Determining vertical application layer (VAL) groups using generative adversarial networks (GANs)

Abstract

Embodiments include a method for determining vertical application layer (VAL) groups operating in a communication network, each VAL group comprising a plurality of user equipments (UEs). Such a method includes determining a first multi-layer graph representation based on a reference VAL group configuration and a historical quality of service (QoS) metric. Such a method includes performing, using a generative adversarial network (GAN), a first operation of determining a second multi-layer graph representation, and determining whether a first condition is satisfied: the second multi-layer graph representation is undistinguishable from the first multi-layer graph representation, and predicting that a QoS metric satisfies one or more service level agreements (SLAs). Such a method includes repeating a first operation based on another random input when a first condition is not satisfied; and after the first condition is satisfied, performing reasoning based on the second multilayer graph representation to generate a configuration of the VAL group for operating in the communication network.

Description

Technical Field

[0001] This disclosure generally relates to the field of communication networks, and more specifically to a technique for determining the vertical application layer (VAL) group of user equipment (UE) operating in a communication network (e.g., a 5G wireless network) based on applied machine learning (ML) techniques (e.g., generative adversarial networks (GANs)). Background Technology

[0002] Fifth-generation (5G) cellular systems (also known as New Radio (NR)) were initially standardized in 3GPP Rel-15 and have continued to evolve in subsequent versions. NR was developed to achieve maximum flexibility to support a wide range of use cases, including enhanced mobile broadband (eMBB), machine-type communication (MTC), ultra-reliable low-latency communication (URLLC), sidelink device-to-device (D2D), and several other use cases. 5G / NR technology shares many similarities with fourth-generation Long Term Evolution (LTE).

[0003] One change in 5G networks is the modification and / or replacement of traditional point-to-point interfaces and protocols found in previous generations of networks (such as LTE) with a service-based architecture (SBA). In this SBA, network functions (NFs) provide one or more services to one or more service consumers. This can be accomplished, for example, through Hypertext Transfer Protocol / Representational State Transfer (HTTP / REST) ​​application programming interfaces (APIs). Typically, the various services are self-contained functions that can be changed and modified in isolation without affecting other services. Furthermore, these services are composed of various "service operations," which are finer-grained divisions of the overall service functionality.

[0004] In 5G SBA, the Network Repository Function (NRF) allows each network function to discover services provided by other network functions, while the Data Storage Function (DSF) allows each network function to store its context. The Unified Data Management (UDM) function supports the generation of 3GPP authentication credentials, user identity processing, subscription-based access authorization, and other subscriber-related functions. The Network Exposure Function (NEF) acts as an entry point into the operator's 5G Core Network (5GC) by securely exposing network capabilities and events provided by other NFs and providing pathways for Application Functions (AFs) to securely provide information to (or receive information from) the 5GC.

[0005] 3GPP has specified the Service Enabler Architecture Layer (SEAL) to support vertical applications on 5G networks. SEAL services include group management, configuration management, location management, identity management, key management, and network resource management. SEAL provides these services to various Vertical Application Layers (VALs) that can run on top of SEAL, corresponding to various applications. Specifically, the UE may include a SEAL group management client that communicates with a corresponding SEAL group management server inside (or outside) the 5GC. The group management client provides group management services to VAL clients in the UE, which communicate with corresponding VAL servers inside (or outside) the 5GC. The VAL servers also communicate with the SEAL group management server.

[0006] SEAL can use various information for VAL group management, including VAL group ID, group member UE ID, general group configuration, group data network name (DNN), single network slice selection assistance information (S-NSSAI), group size, group leader, group location information, enabled VAL services, VAL service-specific information, and Protocol Data Unit (PDU) session type. This information is further specified in Clause 10 of 3GPP TS 23.434 (v17.7.0) and Clause 7.2.1.4 of 3GPP TS 29.549 (v17.8.0). Summary of the Invention

[0007] However, the information specified by 3GPP for VAL group management does not include UE communication characteristics that can be crucial in practical applications. For example, in Industrial Internet of Things (IIoT) applications (such as mining, chemical or food processing, and automotive manufacturing), VAL group management should consider multiple site-specific parameters of communication characteristics. These parameters may include connectivity Quality of Service (QoS) data, guaranteed bit rate and maximum bit rate, packet delay budget, and packet error rate. Furthermore, each of these parameters can relate to intra-group communication as well as communication between users in different groups.

[0008] Furthermore, traditional VAL group management / configuration is done manually by programmers, who rely to some extent on site-specific parameters, historical data, and their own experience. This challenging task becomes even more difficult due to inevitable changes in site specifications and / or physical layout. Therefore, manual VAL group configuration may fail to meet the QoS and / or Quality of Experience (QoE) requirements for intra-group and inter-group communication as mandated by Service Level Agreements (SLAs). Moreover, manual VAL group configuration is unlikely to produce optimal resource allocation within a 5G system, instead leading to traffic congestion, UE and / or network resource unavailability, and QoS degradation.

[0009] One object of embodiments of this disclosure is to address one or more of these and related problems, challenges and / or difficulties by providing techniques and apparatus for automatically configuring VAL groups based on historical and / or predicted communication parameters in a communication network (e.g., a 5G network).

[0010] Some embodiments of this disclosure include methods (e.g., procedures) for determining VAL groups operating in a communication network, wherein each VAL group includes multiple UEs.

[0011] These exemplary methods include determining a first multi-layer graph representation based on: a reference VAL group configuration previously used in the communication network, and historical QoS metrics for multi-layer connections between UEs including the reference VAL group configuration. These exemplary methods also include performing the following first operation using a generative adversarial network (GAN) including a generator and a discriminator: Using a generator, a second multi-layer graph representation is determined based on: candidate VAL group configurations determined based on random input, and QoS metrics predicted for multi-layer connectivity between UEs including reference VAL group configurations. Using a discriminator, determine whether the following first condition is met: the second multi-layer graph representation cannot be distinguished from the first multi-layer graph representation, and the predicted QoS metric associated with the candidate VAL group configuration satisfies one or more Service Level Agreements (SLAs) for the operation of the communication network. When the first condition is not met, these exemplary methods also include repeating the first operation (described above) based on different random inputs. These exemplary methods also include, after the first condition is met and using a generator, performing inference based on a second multi-layer graph representation to generate a configuration for the VAL group operating in the communication network.

[0012] In some embodiments, the exemplary method is performed by the Network Data Analytics Function (NWDAF) of the communication network. In other embodiments, the method is performed by a SEAL Group Management (GM) server external to (or internal to) the communication network. In some of these embodiments, the method is performed by the NWDAF, and the reasoning is further based on the desired architecture of the VAL group obtained from the SEAL GM server.

[0013] In some embodiments, the configuration of VAL groups generated by performing inference includes the following: multiple UEs deployed in multiple VAL groups, and connections between the respective UE pairs. In some embodiments of these embodiments, the connections between the respective UE pairs include the following: connections between UE pairs within the same VAL group, and connections between UE pairs in different VAL groups.

[0014] In some embodiments of these examples, the configuration of the VAL group generated by performing inference also includes QoS settings for the connections between the individual UE pairs. In other embodiments of these examples, these exemplary methods also include obtaining predicted QoS settings for the connections between the individual UE pairs from the NWDAF of the communication network.

[0015] In some embodiments, each of the first multi-layer graph representation and the second multi-layer graph representation includes: Multiple Layer 1s, where each Layer 1 represents a connection between individual UE pairs at a corresponding single protocol layer; and The second layer represents the connection between various UE pairs across multiple protocol layers.

[0016] In some embodiments, the generator and discriminator include a corresponding multi-layer graph neural network (mGNN).

[0017] Other embodiments include various network nodes or functions (NNFs) configured to perform operations corresponding to the exemplary methods described above. Other embodiments include non-transitory computer-readable media storing computer-executable instructions, which, when executed by processing circuitry, configure such NNFs to perform operations corresponding to the exemplary methods described above.

[0018] These and other embodiments described herein offer various benefits and / or advantages. For example, by using a combination of GNN and GAN, the embodiments can simultaneously and efficiently optimize connectivity and QoS parameters for each VAL group, as well as inter-group connectivity within each protocol layer, thereby providing optimal overall network configuration. In this way, the embodiments assign each UE to the optimal VAL group and determine the best connectivity for UEs within and between VAL groups, while satisfying any relevant SLAs. In this way, the embodiments can automate the highly site-specific VAL group configuration process in 5G networks, while providing the necessary QoS and QoE to meet any relevant SLAs. At a high level, the embodiments improve network operation to support vertical applications such as mining, chemical or food processing, and automotive manufacturing.

[0019] These and other objects, features and advantages of this disclosure will become apparent when reading the following detailed description with reference to the accompanying drawings. Attached Figure Description

[0020] Figures 1-2 It illustrates various aspects of the 5G network architecture.

[0021] Figure 3 A functional model block diagram of SEAL-based group management is shown.

[0022] Figure 4 An example implementation of the embodiments of this disclosure in a 5G network is shown.

[0023] Figure 5 An example arrangement of VAL groups in an IIoT environment according to various embodiments of this disclosure is shown.

[0024] Figures 6-7 Example functional architectures for VAL group determination are shown according to various embodiments of this disclosure.

[0025] Figure 8Exemplary methods (e.g., processes) for use in a communication network according to various embodiments of this disclosure are shown.

[0026] Figure 9 Exemplary communication networks that can implement various embodiments of this disclosure are shown.

[0027] Figure 10 Exemplary network nodes that can implement various embodiments of this disclosure are shown.

[0028] Figure 11 An exemplary main computing system that can implement various embodiments of the present disclosure is shown.

[0029] Figure 12 Exemplary virtualization environments are shown that can implement various embodiments of this disclosure. Detailed Implementation

[0030] Some embodiments contemplated herein will now be described more fully with reference to the accompanying drawings. However, other embodiments are included within the scope of the subject matter disclosed herein, and the disclosed subject matter should not be construed as being limited to the embodiments set forth herein; rather, these embodiments are provided by way of example to convey the scope of the subject matter to those skilled in the art.

[0031] Generally, all terms used herein should be interpreted according to their ordinary meaning to one of ordinary skill in the art, unless a different meaning is expressly defined and / or implied in the context of use. Unless otherwise expressly stated or clearly implied in the context of use, all references to elements, devices, components, apparatuses, steps, etc., should be openly interpreted as referring to at least one instance of an element, device, component, apparatus, step, etc. The operation of any methods and / or processes disclosed herein need not be performed in the exact order disclosed, unless an operation is explicitly described as occurring after or before another operation, and / or implies that an operation must occur after or before another operation. Any feature of any embodiment disclosed herein may be applied, where appropriate, to any other disclosed embodiment. Similarly, any advantage of any embodiment described herein may be applied, where appropriate, to any other disclosed embodiment.

[0032] In addition, the following terms are used throughout the description given below: Radio Access Node: As used herein, a “radio access node” (or equivalent “radio network node,” “radio access network node,” or “RAN node”) can be any node in a Radio Access Network (RAN) that operates to wirelessly transmit and / or receive signals. Some examples of radio access nodes include, but are not limited to, base stations (e.g., gNBs in 3GPP 5G / NR networks or enhanced or eNBs in 3GPP LTE networks), base station distributed components (e.g., Central Units (CUs) and Distributed Units (DUs)), high-power or macro base stations, low-power base stations (e.g., micro base stations, pico base stations, femto base stations, or femtocells), integrated access backhaul (IAB) nodes, transmission points (TPs), transmission reception points (TRPs), remote radio units (RRUs or remote radio heads (RRHs)), and relay nodes.

[0033] Core Network Nodes: As used in this document, a “core network node” is any type of node in the core network. Some examples of core network nodes include, for example, a Mobility Management Entity (MME), a Serving Gateway (SGW), a PDN Gateway (P-GW), a Policy and Charging Rules Function (PCRF), an Access and Mobility Management Function (AMF), a Session Management Function (SMF), a User Plane Function (UPF), a Charging Function (CHF), a Policy Control Function (PCF), an Authentication Server Function (AUSF), a Location Management Function (LMF), and so on.

[0034] Wireless Device: As used herein, a “Wireless Device” (or “WD”) is any type of device capable of, configured, positioned, and / or operable to communicate wirelessly with network nodes and / or other wireless devices. Wireless communication may involve transmitting and / or receiving wireless signals using electromagnetic waves, radio waves, infrared light, and / or other signal types suitable for transmitting information through the air. Unless otherwise stated, the term “wireless device” is used interchangeably herein with the term “user equipment” (or “UE”), both of which have a different meaning from the term “network node”.

[0035] Radio node: As used herein, “radio node” can be a “radio access node” (or equivalent term) or a “wireless device”.

[0036] Network Node: As used herein, a “network node” is any node that is part of a radio access network (e.g., a radio access node or equivalent term) or core network (e.g., the core network node discussed above) of a cellular communication network. Functionally, a network node is a device that is capable of, configured, positioned, and / or operable to communicate directly or indirectly with wireless devices and / or with other network nodes or devices in the cellular communication network to enable and / or provide wireless access to the wireless devices, and / or perform other functions (e.g., management) in the cellular communication network.

[0037] Node: As used herein, the term “node” (without prefix) can be any type of node in or with a wireless network (including the RAN and / or core network), including radio access nodes (or equivalent terms), core network nodes, or wireless devices. However, the term “node” may be limited to a particular type (e.g., radio access node, IAB node) based on its specific characteristics in any given context.

[0038] The foregoing definitions are not intended to be exclusive. In other words, the various terms described above may be interpreted and / or described elsewhere in this disclosure using the same or similar terms. However, if any such other interpretation and / or description conflicts with the foregoing definitions, the foregoing definitions shall prevail.

[0039] Please note that the descriptions presented herein focus on 3GPP cellular communication systems; therefore, 3GPP terminology or similar terms are frequently used. However, the concepts disclosed herein are not limited to 3GPP systems but can be applied to any communication system from which one may benefit.

[0040] At a high level, a 5G system (5GS) consists of an access network (AN) and a core network (CN). The AN provides connectivity from the UE to the CN, for example, via base stations (such as gNBs or ng-eNBs). As described in more detail below, the CN includes various network functions (NFs) that provide a wide range of different functions, such as session management, connection management, accounting, authentication, etc.

[0041] Figure 1 A high-level view of an exemplary 5G network architecture is shown, consisting of a Next Generation Radio Access Network (NG-RAN, 199) and a 5G core network (5GC, 198). NG-RAN may include one or more gNodeBs (gNBs) connected to the 5GC via one or more NG interfaces, such as gNBs (100, 150) connected via corresponding interfaces (102, 152). More specifically, gNBs may connect to one or more Access and Mobility Management Functions (AMFs) in the 5GC via corresponding Next Generation Control Plane (NG-C) interfaces and to one or more User Plane Functions (UPFs) in the 5GC via corresponding Next Generation User Plane (NG-U) interfaces. The 5GC may include various other network functions (NFs), such as Session Management Functions (SMFs).

[0042] Furthermore, gNBs can interconnect via one or more Xn interfaces, such as the Xn interface (140) between gNBs (100, 150). NG-RAN's radio technology is commonly referred to as "New Radio" (NR). Regarding the NR interface to the UE, each gNB can support Frequency Division Duplexing (FDD), Time Division Duplexing (TDD), or a combination thereof. Each gNB can serve a geographic coverage area comprising one or more cells, and in some cases, various directional beams can be used to provide coverage within the corresponding cell. Typically, a DL "beam" is the coverage area of ​​a reference signal (RS) transmitted by the network, which the UE can measure or monitor.

[0043] NG-RAN is layered into the Radio Network Layer (RNL) and the Transport Network Layer (TNL). The NG-RAN architecture (i.e., NG-RAN logical nodes and the interfaces between them) is defined as part of the RNL. For each NG-RAN interface (NG, Xn, F1), the associated TNL protocols and functions are specified. The TNL provides user plane transport and signaling transport services.

[0044] An NG RAN logical node (e.g., gNB 100) comprises a central unit (CU or gNB-CU, e.g., 110) and one or more distributed units (DUs or gNB-DUs, e.g., 120, 130). The CU is a logical node that hosts higher-level protocols and performs various gNB functions (e.g., controlling the operation of the DU). The DU is a distributed logical node that hosts lower-level protocols and, depending on the function splitting options, may include various subsets of gNB functions. Each CU and DU may include various circuitry required to perform its respective functions, including processing circuitry, communication interface circuitry (e.g., transceivers), and power supply circuitry.

[0045] gNB-CU communicates through the corresponding F1 logic interface (e.g., Figure 1 As shown in Figures 122 and 132, the interface connects to one or more gNB-DUs. However, a gNB-DU can only connect to a single gNB-CU. The gNB-CU and its connected gNB-DUs are only visible as a single gNB to other gNBs and 5GCs. In other words, the F1 interface is not visible outside of the gNB-CU.

[0046] Figure 2 An exemplary non-roaming reference architecture for a 5G network (200) is shown, including the following NFs and service-based interfaces defined by 3GPP: Application Functions (AFs, with a Naf interface) interact with the 5GC to provide information to the network operator and subscribe to certain events occurring within the operator's network. The AF provides applications whose services are delivered at a different layer (i.e., the transport layer) than the layer requesting the service (i.e., the signaling layer), and controls flow resources based on the results of negotiations with the network. The AF (via the N5 interface) transmits dynamic session information (including a description of the media to be transmitted by the transport layer) to the PCF.

[0047] The Policy Control Function (PCF, with an NPCF interface) supports a unified policy framework for managing network behavior by providing the SMF with PCC rules (e.g., regarding the processing of each service data flow controlled by the PCC) via the N7 reference point. The PCF provides the SMF with policy control decisions and flow-based charging control, including service data flow detection, gating, QoS, and flow-based charging (excluding credit management). The PCF receives session and media-related information from the AF and notifies the AF of traffic plane events (or user plane events).

[0048] User Plane Function (UPF) – Supports processing user plane traffic based on rules received from the SMF, including packet inspection and various execution actions (e.g., event detection and reporting). The UPF communicates with the RAN (e.g., NG-RNA) via reference point N3, with the SMF (discussed below) via reference point N4, and with the external packet data network (PDN) via reference point N6. Reference point N9 is used for communication between two UPFs.

[0049] The Session Management Function (SMF, with an NSMF interface) interacts with the decoupled traffic (or user) plane, including creating, updating, and removing Protocol Data Unit (PDU) sessions, and managing session contexts with User Plane Functions (UPFs), such as for event reporting. For example, the SMF performs flow inspection (based on filter definitions included in PCC rules), online and offline billing interactions, and policy enforcement.

[0050] The Billing Function (CHF, with an Nchf interface) is responsible for both online and offline billing functions. It provides quota management (for online billing), reauthorization triggers, rate conditions, and receives notifications from the SMF regarding usage reports. Quota management involves granting a specific number of units (e.g., bytes, seconds) to a service. The CHF also interacts with the billing system.

[0051] The Access and Mobility Management Function (AMF, with a Namf interface) terminates the RAN CP interface and handles all mobility and connectivity management for the UE (similar to the MME in the EPC). The AMF communicates with the UE via the N1 reference point and with the RAN (e.g., NG-RAN) via the N2 reference point.

[0052] The Network Open Function (NEF), with its Nnef interface, serves as an entry point into the operator's network by securely opening network capabilities and events provided by the 3GPP NF to the AF, and providing the AF with a means to securely provide information to the 3GPP network. For example, the NEF provides services that allow the AF to configure specific subscription data (e.g., expected UE behavior) for various UEs. Typically, the services provided by the NEF are similar to those provided by the SCEF in the EPC.

[0053] The Network Repository Function (NRF), with its Nnrf interface, provides service registration and discovery, enabling NFs to identify the appropriate services available from other NFs.

[0054] Network Slice Selection Function (NSSF), with an NNSSF interface—a “network slice” is a logical partition of a 5G network that provides specific network capabilities and features, such as supporting specific services. A network slice instance is a set of NF instances and the network resources (e.g., compute, storage, communication) required to provide the capabilities and features of the network slice. NSSF enables other NFs (such as AMFs) to identify network slice instances suitable for the services required by the UE.

[0055] The Authentication Server Function (AUSF), with a Nausf interface – is located in the user’s home network (HPLMN) and performs user authentication and calculates security key material for various purposes.

[0056] Network Data Analysis Function (NWDAF) with Nnwdaf interface – provides network analysis information (e.g., statistics and / or predictions of past events) to other NFs at the network slice instance level.

[0057] The Unified Data Management (UDM) function, with a Nudm interface, generates 3GPP authentication credentials, handles user identification, authorizes access based on subscription data, and provides other subscriber-related functions. The UDM retrieves subscription data (including authentication data) from the 5G Unified Data Repository (5GS-UDR, with a Nudr interface). The 5GS-UDR supports PCF for storing and retrieving policy data, and NEF for storing and retrieving application data.

[0058] As briefly mentioned above, 3GPP has specified a Service Enablement Architecture Layer (SEAL) to support vertical applications on 5G networks. SEAL services include group management, configuration management, location management, identity management, key management, and network resource management. SEAL provides these services to the Vertical Application Layers (VALs) that can run on top of SEAL, which correspond to various applications, for example.

[0059] SEAL can use various information for VAL group management, including VAL group ID, group member UE ID, general group configuration, group data network name (DNN), single network slice selection assistance information (S-NSSAI), group size, group leader, group location information, enabled VAL services, VAL service-specific information, PDU session type, etc. This information is further specified in Clause 10 of 3GPP TS 23.434 (v17.7.0) and Clause 7.2.1.4 of 3GPP TS 29.549 (v17.8.0).

[0060] Figure 3 A functional model block diagram of SEAL-based group management is shown. A VAL UE may include a SEAL group management client, which communicates with a corresponding SEAL group management server outside the 3GPP network via a GM-UU interface. This group management client provides group management services to the UE VAL client within the UE via a GM-C interface. The VAL client communicates with a corresponding VAL server outside the 3GPP network via a VAL-UU interface. The VAL server also communicates with the SEAL group management server via a GM-S interface.

[0061] The group management server interacts with the NEF of the 3GPP network via the N33 reference point to perform group management procedures for 5G Virtual Network (5GVN) groups. Group management functions defined by 3GPP SA6 include group creation, group information query, group member update, group declaration and joining, and group member departure.

[0062] Members of a VAL group can communicate with each other, but are "invisible" to members of other VAL groups. Even so, inter-group communication is still possible via the N19 reference point in 5GC, which connects two UPFs for direct routing of traffic between different PDU sessions without using [other methods]. Figure 2 The N6 reference point is shown.

[0063] As briefly mentioned above, NWDAF in 5GC provides network analytics information (e.g., statistics and / or predictive information of past events) to other NFs at the network slice instance level. Note that a "network slice" is a logical partition of a 5G network that provides specific network capabilities and features, for example, to support specific services. A network slice instance is a set of NF instances and the network resources (e.g., compute, storage, communication) required to provide the capabilities and features of the network slice.

[0064] NWDAF can collect data from any 5GC NF as well as from Network Operations, Administration and Maintenance (OAM) and other cloud, edge, or external data sources. 3GPP TS 23.288 (v17.9.0) designates NWDAF as the primary NF for calculating analytics reports and divides NWDAF into two sub-functions (or logical functions): the Analytics Logical Function (AnLF), which executes analytics procedures; and the Model Training Logical Function (MTLF), which performs the training and retraining of the ML model used by AnLF.

[0065] Using rule-based or machine learning (ML) algorithms, NWDAF implements various analytics techniques to support network operations, such as service orchestration, automated network configuration, and automated network operations. NWDAF can provide analytics results based on subscriptions and in response to requests. NWDAF can also be combined with PCF in 5GC to achieve closed-loop policy optimization.

[0066] To further illustrate this with examples, the NWDAF functionality can be used in one or more of the following use cases in 5G networks: Service quality, service experience, SLA monitoring and assurance; Detect and resolve performance issues; Anticipate network problems and proactively adapt to network conditions; Maximize the return on investment in network capacity; Optimize network resources; Network load performance monitoring and analysis and load prediction; Network slicing protection; Subgroup service quality monitoring; Abnormal behavior and anomaly detection; Mobility-related information and predictions; and Congestion control.

[0067] Advanced analytics systems with NWDAF capabilities rely on collecting and correlating basic network events from different network domains, such as core networks, radio networks, and transport networks. These systems calculate end-to-end (E2E) Service Quality Key Performance Indicators (S-KPIs) at the user and session levels, as well as Radio and Network Resource Metrics (R-KPIs) characterizing the radio environment or network operation at the user and session levels. These types of solutions are suitable for session-based troubleshooting and network problem analysis.

[0068] Event-based analytics requires real-time collection and correlation of node and protocol events from different RAN and CN nodes, probe signaling interfaces, and user plane traffic sampling. Furthermore, event-based analytics necessitates advanced databases, rule engines, and artificial intelligence (AI) frameworks that facilitate closed-loop retraining of the ML models being used.

[0069] However, the information specified by 3GPP for VAL group management does not include UE communication characteristics that can be crucial in practical applications. For example, in Industrial Internet of Things (IIoT) applications (such as mining, chemical or food processing, and automotive manufacturing), VAL group management should consider multiple site-specific parameters of communication characteristics. These parameters may include connectivity quality of service (QoS) data, guaranteed bit rate and maximum bit rate, packet delay budget, and packet error rate. Furthermore, each of these parameters can relate to communication within the group as well as to communication between users in different groups.

[0070] Furthermore, traditional VAL group management / configuration is done manually by programmers, who rely to some extent on site-specific parameters, historical data, and their own experience. This challenging task becomes even more difficult due to inevitable changes in site specifications and / or physical layout. Therefore, manual VAL group configuration may fail to meet the QoS and / or Quality of Experience (QoE) requirements for intra- and inter-group communication as mandated by Service Level Agreements (SLAs). Moreover, manual VAL group configuration is unlikely to produce optimal resource allocation within a 5G system, instead leading to traffic congestion, UE and / or network resource unavailability, and QoS degradation.

[0071] Embodiments of this disclosure address one or more of these and related problems, challenges, and / or difficulties by providing techniques and apparatus for automatically configuring VAL groups based on historical and / or predicted communication parameters in a communication network (e.g., a 5G network).

[0072] The implementation can determine the preferred and / or optimal VAL group configuration for any input architecture, including group membership and inter-group connectivity for each UE. In this way, the implementation can automate the highly site-specific VAL group configuration process in 5G networks. Furthermore, the determined VAL group configuration can provide the necessary QoS and QoE for each service to meet any relevant SLAs.

[0073] The embodiments are based on Convolutional Graph Neural Networks with Generative Adversarial Networks (GraphCGAN), which can handle multi-layered (e.g., protocol layers) graph neural networks (GNNs) and any number of graph edges between any two graph nodes. Some embodiments include and / or utilize storage areas for historical VAL group configurations (e.g., group connectivity) and corresponding performance (e.g., connection QoS data).

[0074] Furthermore, the embodiments utilize a novel mapping of analytical data (e.g., in NWDAF) to QoS parameters (e.g., 5QI) for intra- and inter-group VAL connections. Additionally, the embodiments utilize flow-by-flow observation and analysis to obtain, on-demand, Key Performance Indicators (KPIs) for any parameter aggregation, such as by VAL group, by application, or a combination thereof.

[0075] Generative modeling involves automatically identifying and learning regularities or patterns in input data to produce a model, which can then be used to generate new examples that may reasonably originate from the original dataset. Generative adversarial networks (GANs) are a type of generative modeling approach that involves deep learning techniques, such as convolutional neural networks.

[0076] In GANs, two neural networks compete against each other in a zero-sum game, where the gain of one agent is the loss of the other. Given a training set, a GAN learns to generate new data with the same statistical properties as the training set. For example, a GAN trained on photographs can generate new photos that appear at least superficially real to a human observer, possessing many realistic features. Instead of being trained to minimize the distance to a particular image, the generator model is indirectly trained to "fool" a second model called the "discriminator." In other words, the generator model is trained to create new instances, while the discriminator model is trained to classify examples as real (from the training domain) or fake (from the generator). Typically, when both models are trained together in an adversarial zero-sum game, the discriminator model is fooled about half the time. This indicates that the generator model is producing believable examples.

[0077] Graphs can be used to model structured and relational data, such as social networks, citation networks, biological networks, the World Wide Web, and physical networks. Typically, a graph is a collection of objects (nodes) and a set of interactions (edges) between these object pairs. For example, to encode a social network as a graph, we can use nodes to represent individuals and edges to represent that two individuals are friends.

[0078] Graph Neural Networks (GNNs) are a technique used to apply convolutional operators, the core of convolutional neural networks (CNNs) and deep learning, to graph-structured data. While CNNs are well-defined for Euclidean data (e.g., images as a grid of pixels), convolutional operators require new formulas for non-Euclidean domains (as shown in the figure). Within each convolutional layer, a kernel (i.e., a learned tensor) is used to multiply and aggregate the neighborhood of each graph node using some learned function. The output value provided by the aggregation is a higher-level feature that will be fed into the next layer of the model. Thus, by stacking many layers together, higher-level features that capture some higher-order pattern can be obtained.

[0079] A key component of the GNN architecture is pairwise message passing, which enables graph nodes to iteratively update their representations by exchanging information with their neighbors. Other reasons for the success of GNNs include their generally low computational complexity due to the locality of convolution operators, and their ability to generalize to previously unseen nodes and graphs.

[0080] As briefly mentioned above, the embodiments are based on a technique called "Convolutional Graph Neural Network with Generative Adversarial Network (GraphCGAN)," which combines GNN and GAN. The applicant has recognized that this technique is compatible with physical networks that include inter-device connections across multiple protocol layers, and in particular, generates graph attributes for multiple connections across multiple layers.

[0081] The GAN utilized in this disclosure includes a generator and a discriminator working together, wherein the generator is configured to generate an artificial GNN for candidate VAL groups, and the discriminator evaluates the generated artificial GNN. Typically, the generator learns to map from the latent space (e.g., a multivariate normal distribution) to the data distribution of interest, while the discriminator distinguishes between candidate and real data distributions generated by the generator. The training objective of the generator is to improve the error rate of the discriminator, i.e., to "fool" the discriminator by configuring novel artificial GNNs (which the discriminator considers to be part of the real data distribution) for candidate VAL groups.

[0082] The known dataset is used as the initial training data for the discriminator. Training involves presenting the discriminator with samples from the training dataset until it reaches acceptable accuracy. The generator is trained based on whether it successfully fools the discriminator. Candidates synthesized by the generator are evaluated by the discriminator. Independent backpropagation processes are applied to adjust the parameters of the generator and discriminator, so that the generator produces better samples (e.g., with a smaller mean squared error (MSE)) and the discriminator becomes better at recognizing artificial GNNs.

[0083] These technologies can offer a variety of benefits, advantages, and / or solutions to the problems described herein. For example, NWDAF can measure parameters end-to-end to acquire and store historical data aggregated at different levels (e.g., by application, by VAL group, by UE, by tracking area, by cell, etc.). NWDAF can access network-wide data for each VAL group, thereby allowing the configuration of optimal VAL groups across multiple network protocol layers (e.g., L2, L3 / L4, L7, VAL, etc.).

[0084] Furthermore, by using a combination of GNN and GAN, the implementation can simultaneously and efficiently optimize connectivity and QoS parameters for each VAL group, as well as inter-group connectivity within each protocol layer, thereby providing optimal overall network configuration. In this way, the implementation assigns each UE to the optimal VAL group and determines the best connectivity for UEs within and between VAL groups, while satisfying any relevant SLAs.

[0085] Figure 4 An example implementation of this disclosure in a 5G network is illustrated. In this example, VAL groups are determined to be functions or blocks (410) managed by an NWDAF (400), which may also include or have access to a database (420) of historical VAL group data. The 5G network may include various network slices, each of which may include several VAL groups with different QoS requirements. For example, a network slice dedicated to a factory may contain VAL groups for automated guided vehicles, autonomous mobile robots, or mobile control panels. An automotive network slice may utilize its own set of VAL groups.

[0086] The VAL group determination function in the NWDAF communicates with the group management server (GM-Server), which is part of the SEAL function (497), via the NEF. The GM server also communicates with the SEAL AF, which in turn communicates with external data networks (e.g., the Internet) coupled to the RAN (e.g., NG-RAN) via the UPF. The NWDAF also communicates with the SMF, AMF, UPF, and other NFs in the 5GC (498) in a conventional manner.

[0087] Figure 5 This diagram illustrates an example arrangement of VAL groups in an IIoT environment. Each of VAL groups 1-3 includes multiple UEs that can communicate within the group, (in some cases) with UEs in other VAL groups, and with network devices such as edge servers. A VAL layer consists of any number of sublayers, each belonging to a different VAL group. Each VAL group isolates traffic for a specific application. VALs reside above Layer 7 of the Open System Interconnection (OSI) protocol (also known as the application layer). UEs within a VAL group can communicate at Layer 7 using well-known protocols such as HTTP, SNMP, etc.

[0088] Below the application layer are layers 3 and 4, also known as the transport layer and network layer, respectively. UEs in the VAL group can communicate with each other and / or with edge servers at layers 3 and 4 using well-known protocols (e.g., TCP, UDP, IP, etc.). Below that are layers 1 and 2, also known as the data link layer and physical layer, respectively. UEs in the VAL group can use common layer 1 / 2 protocols, such as Medium Access Control (MAC), Radio Link Control (RLC), and the Physical Layer (PHY), for wireless communication with the 5G network. Alternatively, UEs in the VAL group can use common wired communication protocols, such as Ethernet-related protocols, at layers 1 and 2. In some embodiments, each layer in layers 2 and 3 may involve two sublayers to support the GPRS Tunneling Protocol (GTP) in 5GC.

[0089] According to embodiments of this disclosure, the characteristics of connectivity between devices in Layer 2 (e.g., Ethernet, RLC / MAC), Layers 3-4 (e.g., TCP / IP), and Layer 7 (e.g., HTTP) are represented by diagrams in three independent layers. This is in Figure 5The diagram illustrates the nodes and edges of an example graph. Note that graph edges can be within or between protocol layers. Each edge represents a connection between UEs via the network (or directly via a side link) and is assigned attributes (e.g., key performance indicators, KPIs) related to the connection quality between the two nodes (UEs) connected to that edge, as explained in more detail below.

[0090] Based on this diagram, the embodiment determines the optimal arrangement that satisfies the following network-level SLAs: Connectivity of VAL group members in the VAL layer, including the 5G QoS Identifier (5QI) and connection QoS settings; and Member UEs are connected to the UE-network-UE network at layers 2-3 and 4, including each individual VAL group and the connections between VAL groups.

[0091] Figure 6 An example functional architecture for VAL group determination is shown according to some embodiments of this disclosure. This architecture will be interpreted in terms of its various functions or operations, some of which are labeled with numbers. However, unless explicitly stated otherwise, these labels are intended for ease of interpretation and not to require any particular order of operations.

[0092] Initially, the system starts with the training set configured in the historical VAL group. Figure 6 Group configurations 1-3 are shown as examples. These historical group configurations can include group configurations generated by the system itself and / or group configurations manually generated by the operator. For example, each of group configurations 1-3 may correspond to the past configuration of the UE in each VAL group (e.g., Figure 5 This includes connections on layers 2, 3-4, and 7. In Operation 1, select an example VAL group configuration from these configurations.

[0093] In Operation 2, based on the example VAL group configuration from Operation 1 and historical data on connections at layers 2-4 associated with the example VAL group configuration, a real multi-layer attribute graph is calculated or prepared. This historical data can be provided by NWDAF. The table below shows various statistics defined in 3GPP TS 23.288 (v17.9.0), which are output by NWDAF and can be used as historical data for applications corresponding to each VAL group.

[0094]

[0095]

[0096] For the graph prepared in Operation 2, the example VAL group configuration provides the node / edge structure of the graph, while the NWDAF historical data provides the graph's attributes. For example, the graph can be arranged as multiple matrices—one matrix per layer—each matrix entry containing attributes of the connections between two UEs via the data network in the example VAL group configuration. In other words, these attributes are for UE-network-UE (UNU) connections supported by the NWDAF historical data shown in the table above.

[0097] If the training set for the historical VAL group configuration is unavailable, NWDAF historical data can be used to determine the example real VAL group instances needed to obtain the true multilayer attribute graph.

[0098] Initial graph properties can be set in several ways, including the following examples. Let... This indicates the first VAL layer. The first in the sub-layer The integer priority value of the connection represented by each graph edge. This value can be calculated using detailed historical connectivity data according to the following techniques.

[0099] The 5G QoS model is based on QoS flows and supports both QoS flows that require guaranteed flow bit rate (GBR QoS flows) and QoS flows that do not require guaranteed flow bit rate (non-GBR QoS flows). A QoS flow is the finest granular level of QoS differentiation within each UE PDU session, identified by an assigned QoS Flow ID (QFI). UP traffic with the same QFI within a PDU session receives the same traffic forwarding processing (e.g., scheduling, admission thresholds). Each QoS flow is characterized by the following: QoS profiles provided to the RAN by the SMF via the AMF; One or more QoS rules and optional QoS flow-level QoS parameters associated with these QoS rules, which can be provided to the UE by the SMF via the AMF; and One or more UL and DL packet detection rules (PDRs) provided by SMF to UPF.

[0100] For each QoS flow, the QoS profile includes a 5G QoS identifier (5QI) and an Allocation and Retention Priority (ARP). For each non-GBR QoS flow, the QoS profile may also include a Reflective QoS Attribute (RQA). For each GBR QoS flow, the QoS profile also includes Guaranteed Flow Bit Rate (GFBR) - UL and DL parameters and Maximum Flow Bit Rate (MFBR) - UL and DL parameters.

[0101] 5QI is an integer or index referring to the 5G QoS features defined in Clause 5.7 of 3GPP TS 23.501 (v17.9.0), including resource type, priority level, packet delay budget, packet error rate, average window, and maximum data burst capacity. Typically, these are access node-specific parameters that control the QoS forwarding processing of the QoS flow between the UE and UPF for edge-to-edge reception (e.g., scheduling weights, admission thresholds, queue management thresholds, link layer protocol configuration, etc.).

[0102] Therefore, if the training set of the historical VAL group configuration or the historical NWDAF data of the UNU connection in the graph includes 5QI, then The following can be calculated: .

[0103] On the other hand, if the training set of the historical VAL group configuration or the historical NWDAF data of the UNU connection in the graph does not include 5QI, then The calculation can be performed as follows: , , in It uses one or more data network (DN) performance statistics listed in the table above to determine the VAL layer's first... Graph edges in each sub-layer The evaluation priority level of the calculation. AppID identifier mapped to sub-layer. For specific service types, NWDAF performance statistics are obtained and 5QI is calculated. For example, Map the DN performance statistics of maximum packet latency and maximum packet loss rate to the potential 5QI value of AppID within a historical time period T.

[0104] Assuming the UE's subscriber profile allows for edge... For GBR connections, this technology can predict whether GBR QoS is suitable for the edge based on the following comparisons. : , The minimum flow rate is the DN performance statistics in the table above. and These are predefined values. If the above comparison does not hold, then the edge is determined. For non-GBR connections, for example, there are .

[0105] Finally, by analyzing the network status indicators listed in the table above { Percent UL , Percent DL Statistical information, from the shortlisted potential 5QI values, for the edge Assign a specific 5QI value. In other words, assign a specific 5QI value to meet network QoS characteristics, including packet delay budget, packet loss, or priority. An example is given in Table 5.7.4-1, “Standardized 5QI to QoS Feature Mapping,” of 3GPP TS 23.501 (v17.9.0). Subsequently, for the edge ( j , k The 5QI value is assigned and determined.

[0106] In some embodiments, if the edge is determined For GBR, the minimum and maximum flow rates of the DN performance statistics in the table above are also assigned as attributes of that edge. Note that these statistics (e.g., measured by NWDAF) correspond to the Guaranteed Flow Bit Rate (GFBR) and Maximum Flow Bit Rate (MFBR) parameters for the GBR QoS flow, respectively.

[0107] According to Clause 5.8.2.11.4 of 3GPP TS 23.501 (v17.9.0), the QoS Enforcement Rule (QER) defines how packets should be handled in terms of bit rate limiting and packet marking for QoS purposes. This QER definition includes "maximum bit rate" and "minimum bit rate" attributes, which define the maximum and minimum uplink / downlink bit rates to be enforced for the packets. In some embodiments, either or both of these attributes can be assigned as attributes corresponding to graph edges (j, k) of the connection. However, note that the minimum bit rate attribute applies only to non-GBR QoS flows.

[0108] In addition to connections within each VAL group, some embodiments may also include any connections between VAL groups or between sublayers in the graph representation. In 5G networks, inter-group connections are implemented via the N19 interface between UPFs in the 5GC. In such embodiments, attributes can be assigned to edges corresponding to intra-group connections using NWDAF historical data in a manner similar to that described above.

[0109] Operations 3-4 involve generating an artificial multi-level attribute map, which is then compared by a discriminator function with the real multi-level attribute map generated in Operation 2. Operation 3 begins with the generator receiving a random input seed, which is used to generate a random VAL group configuration. Figure 6 The term "configuration X" is used. For example, it generates random values ​​with a uniform distribution in the range (0,1) for each input neuron of a multi-layer GNN. Optionally, a preferred VAL group configuration can be used as input to generate "configuration X". For example, a user-preferred group definition can be introduced as an SLA criterion and / or a general generator strategy. In operation 4, a graph (or GNN) of nodes and edges is generated for configuration X, including nodes and edges similar to the real structure discussed above. The generator takes a random seed and prepares the generated structure for the VAL group.

[0110] Furthermore, NWDAF predictions for QoS, etc., are used to populate the graph attributes determined in Operation 4. In other words, Operation 4 assigns attributes to graph edges in a similar manner to Operation 2, but uses predicted data instead of historical data. For example, a similar technique can be used to convert NWDAF predictions into 5QI values, and then into attributes to be assigned as graph edge attributes. .

[0111] The mapping between NWDAF historical data and predicted edge attributes to be applied to the VAL group configuration in operations 2 and 4. Figure 6 The mapping is indicated as to be performed by NWDAF. However, this mapping may be performed outside of NWDAF, such as by a SEAL group management server, AF, or other NF that can access historical or predicted NWDAF data.

[0112] In Operation 5, the discriminator function of GraphCGAN compares the real multi-level attribute graphs and the artificial multi-level attribute graphs generated in Operations 2 and 4, respectively. For example, the graphs can be compared based on their respective adjacency matrices or adjacency list representations. This operation can be considered part of GAN training. The discriminator also checks whether the VAL group configuration corresponding to the artificial multi-level attribute graph satisfies the relevant network-level SLAs for the VAL group. For example, SLAs include target criteria for QoS parameters such as packet loss, latency, bandwidth, etc. For VAL group configurations, the actual values ​​of these parameters can be monitored by NWDAF per session and stored as part of historical data. In the context of training, NWDAF can compare the (historical) monitored values ​​with the target values.

[0113] If the generated VAL group configuration satisfies the SLA and the discriminator cannot distinguish between the two input graphs, the configuration is considered a VAL group policy for later use, and the training process stops. If either or both of these conditions are not met, a new random input seed is triggered for the generator function, and the generator uses this random input seed to generate different random VAL group configurations in a manner similar to operation 3.

[0114] In operation 6, the VAL group policy is stored for later use by the generator function to infer VAL group configurations in response to requests or queries. In operation 7, the system is queried to obtain the new VAL group structure. This query provides some input regarding the architecture for which the VAL groups should be determined. For example, the architecture input may include a list of UEs to be placed in the VAL group, a list of applications running on the respective UEs, a list of application session IDs, etc. As a more specific example, user-preferred group definitions may be introduced as SLA guidelines and / or general generator policies.

[0115] Based on this input and the stored VAL group policies, the generator performs reasoning to determine the recommended VAL group configuration, including group membership, intra-group and inter-group connectivity configurations, etc. Furthermore, NWDAF predictions for QoS, etc., are used to populate the attributes of the graph determined in Operation 7 in a manner similar to that discussed above for Operation 4.

[0116] Figure 7 Another example functional architecture for VAL group determination according to other embodiments of this disclosure is shown. This architecture is similar to... Figure 6 The architecture shown is different, but it differs in several aspects as described below.

[0117] For example, Figure 7 The generator in the code is configured to generate GNNs with attributes, while Figure 6The generator is configured to generate a GNN without attributes. This is shown as two separate outputs during the training phase: a configuration X of the nodes and edges, and attributes of the nodes and edges. During the inference phase, the generator takes the generated node and edge configurations and their attributes as input, which are determined based on predictions of UNU connectivity QoS parameters provided by NWDAF in a manner similar to that described above. During inference, these predictions are requested from NWDAF based on a back-mapping from the architecture-dependent inputs to the desired NWDAF QoS metric.

[0118] In some embodiments, obtaining predicted attributes from NWDAF during inference can be omitted, and the attributes generated from the configuration can be used instead. However, NWDAF predictions can provide higher quality attributes compared to those generated from the configuration. Furthermore, in some cases, NWDAF predictions can improve the format of attributes according to the generator's requirements.

[0119] The embodiments described above can be referred to Figure 8 To further explain, Figure 8 Exemplary methods (e.g., procedures) for determining VAL groups (each VAL group comprising multiple UEs) operating in a communication network are described according to various embodiments of this disclosure. In other words, various features of the operation described below correspond to the various embodiments described above. Although Figure 8 The diagram illustrates an exemplary method with specific blocks in a particular order, but operations corresponding to these blocks can be performed in a different order than shown, and can be combined and / or divided into blocks with functions different from those shown. Optional blocks and / or operations are indicated by dashed lines.

[0120] Figure 10 The exemplary method shown can be performed by a network node or function (NNF) of a communication network (such as a 5G wireless network), as described in the example above.

[0121] This exemplary method includes the operation of block 810, wherein the NNF can determine a first multi-layer graph representation based on: a reference VAL group configuration previously used in the communication network, and historical quality of service (QoS) metrics for multi-layer connections between UEs including the reference VAL group configuration. This exemplary method also includes performing the first operations of blocks 820-830 using a generative adversarial network (GAN) including a generator and a discriminator, specifically the following first operations: (820) Using a generator, a second multi-layer graph representation is determined based on: candidate VAL group configurations determined based on random input, and QoS metrics predicted for multi-layer connectivity between UEs including reference VAL group configurations; (830) Using a discriminator, determine whether the following first condition is met: the second multi-layer graph representation cannot be distinguished from the first multi-layer graph representation, and the predicted QoS metric associated with the candidate VAL group configuration satisfies one or more service level agreements (SLAs) for the operation of the communication network. The exemplary method also includes the operation of block 850, wherein when the first condition is not met, the NNF can repeat the first operation based on different random inputs. The exemplary method also includes the operation of block 860, wherein after the first condition is met and using a generator, the NNF can perform inference based on a second multi-layer graph representation to generate a configuration for a VAL group operating in the communication network.

[0122] In some embodiments, the method is performed by the NWDAF of the communication network. In other embodiments, the method is performed by a SEAL Group Management (GM) server external to (or internal to) the communication network. In some of these embodiments, the method is performed by the NWDAF, and the reasoning is further based on the desired architecture of the VAL group obtained from the SEAL GM server.

[0123] In some embodiments, the configuration of the VAL groups generated by inference in execution block 860 includes the following: multiple UEs arranged in multiple VAL groups, and connections between the respective UE pairs. In some embodiments of these embodiments, the connections between the respective UE pairs include connections between UE pairs in the same VAL group and connections between UE pairs in different VAL groups.

[0124] In some embodiments of these examples, the configuration of the VAL group generated by performing inference also includes QoS settings for the connections between the individual UE pairs. Figure 7 Examples of these embodiments are shown. In other embodiments of these embodiments, the exemplary method also includes the operation of block 870, wherein the NNF can obtain the predicted QoS settings for the connection between the various UE pairs from the NWDAF of the communication network. Figure 6 Examples of these embodiments are shown.

[0125] In some embodiments, the reference VAL group configuration and each candidate VAL group configuration include the following: multiple UEs deployed in multiple VAL groups, and connections between the various UE pairs at multiple protocol layers. In some embodiments of these embodiments, the multiple VAL groups are associated with corresponding applications or services, and each of the historical QoS metrics and predicted QoS metrics is associated with data traffic in one of the applications or services.

[0126] In some embodiments of these embodiments, each of the first multi-layer graph representation and the second multi-layer graph representation includes: Multiple Layer 1s, where each Layer 1 represents a connection between individual UE pairs at a corresponding single protocol layer; and The second layer represents the connection between various UE pairs across multiple protocol layers.

[0127] In some embodiments of these embodiments, determining the first multi-layer graph representation in block 810 includes the following operations, labeled with corresponding sub-block numbers: (811) Determine the graph structure based on the reference VAL group configuration, wherein the graph structure includes multiple UEs as nodes and connections between each pair of UEs as edges; and (812) Determine the attributes of the edge based on historical QoS metrics.

[0128] In some of these embodiments, determining the attributes of an edge based on historical QoS metrics in sub-block 812 includes the following operations for each connection between UE pairs: Obtain one or more historical QoS metrics for the connection from the NWDAF of the communication network; Based on one or more historical QoS metrics, the attributes corresponding to the edges of the connection are determined as integer priority values.

[0129] In some further variations, each integer precedence value is determined based on the following: 5G QoS Identifier (5QI), when included in one or more historical QoS metrics; or When 5QI is excluded, at least one of the following historical QoS metrics: o The percentage of UL throughput for applications or services in the region of interest out of the total UL throughput; o The percentage of total DL throughput for applications or services in the region of interest; o The average traffic rate of the UE communicating with the application or service observed; o The maximum traffic rate of the UE that is observed communicating with the application or service; o The minimum traffic rate observed by the UE communicating with the application or service; o The average packet latency observed by the UE communicating with the application or service; o The maximum packet delay observed by the UE communicating with the application or service; o The observed average packet loss rate of the UE communicating with the application or service; and o The observed maximum packet loss rate of a UE communicating with an application or service.

[0130] In some embodiments of these embodiments, the generator and discriminator include corresponding multilayer graph neural networks (mGNNs). In some variations of these embodiments, the exemplary method may further include the operation of block 840, wherein when it is determined (e.g., in block 830) that a second multilayer graph representation can be distinguished from a first multilayer graph representation, the NNF may adjust the mGNN of the generator and discriminator based on a metric representing the difference between the first and second multilayer graph representations. In this case, the first operation is repeated using the adjusted mGNN of the generator and discriminator. In some further variations, after determining (e.g., in block 830) that a first condition is met, the inference in block 860 is performed using the adjusted mGNN of the generator.

[0131] In some embodiments of these embodiments, determining the second multi-layer graph representation in block 820 includes the following operations, labeled with corresponding sub-block numbers: (821) The candidate VAL group configuration is determined by applying the generator mGNN to the random input; (822) Determine the graph structure based on the candidate VAL group configuration, wherein the graph structure includes multiple UEs as nodes and connections between each pair of UEs as edges; and (823) Determine the attributes of the edge based on historical QoS metrics.

[0132] Figure 6 Examples of these variations are shown.

[0133] In other variations of these embodiments, determining the second multi-layer graph representation in block 820 includes the operation of sub-block 824, where NNF can determine an attributed graph structure for candidate VAL group configurations based on applying the generator mGNN to random input. The attributed graph structure includes multiple UEs as nodes, connections between the multiple UEs as edges, and corresponding QoS metrics as attributes of the edges. Figure 7 Examples of these variations are shown.

[0134] Although various embodiments have been described above in the form of methods, techniques and / or processes, those skilled in the art will readily understand that such methods, techniques and / or processes can be embodied in various combinations of hardware and software in various systems, communication devices, computing devices, control devices, apparatuses, non-transitory computer-readable media, computer program products, etc.

[0135] Figure 9An example of a communication system 900 according to some embodiments is shown. In this example, the communication system 900 includes a telecommunications network 902, which includes an access network 904 (e.g., a RAN) and a core network 906, the core network 906 including one or more core network nodes 908. The access network 904 includes one or more access network nodes, such as network nodes 910a-b (one or more of which may be collectively referred to as network node 910), or any other similar 3GPP access node or non-3GPP access point. Furthermore, as those skilled in the art will understand, network nodes are not necessarily limited to implementations of radio and baseband portions provided and integrated by a single vendor. Therefore, it will be understood that network nodes include decomposed implementations or portions thereof. For example, in some embodiments, the telecommunications network 902 includes one or more Open-RAN (ORAN) network nodes. An ORAN network node is a node in the telecommunications network 902 that supports ORAN specifications (e.g., specifications published by the O-RAN Alliance or any similar organization) and can operate independently or together with other nodes to perform one or more functions of any node in the telecommunications network 902 (including one or more network nodes 910 and / or core network nodes 908).

[0136] Examples of ORAN network nodes include Open Radio Units (O-RUs), Open Distributed Units (O-DUs), Open Central Units (O-CUs) (including the O-CU Control Plane (O-CU-CP) or the O-CU User Plane (O-CU-UP)), RAN Intelligent Controllers (near real-time or non-real-time) with managed software or software plug-ins (such as near real-time control applications (e.g., xApp) or non-real-time control applications (e.g., rApp)), or any combination thereof (the adjective "open" indicates support for the ORAN specification). Network nodes can support the specification through, for example, interfaces defined by the ORAN specification (e.g., A1, F1, W1, E1, E2, X2, Xn interfaces, Open Fronthaul User Plane Interface, or Open Fronthaul Management Plane Interface). Furthermore, ORAN access nodes can be logical nodes within physical nodes. Additionally, ORAN network nodes can be implemented in a virtualized environment (described further below), where one or more network functions are virtualized. For example, a virtualized environment may include an O-Cloud computing platform orchestrated by a service management and orchestration framework via an O-2 interface or similar technology defined by the O-RAN Alliance. Network node 910 facilitates direct or indirect connections for UEs, such as connecting UEs 912a-d (which may be collectively referred to as UE 912) to the core network 906 via one or more wireless connections.

[0137] Exemplary wireless communication via a wireless connection includes transmitting and / or receiving wireless signals using electromagnetic waves, radio waves, infrared light, and / or other signal types suitable for transmitting information without the need for wires, cables, or other conductors. Furthermore, in various embodiments, the communication system 900 may include any number of wired or wireless networks, network nodes, UEs, and / or any other components or systems that can facilitate or participate in communication of data and / or signals via a wired or wireless connection. The communication system 900 may include, and / or interface with, any type of communication, telecommunications, data, cellular, radio network, and / or other similar type of system.

[0138] UE 912 can be any of a variety of communication devices, including wireless devices that are arranged, configured, and / or operable to communicate wirelessly with network node 910 and other communication devices. Similarly, network node 910 is arranged, capable of, configured, and / or operable to communicate directly or indirectly with UE 912 and / or directly or indirectly with other network nodes or devices in telecommunication network 902 to achieve and / or provide network access (e.g., wireless network access) and / or perform other functions (e.g., management in telecommunication network 902).

[0139] In the depicted example, core network 906 connects network node 910 to one or more hosts, such as host 916. These connections can be direct or indirect (via one or more intermediate networks or devices). In other examples, network nodes can be directly coupled to hosts. Core network 906 includes one or more core network nodes (e.g., 908) constructed with hardware and software components. The characteristics of these components can be substantially similar to those described with respect to UEs, network nodes, and / or hosts, such that the description generally applies to the corresponding components of core network node 908. Exemplary core network nodes include the functions of one or more of the following: Mobile Switching Center (MSC), Mobility Management Entity (MME), Home Subscriber Server (HSS), Access and Mobility Management Function (AMF), Session Management Function (SMF), Authentication Server Function (AUSF), Subscription Identifier De-concealing Function (SIDF), Unified Data Management (UDM), Security Edge Protection Proxy (SEPP), Network Open Function (NEF), and / or User Plane Function (UPF).

[0140] Host 916 may be owned or controlled by a service provider other than the operator or provider of access network 904 and / or telecommunications network 902, and may be operated by or on behalf of that service provider. Host 916 may host various applications to provide one or more services. Examples of such applications include: live and pre-recorded audio / video content, data collection services (e.g., retrieving and compiling data on various environmental conditions detected by multiple UEs), analytics functions, social media, functions for controlling or otherwise interacting with remote devices, functions for alarm and monitoring centers, or any other such functions performed by the server.

[0141] Overall, Figure 9The communication system 900 enables connectivity between the UE, network nodes, and hosts. In this sense, the communication system can be configured to operate according to predefined rules or procedures, such as specific standards, including but not limited to: Global System for Mobile Communications (GSM); Universal Mobile Telecommunications System (UMTS); Long Term Evolution (LTE), and / or other suitable 2G, 3G, 4G, 5G standards, or any applicable next-generation standard (e.g., 6G); Wireless Local Area Network (WLAN) standards, such as the Institute of Electrical and Electronics Engineers (IEEE) 802.11 standard (WiFi); and / or any other suitable wireless communication standards, such as Global Microwave Access Interoperability (WiMax), Bluetooth, Z-Wave, Near Field Communication (NFC), ZigBee, LiFi, and / or any Low-Power Wide-Area Network (LPWAN) standards, such as LoRa and Sigfox.

[0142] In some examples, Telecom Network 902 is a cellular network implementing 3GPP standardized features. Therefore, Telecom Network 902 can support network slicing to provide different logical networks to different devices connected to it. For example, Telecom Network 902 can provide Ultra Reliable Low Latency Communication (URLLC) services to some UEs while providing enhanced mobile broadband (eMBB) services to other UEs, and / or massive machine-type communication (mMTC) / massive IoT services to yet another UE.

[0143] In some examples, UE 912 is configured to send and / or receive information without direct human interaction. For example, the UE may be designed to send information to access network 904 according to a predetermined schedule, or triggered by internal or external events, or in response to a request from access network 904. Furthermore, the UE may be configured to operate in single RAT, multi-RAT, or multi-standard modes. For example, the UE may operate in any or a combination of Wi-Fi, NR (New Radio), and LTE, i.e., configured for Multi-Radio Dual Connectivity (MR-DC), such as E-UTRAN (Evolved UMTS Terrestrial Radio Access Network) New Radio-Dual Connectivity (EN-DC).

[0144] In this example, hub 914 communicates with access network 904 to facilitate indirect communication between one or more UEs (e.g., UEs 912c and / or 912d) and network nodes (e.g., network node 910b). In some examples, hub 914 may be a controller, router, content source, and analytics device, or any other communication device described herein with respect to the UE. For example, hub 914 may be a broadband router that enables the UE to access core network 906. As another example, hub 914 may be a controller that sends commands or instructions to one or more actuators in the UE. Commands or instructions may be received from the UE, network node 910, or by executable code, scripts, processes, or other instructions in hub 914. As another example, hub 914 may be a data collector that acts as temporary storage for UE data, and in some embodiments, analytics or other processing may be performed on the data. As another example, hub 914 may be a content source. For example, for a UE acting as a VR headset, display, speaker, or other media delivery device, hub 914 can retrieve VR assets, video, audio, or other media or data related to sensory information via network nodes, and then hub 914 provides them to the UE directly, or after performing local processing, and / or after adding additional local content. In yet another example, hub 914 acts as a proxy server or orchestrator for the UE, particularly if one or more UEs are low-power IoT devices.

[0145] Hub 914 may have a constant / persistent or intermittent connection to network node 910b. Hub 914 may also allow different communication schemes and / or scheduling between hub 914 and UEs (e.g., UEs 912c and / or 912d) and between hub 914 and core network 906. In other examples, hub 914 is connected to core network 906 and / or one or more UEs via a wired connection. Furthermore, hub 914 may be configured to connect to an M2M service provider via access network 904 and / or be connected to another UE via a direct connection. In some scenarios, a UE may establish a wireless connection to network node 910 while still being connected via hub 914 via a wired or wireless connection. In some embodiments, hub 914 may be a dedicated hub, i.e., a hub whose primary function is to route communication from network node 910b to UE / from UE to network node 910b. In other embodiments, hub 914 may be a non-dedicated hub—that is, a device that can route communication between the UE and network node 910b, but can also operate as a communication start and / or end point for certain data channels.

[0146] In some embodiments, core network node 908 or host 916 can be configured to perform the actions described above. Figures 6-8 The operation corresponding to the exemplary technology described above.

[0147] Figure 10 A network node 1000 according to some embodiments is shown. Examples of network nodes include, but are not limited to, access points (APs) (e.g., radio access points), base stations (e.g., radio base stations, Node B, eNB, gNB), and O-RAN nodes or components of O-RAN nodes (e.g., O-RU, O-DU, O-CU).

[0148] Base stations can be classified based on the coverage they provide (or, in other words, their transmit power level), and thus can be called femtocells, picocells, microcells, or macrocells depending on the coverage provided. A base station can be a relay node or a donor relay node for control relays. Network nodes can also include one or more (or all) portions of a distributed radio base station, such as centralized digital units, distributed units (e.g., in O-RAN access nodes), and / or remote radio units (RRUs), sometimes referred to as remote radio headends (RRHs). Such remote radio units may be integrated with an antenna as an antenna-integrated radio, or they may not be integrated. Portions of a distributed radio base station can also be referred to as nodes in a distributed antenna system (DAS).

[0149] Other examples of network nodes include multi-TRP 5G access nodes, multi-standard radio (MSR) devices (e.g., MSR BS), network controllers (e.g., Radio Network Controller (RNC) or BaseStation Controller (BSC)), Base Transceiver Stations (BTS), transport points, transport nodes, multi-cell / multicast coordination entities (MCE), operation and maintenance (O&M) nodes, operations support system (OSS) nodes, self-organizing network (SON) nodes, location nodes (e.g., Evolved Serving Mobile Location Center (E-SMLC)), and / or minimized drive tests (MDT).

[0150] Network node 1000 includes processing circuitry 1002, memory 1004, communication interface 1006, and power supply 1008. Network node 1000 may consist of multiple physically separate components (e.g., NodeB components and RNC components, or BTS components and BSC components, etc.), each of which may have its own corresponding components. In some scenarios where network node 1000 includes multiple separate components (e.g., BTS and BSC components), one or more of these separate components may be shared among several network nodes. For example, a single RNC can control multiple NodeBs. In such scenarios, each unique NodeB and RNC pair can, in some cases, be considered a single separate network node. In some embodiments, network node 1000 may be configured to support multiple Radio Access Technologies (RATs). In such embodiments, some components may be duplicated (e.g., separate memory 1004 for different RATs), while some components may be reused (e.g., the same antenna 1010 may be shared by different RATs). Network node 1000 may also include various sets of components for different wireless technologies (such as GSM, WCDMA, LTE, NR, WiFi, Zigbee, Z-wave, LoRaWAN, RFID, or Bluetooth wireless technologies) integrated into network node 1000. These wireless technologies may be integrated into the same or different chips or chipsets and other components within network node 1000.

[0151] The processing circuitry 1002 may include a combination of one or more of the following: a microprocessor, a controller, a central processing unit, a digital signal processor, an application-specific integrated circuit, a field-programmable gate array, or any other suitable computing device, resource, or combination of hardware, software, and / or coding logic, operable to provide network node 1000 functionality individually or in combination with other network node 1000 components (e.g., memory 1004).

[0152] In some embodiments, the processing circuitry 1002 includes a system-on-a-chip (SOC). In some embodiments, the processing circuitry 1002 includes one or more of a radio frequency (RF) transceiver circuitry 1012 and a baseband processing circuitry 1014. In some embodiments, the RF transceiver circuitry 1012 and the baseband processing circuitry 1014 may be on separate chips (or chipsets), boards, or units, such as a radio unit and a digital unit. In alternative embodiments, some or all of the RF transceiver circuitry 1012 and the baseband processing circuitry 1014 may be on the same chip or chipset, board, or unit.

[0153] Memory 1004 may include any form of volatile or non-volatile computer-readable memory, including but not limited to: persistent storage devices, solid-state memory, remotely mounted memory, magnetic media, optical media, random access memory (RAM), read-only memory (ROM), mass storage media (e.g., hard disk), removable storage media (e.g., flash drives, compact disks (CDs), or digital video disks (DVDs)), and / or any other volatile or non-volatile, non-transitory device that stores information, data, and / or instructions that can be used by processing circuitry 1002 and is readable and / or computer-executable. Memory 1004 may store any suitable instructions, data, or information, including computer programs, software, and applications including one or more of logic, rules, codes, tables, and / or other instructions (collectively, computer program 1004a, which may be in the form of a computer program product) that can be executed by processing circuitry 1002 and used by network node 1000. The memory 1004 can be used to store any calculations performed by the processing circuitry 1002 and / or any data received via the communication interface 1006. In some embodiments, the processing circuitry 1002 and the memory 1004 are integrated.

[0154] Communication interface 1006 is used for wired or wireless communication of signaling and / or data between network nodes, access networks, and / or UEs. As shown, communication interface 1006 includes a port / terminal 1016 for transmitting and receiving data, for example, data transmission and reception with a network via a wired connection. Communication interface 1006 also includes a radio front-end circuit 1018 that can be coupled to antenna 1010, or in some embodiments, radio front-end circuit 1018 is part of antenna 1010. Radio front-end circuit 1018 includes filter 1020 and amplifier 1022. Radio front-end circuit 1018 can be connected to antenna 1010 and processing circuitry 1002. Radio front-end circuit can be configured to modulate signals communicating between antenna 1010 and processing circuitry 1002. Radio front-end circuit 1018 can receive digital data to be transmitted wirelessly to other network nodes or UEs. Radio front-end circuit 1018 can use a combination of filter 1020 and / or amplifier 1022 to convert digital data into radio signals with appropriate channel and bandwidth parameters. The radio signal can then be transmitted via antenna 1010. Similarly, when receiving data, antenna 1010 can collect the radio signal, which is then converted into digital data by radio front-end circuitry 1018. This digital data can then be passed to processing circuitry 1002. In other embodiments, the communication interface may include different components and / or different combinations of components.

[0155] In some alternative embodiments, network node 1000 does not include a separate radio front-end circuitry 1018; instead, processing circuitry 1002 includes the radio front-end circuitry and is connected to antenna 1010. Similarly, in some embodiments, all or part of RF transceiver circuitry 1012 is part of communication interface 1006. In other embodiments, communication interface 1006 includes one or more ports or terminals 1016, radio front-end circuitry 1018, and RF transceiver circuitry 1012 as part of a radio unit (not shown), and communication interface 1006 communicates with baseband processing circuitry 1014 as part of a digital unit (not shown).

[0156] Antenna 1010 may include one or more antennas or antenna arrays configured to transmit and / or receive wireless signals. Antenna 1010 may be coupled to radio front-end circuitry 1018 and may be any type of antenna capable of wirelessly transmitting and receiving data and / or signals. In some embodiments, antenna 1010 is separate from network node 1000 and may be connected to network node 1000 via an interface or port.

[0157] Antenna 1010, communication interface 1006, and / or processing circuitry 1002 can be configured to perform any receive operation and / or certain acquisition operation described herein as being performed by a network node. Any information, data, and / or signals can be received from the UE, another network node, and / or any other network device. Similarly, antenna 1010, communication interface 1006, and / or processing circuitry 1002 can be configured to perform any transmit operation described herein as being performed by a network node. Any information, data, and / or signals can be transmitted to the UE, another network node, and / or any other network device.

[0158] Power supply 1008 provides power to the various components of network node 1000 in a form suitable for the respective components (e.g., at the voltage and current levels required by each respective component). Power supply 1008 may also include, or be coupled to, power management circuitry to provide power to the components of network node 1000 for performing the functions described herein. For example, network node 1000 may be connected to an external power source (e.g., mains, power outlet) via input circuitry or an interface (e.g., a cable), thereby supplying power to the power circuitry of power supply 1008. As yet another example, power supply 1008 may include a power source in the form of a battery or battery pack, which is connected to or integrated into the power circuitry. The battery can provide backup power if the external power source fails.

[0159] Embodiments of network node 1000 may include Figure 10 Additional components beyond those shown may be used to provide certain aspects of the network node's functionality, including any of the functions described herein and / or any functions necessary to support the topics presented herein. For example, network node 1000 may include a user interface device to allow information to be input into and output from network node 1000. This can allow users to perform diagnostic, maintenance, repair, and other management functions on network node 1000.

[0160] In some embodiments, network node 1000 can be configured to perform the actions described above. Figures 6-8 The operation corresponding to the exemplary technology described above.

[0161] Figure 11 Based on the block diagram of host 1100 described in this document, the host can be... Figure 9 The embodiment of host 916. As used herein, host 1100 can be or include various combinations of hardware and / or software, including processing resources in a standalone server, blade server, cloud-implemented server, distributed server, virtual machine, container, or server cluster. Host 1100 can provide one or more services to one or more UEs.

[0162] Host 1100 includes processing circuitry 1102 operably coupled via bus 1104 to input / output interface 1106, network interface 1108, power supply 1110, and memory 1112. Other embodiments may include additional components. The characteristics of these components may be consistent with those described in the previous figures (e.g., Figure 2 and Figure 10 The features described for the devices are substantially similar, making the description generally applicable to the corresponding components of host 1100.

[0163] Memory 1112 may include one or more computer programs, including one or more host applications 1114 and data 1116. Data 1116 may include user data, such as data generated by the UE for the host 1100 or data generated by the host 1100 for the UE. Embodiments of the host 1100 may utilize all or only a subset of the illustrated components. Host application 1114 may be implemented in a container-based architecture and may provide support for video codecs (e.g., Security Edge Protection Proxy (VVC), High Efficiency Video Coding (HEVC), Advanced Video Coding (AVC), MPEG, VP9) and audio codecs (e.g., FLAC, Advanced Audio Coding (AAC), MPEG, G.711), including transcoding for multiple different categories, types, or implementations of UEs (e.g., mobile phones, desktop computers, wearable display systems, head-up display systems). Host application 1114 can also provide user authentication and authorization checks, and can periodically report health status, routing, and content availability to a central node (such as a device in the core network or at the edge). Therefore, host 1100 can select different hosts and / or instruct different hosts for the UE's over-the-top services. Host application 1114 can support various protocols, such as HTTP Live Streaming (HLS), Real-Time Messaging Protocol (RTMP), Real-Time Streaming Protocol (RTSP), and HTTP-based Dynamic Adaptive Streaming (MPEG-DASH), etc.

[0164] In some embodiments, host 1100 may be configured to perform the actions described above. Figures 6-8 The operation corresponding to the exemplary technology described above.

[0165] Figure 12 This is a block diagram illustrating a virtualization environment 1200 in which some embodiments of functionality can be virtualized. In the current context, virtualization means creating virtual versions of devices or equipment, which can include virtualized hardware platforms, storage devices, and network resources. As used herein, virtualization can be applied to any device or component thereof described herein and involves at least some of its functionality being implemented as an implementation of one or more virtual components. Some or all of the functionality described herein can be implemented as virtual components, which can be executed by one or more virtual machines (VMs) implemented in one or more virtual environments 1200 hosted by one or more hardware nodes (e.g., hardware computing devices operating as network nodes, UEs, core network nodes, or hosts). Furthermore, in embodiments where the virtual node does not require radio connectivity (e.g., core network nodes or hosts), the node can be fully virtualized. In some embodiments, the virtualization environment 1200 includes components defined by the O-RAN Alliance, such as an O-Cloud environment orchestrated via an O-2 interface by a service management and orchestration framework.

[0166] Application 1202 (which may alternatively be referred to as a software instance, virtual device, network function, virtual node, virtual network function, etc.) runs in virtualization environment 1200 to implement certain features, functions, and / or benefits of some embodiments disclosed herein. As a specific example, related to the above regarding... Figures 6-8 The operations corresponding to the exemplary technology described above can be instantiated as application 1402 running in virtualization environment 1400.

[0167] Hardware 1204 includes processing circuitry, memory storing software and / or instructions executable by the hardware processing circuitry (collectively referred to as computer program 1204a, which may be in the form of a computer program product), and / or other hardware devices described herein, such as network interfaces, input / output interfaces, etc. The software may be executed by the processing circuitry to instantiate one or more virtualization layers 1206 (also referred to as a hypervisor or virtual machine monitor (VMM)), provide VMs 1208a and 1208b (one or more of which may be collectively referred to as VM 1208), and / or perform any functions, features, and / or benefits described in relation to some embodiments described herein. Virtualization layer 1206 may present a virtual operating platform to VM 1208 that appears to be network hardware.

[0168] VM 1208 includes virtual processing, virtual memory, virtual network or interface, and virtual storage, and can be run by a corresponding virtualization layer 1206. Different embodiments of instances of virtual device 1202 can be implemented on one or more VMs 1208, and the implementation can be done in different ways. Hardware virtualization is referred to as Network Function Virtualization (NFV) in some contexts. NFV can be used to consolidate many types of network devices onto industry-standard high-capacity server hardware, physical switches, and physical storage devices, which can reside in data centers and customer premises.

[0169] In the context of NFV, each VM 1208 can be a software implementation of a physical machine that runs programs as if they were executing on a physical, non-virtualized machine. Each VM 1208, along with the portion of hardware 1204 that executes that VM (whether dedicated to that VM or shared by that VM with other VMs), forms a separate virtual network element. Still within the NFV context, the virtual network function is responsible for handling specific network functions running on one or more VMs 1208 on hardware 1204, and corresponds to application 1202.

[0170] Hardware 1204 can be implemented in a standalone network node with general or specific components. Hardware 1204 may implement some functions through virtualization. Alternatively, hardware 1204 may be part of a larger hardware cluster (e.g., in a data center or Customer Premise Equipment (CPE)) where many hardware nodes work together and are managed via management and orchestration function 1210, which, among other things, oversees the lifecycle management of application 1202. In some embodiments, hardware 1204 is coupled to one or more radio units, each radio unit including one or more transmitters and one or more receivers that can be coupled to one or more antennas. The radio units may communicate directly with other hardware nodes via one or more suitable network interfaces and may be used in conjunction with virtual components to provide a radio-capable virtual node, such as a radio access node or base station. In some embodiments, some signaling may be provided using control system 1212, which may also be used for communication between hardware nodes and radio units.

[0171] The foregoing only illustrates the principles of this disclosure. Various modifications and variations to the described embodiments will be apparent to those skilled in the art based on the teachings herein. Therefore, it will be understood that those skilled in the art will be able to design numerous systems, arrangements, and processes that, while not expressly shown or described herein, embody the principles of this disclosure and are therefore within the spirit and scope of this disclosure. Various exemplary embodiments may be used in combination with or interchangeably with each other, as will be understood by those skilled in the art.

[0172] As used herein, the term “unit” may have the conventional meaning in the field of electronic, electrical and / or electronic equipment, and may include, for example, electrical and / or electronic circuits, devices, modules, processors, memories, logic solid-state and / or discrete devices, computer programs or instructions for performing corresponding tasks, processes, calculations, outputs and / or display functions, such as those described herein.

[0173] Any suitable steps, methods, features, functions, or benefits disclosed herein may be performed by one or more functional units or modules of one or more virtual devices. Each virtual device may include multiple such functional units. These functional units may be implemented via processing circuitry, which may include one or more microprocessors or microcontrollers, as well as other digital hardware (which may include digital signal processors (DSPs), application-specific digital logic, etc.). The processing circuitry may be configured to execute program code stored in memory, which may include one or more types of memory, such as read-only memory (ROM), random access memory (RAM), cache memory, flash memory devices, optical storage devices, etc. The program code stored in memory includes program instructions for executing one or more telecommunications and / or data communication protocols, and instructions for executing one or more techniques described herein. In some implementations, the processing circuitry may be used to cause corresponding functional units to perform corresponding functions according to one or more embodiments of this disclosure.

[0174] As described herein, devices and / or apparatuses may be represented by semiconductor chips, chipsets, or (hardware) modules including such chips or chipsets; however, this does not preclude the possibility that the functionality of a device or apparatus may be implemented as a software module (e.g., a computer program or a computer program product including executable software code portions for execution or execution on a processor), rather than a hardware implementation. Furthermore, the functionality of a device or apparatus may be implemented by any combination of hardware and software. A device or apparatus may also be considered as a component of multiple devices and / or apparatuses, whether functionally cooperative or independent of each other. Moreover, devices and apparatuses may be implemented in a distributed manner throughout a system, provided that the functionality of the device or apparatus is preserved. Such and similar principles are considered to be known to those skilled in the art.

[0175] Unless otherwise defined, all terms used herein (including technical and scientific terms) shall have the same meaning as commonly understood by one of ordinary skill in the art to which this disclosure pertains. It should also be understood that the terms used herein are to be interpreted as having the meaning consistent with their meaning in the context of this specification and the relevant field, and will not be interpreted in an idealized or overly formal sense unless expressly defined herein.

[0176] Furthermore, certain terms used in this disclosure (including the specification and drawings) may be used as synonyms in certain circumstances (e.g., "data" and "information"). It should be understood that although these terms (and / or other terms that may be synonyms with each other) may be used as synonyms herein, there may be instances where such words may not be intended to be used as synonyms.

Claims

1. A method for determining a Vertical Application Layer (VAL) group operating in a communication network, each VAL group comprising multiple User Equipment (UEs), the method comprising: The first multi-layer diagram representation (810) is determined based on the following: The reference VAL group configuration previously used in the communication network, and Historical Quality of Service (QoS) metrics for multi-layer connections between UEs including the reference VAL group configuration; The following first operation is performed using a Generative Adversarial Network (GAN) that includes a generator and a discriminator: Using the generator of the GAN, the second multilayer graph representation (820) is determined based on the following: Based on the candidate VAL group configuration determined by random input, and Predicted QoS metrics for multi-layer connections between the UEs, including the reference VAL group configuration; Using the discriminator of the GAN, determine whether (830) satisfies the following first condition: The second multi-layer graph representation cannot be distinguished from the first multi-layer graph representation, and The predicted QoS metric associated with the candidate VAL group configuration satisfies one or more Service Level Agreements (SLAs) for the operation of the communication network; When the first condition is not met, repeat the first operation (850) based on different random inputs; and After the first condition is met and the generator is used, (860) reasoning is performed based on the second multi-layer graph representation to generate a configuration for the VAL group to operate in the communication network.

2. The method according to claim 1, wherein, The configuration of the VAL group generated by performing inference includes the following: multiple UEs arranged in multiple VAL groups, and connections between the various UE pairs.

3. The method according to claim 2, wherein, One of the following applies: The configuration of the VAL group generated by performing inference also includes QoS settings for the connections between the various UE pairs; or The method also includes obtaining (870) the predicted QoS settings for the connection between the various UE pairs from the network data analysis function NWDAF of the communication network.

4. The method according to any one of claims 2 to 3, wherein, The connections between the various UE pairs include the following: connections between UE pairs in the same VAL group, and connections between UE pairs in different VAL groups.

5. The method according to any one of claims 1 to 4, wherein, The reference VAL group configuration and each candidate VAL group configuration include the following: multiple UEs arranged in multiple VAL groups, and connections between the various UE pairs at multiple protocol layers.

6. The method according to claim 5, wherein: The multiple VAL groups are associated with corresponding applications or services; and Each of the historical QoS metric and the predicted QoS metric is associated with data traffic in one of the applications or services.

7. The method according to any one of claims 5 to 6, wherein, Each of the first multi-layer graph representation and the second multi-layer graph representation includes: Multiple first layers, wherein each first layer represents a connection between the respective UE pairs at a corresponding single protocol layer; and The second layer represents the connection between the various UE pairs across multiple protocol layers.

8. The method according to any one of claims 5 to 7, wherein, Determining (810) the first multi-layer graph representation includes: A graph structure (811) is determined based on the reference VAL group configuration, wherein the graph structure includes connections between the plurality of UEs as nodes and the respective pairs of UEs as edges; and The attributes of the edge (812) are determined based on the historical QoS metric.

9. The method according to claim 8, wherein, The attributes of the edge (823) determined based on the historical QoS metric include, for each connection between UE pairs: Obtain one or more historical QoS metrics for the connection from the Network Data Analysis Function (NWDAF) of the communication network; and Based on one or more of the historical QoS metrics, the attribute corresponding to the edge of the connection is determined to be an integer priority value.

10. The method according to claim 9, wherein, Each integer priority value is determined based on the following: 5G QoS identifier 5QI, when included in one or more of the historical QoS metrics; or When 5QI is excluded, at least one of the following historical QoS metrics: The percentage of uplink UL throughput of the application or service described in the region of interest out of the total UL throughput; The percentage of downlink DL throughput of the application or service described in the region of interest relative to the total UL throughput; The average traffic rate of the UE communicating with the application or service observed; The observed maximum traffic rate of the UE communicating with the application or service; The minimum traffic rate observed for a UE communicating with the application or service; The average packet latency observed for the UE communicating with the application or service; The observed maximum packet latency for a UE communicating with the application or service; The observed average packet loss rate of UEs communicating with the application or service; and The observed maximum packet loss rate of a UE communicating with the application or service.

11. The method according to any one of claims 5 to 10, wherein, The generator and discriminator include corresponding multi-layer graph neural networks (mGNNs).

12. The method according to claim 11, wherein, It also includes: when it is determined that the second multi-layer graph representation can be distinguished from the first multi-layer graph representation, adjusting (840) the mGNN of the generator and the discriminator based on a metric representing the difference between the first multi-layer graph representation and the second multi-layer graph representation, wherein the first operation is repeated using the adjusted mGNN of the generator and the discriminator.

13. The method according to claim 12, wherein, Once the first condition is met, inference is performed using an adjusted generator mGNN.

14. The method according to any one of claims 11 to 13, wherein, Determining (820) the second multi-layer diagram representation includes: The candidate VAL group configuration is determined by applying the generator's mGNN to the random input; Based on the candidate VAL group configuration, a graph structure (822) is determined, wherein the graph structure includes the connections between the plurality of UEs as nodes and the respective UE pairs as edges; and The attributes of the edge (823) are determined based on the historical QoS metric.

15. The method according to any one of claims 11 to 13, wherein, Determining (820) the second multi-layer graph representation includes: based on applying the generator's mGNN to the random input, determining (824) an attributed graph structure of the candidate VAL group configuration, wherein the attributed graph structure includes the plurality of UEs as nodes, connections between the plurality of UEs as edges, and corresponding QoS metrics as attributes of the edges.

16. The method according to any one of claims 1 to 15, wherein, The method is performed by one of the following: the Network Data Analysis Function (NWDAF) of the communication network; or, the Service Enablement Architecture (SEAL) group management GM server outside the communication network.

17. The method according to claim 16, wherein, The method is performed by the NWDAF, and The reasoning is further based on the expected architecture of the VAL group obtained from the SEAL GM server.

18. A network node or functional NNF (210, 320, 400, 430, 908, 916, 1000, 1100, 1202) configured to determine a vertical application layer (VAL) group for operation in a communication network (100, 200, 498, 902), each VAL group including multiple user equipment (UE) (220, 310, 912), wherein said NNF includes processing circuitry (1002, 1102, 1204) configured to: The first multi-layer graph representation is determined based on the following: The reference VAL group configuration previously used in the communication network, and Historical Quality of Service (QoS) metrics for multi-layer connections between UEs including the reference VAL group configuration; The following first operation is performed using a Generative Adversarial Network (GAN) that includes a generator and a discriminator: Using the generator, a second multi-layer graph representation is determined based on the following: Based on the candidate VAL group configuration determined by random input, and Predicted QoS metrics for multi-layer connections between the UEs, including the reference VAL group configuration; Using the discriminator, determine whether the following first condition is met: The second multi-layer graph representation cannot be distinguished from the first multi-layer graph representation, and The predicted QoS metric associated with the candidate VAL group configuration satisfies one or more Service Level Agreements (SLAs) for the operation of the communication network; When the first condition is not met, the first operation is repeated based on different random inputs; and After the first condition is met and the generator is used, reasoning is performed based on the second multi-layer graph representation to generate a configuration for the VAL group to operate in the communication network.

19. The NNF according to claim 18, wherein, The NNF is the Network Data Analysis Function NWDAF (400) of the communication network, and also includes communication interface circuits (1006, 1108, 1204) configured to communicate with the Service Enablement Architecture Layer SEAL Group Management GM Server (430) outside the communication network.

20. The NNF according to any one of claims 18 to 19, wherein, The processing circuit is also configured to perform operations corresponding to the method of any one of claims 2 to 17.

21. A network node or functional NNF (210, 320, 400, 430, 908, 916, 1000, 1100, 1202) configured to determine vertical application layer (VAL) groups for operation in communication networks (100, 200, 498, 902), each VAL group comprising multiple user equipment (UE) groups (220, 310, 912), wherein said NNF is further configured to: The first multi-layer graph representation is determined based on the following: The reference VAL group configuration previously used in the communication network, and Historical Quality of Service (QoS) metrics for multi-layer connections between UEs including the reference VAL group configuration; The following first operation is performed using a Generative Adversarial Network (GAN) that includes a generator and a discriminator: Using the generator, a second multi-layer graph representation is determined based on the following: Based on the candidate VAL group configuration determined by random input, and Predicted QoS metrics for multi-layer connections between the UEs, including the reference VAL group configuration; Using the discriminator, determine whether the following first condition is met: The second multi-layer graph representation cannot be distinguished from the first multi-layer graph representation, and The predicted QoS metric associated with the candidate VAL group configuration satisfies one or more Service Level Agreements (SLAs) for the operation of the communication network. When the first condition is not met, the first operation is repeated based on different random inputs; and After the first condition is met and the generator is used, reasoning is performed based on the second multi-layer graph representation to generate a configuration for the VAL group to operate in the communication network.

22. The NNF according to claim 21, wherein, It is also configured to perform operations corresponding to the method of any one of claims 2 to 17.

23. A non-transitory computer-readable medium (1004, 1204) storing computer-executable instructions, which, when executed by a processing circuit (1002, 1102, 1204) associated with a network node or functional NNF (210, 320, 400, 430, 908, 916, 1000, 1100, 1202) of a vertical application layer (VAL) group configured to operate in a communication network (100, 200, 498, 902), configures the NNF to perform an operation corresponding to the method of any one of claims 1 to 17, each VAL group comprising a plurality of user equipment (UE) (220, 310, 912).

24. A computer program product (1004a, 1204a) comprising computer-executable instructions, which, when executed by a processing circuit (1002, 1102, 1204) associated with a network node or functional NNF (210, 320, 400, 430, 908, 916, 1000, 1100, 1202) configured to perform an operation corresponding to the method of any one of claims 1 to 17, each VAL group comprising a plurality of user equipment (UE) (220, 310, 912).

Images