WO2023083436A1

WO2023083436A1 - Device and method for ran-based qos prediction

Info

Publication number: WO2023083436A1
Application number: PCT/EP2021/081107
Authority: WO
Inventors: Chan Zhou; Apostolos KOUSARIDAS; Miquel Angel GUTIERREZ ESTEVEZ
Original assignee: Huawei Technologies Co., Ltd.
Priority date: 2021-11-09
Filing date: 2021-11-09
Publication date: 2023-05-19

Abstract

The present disclosure relates to entities and methods for generating a full QoS prediction for a UE consuming a service from one or more access network entities. The disclosure proposes a first core network entity that is configured to: provide, to each of the one or more access network entities, a QoS prediction request, wherein the QoS prediction request comprises prediction configuration information; receive a prediction response from each access network entity of the one or more access network entities, wherein the prediction response comprises QoS prediction information of the access network entity; and derive the full QoS prediction based on the QoS prediction information in the one or more prediction responses. Further, the disclosure proposes other core network entities and an access network entity being configured to operate accordingly.

Description

DEVICE AND METHOD FOR RAN-BASED QoS PREDICTION

TECHNICAL FIELD

The present disclosure relates to communication networks, in particular, to wireless networks with Quality of Service (QoS) requirements. Different types of communication networks are considered including, but not limited to, cellular networks (e.g., with communication through the Uu interface in a 3 GPP communication system), ad-hoc networks (e.g., with communication through the PC5 interface in a 3GPP communication system), satellite, WiFi etc. The disclosure introduces functionalities for core network entities and access network entities to predict the QoS of a specific QoS session and/or QoS flow.

BACKGROUND

Many applications have very strict QoS requirements on wireless systems, e.g., the requirement of a guaranteed bit rate or an ultra-low latency. However, due to the nature of wireless channels, it is difficult to permanently preserve a high QoS performance. The performance may drop rapidly due to the change of a mobile user position, network load, or other impacting factors. Such a sudden performance degradation may lead to service discontinuity, which is critical especially for safety-relevant applications, e.g., for traffic safety services based on vehicular communications. Therefore, there is a need to notify applications about an expected or estimated change of the QoS before an actual change of the QoS occurs. The application will be able to proactively perform an adaptation to the change of QoS. For instance, if a teleoperation application would be aware of an upcoming throughput degradation of uplink video streaming, it ccould reduce the speed of a vehicle or could change the driving behavior, in order to ensure safety.

3 GPP has introduced the requirement that the network must be able to provide an in-advance notification to the application that the QoS is going to degrade. The notification should be sent with a configurable notification period (typically several seconds in advance) before the new QoS takes place.

The effectiveness of this in-advance notification mechanism is determined by the accuracy of the QoS prediction. In order to achieve the prediction accuracy required both in the time domain (update frequency) and the space domain (granularity of the provided information), a prediction function needs to access the radio measurement data, e.g., free space path loss, interference, reference signal received power (RSRP), reference signal received quality (RSRQ), and configuration information, e.g., scheduler information, antennas configuration, etc.

In current 3 GPP systems, QoS predictions are carried out at the core network. However, most radio measurement data and configuration information are only available at the radio access network (RAN). The exchange of this information between RAN andthe core network is difficult. Therefore, there is a need to define a method and corresponding signaling between the core network and the RAN to achieve the required precision of the QoS predictions while preserving isolation and privacy between core network and RAN.

SUMMARY

In view of the above, this disclosure aim to introduce a solution that enables the exchange of predictive QoS information between the RAN, the core network, and application functions (AFs), which are deployed external to a mobile operator network, while preserving isolation and privacy. An objective is to propose a QoS prediction mechanism that provides precise and fresh predictive QoS information to analytics consumers from the RAN. Another objective is to keep the isolation between different domains in the network. A further objective is to allow anonymization of user equipment (UE) information to ensure privacy.

These and other objectives are achieved by the solution of the present disclosure as provided in the enclosed independent claims. Advantageous implementations are further defined in the dependent claims.

A first aspect of the disclosure provides a first core network entity for generating a full QoS prediction for a UE consuming a service from one or more access network entities, the first core network entity being configured to: provide, to each of the one or more access network entities, a QoS prediction request, wherein the QoS prediction request comprises prediction configuration information; receive a prediction response from each access network entity of the one or more access network entities, wherein the prediction response comprises QoS prediction information of the access network entity; and derive the full QoS prediction based on the QoS prediction information in the one or more prediction responses. In this disclosure, a core network entity is proposed that requests RAN QoS prediction information to be provided from the RAN domain. The first core network entity may be a Network Data Analytics Function (NWDAF). It should be noted that the first core network entity may also be any other core network entity or an AF (part of a mobile operator network or external to a mobile operator network) that can request a RAN QoS prediction. After obtaining one or more QoS prediction responses from one or more access network entities (e.g., BSs), the first core network entity prepares the full QoS prediction based on the RAN prediction response, and may optionally also use other information, e.g., core network predictions or other analytics and predictions. Notably, the full QoS prediction may be referred to prediction result of end-to-end (E2E) QoS, which may involve all current or future nework components between the end points of a communication service. End-to-end QoS may comprise one or more access network entites and/or multiple domains (e.g., core network, access, cloud server, Multi-access Edge Computing (MEC), Mobile Edge Computing, private networks, public internet domain, in-vehicle network, local area networks etc).

In an implementation form of the first aspect, the prediction configuration information comprises one or more of: an identifier of the UE; route information of the UE; one or more QoS requirements; a target prediction horizon.

Possibly, the route information of the UE may include location information of the UE, e.g, information about a current and one or more future locations of the targeted UEs, in form of a route. The one or more QoS requirements may include a 5G QoS Identifier (5QI), e.g., a throughput, latency, packet error rate, or reliability, etc. The prediction horizon may refer to a time period indicating how long in advance the QoS prediction consumer expects to receive a QoS prediction notification. In particular, “how long in advance” may relate to the specific time, at which the QoS may actually change.

In an implementation form of the first aspect, the QoS prediction information of the access network entity comprises at least one of a partial QoS prediction performed by the access network entity and/or information from the access network entity for supporting a prediction performed by the first core network entity. Each access network entity may be implemented with a prediction function, that is able to predict the QoS for a specific path or area, according to the configuration received from the core network entity. Possibly, current and expected conditions at the access network entity such as a radio channel, one or more resources, load and traffic, may also be considered.

In an implementation form of the first aspect, the first core network entity is further configured to derive the full QoS prediction by performing the prediction based on the information from the one or more access network entities for supporting the prediction (e.g., resources availability, load conditions, radio channel information etc), and/or by combining the partial QoS predictions performed by the one or more access network entities.

In an implementation form of the first aspect, the partial QoS prediction comprises one or more of one or more predicted values and/or predicted ranges of a QoS parameter; one or more QoS maps; a prediction horizon; information about confidence of the one or more predicted values and/or predicted ranges; information about confidence of the one or more QoS maps.

Optionally, predicted QoS from each access network entity may include an expected or a predicted or an estimated value (or a range, e.g., predicted lower/upper bounds) of a QoS parameter (e.g., of throughput, latency, or packet loss, etc), with or without a specific confidence interval. A QoS map may refer to a geographical map that contains the predicted QoS parameters or predicted QoS indicators (e.g., of throughput, latency, or packet loss, etc) associated with locations of the map. This map can be one-dimensional or multi-dimensional (e.g., to represent a physical location or terrain, road, highway etc). The map may also refer to a path (or route) e.g., of a specific UE. In addition, the QoS map can include or be described by additional dimensions (not only geographical dimensions) to provide information about other parameters that may describe or impact the QoS (e.g., different frequency bands, and/or the time of the day and/or prediction horizon, and/or different beamformers of the MEMO antennas etc). For instance, a QoS map may refer to the location-specific predicted UL throughout of a UE for a specific period of time. The QoS map may represent the output of a prediction function. Possibly, the format of the QoS map can be per pixel (pixels of the map) or can be vectorized. A pixel may be single-dimensional or multi-dimensional.

In an implementation form of the first aspect, the QoS prediction request further comprises configuration information regarding at least one granularity of the one or more QoS maps.

Possibly, the one or more QoS maps may be associated with different granularity values. Alternatively, they may also use the same granularity value.

In an implementation form of the first aspect, providing the QoS prediction request comprises sending the QoS prediction request to each of the one or more access network entities via a second core network entity, or via a dedicated access network entity; and receiving the prediction response from each access network entity comprises receiving a QoS prediction report from the second core network entity, or from the dedicated access network entity, wherein the QoS prediction report comprises the prediction response from each of the one or more access network entities.

Notably, an intermediate entity, e.g., the second core network entity or the dedicated access network entity, may be used by the first core network entity to forward the requests, for providing the RAN QoS prediction information, to the RAN domain.

In an implementation form of the first aspect, deriving the full QoS prediction based on the QoS prediction information comprises deriving the full OoS prediction based on the QoS prediction report.

In this example, the second core network entity or the dedicated access network entity aggregates the QoS prediction responses from different access network entities and provides the QoS prediction report to the first core network entity.

In an implementation form of the first aspect, the first core network entity is further configured to receive an initial QoS prediction request from an application function or another core network entity, wherein the initial QoS prediction request comprises application layer information and/or the prediction configuration information; generate the QoS prediction request based on the initial QoS prediction request; and send a QoS prediction response comprising the full QoS prediction to the application function or the another core network entity in response to the initial QoS prediction request.

The request for the QoS prediction can be provided by an AF and/or by any analytics consumer (e.g., core network entity) to the first core network entity. Accordingly, the first core network entity provides the QoS prediction response to the AF that has requested this information.

In an implementation form of the first aspect, the second core network entity is an access mobility function, and the first core network is further configured to identify the second core network entity based on the initial QoS prediction request.

The intermediate entity may be an access mobility function (AMF). For instance, the AMF can be identified based on the identifier of the UE, e.g., by contacting the Unified Data Management (UDM) or any other core network entity.

In an implementation form of the first aspect, each of the one or more QoS maps comprises one or more of the following QoS indicators:

- throughput; latency; packet loss;

- jitter; availability; mean time to failure; reliability guarantee; accuracy of localization information; sensing accuracy.

A second aspect of the disclosure provides a second core network entity for supporting a QoS prediction for a UE consuming a service from one or more access network entities, the second core network entity being configured to: receive, from a first core network entity, a QoS prediction request, wherein the QoS prediction request comprises prediction configuration information; provide, to each of one or more access network entities, a partial QoS prediction request based on the QoS prediction request; receive a prediction response from each of the one or more access network entities, wherein the prediction response comprises QoS prediction information of the access network entity; and send a QoS prediction report to the first core network entity, wherein the QoS prediction report comprises the prediction response from each of the one or more access network entities.

In this aspect of the disclosure, an intermediate entity, i.e., the second core network entity (e.g., AMF) is porposed, which undertakes the first core network entity to configure the RAN QoS prediction request, and sends the request to the RAN entities. Possibly, the QoS prediction report may be named differently, for instance, it can also be referred to as “QoS prediction notification” or “QoS prediction information” as in the 3 GPP specification.

Optionally, the second core network entity may contact a global RAN prediction function, which can request and collect the data that is relevant for a QoS prediction from each BS. The global RAN prediction function may carry out the QoS prediction aggregation function, e.g., by generating a combined prediction response such as the QoS prediction report.

In an implementation form of the second aspect, the prediction configuration information comprises one or more of: an identifier of the UE; route information of the UE; one or more QoS requirements; and a target prediction horizon.

Possibly, the route information of the UE may include location information of the UE, i.e., information about a current and/orone or more future locations of the targeted UEs in form of a route. The one or more QoS requirements may include 5QI, e.g., a throughput, latency, packet error rate, or reliability, etc. The prediction horizon may refer to a time period indicating how long in advance the QoS prediction consumer expects to receive a QoS prediction notification. In particular, “how long in advance” relates to the specific time, at which the QoS may actually change.

In an implementation form of the second aspect, the QoS prediction information of the access network entity comprises at least one of a partial QoS prediction performed by the access network entity and/or one or more QoS parameters of the access network entity. Each access network entity may be implemented with a prediction function, that is able to predict QoS for a specific path or area, according to the configuration received from the core network entity. Possibly, current and expected conditions at the access network entity, such as radio channel, resources, load, or traffic, may also be considered.

In an implementation form of the second aspect, the partial QoS prediction comprises one or more of one or more predicted values or ranges of a QoS parameter; one or more QoS maps; a prediction horizon; information about confidence of the one or more predicted values and/or predicted ranges; information about confidence of the one or more QoS maps.

The partial QoS prediction received by an individual access network entity may include an expected or a predicted or an estimated value (or a range, e.g., predicted lower/upper bounds) of one or more QoS parameters (e.g., a throughput, latency, or packet loss, etc), with or without a specific confidence interval. The partial QoS prediction is provided for an interface that the access network entity provides to support a comunciation service. QoS maps may refer to predicted QoS maps or paths containing the predicted information. Details of the QoS maps are described in a later part of the application. The partial QoS may refer to interfaces of one or more other network entites.

In an implementation form of the second aspect, the QoS prediction request further comprises configuration information regarding at least one granularity of the one or more QoS maps.

In an implementation form of the second aspect, the second core network entity is further configured to generate the partial QoS prediction request for a particular access network entity of the one or more access network entities, wherein the partial QoS prediction request comprises one or more of an identifier of a flow, wherein the UE is associated with a session of the service, the session comprising the flow; a session identifier of the session; the identifier of the UE; route information of the UE, or partial route information of the UE corresponding to the particular access network entity;

- the at least one granularity of the one or more QoS maps;

- the target prediction horizon.

It should be understood that the partial QoS prediction request, which is provided to an individual access network entity, may indicate route information of the UE in the coverage of that access network entity. In order to further protect the privacy of UE, several measures can be carried out during the QoS prediction request. For instance, the parameters of the prediction horizon, which are sent to each BS, can have a randomly selected overlapping window, in order to hinder the continuous tracing of the UE crosses multiple BS coverage. The partial route information can also be left out from the QoS prediction request.

In an implementation form of the second aspect, the second core network entity is further configured to determine the one or more access network entities based on the QoS prediction request and/or based further on handover knowledge from an access mobility function.

In particular, the list of BSs (e.g., including Long Term Evolution (LTE) BS, 5^th Generation (5G) BS, etc.) which the QoS prediction request should be sent to, may be identified based, e.g., on route information, the AMF‘s knowledge of a handover, UE location information, etc.

In an implementation form of the second aspect, the second core network entity is an access mobility function.

In such a case, the AMF can identify the list of BSs.

In an implementation form of the second aspect, the second core network entity is a network management entity.

Alternatively, the list of BSs can be retrieved by the 0AM.

In an implementation form of the second aspect, the network management entity comprises one or more operations administration and maintenance (0AM) entities. In a particular implementation, the OAM entity may be distributed into multiple entities.

In an implementation form of the second aspect, the second core network entity is further configured to generate the QoS prediction report by performing a QoS estimation based on the prediction response from each of the one or more access network entities.

Possibly, the QoS prediction report can be realized in various ways. For instance, the second core network entity may integrate the prediction response from each access network entity, may process the prediction responses, and may then integrate them, or may concatenate the prediction responses.

A third aspect of the disclosure provides a third core network entity for subscribing to a QoS prediction notification, the third core network entity being configured to: send, to each access network entity of one or more access network entities, a QoS notification subscription, wherein the QoS notification subscription instructs the access network entity to send a QoS notification to the third core network entity, when the access network entity predicts a change on the QoS prediction for a UE consuming a service of the one or more access network entities; and receive the QoS notification from the one or more access network entities.

Notably, an application may subscribe to a core network entity (e.g., a Policy Control Function (PCF), or a Session Management Function (SMF)) to receive notifications for QoS predictions estimated by the RAN for specific applications, services, and/or flows. This disclosure thus further proposes a third core network entity for such purposes.

Optionally, the third core network entity may contact a global function, which can request and collect the data that is relevant for a QoS prediction from each BS. Possibly, such global function may carry out an aggregation function, e.g., by generating a combined QoS notification.

In an implementation form of the third aspect, the QoS notification subscription comprises one or more of: an identifier of a flow, wherein the UE is associated with a session of the service, the session comprising the flow; a session identifier of the session; an identifier of the UE; route information of the UE; one or more QoS requirements; one or more QoS profiles; one or more QoS parameters; at least one granularity of one or more QoS maps; a target prediction horizon.

The QoS notification subscription triggers one or more RAN entities to predict the QoS of a specific UE (or a group of UEs), and notify the third core network entity for potential predicted changes. This can be at least one of a dedicated message and a part of a PDU Session establishment and/or a modification message.

In an implementation form of the third aspect, the QoS notification comprises one or more of: one or more predicted values or ranges of a QoS parameter;

- the one or more QoS maps; a prediction horizon; information about confidence of the one or more predicted values and/or predicted ranges; information about confidence of the one or more QoS maps.

In an implementation form of the third aspect, the third core network entity is further configured to: receive a QoS notification subscription request from an application function, wherein the QoS notification subscription request comprises application layer information and/or configuration information for the QoS prediction notification; generate the QoS notification subscription to a particular access network entity of the one or more access network entities based on the QoS notification subscription request; and send the QoS notification to the application function.

Optionally, an application can subscribe (optionally via NEF) to a third core network entity (e.g., a PCF or SMF) to receive notifications for QoS predictions estimated by RAN for a specific application, service, or flow.

A fourth aspect of the disclosure provides an access network entity for predicting a QoS for a UE consuming a service of one or more access network entities including the access network entity, wherein the UE is associated with a session of the service, the session comprising a flow, and wherein the access network entity is configured to: receive, from a core network entity or from the UE, a QoS prediction request, wherein the QoS prediction request comprises prediction configuration information; generate QoS prediction information by performing a QoS prediction for the flow based on the prediction configuration information and/or by determining information for supporting a prediction based on the prediction configuration information; and send a prediction response comprising the QoS prediction information to the core network entity and/or to the UE.

This disclosure further proposes an access network entity that performs a local QoS prediction in response to the request received from the core network entity, or the UE. Individual predictions will be collected by the core network entity and/or the UE from individual access network entities. The disclosure thus enables the aggregation of predictive analytics from several sources, e.g. several BSs.

In an implementation form of the fourth aspect, the QoS prediction request comprises one or more of an identifier of the flow; a session identifier of the session; the identifier of the UE; partial route information of the UE corresponding to the access network entity;

- the at least one granularity of the one or more QoS maps;

- the target prediction horizon; one or more QoS profiles; one or more QoS parameters.

The QoS prediction request carries prediction configuration information that can be used by the individual access network entity to perform a QoS prediction or estimation.

In an implementation form of the fourth aspect, the QoS prediction information comprises one or more of: one or more predicted values or ranges of a QoS parameter; the one or more QoS maps; a prediction horizon; information about confidence of the one or more predicted values and/or predicted ranges; information about confidence of the one or more QoS maps.

This disclosure proposes to use QoS maps as vectors of the predictive information generated at the RAN side. This allows the RAN entity to provide precise and fresh predictive QoS information to the core network entities, while preserving isolation and privacy between diffenret domains.

In an implementation form of the fourth aspect, each of the one or more QoS maps comprises one or more of the following QoS indicators:

- throughput; latency; packet loss;

Possibly, the format of the QoS map can be per pixel (pixels of the map) or can be vectorized. In either case, the index of a QoS map represents a quantized value of the QoS metric for a predefined size of the grid.

In an implementation form of the fourth aspect, when the QoS prediction request is received from the UE, the access network entity is further configured to: send a further QoS prediction request to a further access network entity; receive a further prediction response from the further access network entity; generate a combined prediction response based on the further prediction response and the prediction response; and send the combined prediction response to the UE.

Optionally, a global RAN prediction function can request and collect the data that is relevant for a QoS prediction from each BS, and can also carry out the QoS prediction aggregation function, e.g., by generating a combined prediction response. For instance, the global RAN prediction function can be co-located at one of the one or more access network entities, e.g., the serving BS to which the UE is attached.

A fifth aspect of the disclosure provides a method for generating a full QoS prediction for a UE consuming a service from one or more access network entities, comprising: providing, to each of one or more access network entities, a QoS prediction request, wherein the QoS prediction request comprises prediction configuration information; receiving a prediction response from each of the one or more access network entities, wherein the prediction response comprises QoS prediction information of the access network entity; and deriving the full QoS prediction based on the QoS prediction information in the one or more prediction responses.

Implementation forms of the method of the fifth aspect may correspond to the implementation forms of the first core network entity of the first aspect described above. The method of the fifth aspect and its implementation forms achieve the same advantages and effects as described above for the first core network entity of the first aspect and its implementation forms.

A sixth aspect of the disclosure provides a method for supporting a QoS prediction for a UE consuming a service from one or more access network entities, comprising: receiving, from a first core network entity, a QoS prediction request, wherein the QoS prediction request comprises prediction configuration information; providing, to each of one or more access network entities, a partial QoS prediction request based on the QoS prediction request; receiving a prediction response from each of the one or more access network entities, wherein the prediction response comprises QoS prediction information of the access network entity; and sending a QoS prediction report to the first core network entity, wherein the QoS prediction report comprises the prediction response from each of the one or more access network entities.

Implementation forms of the method of the sixth aspect may correspond to the implementation forms of the second core network entity of the second aspect described above. The method of the sixth aspect and its implementation forms achieve the same advantages and effects as described above for the second core network entity of the second aspect and its implementation forms.

A seventh aspect of the disclosure provides a method performed by a third core network entity for subscribing to a QoS prediction notification, comprising: sending, to each access network entity of one or more access network entities, a QoS notification subscription, wherein the QoS notification subscription instructs the access network entity to send a QoS notification to the third core network entity, when the access network entity predicts a change on the QoS prediction for a UE consuming a service of the one or more access network entities; and receiving the QoS notification from the one or more access network entities.

Implementation forms of the method of the seventh aspect may correspond to the implementation forms of the third core network entity of the third aspect described above. The method of the seventh aspect and its implementation forms achieve the same advantages and effects as described above for the third core network entity of the third aspect and its implementation forms.

A eighth aspect of the disclosure provides a method performed by an access network entity for predicting a QoS for a UE consuming a service of one or more access network entities including the access network entity, wherein the UE is associated with a session of the service, the session comprising a flow, and wherein the method comprises: receiving, from a core network entity or from the UE, a QoS prediction request, wherein the QoS prediction request comprises prediction configuration information; generating QoS prediction information by performing a QoS prediction on the flow based on the prediction configuration information and/or by determining information for supporting a prediction based on the prediction configuration information; and sending a prediction response comprising the QoS prediction information to the core network entity and/or to the UE.

Implementation forms of the method of the eighth aspect may correspond to the implementation forms of the access network entity of the fourth aspect described above. The method of the eighth aspect and its implementation forms achieve the same advantages and effects as described above for the access network entity of the fourth aspect and its implementation forms.

A ninth aspect of the disclosure provides a computer program product comprising a program code for carrying out, when implemented on a processor, the method according to the fifth aspect and any implementation forms of the fifth aspect, the sixth aspect and any implementation forms of the sixth aspect, the seventh aspect and any implementation forms of the seventh aspect, or the eighth aspect and any implementation forms of the eighth aspect. It has to be noted that all devices, elements, units and means described in the present application could be implemented in software or hardware elements or any kind of combination thereof. All steps which are performed by the various entities described in the present application as well as the functionalities described to be performed by the various entities are intended to mean that the respective entity is adapted to or configured to perform the respective steps and functionalities. Even if, in the following description of specific embodiments, a specific functionality or step to be performed by external entities is not reflected in the description of a specific detailed element of that entity which performs that specific step or functionality, it should be clear for a skilled person that these methods and functionalities can be implemented in respective software or hardware elements, or any kind of combination thereof.

BRIEF DESCRIPTION OF DRAWINGS

The above-described aspects and implementation forms of the present disclosure will be explained in the following description of specific embodiments in relation to the enclosed drawings, in which:

FIG. 1 shows domain isolation requirements for analytics functions;

FIG. 2 shows an example of message flows for a predictive QoS request and reporting;

FIG. 3 shows a first core network entity according to an embodiment of the disclosure;

FIG. 4 shows a first core network entity according to an embodiment of the disclosure;

FIG. 5 shows a second core network entity according to an embodiment of the disclosure;

FIG. 6 shows a third core network entity according to an embodiment of the disclosure; FIG. 7 shows an access network entity according to an embodiment of the disclosure;

FIG. 8 shows an example of input/ output mapping of an exemplary QoS prediction map function;

FIG. 9 shows an exemplary function for computing predictive QoS;

FIG. 10 shows an example of message exchanges between an application and networks according to an embodiment of the disclosure;

FIG. 11 shows an example of message exchanges between an application and networks; FIG. 12 shows an example of aggregating prediction results from different BSs;

FIG. 13 shows an exemple of distributed and centralized RAN prediction functions;

FIG. 14 shows an example of a packets retransmission;

FIG. 15 shows an example of message exchanges between an application and BSs; FIG. 16 shows a method according to an embodiment of the disclosure;

FIG. 17 shows a method according to an embodiment of the disclosure;

FIG. 18 shows a method according to an embodiment of the disclosure; and

FIG. 19 shows a method according to an embodiment of the disclosure.

DETAILED DESCRIPTION OF EMBODIMENTS

Illustrative embodiments of a first core network entity, a second core network entity, a third core network entity, an access network entity and corresponding methods for generating a full QoS prediction for a UE are described with reference to the figures. Although this description provides a detailed example of possible implementations, it should be noted that the details are intended to be exemplary and in no way limit the scope of the application.

Moreover, an embodiment or example may refer to other embodiments or examples. For example, any description including but not limited to terminology, element, process, explanation, and/or technical advantage mentioned in one embodiment/example is applicative to the other embodiments or examples.

As previously discussed, most radio measurement data and configuration information are only available at the RAN, while the QoS prediction request is operated via core network entities, e.g., the NWDAF. The exchange of this information between the RAN and the core network is difficult, on the one hand, because of the high overhead to transfer such an amount of data from the RAN to the CN, and on the other hand, because of isolation requirements between the RAN and the CN, and because of privacy constraints of UEs. FIG. 1 shows exemplarily the requirements of data isolation between the core network and the RAN for an analytics functions.

Typically, the NWDAF predicts the QoS change based on the information provided by an Operations, Administration, and Maintenance (0AM) entity. The 0AM monitors the parameters of the RAN performance on the cell/sector level with an update period in the range of minutes or hours. FIG. 2 shows an example of message flows of the current mechanism in 3GPP used to generate predictive QoS information in a Vehicle-to-Everything (V2X) application scenario. In this example, it can be seen that input data is collected from vehicles, via the RAN entity (NG-RAN), and is provided to the NWDAF by the 0AM entity. This mechanism has the problem of a lack of granularity and freshness of the QoS prediction, due to cell/sector level geographical granularity and temporal freshness. In particular, the QoS prediction uses an input of cell/sector level measurement reports provided by the 0AM, and therefore the prediction accuracy is restricted to the cell/sector level. Further, the update period of the 0AM is in the range of minutes or hours, which impacts the timeliness of the prediction results.

Therefore, a method and corresponding signaling between the core network and the RAN to achieve the required precision of predictions, while preserving isolation and privacy between core network and RAN, are desired.

FIG. 3 shows a first core network entity 301 adapted for generating a full QoS prediction for a UE according to an embodiment of the disclosure. The UE consumes a service from one or more access network entities 401, 402. Notably, the full QoS prediction may be referred to as an E2E QoS prediction result. E2E typically describes a process that takes a system or service from beginning to end and delivers a complete functional solution. The access network entities 401, 402 may include for example a base station (BS) or a RAN entity such as access point, Central Unit (CU)/Distributed Unit (DU), RAN controller.

The first core network entity 301 may comprise processing circuitry (not shown) configured to perform, conduct or initiate the various operations of the first core network entity 301 described herein. The processing circuitry may comprise hardware and software. The hardware may comprise analog circuitry or digital circuitry, or both analog and digital circuitry. The digital circuitry may comprise components such as application-specific integrated circuits (ASICs), field-programmable arrays (FPGAs), digital signal processors (DSPs), or multi-purpose processors. The first core network entity 301 may further comprise memory circuitry, which stores one or more instruction(s) that can be executed by the processor or by the processing circuitry, in particular under the control of the software. For instance, the memory circuitry may comprise a non-transitory storage medium storing executable software code which, when executed by the processor or the processing circuitry, causes the various operations of the first core network entity 301 to be performed. In one embodiment, the processing circuitry comprises one or more processors and a non-transitory memory connected to the one or more processors. The non-transitory memory may carry executable program code which, when executed by the one or more processors, causes first core network entity 301 to perform, conduct or initiate the operations or methods described herein. The first core network entity 301 is configured to provide, to each of the one or more access network entities 401, 402, a QoS prediction request 3011. The QoS prediction request 3011 comprises prediction configuration information. The first core network entity 301 is further configured to receive a prediction response 3012 from each access network entity 401 of the one or more access network entities 401, 402. The prediction response 3012 comprises QoS prediction information 4011 of the access network entity 401. Then, the first core network entity 301 is configured to derive the full QoS prediction 3013 based on the QoS prediction information 4011 in the one or more prediction responses 3012.

Optionally, the prediction configuration information may comprise one or more of an identifier of the UE; route information of the UE; one or more QoS requirements; a target prediction horizon.

Possibly, the route information of the UE may include location information of the UE, for example, information about a current and one or more future locations of the targeted UEs in form of a route. The one or more QoS requirements may include 5G QoS Identifier (5QI), e.g., throughput, latency, packet error rate, reliability, etc. The prediction horizon may refer to a time period indicating how long in advance the QoS prediction consumer expects to receive a QoS prediction notification..

Optionally, the QoS prediction information 4011 of the access network entity 401 may comprise a partial QoS prediction performed by the access network entity 401. The first core network entity 301 may be further configured to derive the full QoS prediction 3013 by combining the partial QoS predictions performed by the one or more access network entities 401, 402.

Optionally, the QoS prediction information 4011 of the access network entity 401 may comprise information from the access network entity 401 for supporting a prediction performed by the first core network entity 301. In such a case, the first core network entity 301 may be further configured to derive the full QoS prediction 3013 by performing the prediction based on the information from the one or more access network entities 401, 402 for supporting the prediction. Notably, the above-mentioned two options may also be combined. That is, the QoS prediction information 4011 of the access network entity 401 may comprise a partial QoS prediction and the information for supporting the prediction.

Optionally, the partial QoS prediction may comprise one or more of: one or more predicted values and/or predicted ranges of a QoS parameter; one or more QoS maps; a prediction horizon; information about a confidence of the one or more predicted values and/or predicted ranges; information about a confidence of the one or more QoS maps.

Predicted QoS from the access network entity 401 may include expected or predicted or estimated value (or range, e.g., predicted lower/upper bounds) of a QoS parameter (e.g., throughput, latency, packet loss, etc), with or without a specific confidence interval. QoS maps may refer to predicted QoS maps or paths containing the predicted information. Details of the QoS maps are described in a later part of the application.

Optionally, the QoS prediction request 3011 may further comprise configuration information regarding at least one granularity of the one or more QoS maps. Possibly, the one or more QoS maps may be associated with different granularity values. However, they may also use the same granularity value.

According to an embodiment of the disclosure, an intermediate entity other than the first core network entity 301 may undertake a configuration of the RAN side to request the QoS prediction, as shown in FIG. 4. FIG. 4 shows the first core network entity 301 based on the first core network entity 301 shown in FIG. 3.

Optionally, providing the QoS prediction request 3011 to each of the one or more access network entities 401, 402 may comprise sending the QoS prediction request 3011 to each of the one or more access network entities 401, 402 via a second core network entity 302, or a dedicated access network entity 403. That is, an intermediate entity, here the second core network entity 302 or the dedicated access network entity 403, may be used by the first core network entity 301 to forward the requests of providing RAN QoS prediction information to the RAN domain.

Accordingly, receiving the prediction response 3012 from each access network entity 401 may comprise receiving a QoS prediction report 3022 from the second core network entity 302, or the dedicated access network entity 403. The QoS prediction report 3022 comprises the prediction response 3012 from each of the one or more access network entities 401, 402.

Possibly, the QoS prediction report 3022 can be realized in various ways. For instance, the intermediate entity may integrate the prediction response 3012 from each access network entity 401, may process the prediction responses 3012, and may then integrate them, or may concatenate the prediction responses 3012.

Optionally, deriving the full QoS prediction 3013 based on the QoS prediction information 4011 may comprise deriving the full OoS prediction 3013 based on the QoS prediction report 3022.

This disclosure introduces functionalities for the first core network entity 301 (e.g., NWDAF) to request one or more access network entities 401, 402 (e.g., BSs) to predict the QoS of a specific QoS session and/or QoS flow and/or QoS class. Possibly, an intermediate entity, e.g., the second core network entity 302 (e.g., an AMF), or the dedicated access network entity 403 (e.g. a stand-alone RAN entity, e.g., a RAN Data Analytical Function), may undertake to configure the RAN QoS prediction request, and may send the request to the RAN entities. The information of predictive QoS may be provided by the RAN to the core network in the form of a QoS map. The intermediate entity aggregates QoS prediction responses to provide map-based QoS prediction. This mechanism allows core network and RAN domains to exchange necessary intermediate information to assist E2E QoS prediction without exposing raw data or sensitive information from RAN to core network (and vice versa). An intermediate entity (e.g., AMF) collects local QoS predictions by various BS to build a RAN QoS map for precise and fresh prediction of QoS metrics (delay, data rate, reliability, etc.).

The first core network entity 301 may be further configured to receive an initial QoS prediction request from an AF or another core network entity. Possibly, the initial QoS prediction request may comprise application layer information and/or the prediction configuration information. The first core network entity 301 may be further configured to generate the QoS prediction request 3011 based on the initial QoS prediction request. Then, the first core network entity 301 may be configured to send a QoS prediction response 3012 comprising the full QoS prediction 3013 to the AF or the another core network entity in response to the initial QoS prediction request.

Optionally, the second core network entity 302 is an AMF, and the first core network is further configured to identify the second core network entity 302 based on the initial QoS prediction request.

This disclosure further proposes a second core network entity 302 as shown in FIG. 5. FIG. 5 shows a second core network entity 302 adapted for supporting a full QoS prediction for a UE according to an embodiment of the disclosure. The UE consumes a service from one or more access network entities 401, 402. The access network entities 401, 402 may be the access network entities shown in FIG. 3 or FIG. 4.

The second core network entity 302 may comprise processing circuitry (not shown) configured to perform, conduct or initiate the various operations of the second core network entity 302 described herein. The processing circuitry may comprise hardware and software. The hardware may comprise analog circuitry or digital circuitry, or both analog and digital circuitry. The digital circuitry may comprise components such as application-specific integrated circuits (ASICs), field-programmable arrays (FPGAs), digital signal processors (DSPs), or multipurpose processors. The second core network entity 302 may further comprise memory circuitry, which stores one or more instruction(s) that can be executed by the processor or by the processing circuitry, in particular under the control of the software. For instance, the memory circuitry may comprise a non-transitory storage medium storing executable software code which, when executed by the processor or the processing circuitry, causes the various operations of the second core network entity 302 to be performed. In one embodiment, the processing circuitry comprises one or more processors and a non-transitory memory connected to the one or more processors. The non-transitory memory may carry executable program code which, when executed by the one or more processors, causes second core network entity 302 to perform, conduct or initiate the operations or methods described herein. The second core network entity 302 is configured to receive, from a first core network entity 301, a QoS prediction request 3011. The QoS prediction request 3011 comprises prediction configuration information. Possibly, the first core network entity 301 may be the first core network entity shown in FIG. 3 or FIG. 4. The second core network entity 302 is further configured to provide, to each of the one or more access network entities 401, 402, a partial QoS prediction request 3021 based on the QoS prediction request 3011. Then, the second core network entity 302 is configured to receive a prediction response 3012 from each of the one or more access network entities 401, 402. Notably, the prediction response 3012 comprises QoS prediction information 4011 of the access network entity 401. Further, the second core network entity 302 is configured to send a QoS prediction report 3022 to the first core network entity 301, wherein the QoS prediction report 3022 comprises the prediction response 3012 from each of the one or more access network entities 401, 402.

Possibly, the QoS prediction report 3022 may be named differently, such as it can also be referred to as “QoS prediction notification” or “QoS prediction information” as in the 3 GPP specification. Optionally, the second core network entity 302 may be configured to generate the QoS prediction report 3022 by performing a QoS estimation based on the prediction response 3012 from each of the one or more access network entities 401, 402.

As discussed in the previous embodiments, the prediction configuration information may comprise one or more of an identifier of the UE; route information of the UE; one or more QoS requirements; and a target prediction horizon.

Optionally, the QoS prediction information 4011 of the access network entity 401 may comprise a partial QoS prediction performed by the access network entity 401. Alternatively, the QoS prediction information 4011 of the access network entity 401 may comprise one or more QoS parameters of the access network entity 401. In another example, the QoS prediction information 4011 may comprise both the partial QoS prediction and the one or more QoS parameters of the access network entity 401. Possibly, the partial QoS prediction comprises one or more of: one or more predicted values or ranges of a QoS parameter; one or more QoS maps; a prediction horizon; information about a confidence of the one or more predicted values and/or predicted ranges; information about a confidence of the one or more QoS maps.

Optionally, the second core network entity 302 may be further configured to generate the partial QoS prediction request 3021 for a particular access network entity of the one or more access network entities 401, 402. The partial QoS prediction request 3021 may comprise one or more of an identifier of a flow, wherein the UE is associated with a session of the service, the session comprising the flow; a session identifier of the session; the identifier of the UE; route information of the UE, or partial route information of the UE corresponding to the particular access network entity;

- the at least one granularity of the one or more QoS maps;

- the target prediction horizon.

Optionally, the second core network entity 302 may be further configured to determine the one or more access network entities 401, 402 based on the QoS prediction request 3011. Optionally, the second core network entity 302 may be further configured to determine the one or more access network entities 401, 402 further based on handover knowledge from an AMF.

In a particular implementation, the second core network entity 302 may be an AMF. Possibly, the determination may also be conducted by another core network entity with similar functions.

Alternatively, the second core network entity 302 may be a network management entity, for instance, an 0AM entity. In a particular implementation, the 0AM entity may be distributed into multiple entities. FIG. 6 shows a third core network entity 303 adapted for subscribing to a QoS prediction notification according to an embodiment of the disclosure.

The third core network entity 303 may comprise processing circuitry (not shown) configured to perform, conduct or initiate the various operations of the third core network entity 303 described herein. The processing circuitry may comprise hardware and software. The hardware may comprise analog circuitry or digital circuitry, or both analog and digital circuitry. The digital circuitry may comprise components such as application-specific integrated circuits (ASICs), field-programmable arrays (FPGAs), digital signal processors (DSPs), or multipurpose processors. The third core network entity 303 may further comprise memory circuitry, which stores one or more instruction(s) that can be executed by the processor or by the processing circuitry, in particular under the control of the software. For instance, the memory circuitry may comprise a non-transitory storage medium storing executable software code which, when executed by the processor or the processing circuitry, causes the various operations of the third core network entity 303 to be performed. In one embodiment, the processing circuitry comprises one or more processors and a non-transitory memory connected to the one or more processors. The non-transitory memory may carry executable program code which, when executed by the one or more processors, causes the third core network entity 303 to perform, conduct or initiate the operations or methods described herein.

The third core network entity 303 is configured to send a QoS notification subscription 3031 to each access network entity 401 of one or more access network entities 401, 402. Possibly, the one or more access network entities 401, 402 may be the one or more access network entities shown in any one of FIG. 3, FIG. 4, and FIG. 5. The QoS notification subscription 3031 instructs the access network entity 401 to send a QoS notification 3032 to the third core network entity 303, when the access network entity 401 predicts a change on the QoS prediction for a UE consuming a service of the one or more access network entities 401, 402. Further, the third core network entity 303 is configured to receive the QoS notification 3032 from the one or more access network entities 401, 402.

An application can subscribe (optionally via Network Exposure Function (NEF)) to a core network entity (e.g., Policy Control Function (PCF), or Session Management Function (SMF)) to receive notifications for QoS predictions estimated by RAN for specific applications, services, and/or flows. This disclosure thus further proposes the third core network entity 303 for such purposes.

Optionally, the QoS notification subscription 3031 may comprise one or more of: an identifier of a flow, wherein the UE is associated with a session of the service, the session comprising the flow; a session identifier of the session; an identifier of the UE; route information of the UE; one or more QoS requirements; one or more QoS profiles; one or more QoS parameters; at least one granularity of one or more QoS maps; a target prediction horizon.

Optionally, the QoS notification 3032 may comprise one or more of: one or more predicted values or ranges of a QoS parameter;

- the one or more QoS maps; a prediction horizon; information about a confidence of the one or more predicted values and/or predicted ranges; information about a confidence of the one or more QoS maps.

The third core network entity 303 may be further configured to receive a QoS notification subscription request from an AF, wherein the QoS notification subscription request comprises application layer information and/or configuration information for the QoS prediction notification 3032. Then, the third core network entity may be configured to generate the QoS notification subscription 3031 to a particular access network entity of the one or more access network entities 401, 402 based on the QoS notification subscription request. The third core network entity may be further configured to send the QoS notification 3032 to the AF.

Accordingly, this disclosure further proposes an access network entity 401 that provides local QoS prediction information and/or QoS notification to the core network entities. FIG. 7 shows an access network entity 401 adapted for predicting a QoS prediction for a UE according to an embodiment of the disclosure. The UE consumes a service from one or more access network entities 401, 402 comprising the access network entity 401. The UE is associated with a session of the service, wherein the session comprises a flow.

Each access network entity 401 may comprise processing circuitry (not shown) configured to perform, conduct or initiate the various operations of the access network entity 401 described herein. The processing circuitry may comprise hardware and software. The hardware may comprise analog circuitry or digital circuitry, or both analog and digital circuitry. The digital circuitry may comprise components such as application-specific integrated circuits (ASICs), field-programmable arrays (FPGAs), digital signal processors (DSPs), or multi-purpose processors. The access network entities 401 may further comprise memory circuitry, which stores one or more instruction(s) that can be executed by the processor or by the processing circuitry, in particular under control of the software. For instance, the memory circuitry may comprise a non-transitory storage medium storing executable software code which, when executed by the processor or the processing circuitry, causes the various operations of the access network entity 401 to be performed. In one embodiment, the processing circuitry comprises one or more processors and a non-transitory memory connected to the one or more processors. The non-transitory memory may carry executable program code which, when executed by the one or more processors, causes the access network entity 401 to perform, conduct or initiate the operations or methods described herein.

The access network entity 401 is configured to receive, from a core network entity 301, 302, 303, or the UE, a QoS prediction request 3011. The QoS prediction request 3011 comprises prediction configuration information. The core network entity may be the first core network entity 301 shown in FIG. 3 or FIG. 4, the second core network entity 302 shown in FIG. 4 or FIG. 5, or the third core network entity 303 shown in FIG. 6.

The access network entity 401 is further configured to generate QoS prediction information 4011 by performing a QoS prediction for the flow based on the prediction configuration information and/or by determining information for supporting a prediction based on the prediction configuration information. Then, the access network entity 401 is configured to send a prediction response 3012 comprising the QoS prediction information 4011 to the core network entity 301, 302, 303, and/or to the UE. The content of the QoS prediction request 3011, and the content of the QoS prediction information 4011, are similar as described in the previous embodiments. Details regarding the QoS maps are discussed as follows.

This disclosure proposes to use QoS maps as vectors of the predictive information generated at the RAN side. A QoS map may contain, but is not limited, to the following predicted QoS indicators:

- throughput; latency; packet loss;

Notably, “mean time to failure” may be the mean time between packet failure events, which represents the mean value of how long the network is available before it becomes unavailable (on a per-packet basis). Consecutive packet failures may be counted as one packet failure event.

The “reliability guarantee” describes the accuracy that the predicted value can guarantee, e.g., the confidence level as defined in the 3 GPP specification. The value of reliability guarantee can be one single scalar between 0 and 1 representing the required Average Reliability (AR). Optionally, the value of reliability guarantee can be a tuple (two-value vector) containing the AR and the Probability Correct Reliability (PCR), that is, a value between 0 and 1 indicating how confident the prediction system is that the AR is over the indicated value.

The “sensing accuracy” shows the difference between sensed and real values for example in range, angle, or velocity of sensed objects.

Possibly, the format of the QoS map can be per pixel or vectorized. In either case, the index of a QoS map represents a quantized value of the QoS metric for a pre-defined size of the grid (e.g., Im x 1m). FIG. 8 shows the input/output mapping of the QoS prediction map function, which is a part of the RAN QoS prediction function.

The function mapping between inputs and outputs can be done with conventional algorithms for 1D/2D/3D signal reconstruction, such as Convolutional Neural Networks (CNNs).

As an implementation example, FIG. 9 shows a function to compute the predictive QoS in the case of throughput. The blocks shown in the figure are described as follows.

Geographical Map (GM) block is the 1D/2D/3D geographical map representing the objects and optionally their impact on the signal propagation. It can be enabled by assigning the pixels corresponding to high attenuation objects a larger value, while those objects with low attenuation such as foliage or humans are assigned a smaller one.

Encoded Local Information (ELI) block encodes the local information to allow the CNN to interpret it. In particular, the ELI can be designed as a 1D/2D/3D signal with the same size as the GM, where the pixel corresponding to the location of the transmitter (BS) has the value of the transmit power. The rest of the pixels are coded according to the product of the (measured) free space path loss between the pixel and the location of the transmitter, multiplied by the beamforming vector used by the BS to transmit to a user located in the corresponding pixel. In turn, the free space path loss depends, among others, on the frequency band and the weather conditions. Optionally, the expected interference experienced at each pixel can be also encoded in the ELI as a negative value subtracted from the previous calculation.

CNN represents the algorithm itself. Depending on the dimensionality of the input features, we can define 1D/2D/3D convolutional filters as the weights of the CNN.

The QoS map represents the output of the CNN, again with the same size and dimensionality as the input features GM and ELI. The QoS map represents, in this case, the spectral efficiency per resource unit, so it needs to be multiplied with the expected number of resources that would be assigned to the user, which may be identified as E[S_av ] in FIG. 9. This information is available at the BS since it depends on the local resource scheduler and can be easily estimated. The Quantizer (Q) quantizes the estimated map to the range of values and format according to the precision request. In the following, details of the previously discussed processes are described, including several implementations.

FIG. 10 shows that the first core network entity 301 (e.g., NWDAF) requests RAN QoS prediction via the second core network entity 302 (e.g., AMF). Possibly, the first core network entity 301 may be the first core network entity 301 shown in FIG. 3, FIG. 4, or FIG. 5. The second core network entity 302 may be the second core network entity shown in FIG. 4 or FIG. 5. It should be noted that alternatively to the NWDAF, any other core network entity or AF (part of a mobile operator network or external to a mobile operator network) may also be the first core network entity 301 that requests the RAN QoS prediction. For ease of description, the NWDAF is used as an example here.

FIG. 10 shows an example of message exchanges between application and networks, when the QoS prediction request is forwarded by the NWDAF to the RAN (e.g., the RAN represents the one or more access network entities 401, 402 as shown in one of FIG. 3 - FIG. 7) via the AMF. In the following, the signaling and interfaces are described.

The request or subscription for QoS prediction can be provided by an AF and or any analytics consumer (e.g., core network entity) to the NWDAF. For example, initially, an AF (e.g., V2X Server) requests the NWDAF to provide QoS prediction information, optionally via a NEF. This request includes application layer information and configuration information for the QoS prediction, including for example Target UE Identifier(s) (e.g., Generic Public Subscription Identifier (GPSI), IP address) or UE group ID, 5QI/QoS Requirement, location information, analytics target period, etc.

The NWDAF requests core network and/or RAN domain predictions for an accurate E2E QoS estimation. Then the NWDAF, based on the received request, identifies the AMF that should be contacted to support the RAN QoS prediction information. For instance, the AMF can be identified based on the UE ID, e.g., by contacting the Unified Data Management (UDM) or any other core network entity.

Then, the NWDAF transmits the RAN QoS prediction request to the AMF. The RAN QoS prediction request includes QoS prediction configuration information provided by the AF (and/or information retrieved with the support of another core network entity) and also (optional) configuration information for granularity of the QoS map, e.g.,

- target UE IDs: identifier(s) of targeted UE(s), application identifier,

5QI QoS requirement, e.g., throughput, latency, packet error rate, reliability, etc, location information: information about current and future locations of the targeted UEs in form of a route, analytics target period: a period of time within which analytics are requested,

- the granularity of the QoS map (optional).

Notably, the granularity can be determined depending on the dynamicity of the RAN environment, which includes one or more of: update periodicity of predicted value, the geographical area where the prediction should be performed, size of the grid which one predicted value applied to, reliability requirements, i.e., the accuracy that the predicted value should guarantee.

The reliability guarantee describes the accuracy that the predicted value can guarantee, e.g., the confidence level as defined in the 3 GPP specification. The value of reliability guarantee can be one single scalar between 0 and 1 representing the required Average Reliability (AR). Alternatively, the value of reliability guarantee can be a tuple (two-value vector) containing the AR and the Probability Correct Reliability (PCR), that is, a value between 0 and 1 indicating how confident the prediction system is that the AR is over the indicated value.

Then the AMF derives information necessary for the request of RAN QoS prediction information by RAN. For instance, the AMF can identify those UE identifiers/information and/or service session identifiers/information (e.g., PDU Session Identifier) and/or QoS flow identifiers/information (e.g., QoS Flow Identifier (QFI)) that can be provided to (and used by) the RAN to identify the RAN UE identifier (e.g., Cell Radio Network Temporary Identifier (RNTI) or any other RNTI). Thus RAN-side information and/or measurements (e.g., QoS, radio, channel) can be derived through the latter for a specific UE that the QoS prediction has been requested. For that purpose, the AMF can use the information provided by the application layer that is forwarded to the AMF by the NWDAF (e.g., IP address, application ID, etc). Alternatively, the NWDAF can associate application layer information with core network information and identity UE identifiers/information and/or session identifiers/information and/or QoS flow identifiers/information that is provided to the AMF. It should be noted that the RAN side should not receive information that can allow the association of RAN side identifiers (that may be temporary) with UE private identifiers or information, e.g., to avoid user tracking, for privacy preservation, etc.

In addition, the AMF can identify the list of BSs (e.g., LTE BS, 5G BS, etc.) that the QoS prediction request should be sent to. The BSs (e.g., RAN Node ID) may be selected based e.g., on route information, AMF‘s knowledge of handover, UE location information, etc. Alternatively, the list of BSs can be retrieved by the 0AM or even be identified by the NWDAF and be part of the RAN QoS prediction request message that the NWDAF provides to the AMF.

The AMF sends the BS QoS prediction request (i.e., the partial QoS prediction request 3021 to one or more BSs. This BS QoS prediction request message may include one or more of the following information that can be used by the BS to configure the prediction determination by each BS:

- PDU Session ID: The PDU Session ID is unique per UE and is the identifier used to uniquely identify one of a UE's PDU Sessions.

QoS Flow Identifier (QFI): The QoS Flow is the finest granularity of QoS differentiation in the PDU Session. A QFI is used to identify a QoS Flow in the 5G System. User Plane traffic with the same QFI within a PDU Session receives the same traffic forwarding treatment (e.g. scheduling, admission threshold).

RAN UE NG Application Protocol (NGAP) ID: Alternatively, or in combination with the QFI, the RAN UE NGAP ID can be also retrieved by AMF and associated with the UE identifiers (provided by the AF, NWDAF). RAN UE NGAP ID can be used by the AMF to indicate to the BS for which UE the RAN and/or channel measurements need to be retrieved in order to be used for QoS prediction determination.

Sub-path: provide route information of the UE to the BS to improve the prediction accuracy (optional).

The granularity of the QoS map (optional).

Analytics target period or prediction horizon (optional).

The QFI ID could be used by a BS to derive online/“live” RAN and channel measurements of a UEthat is attached at the corresponding BS. This information could be used for more accurate QoS prediction estimation. The QFI can be identified at the AMF by retrieving it from the SMF using, e.g., an Application Id or IP filter information or any other identifier(s) of targeted UE(s) and/or the application of interest. A QoS Flow is controlled by the SMF and may be preconfigured, or established via the PDU Session Establishment procedure. Alternatively, the QFI can be identified at the NWDAF, by retrieving it from the SMF using e.g., an Application Id or IP filter information or any other identified s) of targeted UE(s) and/or the application of interest. In that case, it should be added in the RAN QoS prediction request message that the NWDAF provides to the AMF.

The prediction function at each BS predicts QoS for a specific path or area, according to the configuration received by the AMF and considering current and expected conditions, e.g., radio channel, resources, traffic. The prediction function at each BS can request additional information by any other RAN entity or external entities that could help the QoS prediction e.g., expected application-layer traffic.

The BS QoS prediction response sent by each BS to the AMF can include one or more information as follows:

- RAN predicted QoS: expected or predicted or estimated value (or a range e.g., predicted lower/upper bounds) of a QoS parameter (e.g., throughput, latency, packet loss, etc), with or without a specific confidence interval.

QoS Map: predicted QoS map or path containing the predicted information as previously described.

- Prediction horizon: The time period indicating how long in advance the QoS prediction consumer expects to receive a QoS prediction notification. How long in advance relates to the specific time when the QoS may actually change.

The QoS Prediction Aggregation Function at the AMF collects and combines BS QoS predictions responses from BSs including, for instance, the QoS estimation considering map (or path) and time horizon, and then provides such information to the NWDAF entity.

The RAN QoS prediction information from the AMF to the NWDAF can include one or more of the following information: RAN predicted QoS, QoS Map, and prediction horizon. Finally, the NWDAF prepares the QoS prediction response based on the RAN prediction response and optionally further based on other information, e.g., core network predictions or other analytics and predictions. NWDAF provides a QoS prediction response to the AF that has requested this information.

Alternatively, the QoS prediction request from the first core network entity 301 (e.g., NWDAF) can be sent via the 0AM or any other management entity to RAN as shown in FIG. 11. In such a case, the second core network entity 302 may be the 0AM entity. FIG. 11 shows an example of message exchange between an application and networks that is similar to FIG. 10.

At the RAN side, the prediction function at each BS predicts QoS for a specific path or map, according to the configuration received by the 0AM and considering current and expected conditions. The 0AM collects and combines individual predictions from different BSs to provide the QoS estimation considering map (or path) and time horizon and then provides this information to the core network entity (e.g., NWDAF). The information carried by the (RAN/BS) QoS prediction request, BS prediction response, RAN prediction response, is the same as it is described in the previous embodiment discussed in FIG. 10.

FIG. 12 shows a selection of relevant BSs and forwarding of the prediction request (FIG. 12(a)), and aggregation of QoS predictions generated at individual BSs to generate a QoS prediction report (FIG. 12(b)).

The QoS aggregation function in the second core network entity 302 (e.g., AMF or 0AM) is in charge of at least one of the following tasks:

1. Select the BSs that are relevant for the QoS prediction based on the users’ route information (e.g., provided by the NWDAF) and forward via QoS prediction request to those BSs and configure QoS prediction request (as shown in FIG. 12(a)).

2. Collect the prediction results generated by the individuals BSs and generate the RAN QoS prediction report (as shown in FIG. 12(b)).

The QoS prediction request sent to the serving base station (for instance BS1), to which the UE is currently attached, can contain the QFI of the target UE. BS thus can use the QFI to obtain the measurements linked to the particular UE. These measurements can be used as input of the prediction functions to improve the accuracy of the prediction results. In order to further protect the privacy of UE, several measures can be carried out during the QoS prediction request. For instance, the parameters of the prediction horizon sent to each BS can have a randomly selected overlapping window, in order to hinder the continuous tracing of the UE crosses multiple BS coverage. The partial route information can also be left out from the QoS prediction request.

Alternatively, the QoS Prediction Aggregation Function including the collection of QoS prediction by different BS and the preparation of RAN QoS report can take place at the RAN side, e.g., RAN controller or a BS. FIG. 13 shows examples of the location of RAN prediction functions.

According to an embodiment of this disclosure, the prediction function at the RAN side can be implemented at each BS, as shown in FIG. 13(a). In this case, the messages related to RAN QoS prediction (e.g., QoS Prediction Request and QoS Prediction Response) are exchanged between the core network entity (e.g., AMF) (or 0AM, or AF) and each BS directly. Each BS can request additional information from any other RAN entity (e.g., another BS) or external entities that could help the QoS prediction, e.g., expected application layer traffic.

Alternatively, a global RAN prediction function can be implemented as shown in FIG. 13(b). Such a global RAN prediction function can request and collect the data relevant for QoS prediction from each BS, and can also carry out the QoS prediction aggregation function as previously described. The exchange of the messages related to QoS prediction (e.g., QoS Prediction Request and QoS Prediction Response) between core network and RAN then takes place between the core network entity (e.g., AMF) and the global RAN prediction function. For instance, the global RAN prediction function can be placed at a stand-alone RAN entity, e.g., the RAN Data Analytical Function shown in FIG. 13(b). Alternatively, the global RAN prediction function can also be co-located at one of the BSs, e.g., the serving BS where the UE is attached.

According to another embodiment, the request for QoS prediction can be directly sent from the UE to the RAN side, using RRC or NAS signaling. FIG. 14 shows an example of message exchanges between UE and RAN entities (the one or more access network entities 401, 402). The QoS prediction request message can include one or more of the following information: UE ID, 5QI QoS requirement, location information (information about current and future locations of the targeted UEs in form of a route), analytics target period, and the granularity of the QoS map (optional).

Based on the provided configuration by the UE the BS (e.g., BS1, the access network entity 401) that the UE is attached undertakes to collect required local information for RAN prediction (e.g., radio channel, resources, traffic). Also, information by other RAN entities and/or a core network entity can be collected to determine at least one of the following: RAN predicted QoS, QoS Map, prediction horizon.

A BS can request and receive QoS prediction information in any of the above-mentioned forms from other BSs, and then collect the prediction results generated by the individual BSs and generate the RAN QoS prediction report, as presented in the previous embodiment discussing the QoS aggregation function.

The response of the RAN can be provided directly to the UEs using RRC or NAS signaling. In that case, the QoS prediction map is a part of the response message that RAN provides to the UE. Possibly, the UE can forward the response to the UE application layer (and/or to an application server).

Alternatively, a RAN controller (e.g., dedicated RAN entity such as the dedicated access network entity 403 shown in FIG. 4) can undertake the role to receive requests by a UE, and/or provide responses to a UE and/or collect QoS prediction responses by different BSs to build a QoS map.

FIG. 15 shows an example of message flow for notifications of RAN QoS prediction. As previously discussed, an application can subscribe (optionally via NEF) to a third core network entity 303 (e.g., PCF or SMF) to receive notifications for QoS predictions estimated by RAN for a specific application, service, or flow.

As described in FIG. 15, an AF sends the AF QoS Prediction Notification Request. This request includes application layer information and configuration information for the QoS prediction notification including one or more than one of the following information: UE identified s), application identifier, AF identifier, 5QI QoS Requirement, flow description(s), or external application identifier, location information, analytics target period, the granularity of QoS map, etc.

The PCF shall enable predicted QoS notification and include the configuration information in the Policy and Charging Control (PCC) rule sent to the SMF. If the PCF determines that the SMF needs updated policy information, the PCF issues, e.g., a Npcf SMPolicyControl UpdateNotify request with updated policy information about the PDU Session as described in the PCF initiated SM Policy Association Modification procedure of the 3 GPP specification.

The SMF provides a “predicted QoS change notification” subscription to one or more RAN entities that trigger one or more RAN entities to predict the QoS of a specific UE (or group of UEs) and notify the SMF for potential predicted changes. This can be a dedicated message and or part of PDU Session establishment and/or modification message.

The prediction function at RAN (e.g., BS) predicts QoS for a specific path or map, according to the configuration received by the SMF and considering current and expected conditions e.g., radio channel, resources, traffic. A RAN entity, (e.g., BS) shall send a QoS Prediction Notification towards SMF when it is predicted that the QoS of a UE at the cell that is currently attached and/or at a future cell will change (e.g., change of QoS level, change of 5QI or the above/below threshold set by the application). The SMF can combine the RAN prediction with other information and or predictions, e.g., core network QoS prediction information.

The SMF shall also provide to the PCF the predicted QoS information and whether the one more QoS parameters cannot be fulfilled. The PCF can also forward the notification to the application.

The provided notification can be in one of the following forms: RAN predicted QoS, QoS maps including predicted QoS maps , or paths containing the predicted information, and prediction horizon.

SMF undertakes to forward this notification response to the application. The SMF can also send the notification for predicted QoS information of the UE, transparently through NG-RAN (e.g., using NAS message). For instance, about predicted changes in the QoS parameters (i.e. 5QI, Guaranteed Flow Bit Rate (GFBR), Maximum Flow Bit Rate (MFBR)) that the NG-RAN is currently fulfilling.

Alternatively, this QoS prediction notification service can be established also during the PDU session establishment or PDU Session modification phase as described in the 3GPP specification. The SMF can enable the notification of QoS prediction when the QoS Notification Control parameter is set in the rules (received from the PCF) that is bound to the QoS Flow. The Notification control parameter is signaled to the NG-RAN as part of the QoS profile.

To summarize, the proposed prediction mechanism of this disclosure includes four processes: sending QoS report request to RAN, generation of QoS information in RAN, aggregation of QoS information into a single QoS report, and delivery of QoS report.

This disclosure enables the exchange of precise and fresh predictive QoS information between RAN, CN, and AFs deployed external to a mobile operator network while preserving isolation and privacy. It includes the following aspects.

The definition of the RAN/CN signaling and functionality that allows a core network function, e.g., NWDAF, to request RAN QoS prediction information from one or more RAN entities, e.g., BSs. The format specification of how the predictive QoS information can be reported from RAN to core network while preserving isolation and privacy between domains. It is achieved by introducing QoS maps as vectors of information. Finally, the mechanism to combine the predictive QoS information coming from several RAN entities, e.g. BSs, and generate a report for the analytics consumer.

This disclosure thus supports providing precise and fresh predictive QoS information to analytics consumers from RAN. In addition, this disclosure allows to keep isolation between different domains in the network (RAN/CN/AF), and anonymization of UE information to keep privacy. Further, the disclosure also enables aggregation of predictive analytics from several sources, e.g. several BSs.

FIG. 16 shows a method 1600 according to an embodiment of the disclosure, particularly for generating a full QoS prediction for a UE consuming a service from one or more access network entities 401, 402. In a particular embodiment, the method 1600 is performed by the first core network entity 301 shown in FIG. 3 or FIG. 4. The method 1600 comprises a step 1601 of providing a QoS prediction request 3011 to each of one or more access network entities 401, 402. Possibly, the one or more access network entities 401, 402 are the one or more access network entities shown in FIG. 3 or FIG. 4. The QoS prediction request 3011 comprises prediction configuration information. The method 1600 further comprises a step 1602 of receiving a prediction response 3012 from each of the one or more access network entities 401, 402, wherein the prediction response 3012 comprises QoS prediction information 4011 of the access network entity. The method 1600 further comprises a step 1603 of deriving the full QoS prediction 3013 based on the QoS prediction information 4011 in the one or more prediction responses 3012.

FIG. 17 shows a method 1700 according to an embodiment of the disclosure, particularly for supporting a QoS prediction for a UE consuming a service from one or more access network entities 401, 402. In a particular embodiment, the method 1700 is performed by a second core network entity 302 shown in FIG. 4 or FIG. 5. The method 1700 comprises a step 1701 of receiving a QoS prediction request 3011 from a first core network entity 301. The first core network entity 301 may be the first core network entity shown in FIG. 4 or FIG. 5. The QoS prediction request 3011 comprises prediction configuration information. The method 1700 further comprises a step 1702 of providing, to each of one or more access network entities 401, 402, a partial QoS prediction request 3021 based on the QoS prediction request 3011. Then, the method 1700 comprises a step 1703 of receiving a prediction response 3012 from each of the one or more access network entities 401, 402. The prediction response 3012 comprises QoS prediction information 4011 of the access network entity. The method 1700 further comprises a step 1704 of sending a QoS prediction report 3022 to the first core network entity 301, wherein the QoS prediction report 3022 comprises the prediction response 3012 from each of the one or more access network entities 401, 402. Possibly, the one or more access network entities 401, 402 are the one or more access network entities shown in FIG. 4 or FIG. 5.

FIG. 18 shows a method 1800 according to an embodiment of the disclosure, particularly for subscribing to a QoS prediction notification. In a particular embodiment, the method 1800 is performed by a third core network entity 303 shown in FIG. 6. The method 1800 comprises a step 1801 of sending a QoS notification subscription 3031 to each access network entity of one or more access network entities 401, 402. Possibly, the one or more access network entities 401, 402 are the one or more access network entities shown in FIG. 4, FIG. 5, or FIG. 6. The QoS notification subscription 3031 instructs the access network entity to send a QoS notification 3032 to the third core network entity 303, when the access network entity predicts a change on the QoS prediction for a UE consuming a service of the one or more access network entities 401, 402. The method 1800 further comprises a step 1802 of receiving the QoS notification 3032 from the one or more access network entities 401, 402.

FIG. 19 shows a method 1900 according to an embodiment of the disclosure, particularly for predicting a QoS for a UE consuming a service of one or more access network entities 401, 402 including the access network entity 401. Notably, the UE is associated with a session of the service, and the session comprises a flow. In a particular embodiment, the method 1900 is performed by the access network entity 401 shown in any of FIG. 3 - FIG. 7. The method 1900 comprises a step 1901 of receiving, from a core network entity 301, 302, 303 or from the UE, a QoS prediction request 3011. The QoS prediction request 3011 comprises prediction configuration information. Possibly, the core network entity may be the first core network entity 301 shown in FIG. 3 or FIG. 4, the second core network entity 302 shown in FIG. 4 or FIG. 5, or the third core network entity 303 shown in FIG. 6.

The method 1900 comprises a step 1902 of generating QoS prediction information 4011 by performing a QoS prediction on the flow based on the prediction configuration information and/or by determining information for supporting a prediction based on the prediction configuration information. The method 1900 further comprises a step 1903 of sending a prediction response 3012 comprising the QoS prediction information 4011 to the core network entity 301, 302, 303 and/or to the UE.

The present disclosure has been described in conjunction with various embodiments as examples as well as implementations. However, other variations can be understood and effected by those persons skilled in the art and practicing the claimed embodiments of the disclosure, from the studies of the drawings, this disclosure and the independent claims. In the claims as well as in the description the word “comprising” does not exclude other elements or steps and the indefinite article “a” or “an” does not exclude a plurality. A single element or other unit may fulfill the functions of several entities or items recited in the claims. The mere fact that certain measures are recited in the mutual different dependent claims does not indicate that a combination of these measures cannot be used in an advantageous implementation. Furthermore, any method according to embodiments of the disclosure may be implemented in a computer program, having code means, which when run by processing means causes the processing means to execute the steps of the method. The computer program is included in a computer-readable medium of a computer program product. The computer-readable medium may comprise essentially any memory, such as a ROM (Read-Only Memory), a PROM (Programmable Read-Only Memory), an EPROM (Erasable PROM), a Flash memory, an EEPROM (Electrically Erasable PROM), or a hard disk drive.

Moreover, it is realized by the skilled person that embodiments of the first core network entity

301, the second core network entity 302, the third core network entity 303, or the access network entity 401, comprise the necessary communication capabilities in the form of e.g., functions, means, units, elements, etc., for performing the solution. Examples of other such means, units, elements, and functions are: processors, memory, buffers, control logic, encoders, decoders, rate matchers, de-rate matchers, mapping units, multipliers, decision units, selecting units, switches, interleavers, de-interleavers, modulators, demodulators, inputs, outputs, antennas, amplifiers, receiver units, transmitter units, DSPs, trellis-coded modulation (TCM) encoder, TCM decoder, power supply units, power feeders, communication interfaces, communication protocols, etc. which are suitably arranged together for performing the solution.

Especially, the processor(s) of the first core network entity 301, the second core network entity

302, the third core network entity 303, or the access network entity 401 may comprise, e.g., one or more instances of a Central Processing Unit (CPU), a processing unit, a processing circuit, a processor, an Application Specific Integrated Circuit (ASIC), a microprocessor, or other processing logic that may interpret and execute instructions. The expression “processor” may thus represent a processing circuitry comprising a plurality of processing circuits, such as, e.g., any, some or all of the ones mentioned above. The processing circuitry may further perform data processing functions for inputting, outputting, and processing of data comprising data buffering and device control functions, such as call processing control, user interface control, or the like.

Claims

1. A first core network entity (301) for generating a full Quality of Service, QoS, prediction for a user equipment, UE, consuming a service from one or more access network entities (401, 402), the first core network entity (301) being configured to: provide, to each of the one or more access network entities (401, 402), a QoS prediction request (3011), wherein the QoS prediction request (3011) comprises prediction configuration information; receive a prediction response (3012) from each access network entity (401) of the one or more access network entities (401, 402), wherein the prediction response (3012) comprises QoS prediction information (4011) of the access network entity (401); and derive the full QoS prediction (3013) based on the QoS prediction information (4011) in the one or more prediction responses (3012).

2. The first core network entity (301) according to claim 1, wherein the prediction configuration information comprises one or more of: an identifier of the UE; route information of the UE; one or more QoS requirements; a target prediction horizon.

3. The first core network entity (301) according to claim 1 or 2, wherein the QoS prediction information (4011) of the access network entity (401) comprises at least one of a partial QoS prediction performed by the access network entity (401) and/or information from the access network entity (401) for supporting a prediction performed by the first core network entity (301).

4. The first core network entity (301) according to claim 3, further configured to: derive the full QoS prediction (3013) by performing the prediction based on the information from the one or more access network entities (401, 402) for supporting the prediction, and/or by combining the partial QoS predictions performed by the one or more access network entities (401, 402).

42

5. The first core network entity (301) according to claim 3 or 4, wherein the partial QoS prediction comprises one or more of: one or more predicted values and/or predicted ranges of a QoS parameter; one or more QoS maps; a prediction horizon; information about confidence of the one or more predicted values and/or predicted ranges; information about confidence of the one or more QoS maps.

6. The first core network entity (301) according to claim 5, wherein the QoS prediction request (3011) further comprises configuration information regarding at least one granularity of the one or more QoS maps.

7. The first core network entity (301) according to one of the claims 1 to 6, wherein providing the QoS prediction request (3011) comprises: sending the QoS prediction request (3011) to each of the one or more access network entities (401, 402) via a second core network entity (302), or via a dedicated access network entity (403); and wherein receiving the prediction response (3012) from each access network entity (401) comprises: receiving a QoS prediction report (3022) from the second core network entity (302), or from the dedicated access network entity (403), wherein the QoS prediction report (3022) comprises the prediction response (3012) from each of the one or more access network entities (401, 402).

8. The first core network entity (301) according to claim 7, wherein deriving the full QoS prediction (3013) based on the QoS prediction information (4011) comprises: deriving the full OoS prediction (3013) based on the QoS prediction report (3022).

9. The first core network entity (301) according to one of the claims 1 to 8, configured to: receive an initial QoS prediction request from an application function or another core network entity, wherein the initial QoS prediction request comprises application layer information and/or the prediction configuration information;

43 generate the QoS prediction request (3011) based on the initial QoS prediction request; and send a QoS prediction response (3012) comprising the full QoS prediction (3013) to the application function or the another core network entity in response to the initial QoS prediction request.

10. The first core network entity (301) according to claim 9, wherein the second core network entity (302) is an access mobility function, and the first core network is further configured to: identify the second core network entity (302) based on the initial QoS prediction request.

11. The first core network entity (301) according to one of the claims 1 to 10, wherein each of the one or more QoS maps comprises one or more of the following QoS indicators:

- throughput; latency; packet loss;

12. A second core network entity (302) for supporting a Quality of Service, QoS, prediction for a user equipment, UE, consuming a service from one or more access network entities (401, 402), the second core network entity (302) being configured to: receive, from a first core network entity (301), a QoS prediction request (3011), wherein the QoS prediction request (3011) comprises prediction configuration information; provide, to each of one or more access network entities (401, 402), a partial QoS prediction request (3021) based on the QoS prediction request (3011); receive a prediction response (3012) from each of the one or more access network entities (401, 402), wherein the prediction response (3012) comprises QoS prediction information (4011) of the access network entity (401); and

44 send a QoS prediction report (3022) to the first core network entity (301), wherein the QoS prediction report (3022) comprises the prediction response (3012) from each of the one or more access network entities (401, 402).

13. The second core network entity (302) according to claim 12, wherein the prediction configuration information comprises one or more of an identifier of the UE; route information of the UE; one or more QoS requirements; and a target prediction horizon.

14. The second core network entity (302) according to claim 12 or 13, the QoS prediction information (4011) of the access network entity (401) comprises at least one of a partial QoS prediction performed by the access network entity (401) and/or one or more QoS parameters of the access network entity (401).

15. The second core network entity (302) according to claim 14, wherein the partial QoS prediction comprises one or more of one or more predicted values or ranges of a QoS parameter; one or more QoS maps; a prediction horizon; information about confidence of the one or more predicted values and/or predicted ranges; information about confidence of the one or more QoS maps.

16. The second core network entity (302) according to claim 15, wherein the QoS prediction request (3011) further comprises configuration information regarding at least one granularity of the one or more QoS maps.

17. The second core network entity (302) according to one of the claims 13 to 16, and wherein the second core network entity (302) is configured to: generate the partial QoS prediction request (3021) for a particular access network entity of the one or more access network entities (401, 402), wherein the partial QoS prediction request (3021) comprises one or more of an identifier of a flow, wherein the UE is associated with a session of the service, the session comprising the flow; a session identifier of the session; the identifier of the UE; route information of the UE, or partial route information of the UE corresponding to the particular access network entity;

- the at least one granularity of the one or more QoS maps;

- the target prediction horizon.

18. The second core network entity (302) according to one of the claims 12 to 17, further configured to: determine the one or more access network entities (401, 402) based on the QoS prediction request (3011) and/or based further on handover knowledge from an access mobility function.

19. The second core network entity (302) according to claim 18, wherein the second core network entity (302) is an access mobility function.

20. The second core network entity (302) according to one of the claims 12 to 18, wherein the second core network entity (302) is a network management entity.

21. The second core network entity (302) according to claim 20, wherein the network management entity comprises one or more operations administration and maintenance, 0AM, entities.

22. The second core network entity (302) according to one of the claims 12 to 21, further configured to: generate the QoS prediction report (3022) by performing a QoS estimation based on the prediction response (3012) from each of the one or more access network entities (401, 402).

23. A third core network entity (303) for subscribing to a Quality of Service, QoS, prediction notification, the third core network entity (303) being configured to: send, to each access network entity (401) of one or more access network entities (401, 402), a QoS notification subscription (3031), wherein the QoS notification subscription (3031) instructs the access network entity (401) to send a QoS notification (3032) to the third core network entity (303), when the access network entity (401) predicts a change on the QoS prediction for a user equipment, UE consuming a service of the one or more access network entities (401, 402); and receive the QoS notification (3032) from the one or more access network entities (401, 402).

24. The third core network entity (303) according to claim 21, wherein the QoS notification subscription (3031) comprises one or more of an identifier of a flow, wherein the UE is associated with a session of the service, the session comprising the flow; a session identifier of the session; an identifier of the UE; route information of the UE; one or more QoS requirements; one or more QoS profiles; one or more QoS parameters; at least one granularity of one or more QoS maps; a target prediction horizon.

25. The third core network entity (303) according to claim 24, wherein the QoS notification (3032) comprises one or more of one or more predicted values or ranges of a QoS parameter;

26. The third core network entity (303) according to one of the claims 23 to 25, further configured to: receive a QoS notification subscription request from an application function, wherein the QoS notification subscription request comprises application layer information and/or configuration information for the QoS prediction notification (3032);

47 generate the QoS notification subscription (3031) to a particular access network entity of the one or more access network entities (401, 402) based on the QoS notification subscription request; and send the QoS notification (3032) to the application function.

27. An access network entity (401) for predicting a Quality of Service, QoS, for a user equipment, UE, consuming a service of one or more access network entities (401, 402) including the access network entity (401), wherein the UE is associated with a session of the service, the session comprising a flow, and wherein the access network entity (401) is configured to: receive, from a core network entity (301, 302, 303) or from the UE, a QoS prediction request (3011), wherein the QoS prediction request (3011) comprises prediction configuration information; generate QoS prediction information (4011) by performing a QoS prediction for the flow based on the prediction configuration information and/or by determining information for supporting a prediction based on the prediction configuration information; and send a prediction response (3012) comprising the QoS prediction information (4011) to the core network entity (301, 302, 303) and/or to the UE.

28. The access network entity (401) according to claim 27, wherein the QoS prediction request (3011) comprises one or more of an identifier of the flow; a session identifier of the session; the identifier of the UE; partial route information of the UE corresponding to the access network entity;

- the at least one granularity of the one or more QoS maps;

29. The access network entity (401) according to claim 27 or 28, wherein the QoS prediction information (4011) comprises one or more of one or more predicted values or ranges of a QoS parameter;

- the one or more QoS maps;

48 a prediction horizon; information about confidence of the one or more predicted values and/or predicted ranges; information about confidence of the one or more QoS maps.

30. The access network entity (401) according to claim 29, wherein each of the one or more QoS maps comprises one or more of the following QoS indicators:

- throughput; latency; packet loss;

31. The access network entity (401) according to one of the claims 27 to 30, wherein when the QoS prediction request (3011) is received from the UE, the access network entity is further configured to: send a further QoS prediction request to a further access network entity; receive a further prediction response from the further access network entity; generate a combined prediction response based on the further prediction response and the prediction response (3012); and send the combined prediction response to the UE.

32. A method (1600) performed by a first core network entity (301) for generating a full Quality of Service, QoS, prediction for a user equipment, UE, consuming a service from one or more access network entities (401, 402), comprising: providing (1601), to each of one or more access network entities (401, 402), a QoS prediction request (3011), wherein the QoS prediction request (3011) comprises prediction configuration information;

49 receiving (1602) a prediction response (3012) from each of the one or more access network entities (401, 402), wherein the prediction response (3012) comprises QoS prediction information (4011) of the access network entity; and deriving (1603) the full QoS prediction (3013) based on the QoS prediction information (4011) in the one or more prediction responses (3012).

33. A method (1700) performed by a second core network entity (302) for supporting a Quality of Service, QoS, prediction for a user equipment, UE, consuming a service from one or more access network entities (401, 402), comprising: receiving (1701), from a first core network entity (301), a QoS prediction request (3011), wherein the QoS prediction request (3011) comprises prediction configuration information; providing (1702), to each of one or more access network entities (401, 402), a partial QoS prediction request (3021) based on the QoS prediction request (3011); receiving (1703) a prediction response (3012) from each of the one or more access network entities (401, 402), wherein the prediction response (3012) comprises QoS prediction information (4011) of the access network entity; and sending (1704) a QoS prediction report (3022) to the first core network entity (301), wherein the QoS prediction report (3022) comprises the prediction response (3012) from each of the one or more access network entities (401, 402).

34. A method (1800) performed by a third core network entity (303) for subscribing to a Quality of Service, QoS, prediction notification, comprising: sending (1801), to each access network entity of one or more access network entities (401, 402), a QoS notification subscription (3031), wherein the QoS notification subscription (3031) instructs the access network entity to send a QoS notification (3032) to the third core network entity (303), when the access network entity predicts a change on the QoS prediction for a user equipment, UE consuming a service of the one or more access network entities (401, 402); and receiving (1802) the QoS notification (3032) from the one or more access network entities (401, 402).

35. A method (1900) performed by an access network entity (401) for predicting a Quality of Service, QoS, for a user equipment, UE, consuming a service of one or more access network

50 entities (401, 402) including the access network entity (401), wherein the UE is associated with a session of the service, the session comprising a flow, and wherein the method comprises: receiving (1901), from a core network entity (301, 302, 303) or from the UE, a QoS prediction request (3011), wherein the QoS prediction request (3011) comprises prediction configuration information; generating (1902) QoS prediction information (4011) by performing a QoS prediction on the flow based on the prediction configuration information and/or by determining information for supporting a prediction based on the prediction configuration information; and sending (1903) a prediction response (3012) comprising the QoS prediction information (4011) to the core network entity (301, 302, 303) and/or to the UE.

36. A computer program product comprising a program code for carrying out, when implemented on a processor, the method (1600, 1700, 1800, 1900) according to one of the claims 32 to 35.

51