ​​architecture may include multiple nodes. The multiple nodes may be organized in one of two modes: centralized and distributed, both of which may be located in the access network, core network, edge computing system, or third-party network. A centralized training and computing architecture may be limited by significant communication overhead and strict user data privacy. A distributed training and computing architecture may include several frameworks, such as distributed machine learning and federated 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. Novel protocols and signaling mechanisms are desired that allow personalized AI techniques to minimize signaling overhead and maximize system-wide spectral efficiency while personalizing corresponding interface links with customized parameters to meet specific requirements.

[0143] In some embodiments herein, new protocols and signaling mechanisms are provided for operating in and switching between different modes of operation for AI training, including between a training mode and a normal operating mode, and for measurement and feedback to accommodate different measurements and information that may need to be fed back depending on the implementation.

[0144] AI Training 1 and 2, embodiments of the present disclosure can be used to implement AI training involving two or more communication devices in communication system 100. For example, FIG. 5 illustrates four EDs in communication with network device 452 in communication system 100, according to one embodiment. The four EDs are each shown as a different UE and will hereafter be 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., the radio access network 120). The network device 452 can be located in an access network, a core network, an edge computing system, or a third-party network, depending on the implementation. The network device 452 can be (or be part of) a T-TRP or a server. In one example, the network device 452 can be (or be implemented in) the T-TRP 170 or the NT-TRP 172. In another example, the network device 452 can be a T-TRP controller and / or an NT-TRP controller that can manage the T-TRP 170 or the NT-TRP 172. In some embodiments, the components of the network device 452 may be distributed. The 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 serves the UEs 402, 404, 406, and 408. Alternatively, the UEs 402, 404, 406, and 408 may communicate with the network device 452 through one or more intermediate components, such as via a T-TRP and / or an NT-TRP. For example, the network device 452 may transmit and / or receive information (e.g., control signaling, data, training sequences, etc.) to and from one or more of the UEs 402, 404, 406, and 408 via backhaul links and wireless channels intervening between the network device 452 and the UEs 402, 404, 406, and 408.

[0146] Each UE 402, 404, 406, and 408 includes a respective processor 210, memory 208, transmitter 201, receiver 203, and one or more antennas 204 (or alternatively, a panel), as described above. Only the processor 210, memory 208, transmitter 201, receiver 203, and antenna 204 of UE 402 are shown for simplicity, although the other UEs 404, 406, and 408 also include the same respective components.

[0147] For each UE 402, 404, 406, and 408, the communication link between that UE and each TRP in the network is an air interface. The air interface generally includes a number of components and associated parameters that collectively specify how transmissions are sent and / or received over a wireless medium.

[0148] The processor 210 of the UE in FIG. 5 implements one or more air interface components on the UE side. The air interface components configure and / or implement transmission and / or reception over the air interface. Examples of air interface components are described herein. The air interface components may reside in a physical layer, such as a channel encoder (or decoder) that implements a coding component of the air interface for the UE, and / or a modulator (or demodulator) that implements a modulation component of the air interface for the UE, and / or a waveform generator that implements a waveform component of the air interface for the UE. The air interface components may reside in or be part of a higher layer, such as a MAC layer, such as a module that implements channel estimation / tracking and / or a module that implements a retransmission protocol (e.g., that implements a HARQ protocol component of the air interface of the UE). 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, a memory 456, and an input / output device 458. The processor 454 performs or instructs other network devices (e.g., a T-TRP) to perform one or more radio interface components on the network side. The radio interface components may be implemented 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] Processor 454 may be implemented by one or more of the same or different processors configured to execute instructions stored in a memory (e.g., in memory 456). Alternatively, some or all of processor 454 may be implemented using dedicated circuitry, such as a programmed FPGA, GPU, or ASIC. Memory 456 can be implemented with volatile and / or non-volatile storage. Any suitable type of memory may be used, such as RAM, ROM, a hard disk, an optical disk, an on-processor cache, etc.

[0151] Input / output devices 458 enable interaction with other devices by receiving (input) and transmitting (output) information. In some embodiments, input / output devices 458 may be implemented by a transmitter and / or a receiver (or transceiver), and / or one or more interfaces (e.g., a wired interface to an internal network, the Internet, etc.). In some implementations, input / output devices 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, etc., depending on the implementation.

[0152] The network device 452 and the UE 402 are capable of implementing one or more AI-enabled processes. In particular, in the embodiment of FIG. 5, the network device 452 and the UE 402 each include an ML module 410 and 460. The ML module 410 is implemented by the processor 210 of the UE 402, and the ML module 460 is implemented by the processor 454 of the network device 452; thus, in FIG. 5, the ML module 410 is shown within the processor 210, and the ML module 460 is shown with the processor 454. The 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 communications link between the network and the UE 402.

[0153] The ML modules 410 and 460 can be implemented using an AI model. The term AI model can refer to a computer algorithm configured to accept defined input data and output defined inference data, and the parameters (e.g., weights) of the algorithm can be updated and optimized through training (e.g., using a training dataset or using actual collected data). The AI ​​model 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 various neural network architectures (e.g., autoencoders, generative adversarial networks, etc.). Various techniques can be used to train the AI ​​model to update and optimize its parameters. For example, backpropagation is a common technique for training DNNs, in which a loss function is calculated between the inference data generated by the DNN and some target output (e.g., ground truth data). The gradient of the loss function is calculated with respect to the parameters of the DNN, and the calculated gradient is used to update the parameters with the goal of minimizing the loss function (e.g., using a gradient descent algorithm).

[0154] In some embodiments, the AI ​​model encompasses a neural network used in machine learning. A neural network is composed of multiple computational units (also called neurons) arranged in one or more layers. The process of receiving input at the input layer and generating output at the output layer is sometimes called forward propagation. In forward propagation, each layer receives an input (the input can have any suitable data format, such as a vector, matrix, or multidimensional array) and performs a computation to generate an output (the output can have different dimensions than the input). The computation performed by a layer typically involves applying (e.g., multiplying) the input to a set of weights. Except for the first layer (i.e., the input layer) of a neural network, the input to each layer is the output of the previous layer. A neural network can include one or more layers, called internal or hidden layers, between the first layer (i.e., the input layer) and the last layer (i.e., the output layer). For example, FIG. 6A shows an example of a neural network 600 including an input layer, an output layer, and two hidden layers. In this example, it can be seen that the output of each of the three neurons in the input layer of the neural network 600 is included in the input vector to each of the three neurons in the first hidden layer. Similarly, the outputs of each of the three neurons in the first hidden layer are included in the input vectors to each of the three neurons in the second hidden layer, and the outputs of each of the three neurons in the second hidden layer are included in the input vectors 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 at 650 in FIG. 6A. FIG. 6B shows an example of a neuron 650 that can be used as a building block of neural network 600. As shown in FIG. 6B, in this example, neuron 650 takes a vector as input and performs a dot product with the associated vector of weights. The neuron's final output, z, 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 various functions performed by each layer).

[0155] Neural networks are trained to optimize their parameters (e.g., weights). This optimization is performed in an automated manner and is sometimes referred to as machine learning. Training a neural network involves forward-propagating input data samples to generate output values ​​(e.g., called predicted or estimated output values) and comparing the generated output values ​​with 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 neural networks. Backpropagation is used to adjust (also called update) the values ​​of parameters (e.g., weights) in a neural network so that the calculated loss function becomes smaller. Backpropagation calculates the gradient of the loss function with respect to the parameters being optimized and updates the parameters to reduce the loss function using a gradient algorithm (e.g., gradient descent). Backpropagation is performed iteratively over multiple iterations until the loss function converges or is minimized. After training conditions are met (e.g., the loss function has converged or a predefined number of training iterations have been performed), the neural network is considered trained. The trained neural network may be deployed (or run) to generate output data inferred from input data. In some embodiments, training of the neural network may continue even after the neural network has been deployed, such that the neural network's parameters are iteratively updated with the latest training data.

[0156] Referring again to FIG. 5 , in some embodiments, the UE 402 and the network device 452 may exchange information for training purposes. The information exchanged between the UE 402 and the network device 452 is implementation-specific and may not have a human-understandable meaning (e.g., it may be intermediate data generated during the execution of an ML algorithm). Additionally or alternatively, the exchanged information 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 indicate to the UE 402 one or more parameters to be used by the ML module 410 implemented in the UE 402. As an example, the network device 452 may send or indicate updated neural network weights implemented in a neural network executed by the ML module 410 on the UE side to attempt to optimize one or more aspects of the modulation and / or coding used for communications between the UE 402 and the T-TRP or NT-TRP.

[0157] In some embodiments, the UE 402 may implement the AI ​​itself, e.g., perform learning, while in other embodiments, the UE 402 may not perform the learning itself but may instead work in conjunction with a network-side AI implementation, e.g., by receiving from the network the configuration of the AI ​​model (e.g., a neural network or other ML algorithm) implemented by the ML module 410 and / or by assisting other devices (e.g., network devices or other AI-enabled UEs) in training the AI ​​model (e.g., a neural network or other ML algorithm) by providing requested measurements or observations. For example, in some embodiments, the UE 402 may not itself implement learning or training, but may instead receive trained configuration information for the ML model determined by the network device 452 and execute the model.

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

[0159] AI can be used to AI-enable various processes, such as link adaptation, for example, by implementing AI models as described above. To achieve AI-enablement processes according to embodiments of the present disclosure, some 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 FIG. 5 , in the case of wireless federated learning (FL), the network device 452 may initialize a 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 FIG. 5 , and broadcast the parameters of the global AI / ML model to the UEs. Each of the UEs 402, 404, 406, and 408 may then 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. Each of the UEs 402, 404, 406, and 408 may then report the parameters of its updated local AI / ML model to the network device 452. The network device 452 may then aggregate the updated parameters reported from the UEs 402, 404, 406, and 408 and update the global AI / ML model. The foregoing procedure is one iteration of the FL-based AI / ML model training procedure. The network devices and UEs 402, 404, 406, and 408 typically perform multiple iterations until the AI / ML model has converged sufficiently to meet one or more training goals / criteria and the AI / ML model is complete.

[0161] FIG. 7 shows an example of a star topology used in a conventional FL procedure. The star topology shown in FIG. 7 includes a master node / device 700 (e.g., a server, TRP, or BS) and several client nodes / devices 701-708. The master node 700 initializes a 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. The client nodes 701-708 then initialize their 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. The client nodes 701-708 then report their 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] Such a conventional FL-based AI / ML model training procedure has several limitations. In the star topology used in the FL-based AI / ML model training procedure, the master node 700 needs to collect a large amount of training data set (e.g., gradients of client node updates) from each client node 701-708. In the conventional FL-based AI training procedure, the AI / ML model structures (e.g., master AI / ML model and local AI / ML model) must be the same. However, because different client nodes may support different AI / ML model structures, data heterogeneity can be a problem in the conventional FL-based AI training procedure.

[0163] Aspects of the present disclosure provide solutions to overcome the aforementioned limitations, e.g., 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 transfer over the air interface of the wireless communication network. The methods and apparatuses presented in this disclosure can be implemented and deployed taking into account the computational capabilities of each network node and the potential scale-in / scale-out for AI / ML model training and inference.

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

[0165] As shown in FIG. 8 , in the FL-based AI / ML model training scheme, the master model 810 does not move. Instead, a large amount of training data set 815 (e.g., client node update gradients) may be transferred from the client node 811 a to the master node / device 811, which 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 data set 815 follows the AI / ML model 810. In contrast, in the AI / ML model learning scheme of the present disclosure, the AI / ML model 820 follows the training data set that remains on 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 runs locally on each node / device 821.

[0166] In FL-based AI / ML model training, there is only one master node / device 811 that collects AI / ML models and aggregates the collected 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 the present disclosure, any node in the topology can collect AI / ML models and aggregate the collected AI / ML models, so there may be multiple aggregation nodes (e.g., node 821). In practice, unlike FL-based AI / ML model training in which nodes 811 and 811a are arranged in a star topology, multiple nodes / devices 821 are arranged in a self-organizing topology as shown in FIG. 8.

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

[0168] In some embodiments, any node in the network may be a node (e.g., node 821 in FIG. 8 ) that collects one or more AI / ML models and aggregates the collected AI / ML models to generate a new or updated AI / ML model. 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 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 below or elsewhere in this disclosure as an aggregated AI / ML node or an aggregating node. Each aggregating node (or Type 1 node) may be communicatively and / or operatively connected or associated with another aggregating node (or Type 1 node). Each aggregating node (or Type 1 node) may be communicatively and / or operatively 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 second Type AI / ML model) on a set of training data (e.g., local data) without performing an aggregation operation. Type 2 nodes may also be referred to below or elsewhere in this disclosure as basic AI / ML nodes or base nodes. In some embodiments, some aggregate 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 aggregate nodes (or Type 1 nodes)).

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

[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, for example, based on one or more AI / ML models collected by the aggregation node. The first-type AI / ML model may be referred to below, or elsewhere in this disclosure, as a common AI / ML model. A second-type AI / ML model is an AI / ML model trained by a basic AI / ML node without aggregation operations. The second-type AI / ML model may be referred to below, or elsewhere in this disclosure, as a local AI / ML model. Note that the common AI / ML model and the local AI / ML model may have different names. Those skilled in the art will readily understand that even if an AI / ML model is referred to in a different manner, if the AI / ML model has essentially similar characteristics to the common AI / ML model or the local AI / ML model, or is generated / trained / updated in an essentially similar manner as the common AI / ML model or the local AI / ML model as described herein, the AI / ML model should be considered a common AI / ML model or a local AI / ML model.

[0171] Aspects of the present disclosure provide a method for configuring or discovering a topology 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 a first mode, a network device (e.g., a BS or TRP) configures a topology that supports heterogeneous AI / ML model transfer over the air interface of the wireless communication network. Such configuration may be referred to as centralized configuration. In a second mode, the network device (e.g., a BS or TRP) and / or user devices may autonomously configure the topology in the wireless communication network. In some embodiments, an aggregation node declares its aggregation capabilities to other nodes. To do so, the aggregation node may transmit a discovery message containing aggregation information used by other nodes for discovery of the aggregation node. In some embodiments, a node transmits an aggregation request message (e.g., an aggregation request) to other nodes. The node transmitting the aggregation request message may be an aggregation node or a basic node. The aggregation request message may include an aggregation request or solicited aggregation information that other nodes may be interested in discovering.

[0172] In some embodiments, the topology can be configured to allow for the aggregation of AI / ML models to be processed 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 model and the transmitted model are not the same entity, but can have the same or different data, information, and / or neural network structures. The received and transmitted AI / ML models can include multiple parameters, such as one or more parameters associated with a graph model, a parameter model, a model table, a model algorithm, or a database.

[0174] Nodes in a network can be categorized into two node types for training AI / ML models. Figures 9A and 9B illustrate two different types of nodes used to train AI / ML models, according to embodiments of the present disclosure. Figure 9A illustrates one type of node, which is a Type 2 node or basic AI / ML node or base node, and Figure 9B illustrates another type of node, which is a Type 1 node or aggregated AI / ML node or aggregate node. Type 2 / base nodes may also be referred to as local AI / ML nodes or local nodes.

[0175] A basic AI / ML node can receive one or more common AI / ML models from other nodes (e.g., aggregation nodes) and train its own (customized) AI / ML model. The customized AI / ML model may be referred to as a local AI / ML model. The basic node trains its local AI / ML model using its own AI / ML model training algorithm with the assistance of the (current) common AI / ML model (e.g., information regarding distillation and / or expansion) received from one or more aggregation nodes. In some embodiments, the neural network (NN) structure of the local AI / ML model may be the same as the NN structure of at least a portion of the received common AI / ML model. In some embodiments, the NN structure of the local AI / ML model may be different from the NN structure of all of the received common AI / ML model. Once training of the local AI / ML model is complete, the basic node transmits the local AI / ML model to one or more other nodes (e.g., 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 may include one or more local AI / ML models from associated base nodes and / or one or more common AI / ML models from other aggregation nodes. While the aggregated AI / ML node shown in FIG. 9B receives local AI / ML models, in some embodiments, the aggregation node may not receive local AI / ML models from base nodes (i.e., it receives only 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 model, local AI / ML models) and generates a new or updated common AI / ML model. The aggregation node then transmits 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 the local AI / ML models involved in the AI / ML model training 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 six-layer deep neural network (DNN) model, the first local AI / ML model may be a four-layer DNN model, the second local AI / ML model may be an eight-layer DNN model, and the third local AI / ML model may be a convolutional neural network (CNN). The NN structures can be considered as the model structures of the AI / ML models.

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

[0180] Any network appliance / device can act as an aggregation node in a network. An aggregation node may be communicatively and / or operably connected to one or more basic nodes. Such connection may indicate that the aggregation node collects local AI / ML models from associated basic nodes. However, it should be noted that in some embodiments, some aggregation nodes may not be communicatively and / or operably connected to any basic nodes. Such aggregation nodes may not collect local AI / ML models, but may simply receive common AI / ML models from other aggregation nodes.

[0181] It should be noted that any aggregation node and / or any basic node may be, for example, a UE, a relay, a base station (BS), a transmission / reception point (TRP), an edge device, an edge computing system, or a network system.

[0182] FIG. 10 illustrates an exemplary self-organizing topology 1000 according to an embodiment of the present disclosure. The topology 1000 includes connections between aggregation nodes and connections between aggregation nodes and basic nodes. With respect to connections between aggregation nodes, each of aggregation nodes 901 through 908 is communicatively and operatively connected to its adjacent aggregation nodes. For example, as shown in FIG. 10, aggregation node 901 is communicatively and operatively connected to aggregation nodes 902 and 908. In this example, this connection is such that aggregation node 901 can communicate with aggregation node 908 to receive a common AI / ML model M j , indicating that it may send its common AI / ML model M1 to the aggregation node 902.

[0183] Regarding the connections between the aggregation nodes, each of the aggregation nodes 901 to 908 is communicatively and operatively connected to a basic node in this example. Specifically, aggregation node 901 is communicatively and operatively connected to basic nodes 901a and 901b. These connections are established by the aggregation node 901 communicating with the common AI / ML model M received from node 908. jto the basic nodes 901a and 901b, and collect local AI / ML models from the basic nodes 901a and / or 901b. Similarly, the aggregation node 902 is communicatively and operatively connected to the basic nodes 902a and 902b. These connections indicate that the aggregation node 902 may transmit the common AI / ML model M1 received from node 901 to the basic nodes 902a and 902b, and collect local AI / ML models from the basic nodes 902a and / or 902b. Furthermore, the aggregation node 903 is communicatively and operatively connected to the basic nodes 903a and 903b. These connections indicate that the aggregation node 903 may transmit the common AI / ML model M2 received from node 902 to the basic nodes 903a and 903b, and collect local AI / ML models from the basic nodes 903a and / or 903b. Further, the aggregation node 904 is communicatively and operatively connected to the basic nodes 904a and 904b. These connections indicate that the aggregation node 904 may transmit the common AI / ML model M3 received from node 903 to the basic nodes 904a and 904b, and collect local AI / ML models from the basic nodes 904a and / or 904b. Further, the aggregation node 905 is communicatively and operatively connected to the basic nodes 905a and 905b. These connections indicate that the aggregation node 905 may transmit the common AI / ML model M4 received from node 904 to the basic nodes 905a and 905b, and collect local AI / ML models from the basic nodes 905a and / or 905b. Further, the aggregation node 906 is communicatively and operatively connected to the basic nodes 906a and 906b. These connections indicate that aggregation node 906 may transmit the common AI / ML model M5 received from node 905 to basic nodes 906a and 906b, and collect local AI / ML models from basic nodes 906a and / or 906b. Further, aggregation node 907 is communicatively and operatively connected to basic nodes 907a and 907b.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. Further, aggregation node 908 is communicatively and operatively connected to base nodes 908a and 908b. These connections indicate that aggregation node 908 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. i to the basic nodes 908a and 908b, indicating that local AI / ML models may be collected from the basic nodes 908a and / or 908b.

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

[0185] If an aggregation node i exists in the network, the aggregation node i may receive information related to each AI / ML model from one or more aggregation nodes communicatively and operatively connected to the aggregation node i. Each aggregation node that transmits information related to its own AI / ML model may be referred to as a previous aggregation node connected to the aggregation node i. The aggregation node i may transmit information related to its AI / ML model to one or more other aggregation nodes communicatively and operatively connected to the aggregation node i. Each aggregation node that receives information related to the aggregation node i's AI / ML model may be referred to as a next aggregation node connected to the aggregation node i. In the network shown in FIG. 10 , for aggregation node 902, the previous aggregation node is aggregation node 901, and the next aggregation node is aggregation node 903.

[0186] It should be noted that although the topology 1000 of Figure 10 shows that each aggregation node has one previous aggregation node and one next aggregation node, an aggregation node may have one or more previous aggregation nodes and one or more next aggregation nodes, as shown in Figure 16. 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 the present disclosure provide methods for configuring AI / ML topologies. Several methods for configuring AI / ML topologies in wireless communication networks are presented below and elsewhere in this disclosure.

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

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

[0190] 11 illustrates a configured topology supporting an exemplary AI / ML model transfer in accordance with an embodiment of the present disclosure. Referring to FIG. 11, node 1101 i-1 , 1101 i , 1101 i+1 Each of nodes 1101a to 1101f may be a user equipment (UE), a relay, a base station (BS), a transmission / reception point (TRP), an edge device, a network system, or an integrated access backhaul (IAB) node. The network 1100 or the network device 1150 may configure a topology that supports AI / ML model transfer by connecting each node 1101a to 1101f. i-1 , 1101 i , 1101 i+1 , 1101a-1101f, receive information including reports on their AI / ML capabilities. For example, the network 1100 or the network device 1150 may receive information from the nodes, such as node 1101, i Receives information including reports on the AI / ML capabilities of Node 1101 i-1 , 1101 i , 1101 i+1may optionally send 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 including a report on the node's AI / ML capabilities.

[0191] The network 1100 or the network device 1150 includes each node (i.e., node 1101 i-1 , 1101 i , 1101 i+1 , and 1101a through 1101f, respectively). In particular, network 1100 or network device 1150 configures the node type of each node. The node type of each node may be configured based on information received from the 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 (e.g., an aggregate node type) or Type 2 (e.g., a basic node type). For example, in the example shown in FIG. 11, node 1101 i-1 , 1101 i , 1101 i+1 The node type of the nodes 1101a to 1101f may be configured as type 1, and the node type of the nodes 1101a to 1101f may be configured as type 2.

[0192] Node (i.e., node 1101 i-1 , 1101 i , 1101 i+1 , and one of 1101a through 1101f) is configured as described above, network 1100 or network device 1150 may further configure one or more other nodes associated with the configured node. For example, the 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 When the network 1100 or the network device 1150 is configured, the basic node 1101a, the basic node 1101b, the aggregation node 1101 i-1 , or the aggregation node 1101 i+1At least one of the nodes 1101a and 1101b may be configured by the network 1100 or the network device 1150. i The basic nodes 1101a and 1101b may be configured to communicate and operatively connect to the nodes 1101a and 1101b. i The aggregation node 1101 may be configured to transmit the i-1 is node 1101 i , so that node 1101 i-1 is a common AI / ML model CM i-1 Node 1101 i The aggregation node 1101 sends the i+1 is node 1101 i , so that node 1101 i+1 is a common AI / ML model CM i Node 1101 i Receive from.

[0194] In another example, if the basic nodes 1101a and 1101b are configured by the network 1100 or the network device 1150, the aggregation node 1101 i The aggregation node 1101 may also be configured by the network 1100 or the network device 1150. For example, i may be configured to communicatively and operatively connect with basic nodes 1101a and 1101b. FIG. 11 illustrates a case in which basic nodes 1101a and 1101b are connected to one aggregation node (i.e., node 1101 i ), in some embodiments, some basic nodes (e.g., node 1701 in FIG. 17) may be communicatively and operatively connected to multiple aggregation nodes, as shown in FIG. 17. In some embodiments, when basic nodes 1101a and 1101b are configured, aggregation node 1101 i may be configured to collect local AI / ML models from basic nodes 1101a and 1101b, respectively.

[0195] Node 1101 included in network 1100 i-1 , 1101 i , 1101 i+1 , 1101a, 1101b, and other nodes may be communicatively and operatively connected in the topology via sidelinks using device-to-device (D2D) communications, via network devices (e.g., BSs, TRPs, etc.), or via interfaces between network devices. For example, in some embodiments, aggregation node 1101 i The aggregation node 1101 and its associated basic nodes 1101a and 1101b may be communicatively and operatively connected via a sidelink using device-to-device (D2D) communications. i and 1101 i+1 may be communicatively and operatively connected via a network device 1150. In some embodiments, aggregation node 1101 i and 1101 i+1 may be communicatively and operatively connected via network device 1150 and another network device (not shown in FIG. 11). In such a case, aggregation node 1101 i and 1101 i+1 may be communicatively and operatively connected via an interface between the network device 1150 and other network devices.

[0196] According to some aspects of the present disclosure, a network node, such as a user device, can autonomously configure a topology for AI / ML in a wireless communication network. To configure the topology for AI / ML, a node (first node) sends a message for aggregation to another node (second node). The second node determines whether to communicatively and operatively connect with the first node to transmit 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 determination. After the first node receives the response (e.g., an aggregation acknowledgement message), the first node and the second node selectively establish an aggregation connection between them based on the second node's determination. For example, if the second node determines to connect with the first node, the second node can send a positive aggregation acknowledgement message to the first node. After the first node receives the positive aggregation acknowledgement message, the first node and the second node can establish a connection between them. If the second node decides not to connect with the first node, the second node can send a negative aggregate acknowledgment message to the first node. In such a scenario, if the first node receives a negative aggregate acknowledgment message, the connection between the first node and the second node will not be established.

[0197] In this way, a network (e.g., a user device) can autonomously configure an AI / ML topology within a wireless communication network. The configured topology can support AI / ML model transfer over the air 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 distribution of a complete AI / ML model or a partial AI / ML model. The configured topology may include at least one of connections between at least one aggregation node and zero or more basic nodes or connections between at least two aggregation nodes.

[0198] To establish an aggregation connection with another node, an aggregation node may transmit information indicating its aggregation capabilities to the other node. In this case, the aggregation node may be referred to as a declaring node, and the other node may be referred to as a monitoring node. A declaring node is a node that transmits aggregated information that other nodes (e.g., monitoring nodes) can use to discover the declaring node. A monitoring node is a node that monitors the aggregated information transmitted by the declaring node. A monitoring node may be a basic node or another aggregation node (i.e., not a declaring node). For example, if the monitoring node is another aggregation node, the monitoring node may be the declaring node's previous aggregation node. In some embodiments, if the previous aggregation node is not discovered by the discovery procedure, the aggregation node (i.e., the declaring node) may transmit 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] 12 illustrates an example of a procedure for establishing an aggregation connection between an aggregation node and a basic node when configuring an AI / ML topology in a wireless communication network according to an embodiment of the present disclosure. i is a declaration node, and basic nodes 1101a and 1101b are monitoring nodes.

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

[0201] In step 1220, the basic nodes 1101a and 1101b are connected to the aggregation node 1101. i If basic nodes 1101a and 1101b are interested in the discovery message received from node 1101, they can read and process the received discovery message. i The controller may determine whether to communicatively and operatively connect to the device.

[0202] In step 1230, basic nodes 1101a and 1101b each send their response to aggregation node 1101. i Each response can be sent by the basic node 1101a or 1101b to the aggregation node 1101. i In some embodiments, the response is either a positive aggregate acknowledgement message or a negative aggregate acknowledgement message. For example, basic node 1101a indicates whether it has decided to connect with aggregation node 1101. i The basic node 1101b decides to connect to the aggregation node 1101. i In this case, the basic node 1101a may decide not to connect to the aggregation node 1101. i The basic node 1101b sends a positive aggregate acknowledgement message (e.g., ACK) to the aggregation node 1101. isend a negative aggregate acknowledgment message (e.g., NACK) to

[0203] In step 1240, the aggregation node 1101 i After receiving the response, the aggregation node 1101 i Based on the received responses, basic nodes 1101a and 1101b selectively establish aggregation connections with one or both of basic nodes 1101a and 1101b. In the illustrated example, basic nodes 1101a and 1101b both respond with positive aggregation acknowledgement messages in step 1230, and therefore aggregation node 1101 i Assume that in step 1240, establishes an aggregation connection with each of basic nodes 1101a and 1101b. Meanwhile, aggregation node 1101 i may not establish an aggregation connection with associated base nodes that respond with a negative aggregation acknowledgement message in step 1230 .

[0204] Using the discovery procedure above, aggregation node 1101 i After the aggregation connections between the basic nodes 1101a and 1101b are established, the aggregation node 1101 i can collect local AI / ML models from base nodes 1101a and 1101b and perform aggregation operations to generate new or updated common AI / ML models.

[0205] For example, referring again to FIG. 11, aggregation node 1101 i After the aggregation connection between the basic nodes 1101a and 1101b is established, the node 1101 i is node 1101 i-1 From AI / ML model CM i-1 For example, AI / ML model CM i-1 is node 1101 i-1 The AI / ML model CM may communicate by broadcast, groupcast, or unicast signaling, for example, broadcast / groupcast / unicast RRC, MAC-CE, or DCI, or by the TRP / BS interface. i-1 After receivingi assigns an AI / ML model CM to one or more of its associated nodes 1101a and 1101b. i-1 For example, an AI / ML model CM i-1 is node 1101 i This may be communicated via broadcast, groupcast, or unicast signaling, for example, broadcast / groupcast / unicast RRC, MAC-CE, or DCI, or TRP / BS interface.

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

[0207] In some embodiments, node 1101 i may send indicators to collect AI / ML models from associated nodes 1101a and 1101b. For example, the indicators may i The indicator may be communicated by broadcast, groupcast, or unicast signaling, such as broadcast / groupcast / unicast RRC, MAC-CE, or DCI, or by the TRP / BS interface. This indicator may be referred to as a model collection indicator. 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 associated nodes 1101a and 1101b. In some embodiments, 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-triggered indicator. For example, if the model collection indicator is a dynamic indicator, node 1101 i receives reports related to the associated AI / ML models of each of the associated nodes 1101a and 1101b from the associated nodes 1101a and 1101b, and iThe model collection indicator may be received at each preset time by the associated nodes 1101a and 1101b. On the other hand, if the model collection indicator is an event trigger indicator, the associated nodes 1101a and 1101b may send their own reports when a specific event is triggered. The event may be triggered when training of the associated AI / ML model of each of the associated nodes 1101a and 1101b is completed or when the performance of the associated AI / ML model of each of the associated nodes exceeds a specific performance indicator. The specific performance indicator may be preset based on at least one of accuracy or precision. If the performance of one of the associated AI / ML models does not exceed the preset performance indicator (e.g., the accuracy is lower than a predetermined accuracy indicator), the report related to that AI / ML model is not generated and / or the report may be sent by the associated nodes 1101a and 1101b. i (e.g., no local AL / ML model report transmission).

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

[0209] The associated nodes 1101a and 1101b then forward reports related to the associated AI / ML models of the associated nodes 1101a and 1101b to the nodes 1101a and 1101b. iIn 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 associated nodes 1101a and 1101b, information related to training data for the associated AI / ML models of each of associated nodes 1101a and 1101b, or information related to performance of the associated AI / ML models of each of associated nodes 1101a and 1101b. The information related to each associated AI / ML model may include the NN structure (NN algorithm, width, depth, etc.) and / or one or more AI / ML parameters (weights, biases, activation functions, etc.) of associated nodes 1101a and 1101b. The information related to the training data for each associated AI / ML model may include information related to the amount (volume) of training data and / or training data distribution. The information related to the performance of each associated AI / ML model may include accuracy, precision, recall, loss information (e.g., average cross-entropy loss), and the performance of each associated AI / ML model of associated nodes 1101a and 1101b. i (or other aggregator nodes or BSs). Note that in some embodiments, information related to the training data of each associated AI / ML model and / or information related to the performance of each associated AI / ML model can assist the aggregation operations of the aggregator nodes. For example, training data information and / or performance information can be used to determine aggregation weights for the AI / ML models.

[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 Based on the received report, the AI / ML model CM i Generate AI / ML model CM i The AI / ML model CM may be an updated common AI / ML model. i-1 and commercials i may have the same NN structure. In some embodiments, node 1101 iUse aggregation algorithms (e.g., model distillation, model augmentation) to refine AI / ML model CMMs i In some embodiments, node 1200 i may generate its own AI / ML model, which may be a local AI / ML model, which may be generated without a report associated with the associated AI / ML model of each of the associated nodes 1101a and 1101b.

[0211] Next, node 1101 i AI / ML model CM i Node 1101 i+1 AI / ML model CM i may contain one or more AI / ML model parameters (weights, biases, etc.), but the AI / ML model CM i In other words, the AI / ML model CM does not need to include information related to the NN structure. i The NN structure may be pre-configured and / or the AI / ML model CM i The NN structure is node 1101 i+1 Since it may be known in AI / ML model CM i The information related to the NN structure is in node 1101. i From node 1101 i+1 In some embodiments, the node 1200 i AI / ML model sent to CM i may be a complete AI / ML model or a partial AI / ML model.

[0212] As described above, the associated AI / ML models of the associated nodes 1101a and 1101b are represented by the AI / ML model CM i-1 and AI / ML model CM i In other words, heterogeneous AI / ML model aggregation may be enabled by configuring a topology for AI / ML in a wireless communication network in accordance with embodiments of the present disclosure. For example, in some embodiments, node 1101 iBy previous node 1101 i-1 AI / ML model received from CM i-1 and node 1101 i The AI / ML model CM generated by the AI / ML model based on reports received from its associated nodes 1101a and 1101b by i Although both may include a four-layer NN, the local AI / ML models generated by associated node 1101a and / or 1101b may have different NN structures. For example, the local AI / ML model generated by associated node 1101a may include a five-layer NN, and the local AI / ML model generated by associated node 1101b may include a three-layer NN.

[0213] Node 1101 i receives reports related to the local AI / ML models of its associated nodes 1101a and 1101b, and then generates an AI / ML model CMM based on the local AI / ML models generated by the associated nodes 1101a and 1101b. i Each of the local AI / ML models generated by the association nodes 1101a and 1101b can be used to generate an AI / ML model CM. i Since each of the local AI / ML models generated by the associated nodes 1101a and 1101b may have different importance (significance) to the generation of W, when aggregating the local AI / ML models, e.g., W a and W b It may be weighted by each weight such as W a and W b may indicate the importance of the local AI / ML model generated by the associated nodes 1101a and 1101b. i may be an updated common AI / ML model. i The NN structure is the AI / ML model CM i-1 In this case, the AI / ML model CM iand AI / ML model CM i-1 Both may contain, for example, four-layer NNs.

[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 the association nodes 1101a and 1101b are used as the AI / ML model CM. i In order to generate a NN, it is not necessary for the NN structure to be equivalent to that of the common AI / ML model. In other words, it enables the aggregation of 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 associated basic nodes communicatively and operably connected to the aggregation node. The aggregation node may distill and / or extend the received heterogeneous AI / ML models. The aggregation node may then aggregate the distilled and / or extended AI / ML models to generate a new or updated common AI / ML model. In some embodiments, the aggregation node may obtain an average of the distilled AI / ML models and / or an average of the extended AI / ML models. After obtaining the common AI / ML model, the aggregation node may transmit the new or updated common AI / ML model to one or more other nodes, which may be subsequent aggregation nodes.

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

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

[0218] 13 illustrates an example of a 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. i is a declaration node, and the aggregation node 1101 i-1 and 1101 k is a monitoring node. Aggregation node 1101 i-1 is the aggregation node 1101 i It may be an aggregation node before

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

[0220] In step 1320, the aggregation node 1101 i-1 and 1101 k is the aggregation node 1101 i If the aggregation node 1101 is interested in a discovery message received from i-1 and 1101 k can read and process the received discovery message. i-1 and 1101 k is node 1101 i The controller may determine whether to communicatively and operatively connect to the device.

[0221] In step 1330, the aggregation node 1101 i-1 and 1101 k Each node sends its own response to the aggregation node 1101. i Each response can be sent to the aggregation node 1101. i-1 or 1101 k is the aggregation node 1101 i In some embodiments, the response is either a positive aggregate acknowledgement message or a negative aggregate acknowledgement message. For example, aggregation node 1101 i-1 is aggregation node 1101 i and decides to connect to aggregate 1101 k is aggregation node 1101 i In this case, the aggregation node 1101 may decide not to connect to the i-1 is aggregation node 1101 i Node 1101 sends a positive aggregate acknowledgement message (e.g., ACK) to the aggregate node. k is aggregation node 1101 i send a negative aggregate acknowledgment message (e.g., NACK) to

[0222] In some embodiments, an aggregation node may be communicatively and / or operatively connected to 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 acknowledgment to the monitoring aggregation node informing it whether a connection to the declaring node was successfully established. In FIG. 13, aggregation node 1101 i is the aggregation node 1101 i-1 and 1101 k If the aggregation node 1101 can only connect with one of the i In step 1335, optionally, the aggregation node 1101 i-1 and 1101 k each of them sends a conflict confirmation to the aggregation node 1101 i Notifies you whether the connection was successful.

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

[0224] Aggregation node 1101 i and aggregation node 1101 i-1 After the aggregation connection between the aggregation node 1101 and the i is the aggregation node 1101 i-1 You can receive common AI / ML models from

[0225] To establish an aggregation connection with another node, a node may send an aggregation request message (e.g., an aggregation request) to the other node. In this case, the node sending the aggregation request message may be referred to as the discovering node, and the other node receiving the aggregation request message may be referred to as the discovered node. If the discovering node is a basic node, the discovering node may send an aggregation request message to an associated aggregation node. If the discovering node is an aggregation node, the discovering node may send an aggregation request message to the other aggregation node to request the other aggregation node as the next hop (e.g., request the other aggregation node to be the next aggregation node). In some embodiments, if a next aggregation node is not discovered by the discovery procedure, the aggregation node (discovering node) may send the aggregated common AI / ML model to a network device (e.g., BS, TRP).

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

[0227] In step 1410, the basic node 1101a i and 1101 i+1 In some embodiments, the basic node 1101a may broadcast, groupcast, or unicast the aggregation request message at predetermined intervals. The aggregation request message may include requested aggregate information that the basic node 1101a is interested in discovering. If the basic node 1101a is a discovery node, the aggregation request message may include information related to the local AI / ML model of the basic node 1101a. The information related to the local AI / ML model of the basic node 1101a may include information related to the neural network of the local AI / ML model of the basic node 1101a, information related to the size of the local AI / ML model of the basic node 1101a, or information related to the complexity of the local AI / ML model of the basic node 1101a. 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, the aggregation node 1101 i and 1101 i+1 is interested in the aggregation request message received from the basic node 1101a, the aggregation node 1101 i and 1101 i+1 can read and process the received aggregation request message. i and 1101 i+1 may determine whether to communicate and operatively connect with the basic node 1101a. For example, the discovered aggregation node 1101 i has a distillation function, and if the discovery base node 1101a reports an AI / ML model that is larger than the reference AI / ML model, the aggregation node 1101 idetermines to communicatively and operationally connect with the basic node 1101a. i+1 has a distillation function, and if the discovery base node 1101a reports an AI / ML model that is smaller than the reference AI / ML model, the aggregation node 1101 i+1 determines to communicatively and operatively connect with basic node 1101a.

[0229] In step 1430, the aggregation node 1101 i and 1101 i+1 Each of the nodes can send its own response to the aggregation node 1101a. i or 1101 i+1 has decided to connect with basic node 1101a. In some embodiments, the response is either a positive aggregate acknowledgement message or a negative aggregate acknowledgement message. For example, discovered aggregation node 1101 i has a distillation function, and the discovered aggregation node 1101 i+1 has an extended function, and the discovery base node 1101a reports an AI / ML model that is larger than the reference AI / ML model, so the aggregation node 1101 i decides to connect to the basic node 1101a, and the aggregation node 1101 i+1 may decide not to connect to the basic node 1101a. In this case, the aggregation node 1101 i sends a positive aggregate acknowledgement message (e.g., ACK) to basic node 1101a, and aggregation node 1101 i+1 sends a negative aggregate acknowledgement message (eg, NACK) to basic node 1101a.

[0230] In step 1435, the base node 1101a optionally adds the aggregate node 1101 i and 1101 i+1 may send a conflict check to each to check whether they are successfully connected to basic node 1101a.

[0231] In step 1440, the basic node 1101a determines whether the aggregation node 1101 i The aggregation connection between the basic node 1101a and the aggregation node 1101 is established. i+1 No aggregation connection is established between

[0232] The discovery procedure finds aggregation node 1101a and aggregation node 1101 i After the aggregation connection between the aggregation node 1101 and the i can collect local AI / ML models from the base nodes 1101a and perform aggregation operations to generate new or updated common AI / ML models.

[0233] 15 illustrates another example of a 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. i is the discovery node, and the aggregation node 1101 i+1 and 1101 k is the discovered node. Aggregation node 1101 i+1 is the aggregation node 1101 i It may be the next aggregation node of

[0234] In step 1510, the aggregation node 1101 i is the aggregation node 1101 k and 1101 i+1 In some embodiments, aggregation node 1101 i The aggregation node 1101 may broadcast, groupcast or unicast the aggregation request message at predetermined intervals, for example, by broadcast / groupcast / unicast RRC, MAC-CE or DCI, or TRP / BS interface. i It may contain aggregation requests from

[0235] In step 1520, the aggregation node 1101 kand 1101 i+1 is the aggregation node 1101 i If the aggregation node 1101 is interested in the aggregation request message received from k and 1101 i+1 can read and process the received aggregation request message. k and 1101 i+1 is the aggregation node 1101 i The controller may determine whether to communicatively and operatively connect to the device.

[0236] In step 1530, the aggregation node 1101 k and 1101 i+1 Each node sends its own response to the aggregation node 1101. i Each response can be sent to the aggregation node 1101. k or 1101 i+1 is the aggregation node 1101 i In some embodiments, the response is either a positive aggregate acknowledgement message or a negative aggregate acknowledgement message. For example, aggregation node 1101 i+1 is aggregation node 1101 i and decides to connect to aggregate 1101 k is aggregation node 1101 i In this case, the aggregation node 1101 may decide not to connect to the i+1 is aggregation node 1101 i Node 1101 sends a positive aggregate acknowledgement message (e.g., ACK) to the aggregate node. k is aggregation node 1101 i send a negative aggregate acknowledgment message (e.g., NACK) to

[0237] In some embodiments, a discovery aggregation node may be communicatively and / or operatively connected 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 acknowledgment to the discovered aggregation node informing it whether a connection to the discovery node was successfully established. In FIG. 15, discovery aggregation node 1101 i is the discovered aggregation node 1101 k and 1101 i+1 If the aggregation node 1101 can only connect with one of the i In step 1535, optionally, the discovered aggregation node 1101 k and 1101 i+1 each of them sends a conflict confirmation to the aggregation node 1101 i Notifies you whether the connection was successful.

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

[0239] Aggregation node 1101 i and aggregation node 1101 i+1 After the discovery procedure establishes an aggregation connection between the discovered aggregation node 1101 and the i+1 is the discovery aggregation node 1101 i You can receive common AI / ML models from

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

[0241] In some embodiments, a 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 illustrates an example of a configured topology 1600 that supports flexible communication 1600 between aggregation nodes, according to an embodiment 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 common AI / ML model M1 to aggregation nodes 902, 903, 904, and 905. Meanwhile, aggregation node 907 sends common AI / ML model M i 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, e.g., by broadcast / groupcast / unicast RRC, MAC-CE or DCI, or TRP / BS interface.

[0243] In configured topology 1600, an aggregation node can receive a common AI / ML model 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] With respect to aggregating AI / ML models, the operation of aggregation is illustrated 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. Aggregation node 908 then aggregates the received AI / ML models M2, M3, M4, and M5. In some embodiments, aggregation node 908 may calculate an average of AI / ML models M2, M3, M4, and M5 to obtain an aggregated AI / ML model. After obtaining the aggregated AI / ML model, aggregation node 908 transmits the aggregated AI / ML model to its associated basic nodes. In this example, the associated basic nodes include three basic nodes 908a, 908b, and 908c communicatively and operably 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 the aggregation node 908. Once training is complete, the aggregation node 908 collects the local AI / ML models from the base nodes 908a, 908b, and 908c. The aggregation node 908 then aggregates the collected local AI / ML models to generate an updated common AI / ML model M j The aggregation node 908 can send the updated common AI / ML model to its next aggregation node 901.

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

[0246] In the configured topology 1700, each of aggregation nodes 901-905, 907, and 908 is communicatively and operatively connected to one or more basic nodes. For example, aggregation node 901 is communicatively and operatively connected to basic nodes 901a and 1701. However, aggregation node 906 is not communicatively and operatively connected to any basic nodes. Because aggregation node 906 does not have an associated basic node, aggregation node 906 can generate an updated common AI / ML model in different ways. For example, aggregation node 906 receives a common AI / ML model from another aggregation node (e.g., aggregation node 905), and aggregation node 906 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. Once aggregation node 906 obtains the updated common AI / ML model, aggregation node 906 sends the updated AI / ML model to its next aggregation node (e.g., aggregation node 907).

[0247] In the configured topology 1700, each basic node is communicatively and operatively connected to one or more aggregation nodes. For example, basic node 901a is communicatively and operatively connected to aggregation node 901.

[0248] On the other hand, the aggregation node 1701 is communicatively and operatively connected to two aggregation nodes 901 and 908. The connection of the basic node 1701 can be implemented using a network device (e.g., a 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. When the network device indicates the AI / ML model identifiers, the same AI / ML model identifier can be assigned to multiple aggregation nodes. Thus, the identifier of the local AI / ML model associated with the basic node 1701 can be assigned to both aggregation nodes 901 and 908.

[0249] Certain aspects of the present disclosure enable heterogeneous AI / ML capabilities in various network devices and user devices, and can support heterogeneous AI / ML model transfer over the air interface of a wireless communication network.

[0250] According to some aspects of the present disclosure, the topology within a wireless communication network can be centrally configured by a network device (e.g., a base station (BS), a transmit and receive point (TRP)), or a network system.

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

[0252] In accordance with some aspects of the present disclosure, a topology within a wireless communication network configured by various methods described in this disclosure supports parallel AI / ML model aggregation, reduces AI / ML model routing latency, and improves AI / ML model routing robustness.

[0253] Also disclosed are example devices (eg, ED or UE and TRP or network device) for performing the various methods described herein.

[0254] For example, a first device may include a memory storing processor-executable instructions and a processor that executes the processor-executable instructions. Execution of the processor-executable instructions may cause the processor to perform one or more of the device method steps described herein, e.g., in connection with Figures 11-15. For example, the processor may cause the device to communicate over the air interface in an operating mode by performing operations consistent with that operating mode, e.g., performing necessary measurements, generating content from those measurements, preparing uplink transmissions, processing downlink transmissions (e.g., encoding, decoding, etc.), and configuring and / or directing transmission / reception on RF chains and antennas as configured for the operating mode.

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

[0256] While the present invention has been described with respect to specific features and embodiments thereof, various modifications and combinations can be made without departing from the invention. Accordingly, the description and drawings should be considered merely as illustrative of certain embodiments of the invention as defined by the appended claims, and it is intended to cover any and all changes, variations, combinations, or equivalents that are within the scope of the invention. Thus, while the invention and its advantages have been described in detail, various changes, substitutions, and alterations can be made therein without departing from the invention as defined by the appended claims. Moreover, the scope of the present application is not intended to be limited to the particular embodiments of the process, machine, manufacture, composition of matter, means, methods, and steps described in the specification. As those skilled in the art will readily appreciate from this disclosure, any now-existing or later-developed process, machine, manufacture, composition of matter, means, methods, or steps that perform substantially the same function or achieve substantially the same results as the corresponding embodiments described herein can be utilized in accordance with the present invention. Accordingly, the appended claims are intended to include within their scope such processes, machines, manufacture, compositions of matter, means, methods, or steps.

[0257] Additionally, any module, component, or device illustrated herein that executes instructions may include or otherwise be accessible to one or more non-transitory 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-exhaustive list of examples of non-transitory computer / processor-readable storage media includes magnetic cassettes, magnetic tape, magnetic disk storage or other magnetic storage devices, compact disk read-only memory (CD-ROM), digital video disk or digital versatile disk (DVD), optical disk such as Blu-ray Disc™ or other optical storage devices, volatile and non-volatile, removable and non-removable media implemented in any manner or technology, random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory, or other memory technology. Such non-transitory computer / processor storage media may be part of the device or may be accessible or connectable to it. Any applications or modules described herein may be implemented using computer / processor readable / executable instructions stored or maintained by such non-transitory computer / processor readable storage media.

[0258] Abbreviations 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. 1. A method for configuring a topology for artificial intelligence or machine learning (AI / ML) in a wireless communication network, the method comprising: receiving information from a node, the information including a report related to the node's AI / ML capabilities; configuring the node based on the received information, the configuring the node including configuring a node type for the node, the configured node type being one of a plurality of node types, the plurality of node types comprising: Type 1, which indicates a node configured to collect a plurality of AI / ML models and aggregate the collected AI / ML models to obtain an AI / ML model of a first type; or Type 2, which indicates a node configured to obtain a second type of AI / ML model using a set of training data without performing an aggregation operation; and configuring one or more other nodes associated with the configured node based on the configured node type; Including, The configured topology supports AI / ML model transfer over an air interface of the wireless communication network, and includes: a connection between at least one type 1 node and zero or more type 2 nodes, or a connection between at least two type 1 nodes; The method includes at least one of:

2. The configured node type indicates that the node is a Type 1 node, and the one or more other nodes are: one or more Type 2 nodes, a second Type 1 node that provides the first-type AI / ML model of the second Type 1 node to said node; or a third Type 1 node receiving the first Type AI / ML model of said node; The method of claim 1 , comprising at least one of:

3. the one or more other nodes include the one or more Type 2 nodes, and configuring the nodes comprises:

3. The method of claim 2, comprising configuring the one or more type 2 nodes to collect respective second type AI / ML models from the one or more type 2 nodes.

4. 2. The method of 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 of any one of claims 1 to 4, wherein the information further comprises a request by the node to be configured as a Type 1 node.

6. 6. The method of claim 1, wherein the node is connected to the 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.

7. The method according to any one of claims 1 to 6, wherein the node is a user equipment (UE), a relay, a base station (BS), a transmission / reception point (TRP), an edge device, a network system, or an integrated access backhaul (IAB) node.

8. 1. A method for configuring a topology for artificial intelligence or machine learning (AI / ML) in a wireless communication network, the method comprising: establishing, by the first node, an aggregation connection with the second node based on the aggregation acknowledgement message; The configured topology supports AI / ML model transfer over an air interface of the wireless communication network, and includes: a connection between at least one type 1 node and zero or more type 2 nodes, or a connection between at least two type 1 nodes; and A method, wherein a type 1 node is 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, and a type 2 node is a node configured to obtain a second type AI / ML model using a set of training data without performing an aggregation operation.

9. The method of claim 8 , wherein the message for aggregation is a discovery message that includes aggregation information used by the second node to discover the first node.

10. The method of claim 9 , wherein the aggregation information used by the second node includes an aggregation capability of the first node.

11. The method of claim 9 , wherein the discovery message includes at least one of a model collection indicator or a reference AI / ML model.

12. The method of claim 11 , wherein the model collection indicator comprises one of a distillation indicator, an expansion indicator, or a distillation and expansion indicator.

13. 10. The method of claim 8, wherein the message for aggregation is an aggregation request message, and the aggregation request message includes information related to a second-type AI / ML model of the first node or an aggregation request.

14. 14. The method of claim 13, wherein the first node is a Type 2 node, and the aggregation request message includes information related to the second-type AI / ML model of the first node.

15. The information related to the second type AI / ML model of the first node may include: Information related to the neural network of the second type AI / ML model of the first node; Information about 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 of the first node; The method of claim 14 , comprising at least one of:

16. The method of claim 13 , wherein the first node is a Type 1 node and the aggregation request message includes the aggregation request.

17. The second node is one of a plurality of other nodes in the wireless communication network, and the first node can connect to only one node of the plurality of other nodes, and the method includes: sending, by said first node, to at least said one of said plurality of other nodes a respective confirmation informing whether a connection to said first node has been successfully established; The method of claim 8 further comprising:

18. 18. The method of claim 17, wherein the first node is a Type 2 node and the second node is a Type 1 node.

19. 18. The method of claim 17, wherein 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.

20. The method of claim 8 , wherein the first node is a Type 1 node that is only related to one or more other Type 1 nodes.

21. The method according to any one of claims 8 to 20, wherein the step of transmitting the message for aggregation comprises the step of broadcasting, groupcasting or unicasting the message for aggregation at predetermined intervals.

22. 1. A network device for configuring a topology for artificial intelligence or machine learning (AI / ML) in a wireless communication network, the network device including: a processor; and a memory storing processor-executable instructions, the processor-executable instructions, when executed, causing the processor to: receiving information from a node, the information including a report related to the node's AI / ML capabilities; configuring the node based on the received information, wherein configuring the node includes configuring a node type of the node, the configured node type being one of a plurality of node types, the plurality of node types including: Type 1 refers to 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 indicates a node configured to obtain a second-type AI / ML model using a set of training data without performing an aggregation operation; Including, 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 an air interface of the wireless communication network and includes at least one of a connection between at least one Type 1 node and zero or more Type 2 nodes, or a connection between at least two Type 1 nodes.

23. The configured node type indicates that the node is a Type 1 node, and the one or more other nodes are: one or more Type 2 nodes, a second Type 1 node that provides the first-type AI / ML model of the second Type 1 node to said node; or a third Type 1 node receiving the first Type AI / ML model of said node; 23. The network device of claim 22, comprising at least one of:

24. the one or more other nodes include the one or more Type 2 nodes, and configuring the nodes includes:

24. The network device of claim 23, further comprising configuring the node to collect respective second type AI / ML models from the one or more type 2 nodes.

25. 23. The network device of claim 22, 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.

26. A network device according to any one of claims 22 to 25, wherein said information further comprises a request by said node to be configured as a Type 1 node.

27. 27. The network device of claim 22, wherein the node is connected to the 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.

28. The network device according to any one of claims 22 to 29, wherein the node is a user equipment (UE), a repeater, a base station (BS), a transmission / reception point (TRP), an edge device, a network system, or an integrated access backhaul (IAB) node.

29. An apparatus comprising one or more units for carrying out the method according to any one of claims 1 to 7.

30. 1. A device for a node that configures a topology for artificial intelligence or machine learning (AI / ML) in a wireless communication network, the device comprising: a processor; a memory storing processor-executable instructions; wherein the processor-executable instructions, when executed, cause the processor to: establishing an aggregation connection with the second node based on the aggregation acknowledgment message; The configured topology supports AI / ML model transfer over an air interface of the wireless communication network, and includes: a connection between at least one type 1 node and zero or more type 2 nodes, or a connection between at least two type 1 nodes; and A Type 1 node is 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, and a Type 2 node is a node configured to obtain a second-type AI / ML model using a set of training data without performing an aggregation operation, the device.

31. 31. The apparatus of claim 30, wherein the message for aggregation is a discovery message that includes aggregation information used by the second node to discover the first node.

32. 32. The apparatus of claim 31, wherein the aggregation information used by the second node includes an aggregation capability of the first node.

33. 32. The device of claim 31, wherein the discovery message includes at least one of a model collection indicator or a reference AI / ML model.

34. 34. The apparatus of claim 33, wherein the model collection indicator comprises one of a distillation indicator, an expansion indicator, or a distillation and expansion indicator.

35. 31. The apparatus of claim 30, wherein the message for aggregation is an aggregation request message, the aggregation request message including information related to a second-type AI / ML model of the first node or an aggregation request.

36. 36. The apparatus of claim 35, wherein the first node is a Type 2 node, and the aggregation request message includes information related to the second-type AI / ML model of the first node.

37. The information related to the second type AI / ML model of the first node may include: Information related to the neural network of the second type AI / ML model of the first node; Information about 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 of the first node; 37. The device of claim 36, comprising at least one of:

38. 36. The apparatus of claim 35, wherein the first node is a Type 1 node and the aggregation request message includes the aggregation request.

39. the second node is one of a plurality of other nodes in the wireless communication network, the first node can connect to only one node of the plurality of other nodes, and the processor-executable instructions, when executed, cause the processor to:

31. The apparatus of claim 30, further comprising processor-executable instructions for causing the first node to send a respective confirmation to at least the one of the plurality of other nodes informing the at least one node whether a connection to the first node was successfully established.

40. 40. The apparatus of claim 39, wherein the first node is a Type 2 node and the second node is a Type 1 node.

41. An apparatus comprising one or more units for carrying out the method according to any one of claims 8 to 21.

Images