WO2023016653A1

WO2023016653A1 - Method, apparatus, and computer program

Info

Publication number: WO2023016653A1
Application number: PCT/EP2021/072571
Authority: WO
Inventors: Borislava GAJIC; Dario BEGA
Original assignee: Nokia Solutions And Networks Oy
Priority date: 2021-08-13
Filing date: 2021-08-13
Publication date: 2023-02-16
Also published as: EP4384955A1

Abstract

An apparatus comprises means configured to: determine a plan for one or more packet data sessions which are to be established or modified for obtaining data to be used for machine learning; and provide information to a network function to trigger the establishing or modifying of the one or more packet data sessions for obtaining data for the machine learning.

Description

Title

Method, apparatus, and computer program

Field

The present application relates to a method, apparatus, and computer program and in particular but not exclusively a method, apparatus, and computer program relating to obtaining of data for machine learning.

Background

A communication system can be seen as a facility that enables communication sessions between two or more entities such as communication devices, base stations and/or other nodes by providing carriers between the various entities involved in the communications path.

The communication system may be a wireless communication system. Examples of wireless systems comprise public land mobile networks (PLMN) operating based on radio standards such as those provided by 3GPP, satellite based communication systems and different wireless local networks, for example wireless local area networks (WLAN). The wireless systems can typically be divided into cells, and are therefore often referred to as cellular systems.

The communication system and associated devices typically operate in accordance with a given standard or specification which sets out what the various entities associated with the system are permitted to do and how that should be achieved. Communication protocols and/or parameters which shall be used for the connection are also typically defined.

Summary According to an aspect, there is provided an apparatus comprising means configured to: determine a plan for one or more packet data sessions which are to be established or modified for obtaining data to be used for machine learning; and provide information to a network function to trigger the establishing or modifying of the one or more packet data sessions for obtaining data for the machine learning

The means may be configured to provide as part of the plan one or more of: one or more communications devices associated with the one or more packet data sessions; one or more characteristics of one or more of the packet data sessions; a duration for a training/exploration phase for obtaining the data for the machine learning; and a number of packet data sessions.

The means may be configured to determine the plan taking into account one or more of: one or more conditions; and one or more policies.

The one or more conditions and/or one or more policies may be associated with a network in which the one or more packet data sessions which are to be established or modified.

One or more conditions and/or one or more policies comprise one or more of a quality of service requirement, a quality of experience requirement, a network condition, a load threshold, and an interference level.

The one or more packet data sessions which are to be established or modified for obtaining data for the machine learning may be redundant packet data sessions.

The means may be configured to provide information about the plan to a policy control function.

The means may be configured to receive a response from the policy control function in response to the information about the plan and in response providing information to the network function to trigger the establishing or modifying of the one or more packet data sessions for obtaining data for the machine learning. The means may be configured to collect data associated with the one or more of the one or more data packet data sessions which have been established or modified.

The means may be configured to train a machine learning model using said collected data.

The means may be configured to train a reinforcement machine learning model in an exploration phase using said collected data.

The means may be configured to output the trained machine learning model.

The means may be configured to output inferred data.

The apparatus may be provided in a machine learning control function or may be a machine learning control function.

According to another aspect, there is provided an apparatus, the apparatus comprising at least one processor and at least one memory including a computer program code, the at least one memory and computer program code configured to, with the at least one processor, cause the apparatus at least to: determine a plan for one or more packet data sessions which are to be established or modified for obtaining data to be used for machine learning; and provide information to a network function to trigger the establishing or modifying of the one or more packet data sessions for obtaining data for the machine learning

The at least one memory and at least one processor may be configured to cause the apparatus to provide as part of the plan one or more of: one or more communications devices associated with the one or more packet data sessions; one or more characteristics of one or more of the packet data sessions; a duration for a training/exploration phase for obtaining the data for the machine learning; and a number of packet data sessions. The at least one memory and at least one processor may be configured to cause the apparatus to determine the plan taking into account one or more of: one or more conditions; and one or more policies.

The at least one memory and at least one processor may be configured to cause the apparatus to provide information about the plan to a policy control function.

The at least one memory and at least one processor may be configured to cause the apparatus to receive a response from the policy control function in response to the information about the plan and in response providing information to the network function to trigger the establishing or modifying of the one or more packet data sessions for obtaining data for the machine learning.

The at least one memory and at least one processor may be configured to cause the apparatus to collect data associated with the one or more of the one or more data packet data sessions which have been established or modified.

The at least one memory and at least one processor may be configured to cause the apparatus to train a machine learning model using said collected data. The at least one memory and at least one processor may be configured to cause the apparatus to train a reinforcement machine learning model in an exploration phase using said collected data.

The at least one memory and at least one processor may be configured to cause the apparatus to output the trained machine learning model.

The at least one memory and at least one processor may be configured to cause the apparatus to be configured to output inferred data.

The apparatus may be provided in network function or may be a network function.

According to an aspect, there is provided a method comprising: determining a plan for one or more packet data sessions which are to be established or modified for obtaining data to be used for machine learning; and providing information to a network function to trigger the establishing or modifying of the one or more packet data sessions for obtaining data for the machine learning

The at least one memory and at least one processor may be configured to cause the apparatus to provide as part of the plan one or more of: one or more communications devices associated with the one or more packet data sessions; one or more characteristics of one or more of the packet data sessions; a duration for a training/exploration phase for obtaining the data for the machine learning; and a number of packet data sessions.

The method may comprise determining the plan taking into account one or more of: one or more conditions; and one or more policies.

The one or more conditions and/or one or more policies may be associated with a network in which the one or more packet data sessions which are to be established or modified. One or more conditions and/or one or more policies comprise one or more of a quality of service requirement, a quality of experience requirement, a network condition, a load threshold, and an interference level.

The method may comprise providing information about the plan to a policy control function.

The method may comprise receiving a response from the policy control function in response to the information about the plan and in response providing information to the network function to trigger the establishing or modifying of the one or more packet data sessions for obtaining data for the machine learning.

The method may comprise collecting data associated with the one or more of the one or more data packet data sessions which have been established or modified.

The method may comprise training a machine learning model using said collected data.

The method may comprise training a reinforcement machine learning model in an exploration phase using said collected data.

The method may comprise outputting the trained machine learning model.

The method may comprise causing outputting inferred data.

The method may be performed by an apparatus. The apparatus may be provided in network function or may be a network function.

According to another aspect, there is provided an apparatus comprising means configured to: receive information at a policy control function from a machine learning control function about a plan for one or more packet data sessions which are to be established or modified for obtaining data to be used for machine learning; determine if the plan is implementable; and provide a response to the machine learning control function if the plan is implementable.

The information about the plan may comprise information about one or more of: one or more communications devices associated with the one or more packet data sessions; one or more characteristics of one or more of the packet data sessions; a duration for a training/exploration phase for obtaining the data for the machine learning; and a number of the packet data sessions.

The means may be configured to determine the plan if the plan is implementable taking into account one or more of: subscription profile information associated with one or more of the communication devices associated with the one or more packet data sessions; one or more network operator policies; and configuration of one or more network entities.

The means may be configured to update one or more policies and/or one or more rules in response to the received information.

The means may be configured to update communications device route selection policy at one or more communications devices associated with the one or more packet data sessions in response to said received information.

The apparatus may be provided in a policy control function or may be a policy control function.

According to another aspect, there is provided an apparatus, the apparatus comprising at least one processor and at least one memory including a computer program code, the at least one memory and computer program code configured to, with the at least one processor, cause the apparatus at least to: receive information at a policy control function from a machine learning control function about a plan for one or more packet data sessions which are to be established or modified for obtaining data to be used for machine learning; determine if the plan is implementable; and provide a response to the machine learning control function if the plan is implementable.

The at least one memory and at least one processor may be configured to determine the plan if the plan is implementable taking into account one or more of: subscription profile information associated with one or more of the communication devices associated with the one or more packet data sessions; one or more network operator policies; and configuration of one or more network entities.

The at least one memory and at least one processor may be configured to update one or more policies and/or one or more rules in response to the received information.

The at least one memory and at least one processor may be configured to update communications device route selection policy at one or more communications devices associated with the one or more packet data sessions in response to said received information.

The apparatus may be provided in a policy control function or may be a policy control function. According to another aspect, there is provided method comprising: receiving information at a policy control function from a machine learning control function about a plan for one or more packet data sessions which are to be established or modified for obtaining data to be used for machine learning; determining if the plan is implementable; and providing a response to the machine learning control function if the plan is implementable.

The method may comprise determining the plan if the plan is implementable taking into account one or more of: subscription profile information associated with one or more of the communication devices associated with the one or more packet data sessions; one or more network operator policies; and configuration of one or more network entities.

The method may comprise updating one or more policies and/or one or more rules in response to the received information.

The method may comprise updating communications device route selection policy at one or more communications devices associated with the one or more packet data sessions in response to said received information.

The method may be performed by an apparatus. The apparatus may be provided in a policy control function or may be a policy control function.

According to another aspect, there is provided an apparatus comprising means configured to: receive information at an application function from a machine learning control function, about one or more packet data sessions which are to be established or modified for obtaining data to be used for machine learning; and trigger the one or more packet data sessions to be respectively established or modified.

The means may be configured to cause the one or more packet data sessions to be respectively established or modified by causing a respective message to be provided to a respective communications device associated with a respective one of the one or more packet data sessions to be established or modified.

The apparatus may be provided in an application function or may be an application function.

According to another aspect, there is provided an apparatus, the apparatus comprising at least one processor and at least one memory including a computer program code, the at least one memory and computer program code configured to, with the at least one processor, cause the apparatus at least to: receive information at an application function from a machine learning control function, about one or more packet data sessions which are to be established or modified for obtaining data to be used for machine learning; and trigger the one or more packet data sessions to be respectively established or modified.

The at least one memory and at least one processor may be configured to cause the one or more packet data sessions to be respectively established or modified by causing a respective message to be provided to a respective communications device associated with a respective one of the one or more packet data sessions to be established or modified.

The one or more packet data sessions which are to be established or modified for obtaining data for the machine learning may be redundant packet data sessions. The apparatus may be provided in an application function or may be an application function.

According to another aspect, there is provided a method comprising: receiving information at an application function from a machine learning control function, about one or more packet data sessions which are to be established or modified for obtaining data to be used for machine learning; and triggering the one or more packet data sessions to be respectively established or modified.

The method may comprise causing the one or more packet data sessions to be respectively established or modified by causing a respective message to be provided to a respective communications device associated with a respective one of the one or more packet data sessions to be established or modified.

The method may be performed by an apparatus. The apparatus may be provided in an application function or may be an application function.

According an aspect, there is provided an apparatus comprising means configured to: receive a trigger from a network function for a packet data session which is to be established or modified for obtaining data to be used for machine learning; and initiate establishment or modification of a first packet data session and a duplicate of that packet data session, the duplicate packet data session being the packet data session for obtaining the data to be used for machine learning, the first and duplicate packet data session being associated with different identity information.

The apparatus may be provided in a communications device or may be a communications device.

According to another aspect, there is provided an apparatus, the apparatus comprising at least one processor and at least one memory including a computer program code, the at least one memory and computer program code configured to, with the at least one processor, cause the apparatus at least to: receive a trigger from a network function for a packet data session which is to be established or modified for obtaining data to be used for machine learning; and initiate establishment or modification of a first packet data session and a duplicate of that packet data session, the duplicate packet data session being the packet data session for obtaining the data to be used for machine learning, the first and duplicate packet data session being associated with different identity information.

According an aspect, there is provided a method comprising: receiving a trigger from a network function for a packet data session which is to be established or modified for obtaining data to be used for machine learning; and initiating establishment or modification of a first packet data session and a duplicate of that packet data session, the duplicate packet data session being the packet data session for obtaining the data to be used for machine learning, the first and duplicate packet data session being associated with different identity information.

The method may be performed by an apparatus. The apparatus may be provided in a communications device or may be a communications device.

According to another aspect, there is provided an apparatus comprising means configured to: determine that a packet data session which is to be established or modified is to be used for obtaining data for machine learning; and in response provide an indication for the packet data session which is to be established or modified for obtaining data for machine learning.

The indication may comprise one or more of a redundancy sequence number value and an indication that the packet data session is for obtaining machine learning data.. The indication may differentiate the packet data session which is to be established or modified for obtaining data for machine learning from a duplicate packet data session.

The means may be configured to provide the indication to an access node associated with a communications device associated with the packet data session.

The means may be configured to assign a priority to the packet data session established or modified to provide data for machine learning, the priority assigned being based on the packet data session being for providing data for machine learning.

The apparatus may be provided in a session management function or may be a session management function.

According to another aspect, there is provided an apparatus, the apparatus comprising at least one processor and at least one memory including a computer program code, the at least one memory and computer program code configured to, with the at least one processor, cause the apparatus at least to: determine that a packet data session which is to be established or modified is to be used for obtaining data for machine learning; and in response provide an indication for the packet data session which is to be established or modified for obtaining data for machine learning.

The indication may comprise one or more of a redundancy sequence number value and an indication that the packet data session is for obtaining machine learning data..

The indication may differentiate the packet data session which is to be established or modified for obtaining data for machine learning from a duplicate packet data session.

The at least one memory and at least one processor may be configured to provide the indication to an access node associated with a communications device associated with the packet data session.

The at least one memory and at least one processor may be configured to assign a priority to the packet data session established or modified to provide data for machine learning, the priority assigned being based on the packet data session being for providing data for machine learning.

According to another aspect, there is provided a method comprising: determining that a packet data session which is to be established or modified is to be used for obtaining data for machine learning; and in response providing an indication for the packet data session which is to be established or modified for obtaining data for machine learning.

The method may comprise providing the indication to an access node associated with a communications device associated with the packet data session.

The method may comprise assigning a priority to the packet data session established or modified to provide data for machine learning, the priority assigned being based on the packet data session being for providing data for machine learning.

The method may be performed by an apparatus. The apparatus may be provided in a session management function or may be a session management function.

According to an aspect, there is provided an apparatus comprising means configured to: receive from a session management function an indication, the indication indicating for a packet data session which is to be established or modified that the packet data session is to be established or modified to provide data for machine learning. The means may be configured to assign a priority to the packet data session established or modified to provide data for machine learning, the priority assigned being based on the packet data session being for providing data for machine learning to the packet data session established or modified to provide data for machine learning.

The apparatus may be provided in an access node or may be an access node.

According to another aspect, there is provided an apparatus, the apparatus comprising at least one processor and at least one memory including a computer program code, the at least one memory and computer program code configured to, with the at least one processor, cause the apparatus at least to: receive from a session management function an indication, the indication indicating for a packet data session which is to be established or modified that the packet data session is to be established or modified to provide data for machine learning.

The apparatus may be provided in an access node or may be an access node.

According to an aspect, there is provided a method comprising: receiving from a session management function an indication, the indication indicating for a packet data session which is to be established or modified that the packet data session is to be established or modified to provide data for machine learning.

The means may be configured to assign a priority to the packet data session established or modified to provide data for machine learning, the priority assigned being based on the packet data session being for providing data for machine learning. The method may be performed by an apparatus. The apparatus may be provided in an access node or may be an access node.

According to another aspect, there is provided a computer readable medium comprising program instructions for causing an apparatus to perform at least the method according to any of the preceding aspects.

According to an aspect, there is provided a non-transitory computer readable medium comprising program instructions for causing an apparatus to perform at least the method according to any of the preceding aspects.

In the above, many different embodiments have been described. It should be appreciated that further embodiments may be provided by the combination of any two or more of the embodiments described above.

Description of Figures

Embodiments will now be described, by way of example only, with reference to the accompanying Figures in which:

Figure 1 shows an example system architecture;

Figure 2 shows a schematic diagram of an example 5G system;

Figure 3 shows a schematic diagram of an example apparatus;

Figures 4a and b shows a signal flow where PDU (packet data unit) sessions are used for exploration purpose;

Figure 5 shows an example system in which some embodiments may be provided: Figures 6 to 11 show flowcharts of methods performed by an apparatus according to various example embodiments.

Detailed description

In the following, different exemplifying embodiments will be described using, as an example of an access architecture to which the embodiments may be applied, a radio access architecture based on long term evolution advanced (LTE Advanced, LTE-A) or new radio (NR, 5G), without restricting the embodiments to such an architecture, however. The embodiments may also be applied to other kinds of communications networks having suitable means by adjusting parameters and procedures appropriately. Some examples of other options for suitable systems are the universal mobile telecommunications system (UMTS) radio access network (UTRAN), wireless local area network (WLAN or Wi-Fi), worldwide interoperability for microwave access (WiMAX), Bluetooth®, personal communications services (PCS), ZigBee®, wideband code division multiple access (WCDMA), systems using ultra-wideband (UWB) technology, sensor networks, mobile ad-hoc networks (MANETs) and Internet Protocol multimedia subsystems (IMS) or any combination thereof.

Figure 1 depicts examples of simplified system architectures only showing some elements and functional entities, all being logical units, whose implementation may differ from what is shown. The connections shown in Figure 1 are logical connections; the actual physical connections may be different. It is apparent to a person skilled in the art that the system typically comprises also other functions and structures than those shown in Figure 1 .

The embodiments are not, however, restricted to the system given as an example but a person skilled in the art may apply the solution to other communication systems provided with necessary properties.

The example of Figure 1 shows a part of an exemplifying radio access network.

Figure 1 shows devices 100 and 102. The devices 100 and 102 are configured to be in a wireless connection on one or more communication channels with a node 104. The node 104 is further connected to a core network 106. In one example, the node 104 may be an access node such as (eZg)NodeB serving devices in a cell. In one example, the node 104 may be a non-3GPP access node. The physical link from a device to a (eZg)NodeB is called uplink or reverse link and the physical link from the (eZg)NodeB to the device is called downlink or forward link. It should be appreciated that (eZg)NodeBs or their functionalities may be implemented by using any node, host, server, or access point etc. entity suitable for such a usage. A communications system typically comprises more than one (eZg)NodeB in which case the (eZg)NodeBs may also be configured to communicate with one another over links, wired or wireless, designed for the purpose. These links may be used for signalling purposes. The (eZg)NodeB is a computing device configured to control the radio resources of communication system it is coupled to. The NodeB may also be referred to as a base station, an access point or any other type of interfacing device including a relay station capable of operating in a wireless environment. The (eZg)NodeB includes or is coupled to transceivers. From the transceivers of the (eZg)NodeB, a connection is provided to an antenna unit that establishes bi-directional radio links to devices. The antenna unit may comprise a plurality of antennas or antenna elements. The (eZg)NodeB is further connected to the core network 106 (CN or next generation core NGC). Depending on the deployed technology, the (eZg)NodeB is connected to a serving and packet data network gateway (S-GW and P-GW) or user plane function (UPF), for routing and forwarding user data packets and for providing connectivity of devices to one or more external packet data networks, and to a mobile management entity (MME) or access mobility management function (AMF), for controlling access and mobility of the devices.

Exemplary embodiments of a device are a subscriber unit, a user device, a user equipment (UE), a user terminal, a terminal device, a mobile station, a mobile device, etc

The device typically refers to a mobile or static device ( e.g. a portable or non-portable computing device) that includes wireless mobile communication devices operating with or without an universal subscriber identification module (IISIM), including, but not limited to, the following types of devices: mobile phone, smartphone, personal digital assistant (PDA), handset, device using a wireless modem (alarm or measurement device, etc.), laptop andZor touch screen computer, tablet, game console, notebook, a wireless interface card or other wireless interface facility (e.g., USB dongle) and multimedia device. It should be appreciated that a device may also be a nearly exclusive uplink only device, of which an example is a camera or video camera loading images or video clips to a network. The device may be a machine-type communications (MTC) device, an Internet of things (loT) type communication device. The device may be a device having capability to operate in Internet of Things (loT) network which is a scenario in which objects are provided with the ability to transfer data over a network without requiring human-to-human or human-to-computer interaction, e.g. to be used in smart power grids and connected vehicles. The device may also utilise cloud. In some applications, a device may comprise a user portable device with radio parts (such as a watch, earphones, or eyeglasses) and the computation is carried out in the cloud.

The device illustrates one type of an apparatus to which resources on the air interface are allocated and assigned, and thus any feature described herein with a device may be implemented with a corresponding apparatus, such as a relay node. An example of such a relay node is a layer 3 relay (self-backhauling relay) towards the base station. The device (or in some embodiments a layer 3 relay node) is configured to perform one or more of user equipment functionalities.

Various techniques described herein may also be applied to a cyber-physical system (CPS) (a system of collaborating computational elements controlling physical entities). CPS may enable the implementation and exploitation of massive amounts of interconnected information and communications technology, ICT, devices (sensors, actuators, processors microcontrollers, etc.) embedded in physical objects at different locations. Mobile cyber physical systems, in which the physical system in question has inherent mobility, are a subcategory of cyber-physical systems. Examples of mobile physical systems include mobile robotics and electronics transported by humans or animals.

Additionally, although the apparatuses have been depicted as single entities, different units, processors and/or memory units (not all shown in Figure 1 ) may be implemented.

5G enables using multiple input - multiple output (MIMO) antennas, many more base stations or nodes than the LTE (a so-called small cell concept), including macro sites operating in co-operation with smaller stations and employing a variety of radio technologies depending on service needs, use cases and/or spectrum available. 5G mobile communications supports a wide range of use cases and related applications including video streaming, augmented reality, different ways of data sharing and various forms of machine type applications (such as (massive) machine-type communications (mMTC), including vehicular safety, different sensors, and real-time control). 5G is expected to have multiple radio interfaces, e.g. below 6GHz or above 24 GHz, cm Wave and mmWave, and also being integrable with existing legacy radio access technologies, such as the LTE. Integration with the LTE may be implemented, at least in the early phase, as a system, where macro coverage is provided by the LTE and 5G radio interface access comes from small cells by aggregation to the LTE. In other words, 5G is planned to support both inter-RAT operability (such as LTE-5G) and inter-RI operability (inter-radio interface operability, such as below 6GHz - cm Wave, 6 or above 24 GHz - cm Wave and mmWave).

One of the concepts considered to be used in 5G networks is network slicing in which multiple independent and dedicated virtual sub-networks (network instances) may be created within the same infrastructure to run services that have different requirements on latency, reliability, throughput, and mobility.

The current architecture in LTE networks is fully distributed in the radio and fully centralized in the core network. The low latency applications and services in 5G require to bring the content close to the radio which leads to local break out and multiaccess edge computing (MEC). 5G enables analytics and knowledge generation to occur at the source of the data. This approach requires leveraging resources that may not be continuously connected to a network such as laptops, smartphones, tablets, and sensors. MEC provides a distributed computing environment for application and service hosting. It also has the ability to store and process content in close proximity to cellular subscribers for faster response time. Edge computing covers a wide range of technologies such as wireless sensor networks, mobile data acquisition, mobile signature analysis, cooperative distributed peer-to-peer ad hoc networking and processing also classifiable as local cloud/fog computing and grid/mesh computing, dew computing, mobile edge computing, cloudlet, distributed data storage and retrieval, autonomic self-healing networks, remote cloud services, augmented and virtual reality, data caching, Internet of Things (massive connectivity and/or latency critical), critical communications (autonomous vehicles, traffic safety, real-time analytics, time-critical control, healthcare applications).

The communication system is also able to communicate with other networks 112, such as a public switched telephone network, or a VoIP network, or the Internet, or a private network, or utilize services provided by them. The communication network may also be able to support the usage of cloud services, for example at least part of core network operations may be carried out as a cloud service (this is depicted in Figure 1 by “cloud” 114). The communication system may also comprise a central control entity, or a like, providing facilities for networks of different operators to cooperate for example in spectrum sharing.

The technology of Edge cloud may be brought into a radio access network (RAN) by utilizing network function virtualization (NFV) and software defined networking (SDN). Using the technology of edge cloud may mean access node operations to be carried out, at least partly, in a server, host or node operationally coupled to a remote radio head or base station comprising radio parts. It is also possible that node operations will be distributed among a plurality of servers, nodes, or hosts. Application of cloud RAN architecture enables RAN real time functions being carried out at or close to a remote antenna site (in a distributed unit, DU 108) and non-real time functions being carried out in a centralized manner (in a centralized unit, CU 110).

It should also be understood that the distribution of labour between core network operations and base station operations may differ from that of the LTE or even be nonexistent. Some other technology advancements probably to be used are Big Data and all-IP, which may change the way networks are being constructed and managed. 5G (or new radio, NR) networks are being designed to support multiple hierarchies, where MEC servers can be placed between the core and the base station or NodeB (gNB). It should be appreciated that MEC can be applied in 4G networks as well.

5G may also utilize satellite communication to enhance or complement the coverage of 5G service, for example by providing backhauling. Possible use cases are providing service continuity for machine-to-machine (M2M) or Internet of Things (loT) devices or for passengers on board of vehicles, Mobile Broadband, (MBB) or ensuring service availability for critical communications, and future railway/maritime/aeronautical communications. Satellite communication may utilise geostationary earth orbit (GEO) satellite systems, but also low earth orbit (LEO) satellite systems, in particular megaconstellations (systems in which hundreds of (nano)satellites are deployed). Each satellite in the mega-constellation may cover several satellite-enabled network entities that create on-ground cells. The on-ground cells may be created through an on-ground relay node or by a gNB located on-ground or in a satellite.

It is obvious for a person skilled in the art that the depicted system is only an example of a part of a radio access system and in practice, the system may comprise a plurality of (eZg)NodeBs, the device may have access to a plurality of radio cells and the system may comprise also other apparatuses, such as physical layer relay nodes or other network elements, etc. At least one of the (eZg)NodeBs or may be a Home(eZg)NodeB. Additionally, in a geographical area of a radio communication system a plurality of different kinds of radio cells as well as a plurality of radio cells may be provided. Radio cells may be macro cells (or umbrella cells) which are large cells, usually having a diameter of up to tens of kilometres, or smaller cells such as micro-, femto- or picocells. The (eZg)NodeBs of Figure 1 may provide any kind of these cells. A cellular radio system may be implemented as a multilayer network including several kinds of cells. Typically, in multilayer networks, one access node provides one kind of a cell or cells, and thus a plurality of (eZg)NodeBs are required to provide such a network structure.

For fulfilling the need for improving the deployment and performance of communication systems, the concept of “plug-and-play” (eZg)NodeBs has been introduced. Typically, a network which is able to use “plug-and-play” (eZg)Node Bs, includes, in addition to Home (eZg)NodeBs (H(eZg)gNodeBs), a home node B gateway, or HNB-GW (not shown in Figure 1 ). A HNB Gateway (HNB-GW), which is typically installed within an operator’s network may aggregate traffic from a large number of HNBs back to a core network. In the following examples, the communications device is referred to as a UE. However, it should be appreciated that the communication device can any suitable communications device, some examples of which have already been mentioned.

AI/ML (Artificial Intelligence I Machine Learning) is currently studied in different standardization organizations, as ETSI (for example with ZSM (Zero touch network and service management), 3GPP and in different groups and contexts. Artificial Intelligence (Al) including machine learning (ML) algorithms may be considered as enablers for improving the network management and the user experience by providing insights based on autonomous analysis of collected network data.

In a 5G system, there are different ways to collect data for training AI/ML models used for networks optimization.

For example a network data analytics function (NWDAF) may be used. The NWDAF is responsible for providing network analytics information upon request from one or more network functions or apparatus within the network. Network functions can also subscribe to the NWDAF to receive information therefrom. The NWDAF is configured to receive and store network information from one or more network functions or apparatus within the network.

An NWDAF may comprise an analytics logical function (AnLF) and a model Training logical function (MTLF):

An NWDAF containing the analytics logical function, denoted as NWDAF(AnLF), can perform inference, derive analytics information (i.e. derives statistics and/or predictions based on an analytics consumer request) and expose analytics services. The services may be a subscription service or an analytics information service.

An NWDAF containing the model training logical function, denoted as NWDAF(MTLF), trains ML models and exposes new training services (e.g. providing trained model). The NWDAF may collect management data from the services in the OAM (operations, administration, and maintenance). Such data include data about 5G RAN or 5G core performance and/or data about 5G end to end KPIs (key performance indicators). In the case when the target of data collection is a specific UE, the data is collected, for example, by using MDT (minimization of driving test).

The collected data is used for training (in case of supervised/unsupervised learning). In the current specification, RL (Reinforcement Learning) is not considered for NWDAF. RL requires that the network state space be explored by taking random actions that may potentially violate service guarantees given by the system.

RL is generally regarded as being one example of ML.

AI/ML solutions (including the RL) appear to be promising in many use cases, e.g. mobility management where HO (handover) parameters need to be optimized. This may be in a dynamic environment due to changing mobility patterns varying over time. In one use case the RL agent(s) may optimize HO parameters based on the network state by learning the best policy for optimizing cell individual offset (CIOs), and by learning the changes in network mobility and reacting to those changes.

It should be appreciated that in the following references to ML may cover any suitable ML technique such as but not limited to RL. In some embodiments, the ML may be used to provide Al type behaviour. In other words, a model which is provided by the ML process may provide an output which is fed into a computational system which performs an action in an automated fashion. Some embodiments may be address the challenges is obtaining the data required to train the ML model.

One challenge in introducing the AI/ML features into a running mobile network is enabling effective training. In particular, for RL (reinforcement learning), the exploration phase could make the network unstable.

There may be a need to make the E2E (end to end) support of AI/ML features and the workflow transparent, at least in the first deployments in the network. In other words, as the first step of introducing the AI/ML features in mobile network, the AI/ML support should be facultative, i.e. the introduction of AI/ML features on selected network functions shall not hinder the normal network operation and the functionality of the remaining network functions. It is also desirable that the network entities which do not support AI/ML shall not be penalized.

Enabling the AI/ML may be gradual. Enabling the AI/ML may use existing network “tools” for realizing the AI/ML such as providing the appropriate training datasets and exploration opportunities without making network unstable.

Some embodiments may relate to PDU (packet data unit) sessions which are established/modified for ML (including RL) training purposes.

Different UEs may initiate the PDU sessions for AI/ML. The UEs, CN (core network), RAN can modify already established AI/ML PDU sessions. Without proper coordination of establishment/modifications of such PDU sessions the training and/or exploration phase might be inefficient (no proper network states learned) and/or may lead to network instabilities and/or poor network performance. Some embodiments aim to coordinate the AI/ML PDU sessions.

Some embodiments may handle one or more the following issues: which UE, under which conditions and with which parameters are allowed to establish/modify an AI/ML PDU session; admission control for AI/ML PDU sessions: such as can all AI/ML PDU sessions be established/modified as requested, e.g. are there enough resources to accept all of them for the purpose of ML; and avoid an impact of AI/ML PDU sessions on normal network operation, avoid potential congestion in the network, and/or the situation in which the “regular” PDU sessions cannot be admitted due to lack of resources.

Some embodiments may provide coordination of AI/ML PDU sessions. This may be to make use of AI/ML PDU sessions such that the action which improves network performance is learned without violating normal network operation and resource usage during the training and/or exploration phase. This coordination of AI/ML PDU sessions may be to create and modify the AI/ML PDU sessions based on defined ML targets keeping in mind the current network state.

Some embodiments may provide a PDU session coordination function to optimize the establishment, modification and/or management of PDU sessions for RL/DRL (deep reinforced learning) training.

Some embodiments may enable the establishment of PDU sessions alternatively or additionally for supervised/unsupervised ML algorithm training.

Figure 2 shows a schematic representation of a 5G system (5GS). The 5GS may be comprised by a terminal or user equipment (UE), a 5G radio access network (5GRAN) or next generation radio access network (NG-RAN), a 5G core network (5GC), one or more application functions (AF) (not shown) and one or more data networks (DN).

The 5G-RAN may comprise one or more base stations. In 5G the base station may be referred to as a gNodeB (gNB). The RAN may comprise one or more gNodeB (gNB) (or base station ) distributed unit functions connected to one or more gNodeB (gNB) (or base station) centralized unit functions.

The 5GC may comprise the following entities: one or more access management functions (AMF), one or more session management functions (SMF), an authentication server function (AUSF), a unified data management (UDM), one or more user plane functions (UPF), a service communication proxy (SCP), a policy control function (PCF), a network slice selection function (NSSF), a network slice specific authentication and authorization function (NSSAAF), a NWDAF, and/or a network exposure function (NEF).

Some embodiments provide a ML PDU session coordination function - ML PDU CF providing the ML PDU coordination service. This may be part of the architecture shown in Figure 2. The ML PDU CF is shown in Figure 2 as a standalone functionality. The ML PDU CF as a standalone functionality provides the ML PDU coordination service by consuming the data collection and analytics services from other NFs, e.g. from NWDAF. The NWDAF collects the data from various NFs, including RAN and CN, as well as from UE (by means of MDT). Thus, the NWDAF can collect the data relative to the PDU sessions established/modified for ML purposes. Furthermore, the NWDAF may include the MLTF (model training logical function) with training capabilities for supervised and unsupervised learning.

The examples below focused on RL, and the PDU session is referred to as RL PDU session. However it should be appreciated that analogous procedures are applicable to a supervised/unsupervised ML approach using dedicated PDU sessions.

The ML PDU CF may derive the plan for PDU sessions establishment/modification for the purpose of ML training/exploration. This may be in such a way that a best action is learned but potential issues during training and/or exploration phase may be avoided. Such a plan may be use-case specific and includes one or more of the following:

UEs to be involved in the training/exploration; number of PDU sessions to be requested for the purpose of training/exploration;

PDU session characteristics, e.g. 5Qls (5G QoS indicator) to be supported, always-on property, and/or the like; and expected duration of the training and/or exploration phase.

In order not to jeopardize the operation of running network, the ML PDU CF requests the setting up of redundant PDU sessions (whenever the resource conditions allow) for training and/or exploration purposes. A redundant PDU session may be realised by the duplication of data and resources.

The ML PDU CF takes into account the conditions/policies under which the training and/or exploration shall be performed as given by the network operator, e.g. QoS/QoE (quality of service/quality of experience) requirements that should not be violated, network conditions such as load thresholds, interference levels which shall not be exceeded and/or the like. Based on the decision to establish/modify the ML PDU session the input is sent to a consumer network function, i.e. the training assistant network function (TA NF) which will trigger the PDU session establishment/modification. The TA NF may be e.g. an AF, CN or AN (access network) network function and is shown in Figure 2.

The ML PDU CF collects the data relative to the established/modified ML PDU sessions

The ML PDU CF trains the ML (e.g., supervised/unsupervised or RL) model.

The ML PDU CF provides the trained model to one or more interested network functions. The ML PDU CF alternatively or additionally provides the inference output.

The ML PDU CF coordinates the PDU sessions for supervised/unsupervised ML algorithm training and/or for purpose of exploration of a RL algorithm.

The ML PDU CF may provide information related to planned training and/or exploration to the PCF for the potential update of one or more policies.

One or more of the following information may be provided to the PCF: use case, i.e. ML model/purpose for which the ML PDU session is used;

UEs to be involved in the training and/or exploration; number of PDU sessions to be requested for the purpose of training and/or exploration;

PDU session characteristics, e.g. 5Qls to be supported, always-on property, requirements for setting up the redundant PDU session and/or the like; and expected duration of the training and/or exploration phase.

The PCF may perform a feasibility check of such exploration based on subscription profile information (this may be provided by a UDR (unified data repository)), operator policies and/or local configuration (provided by the 0AM), The PCF may update PCC (policy and charging control) rules accordingly. Optionally, if the planned exploration is unacceptable based on one or more of available subscription profile information, operator policies and/or local configuration the PCF may provide such feedback to the ML PDU CF. As a result of this feedback, the ML PDU CF may update its training and/or exploration plan. The updated information may be sent to the PCF.

Based on the updated PCC rules at the PCF, the ML PDU session establishment/modification will take place.

The ML PDU CF decides which PDU sessions, on which UEs need to be established/modified based on the targeted use case and/or based on configured policies/rules/constraints regarding the impact to the rest of the network. The policies/rules/constraints may be configured by the network operator via, for example, the 0AM. For example, the network operator may define that the RL exploration may increase the load up to a given threshold. This may be without violating the QoS/QoE requirements of ongoing PDU sessions.

Based on the decision by the ML PDU CF, the corresponding trigger for PDU session establishment/modification is issued. This may be realized by a dedicated NF, for example, the training assistant AF.

For example, for a PDU session establishment, the training assistant AF (or other CN NF) is triggered based on input from ML PDU CF.

For example, for a PDU session modification, the training assistant AF (or other CN NF) is triggered based on input from ML PDU CF,

The dedicated TA AF can be associated with the ML PDU CF in order to execute the triggers for RL PDU session establishment/modification based on the input from ML PDU CF. Alternatively or additionally, any other suitable NF can be a consumer of this input from ML PDU CF.

The ML PDU CF may use the service provided by NWDAF (AnLF) to collect the PDU session relevant data, analyse it in order to train the model, and provide a trained model. Alternatively or additionally, an inference output to any NF (e.g. SMF) for a specific use case may be provided.

The PDU session establishment may be UE initiated but a network triggered procedure may be used in some embodiments. The training assistant AF consumes the service offered by the ML PDU CF in order to determine the need to trigger the UE to establish the RL PDU session. The TA AF can trigger the RL PDU session establishment based on input from the ML PDU CF.

In the case of network triggered PDU session establishment, the network may send a device trigger message or other suitable message to an application(s) on the UE side. The payload included in the device trigger request message contains information on which application on the UE side is expected to trigger the PDU Session establishment request. The PDU session ID is derived by the UE. For the case of a RL PDU session the UE may derive the PDU session ID based on a predefined rule in order to enable easier differentiation and associated data collection and processing. For example, the RL PDU session ID may be associated to the specific application ID which the TA AF used for triggering.

PDU session modification may be either UE or network requested. For example the PDU session modification may be triggered by one or more of the PCF, SMF, AMF, and AN.

The TA AF may trigger the RL PDU session modification based on ML PDU CF input.

The SMF may decide to modify the PDU Session. This procedure may alternatively or additionally be triggered based on a locally configured policy or triggered from the (R)AN. The SMF may also be a consumer of service from ML PDU CF. Other NFs such as the AMF and/or the PCF may consume the ML PDU CF service.

The ML PDU CF input based on which the ML PDU session establishment/modification is triggered may have input on network condition changes. This may be from the OAM. For example, if the network load exceeds a threshold allowed for RL exploration defined by the network operator, the ML PDU CF may reduce the number of RL PDU sessions and/or update the exploration plan.

The RL PDU session establishment/modification request may be treated based on the PCC rules at PCF. The initial PCC rules (without consideration of RL exploration) may be derived based on the subscription profile information, operator policies and/or local configuration.

In some embodiments, there may be updates of such PCC rules based on the exploration plan information as provided by the ML PDU CF to the PCF.

Figure 4a and b illustrates the signalling where the RL PDU sessions used for exploration are realized using redundant resources in order not to influence the operational network. The example in Figure 4a and b illustrates the traffic steering use case. The traffic steering one example of use cases for which the ML support may be highly beneficial. Other embodiments may be used with other use cases. Other examples of use cases are mobility management optimization (optimization of handover parameters) and mobility load balancing (balancing the load between the cells by handover of UEs on cell borders to less loaded cells) and/or the like.

As referenced 1 , the ML PDU CF provides RL exploration plan details to the PCF. The plan details may include information such as which UEs will be involved and the PDU session requirements.

In some embodiments the exploration plan may comprise one or more of the following: use case: for example traffic steering between different technologies (standards) and/or different cells. This may be in a specific area. This may be in order to optimize the QoS (quality of service). The different cells may be one or more of macro, small, pico, and femto) in specific area. The different technologies may be one or more of LTE, LTE-A, GSM, and 5G. This is one example of a use case and different use cases may alternatively or additionally be supported:

UEs to be involved in the training and/or exploration (For example this may be one or more UEs present in specific area which support the one or more targeted technologies such as discussed above; number of PDU sessions to be requested for the purpose of training and/or exploration based on the required level of data to be collected; in order not to influence the operational network performance the ML PDU CF requests the establishment of redundant PDU sessions for exploration purposes; and expected duration of the training and/or exploration phase, etc.

The exploration plan may take into account one or more conditions and/or policies under which the training and/or exploration shall be performed. These one or more conditions and/or policies may be defined, for example by the network operator. For example, there may be QoS and/or QoE requirements that shall not be violated, network conditions such as load thresholds, interference levels which shall not be exceeded, and/or the like.

As referenced 2, the PCF verifies the feasibility of the planned RL exploration, The PCF may update the PCC rules, accordingly, to support the planned RL exploration. For example, this may allow the establishment of the redundant PDU sessions for the selected UEs.

As referenced 3, the PCF may optionally provide feedback to ML PDU CF indicating the acceptance of the RL exploration plan. This may, for example, be way of an acknowledgement.

As referenced 4, the PCF updates the URSP (UE route selection policy) rules at the selected UEs based on the updated policies and/or rules. As referenced 5, the ML PDU CF provides input to the TA AF on the need to trigger the establishment/modification of the RL PDU session.

As referenced 6, the TA AF triggers the RL PDU session establishment/modification process. This may be in line with the procedures of the corresponding standard. By way of example only, this may be accord with the procedures set out in 3GPP TS 23.502 Sections 4.3.2 and 4.3.3 or any other suitable standard.

As referenced 7, since the establishment of redundant PDU sessions has been requested, the UE initiates two PDU Sessions (one is redundant) with the SMF via the RAN. Different combinations of DNN (data network name) and S-NSSAI (single network slice selection identifier) may be provided for each PDU session.

As referenced 8, the SMF determines whether the PDU Session is to be handled redundantly based on the updated policies from the PCF.

As referenced 9, the SMF uses the S-NSSAI and DNN to determine the RSN (redundancy sequence number) value which differentiates the PDU Sessions. At this point the SMF may differentiate the RL PDU session by for example assigning the RSN value specific for RL exploration. SMF may indicate along with RSN that this session is intended for RL exploration.

As referenced 10, the SMF indicates to the RAN redundant user plane requirements for the PDU sessions in the NG-RAN. Alternatively or additionally, the RL exploration indication may be included in the message.

As referenced 11 , the RSN parameter indicates to the NG-RAN that redundant user plane resources shall be provided for the given PDU Sessions using dual connectivity. Based on the additional information provided, the RAN is aware of the fact that this PDU session is used for RL exploration purpose.

As referenced 12, the NG-RAN determines if the dual connectivity request can be satisfied based on available resources. For example, the RAN node may consider RL exploration PDU session setup as having a lower priority and/or using any other suitable rule or criteria. In other embodiments, the PDU session which is set up for obtaining machine learning data may be given a higher priority or otherwise be treated differently from a PDU session which is not set up to provide machine learning data.

As referenced 13, the RAN may accept or reject the request based on the determination of the NG-RAN (referenced 12) and/or based on available resources. The RAN sends a message to the SMF indicating if the request is accepted or rejected.

As referenced 14, the SMF accepts or rejects PDU session request based on the message received from the RAN and provides an indication of this to the UE. This is based on the determination referenced 12. For example, if the “original” PDU session can be supported but not the redundant RL exploration PDU session, the SMF may decide not to reject the “original” session.

As referenced 15, the RAN may use standardized notification control mechanisms to notify the SMF if the RAN resources indicated by the RSN parameter can no longer be maintained. In addition, the notification may contain the information if it is related to a RL exploration redundant session.

As referenced 16, the SMF can use the received information to determine if the PDU session should be released. For example, if the “original” PDU session can be supported but not the redundant RL exploration PDU session, the SMF may decide not to reject the “original” session.

As the outcome of the exploration process, the data may be collected. For example, the NWDAF services may be used for data collection/analytics and the ML PDU CF can consume such services. This may be done in order to decide on a “best” action for a traffic steering decision. This may be with the goal of maximizing the QoS.

By means of RL PDU sessions, different actions can be explored. An action will be the association between the cell and the PDU session but without performing actual HO to the other cell. In order to decide on the appropriate action one or more of the following inputs may be taken into account: signal strength and interference; UE speed; UE and network capabilities (e.g. supported technologies); requested services and quality of service; cell load for different RATs (radio access technology) and/or cell layers (macro, small, pico, femto);

The reward in the RL algorithm will refer to an improvement or downgrade in QoS. Such a trained RL model (or additionally an inference output) can be provided by ML PDU CF to interested NFs.

The ML PDU CF functionality and associated service may be provided in other architectures such as for example ETSI ZSM.

One example ZSM architecture is shown in Figure 5. The ZSM architecture has an E2E (end to end) service management domain 400. The E2E service management domain may support E2E service orchestration 402, E2E service intelligence 404, E2E service assurance 406 and/or the like.

A management domain 408 is provided. The management domain interacts with the E2E service management domain via a first integration fabric 410. The management domain 408 may comprise an orchestration function 412, an intelligence function 414, a control function 416, an assurance function 418, an intent manager 419 and/or the like. Data services 420 may be provided which communicate via the first integration fabric.

The management domain communicates via a second integration fabric 422 with a physical domain 424, a virtual domain 426 and a cloud domain 428.

A so-called digital store front 430 providing automated customer and business management function interacts with the management domain and/or data services via the first integration fabric. ZSM may support the lifecycle management of communication services. This may involve multiple management processes that rely on the interaction between the end- to-end service management domain (E2E SMD) and the one or more management domains that work together to fulfil the communication services, the SMD sits above various domains. The 5G system may be regarded as being made up a “radio domain” and a “core network” domain. Each of the radio domain and the core domain may be associated with a domain management function which sits between the respective domain and the SMD.

It should be appreciated, that the ZSM architecture is still being developed and other embodiments may be used with any other suitable ZSM architecture

In some embodiments, a ML PDU CF be provided within the ZSM architecture to for example plan the setting up of redundant service sessions to gather data for training the ML model. This ML PDU CF may be provided as part of the management domain 408 and or the E2E service management domain 400.

Figure 3 illustrates an example of an apparatus 200. Ths apparatus may be provided for example in an access node, a communications device, a PCF, a SMF, a ML PDU CF and a TA AF. The apparatus may comprise at least one memory. By way of example only the at least one memory may comprise random access memory (RAM) 211 a and at least on read only memory (ROM) 211 b. Apparatus used by other embodiments may comprise different memory.

The apparatus may comprise at least one processor 212, 213. In this example apparatus, two processors are shown.

The apparatus may comprise an input/output interface 214.

The at least one processor may be coupled to the at least one memory. The at least one processor may be configured to execute an appropriate software code 215. The software code 215 may for example allow the method of some embodiments to be performed. The software code 215 may be stored in the at least one memory, for example ROM 211 b.

A method of some embodiments will now be described with reference to Figure 6. The method may be performed by an apparatus. The apparatus may be provided in an ML PDU CF or be an ML PDU CF. The apparatus may be as shown for example in Figure 3.

As referenced A1 , the method comprises determining a plan for one or more packet data sessions which are to be established or modified for obtaining data to be used for machine learning.

As referenced A2, the method comprises providing information to a network function to trigger the establishing or modifying of the one or more packet data sessions for obtaining data for the machine learning.

A method of some embodiments will now be described with reference to Figure 7. The method may be performed by an apparatus. The apparatus may be provided in a PCF or be a PCF. The apparatus may be as shown for example in Figure 3.

As referenced B1 , the method comprises receiving information at a policy control function from a machine learning control function about a plan for one or more packet data sessions which are to be established or modified for obtaining data to be used for machine learning.

As referenced B2, the method comprises determining if the plan is implementable.

As referenced B3, the method comprises providing a response to the machine learning control function if the plan is implementable. A method of some embodiments will now be described with reference to Figure 8. The method may be performed by an apparatus. The apparatus may be provided in an TA- AF or be an TA- AF. The apparatus may be as shown for example in Figure 3.

As referenced C1 , the method comprises receiving information at an application function from a machine learning control function, about one or more packet data sessions which are to be established or modified for obtaining data to be used for machine learning.

As referenced C2, the method comprises triggering the one or more packet data sessions to be respectively established or modified.

A method of some embodiments will now be described with reference to Figure 9. The method may be performed by an apparatus. The apparatus may be provided in a communications device or be a communications device. The apparatus may be as shown for example in Figure 3.

As referenced D1 , the method comprises receiving a trigger from an application function for a packet data session which is to be established or modified for obtaining data to be used for machine learning.

As referenced D2, the method comprises initiating establishment or modification of a first packet data session and a duplicate of that packet data session, the duplicate packet data session being the packet data session for obtaining the data to be used for machine learning, the first and duplicate packet data session being associated with different identity information.

A method of some embodiments will now be described with reference to Figure 10. The method may be performed by an apparatus. The apparatus may be provided in an SMF or be an SMF. The apparatus may be as shown for example in Figure 3.

As referenced E1 , the method comprises determining that a packet data session which is to be established or modified is to be used for obtaining data for machine learning. As referenced E2, the method comprises in response providing an indication for the packet data session which is to be established or modified for obtaining data for machine learning.

A method of some embodiments will now be described with reference to Figure 11 . The method may be performed by an apparatus. The apparatus may be provided in an access node or be an access node. The apparatus may be as shown for example in Figure 3.

As referenced F1 , the method comprises receiving from a session management function an indication, the indication indicating for a packet data session which is to be established or modified that the packet data session is to be established or modified to provide data for machine learning.

Although the apparatuses have been described as one entity, different modules and memory may be implemented in one or more physical or logical entities.

It is noted that whilst some embodiments have been described in relation to 5G networks, similar principles can be applied in relation to other networks and communication systems. Therefore, although certain embodiments were described above by way of example with reference to certain example architectures for wireless networks, technologies and standards, embodiments may be applied to any other suitable forms of communication systems than those illustrated and described herein.

Some embodiments may be implemented by circuitry. The term “circuitry” may refer to one or more or all of the following:

(a) hardware-only circuit implementations (such as implementations in only analog and/or digital circuitry);

(b) combinations of hardware circuits and software, such as (as applicable):

(i) a combination of analog and/or digital hardware circuit(s) with software/firmware; and (ii) any portions of hardware processor(s) with software (including digital signal processor(s)), software, and memory(ies) that work together to cause an apparatus, such as a mobile phone or server, to perform various functions), and

(c) hardware circuit(s) and or processor(s), such as a microprocessor(s) or a portion of a microprocessor(s), that requires software (e.g., firmware) for operation, but the software may not be present when it is not needed for operation.

This definition of circuitry applies to all uses of this term in this application, including in any claims. As a further example, as used in this application, the term circuitry also covers an implementation of merely a hardware circuit or processor (or multiple processors) or portion of a hardware circuit or processor and its (or their) accompanying software and/or firmware. The term circuitry also covers, for example, an integrated circuit or chiplet.

The embodiments of this disclosure may be implemented by computer software executable by a data processor, such as in the processor entity, or by hardware, or by a combination of software and hardware. Computer software or program, also called program product, including software routines, applets and/or macros, may be stored in any apparatus-readable data storage medium and they comprise program instructions to perform particular tasks. A computer program product may comprise one or more computer-executable components which, when the program is run, are configured to carry out embodiments. The one or more computer-executable components may be at least one software code or portions of it.

Further in this regard it should be noted that any blocks of the logic flow as in the Figures may represent program steps, or interconnected logic circuits, blocks and functions, or a combination of program steps and logic circuits, blocks, and functions. The software may be stored on such physical media as memory chips, or memory blocks implemented within the processor, magnetic media such as hard disk or floppy disks, and optical media such as for example DVD and the data variants thereof, CD or any other suitable physical media. The physical media is a non-transitory media. The memory may be of any type suitable to the local technical environment and may be implemented using any suitable data storage technology, such as semiconductor based memory devices, magnetic memory devices and systems, optical memory devices and systems, fixed memory, and removable memory. The data processors may be of any type suitable to the local technical environment, and may comprise one or more of general purpose computers, special purpose computers, microprocessors, digital signal processors (DSPs), application specific integrated circuits (ASIC), FPGA, gate level circuits and processors based on multi core processor architecture, as nonlimiting examples. Embodiments of the disclosure may be practiced in various components such as integrated circuit modules.

The scope of protection sought for various embodiments of the disclosure is set out by the independent claims. The embodiments and features, if any, described in this specification that do not fall under the scope of the independent claims are to be interpreted as examples useful for understanding various embodiments of the disclosure.

The foregoing description has provided by way of non-limiting examples a full and informative description of the exemplary embodiment of this disclosure. However, various modifications and adaptations may become apparent to those skilled in the relevant arts in view of the foregoing description, when read in conjunction with the accompanying drawings and the appended claims. However, all such and similar modifications of the teachings of this disclosure will still fall within the scope of this invention as defined in the appended claims. Indeed, there is a further embodiment comprising a combination of one or more embodiments with any of the other embodiments previously discussed.

Claims

42 Claims

1 . An apparatus comprising means configured to: determine a plan for one or more packet data sessions which are to be established or modified for obtaining data to be used for machine learning; and provide information to a network function to trigger the establishing or modifying of the one or more packet data sessions for obtaining data for the machine learning.

2. The apparatus as claimed in claim 1 , wherein the means is configured to provide as part of the plan one or more of: one or more communications devices associated with the one or more packet data sessions; one or more characteristics of one or more of the packet data sessions; a duration for a training/exploration phase for obtaining the data for the machine learning; and a number of the one or more packet data sessions.

3. The apparatus as claimed in claim 1 or 2, wherein the means is configured to determine the plan taking into account one or more of: one or more conditions; and one or more policies.

4. The apparatus as claimed in claim 3, wherein one or more conditions and/or one or more policies are associated with a network in which the one or more packet data sessions which are to be established or modified.

5. The apparatus as claimed in claim 3 or 4, wherein one or more conditions and/or one or more policies comprise one or more of a quality of service requirement, a quality of experience requirement, a network condition, a load threshold, and an interference level.

6. The apparatus as claimed in any preceding claim, wherein the one or more packet data sessions which are to be established or modified for obtaining data for the machine learning are redundant packet data sessions. 43

7. The apparatus as claimed in any preceding claim, wherein the means is configured to provide information about the plan to a policy control function.

8. The apparatus as claimed claim 7, wherein the means is configured to receive a response from the policy control function in response to the information about the plan and in response providing information to the network function to trigger the establishing or modifying of the one or more packet data sessions for obtaining data for the machine learning.

9. The apparatus as claimed in any preceding claim, wherein the means is configured to collect data associated with the one or more of the one or more data packet data sessions which have been established or modified.

10. The apparatus as claimed in claim 9, wherein the means is configured to train a reinforcement machine learning model in an exploration phase using said collected data.

11. The apparatus as claimed in claim 10, wherein the means is configured to output the trained machine learning model.

12. The apparatus as any preceding claim, wherein the means is configured to output inferred data.

13. An apparatus comprising means configured to: receive information at a policy control function from a machine learning control function about a plan for one or more packet data sessions which are to be established or modified for obtaining data to be used for machine learning; determine if the plan is implementable; and provide a response to the machine learning control function if the plan is implementable.

14. The apparatus as claimed in claim 13, wherein the information about the plan comprises one or more of: one or more communications devices associated with the 44 one or more packet data sessions; one or more characteristics of one or more of the packet data sessions; a duration for a training/exploration phase for obtaining the data for the machine learning; and a number of the packet data sessions.

15. The apparatus as claimed in claim 13 or 14, wherein the means is configured to determine the plan if the plan is implementable taking into account one or more of: subscription profile information associated with one or more of the communication devices associated with the one or more packet data sessions; one or more network operator policies; and configuration of one or more network entities.

16. The apparatus as claimed in any of claims 13 to 15, wherein the means is configured to update one or more policies and/or one or more rules in response to the received information.

17. The apparatus as claimed in any of claims 13 to 16, wherein the means is configured to update communications device route selection policy at one or more communications devices associated with the one or more packet data sessions in response to said received information.

18. An apparatus comprising means configured to: receive information at an application function from a machine learning control function, about one or more packet data sessions which are to be established or modified for obtaining data to be used for machine learning; and trigger the one or more packet data sessions to be respectively established or modified.

19. The apparatus as claimed in claim 18, wherein the means is configured to cause the one or more packet data sessions to be respectively established or modified by causing a respective message to be provided to a respective communications device associated with a respective one of the one or more packet data sessions to be established or modified.

20. An apparatus comprising means configured to: receive a trigger from a network function for a packet data session which is to be established or modified for obtaining data to be used for machine learning; and initiate establishment or modification of a first packet data session and a duplicate of that packet data session, the duplicate packet data session being the packet data session for obtaining the data to be used for machine learning, the first and duplicate packet data session being associated with different identity information.

21 . An apparatus comprising means configured to: determine that a packet data session which is to be established or modified is to be used for obtaining data for machine learning; and in response provide an indication for the packet data session which is to be established or modified for obtaining data for machine learning.

22. An apparatus comprising means configured to: receive from a session management function an indication, the indication indicating for a packet data session which is to be established or modified that the packet data session is to be established or modified to provide data for machine learning.

23. The apparatus as claimed in claim 21 or 22, wherein the means is configured to assign a priority to the packet data session established or modified to provide data for machine learning, the priority assigned being based on the packet data session being for providing data for machine learning.

24. A method comprising: determining a plan for one or more packet data sessions which are to be established or modified for obtaining data to be used for machine learning; and providing information to a network function to trigger the establishing or modifying of the one or more packet data sessions for obtaining data for the machine learning.

25. A method comprising: receiving information at a policy control function from a machine learning control function about a plan for one or more packet data sessions which are to be established or modified for obtaining data to be used for machine learning; determining if the plan is implementable; and providing a response to the machine learning control function if the plan is implementable.

26. A method comprising: receiving information at an application function from a machine learning control function, about one or more packet data sessions which are to be established or modified for obtaining data to be used for machine learning; and triggering the one or more packet data sessions to be respectively established or modified.

27. A method comprising: receiving a trigger from an application function for a packet data session which is to be established or modified for obtaining data to be used for machine learning; and initiating establishment or modification of a first packet data session and a duplicate of that packet data session, the duplicate packet data session being the packet data session for obtaining the data to be used for machine learning, the first and duplicate packet data session being associated with different identity information.

28. A method comprising: determining that a packet data session which is to be established or modified is to be used for obtaining data for machine learning; and in response providing an indication for the packet data session which is to be established or modified for obtaining data for machine learning.

29. A method comprising: receiving from a session management function an indication, the indication indicating for a packet data session which is to be established or modified that the 47 packet data session is to be established or modified to provide data for machine learning.

30. A computer program comprising computer executable code which when run on at least one processor cause the method of any of claims to be performed.