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

Abstract

Embodiments include methods for determining vertical application layer (VAL) groups for operation in a communication network, each VAL group including multiple user equipment (UEs). Such methods include determining a first multi-layer graph representation based on a reference VAL group configuration and historical quality of service (QoS) metrics. Such methods include performing the following first operations using a generative adversarial network (GAN): determining a second multi-layer graph representation, and determining whether the following first conditions are met: the second multi-layer graph representation cannot be differentiated from the first multi-layer graph representation, and predicted QoS metrics fulfil one or more service level agreements (SLAs). Such methods include repeating the first operations based on another random input when the first conditions are not met and after the first conditions are met, performing inference based on the second multi-layer graph representation to generate a configuration of VAL groups for operation in the communication network.

Description

[0001] DETERMINING VERTICAL APPLICATION LAYER (VAL) GROUPS USING GENERATIVE ADVERSARIAL NETWORKS (GAN)

[0002] TECHNICAL FIELD

[0003] The present disclosure relates generally to the field of communication networks, and more specifically to techniques for determining vertical application layer (VAL) groups of user equipment (UEs) operating in a communication network (e.g., 5G wireless network) based on applying machine learning (ML) techniques such as generative adversarial networks (GANs).

[0004] BACKGROUND

[0005] The fifth generation (5G) of cellular systems, also referred to as New Radio (NR), was initially standardized 3GPP Rel-15 and continues to evolve in subsequent releases. NR is developed for maximum flexibility to support a variety of different use cases including enhanced mobile broadband (eMBB), machine type communications (MTC), ultra-reliable low latency communications (URLLC), side-link device-to-device (D2D), and several other use cases. 5G / NR technology shares many similarities with fourth-generation LTE.

[0006] One change in 5G networks is that traditional peer-to-peer interfaces and protocols found in earlier-generation networks (e.g., LTE) are modified and / or replaced by a Service Based Architecture (SBA) in which Network Functions (NFs) provide one or more services to one or more service consumers. This can be done, for example, by Hyper Text Transfer Protocol / Representational State Transfer (HTTP / REST) application programming interfaces (APIs). In general, the various services are self-contained functionalities that can be changed and modified in an isolated manner without affecting other services. Furthermore, the services are composed of various “service operations”, which are more granular divisions of the overall service functionality.

[0007] In the 5G SBA, network repository functions (NRF) allow every network function to discover the services offered by other network functions, and Data Storage Functions (DSF) allow every network function to store its context. The Unified Data Management (UDM) function supports generation of 3GPP authentication credentials, user identification handling, access authorization based on subscription data, and other subscriber-related functions. A Network Exposure Function (NEF) acts as entry point into an operator's 5G core network (5GC), by securely exposing network capabilities and events provided by other NFs and by providing ways for application functions (AFs) to securely provide information to (or receive information from) the 5GC.

[0008] 3GPP has specified a service enabler architecture layer (SEAL) to support vertical applications over 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. In particular, a UE may include a SEAL group management client that communicates with a corresponding SEAL group management server in (or external to) the 5GC. The group management client provides group management services to VAL client(s) in the UE, which communicate with corresponding VAL server(s) in (or external to) the 5GC. The VAL server(s) also communicate with the SEAL group management server.

[0009] SEAL can use various information for VAL group management, including VAL group IDs, group member UE IDs, generic group configurations, 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 3GPP TS 23.434 (vl7.7.0) clause 10 and 3GPP TS 29.549 (V17.8.0) clause 7.2.I.4.

[0010] SUMMARY

[0011] However, the information specified by 3GPP for VAL group management does not include UE communication characteristics, which can be very important 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 for communication characteristics. These parameters may include connection quality-of-service (QoS) data, guaranteed and maximum bitrates, packet delay budgets, and packet error rates. Moreover, each of these parameters may relate to communications within a group and between users in different groups.

[0012] Furthermore, conventional VAL group management / configuration is done manually by a programmer who relies to some degree on the site-specific parameters along with historical data and own experience. This non-trivial task is made even more difficult by inevitable changes of site specifications and / or physical arrangement. As such, manual VAL group configurations may not fulfil intra- and inter-group communication QoS and / or end user quality of experience (QoE) required under a service level agreement (SLA). Moreover, manual VAL group configurations are unlikely to produce an optimal resource allocation within the 5G system, instead resulting in traffic congestion, UE and / or network resource unavailability, and QoS degradations.

[0013] An object of embodiments of the present disclosure is to address one or more of these and related problems, issues, and / or difficulties by providing techniques and apparatus for automated configuration of VAL groups based on historical and / or predicted communication parameters in a communication network (e.g., 5G network).

[0014] Some embodiments of the present disclosure include methods (e.g., procedures) for determining VAL groups for operation in a communication network, with each VAL group including a plurality of UEs.

[0015] These exemplary methods include determining a first multi-layer graph representation based on the following: a reference VAL group configuration previously used in the communication network, and historical QoS metrics for multi-layer connections between UEs that comprise the reference VAL group configuration. These exemplary methods also include performing the following first operations using a generative adversarial network (GAN) comprising a generator and a discriminator:

[0016] • using the generator, determining a second multi-layer graph representation based on the following: a candidate VAL group configuration determined based on a random input, and predicted QoS metrics for multi-layer connections between the UEs that comprise the reference VAL group configuration;

[0017] • using the discriminator, determining whether the following first conditions are met: the second multi-layer graph representation cannot be differentiated from the first multilayer graph representation, and the predicted QoS metrics associated with the candidate VAL group configuration fulfil one or more service level agreements (SLAs) for operation of the communication network;

[0018] These exemplary methods also include repeating the first operations (above) based on a different random input when the first conditions are not met. These exemplary methods also include, after the first conditions are met and using the generator, performing inference based on the second multi-layer graph representation to generate a configuration of VAL groups for operation in the communication network.

[0019] In some embodiments, the exemplary method is performed by a 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 within) the communication network. In some of these embodiments, the method is performed by the NWDAF and the inference is further based on a desired architecture for VAL groups that is obtained from the SEAL GM server.

[0020] In some embodiments, the configuration of VAL groups generated by performing inference includes the following: a plurality of UEs arranged in multiple VAL groups, and connections between respective pairs of the UEs. In some of these embodiments, the connections between respective pairs of the UEs including the following: connections between pairs of UEs in a same VAL group, and connections between pairs of UEs in different VAL groups.

[0021] In some of these embodiments, the configuration of VAL groups generated by performing inference also includes QoS settings for the connections between respective pairs of the UEs. In other of these embodiments, these exemplary methods also include obtaining, from an NWDAF of the communication network, predicted QoS settings for the connections between respective pairs of the UEs.

[0022] In some embodiments, each of the first and second multi-layer graph representations comprises:

[0023] • a plurality of first layers, with each first layer representing connections on a corresponding single protocol layer between respective pairs of the UEs; and

[0024] • a second layer representing connections across multiple protocol layers between respective pairs of the UEs.

[0025] In some embodiments, the generator and the discriminator include respective multi-layer graph neural networks (mGNNs).

[0026] Other embodiments include various network nodes, or functions (NNFs) configured to perform operations corresponding to the exemplary methods summarized above. Other embodiments include non-transitory, computer-readable media storing computer-executable instructions that, when executed by processing circuitry, configure such NNFs to perform operations corresponding to the exemplary methods summarized above.

[0027] These and other embodiments described herein provide various benefits and / or advantages. For example, by using a combination of GNNs and GANs, embodiments can efficiently optimize connectivity and QoS parameters of each VAL group as well as inter-group connections within the various protocol layers simultaneously, thereby providing an optimal configuration of the overall network. In this manner, embodiments allocate each UE to an optimal VAL group and determine optimal connections for the UEs within a VAL group and between VAL groups, while satisfying any relevant SLAs. In this manner, embodiments can automate the very site-specific process of VAL group configurations in a 5G network while providing necessary QoS and QoE to satisfy any relevant SLAs. At a high level, embodiments improve network operation in support of vertical applications such as mining, chemical or food processing, and automotive manufacturing.

[0028] These and other objects, features, and advantages of the present disclosure will become apparent upon reading the following Detailed Description in view of the Drawings briefly described below. BRIEF DESCRIPTION OF THE DRAWINGS

[0029] Figures 1-2 show various aspects of 5G network architecture.

[0030] Figure 3 shows a block diagram of a functional model for group management based on SEAL.

[0031] Figure 4 shows an example implementation of embodiments of the present disclosure in a 5G network.

[0032] Figure 5 shows an example arrangement of VAL groups in an IIoT environment, according to various embodiments of the present disclosure.

[0033] Figures 6-7 show example functional architectures for VAL group determination, according to various embodiments of the present disclosure.

[0034] Figure 8 illustrates an exemplary method (e.g., procedure) for ... in a communication network, according to various embodiments of the present disclosure.

[0035] Figure 9 shows an exemplary communication network in which various embodiments of the present disclosure can be implemented.

[0036] Figures 10 shows an exemplary network node in which various embodiments of the present disclosure can be implemented

[0037] Figure 11 shows an exemplary host computing system in which various embodiments of the present disclosure can be implemented.

[0038] Figure 12 shows an exemplary virtualization environment in which various embodiments of the present disclosure can be implemented.

[0039] DETAILED DESCRIPTION

[0040] Some of the embodiments contemplated herein will now be described more fully with reference to the accompanying drawings. Other embodiments, however, are contained within the scope of the subject matter disclosed herein, the disclosed subject matter should not be construed as limited to only 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.

[0041] In general, all terms used herein are to be interpreted according to their ordinary meaning to a person of ordinary skill in the relevant technical field, unless a different meaning is expressly defined and / or implied from the context of use. All references to a / an / the element, apparatus, component, means, step, etc. are to be interpreted openly as referring to at least one instance of the element, apparatus, component, means, step, etc., unless explicitly stated otherwise or clearly implied from the context of use. The operations of any methods and / or procedures disclosed herein do not have to be performed in the exact order disclosed, unless an operation is explicitly described as following or preceding another operation and / or where it is implicit that an operation must follow or precede another operation. Any feature of any embodiment disclosed herein can apply to any other disclosed embodiment, as appropriate. Likewise, any advantage of any embodiment described herein can apply to any other disclosed embodiment, as appropriate.

[0042] Furthermore, the following terms are used throughout the description given below:

[0043] • Radio Access Node: As used herein, a “radio access node” (or equivalently “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 a radio access node include, but are not limited to, a base station (e.g., gNB in a 3GPP 5G / NR network or an enhanced or eNB in a 3GPP LTE network), base station distributed components (e.g., CU and DU), a high-power or macro base station, a low-power base station (e.g., micro, pico, femto, or home base station, or the like), an integrated access backhaul (IAB) node, a transmission point (TP), a transmission reception point (TRP), a remote radio unit (RRU or RRH), and a relay node.

[0044] • Core Network Node: As used herein, a “core network node” is any type of node in a core network. Some examples of a core network node include, e.g., 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), or the like.

[0045] • Wireless Device: As used herein, a “wireless device” (or “WD” for short) is any type of device that is capable, configured, arranged and / or operable to communicate wirelessly with network nodes and / or other wireless devices. Communicating wirelessly can involve transmitting and / or receiving wireless signals using electromagnetic waves, radio waves, infrared waves, and / or other types of signals suitable for conveying information through air. Unless otherwise noted, the term “wireless device” is used interchangeably herein with the term “user equipment” (or “UE” for short), with both of these terms having a different meaning than the term “network node”.

[0046] • Radio Node: As used herein, a “radio node” can be either a “radio access node” (or equivalent term) or a “wireless device.”

[0047] • Network Node: As used herein, a “network node” is any node that is either part of the radio access network (e.g., a radio access node or equivalent term) or of the core network e.g., a core network node discussed above) of a cellular communications network. Functionally, a network node is equipment capable, configured, arranged, and / or operable to communicate directly or indirectly with a wireless device and / or with other network nodes or equipment in the cellular communications network, to enable and / or provide wireless access to the wireless device, and / or to perform other functions (e.g., administration) in the cellular communications network.

[0048] • Node: As used herein, the term “node” (without prefix) can be any type of node that can in or with a wireless network (including RAN and / or core network), including a radio access node (or equivalent term), core network node, or wireless device. 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.

[0049] The above definitions are not meant to be exclusive. In other words, various ones of the above terms may be explained and / or described elsewhere in the present disclosure using the same or similar terminology. Nevertheless, to the extent that such other explanations and / or descriptions conflict with the above definitions, the above definitions should control.

[0050] Note that the description given herein focuses on a 3GPP cellular communications system and, as such, 3GPP terminology or terminology similar to 3GPP terminology is oftentimes used. However, the concepts disclosed herein are not limited to a 3GPP system and can be applied to any communication system that may benefit from them.

[0051] At a high level, the 5G System (5GS) consists of an Access Network (AN) and a Core Network (CN). The AN provides UEs connectivity to the CN, e.g., via base stations such as gNBs or ng-eNBs. As described in more detail below, the CN includes a variety of Network Functions (NF) that provide a wide range of different functionalities such as session management, connection management, charging, authentication, etc.

[0052] Figure 1 illustrates a high-level view of an exemplary 5G network architecture, consisting of a Next Generation Radio Access Network (NG-RAN, 199) and a 5G Core (5GC, 198). The NG-RAN can include one or more gNodeB’s (gNBs) connected to the 5GC via one or more NG interfaces, such as gNBs (100, 150) connected via respective interfaces (102, 152). More specifically, the gNBs can be connected to one or more Access and Mobility Management Functions (AMFs) in the 5GC via respective NG-C interfaces and to one or more User Plane Functions (UPFs) in 5GC via respective NG-U interfaces. The 5GC can include various other network functions (NFs), such as Session Management Function(s) (SMF).

[0053] In addition, the gNBs can be connected to each other via one or more Xn interfaces, such as Xn interface (140) between gNBs (100, 150). The radio technology for the NG-RAN is often referred to as “New Radio” (NR). With respect to the NR interface to UEs, each of the gNBs can support frequency division duplexing (FDD), time division duplexing (TDD), or a combination thereof. Each of the gNBs can serve a geographic coverage area including one or more cells and, in some cases, can also use various directional beams to provide coverage in the respective cells. In general, a DL “beam” is a coverage area of a network-transmitted reference signal (RS) that may be measured or monitored by a UE.

[0054] The NG-RAN is layered into a Radio Network Layer (RNL) and a Transport Network Layer (TNL). The NG-RAN architecture, i.e., the NG-RAN logical nodes and interfaces between them, is defined as part of the RNL. For each NG-RAN interface (NG, Xn, Fl) the related TNL protocol and the functionality are specified. The TNL provides services for user plane transport and signaling transport.

[0055] NG RAN logical nodes (e.g., gNB 100) include a Central Unit (CU or gNB-CU, e.g., 110) and one or more Distributed Units (DU or gNB-DU, e.g., 120, 130). CUs are logical nodes that host higher-layer protocols and perform various gNB functions such controlling the operation of DUs. DUs are decentralized logical nodes that host lower layer protocols and can include, depending on the functional split option, various subsets of the gNB functions. Each CU and DU can include various circuitry needed to perform their respective functions, including processing circuitry, communication interface circuitry e.g., transceivers), and power supply circuitry.

[0056] A gNB-CU connects to one or more gNB-DUs over respective Fl logical interfaces (e.g., 122 and 132 shown in Figure 1). However, a gNB-DU can be connected to only a single gNB-CU. The gNB-CU and its connected gNB-DU(s) are only visible to other gNBs and the 5GC as a gNB. In other words, the Fl interface is not visible beyond gNB-CU.

[0057] Figure 2 shows an exemplary non-roaming reference architecture for a 5G network (200) including the following 3GPP-defined NFs and service-based interfaces:

[0058] • Application Function (AF, with Naf interface) interacts with the 5GC to provision information to the network operator and to subscribe to certain events happening in operator's network. An AF offers applications for which service is delivered in a different layer (i.e., transport layer) than the one in which the service has been requested (i.e., signaling layer), the control of flow resources according to what has been negotiated with the network. An AF communicates dynamic session information to PCF (via N5 interface), including description of media to be delivered by transport layer.

[0059] • Policy Control Function (PCF, with Npcf interface) supports unified policy framework to govern the network behavior, via providing PCC rules (e.g., on the treatment of each service data flow that is under PCC control) to the SMF via the N7 reference point. PCF provides policy control decisions and flow based charging control, including service data flow detection, gating, QoS, and flow-based charging (except credit management) towards the SMF. The PCF receives session and media related information from the AF and informs the AF of traffic (or user) plane events. User Plane Function (UPF)- supports handling of user plane traffic based on the rules received from SMF, including packet inspection and different enforcement actions (e.g., event detection and reporting). UPFs communicate with the RAN (e.g., NG-RNA) via the N3 reference point, with SMFs (discussed below) via the N4 reference point, and with an external packet data network (PDN) via the N6 reference point. The N9 reference point is for communication between two UPFs.

[0060] • Session Management Function (SMF, with Nsmf interface) interacts with the decoupled traffic (or user) plane, including creating, updating, and removing Protocol Data Unit (PDU) sessions and managing session context with the User Plane Function (UPF), e.g., for event reporting. For example, SMF performs data flow detection (based on filter definitions included in PCC rules), online and offline charging interactions, and policy enforcement.

[0061] • Charging Function (CHF, with Nchf interface) is responsible for converged online charging and offline charging functionalities. It provides quota management (for online charging), re-authorization triggers, rating conditions, etc. and is notified about usage reports from the SMF. Quota management involves granting a specific number of units (e.g., bytes, seconds) for a service. CHF also interacts with billing systems.

[0062] Access and Mobility Management Function (AMF, with Namf interface) terminates the RAN CP interface and handles all mobility and connection management of UEs (similar to MME in EPC). AMFs communicate with UEs via the N1 reference point and with the RAN (e.g., NG- RAN) via the N2 reference point.

[0063] • Network Exposure Function (NEF) with Nnef interface - acts as the entry point into operator's network, by securely exposing to AFs the network capabilities and events provided by 3GPP NFs and by providing ways for the AF to securely provide information to 3GPP network. For example, NEF provides a service that allows an AF to provision specific subscription data (e.g., expected UE behavior) for various UEs. In general, NEF provides services similar to services provided by SCEF in EPC.

[0064] • Network Repository Function (NRF) with Nnrf interface - provides service registration and discovery, enabling NFs to identify appropriate services available from other NFs.

[0065] • Network Slice Selection Function (NSSF) with Nnssf interface - a “network slice” is a logical partition of a 5G network that provides specific network capabilities and characteristics, e.g., in support of a particular service. A network slice instance is a set of NF instances and the required network resources (e.g., compute, storage, communication) that provide the capabilities and characteristics of the network slice. The NSSF enables other NFs (e.g., AMF) to identify a network slice instance that is appropriate for a UE’s desired service.

[0066] • Authentication Server Function (AUSF) with Nausf interface - based in a user’s home network (HPLMN), it performs user authentication and computes security key materials for various purposes.

[0067] • Network Data Analytics Function (NWDAF) with Nnwdaf interface - provides network analytics information (e.g., statistical information of past events and / or predictive information) to other NFs on a network slice instance level.

[0068] • Unified Data Management (UDM) function with Nudm interface - generation of 3GPP authentication credentials, user identification handling, access authorization based on subscription data, and other subscriber-related functions. The UDM obtains the subscription data (including authentication data) from the 5G unified data repository (5GS- UDR, with Nudr interface). 5GS-UDR supports storage and retrieval of policy data by the PCF, as well as storage and retrieval of application data by NEF.

[0069] As briefly mentioned above, 3GPP has specified a service enabler architecture layer (SEAL) to support vertical applications over 5G networks. SEAL services include group management, configuration management, location management, identity management, key management, and network resource management. SEAL provides these services to vertical application layers (VALs) that can run on top of SEAL, e.g., corresponding to various applications.

[0070] SEAL can use various information for VAL group management, including VAL group IDs, group member UE IDs, generic group configurations, 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 3GPP TS 23.434 (vl7.7.0) clause 10 and 3GPP TS 29.549 (V17.8.0) clause 7.2.I.4.

[0071] Figure 3 shows a block diagram of a functional model for group management based on SEAL. A VAL UE may include a SEAL group management client that communicates via GM- UU interface with a corresponding SEAL group management server external to a 3GPP network. The group management client provides group management services to UE VAL client(s) in the UE via the GM-C interface. The VAL client(s) communicate via VAL-UU interface with corresponding VAL server(s) external to the 3GPP network. The VAL server(s) also communicate with the SEAL group management server via GM-S interface.

[0072] The group management server interacts with a NEF of the 3GPP network via N33 reference point to perform group management procedures for 5G Virtual Network (5GVN) groups. Group management functions defined by 3GPP SA6 involve group creation, query of group information, group membership update, group announcement and join, and group member leave.

[0073] The 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 possible via the N19 reference point in 5GC, which connects two UPFs for direct routing of traffic between different PDU Sessions without using N6 reference point shown in Figure 2.

[0074] As briefly mentioned above, an NWDAF in 5GC provides network analytics information (e.g., statistical information of past events and / or predictive information) to other NFs on a network slice instance level. Note that a “network slice” is a logical partition of a 5G network that provides specific network capabilities and characteristics, e.g., in support of a particular service. A network slice instance is a set of NF instances and the required network resources (e.g., compute, storage, communication) that provide the capabilities and characteristics of the network slice.

[0075] The 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 (vl7.9.0) specifies NWDAF as the main NF for computing analytics reports, and classifies NWDAF into two sub-functions (or logical functions): Analytics Logical Function (AnLF), which performs analytics procedures; and Model Training Logical Function (MTLF), which performs training and retraining of ML models used by AnLF.

[0076] Using rule- or machine learning (ML) based algorithms, NWDAF implements various analytics techniques to support network operations such as service orchestration, automated network configuration, and automated network operation. NWDAF can provide analytics results based on subscription and in response to a request. NWDAF can also implement closed-loop policy optimization in conjunction with the PCF in 5GC.

[0077] To illustrate further by example, NWDAF functionality can be used for one or more of the following use cases in a 5G network:

[0078] • Service quality, service experience, SLA monitoring and assurance;

[0079] • Detect and solve performance issues;

[0080] • Predicting network issues proactively adapting to network conditions;

[0081] • Maximizes return on network capacity investments;

[0082] • Optimize network resources;

[0083] • Network load performance monitoring analytics and load prediction;

[0084] • Network slice assurance;

[0085] • Service quality monitoring of subs groups; • Abnormal behavior and anomaly detection;

[0086] • Mobility-related information and prediction; and

[0087] • Congestion control.

[0088] Advanced analytics systems with NWDAF functionality rely on collecting and correlating elementary network events from different network domains, such as core, radio, and transport networks. Such analytics systems calculate user- and session-level end-to-end (E2E) service quality metrics (S-KPIs) as well as radio and network resource metrics (R-KPIs) that characterize the radio environment or network operation at user and session level. These types of solutions are suitable for session-based troubleshooting and analysis of network issues.

[0089] Event-based analytics require real-time collection and correlation of node and protocol events from different RAN and CN nodes, probing signaling interfaces, and sampling of userplane traffic. Additionally, event-based analytics require an advanced database, a rule engine, and an artificial intelligence (Al) framework that facilitates closed-loop retraining of ML models being used.

[0090] However, the information specified by 3GPP for VAL group management does not include UE communication characteristics, which can be very important 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 for communication characteristics. These parameters may include connection quality-of-service (QoS) data, guaranteed and maximum bitrates, packet delay budgets, and packet error rates. Moreover, each of these parameters may relate to communications within a group and between users in different groups.

[0091] Furthermore, conventional VAL group management / configuration is done manually by a programmer who relies to some degree on the site-specific parameters along with historical data and own experience. This non-trivial task is made even more difficult by inevitable changes of site specifications and / or physical arrangement. As such, manual VAL group configurations may not fulfil intra- and inter-group communication QoS and / or quality of experience (QoE) required under a service level agreement (SLA). Moreover, manual VAL group configurations are unlikely to produce an optimal resource allocation within the 5G system, instead resulting in traffic congestion, UE and / or network resource unavailability, and QoS degradations.

[0092] Embodiments of the present disclosure address one or more of these and related problems, issues, and / or difficulties by providing techniques and apparatus for automated configuration of VAL groups based on historical and / or predicted communication parameters in a communication network (e.g., 5G network). Embodiments can determine a preferred and / or optimum VAL group configuration for an arbitrary input architecture, including group memberships for each UE and inter-group connections. In this manner, embodiments can automate the very site-specific process of VAL group configurations in a 5G network. Moreover, the determined VAL group configuration can provide the necessary QoS and QoE for each service to satisfy any relevant SLAs.

[0093] Embodiments are based on using Convolutional Graph Neural Networks with Generative Adversarial Networks (GraphCGAN), which can handle graph neural networks (GNNs) for multiple layers (e.g., protocol layers) as well as an arbitrary number of graph edges between any two graph nodes. Some embodiments include and / or utilize a storage area for historical VAL group configurations (e.g., group connectivity) and corresponding performance (e.g., connection QoS data).

[0094] Furthermore, embodiments utilize a novel mapping of analytics data (e.g., in NWDAF) to QoS parameters (e.g., 5QI) of connections within and between VAL groups. Also, embodiments utilize per-flow observation and analytics to obtain service quality key performance indicators (KPIs) on demand for any parameter aggregation, e.g., per VAL group, per application, or a combination thereof.

[0095] Generative modelling involves automatically identifying and learning regularities or patterns in input data to produce a model, which can then be used to produce new examples that might have been reasonably derived from the original dataset. Generative adversarial networks (GANs) are an approach to generative modelling that involves deep learning techniques (e.g., convolutional neural networks).

[0096] In a GAN, two neural networks contest with each other in the form of a zero-sum game, where one agent's gain is another agent's loss. Given a training set, a GAN learns to generate new data with the same statistics as the training set. For example, a GAN trained on photographs can generate new photographs that look at least superficially authentic to human observers, with many realistic characteristics. Rather than being trained to minimize the distance to a specific image, the generator model is trained indirectly to “fool” a second model, called the “discriminator”. Put differently, the generator model is trained to create new instances and the discriminator model is trained to categorize examples as either real (from the training domain) or fake (from the generator). In general, when the two 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.

[0097] 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. In general, a graph is a collection of objects (i.e., nodes) and a set of interactions (i.e., edges) between pairs of these objects. For example, to encode a social network as a graph we might use nodes to represent individuals and use edges to represent that two individuals are friends.

[0098] Graph Neural Networks (GNNs) are a technique used to apply to graph-structured data the convolution operator at the core of Convolutional Neural Networks (CNNs) and Deep Learning. While CNNs are well defined for Euclidean data (e.g., images as grids of pixels), the convolution operator needs a new formulation for non-Euclidean domains like graphs. In 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 values provided by the aggregation are higher level features that will be fed to the next layer of the model. Thus, by stacking many layers together, it is possible to get higher level features that capture some higher order pattern.

[0099] An important component of GNN architecture is pairwise message transmission, which enables graph nodes to iteratively update their representations by exchanging information with their neighbors. Other reasons behind the success of GNNs are that they usually have low computational complexity due to the locality of the convolutional operator and they can generalize to previously unseen nodes and graphs.

[0100] As briefly mentioned above, embodiments are based on a technique called Convolutional Graph Neural Networks with Generative Adversarial Networks (GraphCGAN), which combines GNNs and GANs. Applicants have recognized that this technique is compatible with a physical network that includes inter-device connections on multiple protocol layers, specifically to generate graph attributes for multiple connections on multiple layers.

[0101] GANs utilized in embodiments of the present disclosure include a generator and a discriminator that work in tandem, with the generator generating artificial GNNs for candidate VAL group configurations and the discriminator evaluating the generated artificial GNNs. Typically, the generator learns to map from a latent space (e.g., a multivariate normal distribution) to a data distribution of interest, while the discriminator distinguishes candidates produced by the generator from the true data distribution. The generator's training objective is to increase the error rate of the discriminator, i.e., by “fooling” tire discriminator by producing novel artificial GNNs for candidate VAL group configurations that the discriminator thinks are part of the true data distribution.

[0102] A known dataset serves as the initial training data for the discriminator. Training involves presenting the discriminator with samples derived from the training dataset until it achieves acceptable accuracy. The generator is trained based on whether it succeeds in fooling the discriminator. Candidates synthesized by the generator are evaluated by the discriminator. Independent backpropagation procedures are applied to adjust the parameters of both generator and discriminator, so that the generator produces better samples (e.g., with less mean squared error (MSE)) while the discriminator becomes more skilled at identifying artificial GNNs.

[0103] These techniques can provide various benefits, advantages, and / or solutions to problems described herein. For example, NWDAF can measure parameters end-to-end to obtain and store historical data aggregated at various levels, e.g., per application, per VAL group, per UE, per tracking area, per cell, etc. NWDAF has access to network-wide data for each VAL group, allowing the configuration of VAL groups that are optimum across multiple network protocol layers (e.g., L2, L3 / L4, L7, VAL, etc.).

[0104] Moreover, by using a combination of GNNs and GANs, embodiments can efficiently optimize connectivity and QoS parameters of each VAL group as well as inter-group connections within the various protocol layers simultaneously, thereby providing an optimal configuration of the overall network. In this manner, embodiments allocate each UE to an optimal VAL group and determine optimal connections for the UEs within a VAL group and between VAL groups, while satisfying any relevant SLAs.

[0105] Figure 4 shows an example implementation of embodiments of the present disclosure in a 5G network. In this example, VAL group determination can be implemented as a function or block (410) hosted 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 instance, network slices 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.

[0106] The VAL group determination function in NWDAF communicates via NEF with a group management server (GM-Server), which is part of the SEAL function (497). The GM- server also communciates with a SEAL AF, which also communicates with an external data network (e.g., Internet) that is coupled to the RAN (e.g., NG-RAN) via the UPF. NWDAF also communicates with SMF, AMF, UPF, and other NFs in the 5GC (498) in a customary way.

[0107] Figure 5 shows an example arrangement of VAL groups in an IIoT environment. Each of VAL groups 1-3 include multiple UEs that can communicate within the group, with UEs in other VAL groups (in some cases), and with network devices such as an Edge server. The VAL layer consists of an arbitrary number of sub-layers, each of them belonging to separate VAL groups. Each VAL group segregates traffic for one specific Application. VAL sits on top of Open System Inter-connection (OSI) protocol Layer 7, which is also known as the Application layer. The UEs in the VAL groups may communicate on Layer 7 using well-known protocols such as HTTP, SNMP, etc. Beneath the Application layer are Layers 3-4, which are also known respectively as Transport and Network Layers. The UEs in the VAL groups may communicate with each other and / or with the Edge server on Layers 3-4 using well-known protocols such as TCP, UDP, IP, etc. Further down are Layers 1-2, which are also known respectively as Datalink and Physical layers. The UEs in the VAL groups may use common Layer 1 / 2 protocols such as medium access control (MAC), radio link control (RLC), and PHY for wireless communication with the 5G network. Altnerately, the UEs in the VAL groups may use common wired communication protocols on Layers 1-2, such as Ethernet-related protocols. In some embodiments, each of Layers 2-3 may involve two sub-layers to support GPRS Tunnelling Protocol (GTP) in the 5GC.

[0108] According to embodiments of the present disclosure, characteristics of connections between the 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 graphs in three separate layers. This is illustrated in Figure 5, which shows example graph nodes and edges. Note that graph edges can be within a protocol layer or between protocol layers. Each edge represent a connection between UEs via the network (or directly, via sidelink) and is assigned an attribute (e.g., a key performance indicator, KPI) pertaining to connection quality between the two nodes (UEs) connected to the edge, as explained in more detail below.

[0109] Based on this graph representation, embodiments determine an optimal arrangement of the following that satisifes a network level SLA:

[0110] • Connectivity of VAL group members in the VAL layer, including 5G QoS identifier (5QI) and QoS settings of the connections; and

[0111] • UE-network-UE connections of group member UEs at Layers 2-3 and 4, including for each separate VAL group as well as between VAL groups.

[0112] Figure 6 shows an example functional architecture for VAL group determination, according to some embodiments of the present disclosure. This architecture will be explained in terms of its various functions or operations, some of which are given numerical labels. However, these labels are intended to facilitate explanation rather than to require any specific operational order, unless expressly stated otherwise.

[0113] Initially, the system starts with a training set of historical VAL group configurations, with group configurations 1-3 shown in Figure 6 as an example. These historical group configurations can includes ones produced by the system itself and / or ones produced manually by an operator. For example, each of group configurations 1-3 can correspond to a past configuration of UEs in respective VAL groups (e.g., Figure 5), including connections on Layers 2, 3-4, and 7. In operation 1, an example VAL group configuration is selected from among these. In operation 2, a real multi-layered attributed graph is computed or prepared based on the example VAL group configuration from operation 1 and historical data for connections on Layers 2-4 that are relevant for the example VAL group configuration. This historical data can be provided by NWDAF. The following table shows various statistics defined in 3GPP TS

[0114] 23.288 (vl7.9.0) that are output by NWDAF and can be used as historical data for applications corresponding to respective VAL groups.

[0115] For the graph prepared in operation 2, the example VAL group configuration provides the node / edge structure for the graph while the NWDAF historical data provides the attributes for the graph. For example, a graph may be arranged in the form of multiple matrices - one per layer - with each matrix entry containing an attribute of a connection between two UEs via a data network in the example VAL group configuration. In other words, the attributes are for UE-Network-UE (UNU) connections, which are supported by the NWDAF historical data shown in the table above. In case the training set of historical VAL group configurations is not available, the NWDAF historical data can be used to determine an example real VAL group instance needed to obtain the real multi-layer attributed graph.

[0116] Initial graph attributes can be set in various ways, including the following example. Let PrioValyferepresent an integer priority value for the connection represented by the kthgraph edge in the jttlsub-layer of the VAL layer. This value can be calculated using detailed historical connectivity data according to the following technique.

[0117] The 5G QoS model is based on QoS Flows, and supports 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). The QoS Flow is the finest granularity of QoS differentiation in each UE PDU Session, and is identified by an assigned QoS Flow ID (QFI). UP traffic with the same QFI within a PDU Session receives the same traffic forwarding treatment (e.g., scheduling, admission threshold). Each QoS Flow is characterised by the following:

[0118] • a QoS profile provided by SMF to the RAN via AMF;

[0119] • one or more QoS rule(s) and optionally QoS Flow level QoS parameters associated with these QoS rule(s) which can be provided by SMF to UE via AMF; and

[0120] • one or more UL and DL packet detection rules (PDRs) provided by SMF to UPF.

[0121] 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 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.

[0122] 5QI is an integer or index that refers to 5G QoS characteristics defined in 3GPP TS 23.501 (vl7.9.0) clause 5.7, including Resource Type, Priority Level, Packet Delay Budget, Packet Error Rate, Averaging Window, and Maximum Data Burst Volume. In general, these are access nodespecific parameters that control QoS forwarding treatment (e.g., scheduling weights, admission thresholds, queue management thresholds, link layer protocol configuration, etc.) that a QoS Flow receives edge-to-edge between the UE and the UPF.

[0123] Thus, if the training set of historical VAL group configurations or the NWDAF historical data for a UNU connection in the graph includes 5QI, then PrioValyfecan be calculated as follows:

[0124] PrioValyfe= 10 PriorityLevel(5QI)yfe.

[0125] On the other hand, if the training set of historical VAL group configurations or the NWDAF historical data for a UNU connection in the graph does not include 5QI, then PrioValyfecan be calculated as follows: PrioValyfe= 10 EstimatedPriorityLevelyfe, EstimatedPriorityLevel = PriorityLevelNWDAF_DN_hist(AppID, T); fe, where PriorityLevelNWDAF_DN_hist(AppID, T)yfeis an evaluated Priority Level calculated for graph edge k in the jtflsub-layer of the VAL layer, using one or more data network (DN) performance statistics listed in the table above. AppID identifies a specific type of service that maps to sub-layer j, for which the NWDAF performance statistics are obtained and the 5QI is calculated. For example, PriorityLevelNWDAF DN hist(AppID, T); femaps the DN performance statistics of Maximum Packet Delay and Maximum Packet Loss Rate to potential 5QI values for AppID over a historical period T.

[0126] Assuming the subscriber profile of a UE allows a GBR connection for an edge (J, kf the technique can predict whether GBR QoS applies to edge (J, k) based on the comparison: (Minimum Traffic rate) > minTrRate for T > minTrRate, where Minimum Traffic Rate is a DN performance statistic from the table above and minTrRate and minTrRateT are pre-defined values. If the above comparison is not true, then edge (J, k) is determined to be a non-GBR connection, e.g., having minTrRate = INF.

[0127] Finally, by analyzing Network Status Indication {Percentage UL, Percentage DL} statistics listed in the table above, a definite 5QI value is assigned to the edge (J, k) from the shortlisted potential 5QI values. In other words, specific 5QI values are assigned to satisfy network QoS characteristics including packet delay budget, packet loss or priority, etc. An example is given in 3GPP TS 23.501 (vl7.9.0) Table 5.7.4-1, “Standardized 5QI to QoS characteristics mapping." Subsequently, a definite 5QI value is assigned to the edge( / , kf

[0128] In some embodiments, if an edge (J, k) is determined to be GBR, then the DN performance statistics Minimum Traffic Rate and Maximum Traffic Rate in the table above are also assigned as attributes for this edge. Note that these statistics (e.g., measured by NWDAF) correspond to Guaranteed Flow Bit Rate (GFBR) and Maximum Flow Bit Rate (MFBR) parameters, respectively, for a GBR QoS Flow.

[0129] According to 3GPP TS 23.501 (vl7.9.0) clause 5.8.2.11.4, a QoS enforcement rule (QER) defines how a packet shall be treated in terms of bit rate limitation and packet marking for QoS purposes. This QER definition includes “maximum bitrate” and “minimum bitrate” attributes, which define uplink / downlink maximum and minimum bitrates to be enforced for packets. In some embodiments, either or both of these attributes can be assigned as attributes for a graph edge (j, k) corresponding to the connection. Note however that the minimum bitrate attribute is only applicable to non-GBR QoS Flows. In addition to the connections within each VAL group, some embodiments can also include in the graph representation any connections between VAL groups or sub-layers. In a 5G network, inter-group connections are realized via N19 interface between UPFs in the 5GC. In such embodiments, the NWDAF historical data can be used to assign attributes to edges corresponding to intra-group connections in a similar manner as described above.

[0130] Operations 3-4 involve generation of artificial multi-layer attributed graphs that are compared by the discriminator function to the real multi-layer attributed graph generated in operation 2. Operation 3 starts with the generator receiving a random input seed that it uses to generate a random VAL group configuration, which is denoted “Config X” in Figure 6. For example, a random value having uniform distribution in the range of (0, 1) is generated for each input neuron of the multi-layer GNN. Optionally, preferred VAL group configuration(s) can be used as input to generate “Config X”. For example, user-preferred group definitions can be introduced as SLA criteria and / or general generator policies. In operation 4, a graph (or GNN) of nodes and edges is generated for Config X, including nodes and edges similar to the real structure discussed above. The generator takes a random seed and prepares a generated structure for the VAL groups.

[0131] Additionally, NWDAF predictions for QoS, etc. are used to populate the attributes of the graph determined in operation 4. In other words, operation 4 assigns attributes to the graph edges in a similar manner as operation 2, but using predicted data rather than historical data. For example, a similar technique can be used to convert NWDAF predictions to 5QI values and then to PrioValfeassigned as graph edge attributes.

[0132] The mapping of NWDAF historical data and predictions to edge attributes to be applied to VAL group configurations in operations 2 and 4 is shown in Figure 6 as being performed by NWDAF. However, it is possible that this mapping may be performed external to the NWDAF, such as by a SEAL group management server, an AF, or another NF that can access the NWDAF historical data or predictions.

[0133] In operation 5, the discriminator function of the GraphCGAN compares the real and artificial multi-layer attributed 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 party of the GAN training. The discriminator also checks whether the VAL group configuration corresponding to the artificial multi-layer attributed graph fulfils the relevant network-level SLA for the VAL groups. For example, SLAs include target criteria for QoS parameters such as packet loss, delay, bandwidth, etc. Actual values for these parameters can be monitored per session by NWDAF for a VAL group configuration, and stored as part of the historical data. In the context of training, the NWDAF can compare (historical) monitored and target values.

[0134] If the generated VAL group configuration fulfils the SLA and the discriminator cannot differentiate between the two input graphs, then this configuration is considered as a VAL group policy for later use and the training process stops. If either or both of those conditions are not met, a new random input seed is triggered for the generator function, which it uses to generate a different random VAL group configuration in a similar way as operation 3.

[0135] In operation 6, the VAL group policy is stored for later use by the generator function for inference of VAL group configurations in response to a request or query. In operation 7, the system is queried for a new VAL group structure. The query provides some input on the architecture for which VAL groups should be determined. For example, the architecture input can include a list of UEs to be placed into VAL groups, a list of applications running on the respective UEs, a list of session IDs for the applications, etc. As a more specific example, user-preferred group definitions can be introduced as SLA criteria and / or general generator policies.

[0136] Based on this input and the stored VAL group policy, the generator performs inference to determine a proposed VAL group configuration including group membership, configurations of connections within and between groups, etc. Additionally, NWDAF predictions for QoS, etc. are used to populate the attributes of the graph determined in operation 7, in a similar manner as discussed above for operation 4.

[0137] Figure 7 shows another example functional architecture for VAL group determination, according to other embodiments of the present disclosure. This architecture is similar to the architecture shown in Figure 6 but differs in several ways described below.

[0138] For example, the generator in the Figure 7 is arranged to generate attributed GNNs, whereas the generator in Figure 6 is arranged to generate unattributed GNNs. This is shown in the training phase as two separate outputs: Config X including nodes and edges, and the attributes of the nodes and edges. In the inference phase, the generator takes as input a generated configuration of nodes and edges and the attributes of the nodes and edges, which are determined in a similar manner as described above based on predictions of UNU connection QoS parameters provided by the NWDAF. The predictions are requested from NWDAF during inference based on a reverse mapping from the architecture-related inputs to needed NWDAF QoS metrics.

[0139] In some embodiments, obtaining the predicted attributes from NWDAF during inference can be omitted, with the attributes of the generated configuration being used instead. However, the NWDAF predictions may provide higher quality attributes compared to the attributes of the generated configuration. Moreover, in some cases, the NWDAF predictions can improve formatting of the attributes according to the requirements of the generator. The embodiments described above can be further illustrated with reference to Figure 8, which depicts an exemplary method (e.g., procedures) for determining VAL groups for operation in a communication network, with each VAL group including a plurality of UEs, according to various embodiments of the present disclosure. Put differently, various features of the operations described below correspond to various embodiments described above. Although the exemplary method is illustrated in Figure 8 by specific blocks in a particular order, the operations corresponding to the blocks can be performed in a different order than shown and can be combined and / or divided into blocks having different functionality than shown. Optional blocks and / or operations are indicated by dashed lines.

[0140] The exemplary method illustrated by Figure 10 can be performed by a network node or function (NNF) of the communication network (e.g., 5G wireless network), such as described in the examples above.

[0141] The exemplary method includes the operations of block 810, where the NNF can determine a first multi-layer graph representation based on the following: a reference VAL group configuration previously used in the communication network, and historical quality of service (QoS) metrics for multi-layer connections between UEs that comprise the reference VAL group configuration. The exemplary method also includes performing the first operations of blocks 820-830 using a generative adversarial network (GAN) comprising a generator and a discriminator, specifically the first operations listed below:

[0142] • (820) using the generator, determining a second multi-layer graph representation based on the following: a candidate VAL group configuration determined based on a random input, and predicted QoS metrics for multi-layer connections between the UEs that comprise the reference VAL group configuration;

[0143] • (830) using the discriminator, determining whether the following first conditions are met: the second multi-layer graph representation cannot be differentiated from the first multi-layer graph representation, and the predicted QoS metrics associated with the candidate VAL group configuration fulfil one or more service level agreements (SLAs) for operation of the communication network;

[0144] The exemplary method also includes the operations of block 850, where the NNF can repeat the first operations based on a different random input when the first conditions are not met. The exemplary method also includes the operations of block 860, where after the first conditions are met and using the generator, the NNF can perform inference based on the second multi-layer graph representation to generate a configuration of VAL groups for operation in the communication network. In some embodiments, the method is performed by an NWDAF of the communication network. In other embodiments, the method is performed by a SEAL group management (GM) server external to (or within) the communication network. In some of these embodiments, the method is performed by the NWDAF and the inference is further based on a desired architecture for VAL groups that is obtained from the SEAL GM server.

[0145] In some embodiments, the configuration of VAL groups generated by performing inference in block 860 includes the following: a plurality of UEs arranged in multiple VAL groups, and connections between respective pairs of the UEs. In some of these embodiments, the connections between respective pairs of the UEs including the following: connections between pairs of UEs in a same VAL group, and connections between pairs of UEs in different VAL groups.

[0146] In some of these embodiments, the configuration of VAL groups generated by performing inference also includes QoS settings for the connections between respective pairs of the UEs. Figure 7 shows an example of these embodiments. In other of these embodiments, the exemplary method also includes the operations of block 870, where the NNF can obtain, from an NWDAF of the communication network, predicted QoS settings for the connections between respective pairs of the UEs. Figure 6 shows an example of these embodiments.

[0147] In some embodiments, the reference VAL group configuration and each candidate VAL group configuration includes the following: a plurality of UEs arranged in multiple VAL groups, and connections on multiple protocol layers between respective pairs of the UEs. In some of these embodiments, the multiple VAL groups are associated with respective applications or services and each of the historical QoS metrics and the predicted QoS metrics is related to data traffic for one of the applications or services.

[0148] In some of these embodiments, each of the first and second multi-layer graph representations comprises:

[0149] • a plurality of first layers, with each first layer representing connections on a corresponding single protocol layer between respective pairs of the UEs; and

[0150] • a second layer representing connections across multiple protocol layers between respective pairs of the UEs.

[0151] In some of these embodiments, determining the first multi-layer graph representation in block 810 includes the following operations, labelled with corresponding sub-block numbers:

[0152] • (811) determining a graph structure based on the reference VAL group configuration, wherein the graph structure includes the plurality of UEs as nodes and the connections between respective pairs of the UEs as edges; and

[0153] • (812) determining attributes for the edges based on the historical QoS metrics. In some of these embodiments, determining attributes for the edges based on the historical QoS metrics in sub-block 812 includes the following operations, for each connection between a pair of UEs:

[0154] • obtaining the one or more historical QoS metrics for the connection from an NWDAF of the communication network;

[0155] • determining, as the attribute for the edge corresponding to the connection, an integer priority value based on the one or more historical QoS metrics.

[0156] In some further variants, each integer priority value is determined based on the following:

[0157] • a 5G QoS Identifier (5QI), when included in the one or more historical QoS metrics; or

[0158] • at least one of the following historical QoS metrics, when a 5QI is not included: o UL throughput for the application or service as a percentage of total UL throughput in an area of interest; o DL throughput for the application or service as a percentage of total DL throughput in an area of interest; o average traffic rate observed for UEs communicating with the application or service; o maximum traffic rate observed for UEs communicating with the application or service; o minimum traffic rate observed for UEs communicating with the application or service; o average packet delay observed for UEs communicating with the application or service; o maximum packet delay observed for UEs communicating with the application or service; o average packet loss rate observed for UEs communicating with the application or service; and o maximum packet loss rate observed for UEs communicating with the application or service.

[0159] In some of these embodiments, the generator and the discriminator include respective multi-layer graph neural networks (mGNNs). In some variants of these embodiments, the exemplary method can also include the operations of block 840, where when it is determined (e.g., in block 830) that the second multi-layer graph representation can be differentiated from the first multi-layer graph representation, the NNF can adapt the mGNNs of the generator and the discriminator based on a metric representative of differences between the first and second multi-layer graph representations. In such case, the adapted mGNNs of the generator and the discriminator are used to repeat the first operations. In some further variants, the adapted mGNN of the generator is used to perform inference in block 860 after it is determined (e.g., in block 830) that the first conditions are met. In some variants of these embodiments, determining the second multi-layer graph representation in block 820 includes the following operations, labelled with corresponding subblock numbers:

[0160] • (821) determining the candidate VAL group configuration based on applying the generator mGNN to the random input;

[0161] • (822) determining a graph structure based on the candidate VAL group configuration, wherein the graph structure includes the plurality of UEs as nodes and the connections between respective pairs of the UEs as edges; and

[0162] • (823) determining attributes for the edges based on the historical QoS metrics. Figure 6 shows an example of these variants.

[0163] In other variants of these embodiments, determining the second multi-layer graph representation in block 820 includes the operations of sub-block 824, where based on applying the generator mGNN to the random input, the NNF can determine an attributed graph structure for a candidate VAL group configuration. The attributed graph structure includes the plurality of UEs as nodes, the connections among the plurality of UEs as edges, and respective QoS metrics as attributes for the edges. Figure 7 shows an example of these variants.

[0164] Although various embodiments are described above in terms of methods, techniques, and / or procedures, the person of ordinary skill will readily comprehend that such methods, techniques, and / or procedures can be embodied by 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.

[0165] Figure 9 shows an example of a communication system 900 in accordance with some embodiments. In this example, communication system 900 includes a telecommunication network 902 that includes an access network 904 (e.g., RAN) and a core network 906, which includes one or more core network nodes 908. Access network 904 includes one or more access network nodes, such as network nodes 910a-b (one or more of which may be generally referred to as network nodes 910), or any other similar 3GPP access nodes or non-3GPP access points. Moreover, as will be appreciated by those of skill in the art, a network node is not necessarily limited to an implementation in which a radio portion and a baseband portion are supplied and integrated by a single vendor. Thus, it will be understood that network nodes include disaggregated implementations or portions thereof. For example, in some embodiments, telecommunication network 902 includes one or more Open-RAN (ORAN) network nodes. An ORAN network node is a node in telecommunication network 902 that supports an ORAN specification (e.g., a specification published by the O-RAN Alliance, or any similar organization) and may operate alone or together with other nodes to implement one or more functionalities of any node in telecommunication network 902, including one or more network nodes 910 and / or core network nodes 908.

[0166] Examples of an ORAN network node include an open radio unit (O-RU), an open distributed unit (O-DU), an open central unit (O-CU), including an O-CU control plane (O-CU- CP) or an O-CU user plane (O-CU-UP), a RAN intelligent controller (near-real time or non-real time) hosting software or software plug-ins, such as a near-real time control application (e.g., xApp) or a non-real time control application (e.g., rApp), or any combination thereof (the adjective “open” designating support of an ORAN specification). The network node may support a specification by, for example, supporting an interface defined by the ORAN specification, such as an Al, Fl, Wl, El, E2, X2, Xn interface, an open fronthaul user plane interface, or an open fronthaul management plane interface. Moreover, an ORAN access node may be a logical node in a physical node. Furthermore, an ORAN network node may be implemented in a virtualization environment (described further below) in which one or more network functions are virtualized. For example, the virtualization environment may include an O-Cloud computing platform orchestrated by a Service Management and Orchestration Framework via an 0-2 interface defined by the 0-RAN Alliance or comparable technologies. Network nodes 910 facilitate direct or indirect connection of UEs, such as by connecting UEs 912a-d (one or more of which may be generally referred to as UEs 912) to core network 906 over one or more wireless connections.

[0167] Example wireless communications over a wireless connection include transmitting and / or receiving wireless signals using electromagnetic waves, radio waves, infrared waves, and / or other types of signals suitable for conveying information without the use of wires, cables, or other material conductors. Moreover, in different embodiments, communication system 900 may include any number of wired or wireless networks, network nodes, UEs, and / or any other components or systems that may facilitate or participate in the communication of data and / or signals whether via wired or wireless connections. Communication system 900 may include and / or interface with any type of communication, telecommunication, data, cellular, radio network, and / or other similar type of system.

[0168] UEs 912 may be any of a wide variety of communication devices, including wireless devices arranged, configured, and / or operable to communicate wirelessly with network nodes 910 and other communication devices. Similarly, network nodes 910 are arranged, capable, configured, and / or operable to communicate directly or indirectly with UEs 912 and / or with other network nodes or equipment in telecommunication network 902 to enable and / or provide network access, such as wireless network access, and / or to perform other functions, such as administration in telecommunication network 902. In the depicted example, core network 906 connects network nodes 910 to one or more hosts, such as host 916. These connections may be direct or indirect via one or more intermediary networks or devices. In other examples, network nodes may be directly coupled to hosts. Core network 906 includes one or more core network nodes (e.g., 908) that are structured with hardware and software components. Features of these components may be substantially similar to those described with respect to the UEs, network nodes, and / or hosts, such that the descriptions thereof are generally applicable to the corresponding components of core network node 908. Example core network nodes include functions of one or more of a 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 Exposure Function (NEF), and / or a User Plane Function (UPF).

[0169] Host 916 may be under the ownership or control of a service provider other than an operator or provider of access network 904 and / or telecommunication network 902, and may be operated by the service provider or on behalf of the service provider. Host 916 may host a variety of applications to provide one or more service. Examples of such applications include live and pre-recorded audio / video content, data collection services such as retrieving and compiling data on various ambient conditions detected by a plurality of UEs, analytics functionality, social media, functions for controlling or otherwise interacting with remote devices, functions for an alarm and surveillance center, or any other such function performed by a server.

[0170] As a whole, communication system 900 of Figure 9 enables connectivity between the UEs, network nodes, and hosts. In that sense, the communication system may be configured to operate according to predefined rules or procedures, such as specific standards that include, but are 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 future generation standard (e.g., 6G); wireless local area network (WLAN) standards, such as the Institute of Electrical and Electronics Engineers (IEEE) 802.11 standards (WiFi); and / or any other appropriate wireless communication standard, such as the Worldwide Interoperability for Microwave Access (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.

[0171] In some examples, telecommunication network 902 is a cellular network that implements 3GPP standardized features. Accordingly, telecommunication network 902 may support network slicing to provide different logical networks to different devices that are connected to telecommunication network 902. For example, telecommunication network 902 may 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 loT services to yet further UEs.

[0172] In some examples, UEs 912 are configured to transmit and / or receive information without direct human interaction. For instance, a UE may be designed to transmit information to access network 904 on a predetermined schedule, when triggered by an internal or external event, or in response to requests from access network 904. Additionally, a UE may be configured for operating in single- or multi-RAT or multi- standard mode. For example, a UE may operate with any one or combination of Wi-Fi, NR (New Radio) and LTE, i.e. being configured for multi-radio dual connectivity (MR-DC), such as E-UTRAN (Evolved-UMTS Terrestrial Radio Access Network) New Radio - Dual Connectivity (EN-DC).

[0173] In the example, hub 914 communicates with access network 904 to facilitate indirect communication between one or more UEs (e.g., UE 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, or any of the other communication devices described herein regarding UEs. For example, hub 914 may be a broadband router enabling access to core network 906 for the UEs. As another example, hub 914 may be a controller that sends commands or instructions to one or more actuators in the UEs. Commands or instructions may be received from the UEs, network nodes 910, or by executable code, script, process, 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, may perform analysis or other processing of the data. As another example, hub 914 may be a content source. For example, for a UE that is a VR headset, display, loudspeaker or other media delivery device, hub 914 may retrieve VR assets, video, audio, or other media or data related to sensory information via a network node, which hub 914 then provides to the UE either directly, after performing local processing, and / or after adding additional local content. In still another example, hub 914 acts as a proxy server or orchestrator for the UEs, in particular if one or more of the UEs are low energy loT devices.

[0174] Hub 914 may have a constant / persistent or intermittent connection to network node 910b. Hub 914 may also allow for a different communication scheme and / or schedule between hub 914 and UEs (e.g., UE 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. Moreover, hub 914 may be configured to connect to an M2M service provider over access network 904 and / or to another UE over a direct connection. In some scenarios, UEs may establish a wireless connection with network nodes 910 while still connected via hub 914 via a wired or wireless connection. In some embodiments, hub 914 may be a dedicated hub - that is, a hub whose primary function is to route communications to / from the UEs from / to network node 910b. In other embodiments, hub 914 may be a non-dedicated hub - that is, a device which can route communications between the UEs and network node 910b, but which is additionally capable of operating as a communication start and / or end point for certain data channels.

[0175] In some embodiments, core network node 908 or host 916 can be configured to perform operations corresponding to the exemplary techniques described above in relation to Figures 6-8.

[0176] Figure 10 shows a network node 1000 in accordance with some embodiments. 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 Bs, eNBs, gNBs), and O-RAN nodes or components of an O-RAN node (e.g., O-RU, O-DU, O-CU).

[0177] Base stations may be categorized based on the amount of coverage they provide (or, stated differently, their transmit power level) and so, depending on the provided amount of coverage, may be referred to as femto base stations, pico base stations, micro base stations, or macro base stations. A base station may be a relay node or a relay donor node controlling a relay. A network node may also include one or more (or all) parts of a distributed radio base station such as centralized digital units, distributed units (e.g., in an O-RAN access node) and / or remote radio units (RRUs), sometimes referred to as Remote Radio Heads (RRHs). Such remote radio units may or may not be integrated with an antenna as an antenna integrated radio. Parts of a distributed radio base station may also be referred to as nodes in a distributed antenna system (DAS).

[0178] Other examples of network nodes include multiple transmission point (multi-TRP) 5G access nodes, multi-standard radio (MSR) equipment such as MSR BSs, network controllers such as radio network controllers (RNCs) or base station controllers (BSCs), base transceiver stations (BTSs), transmission points, transmission nodes, multi-cell / multicast coordination entities (MCEs), Operation and Maintenance (O&M) nodes, Operations Support System (OSS) nodes, Self-Organizing Network (SON) nodes, positioning nodes (e.g., Evolved Serving Mobile Location Centers (E-SMLCs)), and / or Minimization of Drive Tests (MDTs).

[0179] Network node 1000 includes a processing circuitry 1002, a memory 1004, a communication interface 1006, and a power source 1008. Network node 1000 may be composed of multiple physically separate components (e.g., a NodeB component and a RNC component, or a BTS component and a BSC component, etc.), which may each have their own respective components. In certain scenarios in which network node 1000 comprises multiple separate components (e.g., BTS and BSC components), one or more of the separate components may be shared among several network nodes. For example, a single RNC may control multiple NodeBs. In such a scenario, each unique NodeB and RNC pair, may in some instances 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) and some components may be reused (e.g., a same antenna 1010 may be shared by different RATs). Network node 1000 may also include multiple sets of the various illustrated components for different wireless technologies integrated into network node 1000, for example GSM, WCDMA, LTE, NR, WiFi, Zigbee, Z- wave, LoRaWAN, Radio Frequency Identification (RFID) or Bluetooth wireless technologies. These wireless technologies may be integrated into the same or different chip or set of chips and other components within network node 1000.

[0180] The processing circuitry 1002 may comprise a combination of one or more of a microprocessor, controller, microcontroller, central processing unit, digital signal processor, application-specific integrated circuit, field programmable gate array, or any other suitable computing device, resource, or combination of hardware, software and / or encoded logic operable to provide, either alone or in conjunction with other network node 1000 components, such as memory 1004, to provide network node 1000 functionality.

[0181] 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 radio frequency (RF) transceiver circuitry 1012 and baseband processing circuitry 1014. In some embodiments, RF transceiver circuitry 1012 and baseband processing circuitry 1014 may be on separate chips (or sets of chips), boards, or units, such as radio units and digital units. In alternative embodiments, part or all of RF transceiver circuitry 1012 and baseband processing circuitry 1014 may be on the same chip or set of chips, boards, or units.

[0182] Memory 1004 may comprise any form of volatile or non-volatile computer-readable memory including, without limitation, persistent storage, solid-state memory, remotely mounted memory, magnetic media, optical media, random access memory (RAM), read-only memory (ROM), mass storage media (for example, a hard disk), removable storage media (for example, a flash drive, a Compact Disk (CD) or a Digital Video Disk (DVD)), and / or any other volatile or non-volatile, non-transitory device-readable and / or computer-executable memory devices that store information, data, and / or instructions that may be used by the processing circuitry 1002. Memory 1004 may store any suitable instructions, data, or information, including a computer program, software, an application including one or more of logic, rules, code, tables, and / or other instructions (collected denoted computer program 1004a, which may be in the form of a computer program product) capable of being executed by the processing circuitry 1002 and utilized by network node 1000. Memory 1004 may be used to store any calculations made by the processing circuitry 1002 and / or any data received via communication interface 1006. In some embodiments, the processing circuitry 1002 and memory 1004 is integrated.

[0183] Communication interface 1006 is used in wired or wireless communication of signaling and / or data between a network node, access network, and / or UE. As illustrated, communication interface 1006 comprises port(s) / terminal(s) 1016 to send and receive data, for example to and from a network over a wired connection. Communication interface 1006 also includes radio frontend circuitry 1018 that may be coupled to, or in certain embodiments a part of, antenna 1010. Radio front-end circuitry 1018 comprises filters 1020 and amplifiers 1022. Radio front-end circuitry 1018 may be connected to an antenna 1010 and processing circuitry 1002. The radio front-end circuitry may be configured to condition signals communicated between antenna 1010 and processing circuitry 1002. Radio front-end circuitry 1018 may receive digital data that is to be sent out to other network nodes or UEs via a wireless connection. Radio front-end circuitry 1018 may convert the digital data into a radio signal having the appropriate channel and bandwidth parameters using a combination of filters 1020 and / or amplifiers 1022. The radio signal may then be transmitted via antenna 1010. Similarly, when receiving data, antenna 1010 may collect radio signals which are then converted into digital data by radio front-end circuitry 1018. The digital data may be passed to the processing circuitry 1002. In other embodiments, the communication interface may comprise different components and / or different combinations of components.

[0184] In certain alternative embodiments, network node 1000 does not include separate radio front-end circuitry 1018, instead, the processing circuitry 1002 includes radio front-end circuitry and is connected to antenna 1010. Similarly, in some embodiments, all or some of RF transceiver circuitry 1012 is part of communication interface 1006. In still 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, which is part of a digital unit (not shown).

[0185] Antenna 1010 may include one or more antennas, or antenna arrays, configured to send 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 transmitting and receiving data and / or signals wirelessly. In certain embodiments, antenna 1010 is separate from network node 1000 and connectable to network node 1000 through an interface or port.

[0186] Antenna 1010, communication interface 1006, and / or the processing circuitry 1002 may be configured to perform any receiving operations and / or certain obtaining operations described herein as being performed by the network node. Any information, data and / or signals may be received from a UE, another network node and / or any other network equipment. Similarly, antenna 1010, communication interface 1006, and / or the processing circuitry 1002 may be configured to perform any transmitting operations described herein as being performed by the network node. Any information, data and / or signals may be transmitted to a UE, another network node and / or any other network equipment.

[0187] Power source 1008 provides power to the various components of network node 1000 in a form suitable for the respective components (e.g., at a voltage and current level needed for each respective component). Power source 1008 may further comprise, or be coupled to, power management circuitry to supply the components of network node 1000 with power for performing the functionality described herein. For example, network node 1000 may be connectable to an external power source (e.g., the power grid, an electricity outlet) via an input circuitry or interface such as an electrical cable, whereby the external power source supplies power to power circuitry of power source 1008. As a further example, power source 1008 may comprise a source of power in the form of a battery or battery pack which is connected to, or integrated in, power circuitry. The battery may provide backup power should the external power source fail.

[0188] Embodiments of network node 1000 may include additional components beyond those shown in Figure 10 for providing certain aspects of the network node’s functionality, including any of the functionality described herein and / or any functionality necessary to support the subject matter described herein. For example, network node 1000 may include user interface equipment to allow input of information into network node 1000 and to allow output of information from network node 1000. This may allow a user to perform diagnostic, maintenance, repair, and other administrative functions for network node 1000.

[0189] In some embodiments, network node 1000 can be configured to perform operations corresponding to the exemplary techniques described above in relation to Figures 6-8.

[0190] Figure 11 is a block diagram of a host 1100, which may be an embodiment of host 916 of Figure 9, in accordance with various aspects described herein. As used herein, host 1100 may be or comprise various combinations hardware and / or software, including a standalone server, a blade server, a cloud-implemented server, a distributed server, a virtual machine, container, or processing resources in a server farm. Host 1100 may provide one or more services to one or more UEs.

[0191] Host 1100 includes processing circuitry 1102 that is operatively coupled via a bus 1104 to an input / output interface 1106, a network interface 1108, a power source 1110, and a memory 1112. Other components may be included in other embodiments. Features of these components may be substantially similar to those described with respect to the devices of previous figures, such as Figures QQ2 and 10, such that the descriptions thereof are generally applicable to the corresponding components of host 1100. Memory 1112 may include one or more computer programs including one or more host application programs 1114 and data 1116, which may include user data, e.g., data generated by a UE for host 1100 or data generated by host 1100 for a UE. Embodiments of host 1100 may utilize all or only a subset of the components shown. Host application programs 1114 may be implemented in a container-based architecture and may provide support for video codecs (e.g., Versatile Video Coding (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 classes, types, or implementations of UEs (e.g., handsets, desktop computers, wearable display systems, heads-up display systems). Host application programs 1114 may also provide for user authentication and licensing checks and may periodically report health, routes, and content availability to a central node, such as a device in or on the edge of a core network. Accordingly, host 1100 may select and / or indicate a different host for over-the-top services for a UE. Host application programs 1114 may support various protocols, such as the HTTP Live Streaming (HLS) protocol, Real-Time Messaging Protocol (RTMP), Real- Time Streaming Protocol (RTSP), Dynamic Adaptive Streaming over HTTP (MPEG-DASH), etc.

[0192] In some embodiments, host 1100 can be configured to perform operations corresponding to the exemplary techniques described above in relation to Figures 6-8.

[0193] Figure 12 is a block diagram illustrating a virtualization environment 1200 in which functions implemented by some embodiments may be virtualized. In the present context, virtualizing means creating virtual versions of apparatuses or devices which may include virtualizing hardware platforms, storage devices and networking resources. As used herein, virtualization can be applied to any device described herein, or components thereof, and relates to an implementation in which at least a portion of the functionality is implemented as one or more virtual components. Some or all of the functions described herein may be implemented as virtual components executed by one or more virtual machines (VMs) implemented in one or more virtual environments 1200 hosted by one or more of hardware nodes, such as a hardware computing device that operates as a network node, UE, core network node, or host. Further, in embodiments in which the virtual node does not require radio connectivity (e.g., a core network node or host), then the node may be entirely virtualized. In some embodiments, the virtualization environment 1200 includes components defined by the O-RAN Alliance, such as an O-Cloud environment orchestrated by a Service Management and Orchestration Framework via an O-2 interface.

[0194] Applications 1202 (which may alternatively be called software instances, virtual appliances, network functions, virtual nodes, virtual network functions, etc.) are run in the virtualization environment 1200 to implement some of the features, functions, and / or benefits of some of the embodiments disclosed herein. As a specific example, operations corresponding to the exemplary techniques described above in relation to Figures 6-8 can be instantiated as application(s) 1402 running in virtualization environment 1400.

[0195] Hardware 1204 includes processing circuitry, memory that stores software and / or instructions (collected denoted computer program 1204a, which may be in the form of a computer program product) executable by hardware processing circuitry, and / or other hardware devices as described herein, such as a network interface, input / output interface, and so forth. Software may be executed by the processing circuitry to instantiate one or more virtualization layers 1206 (also referred to as hypervisors or virtual machine monitors (VMMs)), provide VMs 1208a and 1208b (one or more of which may be generally referred to as VMs 1208), and / or perform any of the functions, features and / or benefits described in relation with some embodiments described herein. Virtualization layer 1206 may present a virtual operating platform that appears like networking hardware to the VMs 1208.

[0196] VMs 1208 comprise virtual processing, virtual memory, virtual networking or interface and virtual storage, and may be run by a corresponding virtualization layer 1206. Different embodiments of the instance of a virtual appliance 1202 may be implemented on one or more of VMs 1208, and the implementations may be made in different ways. Virtualization of the hardware is in some contexts referred to as network function virtualization (NFV). NFV may be used to consolidate many network equipment types onto industry standard high volume server hardware, physical switches, and physical storage, which can be located in data centers, and customer premise equipment.

[0197] In the context of NFV, each VM 1208 may 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, and that part of hardware 1204 that executes that VM, be it hardware dedicated to that VM and / or hardware shared by that VM with others of the VMs, forms separate virtual network elements. Still in the context of NFV, a virtual network function is responsible for handling specific network functions that run in one or more VMs 1208 on top of the hardware 1204 and corresponds to the application 1202.

[0198] Hardware 1204 may be implemented in a standalone network node with generic or specific components. Hardware 1204 may implement some functions via virtualization. Alternatively, hardware 1204 may be part of a larger cluster of hardware (e.g. such as in a data center or CPE) where many hardware nodes work together and are managed via management and orchestration function 1210, which, among others, oversees lifecycle management of applications 1202. In some embodiments, hardware 1204 is coupled to one or more radio units that each include one or more transmitters and one or more receivers that may be coupled to one or more antennas. Radio units may communicate directly with other hardware nodes via one or more appropriate network interfaces and may be used in combination with the virtual components to provide a virtual node with radio capabilities, such as a radio access node or a base station. In some embodiments, some signaling can be provided with the use of a control system 1212 which may alternatively be used for communication between hardware nodes and radio units.

[0199] The foregoing merely illustrates the principles of the disclosure. Various modifications and alterations to the described embodiments will be apparent to those skilled in the art in view of the teachings herein. It will thus be appreciated that those skilled in the art will be able to devise numerous systems, arrangements, and procedures that, although not explicitly shown or described herein, embody the principles of the disclosure and can be thus within the spirit and scope of the disclosure. Various exemplary embodiments can be used together with one another, as well as interchangeably therewith, as should be understood by those having ordinary skill in the art.

[0200] The term unit, as used herein, can have conventional meaning in the field of electronics, electrical devices and / or electronic devices and can include, for example, electrical and / or electronic circuitry, devices, modules, processors, memories, logic solid state and / or discrete devices, computer programs or instructions for carrying out respective tasks, procedures, computations, outputs, and / or displaying functions, and so on, as such as those that are described herein.

[0201] Any appropriate steps, methods, features, functions, or benefits disclosed herein may be performed through one or more functional units or modules of one or more virtual apparatuses. Each virtual apparatus may comprise a number of these functional units. These functional units may be implemented via processing circuitry, which may include one or more microprocessor or microcontrollers, as well as other digital hardware, which may include Digital Signal Processor (DSPs), special-purpose digital logic, and the like. The processing circuitry may be configured to execute program code stored in memory, which may include one or several types of memory such as Read Only Memory (ROM), Random Access Memory (RAM), cache memory, flash memory devices, optical storage devices, etc. Program code stored in memory includes program instructions for executing one or more telecommunications and / or data communications protocols as well as instructions for carrying out one or more of the techniques described herein. In some implementations, the processing circuitry may be used to cause the respective functional unit to perform corresponding functions according to one or more embodiments of the present disclosure.

[0202] As described herein, device and / or apparatus can be represented by a semiconductor chip, a chipset, or a (hardware) module comprising such chip or chipset; this, however, does not exclude the possibility that a functionality of a device or apparatus, instead of being hardware implemented, be implemented as a software module such as a computer program or a computer program product comprising executable software code portions for execution or being run on a processor. Furthermore, functionality of a device or apparatus can be implemented by any combination of hardware and software. A device or apparatus can also be regarded as an assembly of multiple devices and / or apparatuses, whether functionally in cooperation with or independently of each other. Moreover, devices and apparatuses can be implemented in a distributed fashion throughout a system, so long as the functionality of the device or apparatus is preserved. Such and similar principles are considered as known to a skilled person.

[0203] Unless otherwise defined, all terms (including technical and scientific terms) used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this disclosure belongs. It will be further understood that terms used herein should be interpreted as having a meaning that is consistent with their meaning in the context of this specification and the relevant art and will not be interpreted in an idealized or overly formal sense unless expressly so defined herein.

[0204] In addition, certain terms used in the present disclosure, including the specification and drawings, can be used synonymously in certain instances (e.g., “data” and “information”). It should be understood, that although these terms (and / or other terms that can be synonymous to one another) can be used synonymously herein, there can be instances when such words can be intended to not be used synonymously.

Claims

CLAIMS1. A method for determining vertical application layer, VAL, groups for operation in a communication network, each VAL group including a plurality of user equipment, UEs, the method comprising: determining (810) a first multi-layer graph representation based on the following: a reference VAL group configuration previously used in the communication network, and historical quality of service, QoS, metrics for multi-layer connections between UEs that comprise the reference VAL group configuration; performing the following first operations using a generative adversarial network, GAN, comprising a generator and a discriminator: using the generator of the GAN, determining (820) a second multi-layer graph representation based on the following: a candidate VAL group configuration determined based on a random input, and predicted QoS metrics for multi-layer connections between the UEs that comprise the reference VAL group configuration; using the discriminator of the GAN, determining (830) whether the following first conditions are met: the second multi-layer graph representation cannot be differentiated from the first multi-layer graph representation, and the predicted QoS metrics associated with the candidate VAL group configuration fulfil one or more service level agreements (SLAs) for operation of the communication network; repeating (850) the first operations based on a different random input when the first conditions are not met; and after the first conditions are met and using the generator, performing (860) inference based on the second multi-layer graph representation to generate a configuration of VAL groups for operation in the communication network.

2. The method of claim 1 , wherein the configuration of VAL groups generated by performing inference includes the following: a plurality of UEs arranged in multiple VAL groups, and connections between respective pairs of the UEs.

3. The method of claim 2, wherein one of the following applies: the configuration of VAL groups generated by performing inference also includes QoS settings for the connections between respective pairs of the UEs; or the method further comprises obtaining (870), from a network data analytics function, NWDAF, of the communication network, predicted QoS settings for the connections between respective pairs of the UEs.

4. The method of any of claims 2-3, wherein the connections between respective pairs of the UEs including the following: connections between pairs of UEs in a same VAL group, and connections between pairs of UEs in different VAL groups.

5. The method of any of claims 1-4, wherein the reference VAL group configuration and each candidate VAL group configuration includes the following: a plurality of UEs arranged in multiple VAL groups, and connections on multiple protocol layers between respective pairs of the UEs.

6. The method of claim 5, wherein: the multiple VAL groups are associated with respective applications or services; and each of the historical QoS metrics and the predicted QoS metrics is related to data traffic for one of the applications or services.

7. The method of any of claims 5-6, wherein each of the first and second multi-layer graph representations comprises: a plurality of first layers, with each first layer representing connections on a corresponding single protocol layer between respective pairs of the UEs; and a second layer representing connections across multiple protocol layers between respective pairs of the UEs.

8. The method of any of claims 5-7, wherein determining (810) the first multi-layer graph representation comprises: determining (811) a graph structure based on the reference VAL group configuration, wherein the graph structure includes the plurality of UEs as nodes and the connections between respective pairs of the UEs as edges; and determining (812) attributes for the edges based on the historical QoS metrics.

9. The method of claim 8, wherein determining (823) attributes for the edges based on the historical QoS metrics comprises, for each connection between a pair of UEs: obtaining the one or more historical QoS metrics for the connection from a network data analytics function, NWDAF, of the communication network; and determining, as the attribute for the edge corresponding to the connection, an integer priority value based on the one or more historical QoS metrics.

10. The method of claim 9, wherein each integer priority value is determined based on the following: a 5G QoS Identifier, 5QI, when included in the one or more historical QoS metrics; or at least one of the following historical QoS metrics, when a 5QI is not included: uplink, UL, throughput for the application or service as a percentage of total UL throughput in an area of interest; downlink, DL, throughput for the application or service as a percentage of total DL throughput in an area of interest; average traffic rate observed for UEs communicating with the application or service; maximum traffic rate observed for UEs communicating with the application or service; minimum traffic rate observed for UEs communicating with the application or service; average packet delay observed for UEs communicating with the application or service; maximum packet delay observed for UEs communicating with the application or service; average packet loss rate observed for UEs communicating with the application or service; and maximum packet loss rate observed for UEs communicating with the application or service.

11. The method of any of claims 5-10, wherein the generator and the discriminator include respective multi-layer graph neural networks, mGNNs.

12. The method of claim 11, further comprising, when it is determined that the second multi-layer graph representation can be differentiated from the first multi-layer graphrepresentation, adapting (840) the mGNNs of the generator and the discriminator based on a metric representative of differences between the first and second multi-layer graph representations, wherein the adapted mGNNs of the generator and the discriminator are used to repeat the first operations.

13. The method of claim 12, wherein the adapted mGNN of the generator is used to perform inference after the first conditions are met.

14. The method of any of claims 11-13, wherein determining (820) the second multi-layer graph representation comprises: determining (821) the candidate VAL group configuration based on applying the generator mGNN to the random input; determining (822) a graph structure based on the candidate VAL group configuration, wherein the graph structure includes the plurality of UEs as nodes and the connections between respective pairs of the UEs as edges; and determining (823) attributes for the edges based on the historical QoS metrics.

15. The method of any of claims 11-13, wherein determining (820) the second multi-layer graph representation comprises, based on applying the generator mGNN to the random input, determining (824) an attributed graph structure for a candidate VAL group configuration, wherein the attributed graph structure includes the plurality of UEs as nodes, the connections among the plurality of UEs as edges, and respective QoS metrics as attributes for the edges.

16. The method of any of claims 1-15, wherein the method is performed by one of the following: a network data analytics function, NWDAF, of the communication network; or a Service Enabler Architecture Layer, SEAL, group management, GM, server external to the communication network.

17. The method of claim 16, wherein the method is performed by the NWDAF, and the inference is further based on a desired architecture for VAL groups that is obtained from the SEAL GM server.

18. A network node or function, NNF (210, 320, 400, 430, 908, 916, 1000, 1100, 1202) configured to determine vertical application layer, VAL, groups for operation in a communication network (100, 200, 498, 902), each VAL group including a plurality of userequipment, UEs (220, 310, 912), wherein the NNF comprises processing circuitry (1002, 1102, 1204) configured to: determine a first multi-layer graph representation based on the following: a reference VAL group configuration previously used in the communication network, and historical quality of service, QoS, metrics for multi-layer connections between UEs that comprise the reference VAL group configuration; perform the following first operations using a generative adversarial network, GAN, comprising a generator and a discriminator: using the generator, determine a second multi-layer graph representation based on the following: a candidate VAL group configuration determined based on a random input, and predicted QoS metrics for multi-layer connections between the UEs that comprise the reference VAL group configuration; using the discriminator, determine whether the following first conditions are met: the second multi-layer graph representation cannot be differentiated from the first multi-layer graph representation, and the predicted QoS metrics associated with the candidate VAL group configuration fulfil one or more service level agreements, SLAs, for operation of the communication network; repeat the first operations based on a different random input when the first conditions are not met; and after the first conditions are met and using the generator, perform inference based on the second multi-layer graph representation to generate a configuration of VAL groups for operation in the communication network.

19. The NNF of claim 18, wherein the NNF is a network data analytics function, NWDAF (400) of the communication network and further comprises communication interface circuitry (1006, 1108, 1204) arranged to communicate with a Service Enabler Architecture Layer, SEAL, group management, GM, server (430) external to the communication network.

20. The NNF of any of claims 18-19, wherein the processing circuitry is further configured to perform operations corresponding to any of the methods of 2-17.

21. A network node or function, NNF (210, 320, 400, 430, 908, 916, 1000, 1100, 1202) configured to determine vertical application layer, VAL, groups for operation in a communication network (100, 200, 498, 902), each VAL group including a plurality of user equipment, UEs (220, 310, 912), wherein the NNF is further configured to: determine a first multi-layer graph representation based on the following: a reference VAL group configuration previously used in the communication network, and historical quality of service, QoS, metrics for multi-layer connections between UEs that comprise the reference VAL group configuration; perform the following first operations using a generative adversarial network, GAN, comprising a generator and a discriminator: using the generator, determine a second multi-layer graph representation based on the following: a candidate VAL group configuration determined based on a random input, and predicted QoS metrics for multi-layer connections between the UEs that comprise the reference VAL group configuration; using the discriminator, determine whether the following first conditions are met: the second multi-layer graph representation cannot be differentiated from the first multi-layer graph representation, and the predicted QoS metrics associated with the candidate VAL group configuration fulfil one or more service level agreements, SLAs, for operation of the communication network; repeat the first operations based on a different random input when the first conditions are not met; and after the first conditions are met and using the generator, perform inference based on the second multi-layer graph representation to generate a configuration of VAL groups for operation in the communication network.

22. The NNF of claim 21, being further configured to perform operations corresponding to any of the methods of 2-17.

23. A non-transitory, computer-readable medium (1004, 1204) storing computer-executable instructions that, when executed by processing circuitry (1002, 1102, 1204) associated with network node or function, NNF (210, 320, 400, 430, 908, 916, 1000, 1100, 1202) configured to determine vertical application layer, VAL, groups for operation in a communication network (100, 200, 498, 902), each VAL group including a plurality of user equipment, UEs (220, 310, 912), configure the NNF to perform operations corresponding to any of the methods of claims 1- 17.

24. A computer program product (1004a, 1204a) comprising computer-executable instructions that, when executed by processing circuitry (1002, 1102, 1204) associated with network node or function, NNF (210, 320, 400, 430, 908, 916, 1000, 1100, 1202) configured to determine vertical application layer, VAL, groups for operation in a communication network (100, 200, 498, 902), each VAL group including a plurality of user equipment, UEs (220, 310, 912), configure the NNF to perform operations corresponding to any of the methods of claims 1- 17.

Images